A method and system for oil cost analysis based on data analysis

Abstract

The application belongs to the technical field of data analysis, and discloses a petroleum cost analysis method and system based on data analysis. The method comprises the following steps: collecting real-time full life cycle data of a petroleum project, and extracting real-time multi-dimensional cost data in the real-time full life cycle data; using a petroleum cost analysis model to perform petroleum cost analysis according to the real-time multi-dimensional cost data, and obtaining real-time petroleum cost analysis results; using a project optimization model to generate a project optimization strategy according to the real-time petroleum cost analysis results, and obtaining real-time project optimization strategies. The application solves the problems of data collection lag and fragmentation, lack of real-time and completeness, single cost dimension, insufficient analysis depth, extensive analysis method, low prediction accuracy, lack of dynamic optimization decision support, and difficulty in dealing with uncertainty in the prior art.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of data analysis, and particularly relates to a petroleum cost analysis method and system based on data analysis. BACKGROUND

[0002] The petroleum industry has the characteristics of long project cycle, multiple links, complex cost composition, and great influence by internal and external factors. The traditional petroleum cost management method often relies on periodic reports and experience judgment, and it is difficult to realize real-time monitoring and fine management of cost. In the whole life cycle of petroleum exploration, transportation and storage, human, equipment, time and other cost factors are intertwined, and there are many uncertainties, resulting in large cost fluctuations, which brings great challenges to enterprise cost control.

[0003] However, the prior art still has many defects, including:

[0004] 1) Data collection is lagging and fragmented, lacking real-time and completeness: in the prior art, the collection of cost data often relies on periodic manual reporting, paper records or scattered, non-integrated information systems, and there is serious lag; the data is scattered in different departments or systems such as exploration, engineering, logistics and warehousing, forming a "data island", lacking a unified data platform for real-time, continuous and comprehensive collection and integration of whole life cycle data, resulting in weak cost analysis foundation and difficulty in capturing the immediate reasons for cost changes;

[0005] 2) Single cost dimension, insufficient analysis depth: traditional methods often only focus on a few explicit costs such as raw material procurement cost and large equipment depreciation, while ignoring implicit or indirect costs such as human efficiency fluctuations, equipment idle time, unplanned downtime losses, logistics delays, storage losses, and the impact of policy and regulation changes; even if some multi-dimensional data is collected, there is a lack of effective correlation analysis means to identify the complex interaction of cost driving factors;

[0006] 3) Analysis method is extensive, and prediction accuracy is low: existing cost analysis mostly uses simple statistical methods, empirical formulas or linear regression based on historical data. These methods are difficult to handle the non-linear, high-dimensional and time-varying characteristics commonly existing in petroleum project cost data; they cannot effectively integrate real-time data and historical big data, and lack the ability to use advanced algorithms such as machine learning or deep learning to mine the internal laws of data; therefore, cost prediction often has large deviation, and cannot provide reliable basis for dynamic decision-making;

[0007] 4) Lack of dynamic optimization decision support, difficult to deal with uncertainty: the results of traditional cost analysis lack dynamic optimization decision support, even if some trend prediction can be made, it often stays at the static optimization level of a single target (such as the lowest cost); oil projects face multiple uncertainties such as changes in geological conditions, market price fluctuations, policy adjustments, safety accidents, etc.; existing technologies are difficult to provide real-time, multi-objective (such as cost, safety, efficiency, environmental protection, etc.) dynamic optimization strategies. SUMMARY

[0008] In order to solve the problems of data collection lag and fragmentation, lack of real-time and completeness, single cost dimension, insufficient analysis depth, rough analysis method, low prediction accuracy, lack of dynamic optimization decision support, and difficulty in dealing with uncertainty in the prior art, the present application aims to provide an oil cost analysis method and system based on data analysis.

[0009] The technical solution adopted by the present application is:

[0010] An oil cost analysis method based on data analysis, comprising the following steps:

[0011] Collecting real-time full life cycle data of oil projects, and extracting real-time multi-dimensional cost data from the real-time full life cycle data;

[0012] According to the real-time multi-dimensional cost data, using an oil cost analysis model to perform oil cost analysis, and obtaining real-time oil cost analysis results;

[0013] According to the real-time oil cost analysis results, using a project optimization model to generate project optimization strategies, and obtaining real-time project optimization strategies.

[0014] Further, the real-time full life cycle data includes real-time oil extraction stage data, real-time oil transportation stage data, and real-time oil storage stage data;

[0015] The real-time oil extraction stage data includes real-time exploration process data, real-time drilling process data, real-time fracturing process data, and real-time oil production process data;

[0016] The real-time oil transportation stage data includes real-time pipeline transportation process data, real-time railway transportation process data, real-time highway transportation process data, and real-time sea transportation process data;

[0017] The real-time oil storage stage data includes real-time oil depot storage process data, real-time storage tank storage process data, and real-time underground storage process data.

[0018] Further, the cost dimensions of the real-time multi-dimensional cost data include a human cost dimension, a device cost dimension, a time cost dimension, a direct cost dimension, an indirect cost dimension, and an external cost dimension.

[0019] Further, the real-time full life cycle data of the oil project is collected, and real-time multi-dimensional cost data in the real-time full life cycle data is extracted, including the following steps:

[0020] Using a data collection device, all data servers of the oil project in the full life cycle are connected, and real-time oil extraction stage data, real-time oil transportation stage data, and real-time oil storage stage data of the oil project are collected;

[0021] The real-time oil extraction stage data, the real-time oil transportation stage data, and the real-time oil storage stage data are integrated to obtain the real-time full life cycle data of the oil project, and pre-processing is performed to obtain pre-processed real-time full life cycle data;

[0022] According to the predetermined cost dimensions, the initial real-time multi-dimensional cost data in the pre-processed real-time full life cycle data is extracted, and the initial real-time multi-dimensional cost data is aligned and synchronized to obtain the final real-time multi-dimensional cost data.

[0023] Further, the oil cost analysis model is constructed based on the MPFEN-TCD-Attention-MLP algorithm, and the oil cost analysis model includes a multi-dimensional feature extraction module constructed based on the MPFEN algorithm, a causal feature extraction module constructed based on the TCD algorithm, a weighted fusion module constructed based on the Attention mechanism, and an oil cost analysis module constructed based on the MLP algorithm. The multi-dimensional feature extraction module and the causal feature extraction module are connected, and the multi-dimensional feature extraction module and the causal feature extraction module are connected to the weighted fusion module. The weighted fusion module is connected to the oil cost analysis module.

[0024] Further, according to the real-time multi-dimensional cost data, the oil cost analysis model is used to perform oil cost analysis to obtain real-time oil cost analysis results, including the following steps:

[0025] Using the multi-dimensional feature extraction module of the oil cost analysis model, real-time multi-dimensional features of the real-time multi-dimensional cost data are extracted;

[0026] According to the real-time multi-dimensional features, the causal feature extraction module of the oil cost analysis model is used to perform causal chain identification to obtain real-time causal chains and extract corresponding real-time causal features;

[0027] According to the dynamic attention weight, the weighted fusion module of the oil cost analysis model is used to perform weighted fusion on the real-time causal features and the real-time multi-dimensional features to obtain real-time fusion features.

[0028] According to the real-time fusion feature, the oil cost analysis module using the oil cost analysis model is used to perform oil cost analysis, and real-time oil cost analysis results are obtained.

[0029] Further, the real-time multi-dimensional feature includes real-time basic features, real-time derived features and real-time external features.

[0030] The real-time oil cost analysis results include real-time overall cost estimation results, real-time phased cost details, real-time multi-dimensional cost analysis results, real-time cost driving factor and influence analysis results, real-time cost anomaly analysis results, and real-time cost trend prediction results.

[0031] Further, the project optimization model is constructed based on the MAPGRPO algorithm, and the project optimization model includes a cycle-level project optimization strategy generation layer, a coordination layer and a stage-level project optimization strategy generation layer connected in sequence, the cycle-level project optimization strategy generation layer is provided with a first multi-optimization target set and a cycle-level agent, the coordination layer is provided with a coordination agent and a multi-target conflict resolution mechanism, and the stage-level project optimization strategy generation layer is provided with a second multi-optimization target set and a plurality of parallel stage-level agents.

[0032] Further, according to the real-time oil cost analysis results, the project optimization model is used to perform project optimization strategy generation, and real-time project optimization strategies are obtained, including the following steps:

[0033] According to the real-time oil cost analysis results, the cycle-level project optimization strategy generation layer of the project optimization model is used to perform project optimization strategy generation, and real-time cycle-level project optimization strategies are obtained.

[0034] Based on the multi-target conflict resolution mechanism, the coordination layer of the project optimization model is used to resolve multi-target conflicts of the real-time cycle-level project optimization strategies, and optimized real-time cycle-level real-time project optimization strategies are obtained.

[0035] According to the real-time oil cost analysis results and the optimized real-time cycle-level real-time project optimization strategies, the stage-level project optimization strategy generation layer of the project optimization model is used to perform project optimization strategy generation, and a plurality of real-time stage-level project optimization strategies are obtained.

[0036] The optimized real-time cycle-level real-time project optimization strategies and the plurality of real-time stage-level project optimization strategies are integrated to obtain real-time project optimization strategies.

[0037] A data analysis-based oil cost analysis system for implementing an oil cost analysis method, the system including a cost data extraction unit, an oil cost analysis unit and a project optimization unit connected in sequence.

[0038] The beneficial effects of the present application are:

[0039] The application provides a petroleum cost analysis method and system based on data analysis, which overcomes the disadvantages of data lag, fragmentation and incompleteness in the prior art. By collecting multi-dimensional data of the whole life cycle of petroleum exploration, transportation and storage in real time and effectively integrating the data, a solid and dynamic data foundation is provided for cost analysis, enabling managers to immediately grasp the real situation of project costs. The application makes up for the problem of single analysis dimension and insufficient depth in the prior art, not only analyzes basic costs such as manpower, equipment and time, but also extracts derived features and considers external features, and uses advanced deep learning models for fine-grained dimension-by-dimension cost analysis to reveal the deep reasons and mutual correlations of cost changes. The application solves the problem of weak prediction ability and large deviation in the prior art, can more accurately predict future cost trends by fusing historical data and real-time data and combining the powerful nonlinear mapping and pattern recognition capabilities of deep learning models, and provides a reliable basis for forward-looking decision-making. The application overcomes the limitations of the prior art, such as lack of dynamic optimization and difficulty in coping with uncertainty, and uses a project process optimization model to quickly generate real-time project optimization strategies that take into account costs, safety, efficiency and other multi-objectives according to real-time cost analysis results, guiding dynamic adjustment of projects in different stages and cycles and effectively coping with market changes and unexpected situations.

[0040] Other beneficial effects of the application will be further described in the specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flowchart of the petroleum cost analysis method based on data analysis in the application.

[0042] Figure 2 is a structural diagram of the petroleum cost analysis system based on data analysis in the application. DETAILED DESCRIPTION

[0043] The application will be further explained below in combination with the drawings and specific embodiments.

[0044] Embodiment 1:

[0045] As shown in Figure 1 , the embodiment provides a petroleum cost analysis method based on data analysis, including the following steps:

[0046] S1: Collecting real-time whole life cycle data of a petroleum project and extracting real-time multi-dimensional cost data in the real-time whole life cycle data;

[0047] The real-time whole life cycle data includes real-time petroleum exploration stage data, real-time petroleum transportation stage data and real-time petroleum storage stage data;

[0048] The real-time oil extraction stage data includes real-time exploration process data, real-time drilling process data, real-time fracturing process data, and real-time oil production process data;

[0049] The real-time oil transportation stage data includes real-time pipeline transportation process data, real-time railway transportation process data, real-time highway transportation process data, and real-time marine transportation process data;

[0050] The real-time oil storage stage data includes real-time oil depot storage process data, real-time storage tank storage process data, and real-time underground storage process data;

[0051] The cost dimensions of the real-time multi-dimensional cost data include human cost dimension, equipment cost dimension, time cost dimension, direct cost dimension, indirect cost dimension, and external cost dimension;

[0052] The human cost dimension includes the salaries, welfare, social security, and training costs of direct production personnel, management personnel, maintenance personnel, etc.; the equipment cost dimension includes equipment purchase, rental, depreciation, repair, maintenance, fuel power consumption, etc.; the time cost dimension includes implicit costs such as project cycle, operation time, waiting time, downtime, and time-related resource occupation costs; the direct cost dimension includes, for example, raw materials (chemicals, pipes, etc.), land use fees, environmental protection treatment fees, safety costs, etc.; the indirect cost dimension includes, for example, management fees, financial fees, and tax fees that are allocated to the project; and the external cost dimension includes, for example, customs duties for transportation routes, insurance fees, oil price fluctuations, and geopolitical risk premiums;

[0053] The real-time full life cycle data of the oil project is collected, and the real-time multi-dimensional cost data in the real-time full life cycle data is extracted, including the following steps:

[0054] S1-1: Using a data collection device, connecting all data servers of the oil project in the full life cycle, and collecting real-time oil extraction stage data, real-time oil transportation stage data, and real-time oil storage stage data of the oil project;

[0055] S1-2: Integrating the real-time oil extraction stage data, the real-time oil transportation stage data, and the real-time oil storage stage data to obtain the real-time full life cycle data of the oil project, and pre-processing to obtain pre-processed real-time full life cycle data;

[0056] The collected raw data is pre-processed, including: removing duplicate data; identifying and processing missing values (such as interpolation method, mean filling, deletion, etc.); identifying and correcting outliers (such as based on statistical methods or business rules); unifying data formats and units (such as unifying currency units to RMB, and unifying time units to hours or days);

[0057] S1-3: According to a plurality of preset cost dimensions, extract initial real-time multi-dimensional cost data in the pre-processed real-time full life cycle data, and align and synchronize the initial real-time multi-dimensional cost data to obtain final real-time multi-dimensional cost data;

[0058] The alignment and synchronization include: establishing unique identifiers for projects, equipment, personnel, job types, etc.; ensuring that data from different sources and different dimensions are aligned in the time dimension (such as daily, weekly, monthly alignment); mapping the cost subjects of different projects to a unified standard cost subject system for horizontal comparison and aggregate analysis;

[0059] S2: According to the real-time multi-dimensional cost data, using the oil cost analysis model, performing oil cost analysis to obtain real-time oil cost analysis results;

[0060] The oil cost analysis model is constructed based on a Multi scale Parallel Feature Extraction Network (MPFEN)-Temporal Causal Discovery (TCD)-Attention-Multi-Layer Perceptron (MLP) algorithm, and the oil cost analysis model includes a multi-dimensional feature extraction module constructed based on the MPFEN algorithm, a causal feature extraction module constructed based on the TCD algorithm, a weighted fusion module constructed based on the Attention mechanism, and an oil cost analysis module constructed based on the MLP algorithm. The multi-dimensional feature extraction module is connected with the causal feature extraction module, and both the multi-dimensional feature extraction module and the causal feature extraction module are connected with the weighted fusion module. The weighted fusion module is connected with the oil cost analysis module.

[0061] According to the real-time multi-dimensional cost data, using the oil cost analysis model, performing oil cost analysis to obtain real-time oil cost analysis results, including the following steps:

[0062] S2-1: Using the multi-dimensional feature extraction module of the oil cost analysis model, extracting real-time multi-dimensional features of the real-time multi-dimensional cost data;

[0063] The real-time multi-dimensional features include real-time basic features, real-time derived features, and real-time external features;

[0064] The real-time basic features include direct labor, equipment, time, material, etc. cost value; the real-time derived features include unit time cost, unit output cost, equipment utilization rate, personnel efficiency, transportation distance, storage turnover rate, etc.; the real-time external features include contemporaneous international oil price, exchange rate, political risk index of a specific region, transportation route congestion index, etc.

[0065] S2-2: According to the real-time multi-dimensional features, using the causal feature extraction module of the oil cost analysis model, the causal chain is identified, the real-time causal chain is obtained, and the corresponding real-time causal features are extracted;

[0066] The formula is:

[0067]

[0068] In the formula, is the real-time causal chain; is the causal discovery algorithm function; is the real-time basic feature, real-time derived feature, and real-time external feature;

[0069]

[0070] In the formula, is the real-time causal feature; is the time series causal reasoning function;

[0071] S2-3: According to the dynamic attention weight, using the weighted fusion module of the oil cost analysis model, the real-time causal features and real-time multi-dimensional features are weighted and fused to obtain real-time fusion features;

[0072] S2-4: According to the real-time fusion features, using the oil cost analysis module of the oil cost analysis model, the oil cost analysis is performed to obtain real-time oil cost analysis results;

[0073] The real-time oil cost analysis results include real-time overall cost estimation results, real-time phased cost details, real-time dimensional cost analysis results, real-time cost driving factor and influence analysis results, real-time cost anomaly analysis results, and real-time cost trend prediction results;

[0074] The real-time overall cost estimation results include the total cost prediction value of the oil full life cycle (exploration, transportation, storage) at the current time point or short period (such as the current hour, the current day, and the current week); compared with the preset budget or the historical average cost of the same period, presented in absolute value (over budget / saving amount) or percentage form;

[0075] The real-time phased cost details include: exploration stage cost: real-time drilling cost, fracturing cost, labor cost, equipment depreciation and maintenance cost, energy consumption cost, environmental governance cost, etc.; transportation stage cost: pipeline transportation cost (including pumping energy consumption, maintenance), highway / railway transportation cost (fuel, road and bridge fee, driver's salary), loading and unloading cost, loss cost, etc.; storage stage cost: storage tank rental / depreciation cost, maintenance cost, management labor cost, storage loss cost, etc.; these details not only show the current value, but also may contain the deviation from the planned value or benchmark value;

[0076] Real-time cost analysis results include: human cost analysis results: real-time human cost of each stage and each post, and its proportion in total cost, and may also include human efficiency indicators; equipment cost analysis results: real-time running cost, maintenance cost and depreciation cost of various key equipment (such as drilling rigs, pumps and storage tanks), and their proportions; time cost analysis results: the influence of each link (such as drilling period, transportation time and storage time) on total cost, which may quantify the specific embodiment of “time is money”, such as additional cost caused by delay; direct cost dimension analysis results: costs directly attributed to specific activities (such as material cost, fuel cost and direct labor cost), such as direct material cost of crude oil, chemicals and fuel in the exploitation stage, direct cost proportion, and proportion of direct cost of each stage in total cost; indirect cost dimension analysis results: total amount of indirect cost: expenses that are difficult to directly attribute but necessary (such as management fee, insurance fee and environmental governance fee), and indirect cost proportion: proportion of indirect cost of each stage in total cost; external cost dimension analysis results: total amount of external cost: cost affected by external factors (such as oil price fluctuation, policy change and weather influence), and external cost proportion: proportion of external cost of each stage in total cost;

[0077] Real-time cost driving factor and influence analysis results include: causal analysis based on the causal feature extraction module, which clearly indicates what the main driving factors of current cost are (for example, is it a certain equipment failure, a certain link inefficiency, or external price fluctuation); and quantifies the specific influence degree of these driving factors on total cost (for example, “the efficiency of equipment X decreases, leading to an increase of Y% in total cost”);

[0078] Real-time cost anomaly analysis results include: identifying cost items or links that deviate significantly from normal mode or expected range, issuing risk warnings for potential cost overruns or inefficiencies according to real-time data and model predictions, and possibly suggesting potential causes;

[0079] Real-time cost trend prediction results include: based on current real-time data and model, predicting the cost trend in the near future (such as the next few hours or days); for example, “if the current transportation efficiency continues, it is predicted that the transportation cost will increase by X% in the next 24 hours”;

[0080] S3: Based on the real-time oil cost analysis results, use the project optimization model to generate project optimization strategies, and obtain real-time project optimization strategies;

[0081] The project optimization model is constructed based on a Multi-Agent Parallel Group Relative Policy Optimization (MAPGRPO) algorithm, and the project optimization model comprises a period-level project optimization strategy generation layer, a coordination layer and a stage-level project optimization strategy generation layer connected in sequence, the period-level project optimization strategy generation layer is provided with a first multi-optimization target set and period-level agents, the coordination layer is provided with a coordination agent and a multi-target conflict resolution mechanism, and the stage-level project optimization strategy generation layer is provided with a second multi-optimization target set and a plurality of parallel stage-level agents;

[0082] According to the real-time oil cost analysis result, the project optimization model is used to generate a project optimization strategy, and a real-time project optimization strategy is obtained, including the following steps:

[0083] S3-1: According to the real-time oil cost analysis result, the period-level agent of the period-level project optimization strategy generation layer of the project optimization model is used to generate a project optimization strategy in combination with the first multi-optimization target set, and a real-time period-level project optimization strategy is obtained;

[0084] In this embodiment, the real-time oil cost analysis result includes: "the cost in the transportation stage is 15% higher than the budget due to temporary driver employment; the storage tank maintenance cost in the storage stage abnormally increases (due to corrosion problems); the delay in the transportation stage due to pipeline blockage results in additional penalties, and the decline in oil prices leads to a decrease in mining income, which indirectly affects the cost";

[0085] The first multi-optimization target set includes: cost minimization, time minimization and environmental protection compliance;

[0086] The real-time period-level project optimization strategy includes: adjusting the driver employment mode in the transportation stage (from temporary workers to long-term contract workers to reduce unit cost); and suggesting increasing the frequency of anticorrosion coating maintenance in the storage stage to reduce unexpected repair costs;

[0087] S3-2: Based on the multi-target conflict resolution mechanism, the coordination agent of the coordination layer of the project optimization model is used to resolve the multi-target conflict of the real-time period-level project optimization strategy, and an optimized real-time period-level real-time project optimization strategy is obtained;

[0088] Conflict identification: in the period-level strategy, optimizing the driver employment mode (long-term contract workers) may increase the indirect cost (such as training fee); and the increase in the frequency of anticorrosion maintenance (reduction in equipment cost) conflicts with the time minimization target (because maintenance needs to occupy storage time);

[0089] Multi-objective conflict resolution mechanism: priority adjustment: reserve manpower and equipment optimization strategy due to the highest cost weight; compromise: arrange anticorrosion maintenance during off-peak hours to reduce the impact on storage time;

[0090] Formula:

[0091]

[0092] In the formula, is the reward function of the coordination agent; is the cost function, time function and environmental pollution function; is the action; is the cost weight, time weight and environmental pollution weight;

[0093]

[0094] In the formula, is the action corresponding to the optimized real-time cycle level real-time project optimization strategy; is the real-time cycle level real-time project optimization strategy; is the real-time cycle level real-time project optimization strategy indication quantity; is the Q function of the coordination agent; is the set of alternative real-time cycle level real-time project optimization strategies; is the state; is the learnable parameter;

[0095] Optimized real-time cycle level real-time project optimization strategy: mixed employment mode (part-time long-term workers + part-time workers) is adopted in the transportation stage; anticorrosion maintenance is carried out at night in the storage stage to ensure that the daytime storage time is not affected;

[0096] S3-3: According to the real-time oil cost analysis result and the optimized real-time cycle level real-time project optimization strategy, combined with the second multi-optimization objective set, use the stage level agent of the project optimization model stage level project optimization strategy generation layer to generate project optimization strategy, and obtain several real-time stage level project optimization strategies;

[0097] The second multi-optimization objective set includes: efficiency maximization in the extraction stage, time minimization in the transportation stage;

[0098] The real-time stage level project optimization strategy includes real-time transportation stage project optimization decision and real-time storage stage project optimization decision;

[0099] Real-time transportation stage project optimization decision: analyze the reason for pipeline blockage (e.g. insufficient flow rate in a certain section of the pipeline), generate a decision (adjust pipeline pressure + activate backup pipeline): adjust the pipeline pressure parameter, increase the flow rate, and reduce the expected delay time by 20%; activate the backup pipeline to avoid late fees;

[0100] Real-time storage stage project optimization decision: combine with the night maintenance strategy to generate a decision (optimize the arrangement of storage tanks); optimize the arrangement of storage tanks to improve storage efficiency and reduce unit storage cost;

[0101] S3-4: Integrate the real-time periodic real-time project optimization strategy and several real-time stage-level project optimization strategies after optimization to obtain a real-time project optimization strategy;

[0102] Real-time project optimization strategy:

[0103] Human cost optimization: mixed employment mode is adopted in the transportation stage, which is expected to save 8%;

[0104] Equipment cost optimization: night corrosion maintenance and storage tank arrangement optimization in the storage stage, which is expected to save 12%;

[0105] Time cost optimization: pipeline pressure adjustment + backup pipeline, which is expected to reduce the delay time by 20% and save late fees;

[0106] Comprehensive benefits: the total cost is expected to be reduced by 15%, meeting the enterprise priority requirements.

[0107] Embodiment 2:

[0108] As shown in Figure 2 , the embodiment provides a petroleum cost analysis system based on data analysis, which is used to realize a petroleum cost analysis method. The system includes a cost data extraction unit, a petroleum cost analysis unit, and a project optimization unit connected in sequence;

[0109] The cost data extraction unit is used to collect real-time full life cycle data of petroleum projects and extract real-time multi-dimensional cost data from the real-time full life cycle data;

[0110] The petroleum cost analysis unit is used to perform petroleum cost analysis using a petroleum cost analysis model according to the real-time multi-dimensional cost data, and obtain real-time petroleum cost analysis results;

[0111] The project optimization unit is used to generate a real-time project optimization strategy using a project optimization model according to the real-time petroleum cost analysis results.

[0112] The application provides a kind of oil cost analysis method and system based on data analysis, overcome the disadvantages of data lag, fragmentation, not complete of prior art, by real-time acquisition petroleum exploitation, transportation, storage and other full life cycle multidimensional data, and effectively integrated, for cost analysis provides solid, dynamic data basis, so that managers can immediately grasp the real situation of project cost;Make up the problem of single analysis dimension and insufficient depth of prior art, not only analyze basic costs such as manpower, equipment, time, but also extract derived features and consider external features, and use advanced deep learning model to carry out fine division of dimension cost analysis, reveal the deep reason and mutual correlation of cost change;Solve the problem of weak prediction ability and large deviation of prior art, by fusing historical data and real-time data, combined with the powerful nonlinear mapping and pattern recognition ability of deep learning model, can more accurately predict future cost trend, provide reliable basis for forward-looking decision;Overcome the limitations of prior art, such as lack of dynamic optimization and difficulty in dealing with uncertainty, use project process optimization model, can generate real-time project optimization strategy considering cost, safety, efficiency and other multi-objective according to real-time cost analysis result, guide project to carry out dynamic adjustment in different stages and cycles, effectively deal with market changes and unexpected situations.

[0113] The application is not limited to the above optional embodiments, and anyone can derive other various forms of products under the inspiration of the application. The above specific embodiments should not be understood as limiting the protection scope of the application, and the protection scope of the application should be defined by the claims, and the specification can be used to explain the claims.

Claims

1. A data-driven method for oil cost analysis, characterized in that: Includes the following steps: Collect real-time full lifecycle data of oil projects and extract real-time multi-dimensional cost data from the real-time full lifecycle data; Based on real-time multi-dimensional cost data, an oil cost analysis model is used to conduct oil cost analysis and obtain real-time oil cost analysis results. The oil cost analysis model is constructed based on the MPFEN-TCD-Attention-MLP algorithm. The oil cost analysis model includes a multi-dimensional feature extraction module constructed based on the MPFEN algorithm of multi-scale parallel feature extraction network, a causal feature extraction module constructed based on the TCD algorithm of temporal causal reasoning, a weighted fusion module constructed based on the attention mechanism, and an oil cost analysis module constructed based on the MLP algorithm of multilayer perceptron. The multi-dimensional feature extraction module is connected to the causal feature extraction module, and both the multi-dimensional feature extraction module and the causal feature extraction module are connected to the weighted fusion module. The weighted fusion module is connected to the oil cost analysis module. Includes the following steps: The multidimensional feature extraction module of the oil cost analysis model is used to extract real-time multidimensional features of real-time multidimensional cost data. The real-time multidimensional features include real-time basic features, real-time derived features, and real-time external features; Based on real-time multidimensional features, the causal feature extraction module of the oil cost analysis model is used to identify causal chains, obtain real-time causal chains, and extract corresponding real-time causal features. Based on the dynamic attention weight, the weighted fusion module of the oil cost analysis model is used to perform weighted fusion of real-time causal features and real-time multidimensional features to obtain real-time fused features. Based on the real-time fusion characteristics, the oil cost analysis module of the oil cost analysis model is used to perform oil cost analysis and obtain real-time oil cost analysis results. The real-time oil cost analysis results include real-time overall cost estimation results, real-time phased cost details, real-time multi-dimensional cost analysis results, real-time cost driver and impact analysis results, real-time cost anomaly analysis results, and real-time cost trend prediction results. Based on the real-time oil cost analysis results, a project optimization strategy is generated using a project optimization model to obtain the real-time project optimization strategy. The project optimization model is constructed based on the MAPGRPO algorithm for relative policy optimization in parallel groups of multiple agents. The project optimization model includes a periodic project optimization policy generation layer, a coordination layer, and a stage-level project optimization policy generation layer connected in sequence. The periodic project optimization policy generation layer is equipped with a first set of multiple optimization objectives and periodic agents. The coordination layer is equipped with a coordinating agent and a multi-objective conflict resolution mechanism. The stage-level project optimization policy generation layer is equipped with a second set of multiple optimization objectives and several parallel stage-level agents. Includes the following steps: Based on the real-time oil cost analysis results, the project optimization strategy is generated using the periodic project optimization strategy generation layer of the project optimization model, resulting in a real-time periodic project optimization strategy. Based on the multi-objective conflict resolution mechanism, the coordination layer of the project optimization model is used to resolve multi-objective conflicts in the real-time periodic project optimization strategy, resulting in the optimized real-time periodic project optimization strategy. Based on the real-time oil cost analysis results and the optimized real-time periodic-level project optimization strategy, the project optimization strategy generation layer of the project optimization model is used to generate several real-time stage-level project optimization strategies. By integrating and optimizing the real-time periodic-level project optimization strategy and several real-time stage-level project optimization strategies, a real-time project optimization strategy is obtained.

2. The oil cost analysis method based on data analysis according to claim 1, characterized in that: The real-time full lifecycle data includes real-time oil extraction stage data, real-time oil transportation stage data, and real-time oil storage stage data; The real-time oil extraction stage data includes real-time exploration process data, real-time drilling process data, real-time fracturing process data, and real-time oil production process data; The real-time oil transportation phase data includes real-time pipeline transportation process data, real-time rail transportation process data, real-time road transportation process data, and real-time maritime transportation process data; The real-time oil storage stage data includes real-time oil depot storage process data, real-time tank storage process data, and real-time underground storage process data.

3. The oil cost analysis method based on data analysis according to claim 2, characterized in that: The cost dimensions of the real-time multi-dimensional cost data include labor cost dimension, equipment cost dimension, time cost dimension, direct cost dimension, indirect cost dimension, and external cost dimension.

4. The oil cost analysis method based on data analysis according to claim 3, characterized in that: Collecting real-time full lifecycle data of oil projects and extracting real-time multi-dimensional cost data from the real-time full lifecycle data includes the following steps: Using data acquisition equipment, connect all data servers of the oil project throughout its entire lifecycle and collect real-time data on the oil extraction phase, real-time oil transportation phase, and real-time oil storage phase of the oil project. By integrating real-time oil extraction stage data, real-time oil transportation stage data, and real-time oil storage stage data, real-time full lifecycle data of an oil project is obtained, and preprocessing is performed to obtain preprocessed real-time full lifecycle data. Based on several preset cost dimensions, the initial real-time multi-dimensional cost data is extracted from the preprocessed real-time full lifecycle data, and the initial real-time multi-dimensional cost data is aligned and synchronized to obtain the final real-time multi-dimensional cost data.

5. A data-based oil cost analysis system for implementing the oil cost analysis method as described in any one of claims 1-4, characterized in that: The system includes a cost data extraction unit, an oil cost analysis unit, and a project optimization unit connected in sequence.

Images