A method for checking and analyzing survey information of a directional drilling crossing section of a long-distance natural gas pipeline

Abstract

The application discloses a kind of for natural gas long-distance pipeline directional drilling crossing section survey information verification analysis method, belong to industrial data processing technical field.It includes: obtaining multidimensional section survey information by data interface and synchronously extracting multidimensional working condition label, constructs survey information vector and working condition label set;With survey information vector and working condition label set as input parameter, after processing, output verification task group;Information verification model is constructed and is independently operated, receives verification task group dynamic adjustment verification sensitivity, carries out multidimensional data verification;Normal working condition data is obtained from historical normal construction working condition database by data interface;Data element-by-element comparison is carried out and confidence is calculated;Obtain comparison result and construct structured analysis result vector group, using MQTT protocol uploads each data of analysis result and real-time monitoring.The present application aims at improving the timeliness, accuracy and safety of section survey information verification.

Description

Technical Field

[0001] This invention relates to the field of industrial data processing technology, and in particular to a method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines. Background Technology

[0002] Natural gas long-distance pipelines are core infrastructure for energy transmission. When pipelines cross rivers, highways, railways, embankments, and areas with dense buildings, directional drilling trenchless crossing technology has become the mainstream construction process for long-distance pipeline crossings due to its advantages such as minimal environmental disturbance, high construction safety, and good pipeline service stability. Currently, directional drilling cross-section surveys encompass multi-source heterogeneous data, including topographic mapping, geological drilling, underground pipeline detection, hydrological monitoring, and in-situ geotechnical testing. These data correspond to key engineering information such as topographic space, stratigraphic structure, geotechnical physical and mechanical parameters, distribution of underground structures, and hydrogeological conditions. Directional drilling crossing projects are characterized by strong concealment, long crossing distances, complex geological conditions, and high construction risks. The results of the cross-section survey are the core basis for pipeline trajectory design, borehole layout, mud system ratio, borehole enlargement and pullback construction, and avoidance of existing pipelines and structures. The completeness, accuracy, and self-consistency of the survey data directly determine the success or failure of directional drilling construction and the long-term operational safety of the pipeline.

[0003] However, the existing verification methods have unresolved issues: 1. Using static task grouping leads to resource idleness and excessively long waiting times for high-risk data; 2. Fixed parameter verification methods make it impossible to balance the detection rate and false alarm rate according to changes in the construction scenario; 3. Fixed prior parameters make it impossible to match gradual changes in equipment performance or changes in data characteristics. Summary of the Invention

[0004] This invention provides a method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines, aiming to solve the problems existing in the prior art.

[0005] The following solutions are provided for this purpose:

[0006] This invention discloses a method for verifying and analyzing cross-section exploration information for directional drilling of long-distance natural gas pipelines. The method includes: acquiring multi-dimensional cross-section exploration information and simultaneously extracting multi-dimensional working condition labels to construct an exploration information vector and a working condition label set; using an improved ant colony algorithm to construct a resource configuration parameter scheduling engine, taking the exploration information vector and working condition label set as input parameters, and outputting a verification task group after processing, wherein the verification task group includes urgency and operation configuration parameters; constructing an information verification model, with each model running concurrently and independently, receiving the verification task group to dynamically adjust the verification sensitivity, performing multi-dimensional data verification, outputting the verification results separately, and merging them into a comprehensive verification vector; acquiring normal working condition data from a historical normal construction working condition database through a data interface to form a normal working condition vector; constructing a data comparison model using an improved weighted Mahalanobis distance, acquiring the normal working condition vector and performing element-by-element comparison with the comprehensive verification vector and calculating the confidence level to obtain a comparison result vector; acquiring the comparison results to construct a structured analysis result vector group, and using a communication protocol to upload each data item in the analysis result vector group to a management platform for monitoring.

[0007] Furthermore, the multi-dimensional cross-sectional exploration information includes formation resistivity, acoustic longitudinal wave velocity, bottom hole annular pressure, rock mass integrity coefficient, actual drilling rig thrust, and drilling rig rotation torque; the working condition label set includes geological stratification labels, risk source proximity distance labels, and drilling speed change indicator labels.

[0008] Furthermore, using an improved ant colony algorithm, a resource configuration parameter scheduling engine is constructed, and the process of generating and verifying task groups includes:

[0009] The data in the exploration information vector is mapped to the verification task, the information verification model is mapped to the execution slot, and the pheromone matrix is ​​constructed, and the number of ant colonies, the maximum number of iterations and the importance factor are set.

[0010] The estimated processing time, urgency, and historical anomaly rate of each verification task are combined and calculated. The weights of each factor are dynamically adjusted by the CPU load ladder function to obtain the heuristic factor used to guide the ant search.

[0011] Using a roulette wheel selection strategy based on pheromone matrix and heuristic factors, and imposing execution slot capacity constraints, each ant determines the information verification model for all verification tasks one by one, dividing the tasks into three concurrent verification task groups: G1, G2, and G3.

[0012] Furthermore, the process of constructing a resource configuration parameter scheduling engine and verifying the generation of task groups using an improved ant colony algorithm also includes:

[0013] The verification task group was quantitatively evaluated, and elite ants were selected to carry out global pheromone volatilization and deposition to obtain an updated pheromone matrix to guide the iterative search.

[0014] The distance to the risk source is extracted from the working condition label set and the risk level is calculated. The pheromone volatilization factor is dynamically adjusted by substituting it into a time-varying formula that decreases with risk, and an adaptive volatilization factor value is obtained to regulate the global volatilization of pheromones.

[0015] Set the iteration termination trigger condition, calculate the urgency of each task group after the iteration is completed, and generate the running configuration parameters and voting weights of the information verification model.

[0016] Furthermore, constructing an information verification model and conducting multi-dimensional data verification includes:

[0017] The information verification model is constructed by including Mahalanobis distance verification model, isolated forest verification model and physical constraint verification model. The verification model extracts data sub-vectors from three concurrent verification task groups G1, G2 and G3 in turn, and loads the running configuration parameters synchronously. The running configuration parameters include shrinkage coefficient, depth limit and residual tolerance threshold.

[0018] Furthermore, constructing an information verification model and conducting multi-dimensional data verification also includes:

[0019] The Mahalanobis distance verification model uses a shrinkage coefficient to perform linear shrinkage estimation of the covariance matrix to calculate the Mahalanobis distance, and compares it with the chi-square threshold to determine anomalies. It quantifies the anomaly contribution of each dimension through Cholesky decomposition and outputs the anomaly labeling result A.

[0020] The isolated forest verification model uses urgency dynamic constraints to limit the depth of the isolated forest, performs verification, calculates the feature split depth contribution index, and outputs anomaly labeling result B.

[0021] The physical constraint verification model adaptively tightens the residual tolerance threshold based on urgency, performs physical inequality constraint verification, and calculates the normalized sensitivity of each dimension to obtain the anomaly labeling result C.

[0022] Furthermore, constructing an information verification model and conducting multi-dimensional data verification also includes:

[0023] Collect anomaly labeling results A, B, and C, and combine them into a comprehensive verification vector;

[0024] The actual processing time feedback from the validation model is smoothed and updated using an exponential moving average to obtain the estimated processing time matrix for subsequent scheduling decisions.

[0025] Furthermore, the construction of the normal operating condition vector, and the element-by-element comparison and confidence calculation include:

[0026] Retrieve abnormality vector samples from the historical normal working condition database during normal drilling, calculate the normal mean and covariance matrix, and replace the identity matrix with the sample when the sample is insufficient, and the mean approaches zero.

[0027] A dynamic weight matrix is ​​constructed based on the urgency, and a weighted Mahalanobis distance value is calculated for the deviation between the comprehensive verification vector and the normal mean, so as to amplify the abnormal contribution value of the high-risk dimension.

[0028] Calculate the average urgency of all verified models in the current period, calculate the confidence level by the ratio of the weighted Mahalanobis distance value to the dynamic comparison threshold, and determine anomalies item by item based on the anomaly contribution value, and output the comparison result vector.

[0029] Furthermore, a structured analysis result vector group is constructed, which includes data item parameter identifier, data item name, current actual survey measurement value, normal working condition mean, lower limit of normal range, upper limit of normal range, comparison anomaly marker, deviation value, and confidence level; the communication protocol uses the MQTT protocol, and the upload cycle is synchronized with the data acquisition.

[0030] Compared with the prior art, the present invention achieves at least one of the following beneficial effects:

[0031] 1. An improved ant colony algorithm is used to dynamically group and allocate resources for the verification tasks of cross-section survey data. This allows key data items in high-risk crossing areas to be pushed to the fastest verification model first, shortening the anomaly detection latency in concurrent verification. At the same time, capacity constraints are used to balance the load of each model, solving the problem of resource idleness and excessively long waiting time for high-risk data caused by the use of static task grouping in existing technologies.

[0032] 2. By dynamically injecting the average urgency of the verification task group into the covariance shrinkage coefficient of Mahalanobis distance, the depth limit of isolated forest, and the residual threshold of physical constraints, the internal sensitivity of the three independent verification models is adaptively adjusted in real time. This allows the verification standards to be automatically tightened under conditions such as weak interlayers or near risk sources, enhancing the ability to capture weak abnormal signals. This solves the problem of existing fixed-parameter verification methods, which cannot balance the detection rate and false alarm rate according to changes in the construction situation.

[0033] 3. In the normal working condition comparison stage, a dynamic weight matrix is ​​constructed using the urgency of each data item generated in the scheduling stage. The judgment threshold of the weighted Mahalanobis distance is adaptively adjusted according to the global average urgency. The comprehensive verification vector is compared element by element and the confidence is calculated. This ensures that the strictness of the comparison matches the current crossing risk level in real time. In high-risk situations, the deviation of key parameters is highlighted, and in low-risk situations, overly sensitive triggering is avoided. This solves the problem that the existing fixed threshold comparison method is prone to generating a large number of invalid alarms or missed alarms under complex geological conditions.

[0034] 4. Construct a closed-loop feedback mechanism that transmits the actual processing time of the verification model back to the scheduling engine. Continuously update the task prediction processing time matrix through exponential moving average and iteratively statistically analyze the historical anomaly rate of each data item. This enables the heuristic factors and allocation decisions of the ant colony algorithm to learn and optimize with the running state, solving the problem that the prior parameters of the existing technology are fixed, which leads to the inability to match the gradual changes in equipment performance or data characteristics. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a flowchart illustrating a method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines, according to the present invention.

[0037] Figure 2 This is a schematic diagram of the execution flow of the Mahalanobis distance verification model algorithm;

[0038] Figure 3 A tree structure diagram for validating an isolated forest model;

[0039] Figure 4 This is a schematic diagram of the execution flow of the physical constraint verification model algorithm. Detailed Implementation

[0040] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0041] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0042] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0043] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0044] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0045] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0046] Example 1:

[0047] See Figure 1 A method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines, comprising:

[0048] S100. Obtain multi-dimensional cross-sectional exploration information and simultaneously extract multi-dimensional working condition labels to construct an exploration information vector and a working condition label set.

[0049] S200. Using an improved ant colony algorithm, a resource configuration parameter scheduling engine is constructed. The engine takes the survey information vector and the working condition label set as input parameters, and outputs a verification task group after processing. The verification task group is accompanied by urgency and running configuration parameters.

[0050] S300: Construct an information verification model. Each model runs concurrently and independently. Each model receives verification task groups to dynamically adjust the verification sensitivity, performs multi-dimensional data verification, outputs the verification results separately, and merges them into a comprehensive verification vector.

[0051] S400: Obtain normal working condition data from the historical normal working condition database through the data interface to form a normal working condition vector;

[0052] S500: Use the improved weighted Mahalanobis distance to build a data comparison model, obtain the normal working condition vector and the comprehensive verification vector, compare them element by element and calculate the confidence level to obtain the comparison result vector;

[0053] S600: Obtain the comparison results and construct a structured analysis result vector group. Use the communication protocol to upload each data item in the analysis result vector group to the management platform for monitoring.

[0054] The key technical points of the above implementation steps include: 1- Improved ant colony algorithm volatile factor adaptive adjustment with risk distance; 2- Heuristic factor fusion of urgency and anomaly rate; 3- Dynamic adjustment of scheduling parameters to improve model verification sensitivity; 4- Adaptive comparison of urgency-weighted Mahalanobis distance threshold; 5- Optimization of verification time feedback loop.

[0055] In specific implementation, the multi-dimensional cross-sectional exploration information includes formation resistivity, acoustic longitudinal wave velocity, bottom hole annular pressure, rock mass integrity coefficient, actual drilling rig thrust, and drilling rig rotation torque; the working condition label set includes geological stratification labels, risk source proximity distance labels, and drilling speed change indicator labels.

[0056] Using a comprehensive data acquisition gateway (such as an industrial-grade edge computing device that supports protocols such as Modbus TCP, OPC-UA, and MQTT) deployed at the directional drilling site, the gateway aggregates data through two independent channels (i.e., the exploration data channel and the working condition label channel).

[0057] The aforementioned geological stratification labels are strings of current strata names (such as "silty clay," "moderately weathered sandstone," and "fracture zone") automatically classified and transmitted by MWD based on natural gamma, resistivity abrupt changes, and drilling speed variations. These labels are directly obtained from the Formation register of the MWD decoder. Risk source proximity distance labels are obtained from the construction information terminal, which pre-stores the engineering longitudinal profile of the traversal path (including precise station numbers for riverbeds, existing pipelines, and important transportation arteries). The PLC provides the current drill bit's station mileage in real time, and the industrial control computer returns the remaining straight-line distance from the current drill bit to the nearest risk source (e.g., 28.5; when the remaining distance is ≤30m, it is marked as "high-risk approach" distance unit / m). Drilling speed abrupt change indicator labels are determined in real time by the drilling rig PLC during sliding drilling. When the instantaneous mechanical drilling rate (ROP) changes by more than a preset threshold within 10 seconds (e.g., a sudden increase from 0.5m / h to 2.0m / h, with a change rate >200%), the label is used. If the current thrust does not have a corresponding step, the Boolean flag is set to 1 (to indicate that a "cavity or fracture" may be encountered), otherwise the Boolean flag is set to 0. The judgment rule is fixed in the PLC function module (the rule is to calculate the initial urgency reference value of each data item instantaneously based on the remaining distance and the drilling speed change flag. When the drilling speed change flag = 1 and the remaining distance ≤ 30m, the urgency of all data items U = 0.9; when only the drilling speed change flag = 1, the urgency U = 0.7; when only the remaining distance ≤ 30m, the urgency U = 0.6; and for other working conditions, the urgency U = 0.2). The software module in the integrated data acquisition gateway performs a packaging operation every 2 seconds (synchronized with the slowest data source wave speed), encapsulating the above 6 data items (i.e., multi-dimensional cross-section exploration information) within the same time window (GPS time error < 0.5 seconds) into a 6-dimensional vector; and constructs a structure to encapsulate the working condition label into a label set object.

[0058] In practical implementation, an improved ant colony algorithm is used to build a resource configuration parameter scheduling engine, and the process of generating and verifying task groups includes:

[0059] The data in the exploration information vector is mapped to the verification task, the information verification model is mapped to the execution slot, and the pheromone matrix is ​​constructed, and the number of ant colonies, the maximum number of iterations and the importance factor are set.

[0060] The estimated processing time, urgency, and historical anomaly rate of each verification task are combined and calculated. The weights of each factor are dynamically adjusted by the CPU load ladder function to obtain the heuristic factor used to guide the ant search.

[0061] Using a roulette wheel selection strategy based on pheromone matrix and heuristic factors, and imposing execution slot capacity constraints, each ant determines the information verification model for all verification tasks one by one, dividing the tasks into three concurrent verification task groups: G1, G2, and G3.

[0062] The pheromone matrix is ​​defined as τ=[τ(i,j)], where τ(i,j) is the empirical advantage (i.e., pheromone concentration) assigned to the verification model j by the verification task i. The initial τ is set to 1 / n (n is the nth independent task). The number of ants is defined as 10, the maximum number of iterations is set to 50, and the importance factor α=1 and β=2 are used to complete the initialization.

[0063] The above-mentioned estimated processing time, urgency, and historical anomaly rate are combined for calculation, including:

[0064] Step 1) Read the historical average execution time of task i on the verification model j from the processing time matrix maintained by the resource configuration parameter scheduling engine (initially, the Mahalanobis distance verification model is 0.05s, the isolated forest verification model is 0.08s, and the physical constraint verification model is 0.04s. In subsequent cycles, it will be continuously updated by exponential moving average based on the actual execution time).

[0065] Step 2) Use the urgency value U mentioned above as the current verification urgency of the current data item (value range [0.2, 0.9]);

[0066] Step 3) Count the total number of times the current data item was marked as an anomaly by any validation model (i.e., Mahalanobis distance validation model, isolated forest validation model, or physical constraint validation model) and the total number of validations in the past 24 hours. Calculate the historical anomaly rate using the formula: R = N 总 / (N 校验 +ε), where R is the historical anomaly rate; N 总 N represents the total number of anomalies. 校验 The total number of checks; ε is a very small positive number (10^3). -6 (To prevent division by zero); the formula for calculating the reciprocal of the historical anomaly rate is: Q i =1 / (r+0.01), where Q i It is the inverse of the historical anomaly rate (the larger the value, the more stable the historical performance of the current data item is, the less likely it is to have anomalies, and the higher the reliability weight in the heuristic factor).

[0067] The method for dynamically adjusting the weights of various factors using the CPU load ladder function includes: reading the current CPU utilization (percentage) of the server to determine the weights of the urgency factor and the data quality factor. The formula set for determining these factors is as follows:

[0068]

[0069] In the formula w 紧迫 and w 质量These are the urgency factor weight and the data quality factor weight, respectively; Load represents CPU utilization. The urgency factor weight controls the contribution ratio of the data item verification urgency U in the heuristic factor. When Load is low, a high value (0.8) is used to prioritize scheduling for high-risk data; when Load is high, a low value (0.2) is used to avoid scheduling oscillations caused by excessive urgency driven by resource contention. The data quality factor weight controls the reciprocal of the historical anomaly rate Q. i The contribution ratio is set to a low value (0.2) when the load is low and a high value (0.8) when the load is high. Priority is given to ensuring that dimensions with high inherent reliability of the data obtain stable verification resources, so as to ensure that the accuracy of basic verification will not deteriorate due to busy load.

[0070] The heuristic factor guiding the ant search is calculated using the following formula: In the formula, η(i,j) is the heuristic factor assigned to the validation model j by task i (the prior expectation value that guides ants to choose a path in the ant colony algorithm); T ij U represents the estimated processing time of task i on the validation model j. i The urgency of validation for data item i.

[0071] The above method involves each ant determining the information verification model for each verification task. In each iteration, each ant independently selects verification model j for each verification task i. The selection probability uses a roulette wheel strategy with pheromones and heuristic factors, and the selection probability formula is as follows: In the formula Let τ(i,j) be the selection probability of the k-th ant assigning verification task i to verification model j; τ(i,j) be the empirical advantage of assigning verification task i to verification model j (i.e., pheromone concentration, obtained from the definition of the pheromone matrix above); η(i,j) be the heuristic factor for assigning task i to verification model j (obtained from the above calculation); α and β are importance factors (values ​​have been defined above, used to control the influence of pheromone concentration on selection probability and the influence of heuristic information on selection probability, respectively); l is the model index; to avoid overloading a single verification model and introducing real-time capacity constraints, the maximum number of tasks that the verification model is allowed to receive is capped at 60%. After each ant completes its selection, an assignment matrix is ​​obtained to determine the composition of the three task groups G1, G2, and G3.

[0072] In practice, the improved ant colony algorithm is used to build a resource configuration parameter scheduling engine, and the process of generating and verifying task groups also includes:

[0073] The verification task group was quantitatively evaluated, and elite ants were selected to carry out global pheromone volatilization and deposition to obtain an updated pheromone matrix to guide the iterative search.

[0074] The distance to the risk source is extracted from the working condition label set and the risk level is calculated. The pheromone volatilization factor is dynamically adjusted by substituting it into a time-varying formula that decreases with risk, and an adaptive volatilization factor value is obtained to regulate the global volatilization of pheromones.

[0075] Set the iteration termination trigger condition, calculate the urgency of each task group after the iteration is completed, and generate the running configuration parameters and voting weights of the information verification model.

[0076] The quantitative evaluation process for the verification task group is as follows: After generating m allocation schemes in each iteration, the fitness of each scheme is calculated (the smaller the fitness, the better). The fitness function is defined as the weighted sum of the maximum completion time and the urgency penalty, and its weighted sum formula is:

[0077]

[0078] In the formula, F is the fitness value (used for quantitative evaluation of the allocation scheme); G j T is the set of tasks assigned to the validation model j; ij Let be the estimated processing time of task i on the validation model j; λ is the penalty coefficient (value 0.8); L ij Assign a boolean flag (i.e., 1 if task i is assigned to model j, 0 otherwise); PT j To verify the total processing time of the currently assigned task on model j; The maximum completion time component; The total time penalty is weighted by the urgency level.

[0079] The process of selecting elite ants to implement global pheromone volatilization and deposition is as follows: Select the elite ant with the smallest F value, update the global pheromone levels along its path, and follow the rules for pheromone volatilization and deposition:

[0080]

[0081] In the formula △τ 精英 (i,j) represents the pheromone increment deposited by the elite ants on the (i,j) path in this round; Q is the pheromone deposition constant (with a value of 10); F 精英 This represents the fitness value of elite ants in this iteration; the global pheromone update formula is:

[0082] τ 更新 (i,j)←(1-ρ)·τ(i,j)+△τ 精英 (i,j)

[0083] In the formula τ 更新 (i,j) represents the updated value of task i and the verification model j position in the pheromone matrix (used to guide ant selection in subsequent iterations); (1-ρ) is the pheromone retention coefficient.

[0084] The formula for calculating the risk level is as follows: In the formula, Risk represents the current construction risk level (used to dynamically adjust the pheromone volatile factor); D risk The remaining straight-line distance between the current drill bit and the nearest risk source (measured data obtained from the working condition label); The distance normalization ratio (when D) risk When the distance decreases from 30 to 0, the ratio decreases linearly from 1 to 0; that is, Risk increases linearly when the distance is ≤30 meters, and Risk=0 when the distance exceeds 30 meters. The update formula for the pheromone volatile factor is:

[0085]

[0086] In the formula, ρ is the pheromone volatile factor; ρ0 is the basic volatile factor (with a value of 0.1); k is the risk sensitivity coefficient (with a value of 0.8); and k·Risk is the risk adjustment term.

[0087] The iteration termination trigger conditions are set as follows: 1- The number of iterations reaches 50; 2- The optimal solution has not improved for 5 consecutive iterations (i.e., the fitness change is <1%). For each task group G, the iteration termination trigger conditions are: 1- The number of iterations reaches 50; 2- The optimal solution has not improved for 5 consecutive iterations (i.e., the fitness change is <1%). j The formula for calculating average urgency is: In the formula, U 平均,j The average urgency of the task group assigned to the verification model j (as the basis for adjusting the verification sensitivity of that group). G is the sum of the urgency levels of all tasks within the task group. j U is the set of tasks assigned to the validation model j; i The urgency of validation for data item i.

[0088] In practice, the construction of an information verification model and the implementation of multi-dimensional data verification include:

[0089] The information verification model is constructed by including Mahalanobis distance verification model, isolated forest verification model and physical constraint verification model. The verification model extracts data sub-vectors from three concurrent verification task groups G1, G2 and G3 in turn, and loads the running configuration parameters synchronously. The running configuration parameters include shrinkage coefficient, depth limit and residual tolerance threshold.

[0090] After completing one iteration of optimization, the resource configuration parameter scheduling engine encapsulates the optimal allocation scheme into a structured message and pushes it to the managers of the three verification models via inter-process communication. The verification model managers parse the message, confirm that the three task groups cover all six data items (i.e., multi-dimensional cross-sectional survey information) without duplication, and then distribute them to the corresponding verification model execution units. For the Mahalanobis distance verification model, the shrinkage coefficient is set in the covariance estimation module, overriding the default value. For the isolated forest verification model, the depth limit is written into the isolated forest constructor, replacing the globally unified depth. For the physical constraint verification model, the residual tolerance threshold is loaded into the constraint check function as a dynamic limit on the right-hand side of the inequality. The binding process is completed through the constructor (or parameter update interface) of each verification model, ensuring that the verification model is in the sensitivity state required for scheduling before the verification begins. The verification model manager creates an independent execution thread (or process) for each verification model and passes in the corresponding data sub-vector and the bound configuration object to ensure that the three verification models start timing execution simultaneously, avoiding time measurement errors caused by differences in initialization order. There is no data interaction between the three threads, and they run completely independently.

[0091] In practice, building an information verification model and conducting multi-dimensional data verification also includes:

[0092] The Mahalanobis distance verification model uses a shrinkage coefficient to perform linear shrinkage estimation of the covariance matrix to calculate the Mahalanobis distance, and compares it with the chi-square threshold to determine anomalies. It quantifies the anomaly contribution of each dimension through Cholesky decomposition and outputs the anomaly labeling result A.

[0093] The steps for calculating the Mahalanobis distance using a linear shrinkage estimation of the covariance matrix with a shrinkage coefficient are as follows:

[0094] Step 1), see Figure 2 The formula for calculating the historical mean vector of the corresponding geological stratification label data item under normal working conditions in the data sub-vector of the verification task group G1 is as follows: In the formula, μ is the mean vector of the dimension subset verified by the Mahalanobis distance verification model under normal working conditions; N is the total number of historical samples. It is the sub-vector of the dimension verified by the Mahalanobis distance verification model in the k-th historical normal sample; The normalization factor is used; the formula for calculating the sample covariance matrix is:

[0095]

[0096] In the formula, S is the sample covariance matrix; N-1 is the degrees of freedom; This is the deviation vector for the k-th sample (used to quantify the direction and magnitude of a single sample's deviation from the mean). This is the transpose of the deviation vector; This is the outer product matrix of the bias (i.e., the contribution of the k-th sample to the covariance);

[0097] Step 2) Use the shrinkage coefficient to perform linear shrinkage estimation on the sample covariance matrix S to obtain the regularized covariance matrix. The linear shrinkage estimation formula is: Σ A =λ A ·S+(1-λ A )·diag(S), where Σ A The shrunken sample covariance matrix (improving the numerical stability of covariance matrix estimation in small sample or high-dimensional scenarios); λ A (1-λ) is the shrinkage coefficient; A ) represents the contraction strength of the diagonal matrix; diag(S) is a diagonal matrix that retains only the diagonal elements of S; to achieve dynamic matching between verification sensitivity and construction risk;

[0098] Step 3) The formula for calculating the squared Mahalanobis distance is: In the formula X is the squared Mahalanobis distance between the current data subvector and the normal operating condition (used to measure the overall degree of anomaly); (A) This is the current subvector of data to be verified; (X) (A) -μ) T This is the transpose of the deviation vector; This is the inverse of the shrunken sample covariance matrix (used to dynamically adjust the scaling in each direction by injecting scheduling sensitivity through the shrinkage coefficient); the confidence level is set at 97.5%, with I degrees of freedom. A The chi-square distribution quantile is used as the overall anomaly threshold and denoted as X. I,A ,when If the condition is met, the current subvector is determined to be normal overall, and a zero-check vector is directly generated; otherwise, the entire subvector is determined to be abnormal, and the Cholesky decomposition and quantization step is initiated.

[0099] The steps described above for quantifying the anomaly contribution of each dimension using Cholesky decomposition are as follows:

[0100] Step 1) Perform Cholesky decomposition on the shrunken sample covariance matrix to obtain the lower triangular matrix L, which satisfies Σ A =L·L T In the formula L T It is the transpose of the lower triangular matrix L (i.e., the upper triangular matrix).

[0101] Step 2) The formula for calculating the normalized deviation vector is: z = L -1 (X (A) -μ), where z is the normalized bias vector; L -1 X is the inverse of the lower triangular matrix L; (A)This is the current subvector of data to be verified;

[0102] Step 3) The formula for calculating the anomaly contribution of each dimension is as follows: In the formula c k is the normalized contribution coefficient of the k-th dimension to the overall Mahalanobis distance; The squared Mahalanobis distance between the previous data subvector and the normal operating condition; The value is the square of the k-th component (i.e., the absolute contribution of that dimension). If the contribution of any dimension exceeds the uniform contribution level, the data of that dimension is judged to be abnormal and marked as 1; otherwise, it is marked as 0, and the mark is used as the abnormal label result A.

[0103] The isolated forest verification model uses urgency dynamic constraints to limit the depth of the isolated forest, performs verification, calculates the feature split depth contribution index, and outputs anomaly labeling result B.

[0104] See Figure 3 The above verification and calculation of the feature splitting depth contribution index includes the following steps:

[0105] Step 1) Load the isolated forest validation model. The validation model parameters include T binary trees (completely randomized), each tree is constructed from a subset of historical normal operating condition samples, and the original maximum depth is H. max =[log2(ψ)], where ψ=256, and loads the average path length statistics of normal samples in each tree for subsequent normalization calculation;

[0106] Step 2) Use the depth limit in the running parameters as the upper limit of the traversal depth of each tree in the current cycle. When the depth limit > H max When the effective depth limit is H, then H is taken as the effective depth limit. max Otherwise, the effective depth limit is taken as the depth limit in the running parameters, which is dynamically calculated by the average urgency of G2.

[0107] Step 3) For the data sub-vectors in the current verification task group G2, traverse each binary tree, starting from the root node and traversing downwards. At each internal node, based on the node's splitting characteristics and splitting value, divide the sample into a left or right child node. Repeat this process until either of the following termination conditions is met: 1. A leaf node is reached; 2. The depth of the current tree has reached the effective depth limit. Record the final stopping depth as the path length of this tree. When traversal stops due to condition 2, a correction term needs to be added to estimate the remaining path length. The correction value uses the standard Isolation Forest method formula: h t =h 深度 +c(n s ), where h th is the estimated final path length of the sample in the t-th binary tree (used to calculate the average path length across all trees); 深度 c(n) represents the number of edges that have actually been traversed from the root node to the terminal node of the sample. s (n) represents the number of node samples. s The average path length correction function for n; s n is the number of training samples contained in the termination node; where n s The set of formulas for determining the value of is as follows:

[0108]

[0109] In the formula, 2H(n-1) is twice the harmonic term; is the linear correction term; n is the number of samples in the subset;

[0110] Step 4) The formula for calculating the average path length of the current data subvector across all T binary trees is: In the formula, E(h) is the average path length; Average factor; h t Let be the estimated final path length of the sample in the t-th binary tree (obtained from the above calculation); using the normalization constant of the expected path length corresponding to the average size of the subsamples in all binary trees as the normalization benchmark, the outlier score is calculated as follows: In the formula s(X) (B) c(ψ) represents the anomaly score of the data subvector in the current verification task group G2; avg ) represents the average subsample size during training of all binary trees; 2 - It is an exponentially decaying function; when s(X) (B) When s(X) ≤ 0.5, the data subvector is judged to be normal overall, and a local check vector of all zeros is generated. (B) If the value is greater than 0.5, it is determined to be an overall anomaly and proceeds to step 5).

[0111] Step 5), the formula for calculating the anomaly contribution index for each feature (i.e., data item) is:

[0112]

[0113] In the formula γ k Let AvgSD(k) be the contribution index of feature k to the overall anomaly; AvgSD(k) is the average depth at which feature k appears as a splitting feature in all T binary tree paths; [AvgSD(k)] -1 It is the reciprocal of the average depth; I B The set of task indices assigned to the isolated forest validation model; AvgSD(l) is the average split depth of feature l across all binary trees; when γ k >1 / I BIf the overall anomaly score s > 0.5, then dimension k is marked as anomaly; otherwise, it is marked as normal. The all-zero local verification vector and the marking result of this step are used together as the anomaly marking result B.

[0114] The physical constraint verification model adaptively tightens the residual tolerance threshold based on urgency, performs physical inequality constraint verification, and calculates the normalized sensitivity of each dimension to obtain the anomaly labeling result C.

[0115] See Figure 4 The adaptive tightening of the residual tolerance threshold based on urgency includes: loading a physical constraint verification model, which stores a set of physical inequality constraints applicable to the current construction area. Each constraint reflects the theoretical or empirical relationship that should be satisfied between two (or more) pieces of exploration information. An example is taken using the typical constraints of pore pressure and acoustic P-wave velocity, with the following inequality constraint formula:

[0116]

[0117] In the formula This represents the constraint deviation function value of the data subvectors in the current verification task group G3 (used to measure the degree of deviation of the current exploration information from geotechnical physical experience); v p ρ is the longitudinal wave velocity of the acoustic wave; p is the formation pore pressure; ρ is the water density; a is an empirical coefficient (taken as 1.2 × 10⁻⁶). -3 (Based on the work area calibration); b is the empirical constant offset (taken as 0.15); ε is the normalized predicted value of the theoretical wave velocity; C ε is the current adaptive residual tolerance threshold; where ε C Based on the dynamic calculation of the average urgency of the task group, the calculation formula is: ε C =ε0×(1+δ×(1-U 平均,C In the formula, ε0 is the basic allowable residual (value 0.15); δ is the relaxation sensitivity coefficient (value 0.5); U 平均,C The average tightness of the physical constraint verification model; when U 平均,C When the threshold approaches 0.9 (i.e., high risk), the restrictions are only slightly relaxed; when U... 平均,C When the risk level is 0.2 (i.e., low risk), the constraint is relaxed; this dynamic mechanism tightens the verification standard in high-risk areas and appropriately tolerates normal fluctuations in low-risk areas, thereby reducing false alarms.

[0118] Performing physical inequality constraint checks and calculating normalized sensitivities for each dimension includes:

[0119] Step 1) Use the inequality constraint formula (removing threshold comparison) to calculate the physical constraint residual of the data subvector in the current verification task group G3, denoted as R;

[0120] Step 2) Set the physical constraint residual R and the adaptive residual tolerance threshold ε. C Anomaly comparison determination, when R < ε C When R > ε, it is determined that the current data subvector as a whole conforms to the physical constraints, and all relevant dimensions are marked normally (the corresponding positions of the local verification vectors are set to 0). C When a violation of physical laws is detected, and the overall system is deemed abnormal, proceed to step 3).

[0121] Step 3) When the overall situation is abnormal, the sensitivity formula for the exploration information item that causes the constraint to be violated due to the location item is:

[0122]

[0123] In the formula S k The standardized sensitivity of variable k (used to measure the degree of influence of unit changes in this exploration information item on constraint deviation); x k is the variable currently involved in the constraint; D is the internal deviation value; To constrain the absolute value of the maximum deviation in historical normal data; The deviation function with respect to variable x k The partial derivatives for the variable x; k The partial derivatives include:

[0124] 1. The formula for calculating pore pressure p is: In the formula v is the first partial derivative of the deviation function D with respect to the pore pressure p (used to quantify the rate and direction of the constraint deviation change caused by a small change in pore pressure); p ρ is the longitudinal wave velocity of the acoustic wave; p is the formation pore pressure; ρ is the water density; a is an empirical coefficient. It is the square root of the product of pore pressure p and water density ρ; is the square of the theoretical prediction value (i.e., the theoretical normalized wave velocity prediction value); b is the empirical constant offset. It is a square inverse factor (which modulates the amplitude of the partial derivative by the theoretical value of the current operating point).

[0125] 2. Regarding v p The formula for calculating the longitudinal wave velocity of sound is: In the formula It is the first partial derivative of the deviation function D with respect to the longitudinal wave velocity of the acoustic wave (used to quantify the rate at which a small change in wave velocity causes a change in the constraint deviation). The normalized wave velocity prediction value is the theoretical value. The sensitivity is normalized. When the normalized sensitivity of any dimension k exceeds the uniform contribution threshold 1 / m (m is the total number of dimensions involved in the constraint), the dimension is marked as an anomaly (the corresponding position in the local verification vector is set to 1), otherwise it is set to 0. The local verification vector is used as the anomaly marking result C.

[0126] In practice, building an information verification model and conducting multi-dimensional data verification also includes:

[0127] Collect anomaly labeling results A, B, and C, and combine them into a comprehensive verification vector;

[0128] The actual processing time feedback from the validation model is smoothed and updated using an exponential moving average to obtain the estimated processing time matrix for subsequent scheduling decisions.

[0129] In practice, the construction of the normal operating condition vector, as well as the element-by-element comparison and confidence calculation, includes:

[0130] Retrieve abnormality vector samples from the historical normal working condition database during normal drilling, calculate the normal mean and covariance matrix, and replace the identity matrix with the sample when the sample is insufficient, and the mean approaches zero.

[0131] A dynamic weight matrix is ​​constructed based on the urgency, and a weighted Mahalanobis distance value is calculated for the deviation between the comprehensive verification vector and the normal mean, so as to amplify the abnormal contribution value of the high-risk dimension.

[0132] Calculate the average urgency of all verified models in the current period, calculate the confidence level by the ratio of the weighted Mahalanobis distance value to the dynamic comparison threshold, and determine anomalies item by item based on the anomaly contribution value, and output the comparison result vector.

[0133] The above calculations of the normal mean and covariance matrix use the following formula for the normal mean: In the formula, μ0 is the mean vector of the anomaly vector under normal operating conditions (serving as the reference center for comparison); N0 is the number of matched samples in the historical normal operating condition database; v k Let be the comprehensive verification vector corresponding to the k-th historical normal sample; its covariance matrix is ​​calculated using the following formula:

[0134]

[0135] In the formula, Σ0 is the covariance matrix of the anomaly vector under normal operating conditions (used as the metric structure in Mahalanobis distance calculation, replacing the standard Euclidean distance to consider the natural correlation between dimensions); (v k -μ0) is the deviation vector of the k-th sample relative to the mean; (v k -μ0) T This is the transpose of the deviation vector.

[0136] The above calculation of the weighted Mahalanobis distance value for the deviation between the comprehensive verification vector and the normal mean includes:

[0137] Step 1) Utilize the urgency U of each data item output by the scheduling process. i Construct a dynamic weight matrix W=diag(wi ), where w i =0.5+0.5×U i ;

[0138] Step 2) The weighted Mahalanobis distance calculation formula for the deviation between the current comprehensive verification vector and the normal mean is as follows:

[0139]

[0140] In the formula D 2 V is the squared weighted Mahalanobis distance (used to measure the deviation of the current integrated verification vector from the normal operating condition baseline in multidimensional space); V is the current integrated verification vector; μ0 is the mean vector of the anomaly vector under normal operating conditions; W T This is the transpose of the dynamic weight matrix W; is the inverse of the normal operating condition covariance matrix; (V-μ0) is the deviation vector (i.e., the deviation between the current comprehensive verification vector and the normal operating condition benchmark).

[0141] The formula for calculating the average urgency of all validation models in the current period is: In the formula, U 全局 The global average urgency; U 平均,j The average urgency of the j-th task group assigned to the j-th validation model; The averaging factor is represented by the formula for determining the dynamic comparison threshold: θ = θ0 + η × (U 全局 -0.5), where θ is the dynamic comparison threshold; θ0 is the basic threshold (taken as the 95th quantile of the chi-square distribution of the dimensions of the comprehensive verification vector V with n degrees of freedom); η is the urgency sensitivity coefficient (taken as 4.0, used to control the magnitude of the threshold change with urgency); when D 2 When the value is greater than θ, the current cross-sectional survey information is determined to deviate from the normal working conditions; otherwise, it is considered to be normal.

[0142] The formula for calculating the confidence level is: In the formula, Conf is the confidence level value; exp is the exponential function; D 2 θ is the square of the weighted Mahalanobis distance; θ is the dynamic comparison threshold; the confidence level ranges from [0,1], with a larger value indicating that the current state is closer to normal operating conditions; for each data item i, the formula for calculating the normalized contribution value is:

[0143]

[0144] In the formula g i The normalized contribution value for the i-th data item (used to locate the data item causing the overall anomaly); w i The dynamic weight of the i-th data item (obtained from the dynamic weight matrix); [(W(V-μ0))i ] 2 The square of the i-th element of the weighted bias vector; [(W(V-μ0)) k ] 2 The square of the k-th component of the weighted bias vector; when g i >1 / n and simultaneously D 2 If the value is greater than θ, then the data comparison is abnormal (marked as 1); otherwise, it is normal (marked as 0). Output the comparison result vector and confidence level for each item.

[0145] In specific implementation, a structured analysis result vector group is constructed, which includes data item parameter identifier, data item name, current actual survey measurement value, normal working condition average value, lower limit of normal range, upper limit of normal range, comparison anomaly mark, deviation value, and confidence level; the communication protocol uses the MQTT protocol, and the upload cycle is synchronized with the data acquisition.

[0146] The analysis result vector group is a structured vector group. All fields are packaged in a fixed order to form a binary (or text) format that is easy to transmit. The upload cycle is initially set to 2 seconds / upload. Each uploaded structured vector group is fully stored in the historical database, and an event record containing timestamps, geological layers, confidence levels, etc. is generated.

[0147] Those skilled in the art will recognize that, in conjunction with the embodiments described herein, the units and algorithm steps of the various examples should be noted. Operations such as data acquisition are conventional and therefore will not be described in detail. Specific implementation methods can be achieved using electronic hardware, computer software, or a combination of both. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

Claims

1. A method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines, characterized in that, include: Acquire multi-dimensional cross-sectional survey information and simultaneously extract multi-dimensional working condition labels to construct survey information vectors and working condition label sets; An improved ant colony algorithm is used to construct a resource configuration parameter scheduling engine. The engine takes the exploration information vector and the working condition label set as input parameters, and outputs a verification task group after processing. The verification task group is accompanied by urgency and running configuration parameters. Construct an information verification model, with each model running concurrently and independently. Each model receives verification task groups, dynamically adjusts the verification sensitivity, performs multi-dimensional data verification, outputs the verification results separately, and merges them into a comprehensive verification vector. Normal working condition data is obtained from the historical normal working condition database through the data interface to form a normal working condition vector; An improved weighted Mahalanobis distance is used to construct a data comparison model. The normal working condition vector and the comprehensive verification vector are compared element by element and the confidence level is calculated to obtain the comparison result vector. The comparison results are used to construct a structured analysis result vector group, and each data item in the analysis result vector group is uploaded to the management platform for monitoring using a communication protocol.

2. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 1, characterized in that, The multi-dimensional cross-sectional exploration information includes formation resistivity, acoustic longitudinal wave velocity, borehole bottom annular pressure, rock mass integrity coefficient, actual drilling rig thrust, and drilling rig rotation torque; the working condition label set includes geological stratification labels, risk source proximity distance labels, and drilling speed change indicator labels.

3. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 1, characterized in that, Using an improved ant colony algorithm, a resource configuration parameter scheduling engine is built, and the process of generating and verifying task groups includes: The data in the exploration information vector is mapped to the verification task, the information verification model is mapped to the execution slot, and the pheromone matrix is ​​constructed, and the number of ant colonies, the maximum number of iterations and the importance factor are set. The estimated processing time, urgency, and historical anomaly rate of each verification task are combined and calculated. The weights of each factor are dynamically adjusted by the CPU load ladder function to obtain the heuristic factor used to guide the ant search. Using a roulette wheel selection strategy based on pheromone matrix and heuristic factors, and imposing execution slot capacity constraints, each ant determines the information verification model for all verification tasks one by one, dividing the tasks into three concurrent verification task groups: G1, G2, and G3.

4. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 3, characterized in that, The process of building a resource configuration parameter scheduling engine using an improved ant colony algorithm and verifying the generation of task groups also includes: The verification task group was quantitatively evaluated, and elite ants were selected to carry out global pheromone volatilization and deposition to obtain an updated pheromone matrix to guide the iterative search. The distance to the risk source is extracted from the working condition label set and the risk level is calculated. The pheromone volatilization factor is dynamically adjusted by substituting it into a time-varying formula that decreases with risk, and an adaptive volatilization factor value is obtained to regulate the global volatilization of pheromones. Set the iteration termination trigger condition, calculate the urgency of each task group after the iteration is completed, and generate the running configuration parameters and voting weights of the information verification model.

5. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 1, characterized in that, Building an information verification model and conducting multi-dimensional data verification includes: The information verification model is constructed by including Mahalanobis distance verification model, isolated forest verification model and physical constraint verification model. The verification model extracts data sub-vectors from three concurrent verification task groups G1, G2 and G3 in turn, and loads the running configuration parameters synchronously. The running configuration parameters include shrinkage coefficient, depth limit and residual tolerance threshold.

6. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 5, characterized in that, Building an information verification model and conducting multi-dimensional data verification also includes: The Mahalanobis distance verification model uses a shrinkage coefficient to perform linear shrinkage estimation of the covariance matrix to calculate the Mahalanobis distance, and compares it with the chi-square threshold to determine anomalies. It quantifies the anomaly contribution of each dimension through Cholesky decomposition and outputs the anomaly labeling result A. The isolated forest verification model uses urgency dynamic constraints to limit the depth of the isolated forest, performs verification, calculates the feature split depth contribution index, and outputs anomaly labeling result B. The physical constraint verification model adaptively tightens the residual tolerance threshold based on urgency, performs physical inequality constraint verification, and calculates the normalized sensitivity of each dimension to obtain the anomaly labeling result C.

7. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 6, characterized in that, Building an information verification model and conducting multi-dimensional data verification also includes: Collect anomaly labeling results A, B, and C, and combine them into a comprehensive verification vector; The actual processing time feedback from the validation model is smoothed and updated using an exponential moving average to obtain the estimated processing time matrix for subsequent scheduling decisions.

8. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 1, characterized in that, The construction of the normal operating condition vector, and the element-by-element comparison and confidence calculation include: Retrieve abnormality vector samples from the historical normal working condition database during normal drilling, calculate the normal mean and covariance matrix, and replace the identity matrix with the sample when the sample is insufficient, and the mean approaches zero. A dynamic weight matrix is ​​constructed based on the urgency, and a weighted Mahalanobis distance value is calculated for the deviation between the comprehensive verification vector and the normal mean, so as to amplify the abnormal contribution value of the high-risk dimension. Calculate the average urgency of all verified models in the current period, calculate the confidence level by the ratio of the weighted Mahalanobis distance value to the dynamic comparison threshold, and determine anomalies item by item based on the anomaly contribution value, and output the comparison result vector.

9. The method for verifying and analyzing information from directional drilling cross-section surveys of long-distance natural gas pipelines according to claim 1, characterized in that, A structured analysis result vector group is constructed, which includes data item parameter identifier, data item name, current actual survey measurement value, normal working condition mean, lower limit of normal range, upper limit of normal range, comparison anomaly marker, deviation value, and confidence level; the communication protocol uses the MQTT protocol, and the upload cycle is synchronized with the data acquisition.

Images