A method and device for detecting and determining corrosion and fouling risks of oil and gas field facilities

Abstract

This specification provides a method and apparatus for detecting and determining the corrosion and scaling risk of oil and gas field facilities. Based on the original dataset, production impact parameters related to corrosion and scaling are extracted. Using a pre-defined facility loss prediction model, multi-dimensional loss prediction values ​​are obtained from the original dataset. The pre-defined facility loss prediction model is then used to process the first and second parameter sets separately to obtain corresponding prediction output differences. Based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, the corresponding dynamic weight sequence of parameters is determined. Based on the multi-dimensional loss prediction values, the dynamic weight sequence of parameters, the material allowable limits of the target oil and gas field facilities, and a pre-defined dynamic threshold, the target risk detection result is determined. This method efficiently and accurately determines the corrosion and scaling risk.

Description

Technical Field

[0001] This manual belongs to the field of oil and gas field development technology, and in particular relates to a method and device for detecting and determining the risk of corrosion and scaling in oil and gas field facilities. Background Technology

[0002] Given that oil and gas field facilities are prone to corrosion and scaling over time, which can affect the normal transportation of oil and gas, we need to predict the risk of corrosion and scaling in advance so that we can deal with it in a timely manner. However, existing methods cannot efficiently and accurately predict the risk of corrosion and scaling.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This specification provides a method and apparatus for detecting and determining the corrosion and scaling risk of oil and gas field facilities, which solves the technical problem that the existing technology lacks multi-physical mechanism hierarchical coupling and dynamic marginal contribution identification of parameters, resulting in low accuracy and poor reliability of early warning for predicting corrosion and scaling risk of oil and gas field facilities.

[0005] This specification provides a method for determining the risk of corrosion and scaling in oil and gas field facilities, including: Obtain the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; Based on the original dataset, production impact parameters related to corrosion and scaling were extracted; Using a pre-defined facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein, the pre-defined facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. Select a target production impact parameter from the plurality of production impact parameters, and construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter; Using the preset facility loss prediction model, the corresponding prediction output differences are obtained by processing the first parameter set and the second parameter set respectively; based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, determine the corresponding dynamic weight sequence of parameters; The target risk detection result is determined based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facilities, and the preset dynamic threshold.

[0006] In one embodiment, after obtaining the raw dataset of the target oil and gas field facilities, the method further includes: Calculate the statistical parameters of each data point in the original dataset within a preset sliding window, and mark data that deviate from the statistical parameters by more than a dispersion threshold as abnormal data to be processed; By identifying the changing trends of the abnormal data to be processed, the deviation type of the abnormal data to be processed is confirmed; wherein, the deviation type includes continuous deviation type and occasional deviation type; When the deviation type is the occasional deviation type, normal feature data in the neighborhood of the abnormal data to be processed is extracted, a replacement value is generated using a preset smooth reconstruction algorithm, and a value replacement strategy is executed on the abnormal data to be processed to obtain the cleaned dataset. When the deviation type is the persistent deviation type, the physical cause is determined based on the evolution trend characteristics of the abnormal data to be processed; and according to the physical cause, the corresponding benchmark deviation correction strategy is executed on the abnormal data to be processed to obtain the cleaned dataset. The cleaned dataset is then filled with missing data and standardized according to a pre-defined set of rules to obtain the processed original dataset.

[0007] In one embodiment, the multiphase flow prediction submodule is a module based on a fluid dynamics mapping structure, the scaling prediction submodule is a module based on a thermodynamic equilibrium calculation structure, and the corrosion prediction submodule is a module based on an electrochemical reaction mapping structure. The multiphase flow prediction submodule includes at least a nonlinear feedforward layer that performs feature dimensionality reduction and flow pattern distribution feature mapping on the production impact parameters in the original dataset to determine the fluid dynamics parameters. The scaling prediction submodule includes at least a feature extraction layer for determining the saturation index and crystal nucleation trend based on the ion concentration characteristics in the production impact parameters. The corrosion prediction submodule includes at least a feedback coupling layer that uses the fluid dynamics parameters and the partial pressure of gas components in the production impact parameters to map the interface mass transfer intensity and extract corrosion rate evolution features.

[0008] In one embodiment, the preset facility loss prediction model is used to obtain corresponding prediction output differences by processing the first parameter set and the second parameter set respectively; based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined, including: Using the facility loss prediction model, the first output response corresponding to the first parameter set and the second output response corresponding to the second parameter set are determined respectively; Calculate the deviation between the first output response and the second output response to obtain the predicted output difference; Traverse the subset spaces of different dimensions composed of the multiple production impact parameters, and determine the corresponding attribution weights based on the arrangement and distribution characteristics of each subset space in the full parameter space; The differences in the corresponding predicted outputs are weighted and summed according to the attribution weights to obtain the marginal contribution increment of each target production impact parameter under different parameter combinations; wherein, the target production impact parameter includes one or more production impact parameters.

[0009] In one embodiment, determining the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations includes: Normalization calculation is performed on the marginal contribution increment corresponding to each of the target production impact parameters to obtain the weighted impact factor of each of the target production impact parameters for the multidimensional loss prediction value. Based on the magnitude of the weighted influence factors, the target production influence parameters are prioritized to obtain an initial weight sequence characterizing the distribution of the contribution intensity of each parameter; The evolution trend features of each data item in the original dataset within a preset time window are extracted, and the initial weight sequence is corrected in a time direction using the evolution trend features to generate the parameter dynamic weight sequence.

[0010] In one embodiment, determining the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and a preset dynamic threshold includes: The risk sensitivity coefficient is determined based on the dynamic weight sequence of the parameters and the material allowable limit data of the target oil and gas field facilities. The preset dynamic threshold is corrected based on the risk adjustment coefficient to determine the target dynamic threshold; The multidimensional loss prediction value is compared and mapped with the target dynamic threshold to determine the target risk detection result.

[0011] In one embodiment, based on the target risk detection results, a set of corresponding candidate protection schemes is determined from a preset protection knowledge base; Based on the on-site cost accounting factors, determine the economic adaptability characteristics of each scheme in the candidate protection scheme set; Based on the economic adaptability characteristics of each scheme in the candidate protection scheme set, the protection scheme corresponding to the target oil and gas field facility is determined.

[0012] This manual provides a device for detecting and determining the corrosion and scaling risk of oil and gas field facilities, including: The data acquisition module is used to acquire the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; The influencing parameter determination module is used to extract production influencing parameters related to corrosion and scaling based on the original dataset. The model prediction module is used to obtain multidimensional loss prediction values ​​based on the original dataset using a preset facility loss prediction model; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. The parameter set determination module is used to select a target production impact parameter from a plurality of production impact parameters, and to construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter. The contribution increment determination module is used to obtain the corresponding prediction output difference by processing the first parameter set and the second parameter set respectively using the preset facility loss prediction model; and to determine the marginal contribution increment of each target production impact parameter under different parameter combinations based on the prediction output difference. The weight sequence determination module is used to determine the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations. The detection result determination module is used to determine the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and a preset dynamic threshold. This specification also provides an electronic device, including a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements a method for determining the corrosion and scaling risk of oil and gas field facilities.

[0013] This specification also provides a computer-readable storage medium having computer instructions stored thereon, which, when executed, implement a method for detecting and determining the risk of corrosion and scaling in oil and gas field facilities.

[0014] Based on the corrosion and scaling risk detection method for oil and gas field facilities provided in this specification, the method involves obtaining the original dataset of the target oil and gas field facilities; wherein the original dataset includes static data, dynamic data, and monitoring data; extracting production impact parameters related to corrosion and scaling based on the original dataset; and obtaining multidimensional loss prediction values ​​using a pre-defined facility loss prediction model based on the original dataset; wherein the pre-defined facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule; and selecting target production impact parameters from multiple production impact parameters and constructing corresponding models. The system includes a first set of parameters containing target production impact parameters and a second set of parameters not containing the target production impact parameters. Using the preset facility loss prediction model, the system processes the first and second parameter sets respectively to obtain corresponding prediction output differences. Based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, the corresponding parameter dynamic weight sequence is determined. Based on the multidimensional loss prediction value, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facilities, and the preset dynamic threshold, the target risk detection result is determined. In this way, by utilizing a facility loss prediction model composed of hierarchical coupling of multiphase flow, scaling, and corrosion sub-modules, quantitative output of multidimensional loss prediction values ​​is achieved. This enables a forward-looking quantitative characterization of the loss status of oil and gas field facilities from multiple physical mechanism dimensions. By calculating the difference in predicted outputs generated by the first parameter set and the second parameter set, the marginal contribution increment of each production impact parameter and the dynamic weight sequence of the parameters are accurately extracted, achieving quantitative identification of the risk-dominant factors. Combining the material allowable limit data and the target risk detection results determined by dynamic thresholds, both facility safety and real-time loss evolution trends are taken into account, providing a scientific basis for the early prediction and timely treatment of corrosion and scaling risks, and ensuring the normal transportation of oil and gas. Attached Figure Description

[0015] To more clearly illustrate the embodiments of this specification, the accompanying drawings used in the embodiments will be briefly introduced below. The drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a method for detecting and determining the corrosion and scaling risk of oil and gas field facilities, provided in one embodiment of this specification. Figure 2 This is a schematic diagram of the electronic device structure provided in one embodiment of this specification; Figure 3 This is a schematic diagram of the structural composition of a corrosion and scaling risk detection and determination device for oil and gas field facilities, provided in one embodiment of this specification. Figure 4 This is a schematic diagram of a process for the whole life cycle management of corrosion and scaling in oil and gas fields, provided by one embodiment of this specification. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0018] See Figure 1 As shown in the embodiments of this specification, a method for detecting and determining the corrosion and scaling risk of oil and gas field facilities is provided, wherein the method is specifically applied to the server side. In specific implementation, the method may include the following: S101: Obtain the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; S102: Based on the original dataset, extract the production impact parameters related to corrosion and scaling; S103: Using a preset facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. S104: Select a target production impact parameter from the plurality of production impact parameters, and construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter; S105: Using the preset facility loss prediction model, by processing the first parameter set and the second parameter set respectively, the corresponding prediction output difference is obtained; based on the prediction output difference, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined; S106: Determine the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations; S107: Determine the target risk detection result based on the multidimensional loss prediction value, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facility, and the preset dynamic threshold.

[0019] The aforementioned oil and gas field facilities can refer to the physical hardware carriers involved in the entire process of oil and gas extraction, gathering, transportation, and processing. Specifically, these include wellbore facilities such as oil production wells, water injection wells, and water source wells (involving well structure and wellbore design specifications), as well as the surface production pipeline network system consisting of surface pipelines, station equipment, and valve chambers (involving pipeline network distribution and pipeline network ledgers). These facilities cover the wellbore and pipelines from the bottom reservoir to the surface processing terminal and are the core production units that carry multiphase flow media and are susceptible to corrosion and scaling risks.

[0020] The aforementioned raw dataset refers to the underlying data set required for risk detection of target oil and gas field facilities, serving as the foundational input for subsequent feature extraction and model calculation. This dataset is compiled through real-time data acquisition from field sensors, export from the production management system, and aggregation from historical design documents, covering all raw information reflecting the physical attributes and operating conditions of the facilities.

[0021] The aforementioned static data refers to facility attribute data that remains unchanged or changes at a very low frequency within a certain production cycle. Specifically, it includes the facility's material parameters (such as model and modulus of elasticity), geometric structure data (such as pipe diameter, wall thickness, length, and inclination angle), geographical location information, and design service life, etc., which are used to provide fixed physical boundary conditions for the model.

[0022] The aforementioned dynamic data refers to a set of operating parameters that fluctuate in real time with production operations. These mainly include fluid pressure, temperature, flow rate (three-phase flow of oil, gas, and water), fluid composition (such as CO2 and H2S content), water quality analysis indicators (such as scale-forming ion concentration and pH value), and chemical reagent dosage monitored by various instruments, reflecting the real-time evolution of the facility's internal environment.

[0023] The aforementioned monitoring and testing data can refer to media detection, tack strip detection, well logging management, probe monitoring, wall thickness detection, and chemical dosing management. This data encompasses media detection data used to assess fluid properties, tack strip and wall thickness detection data reflecting actual metal loss, probe monitoring data used to capture electrochemical signals in real time, well logging management data related to wellbore integrity evaluation, and chemical dosing management data reflecting the implementation of protective measures. These monitoring and testing data, together with static design data and dynamic production data, constitute a high-integrity influencing factor set required for risk prediction, providing direct experimental and measured evidence for accurately identifying the corrosion and scaling status of facilities.

[0024] The aforementioned production impact parameters refer to key characteristic variables extracted from the original dataset that have a significant driving or inhibiting effect on the corrosion and scaling process. These parameters, after data cleaning and feature engineering, can be directly mapped to the kinetic mechanisms of corrosion and scaling, such as phase separation flow rate, supersaturation, and effective reaction concentration, and are the direct operational objects for model calculations.

[0025] The aforementioned facility deterioration prediction model can be a composite prediction architecture built upon hierarchical coupling of physical mechanisms. Through the logical nesting and data transfer of multiple internal sub-modules, this model can simulate the complete physical evolution process from fluid dynamics distribution to chemical deposition and then to electrochemical damage, thereby achieving quantitative prediction of the future deterioration state of the facility.

[0026] The aforementioned multidimensional loss prediction value refers to the quantitative index vector output by the facility loss prediction model, which characterizes the degree of facility damage. This prediction value is not a single-dimensional conclusion, but a multidimensional data set composed of wall shear force reflecting mechanical scouring intensity, predicted scale thickness reflecting the degree of chemical deposition, and predicted corrosion rate reflecting the intensity of electrochemical damage. It objectively depicts the current loss status of the facility from multiple dimensions.

[0027] The aforementioned multiphase flow prediction submodule serves as the foundational layer of the loss prediction model, simulating the flow behavior of multiphase fluids such as oil, gas, and water within pipelines. This module calculates the flow pattern, liquid holdup, and local pressure distribution based on the input dynamic parameters, with a focus on outputting the wall shear force, which is closely related to wall damage, providing a fluid dynamics benchmark for subsequent scaling and corrosion predictions.

[0028] The aforementioned scaling prediction submodule can serve as an intermediate layer in the loss prediction model, used to calculate the precipitation and deposition patterns of scaling components in the fluid under specific operating conditions. This module, combined with the distribution characteristics of the multiphase flow field, predicts and outputs the scaling trend and amount. The aforementioned corrosion prediction submodule can serve as the damage calculation layer in the loss prediction model, used to evaluate the electrochemical thinning process of metallic materials in specific media environments. This module receives parameters such as shear force from the multiphase flow submodule to comprehensively determine the uniform corrosion rate.

[0029] The aforementioned difference in predicted output refers to the change in model output when the input parameter set is changed under controlled conditions using the same prediction model. Specifically, the degree of disturbance to the overall loss prediction result caused by the change in the target parameter is quantified by comparing the multidimensional loss prediction values ​​under two input backgrounds: one with the target parameter included and the other without.

[0030] The aforementioned marginal contribution increment can be the net contribution value of a single production impact parameter to the model's prediction results, calculated based on game theory attribution logic. It reflects the incremental change in the predicted facility loss value caused by the presence of a specific parameter, excluding interference from other parameters, and is a core mathematical indicator for assessing the sensitivity of parameters to risk impact.

[0031] The aforementioned dynamic weight sequence of parameters refers to the real-time ranking of influence weights formed after normalization based on the marginal contribution increment of each production-influencing parameter. This sequence is dynamically updated with fluctuations in production conditions, and can intuitively reflect which factor (such as pressure fluctuations or composition changes) is the dominant cause of increased corrosion and scaling risk at the current moment.

[0032] The aforementioned dynamic threshold refers to a warning judgment boundary that dynamically changes with the facility status and dominant risk factors. Unlike traditional fixed thresholds, dynamic thresholds are scaled and corrected in real time based on the severity of the operating conditions reflected by the parameter weight sequence and the allowable limits of the materials. This makes the warning system more sensitive under aging facilities or high-risk operating conditions, and more fault-tolerant during safe and stable periods.

[0033] The aforementioned target risk detection results can serve as the final comprehensive decision-making instruction output by this method. It is the product obtained by matching and determining multidimensional loss predictions with dynamic thresholds, and includes a clear risk level, the dominant influencing factors triggering early warnings, and targeted maintenance strategy recommendations to guide timely risk management in oil and gas fields.

[0034] In some embodiments, extracting production impact parameters related to corrosion and scaling based on the original dataset may specifically include: S1: Impute missing values ​​and remove outliers from the original dataset, and normalize the processed data to obtain the standard working condition feature matrix. S2: Using a preset correlation analysis algorithm, calculate the correlation between the features of each dimension in the standard operating condition feature matrix and the historical loss data of the facility; S3: Based on the magnitude of the correlation, perform dimensionality reduction screening to extract features that have a significant driving effect on corrosion effect and scaling trend as the production impact parameters; wherein, the production impact parameters include at least fluid temperature, fluid pressure, total liquid production, water content, gas-oil ratio, CO2 content, H2S content and scale-forming ion concentration.

[0035] In some embodiments, the step of obtaining multidimensional loss prediction values ​​based on the original dataset using a preset facility loss prediction model may specifically include: S1: Input the original dataset into the multiphase flow prediction submodule to calculate the wall shear force and local velocity that characterize the fluid dynamics distribution, and pass the wall shear force as an input parameter to the corrosion prediction submodule; S2: Using the scaling prediction submodule, the predicted scaling trend and predicted scaling amount are calculated based on the scale-forming ion concentration and temperature and pressure parameters in the original dataset, combined with the working environment. S3: Using the corrosion prediction submodule, the predicted corrosion rate is calculated based on the corrosive medium composition in the original dataset and the received wall shear force and flow velocity, etc. S4: Combine the output wall shear force, the predicted scale thickness, and the predicted corrosion rate according to a preset vector dimension to generate the multidimensional loss prediction value.

[0036] By establishing a parameterized hierarchical transfer mechanism among the multiphase flow, scaling, and corrosion sub-modules, dynamic coupling of physical field information was achieved. The wall shear force was used to accurately characterize the scouring and damaging effect of fluid mechanics on the material protective film, and the physical evolution mechanism of the "flow-corrosion" interaction in oil and gas production was deeply restored. Compared with the isolated prediction of a single dimension, the multidimensional loss prediction generated by this aggregation not only significantly improves the accuracy and systematicness of the facility loss characterization under complex multiphase flow conditions, but also provides multidimensional and quantitative mechanistic data support for subsequent accurate risk attribution based on marginal contribution.

[0037] In some embodiments, determining the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and a preset dynamic threshold may specifically include: S1: Based on the dynamic weight sequence of the parameters, the wall shear force, predicted scale amount, predicted scale trend and predicted corrosion rate in the multidimensional loss prediction values ​​are weighted and mapped respectively to construct a loss evaluation vector that reflects the severity of the current working conditions. S2: Using the material allowable limit data as the denominator, the loss evaluation vector is normalized to calculate the failure risk evolution value that characterizes the proportion of facility damage. S3: Compare the failure risk evolution value with the preset dynamic threshold. If the failure risk evolution value exceeds the preset dynamic threshold, a risk warning signal is triggered. S4: Extract the number of influence parameters with the highest weights in the dynamic weight sequence of the parameters, identify them as the dominant risk source identifiers, and combine them with the risk warning signal to form the target risk detection result.

[0038] By using a dynamic weight sequence of parameters to perform weighted mapping on multidimensional loss prediction values, dimensionality reduction and fusion of complex damage components are achieved. The constructed loss evaluation vector can comprehensively and highly characterize the severity of operating conditions. By introducing material allowable limit data as the denominator for normalization, the calculated failure risk evolution value accurately reflects the matching relationship between the damage ratio of the facility body and the remaining bearing capacity, significantly improving the accuracy of risk assessment for different individual facilities. Combined with the target risk detection results output by the dominant risk source identifier, a deep correlation is achieved from risk level warning to specific cause tracing, providing a scientific and intuitive decision-making basis for taking precise and targeted prevention and control measures on site and ensuring the normal transportation of oil and gas.

[0039] In some embodiments, after obtaining the raw dataset of the target oil and gas field facilities, the method may further include the following: S1: Calculate the statistical parameters of each data item in the original dataset within a preset sliding window, and mark data that deviate from the statistical parameters by more than the dispersion threshold as abnormal data to be processed; S2: By identifying the changing trend of the abnormal data to be processed, the deviation type of the abnormal data to be processed is confirmed; wherein, the deviation type includes continuous deviation type and occasional deviation type; S3: When the deviation type is the occasional deviation type, extract the normal feature data in the neighborhood of the abnormal data to be processed, generate replacement values ​​using a preset smooth reconstruction algorithm, and execute a value replacement strategy on the abnormal data to be processed to obtain the cleaned dataset. S4: When the deviation type is the continuous deviation type, determine the physical cause based on the evolution trend characteristics of the abnormal data to be processed; and according to the physical cause, execute the corresponding benchmark deviation correction strategy on the abnormal data to be processed to obtain the cleaned dataset; S5: Perform missing data completion on the cleaned dataset and standardize the data according to preset unified standard rules to obtain the processed original dataset.

[0040] Specifically, the original dataset is first scanned in the time domain using a preset sliding window (e.g., using the past 24 hours as the window length). The mean and variance of various indicators (e.g., pressure, displacement) within the window are calculated as statistical parameters. If a sampling point deviates from the mean by more than a preset dispersion threshold (e.g., the 3σ principle), it is marked as outlier data to be processed. The significance of this step is that the dynamic window eliminates baseline fluctuations caused by seasonality or long-term evolution in oil and gas production, enabling more precise identification of local abrupt change points.

[0041] After confirming abnormal data, deletion is not performed directly. Instead, the type of deviation is identified through time-series trends. For intermittent deviations (such as isolated spikes caused by transient electrical signal interference from sensors), the characteristic is a sudden deviation followed by a rapid return to the normal range. For persistent deviations (such as a step-like increase in pressure due to sensor zero-point drift or localized scaling in pipelines), the characteristic is that after the deviation, the data remains stable or evolves with a monotonic trend within the new numerical range. This classification mechanism provides a logical premise for subsequent classification and implementation, avoiding the accidental deletion of data that represents a sudden change in actual operating conditions.

[0042] For occasional deviations, a local information compensation strategy is adopted. Normal feature data before and after the anomaly point are extracted as a reference, and a pre-defined smoothing reconstruction algorithm (such as cubic spline interpolation or weighted moving average) is used to generate a replacement value. This method removes interference noise while ensuring the continuity and smoothness of the data input to the facility loss prediction model over time, preventing spurious signals from causing the model to fail to converge.

[0043] For persistent deviations, the physical causes are determined based on their evolutionary trend characteristics (such as linear increase, exponential convergence, etc.) and a prior knowledge base. If the intercept drift is determined to be caused by sensor aging, a baseline deviation correction strategy is implemented to bring the data back to the logical range through linear shifting. If the change is determined to be a systematic change caused by adjustments in production conditions, its trend is preserved. This ensures that the cleaned dataset is not only mathematically sound but also physically consistent with the actual production conditions of the oil and gas field.

[0044] Finally, the processed dataset undergoes missing data completion (e.g., joint estimation using strongly correlated auxiliary parameters) and standardization (e.g., max-min standardization). The resulting processed original dataset not only eliminates the influence of dimensions but also, through the aforementioned classification and cleaning, retains the true physical characteristics reflecting facility losses, laying a high-quality data foundation for the accurate extraction of subsequent multidimensional loss predictions.

[0045] In some embodiments, the method may further include the following: The multiphase flow prediction submodule is a module based on a fluid dynamics mapping structure, the scaling prediction submodule is a module based on a thermodynamic equilibrium calculation structure, and the corrosion prediction submodule is a module based on an electrochemical reaction mapping structure. The multiphase flow prediction submodule includes at least a nonlinear feedforward layer that performs feature dimensionality reduction and flow pattern distribution feature mapping on the production impact parameters in the original dataset to determine the fluid dynamics parameters. The scaling prediction submodule includes at least a feature extraction layer for determining the saturation index and crystal nucleation trend based on the ion concentration characteristics in the production impact parameters. The corrosion prediction submodule includes at least a feedback coupling layer that uses the fluid dynamics parameters and the partial pressure of gas components in the production impact parameters to map the interface mass transfer intensity and extract corrosion rate evolution features.

[0046] In some embodiments, the multiphase flow prediction submodule is not a general black-box model; it internally constructs a nonlinear feedforward layer that maps to the fluid dynamics within the pipe. This layer specifically performs feature reduction on high-dimensional fluid parameters (such as production rate, gas-oil ratio, pipe diameter, and inclination angle) in the original dataset, mapping them to a low-dimensional space characterizing the flow pattern (such as calculating intermediate variables like liquid holdup and mixed-phase velocity). Through this nonlinear mapping, the module can identify the current flow pattern (such as slug flow, annular flow, or stratified flow) and determine the key fluid dynamic parameter—wall shear force. The significance of this structural design lies in its use of the nonlinear expressive power of the feedforward layer to fit the discontinuity problem of traditional empirical formulas when dealing with complex flow pattern transitions.

[0047] In some embodiments, the scaling prediction submodule employs a thermodynamic equilibrium calculation structure, whose internal feature extraction layer focuses on uncovering the intrinsic logic between ion activity and phase equilibrium. This layer calculates the saturation index under the target operating conditions by jointly processing the ion concentration, temperature, and pressure characteristics among the production-influencing parameters. Furthermore, this feature extraction layer can identify the critical point of solute transformation from a metastable state to a precipitated state, thereby quantifying the crystal nucleation trend. Compared to traditional static chemical equilibrium calculations, the structure of this embodiment can capture the scaling acceleration phenomenon caused by sudden pressure drops (such as through nozzles or pump outlets), achieving real-time characterization of the scaling growth dynamics.

[0048] In some embodiments, the core of the corrosion prediction submodule lies in its feedback coupling layer. This layer simulates the interfacial dynamics in an electrochemical reaction: it receives hydrodynamic parameters from the multiphase flow module, combines them with partial pressure data of gas components, and performs interfacial mass transfer intensity mapping. This mapping process takes into account the reduction effect of fluid erosion on the thickness of the interfacial diffusion layer. Furthermore, this feedback coupling layer also possesses evolutionary feature extraction capabilities, enabling feedback correction of the current corrosion current density based on previously predicted scale thickness. This feedback coupling structure ensures that corrosion prediction is no longer an isolated numerical calculation, but rather an evolutionary logic dynamically balanced with the environment.

[0049] In some embodiments, the method utilizes the preset facility loss prediction model to obtain corresponding prediction output differences by processing the first parameter set and the second parameter set respectively; based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. In specific implementations, the method may further include the following: S1: Using the facility loss prediction model, determine the first output response corresponding to the first parameter set and the second output response corresponding to the second parameter set, respectively; S2: Calculate the deviation between the first output response and the second output response to obtain the predicted output difference; S3: Traverse the subset spaces of different dimensions composed of the multiple production impact parameters, and determine the corresponding attribution weights based on the arrangement and distribution characteristics of each subset space in the full parameter space; S4: The differences in the corresponding predicted outputs are weighted and summed according to the attribution weights to obtain the marginal contribution increment of each target production impact parameter under different parameter combinations; wherein, the target production impact parameter includes one or more production impact parameters.

[0050] Specifically, the extraction of predicted output differences is achieved through dual-path calculation with controlled perturbations. First, a first set of parameters, including the parameters to be evaluated, is input into the facility loss prediction model to obtain a first output response reflecting the current actual operating conditions. Then, a second set of parameters, after benchmark replacement (e.g., mean replacement), is input into the same model to obtain a second output response. The Euclidean distance or component deviation between the two is calculated to obtain the predicted output difference of the parameter under the current specific combination. This step, through the idea of ​​the controlled variable method, initially isolates the direct perturbation of the parameter to multidimensional losses (shear force, scaling tendency, corrosion rate).

[0051] Because of the strong coupling relationships among oil and gas field production parameters (such as pressure, flow rate, and water cut), sensitivity analysis of a single parameter cannot reflect its true role under complex operating conditions. Therefore, this embodiment simulates the performance of a parameter in all possible combinations of "parameter partners" by traversing a subset space composed of multiple production-influencing parameters. For example, the predicted output of the same CO2 partial pressure parameter differs between the "high production + high water cut" subset space and the "low production + low water cut" subset space. Through this traversal of subset spaces in different dimensions, this invention can capture the nonlinear compensation and superposition effects between parameters.

[0052] After obtaining a massive amount of predicted output discrepancies, attribution weights are determined based on the distribution characteristics of each subset space within the full parameter space. Specifically, subset spaces with high combination frequencies and strong physical background representativeness are assigned higher attribution weights; simultaneously, following the fairness principle in game theory, corresponding combination weights are allocated according to the size (number of dimensions) of the subsets. This weighting method ensures that the final determined weight sequence conforms to both statistical laws and the physical logic of the oil and gas field.

[0053] Finally, a weighted summation process is performed on the differences in the predicted outputs obtained across each dimension, ultimately aggregating the marginal contribution increment of the target production impact parameter. This increment represents the net contribution of this parameter to the facility depreciation risk after considering all potential interactions. This transformation from "data discrepancies" to "attribution increments" enables the detection results to accurately pinpoint "who is the primary cause of the current risk increase," achieving a technological leap from correlation analysis to causal identification.

[0054] In some embodiments, the method for determining the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations may further include the following: S1: Perform normalization calculation on the marginal contribution increment corresponding to each of the target production impact parameters to obtain the weighted impact factor of each of the target production impact parameters for the multidimensional loss prediction value; S2: Prioritize each of the target production impact parameters based on the magnitude of the weighted impact factors to obtain an initial weight sequence characterizing the distribution of the contribution intensity of each parameter; S3: Extract the evolution trend features of each data item in the original dataset within a preset time window, and use the evolution trend features to perform time-directed correction on the initial weight sequence to generate the parameter dynamic weight sequence.

[0055] Specifically, the process first receives the marginal contribution increments of each target production impact parameter (such as temperature, pressure, and moisture content) calculated in the previous steps. Due to the significant differences in the physical dimensions and numerical ranges of different parameters, this embodiment uses normalization calculations (such as maximum-minimum standardization or Softmax mapping) to transform the absolute contribution values ​​into dimensionless weighted impact factors. This step ensures that physical parameters of different properties can be horizontally compared in terms of their influence within the same dimension.

[0056] After obtaining the weighting factors of each parameter, the parameters are sorted in descending order of their numerical values ​​to generate an initial weighting sequence. This sequence visually reflects the static intensity distribution of the contribution of each production parameter to facility losses (corrosion, scaling) at the current sampling time. For example, if the CO2 partial pressure has the largest weighting factor under the current operating conditions, it will be ranked first in the initial sequence and identified as the primary cause of the current risk.

[0057] To overcome the random disturbances caused by single-point sampling and capture the kinetic energy of the evolving operating conditions, this embodiment introduces a preset time window (such as the past 1 hour or 4 hours). The evolutionary trend characteristics of the raw data within this window are extracted, specifically including the rate of change (first derivative), acceleration (second derivative), and the standard deviation of the fluctuation. For example, although the current contribution of a certain parameter may not be the largest, it exhibits an exponential upward trend within the window; this "trend characteristic" implies the possibility of a surge in future risk.

[0058] The initial weight sequence is adjusted temporally using the aforementioned evolutionary trend characteristics. By using a preset adjustment factor, the weights of parameters exhibiting an "upward trend" or "high-frequency fluctuation" are increased, while the weights of parameters exhibiting a "decreasing trend" or "stable state" are correspondingly decreased. The resulting dynamic weight sequence is no longer an isolated static ranking, but a deep attribution indicator that integrates "current strength" and "trend of change." This adjustment mechanism effectively solves the lag problem of traditional static weight methods when operating conditions fluctuate drastically, making risk detection results more forward-looking.

[0059] In some embodiments, the method for determining the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and a preset dynamic threshold may further include the following: S1: Determine the risk sensitivity coefficient based on the dynamic weight sequence of the parameters and the material allowable limit data of the target oil and gas field facilities; S2: Correct the preset dynamic threshold according to the risk adjustment coefficient to determine the target dynamic threshold; S3: Compare and map the multidimensional loss prediction value with the target dynamic threshold to determine the target risk detection result.

[0060] Specifically, firstly, the dynamic weight sequence of the parameters is retrieved to identify the leading production parameters (e.g., water cut or CO2 partial pressure) that currently contribute the most to losses. Simultaneously, the material allowable limits of the target oil and gas field facility are extracted from a pre-defined asset database. This data includes the facility's current remaining wall thickness, material corrosion margin, and yield strength reserve. Then, the weight scores of the leading parameters are mapped and compared to the facility's remaining lifespan. If the weight of the leading parameters shows an upward trend, and the remaining wall thickness is already below a pre-defined safe percentage of the original wall thickness (e.g., below 40%), a higher risk sensitivity coefficient is determined using a pre-defined sensitivity mapping table. This process couples and quantifies the risks of "external parameter fluctuations" and "facility internal aging."

[0061] After obtaining the risk sensitivity coefficient, a calibration procedure for the preset dynamic threshold is initiated. Specifically, the preset dynamic threshold is defined as a multi-dimensional benchmark that includes an upper limit for corrosion rate, a critical value for scale accumulation, and an upper limit for hydrodynamic scour.

[0062] Using the aforementioned risk sensitivity coefficient as an adjustment factor, negative scaling or gain adjustment is applied to the above benchmarks. For example, when the sensitivity coefficient is high, the alarm threshold for corrosion rate is automatically lowered (e.g., from 0.1 mm / a to 0.07 mm / a), and the tolerance for scale buildup is tightened. In this way, the originally static warning scale is transformed into a target dynamic threshold bound to the real-time operating status of the facility, ensuring that risk response can be triggered earlier during the facility's sensitive period.

[0063] After generating the target dynamic threshold, the obtained multidimensional loss prediction values ​​(i.e., wall shear force, predicted scale amount, and predicted corrosion rate) are compared and mapped with the target dynamic threshold to determine the target risk detection result.

[0064] The target risk detection results include risk level identification: based on the percentage deviation of the predicted value from the target threshold, it is divided into three levels: low risk, medium risk, and high risk; dominant risk source identification: extracting the name of the parameter with the highest weight in the parameter dynamic weight sequence, and directly indicating the dominant cause of the current risk in the result (e.g., scour risk caused by high-weight flow velocity parameter).

[0065] In some embodiments, firstly, the contribution of the dominant parameter reflected in the dynamic weight sequence of the parameters is logically correlated with the residual strength of the body represented in the material allowable limit data of the target oil and gas field facility. The risk sensitivity coefficient is quantified by determining the potential damage sensitivity of the current dominant parameter fluctuation intensity to a specific material state. Subsequently, the coefficient is used as an adjustment variable to perform sensitivity offset processing on a preset dynamic threshold, that is, when the facility's tolerance is low or the dominant parameter fluctuates drastically, the warning judgment envelope is tightened accordingly, thereby determining a target dynamic threshold that is compatible with the real-time service health status of the facility. Finally, the multidimensional loss prediction value composed of wall shear force, predicted scaling amount, and predicted corrosion rate is compared with the multidimensional components of the target dynamic threshold through multidimensional coordinate mapping. Based on the hierarchical interval of the predicted data points falling within the threshold judgment space and the deviation vector relative to the threshold boundary, the target risk detection result, which includes the risk warning level and the dominant risk cause, is finally determined.

[0066] In some embodiments, the method for determining the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and a preset dynamic threshold may further include the following: First, the dominant parameter with the highest current weight (e.g., CO2 partial pressure, weight 0.65) is extracted from the dynamic weight sequence of parameters. Simultaneously, the material tolerance limits of the target oil and gas field facility are obtained. It is verified that the pipeline has been in service for 8 years, and the remaining wall thickness is only 70% of the initial value. Based on a pre-set logical mapping table, it is determined that the facility has a small tolerance margin in the current high-concentration acidic environment; therefore, a high risk sensitivity coefficient is determined.

[0067] Real-time correction of the target dynamic threshold: After obtaining a high risk sensitivity coefficient, the threshold correction process is initiated. Among the preset dynamic thresholds, the alarm baseline for corrosion rate is 0.2 mm / a. Using the risk sensitivity coefficient, the sensitivity of this baseline is lowered, correcting the alarm threshold from 0.2 mm / a to 0.07 mm / a, thus obtaining the target dynamic threshold. This correction ensures that even with increased facility sensitivity, minute losses can be detected at a more sensitive scale, preventing sudden failures.

[0068] The multidimensional predicted value comparison and result determination involves inputting the currently calculated multidimensional loss prediction value (e.g., predicted corrosion rate of 0.08 mm / a and predicted scale thickness of 0.2 mm) into the judgment matrix.

[0069] According to the comparison mapping logic, although 0.08 mm / a did not reach the original 0.1 mm / a standard, it immediately triggered an early warning because it exceeded the corrected target dynamic threshold (0.07 mm / a).

[0070] The final target risk detection results not only include a "high risk" level label, but also clearly indicate "the risk is dominated by CO2 partial pressure" in the results by tracing back the weight sequence. It also automatically matches and outputs a protection instruction of "increasing the amount of corrosion inhibitor" from the protection knowledge base, thus completing the whole process from risk discovery to cause identification and decision support.

[0071] It should be noted that the risk sensitivity coefficient is used to quantitatively correlate the "internal characteristics" of the facility with the "external threats": the material allowable limit data of the target oil and gas field facility characterizes the facility's bottom-line ability to resist damage, while the dynamic weight sequence of parameters reflects the dominant intensity of each damaging factor in the current environment. The two are used to determine the degree of failure of the facility under the current operating conditions through nonlinear mapping logic.

[0072] Subsequently, the coefficient is used to correct the "sensitivity offset" of the static preset dynamic threshold: when the facility is old or the dominant risk factor is active, the threshold is tightened in a more stringent direction by adjusting the coefficient, thereby generating a target dynamic threshold that is adapted to the real-time service status.

[0073] Finally, through multidimensional coordinate mapping, the predicted multidimensional losses, encompassing mechanical, physical, and chemical dimensions, are compared with the corrected threshold envelope. This mechanism can not only determine whether a single indicator exceeds the limit, but also identify hidden risks under the superposition of multidimensional micro-losses, thereby outputting target risk detection results with attribution analysis capabilities.

[0074] In some embodiments, the method may further include the following: S1: Based on the target risk detection results, determine the corresponding set of candidate protection schemes from the preset protection knowledge base; S2: Determine the economic adaptability characteristics of each scheme in the candidate protection scheme set based on the on-site cost accounting factors; S3: Determine the protection scheme corresponding to the target oil and gas field facility based on the economic adaptability characteristics of each scheme in the candidate protection scheme set.

[0075] Specifically, the target risk detection results determined in the previous steps are first extracted. These results include the specific risk level and the dominant risk source (e.g., high uniform corrosion risk caused by high-weight CO2 partial pressure). Using these risk characteristics as an index, a correlation search is performed in a pre-defined protection knowledge base. This knowledge base pre-stores various protection methods and their corresponding applicable boundaries, such as matching different types of corrosion inhibitors to different concentrations of corrosive media, and matching physical cleaning or chemical descaling to different scale types. The search process outputs a candidate protection scheme set, which contains multiple feasible solutions that can suppress the current risk in terms of technical mechanism, such as increasing the corrosion inhibitor concentration, adjusting downstream back pressure to reduce flow rate, or initiating a predetermined cycle of cleaning operations.

[0076] After obtaining the candidate protection scheme set, on-site cost accounting factors are introduced to evaluate the economic efficiency of each scheme. The on-site cost accounting factors cover the unit price of reagents, energy consumption of the filling equipment, cost of pig consumables, downtime losses (opportunity costs) during the operation, and manpower input for scheme implementation.

[0077] For each candidate solution, its economic suitability characteristics are calculated based on its expected protection period and resource consumption. For example, for the corrosion inhibitor injection solution, the characteristic is the product of the daily consumption of the agent and its unit price, plus equipment maintenance costs; while for the pigging solution, the characteristic focuses on the ratio of the cost per operation to the interval between two operations. This evaluation method transforms abstract protective measures into quantifiable economic expenditure indicators, providing a cost benchmark for subsequent optimization.

[0078] Finally, based on the economic adaptability characteristics of each candidate scheme and the severity of risk in the target risk detection results, the final protection scheme is determined. Specifically, when the risk level is low, the conventional scheme with the smaller economic adaptability characteristics (i.e., the lowest total cost) is prioritized; while when the detection results show a high-risk state, the decision-making logic increases the weight of "protection timeliness." Even if some schemes have higher economic adaptability characteristics (such as high-performance corrosion inhibitors or emergency shutdown and pipeline cleaning), if they can quickly curb asset loss, they are identified as the target protection scheme. The determined protection scheme is output in the form of specific execution instructions, clearly indicating the adjustment range of injection volume, operation frequency, or operating parameters, thereby achieving optimal control of production and operating costs while ensuring the safety of oil and gas facilities.

[0079] As can be seen from the above, the embodiment of this specification provides a method for detecting and determining the corrosion and scaling risk of oil and gas field facilities, which involves obtaining the original dataset of the target oil and gas field facilities; wherein, the original dataset includes static data, dynamic data, and monitoring data; based on the original dataset, production impact parameters related to corrosion and scaling are extracted; using a preset facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule; a target production impact parameter is selected from multiple production impact parameters, and... A first parameter set containing the target production impact parameters and a second parameter set excluding the target production impact parameters are constructed respectively. Using the preset facility loss prediction model, the corresponding prediction output differences are obtained by processing the first parameter set and the second parameter set respectively. Based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, the corresponding dynamic weight sequence of parameters is determined. Based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facilities, and the preset dynamic threshold, the target risk detection result is determined.

[0080] See Figure 2 As shown in the embodiments of this specification, a specific electronic device is also provided, wherein the electronic device includes a network communication port 201, a processor 202 and a memory 203, and the above structures are connected by internal cables so that the various structures can perform specific data interaction.

[0081] Specifically, the network communication port 201 can be used to acquire the raw dataset of the target oil and gas field facilities; wherein the raw dataset includes static data and dynamic data.

[0082] The processor 202 can be specifically used to extract production impact parameters related to corrosion and scaling based on the original dataset; obtain multidimensional loss prediction values ​​based on the original dataset using a preset facility loss prediction model; wherein the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule; select target production impact parameters from multiple production impact parameters, and construct a first parameter set containing the target production impact parameters and a second parameter set not containing the target production impact parameters; obtain corresponding prediction output differences by processing the first parameter set and the second parameter set respectively using the preset facility loss prediction model; determine the marginal contribution increment of each target production impact parameter under different parameter combinations based on the prediction output differences; determine the corresponding parameter dynamic weight sequence based on the marginal contribution increment of each target production impact parameter under different parameter combinations; and determine the target risk detection result based on the multidimensional loss prediction values, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facilities, and a preset dynamic threshold.

[0083] The memory 203 can be used to store the corresponding instruction program.

[0084] Based on the above method, the relevant structural performance of electronic equipment can be effectively utilized to improve the data processing speed of electronic equipment and efficiently realize a method for detecting and determining the corrosion and scaling risk of oil and gas field facilities.

[0085] In this embodiment, the network communication port 201 can be a virtual port bound to different communication protocols, thereby enabling the sending or receiving of different data. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for email data communication. Furthermore, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM or CDMA; it can also be a Wi-Fi chip; or it can be a Bluetooth chip.

[0086] In this embodiment, the processor 202 can be implemented in any suitable manner. For example, the processor can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc. This specification is not limiting.

[0087] In this embodiment, the memory 203 may include a hierarchy. In a digital system, anything that can store binary data can be a memory. In an integrated circuit, a circuit with storage function but no physical form is also called a memory, such as RAM, FIFO, etc. In a system, a storage device with a physical form is also called a memory, such as a memory stick, TF card, etc.

[0088] This specification also provides a computer-readable storage medium based on the above-described method for determining corrosion and scaling risks in oil and gas field facilities, which acquires the original dataset of the target oil and gas field facility; wherein the original dataset includes static data, dynamic data, and monitoring data; based on the original dataset, production impact parameters related to corrosion and scaling are extracted; using a preset facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule; and a target production impact parameter is selected from multiple production impact parameters. The system calculates the impact parameters and constructs a first parameter set containing the target production impact parameters and a second parameter set excluding the target production impact parameters. Using the preset facility loss prediction model, the system processes the first parameter set and the second parameter set to obtain corresponding prediction output differences. Based on the prediction output differences, the system determines the marginal contribution increment of each target production impact parameter under different parameter combinations. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, the system determines the corresponding parameter dynamic weight sequence. Based on the multidimensional loss prediction value, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facility, and the preset dynamic threshold, the system determines the target risk detection result.

[0089] In this embodiment, the storage medium includes, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions. The network communication unit can be an interface configured according to standards specified in the communication protocol for network connection communication.

[0090] In this embodiment, the specific functions and effects implemented by the program instructions stored in the computer-readable storage medium can be explained in comparison with other embodiments, and will not be repeated here.

[0091] See Figure 3At the software level, embodiments of this specification also provide a device for detecting and determining the corrosion and scaling risk of oil and gas field facilities. This device may specifically include the following structural modules: The data acquisition module 301 is used to acquire the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; The influencing parameter determination module 302 is used to extract production influencing parameters related to corrosion and scaling based on the original dataset; The model prediction module 303 is used to obtain multidimensional loss prediction values ​​based on the original dataset using a preset facility loss prediction model; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. The parameter set determination module 304 is used to select a target production impact parameter from a plurality of production impact parameters, and to construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter. The contribution increment determination module 305 is used to obtain the corresponding prediction output difference by processing the first parameter set and the second parameter set respectively using the preset facility loss prediction model; and to determine the marginal contribution increment of each target production impact parameter under different parameter combinations based on the prediction output difference. The weight sequence determination module 306 is used to determine the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations. The detection result determination module 307 is used to determine the target risk detection result based on the multidimensional loss prediction value, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facility, and the preset dynamic threshold.

[0092] In some embodiments, after the data acquisition module 301 described above, in specific implementation, the statistical parameters of each data point in the original dataset within a preset sliding window are calculated, and data that deviates from the statistical parameters by more than a dispersion threshold are marked as abnormal data to be processed; the deviation type of the abnormal data to be processed is confirmed by identifying the changing trend of the abnormal data to be processed; wherein, the deviation type includes a continuous deviation type and an occasional deviation type; when the deviation type is the occasional deviation type, normal feature data in the neighborhood of the abnormal data to be processed is extracted, a substitute value is generated using a preset smoothing reconstruction algorithm, and a value replacement strategy is executed on the abnormal data to be processed to obtain a cleaned dataset; when the deviation type is the continuous deviation type, the physical cause is determined based on the evolution trend characteristics of the abnormal data to be processed; and according to the physical cause, the corresponding benchmark deviation correction strategy is executed on the abnormal data to be processed to obtain a cleaned dataset; the cleaned dataset is processed for missing data completion, and the data is standardized according to a preset unified standard rule to obtain the processed original dataset.

[0093] In some embodiments, the above-described apparatus further includes: the multiphase flow prediction submodule is a module based on a fluid dynamics mapping structure, the scaling prediction submodule is a module based on a thermodynamic equilibrium calculation structure, and the corrosion prediction submodule is a module based on an electrochemical reaction mapping structure; wherein, the multiphase flow prediction submodule includes at least a nonlinear feedforward layer for performing feature dimensionality reduction and flow pattern distribution feature mapping on the production impact parameters in the original dataset, for determining the fluid dynamics parameters; the scaling prediction submodule includes at least a feature extraction layer for determining the saturation index and nucleation trend based on the ion concentration characteristics in the production impact parameters; and the corrosion prediction submodule includes at least a feedback coupling layer for mapping interfacial mass transfer intensity and extracting corrosion rate evolution features using the fluid dynamics parameters and the gas component partial pressure in the production impact parameters.

[0094] In some embodiments, the contribution increment determination module 305, in its specific implementation, utilizes the facility loss prediction model to determine the first output response corresponding to the first parameter set and the second output response corresponding to the second parameter set, respectively; calculates the deviation between the first output response and the second output response to obtain the predicted output difference; traverses the subset spaces of different dimensions composed of the plurality of production impact parameters, and determines the corresponding attribution weights based on the arrangement and distribution characteristics of each subset space in the full parameter space; performs weighted summation processing on the corresponding predicted output differences according to the attribution weights to obtain the marginal contribution increment of each target production impact parameter under different parameter combinations; wherein, the target production impact parameters include one or more production impact parameters. Normalization calculation is performed on the marginal contribution increment corresponding to each of the target production impact parameters to obtain the weighted impact factor of each target production impact parameter for the multidimensional loss prediction value; the target production impact parameters are prioritized based on the magnitude of the weighted impact factor to obtain an initial weight sequence characterizing the contribution intensity distribution of each parameter; the evolution trend features of each data item in the original dataset within a preset time window are extracted, and the initial weight sequence is temporally corrected using the evolution trend features to generate the dynamic weight sequence of the parameters.

[0095] In some embodiments, the detection result determination module 307, in specific implementation, determines the risk sensitivity coefficient based on the parameter dynamic weight sequence and the material allowable limit data of the target oil and gas field facility; corrects the preset dynamic threshold based on the risk adjustment coefficient to determine the target dynamic threshold; and compares and maps the multidimensional loss prediction value with the target dynamic threshold to determine the target risk detection result.

[0096] In some embodiments, the above-described device further includes: determining a set of candidate protection schemes from a preset protection knowledge base based on the target risk detection results; determining the economic adaptability characteristics of each scheme in the candidate protection scheme set based on the on-site cost accounting factor; and determining the protection scheme corresponding to the target oil and gas field facility based on the economic adaptability characteristics of each scheme in the candidate protection scheme set.

[0097] It should be noted that the units, devices, or modules described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above devices are described by dividing them into various modules according to their functions. Of course, in implementing this specification, the functions of each module can be implemented in the same software and / or hardware, or modules that implement the same function can be implemented by a combination of sub-modules or sub-units, etc. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. For example, units or components can be combined or integrated into another system, or some features can be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection between the devices or units shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0098] As can be seen from the above, based on the corrosion and scaling risk detection and determination device for oil and gas field facilities provided in the embodiments of this specification, the original dataset of the target oil and gas field facility is obtained; wherein, the original dataset includes static data, dynamic data, and monitoring data; based on the original dataset, production impact parameters related to corrosion and scaling are extracted; using a preset facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule; a target production impact parameter is selected from multiple production impact parameters, and... A first parameter set containing the target production impact parameters and a second parameter set excluding the target production impact parameters are constructed respectively. Using the preset facility loss prediction model, the corresponding prediction output differences are obtained by processing the first parameter set and the second parameter set respectively. Based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, the corresponding dynamic weight sequence of parameters is determined. Based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facilities, and the preset dynamic threshold, the target risk detection result is determined.

[0099] In a specific scenario example, the corrosion and scaling risk detection method and apparatus for oil and gas field facilities provided in this specification can be applied to solve the technical problem of low accuracy and poor reliability of early warning in predicting corrosion and scaling risks of oil and gas field facilities due to the lack of multi-physical mechanism hierarchical coupling and dynamic marginal contribution identification of parameters. The specific implementation process may include the following:

[0100] In some embodiments, the oil and gas field corrosion and scaling lifecycle management system is deployed in a production center of an operating area. First, the basic dataset of the target facility is acquired through a data acquisition layer: Static data: Import surface production pipeline distribution map, well structure of water injection wells and oil production wells, pipeline design specifications and anti-corrosion layer information in batches using templates; Dynamic data: Real-time access to oil production / water injection production reports and surface pipeline operation data; Monitoring and testing data: Regularly record data on coating inspection, electrochemical probe monitoring, wall thickness testing, and reagent dispensing. This layer supports real-time updates through data interconnection to ensure the completeness of risk influencing factors.

[0101] The data purification and preprocessing implementation layer performs refined cleaning on the collected data. Specifically, this includes: anomaly identification: based on the Raida criterion (calculating the mean and standard deviation of each parameter, automatically marking data that deviates from the error range). The system compares multiple parameters (such as comparing water production and operating pressure when gas production is abnormal) to determine whether the anomaly is a continuous deviation or an occasional deviation, thereby deciding whether to include it in the cleaning process.

[0102] Data completion and standardization: For missing well depth temperature / pressure data, linear interpolation is used to reconstruct the data using known wellhead / bottomhole temperature and pressure and depth gradients; finally, normalization is performed according to preset standard rules to eliminate dimensional differences.

[0103] The calculations are performed using a pre-defined facility loss prediction model. Modular calculation: Hierarchically coupled sub-modules for multiphase flow, scaling, and corrosion prediction are constructed. For example, the medium velocity output from the multiphase flow model is directly used as the input layer for the corrosion prediction model. Each sub-module supports independent parameter settings to ensure computational flexibility. Feature selection and weight ranking: The correlation between parameters such as CO2 partial pressure, H2S partial pressure, and salinity and the prediction results is calculated using the SHAP method, generating a dynamically output weight ranking sequence. Risk assessment: Based on the calculated corrosion rate and scaling trend, combined with the NACE 0775 standard (e.g., a warning is triggered when the corrosion rate > 0.2 mm) and the allowable limits of the facility material, a three-level risk warning is set.

[0104] The application interaction and visualization implementation layer presents the data generated by the logic layer in a visual way: Risk warning: The risk status of different areas is indicated by three colors: red (high), yellow (medium), and green (low).

[0105] Trend Analysis: Generates time series curves for parameters such as pressure, oil production, and corrosion rate, allowing users to backtrack and compare historical data.

[0106] Spatial modeling: Construct a model of the well structure and pipeline elevation distribution to show the specific impact of tubing stress and elevation changes on corrosion, and support users to manually adjust parameters for simulation analysis.

[0107] The protection execution layer is responsible for the selection and feedback of the proposed solution, including protection execution and closed-loop optimization. Solution optimization: The solution recommendation module matches candidate protective measures (such as chemical dosing and pipeline cleaning) according to the risk level, and calculates them in combination with on-site cost factors to automatically recommend the most economically suitable protective solution.

[0108] Feedback and Iteration: The effect feedback module collects monitoring data after the implementation of the solution and compares it with the prediction results. If the prediction deviation is too large, a large deviation warning is triggered, requiring manual re-checking; if there are multiple deviations, the system will initiate the model correction procedure.

[0109] Data traceability: The historical data traceability module stores more than 5 years of original data and solution execution records, providing full lifecycle data support for subsequent continuous model optimization.

[0110] Based on the above embodiments, offline independent operation is achieved: Data processing, prediction, early warning, and solution generation are completed entirely on a single machine without relying on a network or external server, adapting to weak / no network environments in oilfields and solving the problem of "network outage paralysis" in distributed systems. Enhanced data integration capabilities: Through multi-source data fusion technology, the fragmentation of static, dynamic, and monitoring data is broken down, achieving a data fusion accuracy of ≥95%, laying the foundation for accurate prediction. High prediction accuracy: Improved traditional prediction models employ a machine learning hybrid model with a prediction error ≤15%, significantly improving accuracy compared to traditional models, and integrating multi-dimensional data such as structure, production, monitoring, and corrosion protection. Intelligent protection solutions: Combining multi-objective optimization algorithms, while ensuring protection effectiveness, economic cost comparison and optimal recommendation are achieved, balancing security and economy. Data security and controllability: All data is stored locally on a single machine, using encrypted backups to avoid privacy leaks caused by data uploads to the cloud, meeting oilfield data security management requirements. Easy to operate and learn: The visual interface is simple and intuitive (adopting the operating logic commonly used in oilfields), supporting convenient data interaction methods such as importing Excel templates and direct sensor connection. On-site maintenance personnel can operate independently after 1 hour of training. Good scalability: The modular design supports adding reservoir blocks and updating prediction models, adapting to the development needs of oilfields of different sizes, and has a wide range of applications.

[0111] In some embodiments, the crude oil from a certain oil field has a density of 0.85~0.79 g / cm3, a bubble point pressure of 920~3125 psi, a viscosity of 0.55~5.7 cp, and a gas-oil ratio of 228.7~912.5 scf / b; the gas composition is CO2 of 0.03~5 mol%, H2S of 0~8 mol%; the formation water is of CaCl2 type with a salinity of 152~192 g / L. Through software data uploading and retrieval, and using the model, different risk outcomes are predicted.

[0112] S1: Multiphase flow prediction submodule.

[0113] Data parameters are shown in Table 1.

[0114] Table 1

[0115] Specifically, the output parameters for multiphase flow corrosion prediction are shown in Table 2.

[0116] Table 2

[0117] S2: Corrosion prediction submodule.

[0118] The model calls upon system data and performs calculations, outputting a corrosion rate of 0.02337 mm / a, which is compared with the on-site detection result of 0.0272 mm / a. The risk level of multiple sets of on-site data is compared, and the prediction result is consistent with the detection level of the corrosion plate. The accuracy of the corrosion prediction result is 85%, as detailed in Table 3.

[0119] Table 3

[0120] S3: Scaling prediction submodule.

[0121] Scaling was predicted by on-site water quality. The prediction results were compared with the on-site product detection types, and the scaling types were consistent. For specific input parameters, see Table 4. By retrieving the real-time water quality monitoring data shown in Table 4 (including but not limited to the concentrations of scale-forming ions such as calcium, magnesium, barium, and strontium, the concentration of scale-forming anions, the degree of mineralization, and the real-time temperature and operating pressure at the sampling points), this data is used as the basic input parameters and imported into the scaling prediction submodule. The module's internal thermodynamic balance calculation logic simulates the ion activity distribution and phase equilibrium state of the fluid under the current operating conditions, thereby quantitatively producing the saturation index of each mineral component and the predicted scaling trend value. To verify the reliability of the prediction results, solid sediment samples from the target facility are collected simultaneously, and on-site product composition detection is performed (e.g., confirming its physical composition through X-ray diffraction analysis). By comparing and mapping the scaling type output by the system with the actual product type obtained from on-site detection, the results show that the two are highly consistent in terms of the main mineral components. This not only confirms the accurate capture capability of the model described in this invention in handling real water quality environments, but also provides solid technical support and data basis for the subsequent protection execution layer to determine the chemical scale inhibitor model and injection scheme through this "prediction-detection" closed-loop verification method.

[0122] Table 4

[0123] Specifically, the output parameters of the scaling prediction model are shown in Table 5.

[0124] Table 5

[0125] In some embodiments, see Figure 4 As shown, a method for managing corrosion and scaling in oil and gas fields throughout their entire lifecycle is presented, starting with the data management phase. In this phase, the raw dataset of the target facility is acquired through a data acquisition module, including static data such as wellbore structure, dynamic data such as production reports, and monitoring data such as anchor inspection data. Subsequently, the data preprocessing module performs anomaly identification and cleaning on the acquired data based on the Laida criterion, and performs linear interpolation to complete missing values, ensuring that the data entering the data storage module has high-quality attributes. Furthermore, a data interconnection module enables real-time data sharing and verification between various business units, providing complete and standardized data support for subsequent risk prediction.

[0126] After data management is completed, the process enters the core risk prediction stage. This integrates three major modules: multiphase flow prediction, corrosion prediction, and scaling prediction. In practice, the multiphase flow prediction module calculates the medium velocity and wall shear force based on the input real-time flow rate and pipe diameter parameters, and transmits the results as coupling parameters to the corrosion prediction module. The scaling prediction module simultaneously determines the scaling trend based on water ion concentration and temperature and pressure conditions. During this process, the business logic layer uses the SHAP method to dynamically weight and prioritize various production-influencing parameters, ensuring that the model can adjust the calculation priority in real time according to changes in operating conditions, outputting high-precision multidimensional loss prediction values.

[0127] Based on the quantitative output of risk prediction, the process enters the risk warning stage. The predicted corrosion rate, scaling amount, and other indicators are compared with preset dynamic thresholds. In this embodiment, warning limits are set with reference to the NACE 0775 standard, and sensitivity correction is performed in conjunction with the allowable limits of the facility materials. When the predicted value reaches the warning envelope, the application interaction layer uses a visualization module to intuitively display the risk level in three colors: red (high), yellow (medium), and green (low), and generates trend analysis charts to help operators quickly locate high-risk points and the main risk causes.

[0128] Finally, based on the early warning information, the process enters the protective measure execution phase. The solution recommendation module filters matching candidate solutions from the protection knowledge base and performs an economic evaluation based on on-site cost accounting factors, selecting the most suitable protection solution (such as corrosion inhibitor injection or pipeline cleaning) and pushing it to the execution terminal. Figure 4 As indicated by the feedback arrow, the monitoring data after the implementation of the plan will be fed back to the data management stage. If there is a significant deviation between the actual protection effect and the predicted results, the model correction procedure will be initiated. Through this closed-loop iteration throughout the entire life cycle, management accuracy will be continuously optimized to ensure the long-term inherent safety of oil and gas field facilities.

[0129] While this specification provides the steps of operation for the methods described in the embodiments or flowcharts, more or fewer steps may be included based on conventional or non-inventive means. The order of steps listed in the embodiments is merely one possible order of execution among many steps and does not represent the only possible order. In actual device or client product execution, the methods shown in the embodiments or drawings may be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment, or even a distributed data processing environment). The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, product, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, product, or apparatus. Without further limitations, the presence of other identical or equivalent elements in a process, method, product, or apparatus that includes said elements is not excluded. The terms "first," "second," etc., are used to denote names and do not indicate any particular order.

[0130] Those skilled in the art will also know that, besides implementing the controller using purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller function as logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices within it used to implement various functions can also be considered structures within that hardware component. Alternatively, the devices used to implement various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0131] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this specification can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of this specification can essentially be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of this specification.

[0132] Although this specification has been described by way of examples, those skilled in the art will recognize that many variations and modifications are possible without departing from the spirit of this specification, and it is intended that the appended claims cover such variations and modifications without departing from the spirit of this specification.

Claims

1. A method for detecting and determining the risk of corrosion and scaling in oil and gas field facilities, characterized in that, include: Obtain the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; Based on the original dataset, production impact parameters related to corrosion and scaling were extracted; Using a pre-defined facility loss prediction model, multidimensional loss prediction values ​​are obtained based on the original dataset; wherein, the pre-defined facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. Select a target production impact parameter from the plurality of production impact parameters, and construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter; Using the preset facility loss prediction model, the corresponding prediction output differences are obtained by processing the first parameter set and the second parameter set respectively; based on the prediction output differences, the marginal contribution increment of each target production impact parameter under different parameter combinations is determined. Based on the marginal contribution increment of each target production impact parameter under different parameter combinations, determine the corresponding dynamic weight sequence of parameters; The target risk detection result is determined based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facilities, and the preset dynamic threshold.

2. The method according to claim 1, characterized in that, Following the acquisition of the raw dataset of the target oil and gas field facilities, the following is also included: Calculate the statistical parameters of each data point in the original dataset within a preset sliding window, and mark data that deviate from the statistical parameters by more than a dispersion threshold as abnormal data to be processed; By identifying the changing trends of the abnormal data to be processed, the deviation type of the abnormal data to be processed is confirmed; wherein, the deviation type includes continuous deviation type and occasional deviation type; When the deviation type is the occasional deviation type, normal feature data in the domain of the abnormal data to be processed is extracted, a preset smooth reconstruction algorithm is used to generate replacement values, and a value replacement strategy is executed on the abnormal data to be processed to obtain the cleaned dataset. When the deviation type is the persistent deviation type, the physical cause is determined based on the evolution trend characteristics of the abnormal data to be processed; and according to the physical cause, the corresponding benchmark deviation correction strategy is executed on the abnormal data to be processed to obtain the cleaned dataset. The cleaned dataset is then filled with missing data and standardized according to a pre-defined set of rules to obtain the processed original dataset.

3. The method according to claim 2, characterized in that, The multiphase flow prediction submodule is a module based on a fluid dynamics mapping structure, the scaling prediction submodule is a module based on a thermodynamic equilibrium calculation structure, and the corrosion prediction submodule is a module based on an electrochemical reaction mapping structure. The multiphase flow prediction submodule includes at least a nonlinear feedforward layer that performs feature dimensionality reduction and flow pattern distribution feature mapping on the production impact parameters in the original dataset to determine the fluid dynamics parameters. The scaling prediction submodule includes at least a feature extraction layer for determining the saturation index and crystal nucleation trend based on the ion concentration characteristics in the production impact parameters. The corrosion prediction submodule includes at least a feedback coupling layer that uses the fluid dynamics parameters and the partial pressure of gas components in the production impact parameters to map the interface mass transfer intensity and extract corrosion rate evolution features.

4. The method according to claim 3, characterized in that, The method utilizes the preset facility loss prediction model to obtain the corresponding prediction output differences by processing the first parameter set and the second parameter set respectively; Based on the predicted output differences, the marginal contribution increments of each target production impact parameter under different parameter combinations are determined, including: Using the facility loss prediction model, the first output response corresponding to the first parameter set and the second output response corresponding to the second parameter set are determined respectively; Calculate the deviation between the first output response and the second output response to obtain the predicted output difference; Traverse the subset spaces of different dimensions composed of the multiple production impact parameters, and determine the corresponding attribution weights based on the arrangement and distribution characteristics of each subset space in the full parameter space; The differences in the corresponding predicted outputs are weighted and summed according to the attribution weights to obtain the marginal contribution increment of each target production impact parameter under different parameter combinations; wherein, the target production impact parameter includes one or more production impact parameters.

5. The method according to claim 4, characterized in that, The step of determining the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations includes: Normalization calculation is performed on the marginal contribution increment corresponding to each of the target production impact parameters to obtain the weighted impact factor of each of the target production impact parameters for the multidimensional loss prediction value. Based on the magnitude of the weighted influence factors, the target production influence parameters are prioritized to obtain an initial weight sequence characterizing the distribution of the contribution intensity of each parameter; The evolution trend features of each data item in the original dataset within a preset time window are extracted, and the initial weight sequence is corrected in a time direction using the evolution trend features to generate the parameter dynamic weight sequence.

6. The method according to claim 5, characterized in that, The step of determining the target risk detection result based on the multidimensional loss prediction value, the dynamic weight sequence of parameters, the material allowable limit data of the target oil and gas field facility, and the preset dynamic threshold includes: The risk sensitivity coefficient is determined based on the dynamic weight sequence of the parameters and the material allowable limit data of the target oil and gas field facilities. The preset dynamic threshold is corrected based on the risk adjustment coefficient to determine the target dynamic threshold; The multidimensional loss prediction value is compared and mapped with the target dynamic threshold to determine the target risk detection result.

7. The method according to claim 1, characterized in that, The method further includes: Based on the target risk detection results, a set of corresponding candidate protection schemes is determined from a preset protection knowledge base; Based on the on-site cost accounting factors, determine the economic adaptability characteristics of each scheme in the candidate protection scheme set; Based on the economic adaptability characteristics of each scheme in the candidate protection scheme set, the protection scheme corresponding to the target oil and gas field facility is determined.

8. A device for detecting and determining the corrosion and scaling risk of oil and gas field facilities, characterized in that, include: The data acquisition module is used to acquire the raw dataset of the target oil and gas field facilities; wherein, the raw dataset includes static data, dynamic data, and monitoring data; The influencing parameter determination module is used to extract production influencing parameters related to corrosion and scaling based on the original dataset. The model prediction module is used to obtain multidimensional loss prediction values ​​based on the original dataset using a preset facility loss prediction model; wherein, the preset facility loss prediction model includes a hierarchically coupled multiphase flow prediction submodule, a scaling prediction submodule, and a corrosion prediction submodule; the multiphase flow prediction submodule is connected to the corrosion prediction submodule. The parameter set determination module is used to select a target production impact parameter from a plurality of production impact parameters, and to construct a first parameter set containing the target production impact parameter and a second parameter set not containing the target production impact parameter. The contribution increment determination module is used to obtain the corresponding prediction output difference by processing the first parameter set and the second parameter set respectively using the preset facility loss prediction model; and to determine the marginal contribution increment of each target production impact parameter under different parameter combinations based on the prediction output difference. The weight sequence determination module is used to determine the corresponding dynamic weight sequence of parameters based on the marginal contribution increment of each target production impact parameter under different parameter combinations. The detection result determination module is used to determine the target risk detection result based on the multidimensional loss prediction value, the parameter dynamic weight sequence, the material allowable limit data of the target oil and gas field facility, and the preset dynamic threshold.

9. An electronic device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed by a processor, implement the steps of the method according to any one of claims 1 to 7.

Images