Intelligent safety monitoring method for distillation tower based on neural network
Through intelligent safety monitoring methods based on neural networks, real-time collection and analysis of distillation tower data is solved, the safety hazards of distillation units in chemical production is improved, and the safety and efficiency of distillation towers are improved, reducing manual intervention and Accident risk.
Patent Information
- Application Number
- CN202211255785.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-13
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
There are great safety hazards for distillation units during chemical production. The existing safety analysis mainly relies on manual labor, with low accuracy and efficiency, and it is difficult to monitor and adjust process parameters in a timely manner, resulting in high accident risk.
Using intelligent security monitoring method based on neural networks, we can collect the working condition data of the distillation tower in real time, calculate safety indicator parameters, train neural network models, monitor and output the safety status of the distillation tower in real time, free up labor, and improve safety and efficiency. .
Real-time safety monitoring of distillation tower operating conditions is achieved, the safety and efficiency of distillation operations are improved, manual intervention is reduced, and accident risk is reduced.
Abstract
Description
Technical Field
[0001] The invention relates to the field of distillation tower safety monitoring, and in particular to a neural network-based intelligent distillation tower safety monitoring method. Background Art
[0002] Chemical production is more dangerous than other production processes. Most of the raw materials, intermediates and final products in the chemical production process are toxic, flammable, explosive, corrosive and other dangerous. When using, storing and producing these substances, they must be treated with caution, otherwise once an accident occurs, the harm is extremely great.
[0003] Chemical production requires a lot of operations and complex processing steps from raw materials to final products, and multiple separations, purifications or reactions are required to complete production. During the processing, many process parameters are very high, such as temperature, pressure and other parameters. If the process parameters change and cannot be adjusted or stopped in time, it will cause the loss of raw materials at the least, and even explosions and other major accidents that threaten life safety. Therefore, how to intelligently control the safety of the distillation process is of paramount importance.
[0004] At present, modern chemical production processes generally use industrial control systems such as DCS and PLC, and use highly automated operating systems to replace operator operations. The control system directly reads the data of instruments and meters in the production process, and performs start and stop operations on the host computer. The reliability of control has also been effectively improved. However, operating in the production process is not a simple link. In addition to simple operations that do not affect production, every other operation is accompanied by changes in regional or even entire plant equipment. A large amount of materials and equipment need to be processed. Any omissions or imperfect processing may lead to accidents. Therefore, producers attach great importance to operations in the production process, and each step must have a plan before it can be carried out.
[0005] However, due to the widespread safety hazards in distillation unit operations, safety analysis is often done manually, which is inaccurate and inefficient. Summary of the invention
[0006] The present invention provides a neural network-based intelligent safety monitoring method for a distillation tower to solve at least one of the above technical problems.
[0007] To solve the above problems, as one aspect of the present invention, a method for intelligent safety monitoring of a distillation tower based on a neural network is provided, comprising:
[0008] Step 1, data collection: real-time operating condition collection is performed on the distillation tower, and multiple operating condition data are recorded for later neural network model training. Each operating condition data includes multiple operating condition parameters;
[0009] Step 2, data summary: summarize all the collected working condition data;
[0010] Step 3, calculation of key safety indicators: various safety analyses are performed on the distillation tower, and the safety indicator parameters corresponding to each operating condition data are calculated as the distillation tower safety assessment parameters, and stored in the data summary table;
[0011] Step 4, neural network calculation: use the easily measurable data, including the basic parameters of the tower and the material conditions at each sampling point, as the input layer variables, and the data that are not easy to collect and the safety index parameters as the target to train the neural network model and obtain a model that meets the requirements;
[0012] Step 5, distillation tower operating condition input: Based on the industrial Internet, various easily measurable operating parameters of the distillation tower, that is, various input layer variables, are collected in real time and transmitted to the host;
[0013] Step 6, model calculation and output: Use the model obtained by neural network training to calculate the input parameters, obtain the output safety parameters, and print the signal of whether the distillation tower operating condition is in a safe state on the screen to complete the real-time monitoring of the distillation tower safety information.
[0014] Preferably, the operating parameters include the following 41 items: reflux ratio, distillate feed ratio, number of plates, feed position, feed temperature, feed pressure, feed flow rate, feed heavy component molar fraction, feed light component molar fraction, distillate temperature, distillate pressure, distillate flow rate, distillate heavy component molar fraction, distillate light component molar fraction, still liquid temperature, still liquid pressure, still liquid flow rate, still liquid heavy component molar fraction, still liquid light component molar fraction, first plate temperature, first plate pressure, first plate liquid phase flow rate, first plate vapor phase flow rate, 30% plate temperature, 3 Pressure at 0% plate, liquid phase flow rate at 30% plate, vapor phase flow rate at 30% plate, temperature at 60% plate, pressure at 60% plate, liquid phase flow rate at 60% plate, vapor phase flow rate at 60% plate, temperature of last plate, pressure of last plate, liquid phase flow rate of last plate, vapor phase flow rate of last plate, temperature of feed plate, pressure of feed plate, liquid phase flow rate of feed plate, vapor phase flow rate of feed plate, heat load of condenser, heat load of reboiler, among which, temperature, pressure, flow rate etc. can be directly read through instruments on distillation tower equipment, while molar fraction needs to be obtained through sampling and analysis.
[0015] Preferably, the safety indicator parameters in step 3 include a fire and explosion damage index FEDI.
[0016] Preferably, the data that are easy to measure include: basic parameters of the tower and material conditions at each sampling point; the data that are not easy to collect include: molar fraction of each component and safety index parameters.
[0017] Preferably, the basic parameters of the tower include: reflux ratio, distillate feed ratio, number of plates, feed position; the material conditions at each sampling point include: temperature, pressure, flow rate at each sampling point.
[0018] The beneficial effect of the invention is that the safety status of the distillation tower can be output in real time while monitoring the operating parameters of the distillation tower, freeing up manual labor and making the distillation operation safer and more efficient. DETAILED DESCRIPTION
[0019] The following is a detailed description of embodiments of the present invention, but the present invention can be implemented in many different ways as defined and covered by the claims.
[0020] In order to make the distillation process safer, the safety analysis more accurate and efficient, and solve the common safety hazards in the current distillation unit operation, the present invention creates a neural network-based distillation tower intelligent safety monitoring technology, which can not only free the distillation tower operation manual, but also make the distillation process safer.
[0021] The present invention provides a neural network-based intelligent safety monitoring technology for a distillation tower, which relates to a safety monitoring technology for a distillation tower, in particular to a safety monitoring technology for a binary vapor phase feed distillation tower, and specifically comprises the following steps:
[0022] Step 1: Data Collection
[0023] Collect real-time working conditions on the distillation tower and record the data. The data to be collected include the following 41 items:
[0024] (1) Reflux ratio, (2) Distillate feed ratio, (3) Number of plates, (4) Feed position, (5) Feed temperature, (6) Feed pressure, (7) Feed flow rate, (8) Feed heavy component molar fraction, (9) Feed light component molar fraction, (10) Distillate temperature, (11) Distillate pressure, (12) Distillate flow rate, (13) Distillate heavy component molar fraction, (14) Distillate light component molar fraction, (15) Still liquid temperature, (16) Still liquid pressure, (17) Still liquid flow rate, (18) Still liquid heavy component molar fraction, (19) Still liquid light component molar fraction, (20) First stage temperature, (21) First stage pressure, (22) First stage liquid flow rate, (23) ) vapor phase flow rate at the first tower plate, (24) temperature at the 30% tower plate, (25) pressure at the 30% tower plate, (26) liquid phase flow rate at the 30% tower plate, (27) vapor phase flow rate at the 30% tower plate, (28) temperature at the 60% tower plate, (29) pressure at the 60% tower plate, (30) liquid phase flow rate at the 60% tower plate, (31) vapor phase flow rate at the 60% tower plate, (32) temperature of the last tower plate, (33) pressure of the last tower plate, (34) liquid phase flow rate of the last tower plate, (35) vapor phase flow rate of the last tower plate, (36) temperature of the feed tower plate, (37) pressure of the feed tower plate, (38) liquid phase flow rate of the feed tower plate, (39) vapor phase flow rate of the feed tower plate, (40) condenser heat load, (41) reboiler heat load.
[0025] The temperature, pressure, flow rate, etc. can be directly read through the instruments on the distillation tower equipment, while the molar fraction needs to be obtained through sampling and analysis.
[0026] Ultimately, about 500 pieces of operating condition data are needed for subsequent neural network model training.
[0027] Step 2: Data aggregation
[0028] Summarize all the collected working condition data.
[0029] Step 3: Calculation of key safety indicators
[0030] Perform various safety analyses on the distillation tower, and calculate the safety index parameters corresponding to each data (such as the fire and explosion injury index FEDI) as the distillation tower safety assessment parameters, and store them in the data summary table.
[0031] Step 4: Neural Network Calculation
[0032] The easily measurable data, including the basic parameters of the tower [such as reflux ratio (1), distillate feed ratio (2), number of plates (3), feed position (4)], material conditions at each sampling point [such as temperature at each sampling point (5, 10, 15, 20, 24, 28, 32, 36), pressure (6, 11, 16, 21, 25, 29, 33, 37), flow rate (7, 12, 17, 22, 23, 26, 27, 30, 31, 34, 35, 38, 39)], are used as input layer variables. The data that are not easy to collect [such as the molar fraction of each component (8, 9, 13, 14, 18, 19)] and safety index parameters [such as FEDI, etc.] are used as targets for neural network model training to obtain a model that meets the requirements.
[0033] Step 5: Input distillation tower operating conditions
[0034] Based on the industrial Internet, various easily measurable operating parameters of the distillation tower (i.e. various input layer variables) are collected in real time and transmitted to the host.
[0035] Step 6: Model calculation and output
[0036] The model obtained from the previous neural network training is used to calculate the input parameters to obtain the output safety parameters, and a signal indicating whether the distillation tower is in a safe state is printed on the screen to complete real-time monitoring of the distillation tower's safety information.
[0037] The beneficial effect of the invention is that the safety status of the distillation tower can be output in real time while monitoring the operating parameters of the distillation tower, freeing up manual labor and making the distillation operation safer and more efficient.
[0038] Example :
[0039] In the field of ethylene distillation, the output fire and explosion damage index FEDI is calculated as follows:
[0040] FEDI=4.76*(Damage Potential)^0.333
[0041] Based on this calculation, the FEDI index corresponding to each working condition data is used as the output layer variable, which can effectively evaluate the explosion risk in the distillation tower operation process. If there is a safety hazard, the unqualified FEDI index and its corresponding risk level will be immediately output on the screen and an alarm will be triggered.
[0042] In one embodiment:
[0043] The input layer variables in the input acquisition process are (1) reflux ratio, (2) distillate feed ratio, (3) number of plates, (4) feed position, (5) feed temperature, (6) feed pressure, (7) feed flow rate, (8) feed heavy component mole fraction, (9) feed light component mole fraction, (10) distillate temperature, (11) distillate pressure, (12) distillate flow rate, (13) distillate heavy component mole fraction, (14) distillate light component mole fraction, (15) still liquid temperature, (16) still liquid pressure, (17) still liquid flow rate, (18) still liquid heavy component mole fraction, (19) still liquid light component mole fraction, (20) first plate temperature, (21) first plate pressure, (22) first plate liquid phase flow rate, (23) vapor phase flow rate of the first tower plate, (24) temperature at the 30% tower plate, (25) pressure at the 30% tower plate, (26) liquid phase flow rate at the 30% tower plate, (27) vapor phase flow rate at the 30% tower plate, (28) temperature at the 60% tower plate, (29) pressure at the 60% tower plate, (30) liquid phase flow rate at the 60% tower plate, (31) vapor phase flow rate at the 60% tower plate, (32) temperature of the last tower plate, (33) pressure of the last tower plate, (34) liquid phase flow rate of the last tower plate, (35) vapor phase flow rate of the last tower plate, (36) temperature of the feed tower plate, (37) pressure of the feed tower plate, (38) liquid phase flow rate of the feed tower plate, (39) vapor phase flow rate of the feed tower plate, (40) condenser heat load, (41) reboiler heat load.
[0044] The output variables are safety parameters such as FEDI, which can be directly calculated from the process data of the distillation tower. The relationship between the input variables and the output variables is established by training the neural network, and the output form is to print the safety parameters on the screen, which can effectively free up manual labor.
[0045] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A neural network-based intelligent safety monitoring method for a distillation tower, characterized in that: include: Step 1, data collection: real-time operating condition collection is performed on the distillation tower, and multiple operating condition data are recorded for later neural network model training. Each operating condition data includes multiple operating condition parameters; Step 2, data summary: summarize all the collected working condition data; Step 3, safety index parameter calculation: perform various safety analyses on the distillation tower, and calculate the safety index parameters corresponding to each operating condition data as the distillation tower safety assessment parameters, and store them in the data summary table; Step 4, neural network calculation: use the easily measurable data, including the basic parameters of the tower and the material conditions at each sampling point, as input parameters, and the data that are not easy to collect and the safety index parameters as targets to train the neural network model and obtain a model that meets the requirements; The data that are not easy to collect include: the molar fraction of each component; the basic parameters of the tower include: reflux ratio, distillate feed ratio, number of plates, feed position; the material conditions of each sampling point include: temperature, pressure, flow rate of each sampling point; Step 5, distillation tower operating condition input: Based on the industrial Internet, the data that can be easily measured by the distillation tower, that is, various input parameters, are collected in real time and transmitted to the host; Step 6, model calculation and output: Use the model obtained by neural network training to calculate the input parameters, obtain the output safety index parameters, and print the signal of whether the distillation tower operating condition is in a safe state on the screen to complete the real-time monitoring of the distillation tower safety information.
2. The intelligent safety monitoring method for a distillation tower based on a neural network according to claim 1, characterized in that: The operating parameters include the following 41 items: reflux ratio, distillate feed ratio, number of plates, feed position, feed temperature, feed pressure, feed flow rate, feed heavy component mole fraction, feed light component mole fraction, distillate temperature, distillate pressure, distillate flow rate, distillate heavy component mole fraction, distillate light component mole fraction, still liquid temperature, still liquid pressure, still liquid flow rate, still liquid heavy component mole fraction, still liquid light component mole fraction, first plate temperature, first plate pressure, first plate liquid phase flow rate, first plate vapor phase flow rate, 30% plate temperature, 30% Pressure at the tower plate, liquid phase flow rate at 30% of the tower plate, vapor phase flow rate at 30% of the tower plate, temperature at 60% of the tower plate, pressure at 60% of the tower plate, liquid phase flow rate at 60% of the tower plate, vapor phase flow rate at 60% of the tower plate, temperature of the last tower plate, pressure of the last tower plate, liquid phase flow rate of the last tower plate, vapor phase flow rate of the last tower plate, temperature of the feed tower plate, pressure of the feed tower plate, liquid phase flow rate of the feed tower plate, vapor phase flow rate of the feed tower plate, heat load of the condenser, heat load of the reboiler, among which the temperature, pressure and flow rate can be directly read through the instruments on the distillation tower equipment, while the mole fraction needs to be obtained through sampling and analysis.
3. The intelligent safety monitoring method for a distillation tower based on a neural network according to claim 1, characterized in that: The safety indicator parameters in step 3 include the fire and explosion damage index FEDI.
Citation Information
Patent Citations
Multi-sensor data fusion early warning method for monitoring electric transmission line tower
CN103557884A
Radar monitoring method and device for long-term security risk of chemical equipment
CN105389595A
Measuring method of pulverized coal concentration
CN106124373A
Equipment risk early-warning method and device, terminal equipment and storage medium
CN109269556A
Concentration soft measurement method of dividing wall rectifying tower
CN111474856A
