A method for mining association rules of shield tunneling parameters suitable for complex strata
Through the boxing method and probability calculation, the correlation rules are generated, and the problems of large amount of calculation and poor interpretability in the processing of shield excavation parameters of complex formations are solved, and the rapid excavation of long correlation rules and the improvement of construction guidance effect is achieved.
Patent Information
- Application Number
- CN202411270965.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-11
AI Technical Summary
When handling shield excavation parameters of complex formations, the prior art has a large amount of calculation and low efficiency, making it difficult to generate valuable long-term correlation rules, and the rules are poorly interpretable, making it difficult to guide construction.
The data is processed by binning method, and the data is classified into the sub-data set by setting the binning rules, the expected and actual occurrence probability of each combination is calculated, the probability threshold is set to generate association rules, and the physical indicators are used to judge the quality of the rules and increase the Interpretability.
It realizes the rapid excavation of long-related rules in complex strata, improves the operation speed and interpretability of rules, and can better guide shield construction.
Smart Images

Figure SMS_1 
Figure SMS_2 
Figure SMS_3
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of shield tunneling parameter processing, and specifically relates to a shield tunneling parameter association rule mining method suitable for complex strata. Background Art
[0002] During the shield tunneling construction process, due to the interactive feedback between the stratum and the equipment, the correlation between the tunneling parameters and the stratum conditions is very complex, making the control of the shield machine difficult. Since the shield equipment has poor adaptability to the stratum, if the shield tunneling parameters cannot match the current excavation stratum well, it will cause serious wear of the cutter and low tunneling efficiency at the least, and may even cause construction risks such as large surface settlement and collapse.
[0003] In the related technology, there are two main types of shield tunneling parameter mining and analysis methods: one is the conventional mathematical statistical method, such as using the Pearson correlation coefficient to calculate the parameter correlation. This method can only qualitatively characterize the strength of the correlation between parameters, and the conclusion is often only applicable to certain specific working conditions, with poor generalization; the second is to use association rule mining methods to mine association rules between tunneling parameters in different strata, such as the Apriori algorithm and the FP-Growth algorithm. However, the Apriori algorithm and the FP-Growth algorithm require multiple scans of the database to generate candidate item sets and frequent item sets, resulting in huge computational complexity and low efficiency when processing large-scale shield data. In addition, due to the characteristics of the algorithm itself, when the number of feature types increases, it is difficult to generate valuable long association rules. Moreover, these algorithms usually use mathematical indicators such as confidence and support to evaluate the pros and cons of the rules. The rules are poorly interpretable and it is difficult to guide construction in combination with actual engineering problems.
[0004] Therefore, it is necessary to provide a shield tunneling parameter association rule mining method suitable for complex strata to solve the problems raised in the above background technology. Summary of the invention
[0005] The present application provides a shield tunneling parameter association rule mining method suitable for complex strata, which has the characteristics of fast calculation speed, ability to directionally mine long association rules, strong rule interpretability, etc., and can effectively solve at least one technical problem in the background technology.
[0006] In order to solve the above technical problems, the technical solution of this application is:
[0007] A shield tunneling parameter association rule mining method suitable for complex strata includes the following steps:
[0008] Step S1, obtaining a data set of shield tunneling parameters in complex strata, wherein each data in the data set includes a characterization column feature, an excavation column feature and a physical indicator column feature;
[0009] Step S2, taking the characterization column feature as an object, setting a binning rule for each type of feature in the characterization column feature, binning the data in the data set, combining the binning intervals of different types of features of the characterization column feature, and classifying the data in the data set into sub-data sets, wherein each piece of data in the sub-data set includes a mining column feature and a physical indicator column feature;
[0010] Step S3, taking the mining column feature as an object, setting a binning rule for each type of feature in the mining column feature, and performing secondary binning on the data in the sub-data set;
[0011] Step S4, combining the binning intervals of different types of features of the mining column features, and calculating the expected probability of occurrence and the actual probability of occurrence of each combination;
[0012] Step S5, setting a probability threshold. If the actual occurrence probability of the combination in step S4 is not less than the probability threshold and is not less than the expected occurrence probability, it is determined to meet the requirements of the generation rule and the association rule corresponding to the combination is generated; otherwise, it is determined to not meet the requirements of the generation rule and the association rule is not generated.
[0013] As a preferred improvement, the characterization series of features includes the tunnel face elastic modulus and the tunnel face composite ratio; the excavation series of features includes the thrust and torque of the shield machine; and the physical indicator series of features includes surface settlement.
[0014] As a preferred improvement, the binning rules in step S2 and step S3 are formulated using expert experience binning, equal frequency binning, equal distance binning or custom binning.
[0015] As a preferred improvement, in step S4, the calculation process of the expected occurrence probability of the binning interval combination of different types of features of the mining column feature is:
[0016] Calculate the actual occurrence probability of different types of features of the mining column features;
[0017] Multiply the actual occurrence probabilities of all types of features in each combination to get the expected occurrence probability;
[0018] The actual occurrence probability of each type of feature is the ratio of the number of occurrences of each type of feature to the total number of occurrences of all type of features.
[0019] As a preferred improvement, the actual occurrence probability of the binning interval combination of different types of features of the mining column feature is the ratio of the number of times the combination occurs to the total number of times all combinations occur.
[0020] As a preferred improvement, in step S5, the probability threshold is set according to the amount of data and the density of data information in the data set.
[0021] As a preferred improvement, the form of association rules is: mining column feature different types of feature binning interval combination ∪ actual occurrence probability of the combination ∪ mean value of physical indicator column features in the combination.
[0022] The beneficial effects of this application are:
[0023] (1) This application uses a binning method to generate feature classes to solve the problem that continuous data features are difficult to use. Physical indicators and actual probabilities are used as evaluation indicators of rules, so that the quality of rules can be judged from a physical sense through physical indicator columns, which increases the interpretability of rules and better solves practical research problems. It can be effectively applied to many fields such as civil engineering and data mining;
[0024] (2) The physical indicator column is selected according to the research objectives. In addition to being expressed by the actual probability of occurrence, the quality of the rule can also be judged by the mean of the physical indicator column, which enhances the interpretability of the rules and the comparability between rules, and meets the requirements of practical problems to a greater extent. DETAILED DESCRIPTION
[0025] The technical solutions in the embodiments of the present application are described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.
[0026] This embodiment provides a shield tunneling parameter association rule mining method applicable to complex strata, comprising the following steps:
[0027] Step S1, obtaining a data set of shield tunneling parameters in complex strata, wherein each data in the data set includes a characterization column feature, an excavation column feature and a physical indicator column feature;
[0028] Step S2, taking the characterization column feature as an object, setting a binning rule for each type of feature in the characterization column feature, binning the data in the data set, combining the binning intervals of different types of features of the characterization column feature, and classifying the data in the data set into sub-data sets, wherein each piece of data in the sub-data set includes a mining column feature and a physical indicator column feature;
[0029] Step S3, taking the mining column feature as an object, setting a binning rule for each type of feature in the mining column feature, and performing secondary binning on the data in the sub-data set;
[0030] Step S4, combining the binning intervals of different types of features of the mining column features, and calculating the expected probability of occurrence and the actual probability of occurrence of each combination;
[0031] Step S5, setting a probability threshold. If the actual occurrence probability of the combination in step S4 is not less than the probability threshold and is not less than the expected occurrence probability, it is determined to meet the requirements of the generation rule and the association rule corresponding to the combination is generated; otherwise, it is determined to not meet the requirements of the generation rule and the association rule is not generated.
[0032] In this embodiment, the characterization series of features includes the tunnel face elastic modulus and the tunnel face composite ratio; the excavation series of features includes the thrust and torque of the shield machine; and the physical indicator series of features includes surface settlement.
[0033] The binning rules in step S2 and step S3 are formulated using expert experience binning, equal frequency binning, equal distance binning or custom binning.
[0034] In step S4, the calculation process of the expected occurrence probability of the binning interval combination of different types of features of the mining column feature is:
[0035] Calculate the actual occurrence probability of different types of features of the mining column features;
[0036] Multiply the actual occurrence probabilities of all types of features in each combination to get the expected occurrence probability;
[0037] The actual occurrence probability of each type of feature is the ratio of the number of occurrences of each type of feature to the total number of occurrences of all type of features.
[0038] The actual occurrence probability of the binning interval combination of different types of features of the mining column feature is the ratio of the number of times the combination occurs to the total number of times all combinations occur.
[0039] In step S5, the probability threshold is set according to the amount of data and the density of data information in the data set.
[0040] The form of association rules is: the binning interval combination of different types of features of mining column features ∪ the actual occurrence probability of the combination ∪ the mean value of the physical indicator column features within the combination.
[0041] Example 1
[0042] In this embodiment, the selected dataset includes multiple pieces of data. Each piece of data includes 2 characterization column features, 2 mining column features, and 1 physical index column feature. Among them, the characterization column features are the tunnel face elastic modulus and the tunnel face composite ratio. Both the tunnel face elastic modulus and the tunnel face composite ratio are formation characterization parameters. The tunnel face elastic modulus represents the hardness of the soil layer in contact with the tunnel face in front of the tunnel excavation, and the tunnel face composite ratio represents the degree of soil layer composition. The closer the tunnel face composite ratio is to 0, the lower the composite degree and the more homogeneous the soil layer; the mining column features are the thrust and torque of the shield machine; the physical index column feature is the ground settlement of the shield. The dataset is shown in Table 1.
[0043] Table 1 Dataset Table
[0044]
[0045] Binning rules are set for the tunnel face elastic modulus data Z and the tunnel face composite ratio data F respectively, where:
[0046] Binning interval Z1: 10000 < Z ≤ 15000; Binning interval Z2: 15000 < Z ≤ 20000;
[0047] Binning interval F1: 0 < F ≤ 0.5; Binning interval F2: 0.5 < F ≤ 1.
[0048] The dataset after the first binning is shown in Table 2.
[0049] Table 2 Dataset Table after the First Binning
[0050]
[0051] According to the binning intervals of the tunnel face elastic modulus data Z and the tunnel face composite ratio data F, there are a total of four combinations, namely (Z1, F1), (Z1, F2), (Z2, F1), and (Z2, F2). It can be seen from Table 2 that the (Z2, F2) combination does not exist. Therefore, the subset corresponding to (Z2, F2) is an empty set.
[0052] Taking the (Z1, F1) combination as an example, the obtained subset is shown in Table 3.
[0053] Table 3 Subset Table under the (Z1, F1) Combination
[0054]
[0055] Binning rules are set for the shield machine thrust data A and the shield machine torque data B respectively, where:
[0056] A1: 6 < A ≤ 9; A2: 9 < A ≤ 12; A3: A > 12;
[0057] B1:0<B≤2;B2:B> 2.
[0058] The sub-datasets under the combination of (Z1, F1) after secondary binning are shown in Table 4.
[0059] Table 4 Sub-dataset table of the combination of (Z1, F1) after secondary binning
[0060]
[0061] According to the binning intervals of the shield machine thrust data A and the shield machine torque data B, there are a total of 6 combinations, namely: (A1, B1), (A1, B2), (A2, B1), (A2, B2), (A3, B1) and (A3, B2).
[0062] Table 4 contains 10 data in total, including:
[0063] The number of times the combination (A1, B1) appears is 3, so the actual probability of the combination (A1, B1) occurring is 3 / 10=0.3;
[0064] The number of times the combination (A1, B2) occurs is 3, so the actual probability of the combination (A1, B2) occurring is 3 / 10=0.3;
[0065] The number of times the combination (A2, B1) occurs is 2, so the actual probability of the combination (A2, B1) occurring is 2 / 10=0.2;
[0066] The number of times the combination (A2, B2) occurs is 1, so the actual probability of the combination (A2, B2) occurring is 1 / 10=0.1;
[0067] The number of times the combination (A3, B1) occurs is 1, so the actual probability of the combination (A3, B1) occurring is 1 / 10=0.1;
[0068] The number of times the combination (A3, B2) appears is 0, so the actual probability of the combination (A3, B2) occurring is 0.
[0069] Taking the combination (A1, B1) as an example, the calculation process of its expected probability of occurrence includes the following steps:
[0070] Calculate the actual probability of occurrence of bin intervals A1 and B1 in the (A1, B1) combination:
[0071] Table 4 contains 10 data in total, including:
[0072] The number of times bin interval A1 appears is 6, so the actual probability of bin interval A1 occurring is 6 / 10=0.6;
[0073] The number of times bin interval B1 appears is 6, so the actual probability of occurrence of bin interval B1 is 6 / 10=0.6;
[0074] Multiplying the actual occurrence probabilities of interval A1 and interval B1, we get the expected occurrence probability of the combination (A1, B1) = 0.6×0.6=0.36.
[0075] By analogy, bin interval A2 appears 3 times, so the actual probability of bin interval A2 is 0.3; bin interval A3 appears 1 time, so the actual probability of bin interval A3 is 0.1; bin interval B2 appears 4 times, and the actual probability of bin interval B2 is 0.4, which is summarized in Table 5.
[0076] Table 5 Actual occurrence probability of each type of feature in the combination
[0077]
[0078] The actual occurrence probability and expected occurrence probability of all combinations are obtained, as shown in Table 6:
[0079] Table 6 Actual and expected probability of occurrence of combinations
[0080]
[0081] It can be seen from Table 6 that when the probability threshold is set to 0.2, the actual occurrence probability of the combination (A1, B2) and the combination (A2, B1) is not less than the expected occurrence probability and not less than the probability threshold, which meets the requirements of the generation rule, and generates the association rules corresponding to the combination (A1, B2) and the combination (A2, B1); the combinations (A1, B1), (A2, B2), (A3, B1), and (A3, B2) do not meet the requirements of the generation rule.
[0082] The rules associated with the combination (A1, B2) and the combination (A2, B1) are expressed as:
[0083] The association rules under tunnel face elastic modulus (MPa): [10000, 15000] ∪ tunnel face composite ratio: [0, 0.5] are:
[0084] Shield machine thrust (MN): [6,9) ∪ Shield machine torque (MN·m): (2, +∞) ∪ Mean surface settlement (mm): 4.5mm ∪ Actual probability 0.3;
[0085] Shield machine thrust (MN): [9,12) ∪ Shield machine torque (MN·m): [0, 2) ∪ Mean surface settlement (mm): 5.0mm ∪ Actual probability 0.2.
[0086] The embodiments of the present application are described above, but the present application is not limited to the above-mentioned specific implementation modes. The above-mentioned specific implementation modes are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.
Claims
1. A shield tunneling parameter association rule mining method suitable for complex strata, characterized in that: The steps include: Step S1, obtaining a data set of shield tunneling parameters in complex strata, wherein each data in the data set includes a characterization column feature, an excavation column feature and a physical indicator column feature; Step S2, taking the characterization column feature as an object, setting a binning rule for each type of feature in the characterization column feature, binning the data in the data set, combining the binning intervals of different types of features of the characterization column feature, and classifying the data in the data set into sub-data sets, wherein each piece of data in the sub-data set includes a mining column feature and a physical indicator column feature; Step S3, taking the mining column feature as an object, setting a binning rule for each type of feature in the mining column feature, and performing secondary binning on the data in the sub-data set; Step S4, combining the binning intervals of different types of features of the mining column features, and calculating the expected probability of occurrence and the actual probability of occurrence of each combination; Step S5, setting a probability threshold. If the actual occurrence probability of the combination in step S4 is not less than the probability threshold and is not less than the expected occurrence probability, it is determined to meet the requirements of the generation rule and the association rule corresponding to the combination is generated; otherwise, it is determined to not meet the requirements of the generation rule and the association rule is not generated.
2. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: The characterization series of features includes the tunnel face elastic modulus and the tunnel face composite ratio; the excavation series of features includes the thrust and torque of the shield machine; and the physical indicator series of features includes surface settlement.
3. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: The binning rules in step S2 and step S3 are formulated using expert experience binning, equal frequency binning, equal distance binning or custom binning.
4. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: In step S4, the calculation process of the expected occurrence probability of the binning interval combination of different types of features of the mining column feature is: Calculate the actual occurrence probability of different types of features of the mining column features; Multiply the actual occurrence probabilities of all types of features in each combination to get the expected occurrence probability; The actual occurrence probability of each type of feature is the ratio of the number of occurrences of each type of feature to the total number of occurrences of all type of features.
5. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: The actual occurrence probability of the binning interval combination of different types of features of the mining column feature is the ratio of the number of times the combination occurs to the total number of times all combinations occur.
6. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: In step S5, the probability threshold is set according to the amount of data and the density of data information in the data set.
7. The shield tunneling parameter association rule mining method applicable to complex strata according to claim 1 is characterized in that: The form of association rules is: the binning interval combination of different types of features of mining column features ∪ the actual occurrence probability of the combination ∪ the mean value of the physical indicator column features within the combination.
