Hydrofluoric acid purification process state data monitoring method

Abstract

The invention relates to the technical field of data monitoring, and discloses a hydrofluoric acid purification process state data monitoring method. The method comprises the following steps: collecting hydrofluoric acid purification equipment state data of a mixing tank, an evaporator, a rectifying tower and an absorption tower in a hydrofluoric acid purification system; performing state evaluation through the fuzzy logic monitoring model to obtain a real-time state evaluation result; calculating equipment state transition probability, and identifying time-varying delay characteristics of mixed reverse tank temperature control, rectifying tower pressure response and absorption tower concentration adjustment to obtain hydrofluoric acid purification equipment state prediction information; self-adaptive switching is carried out among a conventional mode, an anti-fluctuation mode, a rapid recovery mode and an emergency protection mode, and a control strategy aiming at the characteristics of the hydrofluoric acid purification process is obtained. The asymmetric characteristic of parameter change in the hydrofluoric acid purification process is effectively identified, the detection accuracy of state abnormity is improved, and safe and stable operation of the hydrofluoric acid purification process is guaranteed.

Description

Technical Field

[0001] This application relates to the technical field of data monitoring, and particularly to a method for monitoring the status data of the hydrofluoric acid purification process. Background Art

[0002] The hydrofluoric acid purification process involves multiple key equipment such as mixing and reaction tanks, evaporators, distillation columns, and absorption towers. The process flow is complex and has strong non-linear and time-varying characteristics. Traditional purification processes rely on empirical operations and PID control systems with fixed parameters, and cannot cope with process parameter fluctuations, sudden changes in equipment status, and different requirements in different production stages. Especially the asymmetric response characteristics shown in the parameter rising and falling stages often lead to fluctuations in product purity, increased energy consumption, and safety hazards.

[0003] Existing equipment status monitoring technologies generally have problems such as incomplete data collection, insufficient preprocessing, and single monitoring models. Especially in high-risk processes such as hydrofluoric acid purification, the judgment of parameter fluctuations often uses a single threshold method, which cannot distinguish the different characteristics and risks in the parameter rising process and the falling process. At the same time, the time-varying delay problems in key links such as the temperature control of the mixing and reaction tank, the pressure response of the distillation column, and the concentration adjustment of the absorption tower are ignored, resulting in delayed control responses, increased fluctuations, and even chain reactions. In addition, there is no effective linkage mechanism between the equipment status and the control strategy, and the control method cannot be intelligently adjusted according to the real-time status of the equipment, making the system slow to respond to abnormal situations. Summary of the Invention

[0004] This application provides a method for monitoring the status data of the hydrofluoric acid purification process. This application effectively identifies the asymmetric characteristics of parameter changes in the hydrofluoric acid purification process, improves the detection accuracy of abnormal status, and ensures the safe and stable operation of the hydrofluoric acid purification process.

[0005] In a first aspect, this application provides a method for monitoring the status data of the hydrofluoric acid purification process. The method for monitoring the status data of the hydrofluoric acid purification process includes: Collect the original data of the equipment operation status of the mixing and reaction tank, evaporator, distillation column, and absorption tower in the hydrofluoric acid purification system, and preprocess the original data of the equipment operation status to obtain the status data of the hydrofluoric acid purification equipment; Input the status data of the hydrofluoric acid purification equipment into a fuzzy logic monitoring model for status evaluation to obtain a real-time status evaluation result; Calculate the equipment status transition probability according to the real-time status evaluation result, and at the same time identify the time-varying delay characteristics of the temperature control of the mixing and reaction tank, the pressure response of the distillation column, and the concentration adjustment of the absorption tower to obtain the status prediction information of the hydrofluoric acid purification equipment; Based on the predicted information of the hydrofluoric acid purification equipment status, perform adaptive switching among the normal mode, anti-fluctuation mode, rapid recovery mode, and emergency protection mode to obtain a control strategy for the characteristics of the hydrofluoric acid purification process.

[0006] In the technical solution provided by this application, by installing a variety of sensors in the mixing and reaction tank, evaporator, rectification column, and absorption tower and setting different sampling frequencies, combined with an improved Kalman filter algorithm and a 3σ outlier recognition mechanism, high-quality data acquisition of the key parameters in the hydrofluoric acid purification process is achieved, solving the problems of incomplete data and large noise interference in traditional technologies. The dual fuzzy logic monitoring model is adopted, using the triangular membership function to process the rising interval and the Gaussian membership function to process the falling interval, effectively identifying the asymmetric characteristics of parameter changes in the hydrofluoric acid purification process, improving the detection accuracy of abnormal states, especially the recognition ability of key abnormal conditions such as sudden changes in the pressure of the rectification column and concentration deviation of the absorption tower. By establishing a device state transition probability matrix and a periodic time-varying delay compensation mechanism, combined with fast Fourier transform analysis and a feedforward-feedback combined compensator, the time-varying delay problems existing in the temperature control of the mixing and reaction tank, pressure response of the rectification column, and concentration adjustment of the absorption tower in the hydrofluoric acid purification process are solved, and the accurate prediction of the evolution trend of the device state is realized. Based on the predicted information of the device state, adaptive switching is realized among the normal mode, anti-fluctuation mode, rapid recovery mode, and emergency protection mode, combined with the mode switching condition matrix and the smooth transition mechanism, enhancing the adaptability of the control system to different hydrofluoric acid purification working conditions and avoiding the limitations of traditional single control strategies under different working conditions. By closely combining device state monitoring and control strategies, differential monitoring and predictive control are implemented for key devices in the hydrofluoric acid purification process, realizing early intervention in potential risks, preventing state deterioration and chain reactions, and ensuring the safe and stable operation of the hydrofluoric acid purification process. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0008] Figure 1 It is a schematic diagram of an embodiment of the method for monitoring the status data of the hydrofluoric acid purification process in the embodiments of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0009] The embodiments of the present application provide a method for monitoring the status data of the hydrofluoric acid purification process. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the term "comprising" or "having" and any variation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0010] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the method for monitoring the status data of the hydrofluoric acid purification process in the embodiments of the present application includes: Step S101, collect the original data of the equipment operation status of the mixing and reaction tank, evaporator, rectification column and absorption tower in the hydrofluoric acid purification system, and preprocess the original data of the equipment operation status to obtain the status data of the hydrofluoric acid purification equipment; It can be understood that the execution subject of the present application can be a hydrofluoric acid purification process status data monitoring system, or a terminal or a server, and specific limitations are not made here. The embodiments of the present application will be described by taking the server as the execution subject as an example.

[0011] Specifically, different types of sensors are installed on the key equipment of the hydrofluoric acid purification system, including temperature sensors, pressure sensors, flow sensors, and concentration analyzers. These sensors are used to monitor the core parameters of each device respectively. Among them, the temperature sensors (PT100 type) are installed on the top of the mixing and reaction tank, evaporator, and distillation column to obtain the temperature changes. The pressure sensors (YB-151 type) are arranged at the top of the distillation column, the inlet and outlet of the absorption tower, and in the mixing and reaction tank to monitor the pressure fluctuations of each key device. The flow sensors (LZB-50 type) are installed in the evaporator outlet and the absorption tower circulation pipeline to detect the changes in the hydrofluoric acid flow rate. The concentration analyzer (CM-230 type) is used to monitor the hydrofluoric acid concentration at the outlet of the absorption tower and upload the data in real time. In order to achieve the effective management and synchronization of sensor data, the monitoring point layout is carried out to ensure that the data acquisition covers the key parameters of all key devices. At the same time, different sampling frequencies are set. For example, the temperature data is sampled every 5 seconds, the pressure data is sampled every 3 seconds, the flow data is sampled every 2 seconds, and the concentration data is sampled every 10 seconds to capture the real-time changes in the equipment operation during the hydrofluoric acid purification process to the greatest extent. After the data acquisition is completed, all the collected data will be aggregated and synchronously calibrated through an industrial-grade edge computing unit (MCU-E520). The edge computing unit is responsible for the preliminary processing of the data, including data format conversion, timestamp unification, and synchronization of data between devices, avoiding data misalignment problems caused by different sampling frequencies of different sensors. This process also includes filtering the data using an improved Kalman filtering algorithm to eliminate the measurement noise in the data and improve the accuracy of the data. At the same time, an outlier identification mechanism is introduced during the data aggregation stage, and the 3σ principle is used for outlier identification, that is, when the current sampling value of a certain parameter deviates from the average value of the previous 10 sampling points by more than 3 times the standard deviation, this point is marked as an outlier, and data correction is carried out by interpolating the front and back values to ensure the smoothness of the data. The system operation parameter dataset is transmitted through the industrial Ethernet protocol to ensure the real-time and accuracy of the data. The data transmission rate is maintained above 10Mbps, and at the same time, the transmission delay needs to be controlled within 50ms to ensure the high-speed transmission of key data between the device and the central monitoring system. A dual data storage mechanism is adopted to store the system operation parameter dataset in the local edge device and the central server. Among them, the local edge device stores the high-frequency data of the most recent 72 hours for short-term data traceability and analysis, while the central server stores the compressed long-term historical data and ensures that the data retention period is not less than 3 years. The central monitoring system performs real-time analysis and storage on the received data to form the original data of the complete equipment operation status. Filtering processing and outlier identification are carried out on the original data of the equipment operation status. The filtering processing uses an improved Kalman filtering algorithm to eliminate the measurement noise that appears and improve the data quality.Meanwhile, the outlier recognition mechanism adopts the 3σ principle and ensures the timely recognition of data anomalies in different process states by dynamically adjusting the anomaly recognition threshold. The data after filtering and outlier recognition is the status data of the hydrofluoric acid purification equipment.

[0012] Classify the original data of the equipment operation status. According to the different data sources, the data is divided into the temperature data of the mixing and reaction tank, the outlet temperature data of the evaporator, the pressure data of the rectifying column, and the concentration data of the absorption tower, and independent management and analysis are carried out for different data types. Filter the classified original parameter data. Adopt the Kalman filter algorithm to eliminate the measurement noise in the data through the iterative mechanism of prediction and update, and obtain parameter data that is smoother and more in line with the actual operation of the equipment. Calculate the fluctuation benchmark for the parameter data after noise elimination to provide a reference standard for subsequent outlier recognition. By calculating the average value and standard deviation of the data over a period of time, establish the fluctuation benchmark for each type of parameter. The fluctuation benchmark can reflect the change range of each equipment parameter under normal operation. Based on the fluctuation benchmark, mark the outlier for the parameter data after eliminating the measurement noise. Adopt the 3-fold standard deviation principle to judge the outlier. When the change amplitude of a certain data point exceeds the normal fluctuation range, the system automatically marks this point as an outlier to form an outlier mark set. Perform linear interpolation processing on each outlier in the outlier mark set to eliminate the influence of abnormal data on the system status monitoring. Linear interpolation estimates the reasonable value of the outlier through the adjacent valid data points before and after, thereby replacing the outlier and making the corrected data maintain smoothness and continuity, avoiding misjudgment or malfunction of the control algorithm caused by abnormal data. Perform piecewise linear fitting and data downsampling on the corrected parameter data. Piecewise linear fitting analyzes the data change trend, divides the data into different segments, and fits the data of each segment to enable the system to identify the data change characteristics in different stages. When the data change rate is lower than the set change threshold, the system automatically performs data downsampling to reduce the data volume, calculation pressure, and storage requirements. Obtain the status data of the hydrofluoric acid purification equipment.

[0013] Step S102: Input the status data of the hydrofluoric acid purification equipment into the fuzzy logic monitoring model for status evaluation to obtain the real-time status evaluation result; Specifically, calculate the parameter change rate based on the status data of the hydrofluoric acid purification equipment. For example, calculate the change rate of the temperature in the mixing and reaction tank, the change rate of the temperature at the outlet of the evaporator, the change rate of the pressure at the top of the distillation column, and the change rate of the concentration at the outlet of the absorption tower. Then, divide the equipment status data into a parameter rising interval data set and a parameter falling interval data set according to the parameter change rate. Perform fuzzy processing on the parameter rising interval data set to convert the continuously changing numerical values into fuzzy set input values. The fuzzy processing uses a triangular membership function, defining five fuzzy sets: {extremely small, relatively small, medium, relatively large, extremely large}. According to the normal operating range of the equipment and the range of the parameter change rate, fuzzy the trend of the parameter change to form the fuzzy input values in the rising interval. At the same time, perform fuzzy processing on the parameter falling interval data set. The data in the falling interval uses a Gaussian membership function because there are mutations and non-linear changes during the falling process, and a more sensitive membership function is used to capture these changes to obtain the fuzzy input values in the falling interval. Input the fuzzy input values in the rising interval and the fuzzy input values in the falling interval into the fuzzy logic monitoring model for inference calculation. Map and infer the parameter changes by establishing a 45-rule fuzzy rule base. The fuzzy inference rule is in the form of "if the change rate is X and the current value is Y, then the equipment status is Z", covering various states such as normal, slightly abnormal, moderately abnormal, severely abnormal, and dangerous. The preliminary result obtained from the inference calculation is the first fuzzy state output information. Determine the adaptive factor according to the equipment operation time and batch, and adjust the first fuzzy state output information. The value range of the adaptive factor is between 0.8 and 1.2, and it is dynamically adjusted according to the length of the equipment operation time and the process changes of the batch, so as to improve the accuracy and sensitivity of the fuzzy logic inference result. The adjusted fuzzy state output information is the second fuzzy state output information. Since different equipment has different importance levels in the hydrofluoric acid purification process, weight the second fuzzy state output information through the equipment importance weight coefficient matrix, so that the abnormal states of key equipment can obtain higher sensitivity and priority. After completing the weighting process, defuzzify the fuzzy state information to map the fuzzy set back to the specific numerical space to obtain the real-time status evaluation result.

[0014] Set the universe of discourse ranges for temperature, pressure, flow rate, and concentration to ensure that the parameter change rate and the change in the current parameter value are within a reasonable monitoring range. The universe of discourse space of the parameter change rate is obtained through statistical analysis of the parameter change conditions of the equipment during normal operation. At the same time, determine the normal operation ranges of temperature, pressure, flow rate, and concentration to form the universe of discourse space of the parameter value. Uniformly divide the universe of discourse space of the parameter change rate into five preset levels, and these five levels correspond to the five fuzzy sets of "extremely small, relatively small, medium, relatively large, extremely large". To achieve fuzzification, construct five triangular membership functions, and the vertex coordinates of these functions are located at both ends and the midpoint of the universe of discourse. For example, for the temperature change rate, define five intervals of extremely small, relatively small, medium, relatively large, and extremely large from the lowest to the highest point. The vertices of the triangular membership functions are located at the midpoints of each level interval and drop to zero at both endpoints, forming a continuous fuzzy membership distribution. Similarly, uniformly divide the universe of discourse space of the parameter value into the same five levels and construct five triangular membership functions, and the vertex coordinates of these functions are located at the midpoints of each interval to obtain the fuzzy membership function set of the parameter value. Input each pair of parameter change rates in the parameter rise interval dataset into the fuzzy membership function set of the parameter change rate for calculation to obtain the change rate membership value matrix. During the calculation process, determine which interval range of the membership function the current change rate falls into, and then calculate the membership value of the change rate according to the triangular membership function. According to the linear change relationship of the membership function, obtain the belonging degree of the change rate in different membership functions. Similarly, input the current parameter value in the parameter rise interval dataset into the fuzzy membership function set of the parameter value for calculation to obtain the parameter value membership value matrix. For the fuzzification of the current parameter value, adopt the same mechanism as the change rate. By judging which interval the parameter value falls into and calculating the corresponding membership value, form the parameter value membership matrix. Combine the change rate membership value matrix and the parameter value membership value matrix, and combine them by weighted summation or taking the minimum membership degree of the two membership matrices to ensure that each pair of change rates and current values can be mapped into a fuzzy input value matrix to form the fuzzy input value of the rise interval.

[0015] Step S103: Calculate the equipment state transition probability according to the real-time state evaluation result, and at the same time identify the time-varying delay characteristics of the temperature control of the mixing and reaction tank, the pressure response of the rectification column, and the concentration adjustment of the absorption tower to obtain the state prediction information of the hydrofluoric acid purification equipment; Specifically, based on the real-time status assessment results, the operating statuses of each device are classified, and the current status data is divided into five statuses: "normal, slightly abnormal, moderately abnormal, severely abnormal, and dangerous". The classified data is stored in the device status classification dataset. According to the device status classification dataset, the transfer situations of the operating status of each device to other statuses are statistically analyzed. By analyzing the frequencies of each device transferring from one status to another over a past period of time and performing normalization processing, the device status transfer probability is obtained, which reflects the possibility of the device transferring between different statuses. For example, the probability of the rectification column pressure transferring from the normal status to the slightly abnormal status is greater than the probability of directly jumping from the normal status to the severely abnormal status. The device status transfer probability is dynamically updated to ensure that the device status evolution law always reflects the actual status under the current process conditions. The dynamic update method adopts the sliding time window method to statistically analyze the status transfer data within a recent period of time (such as 500 hours), and perform an update every 100 hours to ensure that the status transfer matrix is always in the latest state, forming a dynamic transfer matrix that reflects the device status change law. The hydrofluoric acid purification device status data is input into the fast Fourier transform analysis algorithm to analyze the delay periods of the temperature control of the mixing and reaction tank, the pressure response of the rectification column, and the concentration adjustment of the absorption tower. The fast Fourier transform converts the data in the time domain into a frequency domain signal, thereby identifying the main delay period characteristics of different devices, which reflects the time lag of the device's response after receiving a control signal. According to the analyzed delay period characteristics, a feedforward-feedback combined delay compensation model is constructed to eliminate the influence of time-varying delay on the device status prediction. The structure of the compensator includes a feedforward part for predicting the future state change of the system and a feedback part for real-time correction of the control error. By combining the feedforward and feedback compensation mechanisms, control adjustments are made in advance within the delay period, thereby reducing the interference of the delay on the system status. The compensation model can dynamically adjust the compensation parameters according to the delay characteristics of different devices, so as to maintain a high control accuracy under different delay periods and ensure the accuracy of the device status prediction. The dynamic transfer matrix is combined with the delay compensation model to form a device status prediction mechanism by calculating the state probability distribution at a future moment and taking the delay factor into consideration. The state probability distribution at a future moment is predicted based on the current state and the state transfer matrix. After combining with the delay compensation model, the state change deviation caused by the control signal delay is corrected, so that the prediction result is closer to the actual operating status of the device, and the hydrofluoric acid purification device status prediction information is obtained.

[0016] Step S104: Based on the hydrofluoric acid purification device status prediction information, adaptively switch between the normal mode, the anti-fluctuation mode, the rapid recovery mode, and the emergency protection mode to obtain a control strategy for the characteristics of the hydrofluoric acid purification process.

[0017] Specifically, a control mode switching condition matrix is constructed based on the device status prediction information. This matrix maps the status classification of the device to the control mode and determines the optimal control mode under different statuses. When the device status gradually transfers from the normal status to the abnormal status, the control mode switching condition matrix dynamically determines the current optimal control mode type according to the device status evaluation result, the state transition probability, and the status prediction information after delay compensation. The conditions in the matrix include the category to which the current status belongs, and also comprehensively consider the evolution trend of the future status, the degree of abnormality, and the possibility of status recovery, ensuring that the appropriate control mode can be switched at the initial stage of the abnormal status to prevent the abnormality from further deteriorating. Select the control algorithm parameters corresponding to the device status based on the currently applicable control mode type. Among them, the PID control algorithm is used in the normal mode. This mode is applicable when the device is in the normal status. After the PID parameters are optimized, the device can be maintained to operate stably under the best working conditions. When the device status transfers to a slightly abnormal or fluctuating trend, the system switches to the anti-fluctuation mode and uses the fuzzy PID control algorithm to suppress fluctuations and stabilize the device operation by adjusting the PID parameters in real time. If the status further deteriorates and enters the moderately abnormal status, the system switches to the fast recovery mode and uses the model predictive control strategy to realize the rapid return of the device to the normal status by predicting the future status and optimizing the control input in advance. When the device status enters the severely abnormal or dangerous status, the system immediately switches to the emergency protection mode and uses the robust H∞ control algorithm. This algorithm has anti-interference ability and robustness to ensure the basic safe operation of the device under extreme conditions. During the process of control mode switching, in order to avoid the sudden change of the control signal and system instability caused by the control switching, the switching process of the control algorithm is smoothed. Input the control algorithm parameters corresponding to the current device status into the state observer. The state observer estimates the current operating status of the device by fusing the actual measurement data and the model prediction data. At the same time, calculate the output values of the old and new control algorithms respectively to obtain the dual-algorithm control quantity during the control mode switching. The dual-algorithm control quantity is weighted and averaged through the transition function to form a smooth control output signal. The transition function adopts the exponential smoothing form and is dynamically adjusted according to the time process and the state change speed during the switching process to ensure the continuity of the control signal during the switching period and avoid the instability of the device operation caused by sudden changes. After completing the control mode switching and applying the new control algorithm, the performance of the smoothly transitioned control output signal is evaluated. By setting the control mode effectiveness evaluation indicators, including the integral of the absolute error, the total change of the control signal, and the system stable time, etc., the actual control effect of the current control strategy is comprehensively evaluated. The performance evaluation result is used to judge the effectiveness of the current control mode and is input into the control mode switching condition matrix as feedback information.If the evaluation result shows that the current control mode fails to effectively suppress the abnormal state or restore the device to the normal operating state, the system further optimizes the control mode switching conditions according to the feedback result, adjusts the thresholds and switching logic in the switching condition matrix, and forms a continuously self-optimizing and adaptable control strategy.

[0018] In the embodiment of the present application, by installing a variety of sensors on the mixing and reaction tank, evaporator, rectification column and absorption tower and setting different sampling frequencies, combined with the improved Kalman filter algorithm and 3σ outlier recognition mechanism, high-quality data acquisition of the key parameters in the hydrofluoric acid purification process is realized, and the problems of incomplete data and large noise interference in the traditional technology are solved. The dual fuzzy logic monitoring model is adopted, the triangular membership function is used to process the rising interval, and the Gaussian membership function is used to process the falling interval, effectively identifying the asymmetric characteristics of the parameter changes in the hydrofluoric acid purification process, improving the detection accuracy of state anomalies, especially the recognition ability of key abnormal conditions such as sudden changes in the pressure of the rectification column and concentration deviation of the absorption tower. By establishing the equipment state transition probability matrix and the periodic time-varying delay compensation mechanism, combined with the fast Fourier transform analysis and the feedforward-feedback combined compensator, the time-varying delay problems existing in the temperature control of the mixing and reaction tank, the pressure response of the rectification column and the concentration adjustment of the absorption tower in the hydrofluoric acid purification process are solved, and the accurate prediction of the equipment state evolution trend is realized. Based on the equipment state prediction information, adaptive switching is realized among the normal mode, anti-fluctuation mode, fast recovery mode and emergency protection mode. Combined with the mode switching condition matrix and the smooth transition mechanism, the adaptability of the control system to different hydrofluoric acid purification working conditions is enhanced, and the limitations of the traditional single control strategy under different working conditions are avoided. By closely combining equipment state monitoring and control strategies, differential monitoring and predictive control are implemented on the key equipment in the hydrofluoric acid purification process, early intervention of potential risks is realized, state deterioration and chain reactions are prevented, and the safe and stable operation of the hydrofluoric acid purification process is guaranteed.

[0019] In a specific embodiment, the process of executing step S101 may specifically include the following steps: Install temperature sensors, pressure sensors, flow sensors and concentration analyzers on the mixing and reaction tank, evaporator, rectification column and absorption tower in the hydrofluoric acid purification system to obtain the monitoring point layout of the hydrofluoric acid purification system; Collect the real-time data of each monitoring point of the hydrofluoric acid purification system through the monitoring point layout of the hydrofluoric acid purification system to obtain the mixing and reaction tank temperature data, evaporator outlet temperature data, rectification column pressure data and absorption tower concentration data; Use an industrial-grade edge computing unit to collect and synchronously calibrate the mixing and reaction tank temperature data, evaporator outlet temperature data, rectification column pressure data and absorption tower concentration data to obtain a system operation parameter data set; The system operation parameter data set is transmitted to the central monitoring system through the industrial Ethernet protocol and stored in the local edge device and the central server to obtain the original device operation status data; The original device operation status data is filtered and outlier identified to obtain the hydrofluoric acid purification device status data.

[0020] Specifically, different types of sensors are installed on key equipment such as mixing tanks, evaporators, distillation towers and absorption towers to monitor the important process parameters of each equipment in real time. For the mixing tank, it is necessary to monitor the temperature, pressure and flow changes during the reaction process. A PT100 temperature sensor is set in the mixing tank to monitor the reaction temperature, a YB-151 pressure sensor is used to monitor the reaction pressure, and an LZB-50 flow sensor is set at the inlet and outlet of the reaction material to monitor the inlet and outlet flow. For the evaporator, it is necessary to monitor the outlet temperature to ensure the temperature control accuracy of the hydrofluoric acid gasification process. A PT100 temperature sensor is set at the outlet of the evaporator to accurately collect temperature data during the evaporation process. As the core equipment in the hydrofluoric acid purification process, the distillation tower needs to monitor the pressure, temperature and flow changes at the top and bottom of the tower. Therefore, a YB-151 pressure sensor and a PT100 temperature sensor are set at the top and bottom of the distillation tower respectively, and an LZB-50 flow sensor is installed at the feed and discharge ports of the distillation tower to grasp the operating status of the distillation tower. The outlet hydrofluoric acid concentration of the absorption tower is a key indicator of the system operation effect. A CM-230 concentration analyzer is installed at the outlet of the absorption tower to monitor the concentration changes of the absorption tower in real time. After completing the layout of the monitoring points, the operating data of each key equipment in the hydrofluoric acid purification system is collected in real time through the monitoring points to obtain the mixing tank temperature data, evaporator outlet temperature data, distillation tower pressure data and absorption tower concentration data. The data collection adopts a layered collection mechanism, and different types of sensors are set with different sampling frequencies to capture subtle changes in the operating status of the equipment. The mixing tank temperature data is collected every 5 seconds to reflect the dynamic change trend of the reaction temperature; the evaporator outlet temperature data also uses a 5-second sampling cycle to ensure that the temperature fluctuation information of the evaporation process is obtained in time; the distillation tower pressure data is collected at a higher frequency of 3 seconds / time; the absorption tower concentration data is a key parameter of the purification process, and the concentration analysis data is collected every 10 seconds to ensure that the concentration change can reflect the operation status of the absorption tower in a timely manner. After these real-time data are collected by sensors, they are directly input into the industrial-grade edge computing unit for data collection and synchronous calibration. The industrial edge computing unit (MCU-E520) is responsible for data collection and synchronization calibration. It aggregates the raw data from different sensors and performs time stamp calibration to ensure that the data from different collection points maintain time consistency. Since different devices and sensors have different sampling periods, in order to avoid data dislocation and time drift, the edge computing unit uses a high-precision time synchronization protocol to time align data with different sampling frequencies. At the same time, during the data collection stage, the edge computing unit converts the data format and unifies the data from different sensors into a standard data format for subsequent data processing and analysis. The edge computing unit also performs preliminary data cleaning on the data, including deleting duplicate data, correcting invalid data, and performing data integrity checks to ensure the integrity and consistency of the system operation parameter data set.After the data aggregation is completed, an operating parameter dataset is generated, which includes key parameters such as the temperature data of the mixing and reaction tank, the outlet temperature data of the evaporator, the pressure data of the distillation column, and the concentration data of the absorption tower. The system operating parameter dataset is transmitted through the industrial Ethernet protocol, which has the advantages of fast data transmission speed, large bandwidth, and strong anti-interference ability, ensuring that a large amount of real-time data can be stably and efficiently transmitted to the central monitoring system. During the data transmission process, data is packed through the TCP / IP protocol, and a data encryption mechanism is adopted to ensure the security of data during transmission. The industrial Ethernet protocol guarantees that the data transmission bandwidth is above 10 Mbps and controls the data transmission delay within 50 ms, ensuring that the system operating data is uploaded to the central monitoring system in a timely manner. To ensure the reliability of data and prevent data loss, a dual data storage mechanism is adopted, and the data is stored in both the local edge device and the central server simultaneously. The local edge device stores high-frequency data within the most recent 72 hours for timely data compensation in case of network interruption, while the central server stores compressed long-term historical data with a data storage period of no less than 3 years for subsequent status analysis and trend prediction. After the data aggregation, transmission, and storage are completed, the original data of the equipment operating status is filtered and outlier identification is performed to obtain the status data of the hydrofluoric acid purification equipment. The filtering process denoises the data through an improved Kalman filtering algorithm. The Kalman filter dynamically predicts the current data based on historical data and fuses the measurement data with the predicted data, thereby effectively eliminating the random noise in the data and improving the smoothness of the data. During the filtering process, outlier identification is performed, and the 3σ principle is used to mark the abnormal data, that is, when a data point deviates from the mean of the historical data by more than 3 times the standard deviation, it is marked as an abnormal point, and data correction is performed through interpolation or data fitting. The data is subjected to piecewise linear fitting and data downsampling processing. The part with a data change rate lower than the threshold is downsampled, thereby reducing the data volume and computational pressure, and finally obtaining the status data of the hydrofluoric acid purification equipment.

[0021] In a specific embodiment, the process of performing filtering processing and outlier identification on the original data of the equipment operating status to obtain the status data of the hydrofluoric acid purification equipment may specifically include the following steps: Classify the original data of the equipment operating status to obtain the classified original parameter data, and use the Kalman filtering algorithm to filter the classified original parameter data to obtain the parameter data after eliminating the measurement noise; Calculate the average value and standard deviation of the parameter data after eliminating the measurement noise to obtain the parameter fluctuation reference value; Based on the parameter fluctuation reference value, mark the abnormal points of the parameter data after eliminating the measurement noise to obtain the abnormal point marking set; Perform linear interpolation on each outlier in the outlier marker set to obtain the parameter data with the outliers corrected; Perform piecewise linear fitting and data downsampling on the parameter data with the outliers corrected to obtain the status data of the hydrofluoric acid purification equipment.

[0022] Specifically, classify the original data of the equipment operating status. According to the different data sources and monitored parameters, the data is divided into four independent data sets: the temperature data of the mixing and reaction tank, the outlet temperature data of the evaporator, the pressure data of the rectifying column, and the concentration data of the absorption tower. The original classified parameter data is refined according to the characteristics of the equipment operating parameters and aligned in time according to the sampling period, so that subsequent data processing can maintain time consistency and prevent time drift of parameters of different equipment during data analysis. Filter the original classified parameter data. Use the Kalman filter algorithm to dynamically estimate the data, and eliminate the measurement noise in the data through the mechanism of prediction and update. The Kalman filter dynamically models the equipment operating status according to the state equation and measurement equation of the system, and uses historical data to predict the current data to obtain the optimal estimated value, making the filtered data smoother and closer to the real state. During the data filtering process, the Kalman filter continuously updates the state estimated value and corrects the prediction result according to the newly collected data. The dynamic update mechanism enables the filtered data to effectively remove random noise, reduce measurement errors, and improve the stability of the data. The data after filtering is the parameter data with the measurement noise eliminated. Calculate the average value and standard deviation of the parameter data with the measurement noise eliminated to obtain the fluctuation reference value of each parameter. Based on the calculated parameter fluctuation reference value, mark the outliers in the parameter data with the measurement noise eliminated and generate an outlier marker set. The outlier identification adopts the 3σ principle, that is, when a data point deviates from the parameter average value by more than 3 times the standard deviation, mark the data point as an outlier and store it in the outlier marker set. By correcting the marked data points, effectively eliminate the data anomalies caused by measurement errors or sudden interferences. Perform linear interpolation on the marked outliers to correct the abnormal data points. Linear interpolation estimates the outliers by using the adjacent valid data points before and after, and replaces the outliers with corrected values that conform to the data change trend. After completing the outlier correction, in order to improve the efficiency of data processing and avoid storing a large amount of redundant data, perform piecewise linear fitting and data downsampling on the corrected parameter data to obtain the status data of the hydrofluoric acid purification equipment. Piecewise linear fitting analyzes the data change trend, divides the data into multiple continuous intervals, and fits the data in each interval, so that the data change within each interval is accurately represented by a linear model. When performing data downsampling, according to the set change rate threshold, downsample the data segments with a change rate lower than the set threshold, thereby reducing the data volume and improving the efficiency of data storage and processing.

[0023] In a specific embodiment, the process of executing step S102 may specifically include the following steps: Calculate the parameter change rate based on the hydrofluoric acid purification equipment status data, and divide the hydrofluoric acid purification equipment status data into a parameter rising interval data set and a parameter falling interval data set according to the parameter change rate; Perform a fuzzification process on the parameter rising interval data set to obtain a rising interval fuzzy input value; Perform a fuzzification process on the parameter falling interval data set to obtain a falling interval fuzzy input value; Input the rising interval fuzzy input value and the falling interval fuzzy input value into the fuzzy logic monitoring model for inference calculation respectively to obtain the first fuzzy state output information; Determine the adaptive factor according to the equipment operation time and batch, and adjust the first fuzzy state output information according to the adaptive factor to obtain the second fuzzy state output information; Through the equipment importance weight coefficient matrix, perform weighted processing and defuzzification on the second fuzzy state output information to obtain the real-time state evaluation result.

[0024] Specifically, monitor the status data of key equipment in the hydrofluoric acid purification process. The data includes the temperature data of the mixing and reaction tank, the outlet temperature data of the evaporator, the pressure data of the distillation column, and the outlet concentration data of the absorption tower. By performing time series analysis on these data, calculate the change rates between consecutive time points for each parameter, such as the temperature change rate, pressure change rate, flow rate change rate, and concentration change rate, etc., and divide the data into an increasing interval data set and a decreasing interval data set according to the positive and negative directions of the change rates. The calculation of the parameter change rate is achieved through the difference method or the sliding window method to capture the minute changes of each key parameter during the operation of the equipment. The increasing interval data set contains the data when the parameter shows an increasing trend, reflecting the states of system heating, pressure increase, or flow rate rise, while the decreasing interval data set contains the data when the parameter is in a decreasing trend, reflecting the states of equipment cooling, pressure drop, or flow rate reduction. Perform fuzzification on the increasing interval data set of the parameter, converting the continuously varying numerical data into fuzzy set input values. The fuzzification process uses the triangular membership function or the Gaussian membership function, and sets the domain range of the membership function according to the normal operation interval of different parameters. The parameter change rate is divided into five fuzzy sets, including "extremely small, relatively small, medium, relatively large, extremely large". The vertices of the membership functions of each fuzzy set are located at both ends and the midpoint of the domain, effectively reflecting the trend and amplitude of data changes. During the fuzzification process, perform fuzzy mapping on each parameter change rate and calculate its membership degree values in each fuzzy set to form the fuzzy input values for the increasing interval. At the same time, perform fuzzification on the decreasing interval data set of the parameter. However, due to the more complex non-linear changes during the decreasing process, the Gaussian membership function is used for the data fuzzification in the decreasing interval to more sensitively capture the sudden changes and fluctuating changes that occur during the decreasing process, obtaining the fuzzy input values for the decreasing interval. Input the fuzzy input values for the increasing interval and the decreasing interval into the fuzzy logic monitoring model for inference calculation. The fuzzy logic monitoring model contains a fuzzy rule base consisting of 45 fuzzy rules. The fuzzy rules perform inference mapping on different parameter change situations in the form of "if the change rate is X and the current value is Y, then the equipment status is Z", covering state categories such as normal, slightly abnormal, moderately abnormal, severely abnormal, and dangerous. During the inference process, the fuzzy model performs fuzzy inference on the fuzzy inputs for the increasing interval and the decreasing interval, and through the fuzzy inference mechanism, combines and matches the fuzzy sets of the input variables to generate the first fuzzy state output information. Since the operation state of the equipment has time-varying characteristics, the first fuzzy state output information is not fixed, but dynamically adjusted according to factors such as the operation time of the equipment and batch changes. To improve the accuracy of the inference results, determine the adaptive factor according to the operation time and batch of the equipment, and adjust the first fuzzy state output information according to the adaptive factor.The adaptive factor is a dynamically adjusted parameter, whose value fluctuates between 0.8 and 1.2 and is dynamically updated according to the running time and batches of different devices, enabling the fuzzy logic monitoring model to automatically adjust the sensitivity and response speed of the inference process as the device state changes. After the adjustment of the adaptive factor is completed, the second fuzzy state output information is obtained. The second fuzzy state output information is weighted and defuzzified through the device importance weight coefficient matrix. The weight coefficient matrix weights the parameter states according to the importance of different devices. For example, the distillation column and the absorption column, as the core devices in the hydrofluoric acid purification process, have higher weight coefficients in the weight matrix, while the weight of the mixing reactor and the evaporator is relatively low. The defuzzification process uses the weighted average method or the maximum membership degree method to convert the fuzzy state information into a specific numerical form, obtaining the final real-time state evaluation result, which reflects the current operating state of the device.

[0025] In a specific embodiment, the process of performing the step of fuzzifying the parameter rising interval data set to obtain the fuzzy input value of the rising interval may specifically include the following steps: Set the domain ranges of temperature, pressure, flow rate, and concentration to obtain the domain space of the parameter change rate, and determine the normal operating intervals of temperature, pressure, flow rate, and concentration to obtain the domain space of the parameter values; Divide the domain space of the parameter change rate evenly into five levels according to a preset, and construct five triangular membership functions with vertex coordinates located at both ends and the midpoint of the domain to obtain the fuzzy membership function set of the parameter change rate; Divide the domain space of the parameter values evenly into five levels according to a preset, and construct five triangular membership functions with vertex coordinates at the midpoints of each interval to obtain the fuzzy membership function set of the parameter values; Calculate each pair of parameter change rates in the parameter rising interval data set into the fuzzy membership function set of the parameter change rate to obtain the change rate membership value matrix. At the same time, input the current parameter value in the parameter rising interval data set into the fuzzy membership function set of the parameter values to obtain the parameter value membership value matrix; Combine the change rate membership value matrix and the parameter value membership value matrix to obtain the fuzzy input value of the rising interval.

[0026] Specifically, according to the process characteristics and historical operation data of the equipment, the change ranges of temperature, pressure, flow rate, and concentration are set, and the change rate range of the parameters is defined as the parameter change rate universe space. The universe space of the parameter change rate is set according to the physical characteristics of each equipment. For example, the temperature change rate of the mixing and reaction tank is set from -10°C / min to 10°C / min. At the same time, the normal operation intervals of temperature, pressure, flow rate, and concentration are determined to obtain the parameter value universe space. The parameter change rate universe space is evenly divided into five preset levels to construct a fuzzy membership function set of the parameter change rate. The five levels are "extremely small, relatively small, medium, relatively large, extremely large". The change situation of the parameter change rate is divided into different degrees of change amplitudes, so as to describe the state change of the system more precisely. The membership function is modeled using a triangular membership function. The vertex coordinates of each membership function are located at both ends and the midpoint of the universe, so that the parameter change rate is evenly mapped into each fuzzy set. At the same time, the parameter value universe space is evenly divided into the same five levels, and a fuzzy membership function set of the parameter value is constructed. The fuzzy membership function of the parameter value is also modeled using a triangular membership function. The vertex coordinates of each membership function are set at the midpoint of each interval and gradually decrease to zero at both ends, thus forming a complete fuzzy mapping. Each pair of parameter change rates in the parameter rising interval dataset is input into the fuzzy membership function set of the parameter change rate for calculation to obtain the change rate membership value matrix. For each data point in the parameter rising interval dataset, judge the fuzzy interval to which its parameter change rate belongs, and calculate its membership value in each fuzzy set according to the triangular membership function. The calculation of the membership value is linearly interpolated according to the distance between the parameter change rate and the vertex of the fuzzy set to obtain the change rate membership value matrix, which reflects the change trend of the equipment parameter change rate and quantifies the influence of different change amplitudes on the system state. At the same time, the current parameter value in the parameter rising interval dataset is input into the fuzzy membership function set of the parameter value for calculation to obtain the parameter value membership value matrix. The parameter value membership value matrix forms the parameter value membership matrix by judging the fuzzy interval where the current parameter value is located and calculating the membership degree according to the membership function. The change rate membership value matrix and the parameter value membership value matrix are combined, and the two are weighted and summed or combined with the minimum membership degree to form the fuzzy input value in the rising interval.

[0027] In a specific embodiment, the process of executing step S103 may specifically include the following steps: Based on the real-time state evaluation result, classify the operating states of each equipment to obtain the equipment state classification dataset; According to the equipment state classification dataset, calculate the frequency of transfer of each equipment operating state to other states and perform normalization processing to obtain the equipment state transition probability; Dynamically update the device state transition probability to obtain a dynamic transition matrix that can reflect the law of device state evolution; Input the hydrofluoric acid purification equipment state data into the fast Fourier transform analysis algorithm to analyze the temperature control delay period of the mixing and reaction tank, the pressure response delay period of the rectification column, and the concentration adjustment delay period of the absorption tower, and obtain the main delay period characteristics of each device; According to the main delay period characteristics of each device, construct a feedforward-feedback combined compensator for the transfer function to obtain a delay compensation model for each device; Combine the dynamic transition matrix and the delay compensation model, and by calculating the state probability distribution at future times and considering the influence of delay factors, obtain the state prediction information of the hydrofluoric acid purification equipment.

[0028] Specifically, based on the real-time status evaluation results, the operating status of each device is classified. The evaluation results are obtained by inferring the fuzzy input values in the rising interval and the falling interval through a fuzzy logic monitoring model, reflecting the current status of the device, including the operating status of devices such as the mixing and reaction tank, evaporator, distillation column, and absorption tower. These statuses are divided into five levels: normal, slightly abnormal, moderately abnormal, severely abnormal, and dangerous status. During the operation of the system, the real-time status evaluation results are continuously updated as the device status changes. Therefore, the status data is sorted according to device classification to form a device status classification data set, recording the status changes of each device at different time periods. According to the device status classification data set, calculate the frequency of each device operating status transitioning to other statuses and perform normalization processing to obtain the device status transition probability, reflecting the possibility of the device evolving from one status to other statuses. During the calculation process, count the number of times each device transitions from one status to other statuses within a recent period of time (e.g., 1000 time steps), and normalize these frequencies to obtain the state transition probability matrix. For example, if the frequency of the distillation column transitioning from the normal status to the slightly abnormal status accounts for 20% of the total transition frequency, then the transition probability is 0.2. These state transition probabilities constitute the preliminary state transition matrix. Since the hydrofluoric acid purification process has non-linear and time-varying characteristics, the change law of the device status will be dynamically adjusted as the process conditions change, and the update of the device status transition probability needs to be dynamic. Introduce the sliding time window method to dynamically update the device status transition probability. Continuously collect new status change data within the set time window (e.g., 500 hours), and update the state transition probability matrix every 100 hours to ensure that the state transition matrix always reflects the evolution law of the current operating status of the device. Through the dynamic update mechanism, a dynamic transition matrix reflecting the evolution law of the device status is generated. Input the hydrofluoric acid purification device status data into the fast Fourier transform analysis algorithm to convert the time-domain data into frequency-domain data, thereby identifying the main delay cycle characteristics in the control systems of each device. For example, the delay cycle of the mixing and reaction tank temperature control system reflects the heat transfer time of the reactants during the mixing process, and its main delay cycle is identified through fast Fourier transform analysis; the pressure response delay cycle of the distillation column reflects the time required for the top pressure to reach a steady state after a load change; the delay cycle of the absorption tower concentration adjustment is related to factors such as liquid absorption efficiency and material transfer speed. According to the main delay cycle characteristics of each device, construct a feedforward-feedback combined compensator for the transfer function to eliminate the system response lag problem caused by the delay and obtain a delay compensation model for each device.The structure of the feedforward-feedback compensator includes a feedforward part and a feedback part. The feedforward part is used to perform control compensation in advance when the system detects a state change, making an advanced adjustment by predicting the future state trend, so as to reduce the impact of delay on the system. The feedback part is used to monitor the actual response of the device state in real time and correct the control quantity according to the actual operating state to form a closed-loop compensation control. The parameters of the compensator are dynamically adjusted according to the delay period of different devices. For example, for the temperature control delay period of the mixing and reaction tank, the feedforward part of the compensator makes an advance compensation according to the temperature change trend in the next 45 minutes, and the feedback part makes real-time correction according to the temperature deviation, so as to ensure that the temperature change can be stably maintained within the set range. The compensation models of the distillation column and the absorption tower are also adjusted according to their respective delay periods, so that the system can maintain a high control accuracy and stability under different working conditions. After combining the dynamic transition matrix and the delay compensation model, by calculating the state probability distribution at future moments and taking the delay factor into account, the state prediction information of the hydrofluoric acid purification equipment is obtained. The state probability distribution is predicted based on the current device state and the dynamic transition matrix. By simulating the state transition situation at multiple future time steps, the future state change trend of the device is predicted. The delay compensation model compares the prediction result with the actual operating state through the feedforward-feedback compensation mechanism and corrects the deviation caused by the control signal delay, making the state prediction information more accurate and reliable. Generate device state prediction information, including the future state changes of each device, the operating state trend after delay compensation, and potential abnormal state warnings.

[0029] In a specific embodiment, the process of executing step S104 may specifically include the following steps: Construct a control mode switching condition matrix according to the state prediction information of the hydrofluoric acid purification equipment; Perform real-time calculation on the control mode switching condition matrix to obtain the currently applicable control mode type; Select the control algorithm parameters corresponding to the device state based on the currently applicable control mode type, where the conventional mode adopts a control algorithm with PID parameters, the anti-fluctuation mode adopts a fuzzy PID control algorithm, the fast recovery mode adopts a model predictive control strategy, and the emergency protection mode adopts a robust H∞ control algorithm; Input the control algorithm parameters corresponding to the device state into the state observer for system current state estimation, and at the same time calculate the output values of the old and new control algorithms to obtain the dual-algorithm control quantity at the time of control mode switching; Perform weighted averaging on the dual-algorithm control quantity at the time of control mode switching through a transition function to obtain a smoothly transitioning control output signal; The performance of the smoothed transition control output signal is evaluated through the control mode effectiveness evaluation index to obtain the performance evaluation result, and the performance evaluation result is fed back to the control mode switching condition matrix to obtain the control strategy for the characteristics of the hydrofluoric acid purification process.

[0030] Specifically, according to the device status prediction information, a control mode switching condition matrix is constructed. This matrix maps the status categories of the device to control modes to ensure automatic switching to the optimal control mode under different device states. The control mode switching condition matrix is a fifth-order matrix. The status categories include "normal, slightly abnormal, moderately abnormal, severely abnormal, and dangerous", while the control modes include "conventional mode, anti-fluctuation mode, rapid recovery mode, and emergency protection mode". By setting the condition thresholds in the matrix, the appropriate control mode is automatically selected according to the device status prediction information. The control mode switching condition matrix is calculated in real time. By comparing the current state of the device with the predicted state, it is judged whether a mode switch is required, and the type of the currently applicable control mode is obtained. The conditions for control mode switching include multiple factors such as state transition probability, parameter change trend, and device importance. For example, if the system detects that the change amplitude of the concentration at the outlet of the absorption tower exceeds the set threshold and the concentration deviation trend continues to increase, the system will switch to the rapid recovery mode; when the temperature of the mixing and reaction tank shows abnormal fluctuations and the device operation state enters the dangerous area, the system will automatically switch to the emergency protection mode. Through the real-time calculation of the multi-dimensional condition matrix, it is ensured that the system automatically selects the optimal control mode type under different states. After determining the type of the currently applicable control mode, the corresponding control algorithm parameters are selected according to different control modes. Among them, the conventional mode is applicable when the device is in a normal state, and the PID control algorithm is used for control. After the PID parameters are precisely tuned, the system operates stably; when the device state turns to slightly abnormal or there is a fluctuation trend, the system switches to the anti-fluctuation mode. In this mode, the fuzzy PID control algorithm is used, and the PID parameters are dynamically adjusted through the fuzzy inference mechanism, so that the control system can better suppress fluctuations and improve the stability of the system; if the device enters the moderately abnormal state, the system selects the rapid recovery mode. This mode adopts the model predictive control strategy, and by predicting the future state and optimizing the control input in advance, the device can be quickly restored to the normal state; when the device state enters the severely abnormal or dangerous state, the system switches to the emergency protection mode. This mode adopts the robust H∞ control algorithm, which has strong anti-interference ability and robustness and can maintain the basic safe operation of the device in extreme cases. After the control mode switching is completed, the control algorithm parameters corresponding to the device state are input into the state observer for the current state estimation of the system, and at the same time, the output values of the old and new control algorithms are calculated to ensure the smooth transition of the system during the control mode switching process. The state observer estimates the current state of the system in real time according to the actual measurement data and the device state model, and calculates the control output values at the current moment by combining the old and new control algorithms respectively.For example, when the system switches from the normal mode to the anti-fluctuation mode, the state observer calculates the PID control output value in the normal mode based on the current device state estimation result, and at the same time calculates the fuzzy PID control output value in the anti-fluctuation mode, generating the dual-algorithm control quantities during the control mode switch. These control quantities are used for the subsequent smooth switching process to prevent the control signal from suddenly changing and causing system instability. To avoid system oscillation or instability caused by the sudden change of the control signal during the control mode switch, the dual-algorithm control quantities during the control mode switch are weighted and averaged through a transition function to obtain a smoothly transitioned control output signal. The transition function adopts an exponential smoothing form and is dynamically adjusted according to the time process of the control mode switch and the state change speed. During the weighted averaging process, the output values of the old and new control algorithms are smoothly transitioned according to the current state estimation result and the mode switch progress. For example, at the initial stage of the mode switch, the weight of the new algorithm is low, and as time goes by, the weight of the new algorithm gradually increases until it completely replaces the old algorithm, realizing the seamless switch of the control signal, thereby maintaining the smoothness of the device operation state and the continuity of the control signal. After completing the control mode switch and the smooth transition of the control output signal, the performance of the smoothly transitioned control output signal is evaluated, and the actual control effect of the current control mode is judged through the control mode effectiveness evaluation index. The performance evaluation includes multiple evaluation indexes such as the integral of absolute error, the total change amount of the control signal, and the system stable time. The system comprehensively evaluates the effect of the current control mode according to these evaluation indexes and generates a performance evaluation result. If the performance evaluation result indicates that the current control mode fails to effectively suppress the abnormal state or restore the device to the normal operation state, the system feeds back the performance evaluation result to the control mode switch condition matrix and dynamically adjusts the condition thresholds and switching logics in the matrix to optimize the control mode switching mechanism. For example, if the system finds that the anti-fluctuation mode cannot effectively suppress fluctuations in a certain specific abnormal situation, the system automatically reduces the trigger condition threshold for switching to the anti-fluctuation mode, or increases the priority for switching to the fast recovery mode, so that it can switch to a more suitable control mode more quickly the next time a similar state appears.

[0031] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described system, system and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0032] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a hydrofluoric acid purification process status data monitoring device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0033] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. A method for monitoring state data of a hydrofluoric acid purification process, characterized in that: include: Collecting original data of equipment operation status of a mixing tank, an evaporator, a distillation tower, and an absorption tower in a hydrofluoric acid purification system, and preprocessing the original data of equipment operation status to obtain hydrofluoric acid purification equipment status data; Inputting the state data of the hydrofluoric acid purification equipment into a fuzzy logic monitoring model for state evaluation to obtain a real-time state evaluation result; Calculate the equipment state transition probability according to the real-time state evaluation result, and simultaneously identify the time-varying delay characteristics of the mixing tank temperature control, the distillation tower pressure response and the absorption tower concentration adjustment to obtain the hydrofluoric acid purification equipment state prediction information; Based on the hydrofluoric acid purification equipment state prediction information, adaptive switching is performed among the normal mode, the anti-fluctuation mode, the fast recovery mode and the emergency protection mode to obtain a control strategy for the characteristics of the hydrofluoric acid purification process.

2. The method for monitoring state data of a hydrofluoric acid purification process according to claim 1, characterized in that: The collecting of the original data of the equipment operation status of the mixing tank, evaporator, distillation tower and absorption tower in the hydrofluoric acid purification system, and preprocessing of the original data of the equipment operation status to obtain the hydrofluoric acid purification equipment status data includes: Installing temperature sensors, pressure sensors, flow sensors and concentration analyzers on a mixing tank, an evaporator, a distillation tower and an absorption tower in a hydrofluoric acid purification system to obtain a monitoring point layout of the hydrofluoric acid purification system; The real-time data of each monitoring point of the hydrofluoric acid purification system are collected through the monitoring point layout of the hydrofluoric acid purification system to obtain the mixing tank temperature data, the evaporator outlet temperature data, the distillation tower pressure data and the absorption tower concentration data; Using an industrial-grade edge computing unit to collect and synchronously calibrate the mixing tank temperature data, the evaporator outlet temperature data, the distillation tower pressure data, and the absorption tower concentration data, to obtain a system operation parameter data set; The system operation parameter data set is transmitted to the central monitoring system through the industrial Ethernet protocol and stored in the local edge device and the central server to obtain the original data of the equipment operation status; The original data of the equipment operation status are filtered and abnormal values ​​are identified to obtain the hydrofluoric acid purification equipment status data.

3. The method for monitoring state data of a hydrofluoric acid purification process according to claim 2, characterized in that: The filtering and outlier identification of the original data of the equipment operation status to obtain the hydrofluoric acid purification equipment status data includes: Classify the original data of the equipment operation status to obtain classified original parameter data, and use a Kalman filter algorithm to filter the classified original parameter data to obtain parameter data after eliminating measurement noise; Calculating the average value and standard deviation of the parameter data after eliminating the measurement noise to obtain a parameter fluctuation reference value; Based on the parameter fluctuation reference value, marking the parameter data after eliminating the measurement noise as abnormal points to obtain an abnormal point marking set; Performing linear interpolation processing on each outlier in the outlier mark set to obtain outlier value corrected parameter data; The parameter data after the abnormal value correction is subjected to piecewise linear fitting and data downsampling to obtain the hydrofluoric acid purification equipment status data.

4. The method for monitoring state data of a hydrofluoric acid purification process according to claim 1, characterized in that: The step of inputting the hydrofluoric acid purification equipment status data into a fuzzy logic monitoring model for status evaluation to obtain a real-time status evaluation result includes: Calculating a parameter change rate according to the hydrofluoric acid purification equipment state data, and dividing the hydrofluoric acid purification equipment state data into a parameter rising interval data set and a parameter falling interval data set according to the parameter change rate; Performing fuzzy processing on the parameter ascending interval data set to obtain an ascending interval fuzzy input value; Performing fuzzy processing on the parameter descending interval data set to obtain a descending interval fuzzy input value; Inputting the ascending interval fuzzy input value and the descending interval fuzzy input value into a fuzzy logic monitoring model for inference calculation to obtain first fuzzy state output information; Determine an adaptive factor according to the equipment operation time and batch, and adjust the first fuzzy state output information according to the adaptive factor to obtain second fuzzy state output information; The second fuzzy state output information is weighted and defuzzified through the equipment importance weight coefficient matrix to obtain a real-time state evaluation result.

5. The method for monitoring state data of a hydrofluoric acid purification process according to claim 4, characterized in that: The fuzzy processing is performed on the parameter ascending interval data set to obtain the ascending interval fuzzy input value, including: Set the domain range of temperature, pressure, flow rate and concentration to obtain the domain space of parameter change rate, and determine the normal operating range of temperature, pressure, flow rate and concentration to obtain the domain space of parameter value; The parameter change rate domain space is evenly divided according to five preset levels, and five triangular membership functions with vertex coordinates located at the two ends and the midpoint of the domain are constructed to obtain a fuzzy membership function set of the parameter change rate; The parameter value domain space is evenly divided according to five preset levels, and five triangle membership functions with vertex coordinates at the midpoints of each interval are constructed to obtain a fuzzy membership function set of the parameter value; Input each pair of parameter change rates in the parameter rising interval data set into the fuzzy membership function set of the parameter change rate to calculate, and obtain a change rate membership value matrix; at the same time, input the current parameter value in the parameter rising interval data set into the fuzzy membership function set of the parameter value to calculate, and obtain a parameter value membership value matrix; The change rate membership value matrix and the parameter value membership value matrix are combined to obtain an ascending interval fuzzy input value.

6. The method for monitoring state data of a hydrofluoric acid purification process according to claim 1, characterized in that: The device state transition probability is calculated according to the real-time state evaluation result, and the time-varying delay characteristics of the mixing tank temperature control, the distillation tower pressure response and the absorption tower concentration adjustment are simultaneously identified to obtain the hydrofluoric acid purification equipment state prediction information, including: Based on the real-time status assessment result, the operating status of each device is classified to obtain a device status classification data set; According to the device state classification data set, the frequency of each device operating state transitioning to other states is calculated and normalized to obtain the device state transition probability; Dynamically updating the device state transition probability to obtain a dynamic transfer matrix that can reflect the device state evolution law; Inputting the state data of the hydrofluoric acid purification equipment into a fast Fourier transform analysis algorithm, analyzing the delay period of the temperature control of the mixing tank, the delay period of the pressure response of the distillation tower, and the delay period of the concentration adjustment of the absorption tower, and obtaining the main delay period characteristics of each equipment; According to the main delay period characteristics of each device, a feedforward-feedback combined compensator of a transfer function is constructed to obtain a delay compensation model for each device; The dynamic transfer matrix and the delay compensation model are combined to obtain the state prediction information of the hydrofluoric acid purification equipment by calculating the state probability distribution at the future moment and considering the influence of the delay factor.

7. The method for monitoring state data of a hydrofluoric acid purification process according to claim 1, characterized in that: Based on the hydrofluoric acid purification equipment state prediction information, adaptive switching is performed between the normal mode, the anti-fluctuation mode, the fast recovery mode and the emergency protection mode to obtain a control strategy for the characteristics of the hydrofluoric acid purification process, including: Constructing a control mode switching condition matrix according to the hydrofluoric acid purification equipment state prediction information; Performing real-time calculation on the control mode switching condition matrix to obtain the currently applicable control mode type; Based on the currently applicable control mode type, the control algorithm parameters corresponding to the device state are selected, wherein the normal mode adopts the control algorithm of PID parameters, the anti-fluctuation mode adopts the fuzzy PID control algorithm, the fast recovery mode adopts the model predictive control strategy, and the emergency protection mode adopts the robust H∞ control algorithm; The control algorithm parameters corresponding to the device state are input into the state observer to estimate the current state of the system, and the output values ​​of the new and old control algorithms are calculated at the same time to obtain the dual-algorithm control quantity when the control mode is switched; The dual algorithm control quantity when the control mode is switched is weighted averaged by a transition function to obtain a control output signal with smooth transition; The performance of the smooth transition control output signal is evaluated through a control mode effectiveness evaluation index to obtain a performance evaluation result, and the performance evaluation result is fed back to a control mode switching condition matrix to obtain a control strategy for the characteristics of the hydrofluoric acid purification process.

Images