Dimethyl sulfoxide process simulation and fault early warning method and system based on digital twinning

Abstract

The invention discloses a dimethyl sulfoxide process simulation and fault early warning method and system based on digital twinning, and relates to the technical field of twinning simulation. The dimethyl sulfoxide process simulation and fault early warning method and system based on digital twinning comprises the following steps: S1, collecting process monitoring data in real time, and preprocessing the process monitoring data; s2, performing pressure fluctuation intensity measurement on the process monitoring data; s3, performing anomaly detection optimization on the process monitoring data; s4, integrating the process monitoring data, the pressure fluctuation intensity measurement result and the anomaly detection optimization result to execute anomaly risk assessment; and S5, realizing process data monitoring, abnormal detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interaction functions. The problems that a traditional method depends on a complex model, sensitivity setting is unreasonable, and it is difficult to timely and accurately detect the pressure abnormal point of the flash condenser, so that false alarms and missing alarms are increased are solved.

Description

Technical Field

[0001] The present invention relates to the field of twin simulation technology, and specifically to a dimethyl sulfoxide process simulation and fault warning method and system based on digital twins. Background Art

[0002] With the continuous advancement of digital technology and intelligent manufacturing, the application of digital twins and twin simulation technologies in industrial production is becoming increasingly extensive. They form highly synchronized virtual factories by mapping physical process flows with data models in real time, and realize dynamic visualization, remote monitoring, and intelligent analysis and management of process flows. They have become an important means to achieve precise process control, intelligent safety assurance, and sustainable optimization, helping the chemical and other process industries to accelerate their progress towards digitalization and intelligence.

[0003] For example, the invention patent with the announcement number CN118625743B discloses a steel bar processing control system and method based on digital twins, which involves the field of digital twin technology. It includes a data acquisition module, a monitoring module, a construction module, an operation module and a control module. By constructing a steel bar processing system, a preset construction library, a basic database and a sample database are established; OSG is used to establish a digital twin model of the steel bar processing system; the virtual-real linkage with the steel bar processing system is realized through the digital twin model; the digital twin model is used for simulation, and the movement of the digital twin model is driven by reading the motion data of each node; the node displacement data is read and saved in the twin database to realize real-time monitoring; a mapping model of the key process parameter factors of steel bar processing and the quality index coefficient of the processed steel bar is established through the regression model, and the quality index coefficient of steel bar processing under different process parameter factor combinations is obtained, and the optimal process parameter factor combination is found.

[0004] For example, patent publication CN118625772B discloses a hydropower station safety measurement and control system and method based on digital twins. The system includes several electromechanical equipment within the hydropower station, field sensors, inspection equipment, a data transmission system, a field control host, and a monitoring cloud platform. The inspection equipment is used to collect infrared thermal data and visual image data from the electromechanical equipment within the hydropower station. The monitoring cloud platform includes a digital twin. The monitoring cloud platform also includes a first diagnostic module and a second diagnostic module. The first diagnostic module is used to provide fault warnings for electromechanical equipment based on field sensor data. The second diagnostic module is used to perform secondary diagnosis of electromechanical equipment that generates fault warning signals based on inspection data and field sensor data. Digital twin technology enables comprehensive, real-time monitoring of the hydropower station. The first and second diagnostic modules enable efficient utilization of measurement and control data and accurate fault diagnosis, thereby improving the safety of the hydropower station.

[0005] Existing DMSO process monitoring and early warning methods primarily rely on threshold judgments based on a single parameter. These methods, also based on simple data monitoring and localized response control, suffer from insufficient real-time performance, sensitivity, and robustness. They fail to fully reflect the dynamic operating conditions and potential risks of condensers and other key equipment, making it difficult to efficiently identify and provide early warnings for abnormal fluctuations and multi-dimensional risks.

[0006] In response to the above problems, there is an urgent need for a dimethyl sulfoxide process simulation and fault warning method and system based on digital twins. Summary of the Invention

[0007] Technical problems solved

[0008] In response to the shortcomings of the existing technology, the present invention provides a dimethyl sulfoxide process simulation and fault warning method and system based on digital twins, which solves the problem that traditional methods rely on complex models and unreasonable sensitivity settings, making it difficult to detect abnormal pressure points in the flash condenser in a timely and accurate manner, thereby leading to an increase in false alarms and missed alarms.

[0009] Technical Solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: a dimethyl sulfoxide process simulation and fault warning method based on digital twins, comprising the following steps: S1: real-time collection of process monitoring data, and preprocessing of the process monitoring data; S2: pressure fluctuation intensity measurement of the process monitoring data, judging the stability of the condenser operating conditions based on the pressure fluctuation intensity measurement results, and triggering the corresponding condensate adjustment, tower kettle heating power adjustment and vacuum pump linkage control strategy; S3: anomaly detection optimization of the process monitoring data, judging the dynamic state of the condenser operating conditions based on the anomaly detection optimization results, and automatically taking adjustment measures for the condensate, tower kettle heating and vacuum pump; S4: performing anomaly risk assessment based on the comprehensive process monitoring data, pressure fluctuation intensity measurement results and anomaly detection optimization results, judging the process safety status based on the anomaly risk assessment results, and performing safety level prompts, sensitivity weight adjustments and historical data record analysis; S5: as an integrated interactive window, realizing process data monitoring, anomaly detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions.

[0011] Furthermore, the specific steps of real-time acquisition of process monitoring data and pre-processing of the process monitoring data are as follows: the process monitoring data includes: time index, condensing pressure, window points, window steps, sampling interval, number of algorithms and weighted anomaly intensity value; according to the time sequence position of the data points in the sliding window, a time index is generated inside the window; the condensing pressure is collected in real time through the condenser pressure sensor; through the system's built-in adaptive algorithm, combined with real-time data volatility analysis and process stage switching, the window length and window steps that adapt to the current working conditions are dynamically generated; the sampling interval of the data acquisition equipment is manually set by the operator according to the process characteristics interval; usually during the system deployment and optimization stage, engineers determine the number of algorithms based on process monitoring requirements; through the various anomaly detection algorithms in the system, weighted anomaly intensity values ​​are automatically generated based on the analysis results of real-time process monitoring data; during the data collection process, each data point is automatically given a timestamp to form time series data; in the preprocessing stage, equipment shutdown, maintenance and other non-production data are eliminated, sensor instantaneous drift and other anomalies are excluded, process monitoring data are standardized and normalized to eliminate dimensional differences between different parameters, and time synchronization operations for multi-dimensional data are performed to ensure the consistency of each process parameter in the time series.

[0012] Furthermore, the specific steps of measuring the pressure fluctuation intensity of process monitoring data are as follows: obtaining the time index, condensation pressure and window points; calculating the relative displacement of the difference between the time index of each data point in the sliding window and the window point number, multiplying it by the pressure exponential attenuation factor and taking the negative value to form an exponential attenuation power term, taking the value of the natural exponential function of the exponential attenuation power term to generate an exponential attenuation factor, and forming a time position attenuation weight; then subtracting the condensation pressure of each data point from the condensation average pressure in the window and squaring it to form a square deviation term, multiplying the square deviation term by the exponential attenuation factor at the time point to generate a weighted square difference, accumulating all weighted square differences in the sliding window to form a weighted square sum, and at the same time accumulating all exponential attenuation factors in the window to obtain a weighted sum, dividing the weighted square sum by the weight sum to form a normalized variance, and finally taking the square root to generate a value which is the pressure fluctuation intensity value.

[0013] Furthermore, the stability of the condenser working condition is judged according to the pressure fluctuation intensity measurement result, and the specific steps of triggering the corresponding condenser water regulation, tower kettle heating power regulation and vacuum pump linkage control strategy are as follows: real-time comparison of the pressure fluctuation intensity value and the fluctuation threshold, the fluctuation threshold includes the first-level fluctuation threshold and the second-level fluctuation threshold; when the pressure fluctuation intensity value is greater than or equal to the first-level fluctuation threshold, the automatic adjustment measures are immediately triggered, including increasing the condenser cooling water flow to quickly suppress pressure fluctuations, reducing the tower kettle heating power to reduce the heat load, and linking the fan and vacuum pump to start and stop to help stabilize the pressure, and starting a red high-risk alarm on the interface to prompt the operator to respond quickly, and at the same time generating a check of the condenser vacuum pump status and cooling water supply stability. operation suggestions and record relevant operation logs; when the pressure fluctuation intensity value is greater than or equal to the secondary fluctuation threshold and less than the primary fluctuation threshold, the condenser cooling water flow and tower kettle heating power are automatically fine-tuned to perform flexible adjustment. At the same time, the interface prompts the operator to pay close attention to the pressure fluctuation in yellow medium-risk, and generates real-time process optimization suggestions. If the medium fluctuation state lasts for more than thirty minutes, the sensitivity is automatically tightened, a higher sensitivity detection mode is started, and the fluctuation trend data is recorded for later analysis; when the pressure fluctuation intensity value is less than the secondary fluctuation threshold, the conventional monitoring mode is maintained, and only real-time data is recorded for process analysis. The operation interface prompts the process stable state in green, and the short-term fluctuation intensity increases, which can automatically switch to the medium fluctuation sensitivity detection mode.

[0014] Furthermore, the specific steps for optimizing abnormality detection of process monitoring data are as follows: obtaining the condensation pressure, window step number and sampling interval; dividing the condensation pressure difference at each adjacent time point in the sliding window by the sampling interval to obtain the pressure change rate, inputting the pressure change rate into the hyperbolic tangent function for nonlinear transformation and taking the absolute value, accumulating all transformed rate values ​​in the sliding window to form a cumulative sum, and the value formed by dividing the cumulative sum by the window step number is the pressure change rate value.

[0015] Furthermore, the dynamic state of the condenser working condition is judged based on the abnormality detection optimization result, and the specific steps of automatically taking adjustment measures for the condenser water, tower kettle heating and vacuum pump are as follows: real-time comparison of the pressure change rate value and the rate threshold, the rate threshold includes the first-level rate threshold and the second-level rate threshold; when the pressure change rate value is greater than or equal to the first-level rate threshold, immediately enter the automatic fast adjustment mode, automatically increase the condenser cooling water flow, reduce the tower kettle heating power, link the vacuum pump and fan start and stop, the interface highlights the red warning and emits sound and light prompts, and automatically generates safe operation suggestions to check the condenser water circulation status, confirm the stability of the vacuum pump working condition, and fully record the operation process and abnormal data to support later process optimization and safety traceability; when the pressure change rate value is greater than the second-level rate threshold and less than When the first-level rate threshold is reached, the automatic fine-tuning adjustment mode is executed to fine-tune the condensate valve opening and the tower kettle heating power. A yellow prompt on the interface reminds the operator to maintain key monitoring, and generates real-time adjustment suggestions to prompt a reasonable adjustment range and operation priority. If the medium-risk state lasts for more than thirty minutes, the sensitivity threshold will be automatically tightened and switched to the high-rate detection mode. At the same time, the fine-tuning process and detection results will be recorded for subsequent process parameter optimization; when the pressure change rate value is less than or equal to the second-level rate threshold, the automatic adjustment measures will not be triggered. Only a green prompt will be displayed on the interface to indicate that production is stable and the status is good. The real-time pressure change rate data will be continuously recorded. If the pressure rate fluctuation increases for a short time, it can be quickly switched to the medium-rate detection threshold to ensure safe production. All data will be continuously updated to support long-term optimization and safety trend analysis of twin simulation.

[0016] Furthermore, the specific steps for performing abnormal risk assessment based on the comprehensive process monitoring data, pressure fluctuation intensity measurement results and abnormality detection optimization results are as follows: obtaining the number of algorithms and weighted abnormality intensity values; inputting the weighted abnormality intensity values ​​into the exponentially weighted sliding average algorithm to calculate the smoothed weighted value, multiplying the smoothed weighted abnormality intensity values ​​of all detection algorithms by their respective sensitivity weight coefficients and adding them up to form a weighted smoothed sum, and at the same time adding all sensitivity weight coefficients to form a weighted sum, and finally dividing the weighted smoothed sum by the weighted sum to generate a value that is the abnormal risk assessment value.

[0017] Furthermore, the specific steps of judging the process safety status according to the abnormal risk assessment results, performing safety level prompts, sensitivity weight adjustments and historical data record analysis are as follows: real-time comparison of the abnormal risk assessment value and the risk threshold, the risk threshold includes the first-level risk threshold and the second-level risk threshold; when the abnormal risk assessment value is greater than or equal to the first-level risk threshold, immediately start the multi-dimensional safety protection mode, automatically increase the condenser cooling water flow, synchronously adjust the vacuum pump vacuum rate, temporarily limit the tower kettle heating power to the safe load lower limit, and link the shutdown signal preparation logic, trigger a red high-risk alarm and emit sound and light and SMS prompts, generate process safety operation guidance suggestions in real time, check the vacuum pump status, ensure sufficient condensate circulation, and fully record the operation process, response records and safety data to form a traceable safety file. ; When the abnormal risk assessment value is greater than the second-level risk threshold and less than the first-level risk threshold, the intelligent flexible adjustment mode is executed, the sensitivity weight coefficient is adaptively adjusted based on historical data and real-time trends, the condensate flow and the tower kettle heating power are fine-tuned, and the interface uses a yellow prompt to remind the operator to maintain monitoring focus, generate real-time safety adjustment suggestions and give priority to automated safety operations. If the medium-risk state lasts for more than thirty minutes, the sensitivity is automatically tightened and switched to the high-risk detection mode. At the same time, the adjustment and safety data are recorded for later optimization; when the abnormal risk assessment value is less than or equal to the second-level risk threshold, the automatic adjustment is not triggered, only a green prompt is displayed on the interface that the production is stable and the status is good, and the risk assessment value and historical data are continuously recorded. If the risk value increases in the short term, it will automatically switch to the medium-risk sensitivity detection mode to ensure safe production.

[0018] Furthermore, the integrated interactive window realizes process data monitoring, abnormal detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions. The specific steps are as follows: real-time display of dynamic curves and digital dashboards of process monitoring data, intuitive presentation of the data relationship of each process section in a dynamic animation manner, real-time display of detection results and abnormality scores, intuitive display of each alarm level using different colors and icons, the interface provides a sensitivity adjustment slider to support operators to fine-tune the sensitivity threshold according to actual working conditions, and real-time visualization of the impact of sensitivity changes on detection results, support for historical data query, display of alarm history records, response records and process fluctuation trends, and generation of a comparative analysis interface to assist in operation and maintenance summary and process optimization, while achieving multi-terminal adaptation, supporting large screens, PCs, Web terminals and other operating scenarios.

[0019] The second aspect of the present invention provides a dimethyl sulfoxide process simulation and fault warning system based on digital twins, comprising a real-time data acquisition module, a process twin simulation module, an anomaly detection and optimization module, a hierarchical alarm response module, and an interactive visual interface module, characterized in that: the real-time data acquisition module is used to collect process monitoring data in real time and pre-process the process monitoring data; the process twin simulation module is used to measure the pressure fluctuation intensity of the process monitoring data, judge the stability of the condenser working condition according to the pressure fluctuation intensity measurement result, and trigger the corresponding condensate water regulation, tower kettle heating power regulation and vacuum pump linkage control strategy; the anomaly detection and optimization module, It is used to optimize the anomaly detection of process monitoring data, judge the dynamic state of the condenser working condition based on the anomaly detection optimization results, and automatically take adjustment measures for the condenser water, tower kettle heating and vacuum pump; the hierarchical alarm response module is used to perform anomaly risk assessment based on the comprehensive process monitoring data, pressure fluctuation intensity measurement results and anomaly detection optimization results, judge the process safety status according to the anomaly risk assessment results, and perform safety level prompts, sensitivity weight adjustment and historical data record analysis; the interactive visual interface module is used as an integrated interactive window to realize process data monitoring, anomaly detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) The present invention effectively eliminates equipment maintenance, instantaneous drift and other invalid data through real-time acquisition and dynamic preprocessing of process monitoring data, achieves high-quality data input, significantly improves the accuracy and reliability of subsequent pressure fluctuation intensity measurement and abnormality detection results, and enhances the applicability of the system and its engineering promotion value.

[0023] (2) The present invention combines the sliding window weighted pressure fluctuation intensity measurement with dynamic sensitivity optimization to achieve high-sensitivity monitoring and adaptive response to condenser operating condition fluctuations, quickly triggering automated adjustment measures to ensure stable process operation, and significantly improving the safety of the process and the dynamic adjustment efficiency.

[0024] (3) The present invention integrates the results of multiple anomaly detection algorithms to conduct comprehensive anomaly risk assessment. The sensitivity weight can be dynamically adjusted to adapt to the risk level under different working conditions, complete hierarchical alarms and intelligent early warnings, reduce false alarms and missed alarms, and improve the system's intelligent and preventive operation and maintenance capabilities.

[0025] (4) The present invention uses an interactive visual interface to integrate data monitoring, abnormality detection result display and sensitivity fine-tuning functions, supporting operators to intuitively grasp the process status on multiple devices, realizing efficient human-machine interaction and multi-dimensional information visualization, and improving the intelligence level of operation and maintenance management and the reliability of digital operation and maintenance decision-making.

[0026] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of the dimethyl sulfoxide process simulation and fault warning method based on digital twins of the present invention;

[0028] Figure 2 This is a structural diagram of the dimethyl sulfoxide process simulation and fault warning system based on digital twins of the present invention;

[0029] Figure 3 It is a line graph of the pressure fluctuation intensity value of the present invention. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0031] See also Figure 1-Figure 3 , an embodiment of the present invention provides a technical solution: a dimethyl sulfoxide process simulation and fault warning method based on digital twins, comprising the following steps: S1: real-time collection of process monitoring data, and preprocessing of the process monitoring data; S2: pressure fluctuation intensity measurement of the process monitoring data, judging the stability of the condenser working condition based on the pressure fluctuation intensity measurement result, triggering the corresponding condensate adjustment, tower kettle heating power adjustment and vacuum pump linkage control strategy; S3: anomaly detection optimization of the process monitoring data, judging the dynamic state of the condenser working condition based on the anomaly detection optimization result, and automatically taking adjustment measures for the condensate, tower kettle heating and vacuum pump; S4: performing anomaly risk assessment based on the comprehensive process monitoring data, pressure fluctuation intensity measurement result and anomaly detection optimization result, judging the process safety status according to the anomaly risk assessment result, and performing safety level prompts, sensitivity weight adjustment and historical data record analysis; S5: serving as an integrated interactive window, realizing process data monitoring, anomaly detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions.

[0032] Specifically, the process monitoring data is collected in real time and preprocessed as follows: the process monitoring data includes: time index, condensing pressure, number of window points, number of window steps, sampling interval, number of algorithms and weighted anomaly intensity value, covering multi-dimensional parameters in the operation process of the condenser; according to the time sequence position of the data points in the sliding window, a time index is generated inside the window to ensure orderly connection of data; the condensing pressure is collected in real time through the condenser pressure sensor to ensure the accuracy and continuity of the pressure data; through the system's built-in adaptive algorithm, combined with real-time data volatility analysis and process stage switching, the window length and window step number that adapt to the current working conditions are dynamically generated; the operator manually sets the sampling interval of the data acquisition equipment according to the process characteristics to ensure the sampling frequency Reasonable; usually during the system deployment and optimization stage, engineers determine the number of algorithms based on process monitoring requirements to achieve flexible configuration of multiple algorithm combinations; through the various anomaly detection algorithms in the system, weighted anomaly intensity values ​​are automatically generated based on the analysis results of real-time process monitoring data to improve the accuracy of anomaly detection; during the data collection process, each data point is automatically assigned a timestamp to form time series data; in the preprocessing stage, equipment shutdown, maintenance and other non-production data are eliminated, sensor instantaneous drift and other anomalies are excluded, process monitoring data are standardized and normalized, dimensional differences between different parameters are eliminated, and time synchronization operations on multi-dimensional data are performed to ensure the consistency of each process parameter in the time series, ultimately providing a high-quality data foundation for subsequent simulation and early warning analysis.

[0033] In this implementation plan, based on the multi-dimensional process monitoring data collected in real time, dynamic sensitivity window configuration and flexible algorithm combination are used to eliminate non-production data and outliers, and standardization, normalization, time synchronization and other multiple preprocessing are performed to ensure the high quality, comparability and traceability of the data. Ultimately, a complete time-serialized data foundation is formed that can be directly used for process twin simulation and dynamic fault warning analysis, significantly improving the accuracy and real-time performance of subsequent process status monitoring and fault prediction.

[0034] Specifically, the specific steps for measuring the pressure fluctuation intensity of process monitoring data are as follows: obtain the time index, condensation pressure and window points to ensure that the data are arranged in time series order within the sliding window; obtain the pressure exponential decay factor through the adaptive fitting result of the dynamic characteristics change trend of the data within the sliding window, which usually ranges from 0.05 to 0.2; calculate the relative displacement of the difference between the time index and the window points of each data point in the sliding window, multiply it by the pressure exponential decay factor and take the negative value to form an exponential decay power term, which reflects the higher attention to the recent pressure data. ; The exponential decay power exponential term takes the value of the natural exponential function to generate an exponential decay factor, forming a weighted weight that decays over time; then the condensing pressure of each data point is subtracted from the average condensing pressure in the window and squared to form a square deviation term, and multiplied by the exponential decay factor to generate a weighted square difference; all weighted square differences in the sliding window are accumulated to form a weighted square sum, and all exponential decay factors are accumulated to obtain the weighted sum; the weighted square sum is divided by the weighted sum to form a normalized variance, and finally the square root is taken to obtain the pressure fluctuation intensity value, which is used to dynamically reflect the amplitude and potential risk of the condenser pressure fluctuation.

[0035] The specific calculation method of the pressure fluctuation intensity value is:

[0036]

[0037] Where S p (t) represents the pressure fluctuation intensity value, P i Indicates the condensing pressure, represents the condensation average pressure, N represents the number of window points, β represents the pressure exponential decay factor, and i represents the time index.

[0038] As shown in Table 1, this is a data table of pressure fluctuation intensity values ​​provided in an embodiment of the present application. In this embodiment, the condensing pressure sequence of sampling 1 is set to 102.1, 101.9, 102.0, 102.2, 101.8, the condensing average pressure is set to 102.00, and the number of window points is set to 5; the condensing pressure sequence of sampling 2 is set to 101.5, 102.8, 101.7, 102.9, 101.5, the condensing average pressure is set to 102.08, and the number of window points is set to 5; the condensing pressure sequence of sampling 3 is set to 1 00.0, 103.0, 99.5, 104.0, 99.0, the condensing average pressure is set to 101.10, and the number of window points is set to 5; the condensing pressure sequence of sampling 4 is set to 102.2, 102.1, 102.3, 102.0, 102.2, the condensing average pressure is set to 102.16, and the number of window points is set to 5; the condensing pressure sequence of sampling 5 is set to 101.5, 103.0, 100.0, 104.5, 99.0, the condensing average pressure is set to 101.60, and the number of window points is set to 5.

[0039] Table 1 Pressure fluctuation intensity value data table

[0040]

[0041]

[0042] like Figure 3 As shown, it is a line graph of the pressure fluctuation intensity value provided in the embodiment of the present application. According to the data in the table and the image, the set first-level fluctuation threshold is 0.8, and the second-level fluctuation threshold is 0.5. The distribution range of the pressure fluctuation intensity value ranges from 0.1038 to 2.0821, and significant peaks and troughs are shown in the image, which intuitively reflects the dynamic fluctuation characteristics of the condensation process state under different sampling numbers. The pressure fluctuation intensity values ​​of sampling 3 and sampling 5 are relatively high, reaching 2.0632 and 2.0821 respectively, exceeding the first-level threshold, indicating that the condenser pressure fluctuation is significant at this stage and high-risk response measures need to be triggered in time. The value of sampling 2 is 0.6360, which is between the second-level and first-level thresholds, indicating a medium-risk state, suggesting that flexible adjustment is needed to balance fluctuations and energy efficiency. The pressure fluctuation intensity values ​​of sampling 1 and sampling 4 are relatively low, at 0.1478 and 0.1038 respectively, which are far below the second-level threshold, indicating that the condenser operating conditions are stable during these sampling stages. The overall broken line trend clearly shows the distribution pattern of pressure fluctuation intensity and risk level changes in each stage of the condensation process, providing important data support and visualization basis for subsequent process simulation and safety early warning strategies.

[0043] In this implementation scheme, through a sliding window weighted method based on time series, combined with a comprehensive analysis of the dynamic changes of condensing pressure and the exponential decay factor, the core indicators reflecting the stability of process conditions and the pressure fluctuation risk level are comprehensively and dynamically extracted, and finally a quantifiable pressure fluctuation intensity value is formed, which provides a high-precision dynamic feature foundation for subsequent process twin simulation, intelligent adjustment and fault warning, and significantly enhances the system's sensitive perception of the safety risks of the condensing process.

[0044] Specifically, the stability of the condenser operating condition is judged according to the pressure fluctuation intensity measurement results, and the corresponding condensate water regulation, tower kettle heating power regulation and vacuum pump linkage control strategy is triggered. The specific steps are as follows: real-time comparison of the pressure fluctuation intensity value and the fluctuation threshold, the fluctuation threshold includes the first-level fluctuation threshold and the second-level fluctuation threshold, to ensure that the sensitivity matches the fluctuation characteristics of different working conditions; when the pressure fluctuation intensity value is greater than or equal to the first-level fluctuation threshold, the automatic adjustment measures are immediately triggered, including rapidly increasing the condenser cooling water flow to suppress high-amplitude pressure fluctuations, reducing the tower kettle heating power to reduce the heat load, and at the same time linking the start and stop of the fan and vacuum pump to assist in stabilizing the pressure. The interface starts a red high-risk alarm to prompt the operator to respond quickly, and simultaneously generates operational suggestions for checking the status of the condenser vacuum pump and the stability of the cooling water supply. All response data are recorded in real time to form a traceable operation log. ; When the pressure fluctuation intensity value is greater than or equal to the secondary fluctuation threshold and less than the primary fluctuation threshold, the condenser cooling water flow and the tower kettle heating power are automatically fine-tuned, and flexible adjustment is performed to balance energy efficiency and safety. The interface prompts the operator to pay attention with a yellow medium-risk indicator, and generates process optimization suggestions in real time for reference. If the medium fluctuation state lasts for more than thirty minutes, the system will automatically tighten the sensitivity threshold and switch to a higher sensitivity detection mode to ensure that potential risks are captured more quickly and record fluctuation trend data to support later process optimization; when the pressure fluctuation intensity value is less than the secondary fluctuation threshold, the normal monitoring mode is maintained, and automatic adjustment is not triggered. Only real-time data is recorded for subsequent process analysis. The interface prompts in green that the process is in a stable state. If the short-term fluctuation intensity increases, it can be quickly switched to the medium fluctuation sensitivity detection mode to ensure process safety and flexible switching of monitoring sensitivity.

[0045] In this implementation plan, combined with the real-time monitoring results of the pressure fluctuation intensity value, different level thresholds are dynamically matched, and the automatic adjustment mode of the condenser cooling water flow, the tower kettle heating power and the vacuum pump is flexibly triggered, forming a fast, flexible and efficient dynamic control mechanism. At the same time, the operating risk level is prompted in a multi-dimensional visual manner on the operation interface, process optimization and safe operation suggestions are generated in real time, and fluctuation data and response logs are recorded, which significantly improves the sensitivity, stability and safety of the system when facing pressure fluctuation risks.

[0046] Specifically, the specific steps for optimizing anomaly detection of process monitoring data are as follows: obtain the condensing pressure, window step number and sampling interval to ensure that the data in the sliding window is dynamically updated and the time series is complete; divide the condensing pressure difference at each adjacent time point in the sliding window by the sampling interval to obtain the pressure change rate that reflects the rate of change of the operating condition, reflecting the dynamic sensitivity of the process operation; input the pressure change rate into the hyperbolic tangent function for nonlinear transformation and take the absolute value to suppress noise amplification and enhance the sensitivity to anomalies; accumulate all transformed rate values ​​in the sliding window to form the cumulative sum of the overall fluctuation intensity, and the value formed by dividing the cumulative sum by the window step number is the pressure change rate value, which serves as the key basis for subsequent dynamic operating condition judgment and sensitivity optimization.

[0047] The specific calculation method of the pressure change rate value is:

[0048]

[0049] Where V p (t) represents the pressure change rate value, M represents the window step number, P i Indicates the condensing pressure, P i-1 It represents the condensation pressure of the previous window, and Δt represents the sampling interval.

[0050] In this implementation scheme, through dynamic analysis of the condensation pressure data, a sliding window combined with hyperbolic tangent nonlinear transformation is adopted to accurately extract the pressure change rate value reflecting the dynamic fluctuations of the process, forming sensitivity optimization and dynamic monitoring of changes in the condensation working conditions, which significantly improves the system's rapid response capability and comprehensive recognition capability to instantaneous abnormal fluctuations.

[0051] Specifically, based on the abnormality detection optimization results, the dynamic state of the condenser working condition is judged, and the adjustment measures of the condenser water, tower kettle heating and vacuum pump are automatically carried out. The specific steps are as follows: real-time comparison of the pressure change rate value and the rate threshold, the rate threshold includes the first-level rate threshold and the second-level rate threshold, to ensure that the sensitivity covers the risk changes under different working conditions; when the pressure change rate value is greater than or equal to the first-level rate threshold, it immediately enters the automatic rapid adjustment mode, automatically increases the condenser cooling water flow to enhance the condensation effect, reduces the tower kettle heating power to reduce the heat load, and links the vacuum pump and fan to start and stop at the same time. The interface highlights the red warning and emits sound and light prompts to remind the operator to respond quickly, and generates safe operation suggestions, prompting to check the condenser water circulation status, confirm the key points of the vacuum pump working condition stability, and fully record the operation process and abnormal data to form a traceable safety response file; when the pressure changes When the pressure change rate value is greater than the secondary rate threshold and less than the primary rate threshold, the automatic fine-tuning adjustment mode is executed to fine-tune the condensate valve opening and the tower kettle heating power to balance fluctuations and energy efficiency. A yellow prompt on the interface reminds the operator to maintain key monitoring and generates real-time adjustment suggestions to indicate a reasonable operating range and priority. If the medium-risk state lasts for more than thirty minutes, the sensitivity threshold is automatically tightened and switched to the high-rate detection mode. At the same time, the fine-tuning process and detection results are recorded to support later process optimization; when the pressure change rate value is less than or equal to the secondary rate threshold, the automatic adjustment measures are not triggered. Only a green prompt will appear on the interface to indicate that production is stable and in good condition. The real-time pressure change rate data is continuously recorded. If the short-term pressure rate fluctuation increases, it can be quickly switched to the medium-rate detection threshold to ensure safe production. All data are continuously updated to support long-term optimization and safety trend analysis of twin simulation.

[0052] In this implementation plan, by comparing the pressure change rate value with the classification threshold in real time, the condenser operating status is dynamically identified and the multi-level automatic adjustment mode is flexibly triggered. It can not only quickly respond and automatically adjust key process parameters under high-risk conditions, but also flexibly fine-tune and maintain stability under medium-risk and low-risk conditions. The operation interface synchronizes visual alarms and operation suggestions to ensure the real-time and traceability of data, thereby significantly improving the safety, intelligence and dynamic adaptability of the condensation process.

[0053] Specifically, the specific steps for performing anomaly risk assessment based on comprehensive process monitoring data, pressure fluctuation intensity measurement results and anomaly detection optimization results are as follows: obtain the number of algorithms and weighted anomaly intensity values ​​to ensure that the sensitivity differences and operating conditions of different algorithms are considered in the multi-algorithm fusion analysis; input the weighted anomaly intensity value into the exponentially weighted moving average algorithm for smoothing to weaken the interference of short-term fluctuations on risk judgment, while maintaining a keen response to major fluctuations; obtain the sensitivity weight coefficient by normalizing the output credibility scores of different anomaly detection algorithms, which usually ranges from 0.1 to 1.0; multiply the smoothed weighted anomaly intensity values ​​of all detection algorithms by their respective sensitivity weight coefficients, and perform comprehensive weighted accumulation to form a weighted smoothed sum reflecting the overall process status; at the same time, accumulate all sensitivity weight coefficients to obtain the final weighted sum; finally, divide the weighted smoothed sum by the weighted sum, and the generated value is the anomaly risk assessment value, which is used to subsequently trigger graded alarms and operation strategies to ensure the comprehensiveness and dynamic adaptability of process monitoring.

[0054] The specific calculation method of abnormal risk assessment value is:

[0055]

[0056] Where R s (t) represents the abnormal risk assessment value, K represents the number of algorithms, and w k Represents the sensitivity weight coefficient, S k (t) 1.5 Represents the weighted anomaly intensity value.

[0057] In this implementation plan, by integrating process monitoring data, pressure fluctuation intensity measurement results and various anomaly detection optimization results, and comprehensively utilizing the exponentially weighted moving average algorithm and sensitivity weight adjustment mechanism, a dynamic, smooth and sensitivity-adaptive anomaly risk assessment value is formed, which realizes accurate identification and real-time response to comprehensive process risks, provides a reliable risk classification basis for subsequent graded alarms, automated adjustments and operation suggestions, and significantly improves the intelligence level of process safety early warning.

[0058] Specifically, the process safety status is judged according to the abnormal risk assessment results, and the safety level prompt, sensitivity weight adjustment and historical data record analysis are carried out. The specific steps are as follows: real-time comparison of the abnormal risk assessment value and the risk threshold, the risk threshold includes the first-level risk threshold and the second-level risk threshold, to ensure that the risk judgment accuracy covers different levels of working conditions; when the abnormal risk assessment value is greater than or equal to the first-level risk threshold, immediately start the multi-dimensional safety protection mode, automatically increase the condenser cooling water flow, quickly suppress potential abnormalities, dynamically adjust the vacuum pump vacuum rate, temporarily limit the tower kettle heating power to the lower limit of the safe load, link the shutdown signal preparation logic, trigger the red high-risk alarm and issue sound and light and SMS prompts, the interface generates safety operation suggestions such as checking the vacuum pump status, confirming the stability of the cooling water supply, and fully record all operation processes and response information to form a traceable safety file; when the abnormal risk assessment value is greater than or equal to the first-level risk threshold, the multi-dimensional safety protection mode is immediately started, the condenser cooling water flow is automatically increased, the potential abnormalities are quickly suppressed, the vacuum pump vacuum rate is dynamically adjusted, the tower kettle heating power is temporarily limited to the lower limit of the safe load, the shutdown signal preparation logic is linked, the red high-risk alarm is triggered, and sound and light and SMS prompts are issued, and the interface generates safety operation suggestions such as checking the vacuum pump status and confirming the stability of the cooling water supply. All operation processes and response information are fully recorded to form a traceable safety file; when the abnormal risk assessment When the value is greater than the second-level risk threshold and less than the first-level risk threshold, the intelligent flexible adjustment mode is executed, and the sensitivity weight coefficient is adaptively adjusted based on historical data and real-time trends, and the condensate flow and tower kettle heating power are fine-tuned. The interface uses a yellow prompt to remind the operator to focus on monitoring, and generates safety adjustment suggestions in real time and recommends automated safety operations as a priority. If the medium-risk state lasts for more than thirty minutes, the sensitivity is automatically tightened and switched to the high-risk detection mode. At the same time, all adjustments and safety data are fully recorded for later process optimization and safety analysis; when the abnormal risk assessment value is less than or equal to the second-level risk threshold, the automatic adjustment is not triggered, and the interface prompts green that the production is stable and in good condition. Only the risk assessment value and historical data are continuously recorded. If the short-term risk value increases, it can be quickly switched to the medium-risk sensitivity detection mode to ensure the safety and stability of the process operation and the reliability of intelligent dynamic monitoring.

[0059] In this implementation plan, by comparing the abnormal risk assessment value with the graded risk threshold in real time, high-risk, medium-risk and low-risk states can be sensitively distinguished, and multi-dimensional safety protection modes and intelligent flexible adjustment measures can be automatically triggered accordingly. The condenser cooling water flow, tower kettle heating power, vacuum pump and other equipment parameters can be dynamically adjusted to complete the visual prompt of the process safety level, dynamic optimization of the sensitivity weight and complete recording of historical data, which comprehensively improves the safety assurance and intelligent monitoring capabilities of the process.

[0060] Specifically, as an integrated interactive window, it realizes process data monitoring, abnormal detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions. The specific steps are as follows: real-time display of dynamic curves and digital dashboards of process monitoring data to ensure that operators can see the changes in working conditions at a glance; the process flow uses dynamic animation to intuitively present the data relationship of each process section to help understand the overall process; real-time display of detection results and abnormality scores, using different colors and icons to intuitively distinguish each alarm level, facilitating rapid identification of risks; the interface provides a sensitivity adjustment slider to support operators to manually fine-tune the sensitivity threshold according to actual working conditions, and visualize the impact of sensitivity changes on detection results in real time to ensure accurate sensitivity settings; at the same time, it supports historical data query, displays alarm history records, response records and process fluctuation trends, and forms a comparative analysis interface to assist operators in operation and maintenance summaries and process optimization; the system is multi-terminal adaptable, supporting large screens, PCs, Web terminals and other operating scenarios to meet the flexible application and visual interaction needs of different production environments.

[0061] In this implementation plan, an integrated operator interaction window is constructed. Through dynamic visualization of process data and test results, sensitivity adjustment and multi-dimensional historical record analysis, real-time monitoring of the entire condensation process and intuitive prompts of abnormal fluctuations are achieved. This not only improves the operator's process perception and operating efficiency, but also significantly enhances the intelligent level of operation and maintenance and the visualization capability of safety warnings.

[0062] like Figure 2As shown, it is a structural diagram of the dimethyl sulfoxide process simulation and fault warning system based on digital twin provided in the embodiment of the present application. The dimethyl sulfoxide process simulation and fault warning system based on digital twin provided in the embodiment of the present application applies the dimethyl sulfoxide process simulation and fault warning method based on digital twin, including: a real-time data acquisition module, a process twin simulation module, an anomaly detection optimization module, a hierarchical alarm response module and an interactive visual interface module: a real-time data acquisition module is used to collect process monitoring data in real time and pre-process the process monitoring data; a process twin simulation module is used to measure the pressure fluctuation intensity of the process monitoring data, judge the stability of the condenser working condition according to the pressure fluctuation intensity measurement result, and trigger the corresponding condenser water Regulation, tower kettle heating power regulation and vacuum pump linkage control strategy; anomaly detection optimization module, which is used to perform anomaly detection optimization on process monitoring data, judge the dynamic state of condenser working condition based on the anomaly detection optimization result, and automatically perform adjustment measures for condensed water, tower kettle heating and vacuum pump; a graded alarm response module, which is used to perform anomaly risk assessment based on the comprehensive process monitoring data, pressure fluctuation intensity measurement results and anomaly detection optimization results, judge the process safety status according to the anomaly risk assessment results, and perform safety level prompts, sensitivity weight adjustment and historical data record analysis; an interactive visual interface module, which is used as an integrated interactive window to realize process data monitoring, anomaly detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions.

[0063] In this implementation plan, by integrating five modules of real-time data acquisition, process twin simulation, anomaly detection optimization, graded alarm response and interactive visualization under the digital twin framework, a complete closed-loop system of dimethyl sulfoxide process simulation and dynamic fault warning driven by real-time data is formed, completing the digital and intelligent management of the entire process from data acquisition to sensitivity adaptive adjustment, from dynamic anomaly identification to safety graded response, significantly improving the visualization level, dynamic adjustment capability and intelligent level of safe operation and maintenance of the condensation process.

[0064] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0065] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A dimethyl sulfoxide process simulation and fault warning method based on digital twins is characterized by: The following steps are involved: S1: Real-time collection of process monitoring data and pre-processing of the process monitoring data; S2: Measure the pressure fluctuation intensity of process monitoring data, judge the stability of the condenser operating conditions based on the pressure fluctuation intensity measurement results, and trigger the corresponding condensate water regulation, tower kettle heating power regulation and vacuum pump linkage control strategy; S3: Optimize anomaly detection of process monitoring data, determine the dynamic state of the condenser operating conditions based on the anomaly detection optimization results, and automatically adjust the condensate water, tower kettle heating, and vacuum pump; S4: Perform abnormal risk assessment based on the process monitoring data, pressure fluctuation intensity measurement results, and abnormality detection optimization results. The process safety status is determined based on the abnormal risk assessment results, and safety level prompts, sensitivity weight adjustments, and historical data record analysis are performed. S5: As an integrated interactive window, it realizes process data monitoring, abnormal detection result display, alarm response information and sensitivity adjustment, and other multi-dimensional interactive functions.

2. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of real-time acquisition of process monitoring data and preprocessing of the process monitoring data are as follows: Process monitoring data includes: time index, condensation pressure, window points, window steps, sampling interval, number of algorithms and weighted anomaly intensity value; Based on the chronological position of data points in the sliding window, a time index is generated within the window. The condensing pressure is collected in real time via the condenser pressure sensor. The system's built-in adaptive algorithm, combined with real-time data volatility analysis and process stage switching, dynamically generates a window length and number of steps adapted to the current operating conditions. The operator manually sets the sampling interval of the data acquisition equipment based on process characteristics. During the system deployment and optimization phase, engineers typically determine the number of algorithms based on process monitoring requirements. The system's multiple anomaly detection algorithms automatically generate weighted anomaly intensity values ​​based on the analysis results of real-time process monitoring data. During the data collection process, a timestamp is automatically assigned to each data point to form time-series data. In the preprocessing stage, equipment shutdown, maintenance and other non-production data are eliminated, sensor instantaneous drift and other abnormal values ​​are excluded, process monitoring data is standardized and normalized to eliminate dimensional differences between different parameters, and time synchronization operations are performed on multi-dimensional data to ensure the consistency of each process parameter in the time series.

3. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of measuring the pressure fluctuation intensity of process monitoring data are as follows: Get time index, condensation pressure and window points; The relative displacement is calculated by the difference between the time index of each data point in the sliding window and the number of window points, which is multiplied by the pressure exponential attenuation factor and then negatively calculated to form an exponential attenuation power term. The exponential attenuation power term is converted to the value of the natural exponential function to generate an exponential attenuation factor, forming the time position attenuation weight. The condensation pressure of each data point is subtracted from the average condensation pressure in the window and squared to form a square deviation term. The square deviation term is multiplied by the exponential attenuation factor at the time point to generate a weighted square difference. All weighted square differences in the sliding window are accumulated to form a weighted square sum. At the same time, all exponential attenuation factors in the window are also accumulated to obtain a weighted sum. The weighted square sum is divided by the weighted sum to form a normalized variance. Finally, the square root is taken to generate the value of the pressure fluctuation intensity value.

4. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of judging the stability of the condenser working condition based on the pressure fluctuation intensity measurement result and triggering the corresponding condensate water regulation, tower kettle heating power regulation and vacuum pump linkage control strategy are as follows: Comparing the pressure fluctuation intensity value with the fluctuation threshold in real time, wherein the fluctuation threshold includes a primary fluctuation threshold and a secondary fluctuation threshold; When the pressure fluctuation intensity value is greater than or equal to the first-level fluctuation threshold, automated adjustment measures are immediately triggered, including increasing the condenser cooling water flow to quickly suppress pressure fluctuations, reducing the tower kettle heating power to reduce the heat load, and linking the fan and vacuum pump to start and stop to help stabilize the pressure. A red high-risk alarm is activated on the interface to prompt the operator to respond quickly. At the same time, operational suggestions are generated to check the status of the condenser vacuum pump and the stability of the cooling water supply, and relevant operation logs are recorded; When the pressure fluctuation intensity is greater than or equal to the second-level fluctuation threshold and less than the first-level fluctuation threshold, the system automatically fine-tunes the condenser cooling water flow and the tower kettle heating power to perform flexible adjustments. At the same time, a yellow medium-risk indicator is displayed on the interface to remind the operator to pay close attention to the pressure fluctuation and generate real-time process optimization suggestions. If the medium fluctuation state lasts for more than 30 minutes, the system automatically tightens the sensitivity, activates a higher-sensitivity detection mode, and records the fluctuation trend data for later analysis. When the pressure fluctuation intensity value is less than the secondary fluctuation threshold, the normal monitoring mode is maintained and only real-time data is recorded for process analysis. The operation interface prompts the process stable state in green. When the short-term fluctuation intensity increases, it can automatically switch to the medium fluctuation sensitivity detection mode.

5. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of optimizing anomaly detection of process monitoring data are as follows: Get condensation pressure, window steps and sampling interval; The pressure change rate is obtained by dividing the condensation pressure difference at each adjacent time point in the sliding window by the sampling interval. The pressure change rate is input into the hyperbolic tangent function for nonlinear transformation and the absolute value is taken. All transformed rate values ​​in the sliding window are accumulated to form a cumulative sum. The value formed by dividing the cumulative sum by the number of window steps is the pressure change rate value.

6. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of judging the dynamic state of the condenser working condition based on the abnormal detection optimization result and automatically adjusting the condensate water, tower kettle heating and vacuum pump are as follows: Comparing the pressure change rate value with a rate threshold in real time, wherein the rate threshold includes a primary rate threshold and a secondary rate threshold; When the pressure change rate is greater than or equal to the first-level rate threshold, the system immediately enters the automatic rapid adjustment mode, automatically increasing the condenser cooling water flow, reducing the tower kettle heating power, and linking the vacuum pump and fan to start and stop. The interface highlights a red warning and emits an audible and visual prompt. Safe operation suggestions are automatically generated to check the condenser water circulation status and confirm the stability of the vacuum pump working condition. The operation process and abnormal data are fully recorded to support later process optimization and safety traceability. When the pressure change rate value is greater than the secondary rate threshold and less than the primary rate threshold, the automatic fine-tuning adjustment mode is executed to fine-tune the condensate valve opening and the tower kettle heating power. A yellow prompt on the interface reminds the operator to maintain key monitoring, and generates real-time adjustment suggestions to indicate the reasonable adjustment range and operation priority. If the medium-risk state lasts for more than 30 minutes, the sensitivity threshold is automatically tightened and the high-rate detection mode is switched. At the same time, the fine-tuning process and detection results are recorded for subsequent process parameter optimization; When the pressure change rate value is less than or equal to the secondary rate threshold, the automatic adjustment measures will not be triggered. Only a green prompt will be displayed on the interface indicating that production is stable and in good condition. The real-time pressure change rate data will be continuously recorded. If the pressure rate fluctuates for a short time, it can be quickly switched to the medium rate detection threshold to ensure safe production. All data will be continuously updated to support long-term optimization and safety trend analysis of twin simulation.

7. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps for performing abnormality risk assessment based on the integrated process monitoring data, pressure fluctuation intensity measurement results, and abnormality detection optimization results are as follows: Get the number of algorithms and weighted anomaly intensity values; The weighted anomaly intensity value is input into the exponentially weighted moving average algorithm to calculate the smoothed weighted value. The smoothed weighted anomaly intensity values ​​of all detection algorithms are multiplied by their respective sensitivity weight coefficients and accumulated to form a weighted smoothed sum. At the same time, all sensitivity weight coefficients are accumulated to form a weighted sum. Finally, the weighted smoothed sum is divided by the weighted sum to generate the value that is the anomaly risk assessment value.

8. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps of judging the process safety status based on the abnormal risk assessment results, performing safety level prompts, sensitivity weight adjustment, and historical data record analysis are as follows: Comparing the abnormal risk assessment value with the risk threshold in real time, wherein the risk threshold includes the first-level risk threshold and the second-level risk threshold; When the abnormal risk assessment value is greater than or equal to the first-level risk threshold, the multi-dimensional safety protection mode is immediately activated, automatically increasing the condenser cooling water flow, synchronously adjusting the vacuum pump vacuum rate, temporarily limiting the tower kettle heating power to the safe load lower limit, and linking the production stop signal preparation logic to trigger a red high-risk alarm and issue an audible, visual, and SMS prompt. In addition, process safety operation guidance suggestions are generated in real time to check the vacuum pump status and ensure sufficient condensate water circulation. The operation process, response records, and safety data are fully recorded to form a traceable safety file. When the abnormal risk assessment value is greater than the second-level risk threshold and less than the first-level risk threshold, the intelligent flexible adjustment mode is executed. The sensitivity weight coefficient is adaptively adjusted based on historical data and real-time trends, and the condensate flow rate and the tower kettle heating power are fine-tuned. The interface prompts the operator to maintain monitoring focus with a yellow prompt, generates real-time safety adjustment suggestions, and prioritizes automated safety operations. If the risk state persists for more than 30 minutes, the sensitivity is automatically tightened and switched to high-risk detection mode. At the same time, the adjustment and safety data are recorded for later optimization. When the abnormal risk assessment value is less than or equal to the secondary risk threshold, automatic adjustment will not be triggered. Only a green prompt will be displayed on the interface indicating that production is stable and in good condition. The risk assessment value and historical data will be continuously recorded. If the risk value increases in the short term, it will automatically switch to the medium-risk sensitivity detection mode to ensure safe production.

9. The dimethyl sulfoxide process simulation and fault warning method based on digital twin according to claim 1 is characterized in that: The specific steps for implementing process data monitoring, abnormality detection result display, alarm response information and sensitivity adjustment, and other multi-dimensional interactive functions as an integrated interactive window are as follows: The dynamic curves and digital dashboards of process monitoring data are displayed in real time. The process flow is presented intuitively in a dynamic animation manner. The data relationship of each process section is presented intuitively. The detection results and abnormal scores are displayed in real time. Different colors and icons are used to intuitively display each alarm level. The interface provides a sensitivity adjustment slider to support operators to fine-tune the sensitivity threshold according to actual working conditions. The impact of sensitivity changes on the detection results is visualized in real time. Historical data query is supported, alarm history records, response records and process fluctuation trends are displayed, and a comparative analysis interface is generated to assist in operation and maintenance summary and process optimization. At the same time, multi-terminal adaptation is achieved, supporting large screens, PC terminals, Web terminals and other operation scenarios.

10. A digital twin-based dimethyl sulfoxide process simulation and fault warning system: This system includes a real-time data acquisition module, a process twin simulation module, an anomaly detection and optimization module, a hierarchical alarm response module, and an interactive visual interface module. Its features include: The real-time data acquisition module is used to collect process monitoring data in real time and pre-process the process monitoring data; The process twin simulation module is used to measure the pressure fluctuation intensity of process monitoring data, judge the stability of the condenser working condition based on the pressure fluctuation intensity measurement results, and trigger the corresponding condensate adjustment, tower kettle heating power adjustment and vacuum pump linkage control strategy; The anomaly detection and optimization module is used to perform anomaly detection and optimization on process monitoring data, judge the dynamic state of the condenser working condition based on the anomaly detection and optimization results, and automatically perform adjustment measures for condensate water, tower kettle heating and vacuum pump; The hierarchical alarm response module is used to perform abnormal risk assessment based on the process monitoring data, pressure fluctuation intensity measurement results and abnormal detection optimization results, determine the process safety status based on the abnormal risk assessment results, and provide safety level prompts, sensitivity weight adjustment and historical data record analysis; The interactive visual interface module is used to integrate interactive windows to realize process data monitoring, abnormality detection result display, alarm response information and sensitivity adjustment and other multi-dimensional interactive functions.

Images