A crude oil centralized purchasing, warehousing and pipeline transportation integrated scheduling method for multiple refineries

Abstract

The application provides a crude oil centralized purchasing, storage and pipeline transportation integrated scheduling method for multiple refineries, which comprises the following steps: S1, demand collection and analysis, obtaining crude oil types, total demand, time window and demand priority parameters; S2, constructing a multi-dimensional resource model based on the demand parameters obtained in S1 and full-link resources, wherein the full-link resources comprise purchasing resources, transportation resources, port resources, storage resources, pipeline transportation resources and emergency resources; and S3, constructing a multi-objective optimization function to realize multi-constraint collaborative optimization. Since the application adopts full-link collaborative modeling, breaks the limitations of the segmented scheduling in the prior art, and realizes information sharing and plan linkage in each link, the application can effectively eliminate information islands, solve problems such as unloading congestion and pipeline idling caused by disconnection among links, and improve the collaborative efficiency of the supply chain.

Description

Technical Field

[0001] This invention relates to the field of intelligent crude oil refining, belonging to the new generation of information technology, specifically a method for the integrated scheduling of centralized crude oil procurement, storage, pipeline transportation, and distribution for multiple refineries. Background Technology

[0002] In the crude oil processing industry, the multi-refinery layout is the mainstream model. Its crude oil supply relies on the coordinated operation of the entire chain of "centralized procurement - ocean transportation - port unloading - multi-level storage - pipeline network - refinery receiving". This supply chain system is characterized by many links, complex constraints, great difficulty in coordination, and many influencing factors (such as international oil prices, shipping schedules, pipeline capacity, and inventory levels).

[0003] Currently, the core requirement for crude oil supply chain scheduling is to achieve "stable supply, optimal cost, and safety and compliance," which means ensuring that the crude oil demand of each refinery is met on time, in terms of quality and quantity, while minimizing the operating costs of the entire chain (procurement, transportation, storage, pipeline transportation, etc.) and strictly adhering to the rigid constraints of each link (such as pipeline transportation safety, upper and lower limits of inventory, and compatibility of crude oil properties).

[0004] With the expansion of refinery capacity and the increase in crude oil imports, the existing decentralized scheduling model can no longer meet the needs of full-chain coordination. The industry urgently needs an integrated scheduling method that can achieve multi-link linkage, multi-constraint adaptation, and multi-objective optimization to solve existing technical pain points and improve the overall resilience and economy of the supply chain.

[0005] Existing technologies propose a segmented scheduling method for the crude oil supply chain. This method mainly optimizes the scheduling of a single link or two adjacent links in the crude oil supply chain, but it does not achieve integrated planning across the entire chain. The specific implementation steps are as follows:

[0006] S1. Demand Acquisition: Collects basic information such as total crude oil demand and oil type from each refinery. No demand standardization analysis or priority classification is performed. It is only used as the basic input for scheduling.

[0007] S2. Segmented scheduling plan formulation: It is divided into three independent scheduling modules, which formulate procurement-transportation scheduling plan, warehousing scheduling plan, and pipeline-refinery receiving scheduling plan respectively. Each module operates independently and there is no linkage adjustment mechanism.

[0008] S3. Constraint Adaptation: Each segment scheduling module only considers the single constraint of its own link (such as the procurement module considering procurement cost constraints, and the pipeline module considering pipeline capacity constraints), without integrating the constraints of the entire link, and without considering the mutual influence between constraints.

[0009] S4. Multi-objective optimization: It adopts a fixed weight and simple linear normalization method to integrate the two core objectives of demand satisfaction rate and operating cost. It lacks quantitative consideration of inventory safety objectives and has no constraint penalty mechanism for violations. The optimization dimension is single and the adaptability is poor.

[0010] S5. Plan Execution and Correction: The plan execution is monitored manually. After deviations are found, the corresponding segment's scheduling plan is manually adjusted. The correction is lagging, inefficient, and cannot achieve full-link linkage correction.

[0011] S6. Emergency Response: When an emergency occurs, emergency resources (commercial oil storage, cross-enterprise oil borrowing) are manually called up. Emergency dispatch is disconnected from regular dispatch, there is no closed-loop management mechanism, and the emergency response efficiency is low.

[0012] The core drawbacks of existing technologies are as follows:

[0013] (1) Severe information silos and low collaboration efficiency: The existing technology adopts a segmented scheduling mode, with each link's scheduling module operating independently. The lack of a full-link information sharing and linkage mechanism has led to poor coordination between each link's plans (such as the disconnect between procurement plans and pipeline transportation plans, and the disconnect between warehousing plans and refinery demands), which in turn causes problems such as unloading congestion, pipeline idleness, and inability to meet demands on time. The root cause is the lack of full-link collaborative modeling and the absence of an end-to-end coordination mechanism.

[0014] (2) Insufficient adaptation to multiple constraints and lack of global optimization: Existing technologies only perform local optimization for a single link and a single constraint, without integrating rigid constraints across the entire chain (pipeline transportation can be stopped / cannot be stopped, upper and lower limits of inventory, number of berths, physical property matching, etc.), resulting in the problem of "local optima and global imbalance" in the optimization results (such as increasing the purchase batch to reduce procurement costs, leading to warehouse over-limit and excessive pipeline pressure). The essential reason is the lack of a multi-constraint collaborative optimization framework.

[0015] (3) Delayed emergency response and weak supply chain resilience: In the existing technology, emergency resource call is disconnected from regular scheduling, there is no automated emergency identification and linkage mechanism, manual intervention is required to call emergency resources in case of emergencies, and there is no closed-loop management of resource return after the emergency scenario is lifted, which makes it impossible to guarantee the continuity of refinery production. The essential reason is that emergency resources have not been integrated into the whole-chain integrated scheduling system.

[0016] (4) Low accuracy of demand matching and frequent supply and demand imbalance: Existing technology only collects basic demand information from refineries, without standardized analysis and priority division (rigid demand and flexible demand). The scheduling plan is resource-centric rather than demand-centric, resulting in oil type mismatch and inability to guarantee demand priority. The essential reason is the lack of demand-driven full-link scheduling logic.

[0017] (5) The multi-objective optimization logic is unreasonable and has poor adaptability: The existing technology uses fixed weights and simple linear normalization to integrate optimization objectives without considering the adaptation requirements of different scheduling scenarios (regular, emergency, cost priority) and without a constraint penalty mechanism for violations. This results in the optimization results being out of touch with actual operational needs and failing to balance the three major objectives of supply guarantee, cost and safety. The essential reason is the lack of refined and dynamic multi-objective integration logic. Summary of the Invention

[0018] To address the aforementioned problems in the background art, this invention proposes an integrated scheduling method for centralized crude oil procurement, storage, pipeline transportation, and distribution across multiple refineries.

[0019] Technical solution:

[0020] A method for integrated scheduling of centralized crude oil procurement, storage, and pipeline transportation for multiple refineries includes the following steps:

[0021] S1. Demand collection and analysis: Obtain parameters such as crude oil type, total demand, time window, and demand priority.

[0022] S2. Based on the demand parameters and end-to-end resources obtained in S1, construct a multi-dimensional resource model, wherein the end-to-end resources include procurement resources, transportation resources, port resources, warehousing resources, pipeline resources, and emergency resources;

[0023] S3. Construct a multi-objective optimization function to achieve multi-constraint collaborative optimization;

[0024] S4. Perform constraint verification on the scheduling plan obtained in S3. If the verification fails, return to S3 for re-optimization; if the verification is compliant, output the scheduling plan.

[0025] S5, Emergency Dispatch and Coordination under the Correction Mechanism.

[0026] Preferably, in S1, the collected demand information is standardized, cleaned, and parsed to remove invalid data, classify rigid and flexible demands, and establish a unified refinery demand database to provide accurate input for the multi-dimensional resource model.

[0027] Preferably, in S2, the rigid constraint conditions of each link are determined.

[0028] Preferably, in S2, based on the collected refinery demand parameters and the full-chain resource data of procurement, warehousing, pipeline transportation, and distribution, a multi-dimensional resource model coupled with "demand dimension - resource dimension - constraint dimension" is constructed. This model achieves deep integration of demand and full-chain resources through data association mapping, taking refinery demand parameters as target input and full-chain resource parameters as supply input, and simultaneously embedding process, cost, and time constraints of each link to form a three-dimensional linkage resource matching basic model of "demand-supply-constraint", providing a quantitative analysis carrier for the subsequent integrated scheduling plan generation.

[0029] Preferably, in S3, the multi-objective optimization function is:

[0030]

[0031] In the formula, This represents the normalized demand satisfaction rate. This represents the normalized operating cost. This represents the normalized inventory over-limit rate. The penalty coefficient is... To constrain violations and penalties, These are the weighting coefficients for demand fulfillment rate, operating costs, and inventory overrun rate, respectively.

[0032] Preferably, the solution is obtained by using a genetic algorithm, particle swarm optimization algorithm, simulated annealing algorithm, or ant colony optimization algorithm for iterative solution.

[0033] Preferred, The values ​​are automatically adjusted according to the scenario. In a preferred embodiment, the adjustment of the weight coefficient is triggered by a target threshold, or by using operational data trend prediction (adjusting the weight coefficient in advance based on the changing trend of the entire operational data to achieve more accurate scenario adaptation).

[0034] Preferably, demand fulfillment rate, operating costs, and inventory overrun rate are normalized using the following formula:

[0035]

[0036] In the formula, To adapt the adjustment factor to the demand, Demand fulfillment rate; deviation correction factor for normalized demand fulfillment rate. Linear or piecewise correction methods can be used to replace the existing exponential correction. As long as the difference in the adaptation between rigid and flexible demand can be compensated and the demand priority can be adapted, the same effect can be achieved.

[0037]

[0038] In the formula, This is the lower limit threshold for end-to-end operating costs. The upper limit threshold, For operating costs;

[0039]

[0040] In the formula, This is the influence coefficient exceeding the limit. For inventory over-limit rate; the exponential decay correction of normalized inventory over-limit rate can be replaced by piecewise linear penalty correction. As long as the difference in the impact of slight over-limit and severe over-limit can be distinguished, the penalty intensity can be strengthened to achieve the same purpose.

[0041] And calculate the consistency coefficient. ,like If the normalization is successful, then the normalization is effective; otherwise, the demand adaptation correction coefficient should be readjusted. Full-chain operating cost lower limit threshold and upper limit threshold Over-limit influence coefficient .

[0042] Preferred demand satisfaction rate Operating costs Inventory over-limit rate Calculate using the following formulas respectively:

[0043]

[0044] In the formula,

[0045] For the first Total crude oil demand of each refinery (corresponding to the resources required by the refinery and reflecting the demand of end-user production).

[0046] For the first The total amount of crude oil actually obtained by each refinery This represents the total number of refineries (corresponding to pipeline / distribution resources, reflecting the resource allocation results);

[0047]

[0048] In the formula,

[0049] This refers to procurement costs (corresponding to resources in the procurement process, including crude oil procurement price, batch size, and supplier fulfillment costs).

[0050] This includes transportation and demurrage costs (corresponding to resources in the ocean / pipeline transportation segment, including vessel / pipeline capacity, cycle time, and demurrage fees).

[0051] This refers to warehousing and pipeline transportation costs (corresponding to resources in the warehousing and pipeline transportation links, including oil depot capacity, storage costs, pipeline transportation capacity, energy consumption and maintenance costs).

[0052] Emergency resource mobilization cost (corresponding to emergency resource links, including commercial oil storage, cross-enterprise oil borrowing, temporary dispatch changes and other emergency support resources);

[0053]

[0054] In the formula, This represents the total number of monitoring points for oil depot and refinery inventory (corresponding to resources in the storage and refinery inventory stages, covering multiple levels of inventory nodes).

[0055] For the first The duration of inventory exceeding the limit at each monitoring point (corresponding to warehouse / refinery inventory constraints, reflecting the occupancy of inventory resources and the safety boundary).

[0056] The scheduling cycle (corresponding to the entire time dimension of resources, with a unified time benchmark for procurement, transportation, warehousing, pipeline transportation, and emergency response).

[0057] Preferably, in S5, real-time collection of full-link operation data is performed, the preliminary scheduling plan is compared with the actual operation status, the plan deviation is automatically identified and the correction mechanism is triggered; the scheduling plans of each link are adjusted in conjunction, and the multi-objective optimization function is re-substituted for verification to ensure that the plan is adaptive to the actual operation scenario.

[0058] Preferably, in S5, after the correction mechanism is triggered, emergency resources (commercial oil storage, cross-enterprise oil borrowing, temporary procurement, and crude oil substitution) are invoked, and the weight coefficients of the multi-objective optimization function and the scheduling plan are adjusted simultaneously to ensure the continuity of refinery demand; after the correction mechanism is lifted, the emergency resource return demand is integrated into the regular scheduling plan.

[0059] Step 5: Emergency Dispatch and Coordination (see Appendix 3) Establish an automated coordination mechanism of "full-link emergency perception - graded response - dynamic correction - closed-loop replenishment" to collect four core indicators in real time: shipping delay amount ΔT, inventory deviation rate ΔI, pipeline flow abnormality rate ΔF, and refinery demand gap ΔD. Automatically identify four emergency scenarios: procurement gap, inventory over-limit, pipeline failure, and sudden demand change by using preset thresholds (ΔT≥10%, ΔI≥20%, ΔF≥15%, ΔD≥10%).

[0060] Emergency resources will be allocated in a tiered manner based on the severity of the situation:

[0061] Mild emergency response (single indicator trigger): Activate regional commercial oil reserves and adjust pipeline flow distribution;

[0062] Moderate emergency response (triggered by two indicators): Activate cross-enterprise oil borrowing channels and switch to backup pipeline routes;

[0063] Severe emergency (triggered by multiple indicators): Activate the refinery's phased load reduction plan to ensure continuous production at core refineries.

[0064] Synchronously and dynamically update the weights of the multi-objective optimization function: , The weight of "demand fulfillment rate" was increased, while the weight of "cost" was decreased, prioritizing continuous refinery production above all else, and a new emergency dispatch plan was generated. : Target weight of refinery demand fulfillment rate (value range [0,1]). The higher the weight, the higher the priority of the scheduling plan to ensure that the refinery's crude oil demand is met in full and on time. : The target weight of the end-to-end scheduling cost (value range [0,1]). The higher the weight, the more the scheduling plan prioritizes minimizing the end-to-end cost of procurement, warehousing, and pipeline transportation. The weight adjustment increment (a fixed quantitative value, such as 0.1 / 0.2, which can be adapted to the actual scenario of the refining and chemical industry) is a positive constant.

[0065] Beneficial effects:

[0066] (1) Since the present invention adopts full-link collaborative modeling, it breaks the limitations of segmented scheduling in the existing technology and realizes information sharing and planning linkage in each link. Therefore, it can effectively eliminate information silos, solve problems such as unloading congestion and pipeline idleness caused by link disconnection, and improve supply chain collaborative efficiency.

[0067] (2) Because the present invention constructs a multi-constraint collaborative optimization framework and a refined multi-objective integration logic, and combines dynamic weights and constraint violation penalty mechanisms, it can achieve full-link global optimization, balance the three major objectives of supply guarantee, cost and security, and solve the problems of local optimization and dimensional imbalance in the existing technology;

[0068] (3) Since the present invention integrates emergency resources into an integrated scheduling system and establishes an automated emergency linkage and closed-loop correction mechanism, it can quickly respond to emergencies, ensure the continuity of refinery production, and enhance the resilience of the supply chain. Attached Figure Description

[0069] Figure 1 A schematic diagram of the multi-objective optimization logic.

[0070] Figure 2 This is a schematic diagram illustrating the collaboration across the entire crude oil supply chain.

[0071] Figure 3 This is a schematic diagram illustrating the coordination between emergency and routine dispatching. Detailed Implementation

[0072] The present invention will be further described below with reference to embodiments, but the scope of protection of the present invention is not limited thereto:

[0073] The integrated scheduling method for centralized crude oil procurement, storage, and pipeline transportation for multiple refineries provided by this invention focuses on achieving coordinated planning across the entire chain. It centers on multi-objective quantitative optimization, taking into account multiple constraints and emergency needs, without delving into the detailed execution steps of each module. Emphasizing overall control, the complete closed-loop steps are as follows, with corresponding appendices. Figure 1-3 Logical connections:

[0074] Step 1: Requirements Gathering and Analysis (see attached document) Figure 2 (Demand from Chinese refineries)

[0075] Collect crude oil demand information from multiple refineries, including but not limited to core parameters such as crude oil type, total demand, time window, and demand priority (rigid / flexible); standardize, clean, and parse the collected demand information, remove invalid data, classify rigid and flexible demand, and establish a unified refinery demand database to provide accurate input for end-to-end collaborative scheduling; this step solves the problem of low accuracy in demand adaptation of existing technologies and ensures that scheduling plans are demand-centric.

[0076] Step 2: End-to-end resource modeling (see attached document) Figure 2 (Mid-to-end link)

[0077] This process involves a comprehensive analysis of all resources across the crude oil supply chain, including procurement resources (crude oil suppliers, procurement prices, procurement volumes), transportation resources (ocean-going vessels, transportation cycles), port resources (number of berths, unloading capacity), storage resources (multi-level oil depot capacity, storage costs), pipeline resources (pipeline network, transport capacity, and whether or not it can be shut down), and emergency resources (commercial oil reserves, cross-enterprise oil borrowing channels). A unified, multi-dimensional resource model is constructed, clearly defining the rigid constraints and cost benchmarks for each stage (such as upper and lower limits of inventory, upper limits of pipeline capacity, and berth unloading efficiency). This model is then linked to the demand database and constraints, forming the foundation for collaborative scheduling across the entire supply chain. This step addresses the issues of fragmented resource modeling and insufficient constraint adaptation in existing technologies.

[0078] Step 3: Multi-constraint collaborative optimization (see attached diagram) Figure 1 (Complete process)

[0079] Based on the refinery demand database, the end-to-end resource model, and all rigid constraints, the multi-objective optimization function constructed in this invention is substituted into the solution. Iteratively solved using an adapted optimization algorithm (such as genetic algorithm or particle swarm optimization algorithm), an integrated preliminary scheduling plan covering the entire process of procurement, transportation, port unloading, warehousing, and pipeline transportation is generated. In this step, through refined normalization processing, dynamic scenario adaptation weights, and constraint violation penalty mechanisms, the scheduling plan is ensured to meet multiple constraint requirements, achieving global optimization of the three major objectives of supply guarantee, cost, and safety. This step solves the problems of lack of global optimization and poor multi-objective adaptability in existing technologies.

[0080] Step 4: Dynamically adjust the scheduling plan (see attached document) Figure 2 End-to-end linkage, attached Figure 1 (Constraint verification process)

[0081] Perform constraint verification on the scheduling plan obtained in step 3. If the verification fails, return to step 3 for re-optimization; if the verification is compliant, output the scheduling plan.

[0082] Step 5: Emergency Dispatch and Coordination (see attached document) Figure 3 (Complete closed loop)

[0083] Real-time collection of end-to-end operational data (such as actual procurement volume, transportation arrival time, oil depot inventory levels, and actual pipeline flow) is used to compare the preliminary scheduling plan with the actual operating status. This automatically identifies plan deviations (such as arrival delays and inventory anomalies) and triggers a correction mechanism. The system also coordinates adjustments to the scheduling plans at each stage (such as adjusting pipeline flow and optimizing warehouse allocation), re-submitting the data into a multi-objective optimization function for verification. This forms a closed loop of "monitoring-identification-correction-verification," ensuring the plan adapts to the actual operating scenario. This step addresses the problems of delayed and inefficient plan correction in existing technologies.

[0084] Real-time identification of emergency scenarios across the entire supply chain (such as procurement gaps due to shipping delays, inventory overruns, and pipeline failures) automatically triggers emergency mechanisms, mobilizes emergency resources (commercial oil storage, cross-enterprise oil borrowing), and simultaneously adjusts the weight coefficients of the multi-objective optimization function (prioritizing the increase of demand fulfillment rate weight) and the scheduling plan to ensure continuous refinery demand. After the emergency scenario is resolved, the return of emergency resources is integrated into the regular scheduling plan, achieving closed-loop coordination between emergency and regular scheduling. This step solves the problem of existing technologies being disconnected from emergency and regular scheduling and having weak risk resistance capabilities.

[0085] To ensure the feasibility of the technical solution, this invention defines core variables, optimization normalization processing logic, dynamic weight setting logic, and multi-objective optimization functions. The application process is demonstrated with exemplary cases, as follows:

[0086] Core variable definition

[0087] Variables related to demand fulfillment rate:

[0088] To maximize the refinery's demand fulfillment rate (target value). For the first Total crude oil demand of each refinery For the first The total amount of crude oil actually obtained by each refinery Total number of refineries; Demand fulfillment rate adaptation value:

[0089]

[0090] Operating cost related variables:

[0091] The total cost of operation across the entire value chain (minimizing the target value). For procurement costs, For transportation and demurrage costs, For warehousing and pipeline transportation costs, Cost of emergency resource mobilization; Cost adjustment value:

[0092]

[0093] Inventory safety-related variables:

[0094] The target is to minimize the inventory over-limit rate. This represents the total number of monitoring points for oil depot and refinery inventory. For the first The duration of inventory exceeding the limit at each monitoring point For scheduling cycle; inventory overrun rate adaptation value:

[0095]

[0096] Normalization processing logic

[0097] Due to demand satisfaction rate Operating costs Inventory over-limit rate The dimensions and orders of magnitude differ significantly, and direct integration would lead to an imbalance in the optimization objective. This invention employs a dual processing logic of "sub-objective normalization + dynamic correction," as detailed below:

[0098] Demand satisfaction rate normalization:

[0099] Its value range is No additional linear scaling is required; optimization is achieved using a "bias correction factor," with the correction formula as follows: ,in Adjustment factor to suit demand ( This is used to compensate for the difference in the adaptation between flexible and rigid demand. The higher the proportion of rigid demand in a refinery, The larger the value, the more likely it is to be prioritized for meeting essential needs.

[0100] Operating cost normalization:

[0101] Using the "interval scaling method + cost weight adjustment", the formula is: ,in This is the lower limit threshold for end-to-end operating costs. The upper limit threshold; if ,but (Optimal cost); if ,but (Cost exceeded the limit, triggering optimization and adjustment).

[0102] Normalized inventory overage rate:

[0103] Using "exponential decay correction", the formula is: ,in The influence coefficient exceeding the limit ( Strengthen the penalties for exceeding inventory limits and differentiate the impact of minor and severe over-limits.

[0104] Normalized consistency check:

[0105] Calculate the consistency coefficient ,like If the normalization is successful, then the normalization is effective; otherwise, the correction coefficients and thresholds are readjusted to ensure consistency across multiple objectives.

[0106] Weight setting logic (dynamically adapting to different scenarios)

[0107] The logic of "dynamic scene adaptation weight" is adopted. These are the weighting coefficients for demand fulfillment rate, operating costs, and inventory overrun rate, respectively. Automatically adjust based on the scheduling scenario:

[0108] Typical scenarios: To balance the three major objectives;

[0109] Emergency Scenario: Prioritize supply;

[0110] Cost-first scenario: Prioritize cost reduction;

[0111] Note: This only indicates that multiple scenarios are supported as examples; it does not mean that only the above three scenarios are allowed, or that the weight values ​​for each scenario must be as shown above. Specific scenarios and weights can be configured by the customer.

[0112] It also supports dynamic weight adjustment. When a target fails to reach the preset threshold, the corresponding weight is automatically adjusted until the requirements are met.

[0113] Multi-objective optimization function

[0114] The optimization function is constructed using a "weighted summation + penalty term" formula, as follows:

[0115]

[0116] in, As a reverse indicator of demand satisfaction rate, Penalty coefficient ( ), To constrain violations and ensure that optimization results balance overall optimality with compliance.

[0117] Exemplary Case

[0118] To facilitate understanding of the application of the above quantitative logic, an exemplary case is provided in conjunction with a typical scheduling scenario. All data are exemplary values ​​and do not involve specific enterprises or oil depots; they are only used to demonstrate the calculation process.

[0119] (1) Basic parameters of the case:

[0120] Number of refineries , (100% rigidity) (Rigidity 80%) (Rigidity 60%); Inventory monitoring points Scheduling cycle ; , ; , , Weights for typical scenarios Unrestrained violations ( ).

[0121] (2) Actual operating data:

[0122] , , ; , , , Only one monitoring point exceeded the limit for one day. ).

[0123] (3) Calculation demonstration:

[0124] Demand fulfillment rate: ;

[0125] Operating costs: ;

[0126] Inventory overage rate: ;

[0127] Normalization: , , Consistency check (efficient);

[0128] Optimize function computation: (Approaching 0, satisfying optimization requirements).

[0129] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A crude oil centralized procurement, storage and pipeline integrated scheduling method for multiple refineries, characterized in that Includes the following steps: S1. Demand collection and analysis: Obtain parameters such as crude oil type, total demand, time window, and demand priority. S2. Based on the demand parameters, end-to-end resources, and constraints of each link obtained in S1, a multi-dimensional resource model is constructed. The multi-dimensional resource model includes demand dimension, resource dimension, and constraint dimension. The model integrates demand and end-to-end resources through data association mapping, taking refinery demand parameters as target input and end-to-end resource parameters as supply input, and simultaneously embedding process, cost, and time constraints of each link. S3. Construct a multi-objective optimization function to achieve multi-constraint collaborative optimization. The multi-objective optimization function is as follows: In the formula, denotes the normalized demand satisfaction rate, denotes the normalized operating cost, denotes the normalized out-of-stock rate, is a penalty coefficient, is a constraint violation penalty term, are weight coefficients of the demand satisfaction rate, the operating cost, and the out-of-stock rate, respectively; S4. Perform constraint verification on each stage of the scheduling plan obtained in S3. If the verification fails to meet the rules, return to S3 for re-optimization. Verify compliance before outputting the scheduling plan; S5 collects real-time operational data across the entire chain, compares the preliminary scheduling plan with the actual operational status, and automatically identifies plan deviations and triggers correction mechanisms. The scheduling plans of each stage are adjusted in a coordinated manner, and the multi-objective optimization function is re-substituted for verification to ensure that the plan adapts to the actual operation scenario. The correction mechanism includes: First layer: Real-time data monitoring triggers correction. Monitor crude oil arrival delay rate, refinery demand change rate, pipeline pressure anomaly, and storage capacity occupancy rate. If any indicator reaches the threshold, correction is triggered. The second layer: a tiered correction strategy, which takes different correction measures based on the degree of deviation from the threshold. Minor adjustments: Adjusting pipeline flow distribution ratio and optimizing warehouse outbound sequence; Moderate adjustments: adjusting pipeline routes, changing some procurement batches, and activating regional emergency reserves; Major revisions: restructuring scheduling plans, initiating cross-regional emergency coordination, and adjusting refinery production priorities; The third layer: After the closed-loop feedback iterative correction is executed, the recovery status of the indicators is monitored regularly until all indicators return to the threshold; the correction records are summarized regularly, the trigger threshold and correction strategy parameters are optimized, and the correction response speed is iteratively improved.

2. The method according to claim 1, characterized in that... In S1, the collected demand information is standardized, cleaned, and parsed to remove invalid data, classify rigid and flexible demands, and establish a unified refinery demand database to provide accurate input for multi-dimensional resource models.

3. The method according to claim 1, characterized in that... In S3, a genetic algorithm or particle swarm optimization algorithm is used for iterative solution.

4. The method according to claim 1, characterized in that... In S3 The values ​​are automatically adjusted based on the scenario.

5. The method according to claim 1, characterized in that... In S3, demand fulfillment rate, operating costs, and inventory overrun rate are normalized using the following formula: In the formula, To adapt the adjustment factor to the demand, For demand satisfaction rate; In the formula, This is the lower limit threshold for end-to-end operating costs. The upper limit threshold, For operating costs; In the formula, This is the influence coefficient exceeding the limit. Inventory overage rate; Calculate the consistency coefficient ,like If so, then normalization is effective; Otherwise, readjust the demand adaptation correction factor. Full-chain operating cost lower limit threshold and upper limit threshold Over-limit influence coefficient .

6. The method according to claim 5, characterized in that... Demand fulfillment rate Operating costs Inventory over-limit rate Calculate using the following formulas respectively: In the formula, For the first Total crude oil demand of each refinery For the first The total amount of crude oil actually obtained by each refinery This represents the total number of refineries. In the formula, For procurement costs, For transportation and demurrage costs, For warehousing and pipeline transportation costs, Costs associated with emergency resource allocation; In the formula, This represents the total number of monitoring points for oil depot and refinery inventory. For the first The duration of inventory exceeding the limit at each monitoring point The scheduling period is [number].

7. The method according to claim 1, characterized in that... After the correction mechanism is lifted, the need to return emergency resources will be integrated into the regular dispatch plan.

Images