Shield deviation rectification parameter recommendation method and system based on association rules and machine learning

Abstract

The invention provides a shield deviation rectification parameter recommendation method and system based on association rules and machine learning, and belongs to the technical field of shield deviation rectification parameter recommendation, and the method comprises the steps: obtaining required shield construction deviation rectification section data, and constructing a shield deviation rectification data set based on the obtained data; the shield deviation correction data set is preprocessed; discretization processing is conducted on data in the preprocessed shield deviation correction data set, and a tunneling data interval and a deviation correction amount interval of each ring in the shield deviation correction process are obtained; screening an association rule meeting a set minimum confidence threshold value from the obtained tunneling data interval and the deviation correction interval of each ring; and performing equal division on each shield deviation correction interval, permutation and combination of the data, inputting the data into a shield axis deviation correction model, performing statistical analysis on a model prediction result, and selecting a group of input operation data with the minimum distance true value as a ring deviation correction recommended value.

Description

Technical Field

[0001] The present invention belongs to the technical field of shield deviation correction parameter recommendation, and in particular relates to a shield deviation correction parameter recommendation method and system based on association rules and machine learning. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the long-term stable development of the economy, the urbanization rate has increased year by year, the intrinsic demand for underground space has continued to increase, and the volume and complexity of underground engineering construction have also continued to rise. In the urban development blueprint that focuses on the deep development and utilization of underground space, tunnel engineering construction is bound to maintain a growth trend for a long period of time in the future. Among the many tunnel construction methods, the shield method has become the main construction method for tunnel construction due to its advantages such as short construction period, low carbon and environmental protection, high safety, and less disturbance to the surface environment.

[0004] During the tunneling process of the shield machine, due to the complex geological environment and stress conditions of the underground construction site, the deviation of the shield axis is inevitable. Once a large deviation occurs and is not corrected in time, it may cause serious consequences such as segment misalignment and rupture, which in turn affects the efficiency and quality of construction. At present, the correction method of the shield axis is mainly controlled by manual operation. The driver obtains the current position and posture information of the shield machine through the guidance system. When the deviation between the shield machine posture and the tunnel design axis exceeds the allowable range, the system will issue an alarm signal. According to the actual deviation, the driver adjusts the relevant construction parameters to make the shield machine excavation path return to the tunnel design axis. However, manual operation control is highly subjective, the proficiency of shield drivers is different, and the effect is not good under complex geological conditions. The actual correction route often shows a "snake-shaped" trend. Therefore, it is urgent to study more efficient and accurate shield correction methods.

[0005] In recent years, the reverse circular geometric correction curve and segment arrangement correction method have been widely studied. However, these two methods rely heavily on manual experience in the process of setting shield machine parameters, resulting in uneven correction accuracy. Automation technology has also been deeply discussed in the field of shield machine construction, but when faced with complex and changeable geological conditions, it is necessary to continuously adjust and optimize the algorithm to adapt to different construction environments, resulting in low correction efficiency. With the rapid development of artificial intelligence technology, machine learning has shown great advantages in processing the complex relationship between input features and output values, making it gradually used in the field of shield construction, but there is still little research on shield excavation posture correction.

[0006] The inventors discovered in their research that there are related solutions in the prior art for obtaining shield correction parameters using algorithms, but the technical problems with the above technologies are: current technologies generally pursue accuracy, so the calculation efficiency is often low, and cannot meet the needs of accuracy and efficiency at the same time, resulting in poor actual on-site application effects; and they fail to effectively explore the correlation between parameters and the correction amount, and cannot combine multi-parameter recommendations to achieve flexible correction. Summary of the invention

[0007] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a shield correction parameter recommendation method based on association rules and machine learning, and the optimal correction excavation parameter combination is used as the recommended parameter of the correction method, thereby significantly improving the quality and efficiency of the shield machine when performing axis correction.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] First, a method for recommending shield deviation correction parameters based on association rules and machine learning is disclosed, including:

[0010] Obtain the required shield construction correction section data and construct a shield correction data set based on the acquired data;

[0011] Preprocess the shield deviation correction data set;

[0012] Discretize the data in the preprocessed shield deviation correction data set to obtain the tunneling data interval and the deviation correction interval of each ring during the shield deviation correction process;

[0013] A minimum confidence threshold is set, and association rules that meet the set minimum confidence threshold are selected from the obtained excavation data interval and the deviation correction interval of each ring;

[0014] All association rules that meet the minimum confidence threshold are integrated to obtain the shield machine deviation correction interval corresponding to the tunneling data interval during the final shield machine deviation correction;

[0015] Each shield correction interval is divided into equal parts and the data is arranged and combined, input into the shield axis correction model, and the model prediction results are statistically analyzed. The set of input operation data with the smallest distance from the true value is selected as the recommended correction value for this loop.

[0016] As a further technical solution, the correction section data is extracted from the key excavation parameters during the shield machine construction process and integrated into the shield correction data set.

[0017] As a further technical solution, the thrust and stroke of each group of propulsion cylinders and the correction amount of the previous ring are used as input data of the shield axis correction model, and the horizontal deviation of the shield machine's incision, the vertical deviation of the incision, the horizontal deviation of the shield machine's shield tail, and the vertical deviation of the shield tail are used as output data of the machine learning regression model.

[0018] As a further technical solution, the shield deviation correction data set is preprocessed, including:

[0019] Taking ring as unit, all data in the shield machine deviation correction data set are divided according to the ring number, and the deviation correction amount of each ring of the shield machine is obtained through analysis and calculation, and the average value of each ring is calculated so that each ring corresponds to one of the data;

[0020] The box plot method is used to detect parameter outliers. The upper and lower bounds exceeding the gray dotted lines in the figure are regarded as outliers and are removed.

[0021] As a further technical solution, the preprocessed shield correction data set is discretized. Specifically, the equal-width method is used to convert the continuous data of the shield correction data set into discrete data, and the excavation data interval and the correction amount interval of each ring during the shield correction process are obtained.

[0022] As a further technical solution, the obtained excavation data interval and the deviation correction interval of each ring are screened to find the association rules that meet the set minimum confidence threshold, including:

[0023] The minimum support threshold of the FP-Growth association rule algorithm is set, and each data below the support threshold is eliminated to mine the frequent item sets in the shield deviation correction data set;

[0024] The confidence of each frequent item set is calculated to generate association rules, and the minimum confidence threshold is set to filter out association rules that meet the set minimum confidence threshold.

[0025] As a further technical solution, after determining the association rules between the final shield machine correction amount and the operating parameters, all association rules that meet the threshold mined by the FP-Growth algorithm are integrated to obtain the shield machine correction amount interval corresponding to the excavation data interval during the final shield machine correction.

[0026] As a further technical solution, when the shield machine performs axis correction, the total number of correction rings and the correction amount of each ring are calculated through data analysis, and the correction interval of the ring correction amount is determined according to the association rules, and the excavation data interval corresponding to each ring correction amount is obtained.

[0027] Secondly, a shield deviation correction parameter recommendation system based on association rules and machine learning is disclosed, including:

[0028] The data acquisition and analysis module is configured to: obtain the required shield construction correction section data and construct a shield correction data set based on the obtained data;

[0029] The data preprocessing module is configured to: preprocess the shield deviation correction data set;

[0030] The association rule mining and analysis module is configured to: discretize the data in the preprocessed shield deviation correction data set to obtain the tunneling data interval and the deviation correction amount interval of each ring during the shield deviation correction process;

[0031] A minimum confidence threshold is set, and association rules that meet the set minimum confidence threshold are selected from the obtained excavation data interval and the deviation correction interval of each ring;

[0032] The shield machine deviation correction parameter recommendation module is configured to: integrate all association rules that meet the minimum confidence threshold, and obtain the shield machine deviation correction amount interval corresponding to the tunneling data interval during the final shield machine deviation correction;

[0033] Each shield correction interval is divided into equal parts and the data is arranged and combined, input into the shield axis correction model, and the model prediction results are statistically analyzed. The set of input operation data with the smallest distance from the true value is selected as the recommended correction value for this loop.

[0034] One or more of the above technical solutions have the following beneficial effects:

[0035] The technical solution of the present invention extracts the correction section data during the excavation process of the shield machine, constructs a shield historical correction data set after data preprocessing, and uses the FP-Growth association rule algorithm to mine the association rules between the shield deviation parameters and the operating parameters. This is used as the basis for recommending operating parameters for training the shield correction prediction model, thereby improving the efficiency and further improving the correction accuracy of the shield machine.

[0036] The technical solution of the present invention analyzes the excavation parameters that the shield machine driver mainly adjusts when correcting the axis, thereby optimizing the input data of the model, and takes whether the excavation route of the incision and the shield tail fits the design axis of the tunnel as the basis for judging whether the shield machine has deviated. A shield correction parameter recommendation method based on the lightweight gradient boosting decision tree (LightGBM) machine learning model is established, and a genetic algorithm (GA) is used to search for the global optimal solution of the model hyperparameters in parallel and efficiently. Through the LightGBM algorithm with relatively lower memory consumption and efficient parallel training, accurate and rapid prediction of the shield machine axis correction amount is achieved.

[0037] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0039] Figure 1 A schematic diagram of the structure of a method for recommending shield deviation correction parameters for an example of the present invention;

[0040] Figure 2 A flow chart of a method for recommending shield deviation correction parameters for an example implementation of the present invention;

[0041] Figure 3 A data schematic diagram of the deviation section and the deviation correction section of the horizontal deviation of the incision in an example of the present invention;

[0042] Figure 4 This is a diagram showing the effect of detecting data outliers using the box plot method in an example of the present invention;

[0043] Figure 5 This is a schematic diagram of the genetic algorithm optimization model hyperparameters for an implementation example of the present invention. DETAILED DESCRIPTION

[0044] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0045] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0046] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0047] Glossary:

[0048] Genetic Algorithm, GA: genetic algorithm;

[0049] Light Gradient Boosting Machine, LightGBM: Lightweight gradient boosting decision tree.

[0050] Embodiment 1

[0051] The shield correction parameter recommendation method based on association rule mining and machine learning has the following workflow: Figure 2 As shown, the specific steps include:

[0052] Step 1: Select the thrust of the propulsion cylinder and the stroke of the propulsion cylinder that the shield driver mainly adjusts when correcting the shield axis as the model input data; the specific correction amount of a certain ring of the shield machine will also take into account the correction amount of the previous ring, so the correction amount of the previous ring is also included as the model input data. Select the horizontal deviation of the cut, the vertical deviation of the cut, the horizontal deviation of the shield tail, and the vertical deviation of the shield tail as the model output data. Figure 3 As shown in the figure, the correction section data is analyzed and extracted from the excavation process and integrated into the shield correction data set.

[0053] In this step, since it is concluded according to the investigation and analysis that the shield driver mainly adjusts the thrust cylinder when correcting the deviation, the deviation of the shield is mainly judged through the VMT guide system. The guide system mainly measures four values: horizontal deviation of the incision, vertical deviation of the incision, horizontal deviation of the shield tail, and vertical deviation of the shield tail to judge the deviation of the shield machine.

[0054] Since the correction operation of the shield machine generally requires multiple rings to solve, it is necessary to extract the data in ring units so that the correction amount of the previous ring of the current ring can be known. The first ring can be set to 0.

[0055] The horizontal deviation of the incision, the vertical deviation of the incision, the horizontal deviation of the shield tail and the vertical deviation of the shield tail are collected and subsequently used to calculate the mean deviation of each ring. During the correction process, these values ​​should be as close to 0 as possible.

[0056] In this implementation example, the collected data is visualized to find the deviation interval, extract the deviation interval and splice it into a data set. The shield correction data set includes: 12 sets of data for the thrust and stroke of six propulsion cylinders, plus the correction amount of the previous ring, a total of 13 sets of input data, and 4 sets of output data including the horizontal deviation of the cut, the vertical deviation of the cut, the horizontal deviation of the shield tail and the vertical deviation of the shield tail, a total of 17 sets of data.

[0057] Step 2: Preprocess the shield machine deviation correction data set to remove some noise and abnormal data, ensure data accuracy and improve data quality, thereby improving the accuracy of model prediction. Extract data in ring units, calculate the deviation correction amount of each ring of the shield machine through analytical calculations, and calculate the average value of each ring. Figure 4 As shown in the figure, the box plot method is used to detect outliers in the data. The upper and lower thresholds of the box plot method are set. If the upper and lower limits exceed the gray dotted line in the figure, that is, the threshold, they are regarded as outliers and are removed.

[0058] In one or implementation examples, when calculating the deviation correction amount of each ring of the shield machine, the deviation amount at the beginning of the current ring is subtracted from the deviation amount at the end, that is, the deviation correction amount for each ring.

[0059] In addition, the input parameters calculate the mean of each ring, and the output parameters calculate the variation of each ring, so that each ring corresponds to a set of data in the data set.

[0060] Step 3: Use the equal-width method to convert the continuous data of the shield correction data set into discrete data, and obtain the excavation data interval and the correction amount interval of each ring during the shield correction process, so as to simplify the data structure and improve the efficiency of the subsequent association rule algorithm.

[0061] In this step of the present embodiment, the shield correction data set is concentrated: each ring has a corresponding set of data, and the specific conversion process is to disperse the data into equal distances. For example, if the original data is 672.5 and the set distance is 500, then the interval it is in is [500, 1000].

[0062] It should be noted that the data is discrete, so subsequent search and merging operations will become simpler, and the mined association rules will be more intuitive and easier to analyze, which can significantly improve the efficiency of the method.

[0063] After the data is discretized, it becomes an interval form. It can be discretized by equal width, equal frequency, centroid method, etc. This example adopts equal width discretization.

[0064] After obtaining the deviation correction value of each link, the deviation correction value interval is obtained by the same discretization method. For example, -8 is in the interval [-9, -6]. After discretization, the data becomes the form of interval, which is equivalent to each data becoming the form of interval.

[0065] Step 4: Set the minimum support threshold of the FP-Growth association rule algorithm, remove each data below the support threshold, and mine the frequent item sets in the shield correction data set processed in step 3. The frequent item set is a collection of items that meet the support threshold. Support is used to measure the frequency of an item set appearing in the data set. Support is used to filter out frequent item sets. The calculation formula is as follows:

[0066]

[0067] In the formula: Support represents support, which indicates the frequency of occurrence of a certain item set; σ(X∪Y) represents the number of item sets containing both X and Y; N represents the total number of sets.

[0068] Calculate the confidence of each frequent item set to generate association rules, set the minimum confidence threshold, and filter out the association rules that meet the set minimum confidence threshold. The association rules are obtained based on the frequent item sets found and the item sets that meet the confidence threshold. For example, the input data of the propulsion cylinder of group A in the interval [4000,7000] corresponds to the vertical deviation interval of the cutout [-9,-6]. Confidence represents an evaluation of the credibility of an association rule. The calculation formula is as follows:

[0069]

[0070] Where: Confidence represents confidence, which indicates the frequency of Y appearing when X appears; σ(X) represents the number of item sets containing X.

[0071] Step 5: Integrate all the association rules that meet the threshold mined by the FP-Growth algorithm to obtain the shield correction amount interval corresponding to the excavation data interval during the final shield machine correction; the subsequent corresponding step 9, finds the corresponding correction interval through the predicted correction amount, and gives the recommended values ​​of each group of thrust cylinders through the obtained association rules.

[0072] In the above steps of this embodiment, different correction amount intervals correspond to different input data intervals; integration refers to arranging each obtained association rule together.

[0073] Step 6: After normalizing the shield deviation correction data set obtained in step 2, 70% of it is used as a training set for training the model, and the LightGBM shield axis deviation correction training model is obtained to establish a nonlinear mathematical relationship between the tunneling data (thrust and stroke of each group of propulsion cylinders) and the deviation correction amount of each ring;

[0074] Step 7: Figure 5 As shown in the figure, a genetic algorithm (GA) is used to optimize the hyperparameters of the LightGBM model, initialize the hyperparameters of the LightGBM model (tree depth, learning rate, number of leaf nodes, subsample ratio, etc.), and construct the initial population of the genetic algorithm, where each individual represents a combination of hyperparameters. Through genetic operations such as selection, crossover, and mutation, a new population is gradually generated to explore possible better hyperparameter configurations. 10% of the shield correction data set is used as a validation set. In each generation, the validation set is used to evaluate its performance to obtain the fitness value. After several generations of evolution, the individual with the highest fitness value is selected as the optimal hyperparameter combination, thereby effectively improving the robustness of the LightGBM model and avoiding overfitting of the model to the training data.

[0075] Step 8: Use 20% of the shield correction data set as the test set to test the model performance, and quantitatively evaluate the accuracy of the model prediction through regression analysis indicators. Determine the coefficient of determination (R 2 ) is greater than the preset threshold to evaluate the model prediction effect. 2 If the threshold is exceeded, save the model prediction results. Otherwise, adjust the model's hyperparameters and return to step 6;

[0076] Coefficient of determination (R 2 ) represents the ratio of the regression sum of squares to the total deviation sum of squares, reflecting the model's ability to explain data variation. 2 The closer the value is to 1, the stronger the model's ability to explain the data and the higher the prediction accuracy. The calculation formula is as follows:

[0077]

[0078] Where: f i Represents the predicted value calculated by the model; y i Represents the real data of the i-th test sample; Represents the average value of the true data of the test sample.

[0079] Step 9: When the shield machine performs axis correction, the total number of correction rings and the correction amount of each ring are obtained through data analysis. The correction interval of the ring correction amount is determined according to the association rule, and the excavation data interval corresponding to each ring correction amount is obtained. Each data interval is divided into equal parts and the data is arranged and combined. The GA-LightGBM shield axis correction model trained in step 8 is input, and the model prediction results are statistically analyzed. The set of input operation data with the smallest distance from the true value is selected as the recommended value for this ring correction, thereby realizing efficient and accurate correction of the excavation posture of the shield machine.

[0080] The total number of correction rings and the correction amount of each ring are extracted in the form of rings to see which ring he starts to correct. Figure 3 The beginning part of the correction section, the end of the correction is when it is infinitely close to or equal to 0 value.

[0081] Assuming that the correction value of this loop is predicted to be -2, it is in the interval [-3,0].

[0082] The selected excavation parameters and correction amounts are discretized, the correction amounts are predicted, and the corresponding excavation parameter intervals can be found through association rules, and then subsequent operations can be performed.

[0083] For the data discretized in step 1, the interval is divided equally to satisfy the data interval of the association rule. This way, the recommendation can be more precise. After equal division, each data interval is very close. The interval after equal division is no longer in the form of an interval, but a specific value, such as [500,1000]. If it is divided into ten equal parts, it will be 500, 550, 600, ..., 950, 1000.

[0084] Regarding statistical analysis: For example, for the horizontal deviation of the incision, after dividing my data into ten equal parts, the correction amounts predicted by these ten data are different. Then they are compared with the actual correction amounts, and the closest one is selected as the recommended value to guide the correction.

[0085] Select a set of input operation data with the smallest distance from the true value: for example, the data set has 6 sets of propulsion cylinders, that is, 6 sets of propulsion cylinder pressures and 6 sets of propulsion cylinder strokes.

[0086] Embodiment 2

[0087] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0088] Embodiment 3

[0089] The purpose of this embodiment is to provide a computer-readable storage medium.

[0090] A computer-readable storage medium stores a computer program, which executes the steps of the above method when executed by a processor.

[0091] Embodiment 4

[0092] The purpose of this embodiment is to provide a shield correction parameter recommendation method based on association rule mining and machine learning. Its system modules are as follows: Figure 1 As shown, specifically including the following:

[0093] Data collection and analysis module, data set preprocessing module, association rule mining and analysis module, shield deviation correction parameter recommendation module;

[0094] Data collection and analysis module: used to collect and filter data and obtain key tunneling parameters during the construction of the shield machine; whether the tunneling route of the shield machine cutout and the shield tail fits the tunnel design axis is the main basis for judging whether the shield machine has deviated, so the horizontal deviation of the shield machine cutout, the vertical deviation of the cutout, the horizontal deviation of the shield machine tail and the vertical deviation of the shield tail are selected as the model output data. When the axis deviates, the shield driver mainly adjusts the thrust cylinder to correct the tunneling posture of the shield machine, so the thrust of each group of thrust cylinders and the stroke of the thrust cylinder are selected as the model input data, and the correction amount of the previous link is used as a guide. Finally, the correction segment data is extracted from the tunneling parameters and integrated into the shield correction data set.

[0095] The data set preprocessing module is used to process the shield correction data set; calculate the correction amount of each ring of the shield axis deviation, divide all data by ring number and calculate the average value of each ring, and use the box plot method to test the data for outliers and eliminate them.

[0096] The association rule mining and analysis module is used to mine the association relationship between the excavation parameters and the correction amount; the equal-width method is used to discretize the data set, and the FP-Growth association rule algorithm is used to mine the frequent item sets of the shield deviation parameter interval and the correction amount interval of each ring. The key parameters such as support and confidence in the FP-Growth association rule algorithm are set to determine the association rules between the final shield deviation interval and the operation parameter interval.

[0097] The shield correction parameter recommendation module is used to predict the optimal recommended operating parameter values ​​corresponding to the correction amount; for each correction amount, the corresponding discretized parameter interval is identified, and the operating parameter interval matching the correction amount is selected through the obtained association rules. The operating parameters are iteratively input into the LightGBM machine learning regression model, and the genetic algorithm (GA) is used to optimize the model's hyperparameters, so as to efficiently and accurately predict the recommended values ​​of key operating parameters during shield machine correction.

[0098] The technical solution of the present invention selects the key operating parameters that the shield driver mainly adjusts when correcting the axis as input data, selects the correction segment data during the excavation process, and integrates them into a shield correction data set after preprocessing. The correction segment data is discretized with equal width, and the association rules between the operating parameters and the correction amount of each ring are mined through the FP-Growth association analysis algorithm, which is used as the basis for recommending operating parameters. For a specific deviation amount, find out the discretized data interval where it is located, use the obtained association rules, select the corresponding operating parameter interval as the input data of the machine learning regression model, and predict the output value closest to the target correction amount through the GA-LightGBM model. The corresponding input parameters are the best recommended operating parameter values.

[0099] Embodiment 5

[0100] The purpose of this embodiment is to provide a computer program product containing instructions, which, when running on a computer, enables the computer to execute the methods and functions involved in any of the above embodiments.

[0101] The steps involved in the apparatus of the above embodiment correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0102] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0103] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A shield correction parameter recommendation method based on association rules and machine learning, characterized by: include: Obtain the required shield construction correction section data and construct a shield correction data set based on the acquired data; Preprocess the shield deviation correction data set; Discretize the data in the preprocessed shield deviation correction data set to obtain the tunneling data interval and the deviation correction interval of each ring during the shield deviation correction process; A minimum confidence threshold is set, and association rules that meet the set minimum confidence threshold are selected from the obtained excavation data interval and the deviation correction interval of each ring; All association rules that meet the minimum confidence threshold are integrated to obtain the shield machine deviation correction interval corresponding to the tunneling data interval during the final shield machine deviation correction; Each shield correction interval is divided into equal parts and the data is arranged and combined, input into the shield axis correction model, and the model prediction results are statistically analyzed. The set of input operation data with the smallest distance from the true value is selected as the recommended correction value for this loop.

2. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1 is characterized in that: The correction section data is extracted from the key excavation parameters during the shield machine construction process and integrated into the shield correction data set.

3. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1 is characterized in that: The thrust and stroke of each group of propulsion cylinders and the correction amount of the previous ring are used as the input data of the shield axis correction model. The horizontal deviation of the shield machine's incision, the vertical deviation of the incision, the horizontal deviation of the shield machine's shield tail, and the vertical deviation of the shield tail are used as the output data of the machine learning regression model.

4. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1, characterized in that: Preprocess the shield deviation correction data set, including: Taking ring as unit, all data in the shield machine deviation correction data set are processed according to the ring number, and the deviation correction amount of each ring of the shield machine is obtained through analysis and calculation, and the average value of each ring is calculated so that each ring corresponds to one of the data; The box plot method is used to detect parameter outliers. The upper and lower bounds exceeding the gray dotted lines in the figure are regarded as outliers and are removed.

5. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1, characterized in that: The preprocessed shield correction data set is discretized. Specifically, the equal-width method is used to convert the continuous data of the shield correction data set into discrete data, and the excavation data interval and the correction amount interval of each ring during the shield correction process are obtained.

6. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1, characterized in that: The obtained excavation data interval and the deviation correction interval of each ring are used to select the association rules that meet the set minimum confidence threshold, including: Set the minimum support threshold of the FP-Growth association rule algorithm, remove each data below the support threshold, and mine the frequent item sets in the shield deviation correction data set; The confidence of each frequent item set is calculated to generate association rules, and the minimum confidence threshold is set to filter out association rules that meet the set minimum confidence threshold.

7. The method for recommending shield deviation correction parameters based on association rules and machine learning as claimed in claim 1, characterized in that: After determining the association rules between the final shield machine deviation correction amount and the operating parameters, all the association rules that meet the threshold mined by the FP-Growth algorithm are integrated to obtain the shield machine deviation correction amount interval corresponding to the tunneling data interval during the final shield machine deviation correction; Preferably, when the shield machine performs axis correction, the total number of correction rings and the correction amount of each ring are calculated through data analysis, the correction interval of the ring correction amount is determined according to the association rule, and the excavation data interval corresponding to each ring correction amount is obtained.

8. The shield deviation correction parameter recommendation system based on association rules and machine learning is characterized by: include: The data acquisition and analysis module is configured to: obtain the required shield construction correction section data and construct a shield correction data set based on the obtained data; The data preprocessing module is configured to: preprocess the shield deviation correction data set; The association rule mining and analysis module is configured to: discretize the data in the preprocessed shield deviation correction data set to obtain the tunneling data interval and the deviation correction amount interval of each ring during the shield deviation correction process; A minimum confidence threshold is set, and association rules that meet the set minimum confidence threshold are selected from the obtained excavation data interval and the deviation correction interval of each ring; The shield machine deviation correction parameter recommendation module is configured to: integrate all association rules that meet the minimum confidence threshold, and obtain the shield machine deviation correction amount interval corresponding to the tunneling data interval during the final shield machine deviation correction; Each shield correction interval is divided into equal parts and the data is arranged and combined, input into the shield axis correction model, and the model prediction results are statistically analyzed. The set of input operation data with the smallest distance from the true value is selected as the recommended correction value for this loop.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are performed.

Images