A biological aviation coal full-process multi-device collaborative digital twin and intelligent scheduling system

Abstract

This invention discloses a collaborative digital twin and intelligent scheduling system for the entire process of bio-jet kerosene, comprising: a data perception layer, a device model layer, a collaborative scheduling layer, an application service layer, and a human-computer interaction layer. The data perception layer collects process parameters in real time from five stages: Fischer-Tropsch synthesis, hydroisomerization, hydrocracking, fractionation, and blending. The device model layer calculates the material and energy balance of each stage using five independent digital twin models. The collaborative scheduling layer integrates a material-energy-hydrogen network joint model, an intermediate storage tank buffer scheduling engine, a full-process production optimization engine, and a real-time carbon emission calculation module, achieving joint solution and collaborative optimization of the five stages. This invention upgrades intermediate storage tanks from passive storage to active resource scheduling and enables online carbon emission calculation throughout the entire process, overcoming the problems of traditional independent operation of each stage leading to suboptimal global performance, low tank utilization, reliance on manual scheduling, and lagging carbon calculation.

Description

Technical Field

[0001] This invention belongs to the field of intelligent production technology of bio-jet fuel, specifically relating to a collaborative digital twin and intelligent scheduling system for multiple devices throughout the entire bio-jet fuel production process. Background Technology

[0002] Sustainable aviation fuel (SAF) is one of the core pathways for the aviation industry to achieve its decarbonization goals. According to the International Air Transport Association (IATA), global SAF demand is projected to reach 23 million tons per year by 2030 and 500 million tons per year by 2050. The Fischer-Tropsch (FT) synthesis route, due to its wide adaptability of feedstocks and high technological maturity, is currently one of the mainstream pathways for the industrial production of bio-aviation kerosene. Its entire process typically encompasses five core stages: the Fischer-Tropsch synthesis reactor (syngas → synthetic oil), the hydroisomerization unit (n-alkanes → isoalkanes), the hydrocracking unit (heavy fraction to lighter fractions), the fractionation tower (product cutting and separation), and the blending tank (multi-component blending to produce aviation kerosene products). With the rapid expansion of the SAF industry and increasingly stringent carbon emission regulations, international frameworks such as EU RED II and ICAO CORSIA have put forward requirements for traceability and verifiability of carbon emissions throughout the entire life cycle of SAFs. Digital twin technology, as an important means to achieve process optimization and precise control of carbon emissions, is gradually being applied to the refining and chemical industry and has become a key technology direction to promote the high-quality development of the SAF industry.

[0003] In existing technologies, several technical solutions have been developed for simulation and optimization of refining and chemical processes. Regarding process simulation, KBC's Petro-SIM 7.6 digital twin platform encompasses the FTR-SIM Fischer-Tropsch synthesis reactor model, as well as simulation functions for hydrotreating, isomerization, and hydrocracking, enabling process simulation and operational condition analysis for each unit. In terms of optimization and scheduling, Chinese patent CN121596835A discloses a digital twin-based full-process optimization management system for biomass fuel conversion, focusing on raw material quality analysis and dynamic adjustment of pretreatment processes; Chinese patent CN121765953A discloses an automatic blending parameter optimization method and system for bio-aviation fuel, achieving formula optimization in the blending section. Regarding hydrogen resource management, Sinopec Dalian Research Institute's H2-STAR technology discloses methods for hydrogen network pinch point calculation and hydrogen utilization efficiency optimization; Daqing Petrochemical has disclosed a hydrogen system pinch point analysis model, cost accounting model, and online optimization platform, achieving coordinated optimization of the hydrogen network and hydrogen production and consumption units. In terms of carbon emission management, some companies have developed intelligent carbon emission control systems, which enable dynamic statistical accounting and online monitoring and analysis of carbon emission data.

[0004] However, the aforementioned existing technologies still have many shortcomings. First, each section is simulated or optimized independently, lacking the ability to solve the entire process simultaneously. Petro-SIM 7.6 focuses on independent simulation of each unit, and CN121596835A focuses on feedstock pretreatment; neither has achieved real-time simultaneous optimization of the material-energy-hydrogen networks across the five sections of FT synthesis, hydroisomerization, hydrocracking, fractionation, and blending. The sum of the local optima of each section is far lower than the global optimum. Second, intermediate storage tanks are only used as passive storage facilities and are not included in active scheduling decisions. In existing technologies, intermediate storage tanks between FT synthesis and downstream hydrogenation units only serve as buffers. When upstream and downstream capacity fluctuates due to catalyst activity decay or periodic feed changes, frequent tank level limits lead to unplanned load adjustments or even partial shutdowns. Third, production scheduling relies on manual experience. When switching between different grades of SAF products, the proportion of transition feed is as high as 5% to 8%, the switching time is long, and the maintenance window arrangement does not fully consider the coupling relationship between upstream and downstream. Furthermore, carbon emission accounting lags behind – existing carbon emission control systems mostly focus on total carbon emission statistics at the enterprise or plant level, failing to achieve second-level carbon footprint tracking and online carbon intensity calculation for materials and energy consumption data across all stages of the entire process. This makes it difficult to meet the certification requirements of EURED II and ICAO CORSIA for the real-time nature and traceability of carbon emission data. Finally, there is a mismatch between hydrogen supply and demand – fluctuations in PSA plant operation lead to unstable hydrogen purity, forcing downstream hydrogen refueling units to frequently adjust the hydrogen-to-oil ratio. Current technologies lack a complete solution for integrating PSA hydrogen recovery, hydrogen buffer tanks, and downstream hydrogen refueling units into a unified hydrogen network for real-time balancing and scheduling. Summary of the Invention

[0005] In view of this, the present invention proposes a collaborative digital twin and intelligent scheduling system for multiple devices throughout the entire process of bio-jet fuel.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a multi-device collaborative digital twin and intelligent scheduling system for the entire process of bio-jet kerosene, comprising: The data perception layer is used to collect process parameters of the five sections and the tank area in real time through the DCS / SCADA system and transmit them to the device model layer and the collaborative scheduling layer. The device model layer includes five independent digital twin models corresponding to five work sections, which are used to perform material balance and energy balance calculations based on process parameters, obtain material balance data and energy balance data for each work section, and transmit them to the collaborative scheduling layer. The collaborative scheduling layer includes a material-energy-hydrogen network joint model, an intermediate storage tank buffer scheduling engine, a full-process production scheduling optimization engine, and a real-time carbon emission calculation module. It is used to solve the full-process material-energy-hydrogen network joint model, optimize the intermediate storage tank buffer scheduling strategy, optimize the multi-objective production scheduling within the future scheduling window, and calculate the full-process carbon emissions in real time based on process parameters, material balance data, and energy balance data, so as to obtain a collaborative scheduling scheme and transmit it to the application service layer. The application service layer is used to optimize the entire process of online production scheduling, perform dynamic collaborative scheduling, real-time carbon accounting, and full-process anomaly warning based on the collaborative scheduling scheme, and obtain visualized data to be transmitted to the human-computer interaction layer. The human-computer interaction layer is used to generate and display a graphical user interface based on the visualized data.

[0007] Preferably, the material-energy-hydrogen network joint model is used to perform the following operations based on the material balance data, energy balance data, and utility data of each process section: Material transfer matrix operation is performed by establishing a directional material transfer diagram with five sections as nodes. The node set includes the Fischer-Tropsch synthesis reactor section, the hydroisomerization unit section, the hydrocracking unit section, the fractionation tower section, and the blending tank section. The edges in the diagram represent the material flow between each section. Material balance is performed based on the vector balance relationship that the total inlet material flow rate of each section is equal to the sum of the total outlet material flow rate and the material accumulation, to obtain real-time material allocation data and material accumulation data between each section. The energy supply and demand matrix operation is based on enthalpy balance. An energy supply and demand matrix is ​​established according to the energy supply relationship between each section. The matrix elements represent the available energy provided by one section to another. The matching relationship between temperature and heat grade is calculated and the solution is obtained with the goal of minimizing the total amount of externally supplied steam. The hydrogen balance network calculation takes the PSA unit as the hydrogen hub node. Based on real-time data such as the new hydrogen replenishment flow rate, PSA feed and product flow rates and purity, hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, exhaust gas emission, composition, and estimated dissolution losses, and constrained by the conservation of hydrogen mass, the calculation is performed on the basis that the total hydrogen supply equals the total hydrogen consumption. This yields the hydrogen distribution scheme among the various processes and data on hydrogen resource gaps or surpluses. The total hydrogen supply is the sum of the new hydrogen replenishment flow rate and the PSA recovered hydrogen flow rate, while the total hydrogen consumption is the sum of hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, hydrogen carried by exhaust gas emissions, and dissolution losses.

[0008] Preferably, the intermediate storage tank buffer scheduling engine is used to perform the following operations based on the upstream inflow rate, the downstream outflow rate, and the current tank level data: Based on the real-time difference between the upstream inflow rate and the downstream outflow rate, the current tank level and its trend data are obtained by integrating the rate of change of liquid level over time. Based on tank level and its changing trend data, scheduling decisions are made according to the following strategies: Under normal operating conditions, when the downstream material intake flow rate equals the moving average of the upstream material intake flow rate and the tank level is within the normal range, the current operation is maintained; Under liquid accumulation warning conditions, when the tank level approaches the upper limit and the upstream capacity is greater than the downstream capacity, priority is given to determining whether the downstream section can increase its load. If it can increase its load, a scheduling instruction to increase the downstream material intake is generated; otherwise, a scheduling instruction to reduce the upstream load or activate the backup storage tank is generated; Under liquid discharge warning conditions, when the tank level approaches the lower limit and the downstream capacity is greater than the upstream capacity, priority is given to determining whether the upstream section can increase its load. If it can increase its load, a scheduling instruction to increase the upstream output rate is generated; otherwise, a scheduling instruction to reduce the downstream load or activate the backup feed source is generated; Under emergency operating conditions, when the tank level exceeds the upper limit threshold or falls below the lower limit threshold, a full-process collaborative emergency scheduling instruction is generated. With the objective of minimizing the sum of the absolute values ​​of the ratios of the load adjustment range of each section to its baseline load, and under the constraints of the load adjustment range of each section and the safe zone of the tank position, a rolling time-domain optimization method is adopted. Within the set optimization window, rolling optimization calculations are performed with a set decision step size to obtain the optimal load adjustment scheme.

[0009] Preferably, the end-to-end production scheduling optimization engine performs the following operations based on current equipment operating data, order data, and material balance data: The load rate per unit time of each section within the future scheduling window, the product grade switching time and switching sequence, the online regeneration time of the catalyst in the Fischer-Tropsch synthesis reactor, the catalyst replacement time of the hydroisomerization unit and the hydrocracking unit, and the fractionation tower cut-off point adjustment scheme are used as decision variables. A multi-objective optimization model is constructed with the following objectives: maximizing total product yield, minimizing unit product energy consumption, minimizing transition material ratio, and maximizing hydrogen utilization efficiency. The constraints are the upper and lower limits of load for each section, the safety constraints of storage tank liquid level, product quality, maintenance window constraints, order constraints, and catalyst life constraints. An improved NSGA-II algorithm is adopted, which iteratively solves the problem with a preset population size and a preset maximum number of iterations. The constraint dominance principle is introduced to handle the constraint conditions. The Pareto front solution set is obtained through iterative calculation, and the optimal production scheduling scheme is selected from the solution set.

[0010] Preferably, the real-time carbon emission calculation module is used to perform the following operations based on the material consumption data, energy consumption data, and preset emission factor database of each work section: Based on the product of the emission factors of each emission category and the corresponding activity data, the total carbon emissions of the entire process are calculated by summing them according to the mass balance method and then allocated to each section. The total carbon emissions of the entire process are the sum of the carbon footprint of raw materials, carbon emissions of public works, carbon emissions of process technology, and carbon emissions of transportation. The system receives DCS data from each work section and updates it on a rolling basis according to the set data acquisition cycle. It calculates and updates the instantaneous and cumulative carbon emission values ​​of the entire process according to the set accounting cycle, and automatically generates carbon emission reports according to the preset time cycle. The calculated carbon emission data per unit product is compared with a preset threshold, and different levels of carbon emission warnings are triggered based on the comparison results; the sudden increase in carbon emissions in a single work section is compared with a preset percentage, and trend warnings are triggered based on the comparison results, and inspection suggestions are output in conjunction with the comparison results.

[0011] Preferably, the full-process anomaly early warning function is used to perform the following operations based on the process parameters and material balance data of each section: The propagation early warning operation involves analyzing the downstream impact based on the material flow propagation path when an anomaly occurs in a certain section, thereby obtaining the impact time and degree data for each downstream section. Cascaded early warning operation: When the same disturbance triggers early warnings in two or more work sections at the same time, the multiple early warning messages are merged into a single cascaded early warning event for output. The root cause analysis operation uses Bayesian inference to calculate the root cause based on the full-process cause-effect graph model, and obtains the results of abnormal root cause localization.

[0012] Preferably, each digital twin model in the device model layer is used to perform the following operations based on the process parameters of the corresponding section: The Fischer-Tropsch synthesis reactor model uses the ASF product distribution model and the tube temperature field model. Based on the syngas feed parameters, the product distribution and reactor temperature field are calculated to obtain the product distribution data and reactor temperature distribution data of the Fischer-Tropsch synthesis section. The hydroisomerization unit model uses the LH kinetic model and isomer selectivity model. The isomerization reaction is calculated based on the feed composition and operating conditions to obtain isomerization product yield data and isomer selectivity data. The hydrocracking unit model adopts a lumped kinetic model segmented by carbon number. The cracking reaction is calculated based on the heavy distillate feed and operating conditions to obtain cracking conversion rate data and cracking product distribution data. The fractionation tower model uses the MESH equation and combines it with tray efficiency correction. Based on the feed composition, distillation calculations are performed to obtain the side stream outflow data and product cut-off point data. The blending tank model employs multi-component linear or nonlinear blending rules, performs blending calculations based on the properties and proportions of each component, and obtains the quality index data of the blended aviation kerosene product, including its freezing point, flash point, density, and distillation range.

[0013] Preferably, the intermediate storage tank includes a synthetic oil intermediate tank located between the Fischer-Tropsch synthesis section and the downstream hydrogenation unit, an intermediate product tank area located downstream of the fractionation tower, and a hydrogen buffer tank; the intermediate storage tank buffer scheduling engine makes scheduling decisions based on the collected intermediate storage tank operation data. The intermediate tank for synthetic oil and the intermediate product tank area each have a set buffer volume range, and the hydrogen buffer tank has a set volume range and operating pressure range. It is used to adjust the hydrogen buffer based on the hydrogen resource gap or surplus data calculated by the hydrogen balance network when the hydrogen supply and demand of the whole process is mismatched.

[0014] Preferably, the application service layer is used to perform the following operations according to the collaborative scheduling scheme: The entire process is online and optimized. Based on the Pareto frontier solution set output by the full-process production optimization engine, the production plan is visualized and processed to generate a production Gantt chart and data on the expected transition material quantity, yield, energy consumption and carbon emission curves. Dynamic collaborative scheduling operation generates scheduling instructions based on the optimal load adjustment scheme output by the intermediate storage tank buffer scheduling engine, monitors the liquid level trend of the storage tank in real time, and automatically generates load adjustment suggestions when the liquid level deviates from the normal range; Real-time carbon accounting operation generates carbon emission dashboards and carbon emission reports based on the total carbon emission data of the entire process calculated by the real-time carbon emission accounting module. The entire process of abnormal early warning operation pushes early warning information based on the affected time, affected degree and abnormal root cause location results output by the entire process abnormal early warning function; The human-computer interaction layer is used to monitor the entire process flow diagram, display the production scheduling Gantt chart, display the carbon footprint dashboard, display the collaborative scheduling parameter configuration interface, and display the early warning panel based on the visualization display data generated by the application service layer through a web graphical interface.

[0015] The present invention has achieved at least the following beneficial effects: 1. This invention achieves real-time simultaneous solution and global optimization of five processes through a material-energy-hydrogen three-network joint model, overcoming the technical defect that the sum of local optima is far lower than the global optimum due to the independent operation of each process in the traditional method, and realizing a significant improvement in the overall yield and energy consumption of the entire process.

[0016] 2. This invention upgrades intermediate storage tanks from passive storage to active resource scheduling. By combining a dynamic liquid level model with rolling time-domain optimization, it can predict tank level risks in advance and intervene proactively, significantly reducing tank limit events and unplanned shutdown recovery time. At the same time, through multi-objective optimized production scheduling, it can significantly reduce the proportion of transition materials and the switching time for different grades.

[0017] 3. This invention achieves real-time carbon emission accounting for the entire process based on the mass balance-emission factor method, improving the accounting response from offline mode to online second-level, significantly improving data reliability and authentication efficiency; at the same time, it realizes propagation early warning, cascading early warning and root cause analysis based on causal graph model and Bayesian inference, significantly improving the accuracy of fault diagnosis and reducing the false alarm rate.

[0018] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0019] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration: Figure 1 This is an architecture diagram of the multi-device collaborative digital twin and intelligent scheduling system for the entire process of bio-jet fuel in this embodiment of the invention; Figure 2 This is a flowchart illustrating the calculation process of the mechanism model in an embodiment of the present invention; Figure 3 This is a flowchart of the NSGA-II multi-objective genetic algorithm in an embodiment of the present invention. Detailed Implementation

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0021] This invention provides a multi-device collaborative digital twin and intelligent scheduling system for the entire process of bio-jet kerosene, referring to... Figure 1 ,include: The data perception layer is used to collect process parameters of the five sections and the tank area in real time through the DCS / SCADA system and transmit them to the device model layer and the collaborative scheduling layer. The device model layer includes five independent digital twin models corresponding to five work sections, which are used to perform material balance and energy balance calculations based on process parameters, obtain material balance data and energy balance data for each work section, and transmit them to the collaborative scheduling layer. The collaborative scheduling layer includes a material-energy-hydrogen network joint model, an intermediate storage tank buffer scheduling engine, a full-process production scheduling optimization engine, and a real-time carbon emission calculation module. It is used to solve the full-process material-energy-hydrogen network joint model, optimize the intermediate storage tank buffer scheduling strategy, optimize the multi-objective production scheduling within the future scheduling window, and calculate the full-process carbon emissions in real time based on process parameters, material balance data, and energy balance data, so as to obtain a collaborative scheduling scheme and transmit it to the application service layer. The application service layer is used to optimize the entire process of online production scheduling, perform dynamic collaborative scheduling, real-time carbon accounting, and full-process anomaly warning based on the collaborative scheduling scheme, and obtain visualized data to be transmitted to the human-computer interaction layer. The human-computer interaction layer is used to generate and display a graphical user interface based on the visualized data.

[0022] In this embodiment, the five sections are specifically the Fischer-Tropsch synthesis reactor section, the hydroisomerization unit section, the hydrocracking unit section, the fractionation tower section, and the blending tank section, which together constitute the core processing link of the entire bio-jet kerosene process. The data perception layer collects key process parameters of each section at a sampling frequency of 10–30 seconds through a distributed control system (DCS) and a monitoring and data acquisition system (SCADA), and transmits the collected data in real time to the unit model layer and the collaborative scheduling layer via industrial Ethernet. The digital twin models in the unit model layer adopt a combination of mechanistic modeling and data-driven approaches, performing online calculations of material and energy balances based on real-time process parameters, referencing… Figure 2 The calculation results are transmitted to the collaborative scheduling layer in a structured data format. As the core innovative layer of this invention, the collaborative scheduling layer solves the material, energy, and hydrogen networks simultaneously, while also integrating tank buffer scheduling, production optimization, and carbon emission accounting functions to achieve online collaborative optimization throughout the entire process. The application service layer transforms the optimization results into production scheduling plans, scheduling instructions, carbon accounting reports, and early warning information, and presents them to operators and managers through a web-based graphical interface via the human-computer interaction layer.

[0023] In this embodiment, the system adopts a five-layer architecture, consisting of a data perception layer, a device model layer, a collaborative scheduling layer, an application service layer, and a human-computer interaction layer, from bottom to top. Data interaction between layers occurs through standardized application programming interfaces (APIs), and inter-layer communication uses a publish-subscribe model to ensure data real-time performance and system scalability. The process parameters collected by the data sensing layer include: syngas flow rate (0–80,000 Nm³ / h), H₂ / CO molar ratio (1.5–2.5), reaction temperature (200–250 °C), reaction pressure (2.0–3.0 MPa), reactor tube temperatures (293 tubes), tail gas composition (CO, H₂, CO₂, CH₄, C₂–C₄), and three-phase separator level in the Fischer-Tropsch synthesis section; feed flow rate (5–30 t / h), reaction temperature (280–380 °C), hydrogen partial pressure (2–5 MPa), hydrogen-to-oil volume ratio (300–800), reactor outlet temperature, and online analysis data of product freezing point / flash point in the hydroisomerization section; and feed flow rate in the hydrocracking section. (2-15 t / h), reaction temperature (320-400℃), hydrogen partial pressure (3-8 MPa), online estimated conversion rate and cracked gas composition; feed flow rate, top / bottom temperature, flow rate of each side stream, reflux ratio (2-4), tower pressure and product cut-off point temperature of the fractionation tower section; inlet flow rate of each component in the blending tank section, tank level, online freezing point / flash point / density / distillation range analyzer data and circulation stirring status; intermediate tank level (sampling every 30-60 seconds), temperature and nitrogen sealing pressure of the storage tank area; steam (high / medium / low pressure) flow rate, circulating water flow rate and temperature, fuel gas flow rate and calorific value, PSA hydrogen purity (99.0%-99.9%) and flow rate of utilities.

[0024] In a preferred embodiment, the material-energy-hydrogen network joint model is used to perform the following operations based on the material balance data, energy balance data, and utility data of each process segment: Material transfer matrix operation is performed by establishing a directional material transfer diagram with five sections as nodes. The node set includes the Fischer-Tropsch synthesis reactor section, the hydroisomerization unit section, the hydrocracking unit section, the fractionation tower section, and the blending tank section. The edges in the diagram represent the material flow between each section. Material balance is performed based on the vector balance relationship that the total inlet material flow rate of each section is equal to the sum of the total outlet material flow rate and the material accumulation, to obtain real-time material allocation data and material accumulation data between each section. The energy supply and demand matrix operation is based on enthalpy balance. An energy supply and demand matrix is ​​established according to the energy supply relationship between each section. The matrix elements represent the available energy provided by one section to another. The matching relationship between temperature and heat grade is calculated and the solution is obtained with the goal of minimizing the total amount of externally supplied steam. The hydrogen balance network calculation takes the PSA unit as the hydrogen hub node. Based on real-time data such as the new hydrogen replenishment flow rate, PSA feed and product flow rates and purity, hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, exhaust gas emission, composition, and estimated dissolution losses, and constrained by the conservation of hydrogen mass, the calculation is performed on the basis that the total hydrogen supply equals the total hydrogen consumption. This yields the hydrogen distribution scheme among the various processes and data on hydrogen resource gaps or surpluses. The total hydrogen supply is the sum of the new hydrogen replenishment flow rate and the PSA recovered hydrogen flow rate, while the total hydrogen consumption is the sum of hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, hydrogen carried by exhaust gas emissions, and dissolution losses.

[0025] In this embodiment, the material transfer matrix is ​​a directed graph. Represents the set of nodes. These correspond to five stages: Fischer-Tropsch synthesis, hydroisomerization, hydrocracking, fractionation, and blending, respectively, with directed edges. This indicates the direction of material flow between different work sections. The main material flows include: Fischer-Tropsch synthesis products (… The full-fraction synthetic oil is transported from intermediate storage tank T1 to the hydroisomerization unit, where the heavy fraction (after isomerization) is processed. The product enters the hydrocracking unit, and the full fraction of the cracking products enters the fractionation tower. The jet fuel fraction separated in the fractionation tower... The product enters the blending tank, while the fractionation tower also produces naphtha. ) and diesel ( Sideline products. Material balance for each section is expressed in vector form: in This is the vector of total material flow at the inlet. This represents the total material flow vector at the export end. This represents the cumulative material volume vector within the unit. A set of material balance equations for the five work sections and three intermediate storage tanks is established to form a complete material transfer matrix. Material allocation data for each node is obtained through numerical solution.

[0026] In this embodiment, the energy supply and demand matrix The dimension is ,in Indicating work section To the work section Available energy provided (in MW). Matrix elements are determined based on the reaction heat effect and thermodynamic calculations for each stage: the heat of reaction exothermic by the Fischer-Tropsch synthesis is... The byproduct high-pressure steam (approximately 4.0 MPa, 400℃) can be used to drive a recirculating gas compressor and preheat the feed; the hydroisomerization reaction releases heat. Hydrocracking reaction releases heat Both processes produce medium-pressure steam (approximately 1.0 MPa, 200℃) as a byproduct, which can be used in the reboiler of the fractionation tower and the heat tracing system. The energy supply and demand matrix is ​​based on enthalpy balance. By calculating the temperature-heat grade matching relationship, the cascade utilization scheme of steam and heat streams between each stage is determined, and a linear programming solution is performed with the objective of minimizing the total amount of externally supplied steam. The constraints are the heat balance equations for each work section: in For the section The received external heat, For the section The reaction itself generates heat. For the section The output heat (including heat carried away by the product and heat from by-product steam). This represents heat loss. The steam replenishment amount and heat exchange scheme for each section are obtained by solving the problem.

[0027] In this embodiment, the hydrogen balance network model uses the PSA (Pressure Swing Adsorption) hydrogen purification unit as the hub node, and collects data in real time, including PSA feed tail gas flow rate (5000–10000 Nm³ / h), PSA product hydrogen purity (99.0%–99.9%), product flow rate, and hydrogen recovery rate (85%–90%). A hydrogen balance equation is established based on the law of conservation of hydrogen mass. The hydrogen consumption for hydroisomerization and hydrocracking is calculated online based on their respective reaction kinetics and feed composition. The amount of hydrogen carried in the exhaust gas is determined by exhaust gas flow and composition analyzers, and the dissolution loss is estimated based on a model of hydrogen solubility in the liquid phase under high pressure (such as Henry's Law). When the calculated total hydrogen supply is less than the total hydrogen consumption, the system outputs hydrogen resource gap data and triggers the hydrogen release operation of the hydrogen buffer tank or the hydrogen supply adjustment decision of the production optimization engine. When the total hydrogen supply is greater than the total hydrogen consumption, the system outputs hydrogen resource surplus data and triggers the hydrogen storage operation of the hydrogen buffer tank or adjusts the PSA operating parameters to improve the hydrogen recovery rate.

[0028] In a preferred embodiment, the intermediate tank buffer scheduling engine is used to perform the following operations based on the upstream inflow rate, the downstream outflow rate, and the current tank level data: Based on the real-time difference between the upstream inflow rate and the downstream outflow rate, the current tank level and its trend data are obtained by integrating the rate of change of liquid level over time. Based on tank level and its changing trend data, scheduling decisions are made according to the following strategies: Under normal operating conditions, when the downstream material intake flow rate equals the moving average of the upstream material intake flow rate and the tank level is within the normal range, the current operation is maintained; Under liquid accumulation warning conditions, when the tank level approaches the upper limit and the upstream capacity is greater than the downstream capacity, priority is given to determining whether the downstream section can increase its load. If it can increase its load, a scheduling instruction to increase the downstream material intake is generated; otherwise, a scheduling instruction to reduce the upstream load or activate the backup storage tank is generated; Under liquid discharge warning conditions, when the tank level approaches the lower limit and the downstream capacity is greater than the upstream capacity, priority is given to determining whether the upstream section can increase its load. If it can increase its load, a scheduling instruction to increase the upstream output rate is generated; otherwise, a scheduling instruction to reduce the downstream load or activate the backup feed source is generated; Under emergency operating conditions, when the tank level exceeds the upper limit threshold or falls below the lower limit threshold, a full-process collaborative emergency scheduling instruction is generated. With the objective of minimizing the sum of the absolute values ​​of the ratios of the load adjustment range of each section to its baseline load, and under the constraints of the load adjustment range of each section and the safe zone of the tank position, a rolling time-domain optimization method is adopted. Within the set optimization window, rolling optimization calculations are performed with a set decision step size to obtain the optimal load adjustment scheme.

[0029] In this embodiment, the intermediate storage tank includes a synthetic oil intermediate tank (T1) located between the Fischer-Tropsch synthesis section and the downstream hydrotreating unit. The intermediate product tank area (T2, naphtha) located downstream of the fractionation tower + Aviation kerosene blending components +Diesel ) and hydrogen buffer tank ( Operating pressure The dynamic model of the tank liquid level is as follows: in The liquid level inside the tank ( ), For upstream material flow ( ), For downstream material intake flow rate ( The above differential equation is discretized, and numerical integration is performed with a time step of 30 seconds to obtain the current tank level and its future trend. Under normal operating conditions, the downstream material intake flow rate is set as the moving average of the upstream material intake flow rate (window 30 minutes), and the tank level is... Natural fluctuations within the range of twice the tank capacity.

[0030] In this embodiment, the trigger condition for the liquid storage warning state is the tank level. Furthermore, the 30-minute moving average of the upstream inlet flow rate consistently exceeds the downstream outlet flow rate for more than 15 minutes. In this case, the system first checks if the current load rate of the downstream section is below its maximum allowable load (110%). If so, it generates a scheduling suggestion to increase the downstream outlet flow rate by 10%–20% and re-predicts the tank position 4 hours later. If the downstream has reached its maximum load, it generates an upstream load reduction instruction (the reduction should not exceed 5% to avoid significant fluctuations in product distribution), or activates a backup storage tank (if configured) and switches the feed path. The trigger condition for the liquid discharge warning state is the tank position. Furthermore, if the downstream feed rate consistently exceeds the upstream feed rate for more than 15 minutes, the system will first check if the current load rate of the upstream section is lower than its maximum allowable load. If so, it will generate a scheduling instruction to increase the upstream yield by 5% to 15% (based on the adjustment margin of the upstream reactor). Otherwise, it will generate a downstream load reduction instruction (with a reduction not exceeding 10%) or activate a backup feed source (such as replenishment with purchased synthetic oil). In emergency situations, when the tank level exceeds... or below At that time, the system generates a full-process collaborative emergency dispatch instruction, synchronously adjusts the load of multiple upstream and downstream sections, and triggers audible and visual alarms to prompt operators for manual confirmation.

[0031] In this embodiment, the buffer scheduling strategy takes minimizing the sum of the absolute values ​​of the ratios of the load adjustment magnitude of each work section to its baseline load as its objective function: in For the section Load adjustment amount ( or ), For the section The baseline load (design load). Constraints include that the load rate of each section must be within the allowable range (FT 60%~110%, HISO 50%~110%, HCR 40%~110%, DIST 50%~110%), and the tank level must be within the safe range. ), and the rate of load change in each section does not exceed the equipment safety limit (such as the temperature rise rate of the FT reactor). The rolling time-domain optimization method is adopted, with an optimization window of 4 to 8 hours and a decision step of 30 minutes. The calculation is rolled every 5 minutes. Based on the real-time updated material receiving and retrieving data, the rolling optimization is continuously performed. It can predict the risk of tank level reaching the limit 2 to 4 hours in advance and provide adjustment suggestions, effectively avoiding unplanned load fluctuations and shutdown events.

[0032] In a preferred embodiment, the end-to-end production scheduling optimization engine performs the following operations based on current equipment operating data, order data, and material balance data: The load rate per unit time of each section within the future scheduling window, the product grade switching time and switching sequence, the online regeneration time of the catalyst in the Fischer-Tropsch synthesis reactor, the catalyst replacement time of the hydroisomerization unit and the hydrocracking unit, and the fractionation tower cut-off point adjustment scheme are used as decision variables. A multi-objective optimization model is constructed with the following objectives: maximizing total product yield, minimizing unit product energy consumption, minimizing transition material ratio, and maximizing hydrogen utilization efficiency. The constraints are the upper and lower limits of load for each section, the safety constraints of storage tank liquid level, product quality, maintenance window constraints, order constraints, and catalyst life constraints. The improved NSGA-II algorithm is adopted, referring to Figure 3 The solution is obtained iteratively with a preset population size and a preset maximum number of iterations. The constraint dominance principle is introduced to handle the constraints. The Pareto front solution set is obtained through iterative calculation, and the optimal production scheduling scheme is selected from the solution set.

[0033] In this embodiment, the scheduling window is set to the next 1 to 4 weeks (in hours). Decision variables include continuous variables (hourly load rate of each work section). ,in Indicates the work section, (representing hours), discrete variables (product brand switching time) and switching sequence, catalyst replacement time (Fracturing tower cut-off point adjustment scheme). Specifically, the online regeneration cycle of the FT reactor catalyst is once every 2000 hours (each regeneration takes 12 hours), and the replacement cycle of the hydroisomerization and hydrocracking catalysts is once every 8000 hours (each replacement takes 48 hours). These time points are all decision variables and need to be optimized uniformly within the scheduling window.

[0034] In this embodiment, the multi-objective function is: (Maximize total product yield) (Minimize energy consumption per unit of product) (Minimize the proportion of transitional materials, where transitional materials refer to non-conforming products during the brand switching period) (Maximize hydrogen utilization efficiency, i.e., reduce the purchase of new hydrogen from external sources) In this embodiment, the constraints specifically include: upper and lower load limits for each work section (FT 60%–110%, HISO 50%–110%, HCR 40%–110%, DIST 50%–110%); and tank level safety constraints (…). Applicable to T1, T2 and hydrogen buffer tanks); Product quality constraints (jet kerosene freezing point) ,Flash point ,density According to ASTM D7566 standard); maintenance window constraints (production cannot be scheduled during planned unit downtime, and catalyst replacement cannot be scheduled during the order's emergency delivery period); order constraints (all grades of SAF products must be produced before the delivery date); catalyst life constraints (cumulative operating time of FT catalyst). Hours, cumulative operating time of hydroisomerization and hydrocracking catalysts Hour).

[0035] In this embodiment, an improved NSGA-II algorithm is used to solve the above multi-objective optimization model. The population size is set to 100-200, the maximum number of iterations is set to 200-300 generations, the simulated binary crossover (SBX) probability is 0.9, and the distribution index is... The polynomial mutation probability is 0.1, and the distribution exponent is... The improvement lies in introducing a remediation strategy based on the characteristics of the bio-jet fuel process: when individuals generated by genetic manipulation violate tank level constraints, a heuristic remediation operator is used to adjust the level to a safe range; when catalyst lifetime constraints are violated, the catalyst replacement time is automatically adjusted to meet the lifetime limit. The constraint dominance principle stipulates: when both individuals are infeasible, the individual with the smaller constraint violation is selected; when one is feasible and the other infeasible, the feasible individual dominates the infeasible individual; when both are feasible, they are sorted according to Pareto non-dominance relationships. The Pareto front solution set output by the algorithm contains 8–15 non-dominant production scheduling schemes, each corresponding to a set of decision variable values ​​and its objective function value. The scheduling decision-maker can select the final execution scheme from the solution set based on the current plant's actual priorities (such as urgent orders, energy costs, carbon quotas, etc.).

[0036] In a preferred embodiment, the real-time carbon emission calculation module is used to perform the following operations based on the material consumption data, energy consumption data, and a preset emission factor database for each work section: Based on the product of the emission factors of each emission category and the corresponding activity data, the total carbon emissions of the entire process are calculated by summing them according to the mass balance method and then allocated to each section. The total carbon emissions of the entire process are the sum of the carbon footprint of raw materials, carbon emissions of public works, carbon emissions of process technology, and carbon emissions of transportation. The system receives DCS data from each work section and updates it on a rolling basis according to the set data acquisition cycle. It calculates and updates the instantaneous and cumulative carbon emission values ​​of the entire process according to the set accounting cycle, and automatically generates carbon emission reports according to the preset time cycle. The calculated carbon emission data per unit product is compared with a preset threshold, and different levels of carbon emission warnings are triggered based on the comparison results; the sudden increase in carbon emissions in a single work section is compared with a preset percentage, and trend warnings are triggered based on the comparison results, and inspection suggestions are output in conjunction with the comparison results.

[0037] In this embodiment, the formula for calculating the total carbon emissions of the entire process is as follows: in, The carbon footprint of raw materials (emissions during syngas production and transportation) is calculated by multiplying the default emission factor from a life cycle database (such as Ecoinvent) by the syngas consumption, based on the syngas source (such as natural gas reforming or biomass gasification). Carbon emissions from public utilities are allocated by work section: in For the first Consumption of various public utilities (such as steam, electricity, circulating water, and fuel gas). This corresponds to the emission factor. The emission factor value is determined according to the latest standards, for example: power grid electricity. (China's 2024 power grid baseline), high-pressure steam (coal-fired boiler) ,natural gas Circulating water . Carbon emissions from the process include CO2 emissions from the Fischer-Tropsch synthesis reaction (based on the stoichiometric formula, approximately 1 mol of CO is converted as a byproduct). Emissions include molCO2, PSA exhaust combustion emissions (combustion of combustible components such as CH4 and H2 in the exhaust gas to generate CO2), and flare emissions. Carbon emissions from the transportation of raw materials and products, calculated using the CORSIA default path factor (e.g., emissions per ton-kilometer). Multiply by the estimated transport distance and volume.

[0038] In this embodiment, the data acquisition cycle is 30 seconds (synchronized with the DCS sampling frequency), and the calculation cycle is 5 minutes, that is, the instantaneous carbon emission values ​​of the entire process are updated on a rolling basis at 5-minute intervals. ) and cumulative value ( Each batch (each batch is switched to a different brand and lasts for about 3 to 5 hours) automatically generates a batch carbon emission report. At the same time, carbon emission reports for the corresponding period are automatically generated by shift (8 hours), day (24 hours), and month (720 hours), and the format conforms to CORSIA MRV (Monitoring, Reporting, Verification) requirements.

[0039] In this embodiment, the early warning function includes three threshold levels: when the carbon emissions per unit of product exceed the CORSIA default value ( ) 80% (i.e. When the energy consumption exceeds 100%, a yellow alert is triggered, prompting operators to pay attention to the current energy consumption and emissions; when the energy consumption exceeds 100% (i.e., ... When a red alert is triggered, operators are required to immediately check and take carbon reduction measures. The trigger condition for a trend alert is: the instantaneous carbon emission value of any section suddenly increases by more than 20% compared with its moving average value of the previous hour and lasts for more than 10 minutes. At this time, the system automatically outputs inspection suggestions, such as "Is the FT exhaust flare abnormally turned on?", "Is the PSA exhaust venting volume too large?", "Is steam consumption abnormally increased?", etc., to help operators quickly locate the source of the abnormality.

[0040] In a preferred embodiment, the full-process anomaly early warning function is used to perform the following operations based on the process parameters and material balance data of each section: The propagation early warning operation involves analyzing the downstream impact based on the material flow propagation path when an anomaly occurs in a certain section, thereby obtaining the impact time and degree data for each downstream section. Cascaded early warning operation: When the same disturbance triggers early warnings in two or more work sections at the same time, the multiple early warning messages are merged into a single cascaded early warning event for output. The root cause analysis operation uses Bayesian inference to calculate the root cause based on the full-process cause-effect graph model, and obtains the results of abnormal root cause localization.

[0041] In this embodiment, the propagation warning is based on a directed graph model to analyze the abnormal propagation path. The system pre-constructs a full-process causal graph, where nodes are key process variables for each stage (such as the FT reactor outlet temperature, hydroisomerization feed flow rate, fractionation tower top temperature, etc.), and directed edges are defined according to the material flow and energy flow directions (anomalies in upstream nodes affect downstream nodes along the material flow direction). When a single-variable or multi-variable anomaly occurs in a certain stage (such as the FT reactor single-tube temperature exceeding 250°C or the CO conversion rate decreasing by more than 5%), the system predicts the impact time and degree of impact on downstream stages along the material flow propagation path. For example, if the FT reactor temperature is too high, leading to a decrease in CO conversion rate, the system predicts a reduction in the hydroisomerization feed rate (with a lag time approximately equal to the residence time in intermediate storage tank T1, 8–16 hours), which in turn predicts a possible 10%–15% decrease in isomerization product yield, a reduction in cracking feed rate, a shift in fractionation tower feed composition, and changes in jet fuel blending composition. The system outputs the impact time (delay hours based on the current time) and the degree of impact (estimated shift of key variables) for each downstream stage.

[0042] In this embodiment, cascaded early warning operations are used to avoid redundant alarm information from multiple processes interfering with operator judgment. When the same disturbance (such as an abnormal temperature event in the same FT reactor) triggers early warnings in two or more processes simultaneously (such as an alarm in the FT process itself, a low flow rate alarm for hydroisomer feed, or a composition deviation alarm for the fractionation tower feed), the system merges these early warnings into a single cascaded early warning event, providing a unified event number, disturbance source, propagation path, and comprehensive impact assessment, rather than pushing out independent alarms for each process individually, thereby significantly reducing information redundancy.

[0043] In this embodiment, root cause analysis is based on a full-process causal graph model, using Bayesian inference to calculate the posterior probability of each possible root cause. The system pre-establishes a conditional probability table based on historical anomaly data and expert knowledge, for example... , When multiple alarms (such as "high freezing point of hydroisomerized product" + "low purity of PSA hydrogen" + "high CH4 content in FT tail gas") are detected simultaneously, the system uses these alarms as observational evidence, inputs them into a Bayesian network, and calculates the conditional probability of each potential root cause. The inference result is output as a root cause ranking list, for example: , , , The system automatically identifies the most probable root cause (low FT reactor temperature) as the diagnostic conclusion and recommends corresponding treatment measures (such as increasing the FT reactor inlet temperature or adjusting the H2 / CO ratio). After 12 months of actual operation, the accuracy of the full-process propagation early warning system has been verified. Root cause localization accuracy Compared to independent work sections, the false alarm rate of early warning has been reduced by 40%.

[0044] In a preferred embodiment, each digital twin model in the device model layer is used to perform the following operations based on the process parameters of the corresponding section: The Fischer-Tropsch synthesis reactor model uses the ASF product distribution model and the tube temperature field model. Based on the syngas feed parameters, the product distribution and reactor temperature field are calculated to obtain the product distribution data and reactor temperature distribution data of the Fischer-Tropsch synthesis section. The hydroisomerization unit model uses the LH kinetic model and isomer selectivity model. The isomerization reaction is calculated based on the feed composition and operating conditions to obtain isomerization product yield data and isomer selectivity data. The hydrocracking unit model adopts a lumped kinetic model segmented by carbon number. The cracking reaction is calculated based on the heavy distillate feed and operating conditions to obtain cracking conversion rate data and cracking product distribution data. The fractionation tower model uses the MESH equation and combines it with tray efficiency correction. Based on the feed composition, distillation calculations are performed to obtain the side stream outflow data and product cut-off point data. The blending tank model employs multi-component linear or nonlinear blending rules, performs blending calculations based on the properties and proportions of each component, and obtains the quality index data of the blended aviation kerosene product, including its freezing point, flash point, density, and distillation range.

[0045] In this embodiment, the Fischer-Tropsch synthesis reactor model is based on the Anderson-Schulz-Flory product distribution model and chain growth probability. Updated in real time based on actual operating conditions (temperature, pressure, H2 / CO ratio, catalyst activity factor). The product distribution calculation formula is: in For carbon number The hydrocarbon mass fraction. The tube temperature field model is a one-dimensional heterogeneous model, divided into 20 grids along the tube length, to calculate the radial temperature distribution. The data is then verified against the actual temperature measurement points of 293 tubes (each tube is equipped with a thermocouple), and the heat transfer coefficient and reaction rate constant are corrected in real time.

[0046] In this embodiment, the hydroisomerization unit model employs Langmuir-Hinshelwood bimolecular reaction kinetics. The rate constants for the parallel reactions of n-alkane isomerization and cracking are calculated based on the feed carbon number distribution and operating temperature, pressure, and hydrogen-to-oil ratio. The isomer selectivity model is defined as the molar ratio of isomerization products to cracking products, which decreases with increasing reaction temperature and increases with increasing hydrogen partial pressure. The model parameters are obtained from regression analysis of industrial unit calibration data and are automatically updated every 24 hours.

[0047] In this embodiment, the hydrocracking unit model adopts a five-lumped kinetic model, segmented according to the number of carbon atoms as follows: (gas), (Naphtha) (Jet fuel) (diesel fuel), (Unconverted heavy oil). The reaction rate constant matrix between each lumped group is: in Representing carbon number segments Pyrolysis into carbon fractions The rate constant, considering only downward cracking ( hour Each rate constant varies with temperature according to the Arrhenius equation. The model inputs are the feed composition and operating conditions (temperature). hydrogen partial pressure The output is the overall quality yield and conversion rate of each set.

[0048] In this embodiment, the fractionation tower model employs the MESH equation set (material balance, phase balance, normalization, and heat balance equations), with the number of trays set to 60 (actual floating valve trays). The feed position is dynamically adjusted between the 25th and 35th trays based on the feed composition. The tray efficiency correction coefficient is obtained by regressing actual operating data (flow rates and temperatures from each side stream) with model predictions, and is updated every 4 hours to ensure model accuracy. The model inputs are the feed flow rate and composition (from the cracking product distribution), and the outputs are the flow rates from each side stream (naphtha, jet fuel, diesel) and the cut-off point temperature.

[0049] In this embodiment, the blending tank model uses multi-component blending rules to calculate the key quality indicators of jet fuel products. Freezing point blending employs a non-linear rule based on the logarithmic linear relationship between the freezing point value of each component and its volume fraction; flash point blending uses a volume fraction weighted average method; density blending uses a mass fraction weighted average method; and distillation range (initial boiling point, 10%, 50%, 90% distillation temperature, and final boiling point) uses a volume fraction weighted average method. Taking density blending as an example, the calculation formula is: in For the adjusted density, Components mass fraction, Components The density of the product. The model input consists of the flow rate and measured properties of each blending component (jet fuel fraction, a small amount of naphtha or diesel component, depending on the product grade). The output is all quality indicators of the blended product, which are compared in real time with the measured values ​​from the online analyzer. The error exceeds [a certain threshold]. (Freezing point) or The harmonic parameters are automatically adjusted when the flash point is reached.

[0050] In a preferred embodiment, the intermediate storage tank includes a synthetic oil intermediate tank located between the Fischer-Tropsch synthesis section and the downstream hydrotreating unit, an intermediate product tank area located downstream of the fractionation tower, and a hydrogen buffer tank; the intermediate storage tank buffer scheduling engine makes scheduling decisions based on the collected intermediate storage tank operation data. The intermediate tank for synthetic oil and the intermediate product tank area each have a set buffer volume range, and the hydrogen buffer tank has a set volume range and operating pressure range. It is used to adjust the hydrogen buffer based on the hydrogen resource gap or surplus data calculated by the hydrogen balance network when the hydrogen supply and demand of the whole process is mismatched.

[0051] In this embodiment, there are two synthetic oil intermediate tanks (T1). The storage tanks are operated alternately, and the normal liquid level is controlled at 40%–60% (i.e., The residence time (volume / average flow rate) is 8–16 hours. When the capacity of the FT synthesis section and the downstream hydroisomerization section is mismatched, the T1 tank can be filled to a maximum of 85% ( Or drain to a minimum of 20% ( ), buffer capacity is This is equivalent to a buffer of approximately 13 hours for production capacity fluctuations. The intermediate product tank area (T2) includes naphtha storage tanks (…). ), aviation kerosene blending component storage tank ( ) and diesel storage tanks ( The side stream products from the fractionation tower are received separately, and the normal liquid level is controlled at 30% to 70% to coordinate the production-consumption rhythm between the fractionation tower and the blending tank.

[0052] In this embodiment, the volume of the hydrogen buffer tank is Operating pressure set to (Design Scope) The formula for calculating hydrogen storage capacity is: in Operating pressure (Pa). For the tank volume ( ), The compression factor is 0.95. The gas constant ( ), Temperature (K). , Under these conditions, the hydrogen storage capacity is approximately This is equivalent to approximately 2 hours of total hydrogen consumption in a hydroisomerization and hydrocracking unit (based on average hydrogen consumption). (Calculation). When the hydrogen balance network calculates a hydrogen resource deficit (such as PSA load reduction or FT tail gas production reduction), the hydrogen buffer tank automatically depressurizes and releases hydrogen (to a minimum of [value missing]). Release approximately Hydrogen gas is supplied to provide operators with 1.5 to 2 hours of response time; when hydrogen resources are abundant, the hydrogen buffer tank is pressurized to store hydrogen (up to a maximum of...). , more savings (Hydrogen), reducing hydrogen venting waste. This dynamic adjustment strategy is uniformly coordinated by the intermediate storage tank buffer scheduling engine and is executed in conjunction with the load adjustment schemes of upstream and downstream units.

[0053] In a preferred embodiment, the application service layer is configured to perform the following operations based on the collaborative scheduling scheme: The entire process is online and optimized. Based on the Pareto frontier solution set output by the full-process production optimization engine, the production plan is visualized and processed to generate a production Gantt chart and data on the expected transition material quantity, yield, energy consumption and carbon emission curves. Dynamic collaborative scheduling operation generates scheduling instructions based on the optimal load adjustment scheme output by the intermediate storage tank buffer scheduling engine, monitors the liquid level trend of the storage tank in real time, and automatically generates load adjustment suggestions when the liquid level deviates from the normal range; Real-time carbon accounting operation generates carbon emission dashboards and carbon emission reports based on the total carbon emission data of the entire process calculated by the real-time carbon emission accounting module. The entire process of abnormal early warning operation pushes early warning information based on the affected time, affected degree and abnormal root cause location results output by the entire process abnormal early warning function; The human-computer interaction layer is used to monitor the entire process flow diagram, display the production scheduling Gantt chart, display the carbon footprint dashboard, display the collaborative scheduling parameter configuration interface, and display the early warning panel based on the visualization display data generated by the application service layer through a web graphical interface.

[0054] In this embodiment, the entire process online scheduling optimization operation runs automatically once every 8 hours (synchronized with shift handover). The scheduling Gantt chart uses time as the horizontal axis (future 1-4 weeks) and five work sections as the vertical axis, using different colored blocks to represent the load rate (60%-110%), product grade (ordinary jet fuel / high-density jet fuel / cold region jet fuel) and catalyst regeneration / replacement period for each work section. The expected transition feed volume is superimposed on the switching time point in the form of a bar chart, and the yield, energy consumption, and carbon emission curves are displayed as line graphs to show the predicted trend. In the event of major disturbances (such as order changes, abnormal unit shutdowns), manual re-optimization can be triggered, and a new plan will be generated within 10 minutes.

[0055] In this embodiment, the dynamic collaborative scheduling operation updates the intermediate storage tank level prediction (for the next 4-8 hours) every 5 minutes. When the system predicts that the tank level will reach its limit within the next 2-4 hours ( or When a disturbance occurs, the system automatically generates load adjustment suggestions and notifies relevant operators via pop-up windows and SMS. After operator confirmation, the adjustment command is automatically sent to the DCS system of each section for execution. In disturbance response mode, when a unit suddenly reduces its load or shuts down, the system completes the impact propagation path analysis within 30 seconds and provides a full-process emergency dispatch plan (which units reduce load, by how much, how hydrogen is redistributed, and storage / discharge strategies). In recovery mode, after the disturbance is eliminated, the system provides the optimal path for full-process load recovery (gradually increasing load to avoid secondary disturbances), reducing the full-process recovery time from 24-48 hours in the traditional mode to 12-18 hours.

[0056] In this embodiment, the real-time carbon accounting dashboard displays the trend of total carbon emissions. Carbon emissions per unit of product ( Benchmark against CORSIA default value The system displays the carbon emission percentage of each production segment (pie chart) and the predicted carbon emission trend (for the next 4-8 hours). Warning information is displayed with color changes (yellow / red) and flashing when the carbon emission per unit of product exceeds the set threshold, and inspection suggestions are automatically generated. Carbon emission reports are automatically generated according to EUREDIIMRV requirements, including monthly carbon emission reports, annual greenhouse gas reports, and sustainability statements (optional fields are automatically filled), meeting all data requirements for CORSIA Qualified Fuels (CEF). All reports can be exported to PDF and Excel formats for easy submission to certification bodies for review.

[0057] In this embodiment, the human-machine interface adopts a B / S architecture (browser / server), allowing operators and managers to access the system via a web browser. The process flow diagram monitoring interface displays the real-time operating status of the five work sections and the tank area in dynamic animation, with key parameters updated in real time and abnormal areas highlighted. The production scheduling Gantt chart supports timeline zooming, drag-and-drop adjustment (manual fine-tuning), and scheme comparison. The carbon footprint dashboard allows users to drill down to detailed emission data for each work section. The collaborative scheduling parameter configuration interface allows users to set key parameters such as optimization window, decision step size, warning threshold, and emission factors. The warning panel arranges all current and historical warning events in chronological order, with each event including trigger time, work section, parameter name, alarm level, and related suggestions. The system interface supports switching between Chinese and English, and its responsive design adapts to desktop computers, tablets, and large-screen monitoring displays.

[0058] In this embodiment, the system was deployed in a 100,000-ton / year bio-jet fuel industrial plant. After 12 months of continuous operation and verification, significant technical effects were achieved: the total SAF yield of the entire process increased from 76%–78% in the traditional independent operation mode to 83.5%; unit product energy consumption decreased by 10.5%; product qualification rate increased from 85%–88% to 96.2%; the proportion of transition material for grade switching decreased from 5%–8% to 1.8%; the switching time was shortened from 6–12 hours to 3.5 hours; the response time for carbon emission accounting of the entire process was shortened from monthly / quarterly to less than 30 seconds; the recovery time for unplanned shutdowns was shortened from 24–48 hours to 14 hours; hydrogen utilization efficiency increased from 82%–85% to 91.5%; the number of storage tank limit events decreased from an average of 8–12 times per month to 1–2 times; the production scheduling plan generation time was shortened from 2–4 hours (manual) to less than 10 minutes; and the reliability of carbon emission data was reduced from manual reporting errors. Upgrade to automatic error calculation The above data verifies the technical effects of the preferred embodiments of the present invention.

[0059] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A biological aviation coal full-process multi-device collaborative digital twin and intelligent scheduling system, characterized in that, include: The data perception layer is used to collect process parameters of the five sections and the tank area in real time through the DCS / SCADA system and transmit them to the device model layer and the collaborative scheduling layer. The device model layer includes five independent digital twin models corresponding to five work sections, which are used to perform material balance and energy balance calculations based on process parameters, obtain material balance data and energy balance data for each work section, and transmit them to the collaborative scheduling layer. The collaborative scheduling layer includes a material-energy-hydrogen network joint model, an intermediate storage tank buffer scheduling engine, a full-process production scheduling optimization engine, and a real-time carbon emission accounting module; It is used to solve the material-energy-hydrogen network simultaneously throughout the entire process based on process parameters, material balance data and energy balance data, optimize the intermediate storage tank buffer scheduling strategy, optimize the multi-objective production scheduling within the future scheduling window and calculate the carbon emissions of the entire process in real time, obtain the collaborative scheduling scheme and transmit it to the application service layer. The application service layer is used to optimize the entire process of online production scheduling, perform dynamic collaborative scheduling, real-time carbon accounting, and full-process anomaly warning based on the collaborative scheduling scheme, and obtain visualized data to be transmitted to the human-computer interaction layer. The human-computer interaction layer is used to generate and display a graphical user interface based on the visualized data.

2. The bio-synthetic coal full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The material-energy-hydrogen network joint model is used to perform the following operations based on the material balance data, energy balance data, and utility data of each process section: Material transfer matrix operation is performed by establishing a directional material transfer diagram with five sections as nodes. The node set includes the Fischer-Tropsch synthesis reactor section, the hydroisomerization unit section, the hydrocracking unit section, the fractionation tower section, and the blending tank section. The edges in the diagram represent the material flow between each section. Material balance is performed based on the vector balance relationship that the total inlet material flow rate of each section is equal to the sum of the total outlet material flow rate and the material accumulation, to obtain real-time material allocation data and material accumulation data between each section. The energy supply and demand matrix operation is based on enthalpy balance. An energy supply and demand matrix is ​​established according to the energy supply relationship between each section. The matrix elements represent the available energy provided by one section to another. The matching relationship between temperature and heat grade is calculated and the solution is obtained with the goal of minimizing the total amount of externally supplied steam. The hydrogen balance network calculation takes the PSA unit as the hydrogen hub node. Based on real-time data such as the new hydrogen replenishment flow rate, PSA feed and product flow rates and purity, hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, exhaust gas emission, composition, and estimated dissolution losses, and constrained by the conservation of hydrogen mass, the calculation is performed on the basis that the total hydrogen supply equals the total hydrogen consumption. This yields the hydrogen distribution scheme among the various processes and data on hydrogen resource gaps or surpluses. The total hydrogen supply is the sum of the new hydrogen replenishment flow rate and the PSA recovered hydrogen flow rate, while the total hydrogen consumption is the sum of hydrogen consumption in hydroisomerization, hydrogen consumption in hydrocracking, hydrogen carried by exhaust gas emissions, and dissolution losses.

3. The bio-synthetic coal full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The intermediate storage tank buffer scheduling engine performs the following operations based on upstream inflow rate, downstream outflow rate, and current tank level data: Based on the real-time difference between the upstream inflow rate and the downstream outflow rate, the current tank level and its trend data are obtained by integrating the rate of change of liquid level over time. Based on tank level and its changing trend data, scheduling decisions are made according to the following strategies: Under normal operating conditions, when the downstream material intake flow rate equals the moving average of the upstream material intake flow rate and the tank level is within the normal range, the current operation is maintained; Under liquid accumulation warning conditions, when the tank level approaches the upper limit and the upstream capacity is greater than the downstream capacity, priority is given to determining whether the downstream section can increase its load. If it can increase its load, a scheduling instruction to increase the downstream material intake is generated; otherwise, a scheduling instruction to reduce the upstream load or activate the backup storage tank is generated; Under liquid discharge warning conditions, when the tank level approaches the lower limit and the downstream capacity is greater than the upstream capacity, priority is given to determining whether the upstream section can increase its load. If it can increase its load, a scheduling instruction to increase the upstream output rate is generated; otherwise, a scheduling instruction to reduce the downstream load or activate the backup feed source is generated; Under emergency operating conditions, when the tank level exceeds the upper limit threshold or falls below the lower limit threshold, a full-process collaborative emergency scheduling instruction is generated. With the objective of minimizing the sum of the absolute values ​​of the ratios of the load adjustment range of each section to its baseline load, and under the constraints of the load adjustment range of each section and the safe zone of the tank position, a rolling time-domain optimization method is adopted. Within the set optimization window, rolling optimization calculations are performed with a set decision step size to obtain the optimal load adjustment scheme.

4. The bio-synthetic coal full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The end-to-end production scheduling optimization engine performs the following operations based on current equipment operating data, order data, and material balance data: The load rate per unit time of each section within the future scheduling window, the product grade switching time and switching sequence, the online regeneration time of the catalyst in the Fischer-Tropsch synthesis reactor, the catalyst replacement time of the hydroisomerization unit and the hydrocracking unit, and the fractionation tower cut-off point adjustment scheme are used as decision variables. A multi-objective optimization model is constructed with the following objectives: maximizing total product yield, minimizing unit product energy consumption, minimizing transition material ratio, and maximizing hydrogen utilization efficiency. The constraints are the upper and lower limits of load for each section, the safety constraints of storage tank liquid level, product quality, maintenance window constraints, order constraints, and catalyst life constraints. An improved NSGA-II algorithm is adopted, which iteratively solves the problem with a preset population size and a preset maximum number of iterations. The constraint dominance principle is introduced to handle the constraint conditions. The Pareto front solution set is obtained through iterative calculation, and the optimal production scheduling scheme is selected from the solution set.

5. The bio-synthetic coal full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The real-time carbon emission calculation module is used to perform the following operations based on material consumption data, energy consumption data, and a preset emission factor database for each work section: Based on the product of the emission factors of each emission category and the corresponding activity data, the total carbon emissions of the entire process are calculated by summing them according to the mass balance method and then allocated to each section. The total carbon emissions of the entire process are the sum of the carbon footprint of raw materials, carbon emissions of public works, carbon emissions of process technology, and carbon emissions of transportation. The system receives DCS data from each work section and updates it on a rolling basis according to the set data acquisition cycle. It calculates and updates the instantaneous and cumulative carbon emission values ​​of the entire process according to the set accounting cycle, and automatically generates carbon emission reports according to the preset time cycle. The calculated carbon emission data per unit of product is compared with a preset threshold, and different levels of carbon emission warnings are triggered based on the comparison results. The sudden increase in carbon emissions in a single work section is compared with a preset percentage. Based on the comparison results, a trend warning is triggered, and inspection suggestions are output.

6. The bio-jet fuel full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The full-process anomaly warning function is used to perform the following operations based on the process parameters and material balance data of each section: The propagation early warning operation involves analyzing the downstream impact based on the material flow propagation path when an anomaly occurs in a certain section, thereby obtaining the impact time and degree data for each downstream section. Cascaded early warning operation: When the same disturbance triggers early warnings in two or more work sections at the same time, the multiple early warning messages are merged into a single cascaded early warning event for output. The root cause analysis operation uses Bayesian inference to calculate the root cause based on the full-process cause-effect graph model, and obtains the results of abnormal root cause localization.

7. The bio-jet fuel full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The digital twin models in the device model layer are used to perform the following operations based on the process parameters of the corresponding section: The Fischer-Tropsch synthesis reactor model uses the ASF product distribution model and the tube temperature field model. Based on the syngas feed parameters, the product distribution and reactor temperature field are calculated to obtain the product distribution data and reactor temperature distribution data of the Fischer-Tropsch synthesis section. The hydroisomerization unit model uses the LH kinetic model and isomer selectivity model. The isomerization reaction is calculated based on the feed composition and operating conditions to obtain isomerization product yield data and isomer selectivity data. The hydrocracking unit model adopts a lumped kinetic model segmented by carbon number. The cracking reaction is calculated based on the heavy distillate feed and operating conditions to obtain cracking conversion rate data and cracking product distribution data. The fractionation tower model uses the MESH equation and combines it with tray efficiency correction. Based on the feed composition, distillation calculations are performed to obtain the side stream outflow data and product cut-off point data. The blending tank model employs multi-component linear or nonlinear blending rules, performs blending calculations based on the properties and proportions of each component, and obtains the quality index data of the blended aviation kerosene product, including its freezing point, flash point, density, and distillation range.

8. The bio-jet fuel full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, Intermediate storage tanks include a synthetic oil intermediate tank located between the Fischer-Tropsch synthesis section and the downstream hydrogenation unit, an intermediate product tank area located downstream of the fractionation tower, and a hydrogen buffer tank; the intermediate storage tank buffer scheduling engine makes scheduling decisions based on the collected intermediate storage tank operating data. The intermediate tank for synthetic oil and the intermediate product tank area each have a set buffer volume range, and the hydrogen buffer tank has a set volume range and operating pressure range. It is used to adjust the hydrogen buffer based on the hydrogen resource gap or surplus data calculated by the hydrogen balance network when the hydrogen supply and demand of the whole process is mismatched.

9. The bio-jet fuel full-process multi-device collaborative digital twin and intelligent scheduling system according to claim 1, characterized in that, The application service layer is used to perform the following operations based on the collaborative scheduling scheme: The entire process is online and optimized. Based on the Pareto frontier solution set output by the full-process production optimization engine, the production plan is visualized and processed to generate a production Gantt chart and data on the expected transition material quantity, yield, energy consumption and carbon emission curves. Dynamic collaborative scheduling operation generates scheduling instructions based on the optimal load adjustment scheme output by the intermediate storage tank buffer scheduling engine, monitors the liquid level trend of the storage tank in real time, and automatically generates load adjustment suggestions when the liquid level deviates from the normal range; Real-time carbon accounting operation generates carbon emission dashboards and carbon emission reports based on the total carbon emission data of the entire process calculated by the real-time carbon emission accounting module. The entire process of abnormal early warning operation pushes early warning information based on the affected time, affected degree and abnormal root cause location results output by the entire process abnormal early warning function; The human-computer interaction layer is used to monitor the entire process flow diagram, display the production scheduling Gantt chart, display the carbon footprint dashboard, display the collaborative scheduling parameter configuration interface, and display the early warning panel based on the visualization display data generated by the application service layer through a web graphical interface.

Images