Continuous reforming device equipment fault prediction and maintenance optimization modeling method based on AvenPlus

Abstract

The invention relates to the technical field of petrochemical engineering, and discloses an AvenPlus-based continuous reforming device equipment fault prediction and maintenance optimization modeling method, which comprises the following steps of: establishing a continuous reforming device AvenPlus steady-state model: collecting device data covering feed composition, flow, temperature, pressure and process conditions of each reaction unit and separation unit, and establishing a continuous reforming device AvenPlus steady-state model; the method comprises the following steps of: selecting and connecting a reactor module, a distillation tower module, a heat exchanger module and a compressor module suitable for unit operation modules according to the technological process of a continuous reforming device in AspenPlus software, rotating a physical property method suitable for a continuous reforming device system from an AspenPlus physical property database, and establishing a device model framework; by establishing an accurate steady-state model of the continuous reforming device based on AvenPlus and combining an advanced data analysis method and a fault prediction model, the fault characteristic parameters of the equipment can be extracted more accurately, the running state of the equipment can be monitored in real time, the fault of the equipment can be predicted in advance, and a reliable basis is provided for equipment maintenance.

Description

Technical Field

[0001] This invention relates to the field of petrochemical technology, specifically to a method for predicting equipment failures and optimizing maintenance of continuous reforming units based on AspenPlus. Background Technology

[0002] Continuous reforming units are crucial production facilities for oil refining and petrochemical companies, producing high-octane gasoline blending components and aromatics. Their operational stability and efficiency directly impact the company's economic benefits and product quality. However, continuous reforming units are complex processes, comprising multiple reaction and separation units. The equipment operates under harsh conditions such as high temperature, high pressure, and hydrogen exposure for extended periods, making it prone to equipment failures, such as reactor catalyst deactivation, heat exchanger scaling, and compressor malfunctions. These failures not only lead to unplanned shutdowns and increased maintenance costs but also potentially cause safety accidents and environmental pollution.

[0003] Traditional equipment maintenance strategies primarily rely on periodic inspections. This approach lacks real-time monitoring and accurate assessment of the actual operating status of equipment, easily leading to over-maintenance or under-maintenance. With the development of computer technology and process simulation technology, using process simulation software to model and analyze continuous reforming units provides new avenues for equipment failure prediction and maintenance optimization. Aspen Petrochemical Plus, a professional software widely used in chemical process simulation, boasts a powerful physical property database, rich unit operation models, and an efficient solver, enabling accurate simulation of the steady-state operation of continuous reforming units. However, a systematic modeling method combining Aspen Petrochemical Plus software with continuous reforming unit equipment failure prediction and maintenance optimization still requires further development. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a method for predicting equipment failures and optimizing maintenance of continuous reforming units based on AspenPlus. This method has the advantages of accurate prediction and efficient maintenance, and solves the problems of over- or under-maintenance of traditional continuous reforming units and the difficulty in responding to sudden failures in advance.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for equipment fault prediction and maintenance optimization modeling of continuous reforming units based on AspenPlus, comprising the following steps:

[0008] I. Establishing the AspenPlus steady-state model of the continuous reforming unit:

[0009] Collect device data, including feed composition, flow rate, temperature, pressure, process conditions of each reaction unit and separation unit, and equipment size and material information;

[0010] In AspenPlus software, according to the continuous recombination unit process flow, select appropriate unit operation modules such as reactor module, distillation column module, heat exchanger module, and compressor module and connect them to build the unit model framework;

[0011] Rotate the property method of the appropriate continuous reforming system from the AspenPlus property database and input the property parameters of each component;

[0012] Initialize the built model;

[0013] The simulation results are compared and analyzed with the actual operating data of the device, and the model parameters are adjusted to make the simulation results match the actual data.

[0014] II. Extraction of equipment fault characteristic parameters:

[0015] Select key equipment in the continuous reforming unit;

[0016] For each type of critical equipment, analyze its possible failure modes;

[0017] Based on the equipment failure mode, select parameters that can reflect the equipment's operating status and failure characteristics;

[0018] The device's distributed control system and online monitoring instruments collect data on the selected characteristic parameters in real time, and preprocess the collected data.

[0019] III. Model-based Fault Prediction Methods:

[0020] Using preprocessed equipment feature parameter data and combined with machine learning algorithms, an equipment fault prediction model is established in the AspenPlus software platform.

[0021] The real-time collected equipment characteristic parameter data is input into the fault prediction model. The model calculates the failure probability of the equipment in real time and feeds the prediction results back to the AspenPlus simulation model. When the prediction result indicates that the equipment has a failure risk, an early warning signal is generated.

[0022] IV. Maintenance and Optimization Strategy Formulation:

[0023] Develop an equipment maintenance plan based on the fault prediction results and the importance level of the equipment;

[0024] Using the AspenPlus simulation model, the operating performance of the device under different scenarios is simulated and analyzed. When new signs of failure appear in the equipment, the failure risk of the equipment is reassessed and the maintenance strategy is adjusted.

[0025] Preferably, when establishing the AspenPlus steady-state model of the continuous reforming unit, the feed composition includes C5-C11 alkanes, cycloalkanes, and mixed aromatics; the reactor module includes a fixed-bed reactor; the distillation column module includes a pre-fractionation column, a depentanizer column, and an aromatics separation column; the heat exchanger module includes a shell-and-tube heat exchanger and a plate heat exchanger; and the Peng-Robinson equation is used as the property method.

[0026] Preferably, during the extraction of equipment fault characteristic parameters, the reaction temperature distribution is obtained by setting multiple temperature measurement points along the axial direction of the reforming reactor, the amount of catalyst carbon deposit is obtained by periodic sampling and analysis, the surface temperature of the furnace tube is measured by thermocouples installed on the outer wall of the furnace tube, the fuel gas consumption is measured by flow meters, the CO content in the flue gas is detected by online gas analyzers, the pressure difference between the inlet and outlet of the heat exchanger and the compressor exhaust pressure are measured by pressure sensors, the compressor vibration value is measured by vibration sensors, and the compressor oil temperature is measured by temperature sensors.

[0027] Preferably, in the model-based fault prediction method, the machine learning algorithm includes support vector machine and neural network, and the statistical analysis method includes multiple linear regression and principal component analysis.

[0028] Preferably, when formulating maintenance optimization strategies, for the maintenance of catalyst deactivation in reforming reactors, the AspenPlus simulation model is used to evaluate the impact of different catalyst regeneration processes and replacement schemes on the unit's aromatics yield and energy consumption indicators, and to determine the optimal scheme; for heating furnaces, the optimal cleaning cycle is determined through simulation based on the changing trends of furnace tube surface temperature and fuel gas consumption.

[0029] Preferably, when establishing the AspenPlus steady-state model of the continuous reforming unit, during the model calibration and verification process, if the error between the model results and the actual data exceeds a preset threshold, the kinetic parameters of the reaction unit are adjusted sequentially, and then the efficiency parameters of the separation unit and the heat transfer and mass transfer parameters of the equipment are adjusted in turn until the error meets the requirements.

[0030] Preferably, in the data preprocessing for extracting equipment fault characteristic parameters, missing data is supplemented using two methods: mean filling and interpolation based on adjacent time-time data; outliers are identified and removed using the 3σ criterion and box plots to ensure data integrity.

[0031] Preferably, in the model-based fault prediction method, the monitoring frequency is determined according to the importance of the equipment and the probability of fault occurrence. The monitoring frequency of critical equipment is no less than once every ten minutes, and the monitoring frequency of non-critical equipment can be set to once every thirty minutes.

[0032] Preferably, during the dynamic adjustment process of maintenance optimization strategy formulation, when the device experiences two situations, namely, a fluctuation in feed rate exceeding 10% and a change in the content of key components in the feed composition exceeding 8%, the fault prediction model is automatically retrained and the maintenance plan is re-evaluated.

[0033] Preferably, this method is applied to continuous reforming devices of different processing scales. For devices of different scales, the model can be quickly adapted in real time by adjusting the size parameters and processing capacity parameters of the equipment in the model.

[0034] (III) Beneficial Effects

[0035] Compared with existing technologies, this invention provides a method for predicting equipment failures and optimizing maintenance of continuous reforming units based on AspenPlus, which has the following beneficial effects:

[0036] 1. By establishing an accurate steady-state model of the continuous reforming unit based on AspenPlus, and combining it with advanced data analysis methods and fault prediction models, we can more accurately extract equipment fault characteristic parameters, monitor equipment operating status in real time, predict equipment faults in advance, and provide a reliable basis for equipment maintenance.

[0037] 2. Personalized maintenance plans are developed based on fault prediction results, and the maintenance schemes are optimized using the AspenPlus simulation model. This avoids the blindness of traditional periodic maintenance methods, reduces over-maintenance and under-maintenance, lowers maintenance costs, and improves equipment reliability and service life.

[0038] 3. By promptly identifying and addressing potential equipment failures, unplanned downtime was reduced, ensuring stable operation and improving production efficiency. Simultaneously, the optimized maintenance strategy helped reduce energy consumption, improve product quality and output, thereby enhancing the company's economic benefits. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the process for the continuous reforming device equipment fault prediction and maintenance optimization modeling method based on AspenPlus proposed in this invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1:

[0042] See attached document Figure 1 As shown, the method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units includes the following steps:

[0043] I. Establishing the AspenPlus steady-state model of the continuous reforming unit:

[0044] Collect device data, including feed composition, flow rate, temperature, pressure, process conditions of each reaction unit and separation unit, and equipment size and material information;

[0045] In AspenPlus software, according to the continuous recombination unit process flow, select appropriate unit operation modules such as reactor module, distillation column module, heat exchanger module, and compressor module and connect them to build the unit model framework;

[0046] Rotate the property method of the appropriate continuous reforming system from the AspenPlus property database and input the property parameters of each component;

[0047] The established model is initialized, including parameters such as convergence conditions for iterative calculation and initial values. The model is then run to solve the problem, and the simulation results of the composition, temperature, pressure and flow of each material under the current operating conditions are obtained.

[0048] The simulation results are compared and analyzed with the actual operating data of the device, and the model parameters are adjusted to make the simulation results match the actual data. The adjusted model parameters include reaction kinetic parameters and equipment efficiency parameters.

[0049] II. Extraction of equipment fault characteristic parameters:

[0050] Key equipment in the continuous reforming unit is selected, including reforming reactor, heater, and compressor, etc.

[0051] For each type of key equipment, analyze the possible failure modes, such as catalyst deactivation in reforming reactor leading to a decrease in reaction conversion rate, coking of furnace tubes in heating furnace causing a decrease in thermal efficiency, scaling in heat exchanger causing a decrease in heat transfer coefficient, and mechanical failure of compressor leading to insufficient exhaust volume.

[0052] Based on the equipment failure mode, parameters that can reflect the equipment's operating status and failure characteristics are selected. For example, for a reforming reactor, parameters such as reaction heat temperature distribution, product composition, and catalyst carbon deposition are selected; for a heating furnace, parameters such as furnace tube surface temperature, fuel gas consumption, and flue gas composition are selected; for a heat exchanger, parameters such as heat transfer temperature difference and pressure drop are selected; and for a compressor, parameters such as exhaust pressure, vibration value, and oil temperature are selected.

[0053] The device's distributed control system and online monitoring instruments collect data of the selected characteristic parameters in real time, and preprocess the collected data, including data cleaning and data normalization.

[0054] III. Model-based Fault Prediction Methods:

[0055] Using preprocessed equipment feature parameter data and combined with machine learning algorithms, an equipment fault prediction model is established in the AspenPlus software platform.

[0056] The real-time collected equipment characteristic parameter data is input into the fault prediction model. The model calculates the failure probability of the equipment in real time and feeds the prediction results back to the AspenPlus simulation model. When the prediction result indicates that the equipment has a failure risk, an early warning signal is generated.

[0057] IV. Maintenance and Optimization Strategy Formulation:

[0058] Based on the fault prediction results and the importance level of the equipment, formulate an equipment maintenance plan, including information such as maintenance time, maintenance content and maintenance personnel arrangement. For equipment with a high probability of failure or that has a significant impact on the operation of the unit, priority should be given to maintenance and repair work, and for equipment with a low probability of failure, the maintenance cycle should be appropriately extended.

[0059] Using the AspenPlus simulation model, the operating performance of the device under different scenarios is simulated and analyzed. When new signs of failure appear in the equipment, the failure risk of the equipment is reassessed and the maintenance strategy is adjusted.

[0060] Furthermore, when establishing the AspenPlus steady-state model for the continuous reforming unit, the feed composition includes C5-C11 alkanes, cycloalkanes, and mixed aromatics. The reactor module includes a fixed-bed reactor, the distillation column module includes a pre-fractionation column, a depentanizer column, and an aromatics separation column, and the heat exchanger module includes a shell-and-tube heat exchanger and a plate heat exchanger. The Peng-Robinson equation is used as the property method.

[0061] Furthermore, during the extraction of equipment fault characteristic parameters, the reaction temperature distribution is obtained by setting multiple temperature measurement points along the axial direction of the reforming reactor, the amount of catalyst carbon deposit is obtained by periodic sampling and analysis, the surface temperature of the furnace tube is measured by thermocouples installed on the outer wall of the furnace tube, the fuel gas consumption is measured by flow meters, the CO content in the flue gas is detected by online gas analyzers, the pressure difference between the inlet and outlet of the heat exchanger and the compressor exhaust pressure are measured by pressure sensors, the compressor vibration value is measured by vibration sensors, and the compressor oil temperature is measured by temperature sensors.

[0062] Furthermore, in the model-based fault prediction method, the machine learning algorithm includes support vector machine and neural network, and the statistical analysis method includes multiple linear regression and principal component analysis. For example, when constructing a catalyst deactivation prediction model for a reforming reactor using the neural network algorithm, reaction temperature, feed composition, running time, etc. are used as input variables, and catalyst activity is used as the output variable.

[0063] Furthermore, when formulating maintenance optimization strategies, for the maintenance of catalyst deactivation in reforming reactors, the AspenPlus simulation model is used to evaluate the impact of different catalyst regeneration processes and replacement schemes on the unit's aromatics yield and energy consumption indicators, and to determine the optimal scheme; for the heater, the optimal cleaning cycle is determined through simulation based on the changing trends of furnace tube surface temperature and fuel gas consumption.

[0064] Furthermore, during the establishment of the AspenPlus steady-state model for the continuous reforming unit, when the error between the model results and the actual data exceeds a preset threshold during the model calibration and verification process, the kinetic parameters of the reaction unit are adjusted sequentially, and then the efficiency parameters of the separation unit and the heat transfer and mass transfer parameters of the equipment are adjusted in turn until the error meets the requirements.

[0065] Furthermore, in the data preprocessing for extracting equipment fault characteristic parameters, missing data is supplemented using two methods: mean imputation and interpolation based on adjacent time-time data; outliers are identified and removed using the 3σ criterion and box plots to ensure data integrity.

[0066] Furthermore, in the model-based fault prediction method, the monitoring frequency is determined according to the importance of the equipment and the probability of fault occurrence. The monitoring frequency of critical equipment is no less than once every ten minutes, while the monitoring frequency of non-critical equipment can be set to once every thirty minutes.

[0067] Furthermore, during the dynamic adjustment process of maintenance optimization strategy formulation, when the device experiences two situations—feed fluctuations exceeding 10% and changes in the content of key components in the feed composition exceeding 8%—the fault prediction model is automatically retrained and the maintenance plan is re-evaluated.

[0068] Furthermore, this method is applied to continuous reforming units of different processing scales. For units of different scales, the model can be quickly adapted in real time by adjusting the size parameters and processing capacity parameters of the equipment in the model.

[0069] Equipment Failure Prediction and Maintenance Optimization of a Continuous Reforming Unit in an Oil Refinery

[0070] Establish AspenPlus steady-state model

[0071] Detailed design data and one year of actual operation data for a 1 million tons / year continuous reforming unit at a refinery were collected. This included the composition of the feed naphtha (mainly C5-C11 alkanes, cycloalkanes, and aromatics), flow rate (average 120 t / h), temperature (35℃), pressure (1.5 MPa), type of reforming reactor (moving bed reactor), dimensions (3m diameter, 15m height), number of trays in each distillation column (30 trays in the pre-fractionation column, 40 trays in the depentanizer column), and type of heat exchanger (shell-and-tube heat exchanger) and heat exchange area (500 m²). 2 ), compressor model (centrifugal compressor), rated displacement (5000m³ / h) 3 Parameters such as / h).

[0072] In AspenPlus software, following the unit's process flow, the following modules were added sequentially: feed stream, pre-fractionation tower, reforming reactor, depentanizer, a series of heat exchangers, and compressor. All modules were then correctly connected. The Peng-Robinson property method was selected, and the property parameters of each component of the feed naphtha were input. The model's convergence condition was set to a maximum of 50 iterations and a convergence accuracy of 0.001. The model was run, and the parameters were calibrated based on actual operating data. After multiple adjustments, the error between the model simulation results and the actual data was brought within an acceptable range; for example, the error between the simulated and actual values ​​of aromatics content in the reformed oil was less than 3%.

[0073] Equipment fault characteristic parameter extraction

[0074] The reforming reactor, heater, main heat exchanger, and compressor were selected as key equipment. For the reforming reactor, the reaction temperature distribution (five temperature measurement points were set along the reactor axis), the yield of aromatics in the products, and the amount of catalyst carbon deposition (analyzed through periodic sampling) were identified as fault characteristic parameters. For the heater, the furnace tube surface temperature (measured by thermocouples installed on the outer wall of the furnace tubes), fuel gas consumption (measured by flow meters), and CO content in the flue gas (detected by an online gas analyzer) were selected as characteristic parameters. For the heat exchanger, the heat transfer temperature difference (temperature difference between inlet and outlet streams) and pressure drop (pressure difference between inlet and outlet measured by pressure sensors) were selected as characteristic parameters. For the compressor, the exhaust pressure (measured by pressure sensors), vibration value (measured by vibration sensors), and oil temperature (measured by temperature sensors) were selected as characteristic parameters. Data for these characteristic parameters were collected every 15 minutes through the DCS system, and the data was cleaned and normalized.

[0075] Model-based fault prediction methods

[0076] Using a support vector machine algorithm, taking a reforming reactor as an example, data such as reaction temperature, feed composition, and operating time from the past year were used as training samples. Catalyst activity (obtained through laboratory analysis) was used as the target value. A catalyst deactivation prediction model was established using the Scikit-learn library in a Python environment. The model was integrated into the AspenPlus software platform. Real-time data such as the current reaction temperature and feed composition were obtained through the AspenPlus interface and input into the prediction model to calculate the current predicted catalyst activity and remaining service life. When the predicted catalyst activity falls below a set threshold (e.g., 0.8), a catalyst deactivation warning signal is issued.

[0077] Maintenance and optimization strategy formulation

[0078] Based on the fault prediction results, a maintenance plan is formulated after the catalyst deactivation warning signal is issued, arranging for catalyst regeneration or replacement of the reformer during the next planned shutdown of the unit. Using the AspenPlus simulation model, different catalyst regeneration processes and replacement schemes are simulated and analyzed to evaluate the changes in aromatics yield, energy consumption, and other indicators of the unit after regeneration or catalyst replacement. Through comparison, the optimal catalyst regeneration process is selected, resulting in a 3% increase in aromatics yield and a 5% reduction in unit energy consumption. Simultaneously, for the heater, a maintenance plan for regular heater tube cleaning is formulated based on the changing trends of furnace tube surface temperature and fuel gas consumption. Simulations determine the optimal cleaning cycle to be 3 months, which can improve the heater thermal efficiency by 8%.

[0079] Equipment Failure Prediction and Maintenance Optimization of a Continuous Reforming Unit in an Oil Refinery

[0080] Establish AspenPlus steady-state model

[0081] Detailed design data and one year of actual operation data for a 1 million tons / year continuous reforming unit at a refinery were collected. This included the composition of the feed naphtha (mainly C5-C11 alkanes, cycloalkanes, and aromatics), flow rate (average 120 t / h), temperature (35℃), pressure (1.5 MPa), type of reforming reactor (moving bed reactor), dimensions (3m diameter, 15m height), number of trays in each distillation column (30 trays in the pre-fractionation column, 40 trays in the depentanizer column), and type of heat exchanger (shell-and-tube heat exchanger) and heat exchange area (500 m²). 2 ), compressor model (centrifugal compressor), rated displacement (5000m³ / h) 3 Parameters such as / h).

[0082] In AspenPlus software, following the unit's process flow, the following modules were added sequentially: feed stream, pre-fractionation tower, reforming reactor, depentanizer, a series of heat exchangers, and compressor. All modules were then correctly connected. The Peng-Robinson property method was selected, and the property parameters of each component of the feed naphtha were input. The model's convergence condition was set to a maximum of 50 iterations and a convergence accuracy of 0.001. The model was run, and the parameters were calibrated based on actual operating data. After multiple adjustments, the error between the model simulation results and the actual data was brought within an acceptable range; for example, the error between the simulated and actual values ​​of aromatics content in the reformed oil was less than 3%.

[0083] Equipment fault characteristic parameter extraction

[0084] The reforming reactor, heater, main heat exchanger, and compressor were selected as key equipment. For the reforming reactor, the reaction temperature distribution (five temperature measurement points were set along the reactor axis), the yield of aromatics in the products, and the amount of catalyst carbon deposition (analyzed through periodic sampling) were identified as fault characteristic parameters. For the heater, the furnace tube surface temperature (measured by thermocouples installed on the outer wall of the furnace tubes), fuel gas consumption (measured by flow meters), and CO content in the flue gas (detected by an online gas analyzer) were selected as characteristic parameters. For the heat exchanger, the heat transfer temperature difference (temperature difference between inlet and outlet streams) and pressure drop (pressure difference between inlet and outlet measured by pressure sensors) were selected as characteristic parameters. For the compressor, the exhaust pressure (measured by pressure sensors), vibration value (measured by vibration sensors), and oil temperature (measured by temperature sensors) were selected as characteristic parameters. Data for these characteristic parameters were collected every 15 minutes through the DCS system, and the data was cleaned and normalized.

[0085] Model-based fault prediction methods

[0086] Using a support vector machine algorithm, taking a reforming reactor as an example, data such as reaction temperature, feed composition, and operating time from the past year were used as training samples. Catalyst activity (obtained through laboratory analysis) was used as the target value. A catalyst deactivation prediction model was established using the Scikit-learn library in a Python environment. The model was integrated into the AspenPlus software platform. Real-time data such as the current reaction temperature and feed composition were obtained through the AspenPlus interface and input into the prediction model to calculate the current predicted catalyst activity and remaining service life. When the predicted catalyst activity falls below a set threshold (e.g., 0.8), a catalyst deactivation warning signal is issued.

[0087] Maintenance and optimization strategy formulation

[0088] Based on the fault prediction results, a maintenance plan is formulated after the catalyst deactivation warning signal is issued, arranging for catalyst regeneration or replacement of the reformer during the next planned shutdown of the unit. Using the AspenPlus simulation model, different catalyst regeneration processes and replacement schemes are simulated and analyzed to evaluate changes in aromatics yield, energy consumption, and other indicators of the unit after regeneration or catalyst replacement. Through comparison, the optimal catalyst regeneration process is selected, increasing aromatics yield by 3% and reducing unit energy consumption by 5%. Simultaneously, for the heater, a maintenance plan for regular heater tube cleaning is formulated based on the changing trends of furnace tube surface temperature and fuel gas consumption. Simulations determine the optimal cleaning cycle to be 3 months, which can improve heater thermal efficiency by 8%.

[0089] It should be noted that the term "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0090] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting equipment failures and optimizing maintenance of continuous reforming units based on AspenPlus, characterized in that, Includes the following steps: I. Establishing the AspenPlus steady-state model of the continuous reforming unit: Collect device data, including feed composition, flow rate, temperature, pressure, process conditions of each reaction unit and separation unit, and equipment size and material information; In AspenPlus software, according to the continuous recombination unit process flow, select appropriate unit operation modules such as reactor module, distillation column module, heat exchanger module, and compressor module and connect them to build the unit model framework; Rotate the property method of the appropriate continuous reforming system from the AspenPlus property database and input the property parameters of each component; Initialize the built model; The simulation results are compared and analyzed with the actual operating data of the device, and the model parameters are adjusted to make the simulation results match the actual data. II. Extraction of equipment fault characteristic parameters: Select key equipment in the continuous reforming unit; For each type of critical equipment, analyze its possible failure modes; Based on the equipment failure mode, select parameters that can reflect the equipment's operating status and failure characteristics; The device's distributed control system and online monitoring instruments collect data on the selected characteristic parameters in real time, and preprocess the collected data. III. Model-based Fault Prediction Methods: Using preprocessed equipment feature parameter data and combined with machine learning algorithms, an equipment fault prediction model is established in the AspenPlus software platform. The real-time collected equipment characteristic parameter data is input into the fault prediction model. The model calculates the failure probability of the equipment in real time and feeds the prediction results back to the AspenPlus simulation model. When the prediction result indicates that the equipment has a failure risk, an early warning signal is generated. IV. Maintenance and Optimization Strategy Formulation: Develop an equipment maintenance plan based on the fault prediction results and the importance level of the equipment; Using the AspenPlus simulation model, the operating performance of the device under different scenarios is simulated and analyzed. When new signs of failure appear in the equipment, the failure risk of the equipment is reassessed and the maintenance strategy is adjusted.

2. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: When establishing the AspenPlus steady-state model for the continuous reforming unit, the feed composition includes C5-C11 alkanes, cycloalkanes, and mixed aromatics. The reactor module includes a fixed-bed reactor, the distillation column module includes a pre-fractionation column, a depentanizer column, and an aromatics separation column, and the heat exchanger module includes a shell-and-tube heat exchanger and a plate heat exchanger. The Peng-Robinson equation is used as the property method.

3. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: During the extraction of equipment fault characteristic parameters, the reaction temperature distribution is obtained by setting multiple temperature measurement points along the axial direction of the reforming reactor, the amount of catalyst carbon deposit is obtained by periodic sampling and analysis, the surface temperature of the furnace tube is measured by thermocouples installed on the outer wall of the furnace tube, the fuel gas consumption is measured by flow meters, the CO content in the flue gas is detected by online gas analyzers, the pressure difference between the inlet and outlet of the heat exchanger and the compressor exhaust pressure are measured by pressure sensors, the compressor vibration value is measured by vibration sensors, and the compressor oil temperature is measured by temperature sensors.

4. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: In the model-based fault prediction method, the machine learning algorithms include support vector machines and neural networks, and the statistical analysis methods include multiple linear regression and principal component analysis.

5. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: When formulating maintenance optimization strategies, for the maintenance of catalyst deactivation in reforming reactors, the AspenPlus simulation model is used to evaluate the impact of different catalyst regeneration processes and replacement schemes on the unit's aromatics yield and energy consumption indicators, and to determine the optimal scheme; for the heater, the optimal cleaning cycle is determined by simulation based on the changing trends of furnace tube surface temperature and fuel gas consumption.

6. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: When establishing the AspenPlus steady-state model of the continuous reforming unit, during the model calibration and verification process, if the error between the model results and the actual data exceeds the preset threshold, the kinetic parameters of the reaction unit are adjusted first, and then the efficiency parameters of the separation unit and the heat transfer and mass transfer parameters of the equipment are adjusted in sequence until the error meets the requirements.

7. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: In the data preprocessing for extracting equipment fault characteristic parameters, missing data is supplemented using two methods: mean imputation and interpolation based on adjacent time-time data; outliers are identified and removed using the 3σ criterion and box plots to ensure data integrity.

8. The method for fault prediction and maintenance optimization modeling of continuous reforming equipment based on AspenPlus according to claim 1, characterized in that: In model-based fault prediction methods, the monitoring frequency is determined based on the importance of the equipment and the probability of fault occurrence. The monitoring frequency for critical equipment is no less than once every ten minutes, while the monitoring frequency for non-critical equipment can be set to once every thirty minutes.

9. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: During the dynamic adjustment process of maintenance optimization strategy formulation, when the device experiences two situations, namely, a fluctuation in feed rate exceeding 10% and a change in the content of key components in the feed composition exceeding 8%, the fault prediction model will be automatically retrained and the maintenance plan will be re-evaluated.

10. The method for equipment fault prediction and maintenance optimization modeling based on AspenPlus for continuous reforming units according to claim 1, characterized in that: This method is applied to continuous reforming units of different processing scales. For units of different scales, the model can be quickly adapted in real time by adjusting the size parameters and processing capacity parameters of the equipment in the model.

Images