Fractionating tower production operation optimization method, device and equipment and storage medium
By integrating prediction and optimization models and utilizing sub-prediction models based on multiple machine learning algorithms, the operating status of the fractionation tower is evaluated in real time and process parameters are optimized. This solves the problem of the inability to optimize the production operation of the fractionation tower in a timely and accurate manner, thereby improving production efficiency and product quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- PETROCHINA CO LTD
- Filing Date
- 2024-11-04
- Publication Date
- 2026-05-08
AI Technical Summary
In existing technologies, the production and operation of fractionation towers cannot be optimized in a timely and accurate manner, resulting in high-value products being mixed with low-value products, which affects production efficiency.
By establishing an integrated prediction model, utilizing sub-prediction models of multiple machine learning algorithms, and combining process parameter rating and optimization models, the operating status of the fractionation tower is evaluated in real time, and the target process parameters are optimized to minimize the carbon number conversion rate.
It enables timely and accurate optimization of the fractionation tower production process, improves production efficiency and product quality, and solves the problem that existing technologies cannot optimize in a timely and accurate manner.
Smart Images
Figure CN121998791A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of fractionation tower technology, and in particular to a method, apparatus, equipment and storage medium for optimizing the production and operation of a fractionation tower. Background Technology
[0002] Currently, in the fractionation towers for naphtha and jet fuel, the overlapping of distillation ranges and the mixing or overlap of products with different prices result in high-value products being mixed with low-value products, affecting production efficiency. Therefore, the precise control of the extraction temperature of side-stream products and the operating parameters at the top of the tower has become crucial.
[0003] In related technologies, expert subjective methods are used to optimize the production and operation of distillation towers. However, due to the complexity of distillation tower process control, there is a delay in the target response, which means that optimization space needs to be identified in a timely manner, otherwise the best adjustment time will be missed.
[0004] However, this technical method is difficult for ordinary operators to master and apply, which greatly reduces the accuracy and feasibility of process optimization, resulting in the inability to optimize the production and operation of the distillation tower in a timely and accurate manner. Summary of the Invention
[0005] This invention provides a method, apparatus, equipment, and storage medium for optimizing the production and operation of a fractionation tower, which can solve the technical problem in the prior art that it is impossible to optimize the production and operation of a fractionation tower in a timely and accurate manner.
[0006] To achieve the above objectives, the present invention provides the following technical solution:
[0007] In a first aspect, embodiments of the present invention provide a method for optimizing the production and operation of a distillation tower, the method comprising:
[0008] Based on the process parameters corresponding to m fractionation towers, determine the production operation status rating of m fractionation towers, where m is a positive integer;
[0009] Based on the production operation status rating, the target process parameters corresponding to the target fractionation tower to be optimized are determined.
[0010] The target process parameters are input into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model. The integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms.
[0011] An optimization model is established with the goal of minimizing the carbon number conversion rate. Based on the optimization model, the production operation of the target fractionation tower to be optimized is optimized.
[0012] Optionally, the step of determining the production operation status rating of m fractionation towers based on the process parameters corresponding to m fractionation towers includes:
[0013] Determine the index type corresponding to each process parameter for each of the m fractionation towers;
[0014] For each type of indicator, the process parameters are subjected to benefit-type indicator homogenization and vector normalization to obtain a normalized decision matrix;
[0015] The index weights corresponding to each process parameter are determined based on the objective weighting method, and the weighted decision matrix is determined based on the index weights and the normalized decision matrix.
[0016] The optimal and worst solutions are determined based on the weighted decision matrix, and the Euclidean distance between the process parameters corresponding to each fractionation tower and the optimal and worst solutions is calculated.
[0017] The relative proximity of each fractionation tower is calculated based on the Euclidean distance, and the production operation status rating of each fractionation tower is determined based on the relative proximity.
[0018] Optionally, the process parameters include: bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate;
[0019] The indicator types include: benefit-type indicators, cost-type indicators, intermediate indicators, and range-type indicators.
[0020] Optionally, the step of performing benefit-type index normalization and vector normalization on the process parameters corresponding to each index type to obtain a normalized decision matrix includes:
[0021] By using the conversion formulas corresponding to cost-type indicators, intermediate indicators, and range-type indicators, the process parameters corresponding to each indicator type are converted into benefit-type indicators.
[0022] The original data matrix after the benefit-type indicators were aligned was subjected to vector normalization to obtain the normalized decision matrix.
[0023] Optionally, the sub-prediction model includes at least one of the following: decision tree model, random forest model, and support vector machine model.
[0024] Optionally, the step of inputting the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model includes:
[0025] The target process parameters are input into each sub-prediction model included in the integrated prediction model to obtain the sub-carbon number conversion rate corresponding to the sub-sideline extraction temperature output by each sub-prediction model.
[0026] The carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
[0027] Optionally, the step of weighted averaging the sub-carbon number conversion rates corresponding to the sub-sideline extraction temperatures output by each sub-prediction model to obtain the carbon number conversion rate corresponding to the sideline extraction temperatures output by the integrated prediction model includes:
[0028] The evaluation metrics for each sub-prediction model are determined based on the test set;
[0029] The weights corresponding to each sub-prediction model are determined based on the evaluation indicators.
[0030] Based on the weights, the carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
[0031] Secondly, embodiments of the present invention provide a fractionation tower production operation optimization device, the device comprising:
[0032] The information determination module is configured to determine the production operation status rating of m fractionation towers based on the process parameters corresponding to the m fractionation towers, where m is a positive integer; and to determine the target process parameters corresponding to the target fractionation tower to be optimized based on the production operation status rating.
[0033] The information acquisition module is configured to input the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model, wherein the integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms.
[0034] An optimization model building module is configured to establish an optimization model with the goal of minimizing the carbon number conversion rate, and to optimize the production operation of the target fractionation tower to be optimized based on the optimization model.
[0035] Thirdly, embodiments of the present invention also provide an electronic device, including: a memory and a processor; the processor is used to read and execute a computer program stored in the memory to implement the steps of the aforementioned method for optimizing the production and operation of a distillation tower.
[0036] Fourthly, embodiments of the present invention also provide a computer storage medium storing computer-executable instructions, which, when executed, implement the steps of the aforementioned method for optimizing the production and operation of a distillation tower.
[0037] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the aforementioned method for optimizing the production and operation of a distillation tower.
[0038] The beneficial effects of the technical solutions provided by the embodiments of the present invention include:
[0039] This invention first assesses the production and operation status of each fractionation tower based on real-time collected process parameters, allowing for the timely selection of towers with optimization potential. Then, an integrated prediction model composed of multiple sub-prediction models accurately predicts the carbon conversion rate corresponding to the side-stream extraction temperature of the fractionation tower to be optimized. This leads to the determination of the optimal process parameters corresponding to the minimum carbon conversion rate. This enables staff to manage and optimize the fractionation tower's production process in a timely and accurate manner, achieving more efficient production and higher-quality product output. It solves the technical problem in related technologies where timely and accurate optimization of fractionation tower production and operation is impossible. Attached Figure Description
[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0041] Figure 1 This is a schematic flowchart of an embodiment of the fractionation tower production and operation optimization method of the present invention;
[0042] Figure 2 for Figure 1 A detailed flowchart of step S10;
[0043] Figure 3 for Figure 1 A detailed flowchart of step S30;
[0044] Figure 4 This is a functional module diagram of an embodiment of the fractionation tower production operation optimization device of the present invention;
[0045] Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation
[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0047] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0048] In a first aspect, embodiments of the present invention provide a method for optimizing the production and operation of a distillation tower.
[0049] In one embodiment, reference is made to Figure 1 , Figure 1 This is a schematic flow diagram of an embodiment of the fractionation tower production operation optimization method of the present invention. Figure 1 As shown, the optimization methods for the production and operation of the distillation tower include:
[0050] Step S10: Determine the production operation status rating of m fractionation towers based on the process parameters corresponding to m fractionation towers, where m is a positive integer;
[0051] In some specific embodiments, reference is made to Figure 2 , Figure 2 for Figure 1 A detailed flowchart of step S10. (See attached diagram.) Figure 2 As shown, step S10 includes:
[0052] Step S101: Determine the index type corresponding to each process parameter of each of the m fractionation towers;
[0053] In this embodiment, a certain oil refining process may include multiple fractionation towers. For each of the m fractionation towers, it is first necessary to obtain multiple process parameters characterizing its production process. These process parameters include: bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate. These process parameters can be obtained from multiple sensors.
[0054] If there are m fractionation columns, and each fractionation column has n process parameters x j (j=1,2,…,n), then the original data matrix corresponding to the decision problem of n process parameters for m fractionation towers is as follows:
[0055]
[0056] It is important to note that when obtaining multiple process parameters of a fractionation column, these process parameters need to be preprocessed first. For example, based on the actual operation, maximum, minimum, empty, and abnormal process values should be removed. Abnormal process values refer to values outside the reasonable range of process parameters, which are usually errors caused by instrument drift or abnormal production.
[0057] Next, it is necessary to determine the index type for each process parameter. Typically, index types include: benefit-oriented indexes, cost-oriented indexes, intermediate-level indexes, and range-oriented indexes. The characteristics of each index type are shown in Table 1. In this embodiment, the bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate are all intermediate-level indexes.
[0058] Table 1
[0059] Indicator Type Indicator characteristics Benefit indicators The bigger the better Cost indicators The smaller the better intermediate indicators The more stable it is to a certain value, the better. Interval indicators The better it falls within a certain range
[0060] Step S102: Perform benefit-type index homogenization and vector normalization on the process parameters corresponding to each index type to obtain a normalized decision matrix;
[0061] In some specific embodiments, step S102 includes:
[0062] By using the conversion formulas corresponding to cost-type indicators, intermediate indicators, and range-type indicators, the process parameters corresponding to each indicator type are converted into benefit-type indicators.
[0063] The original data matrix after the benefit-type indicators were aligned was subjected to vector normalization to obtain the normalized decision matrix.
[0064] In this embodiment, after determining the index type corresponding to each process parameter of each of the m fractionation towers, the optimal ordering method (Technique for Order Preference by Similarity to Ideal Solution, or TOPSIS) is used to evaluate the production and operation status of each fractionation tower. However, the TOPSIS method uses a distance scale to measure the difference between samples. Since the numerical values of different index attributes fluctuate significantly, scale confusion can occur during attribute comparison. Therefore, it is necessary to homogenize the index attributes. In this embodiment, a benefit-type index homogenization method is used to convert cost-type, intermediate, and interval-type indicators into benefit-type indicators.
[0065] Specifically, the formula for converting intermediate indicators into benefit-type indicators is as follows:
[0066]
[0067] The formula for converting cost-based indicators into benefit-based indicators is shown below:
[0068] x ih =x max -x ij i = 1, 2, ..., m, j = 1, 2, ..., n
[0069] The formula for converting interval-type indicators into benefit-type indicators is shown below:
[0070]
[0071] M = max(ax) min ,x max -b)
[0072] In the formula, x ij Let x represent the j-th process parameter corresponding to the i-th fractionation column. best Let x represent the optimal stable value of the j-th process parameter for all fractionation columns. max x represents the maximum value of the j-th process parameter corresponding to all fractionation columns. min Let represent the minimum value of the j-th process parameter corresponding to all fractionation columns, where a and b are the optimal ranges for the j-th process parameter corresponding to all fractionation columns, and M is a constant.
[0073] Then, vector normalization is performed on each data point in the original data matrix after the benefit-type indicator normalization process to obtain the normalized decision matrix. Since the indicators have different units, they need to be standardized to address the problem of direct comparison due to different units. The normalized decision matrix Y can be obtained through vector normalization, and its calculation formula is as follows:
[0074]
[0075] In the formula, y ij Let x represent the data in the i-th row and j-th column of the normalized decision matrix. ij This represents the data in the i-th row and j-th column of the original data matrix, where m represents the number of fractionation towers and n represents the number of process parameters.
[0076] The normalized decision matrix is as follows:
[0077]
[0078] Step S103: Determine the index weight corresponding to each process parameter based on the objective weighting method, and determine the weighted decision matrix according to the index weight and the normalized decision matrix;
[0079] In this embodiment, considering the large number of process parameters, objective weights are used instead of subjective weights in the traditional TOPSIS method to improve the scientific rigor and rationality of the TOPSIS comprehensive evaluation. The specific calculation steps are as follows:
[0080] Calculate the proportion of each process parameter in all fractionation columns, and for the j-th parameter x in the i-th fractionation column... ij The proportion p of the sum of the j-th index of all fractionation columns ij The formula is shown below:
[0081]
[0082] Similarly, the proportion of each process parameter in all fractionation towers can be calculated.
[0083] Calculate the entropy value E of the j-th process parameter. j :
[0084] Among them, E j >0, and when p ij =0 lnp ij =0.
[0085] Calculate the entropy weight w of the j-th process parameter index j :
[0086]
[0087] For the j-th process parameter, the greater the difference in the process parameter value, the greater the impact on the production operation status rating, and the smaller the entropy value.
[0088] Finally, the entropy weight vector of each process parameter is obtained as: W = (w1, w2, ..., w n ) T Multiplying the entropy weight vector W by the normalized matrix Y yields the weighted decision matrix Z:
[0089]
[0090] Where T represents the inverted matrix, i = 1, 2, ..., m, j = 1, 2, ..., n.
[0091] Step S104: Determine the optimal and worst solutions based on the weighted decision matrix, and calculate the Euclidean distance between the process parameters corresponding to each fractionation tower and the optimal and worst solutions;
[0092] In this embodiment, the optimal and worst solutions are determined from the weighted decision matrix. The j-th process parameter index of the optimal solution is the maximum value among the j-th process parameter index values of all fractionation columns, and the j-th process parameter index of the worst solution is the minimum value among the j-th process parameter index values of all fractionation columns. The formula is as follows:
[0093]
[0094] in, This represents the j-th process parameter index of the optimal solution. This represents the j-th process parameter index of the worst solution.
[0095] Then, the Euclidean distance between the process parameters corresponding to each fractionation tower and the optimal and worst solutions is calculated, as shown in the following formula:
[0096]
[0097] in, This represents the Euclidean distance between the process parameters corresponding to the i-th fractionation column and the optimal solution. Z represents the Euclidean distance between the process parameters corresponding to the i-th fractionation column and the worst solution. ij This represents the data in the i-th row and j-th column of the weighted decision matrix.
[0098] Step S105: Calculate the relative proximity of each fractionation tower based on the Euclidean distance, and determine the production operation status rating of each fractionation tower based on the relative proximity.
[0099] In this embodiment, according to and The formula for calculating the relative closeness between the process parameters of each fractionation column and the optimal solution is shown below:
[0100]
[0101] The production and operation status of each fractionation tower is ranked according to the relative proximity value Ci. i A higher Ci value indicates a better operating condition for the fractionation column, while a lower Ci value indicates a worse operating condition. This can be determined based on the Ci value. i The fractionation columns were classified into three levels: normal, good, and abnormal, to conduct a qualitative rating of each fractionation column, as shown in Table 2.
[0102] Table 2
[0103] normal good abnormal [1,0.6) [0.6,0.4) [0.4,0.0}
[0104] Building upon the aforementioned embodiments, multiple sensors are used to collect process parameters such as bottom liquid level, pressure, temperature, top temperature, and feed flow rate during the operation of the fractionation towers. This allows for a comprehensive understanding of the towers' production and operational status. Subsequently, the aforementioned operational monitoring data is used to construct a standardized decision matrix, with weighted coefficients assigned to each indicator. A TOPSIS evaluation model is then established. Through these evaluation indicators, an accurate production and operational status rating for each fractionation tower can be obtained. These ratings can be visually displayed to staff using a radar chart, enabling them to monitor the real-time operational status of each tower and promptly identify towers exhibiting abnormal operating conditions, allowing for the implementation of optimization measures.
[0105] Step S20: Based on the production operation status rating, determine the target process parameters corresponding to the target fractionation tower to be optimized;
[0106] In this embodiment, after obtaining the production operation status rating of each fractionation column, the fractionation columns with an abnormal production operation status rating are identified as target fractionation columns to be optimized, and optimization measures are taken for these target fractionation columns. The corresponding process parameters for the target fractionation columns to be optimized are determined as the target process parameters.
[0107] Step S30: Input the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model. The integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms.
[0108] In some specific embodiments, reference is made to Figure 3 , Figure 3 for Figure 1 A detailed flowchart of step S30. (See attached diagram.) Figure 3 As shown, step S30 includes:
[0109] Step S301: Input the target process parameters into each sub-prediction model included in the integrated prediction model to obtain the sub-carbon number conversion rate corresponding to the sub-sideline extraction temperature output by each sub-prediction model;
[0110] In this embodiment, the ensemble prediction model consists of multiple sub-prediction models based on different machine learning algorithms. In some embodiments, the sub-prediction models include at least one of the following: decision tree model, random forest model, and support vector machine model. Decision tree model performs well in the initial feature selection due to its clear structure and robustness to noisy data; random forest model improves the stability and accuracy of prediction by integrating multiple decision trees; and support vector machine shows advantages in handling nonlinear problems due to its good generalization ability.
[0111] Each sub-prediction model, such as the decision tree model, random forest model, and support vector machine model, is pre-set with corresponding weights. The target process parameters (i.e., bottom temperature, top temperature, pressure, bottom level, top level, feed flow rate, etc.) of the target distillation column to be optimized are input into the decision tree model, random forest model, and support vector machine model, respectively. The decision tree model, random forest model, and support vector machine model output the carbon conversion rate corresponding to the sub-sideline extraction temperature of each target process parameter.
[0112] It is important to understand that process parameters such as bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate of the fractionation column will affect the tapping temperature of the fractionation column, and the tapping temperature will affect the carbon number conversion rate.
[0113] Step S302: Perform a weighted average of the carbon number conversion rates corresponding to the sub-side extraction temperatures output by each sub-prediction model to obtain the carbon number conversion rates corresponding to the side extraction temperatures output by the integrated prediction model.
[0114] In some specific embodiments, step S302 includes:
[0115] The evaluation metrics for each sub-prediction model are determined based on the test set;
[0116] The weights corresponding to each sub-prediction model are determined based on the evaluation indicators.
[0117] Based on the weights, the carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
[0118] In this embodiment, the sub-sideline extraction temperature and the corresponding carbon number conversion rate output by the sub-prediction model are weighted and averaged to obtain the final sideline extraction temperature and carbon number conversion rate.
[0119] In some embodiments, the method further includes: determining at least one evaluation index for each sub-prediction model based on a test set, and determining the weight corresponding to each sub-prediction model according to the at least one evaluation index for each sub-prediction model, wherein the weight corresponding to each sub-prediction model is used to perform a weighted average of the sub-carbon number conversion rates corresponding to the sub-sideline extraction temperatures output by multiple sub-prediction models.
[0120] Specifically, before applying the ensemble prediction model, each sub-prediction model within the ensemble prediction model needs to be trained and tested. First, a historical sample dataset of the fractionation column can be obtained, where each sample data includes historical process parameters and the corresponding historical side-stream extraction temperature and historical carbon number conversion rate. Then, the historical sample dataset is divided into a training dataset and a test dataset. The training dataset is used to train each sub-prediction model, and the test dataset is used to test each trained sub-prediction model.
[0121] During the testing phase, based on the test results of each sub-prediction model, its evaluation metrics can be assessed, such as accuracy, recall, and F1 score. The weights of each sub-prediction model can be determined based on one or more of these metrics. For example, assuming that the accuracies of decision trees, random forests, and support vector machines are 0.6, 0.85, and 0.95 respectively under the accuracy evaluation metric, the weights of each model can be calculated as follows:
[0122] The weights of the decision tree model are 0.6 / (0.6+0.85+0.95)=0.25;
[0123] The weights of the random forest model are 0.85 / (0.6+0.85+0.95)=0.354;
[0124] The weights of the support vector machine model are 0.95 / (0.6+0.85+0.95)=0.396.
[0125] In this way, appropriate weights can be assigned to each sub-prediction model based on its performance, so that their predictive capabilities can be comprehensively utilized in the integrated prediction model.
[0126] Based on the aforementioned embodiments, an integrated prediction model is constructed, which includes multiple sub-prediction models. Each sub-prediction model can predict the fractionation column separately, and the final optimization results are integrated. That is, after determining the weight corresponding to each sub-prediction model, the carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged based on the weight corresponding to each sub-prediction model, so as to obtain the carbon number conversion rate corresponding to the side extraction temperature output by the integrated prediction model. This solves the problem that traditional single models may perform poorly on certain data distributions, and provides more accurate guidance for fractionation column operation.
[0127] Step S40: Establish an optimization model with the goal of minimizing the carbon number conversion rate, and optimize the production operation of the target fractionation tower to be optimized based on the optimization model.
[0128] In this embodiment, an optimization model is established with the goal of minimizing the carbon number conversion rate. The optimization model also includes constraints, namely, upper and lower limit constraints on each process parameter. By solving this optimization model, the optimized target process parameters corresponding to minimizing the carbon number conversion rate can be determined. These optimized target process parameters can be displayed so that staff can optimize the target fractionation column to be optimized.
[0129] In this embodiment, the production operation status rating of m fractionation towers is determined according to the process parameters corresponding to m fractionation towers, where m is a positive integer; based on the production operation status rating, the target process parameters corresponding to the target fractionation tower to be optimized are determined; the target process parameters are input into an integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model, wherein the integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms; an optimization model is established with the goal of minimizing the carbon number conversion rate, and the production operation of the target fractionation tower to be optimized is optimized based on the optimization model. This embodiment first assesses the production and operation status of each fractionation tower based on real-time collected process parameters, allowing for the timely selection of towers with optimization potential. Then, an integrated prediction model composed of multiple sub-prediction models accurately predicts the carbon conversion rate corresponding to the side-stream extraction temperature of the fractionation tower to be optimized. This leads to the determination of the optimal process parameters corresponding to the minimum carbon conversion rate. This enables staff to manage and optimize the fractionation tower's production process in a timely and accurate manner, achieving more efficient production operation and higher-quality product output. It solves the technical problem in related technologies where timely and accurate optimization of fractionation tower production and operation is impossible.
[0130] Secondly, embodiments of the present invention also provide a fractionation tower production operation optimization device.
[0131] In one embodiment, reference is made to Figure 4 , Figure 4 This is a functional module diagram of an embodiment of the fractionation tower production operation optimization device of the present invention. Figure 4 As shown, the fractionation tower production operation optimization device includes:
[0132] The information determination module 10 is configured to determine the production operation status rating of m fractionation towers based on the process parameters corresponding to m fractionation towers, where m is a positive integer; and to determine the target process parameters corresponding to the target fractionation tower to be optimized based on the production operation status rating.
[0133] The information acquisition module 20 is configured to input the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model, wherein the integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms.
[0134] The optimization model building module 30 is configured to establish an optimization model with the goal of minimizing the carbon number conversion rate, and to optimize the production operation of the target fractionation tower to be optimized based on the optimization model.
[0135] Optionally, in one embodiment, the information determination module 10 is configured to:
[0136] Determine the index type corresponding to each process parameter for each of the m fractionation towers;
[0137] For each type of indicator, the process parameters are subjected to benefit-type indicator homogenization and vector normalization to obtain a normalized decision matrix;
[0138] The index weights corresponding to each process parameter are determined based on the objective weighting method, and the weighted decision matrix is determined based on the index weights and the normalized decision matrix.
[0139] The optimal and worst solutions are determined based on the weighted decision matrix, and the Euclidean distance between the process parameters corresponding to each fractionation tower and the optimal and worst solutions is calculated.
[0140] The relative proximity of each fractionation tower is calculated based on the Euclidean distance, and the production operation status rating of each fractionation tower is determined based on the relative proximity.
[0141] Optionally, in one embodiment, the process parameters include: bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate;
[0142] The indicator types include: benefit-type indicators, cost-type indicators, intermediate indicators, and range-type indicators.
[0143] Optionally, in one embodiment, the information determination module 10 is further configured to:
[0144] By using the conversion formulas corresponding to cost-type indicators, intermediate indicators, and range-type indicators, the process parameters corresponding to each indicator type are converted into benefit-type indicators.
[0145] The original data matrix after the benefit-type indicators were aligned was subjected to vector normalization to obtain the normalized decision matrix.
[0146] Optionally, in one embodiment, the sub-prediction model includes at least one of the following: decision tree model, random forest model, and support vector machine model.
[0147] Optionally, in one embodiment, the information acquisition module 20 is configured to:
[0148] The target process parameters are input into each sub-prediction model included in the integrated prediction model to obtain the sub-carbon number conversion rate corresponding to the sub-sideline extraction temperature output by each sub-prediction model.
[0149] The carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
[0150] Optionally, in one embodiment, the information acquisition module 20 is further configured to:
[0151] The evaluation metrics for each sub-prediction model are determined based on the test set;
[0152] The weights corresponding to each sub-prediction model are determined based on the evaluation indicators.
[0153] Based on the weights, the carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
[0154] The functions of each module in the above-mentioned fractionation tower production operation optimization device correspond to the steps in the above-mentioned fractionation tower production operation optimization method embodiment, and their functions and implementation processes will not be described in detail here.
[0155] Thirdly, embodiments of the present invention also provide an electronic device, the structure of which is as follows: Figure 5 As shown, it includes: a memory and a processor, wherein the processor is used to read and execute the computer program stored in the memory to implement the aforementioned method for optimizing the production and operation of a distillation tower.
[0156] Fourthly, embodiments of the present invention also provide a computer storage medium storing computer-executable instructions, which, when executed, implement the aforementioned method for optimizing the production and operation of a distillation tower.
[0157] Fifthly, embodiments of the present invention provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the fractionation tower production operation optimization method embodiments described above, and can achieve the same technical effects. To avoid repetition, it will not be described again here.
[0158] Finally, it should be noted that while some processes described in the embodiments of the present invention include multiple operations or steps that appear in a specific order, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of the present invention, or may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.
[0159] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for optimizing the production and operation of a distillation tower, characterized in that, The method includes: Based on the process parameters corresponding to m fractionation towers, determine the production operation status rating of m fractionation towers, where m is a positive integer; Based on the production operation status rating, the target process parameters corresponding to the target fractionation tower to be optimized are determined. The target process parameters are input into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model. The integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms. An optimization model is established with the goal of minimizing the carbon number conversion rate. Based on the optimization model, the production operation of the target fractionation tower to be optimized is optimized.
2. The method for optimizing the production and operation of a fractionation tower according to claim 1, characterized in that, The step of determining the production operation status rating of m fractionation towers based on the process parameters corresponding to m fractionation towers includes: Determine the index type corresponding to each process parameter for each of the m fractionation towers; For each type of indicator, the process parameters are subjected to benefit-type indicator homogenization and vector normalization to obtain a normalized decision matrix; The index weights corresponding to each process parameter are determined based on the objective weighting method, and the weighted decision matrix is determined based on the index weights and the normalized decision matrix. The optimal and worst solutions are determined based on the weighted decision matrix, and the Euclidean distance between the process parameters corresponding to each fractionation tower and the optimal and worst solutions is calculated. The relative proximity of each fractionation tower is calculated based on the Euclidean distance, and the production operation status rating of each fractionation tower is determined based on the relative proximity.
3. The method for optimizing the production and operation of a distillation tower according to claim 2, characterized in that, The process parameters include: bottom temperature, top temperature, pressure, bottom liquid level, top liquid level, and feed flow rate; The indicator types include: benefit-type indicators, cost-type indicators, intermediate indicators, and range-type indicators.
4. The method for optimizing the production and operation of a fractionation tower according to claim 2, characterized in that, The step of performing benefit-type index normalization and vector normalization on the process parameters corresponding to each index type to obtain a normalized decision matrix includes: By using the conversion formulas corresponding to cost-type indicators, intermediate indicators, and range-type indicators, the process parameters corresponding to each indicator type are converted into benefit-type indicators. The original data matrix after the benefit-type indicators were aligned was subjected to vector normalization to obtain the normalized decision matrix.
5. The method for optimizing the production and operation of a fractionation tower according to claim 1, characterized in that, The sub-prediction model includes at least one of the following: decision tree model, random forest model, and support vector machine model.
6. The method for optimizing the production and operation of a fractionation tower according to claim 5, characterized in that, The step of inputting the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model includes: The target process parameters are input into each sub-prediction model included in the integrated prediction model to obtain the sub-carbon number conversion rate corresponding to the sub-sideline extraction temperature output by each sub-prediction model. The carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
7. The method for optimizing the production and operation of a distillation tower according to claim 6, characterized in that, The step of weighted averaging the sub-carbon number conversion rates corresponding to the sub-sideline extraction temperatures output by each sub-prediction model to obtain the carbon number conversion rate corresponding to the sideline extraction temperatures output by the integrated prediction model includes: The evaluation metrics for each sub-prediction model are determined based on the test set; The weights corresponding to each sub-prediction model are determined based on the evaluation indicators. Based on the weights, the carbon number conversion rate corresponding to the sub-side extraction temperature output by each sub-prediction model is weighted and averaged to obtain the carbon number conversion rate corresponding to the side-line extraction temperature output by the integrated prediction model.
8. A device for optimizing the production and operation of a distillation tower, characterized in that, The device includes: The information determination module is configured to determine the production operation status rating of m fractionation towers based on the process parameters corresponding to the m fractionation towers, where m is a positive integer; and to determine the target process parameters corresponding to the target fractionation tower to be optimized based on the production operation status rating. The information acquisition module is configured to input the target process parameters into the integrated prediction model to obtain the carbon number conversion rate corresponding to the side-stream extraction temperature output by the integrated prediction model, wherein the integrated prediction model includes multiple sub-prediction models based on different machine learning algorithms. An optimization model building module is configured to establish an optimization model with the goal of minimizing the carbon number conversion rate, and to optimize the production operation of the target fractionation tower to be optimized based on the optimization model.
9. An electronic device, characterized in that, include: Memory and processor; The processor is configured to read and execute the computer program stored in the memory to implement the steps of the fractionation tower production operation optimization method as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed, implement the steps of the fractionation tower production operation optimization method as described in any one of claims 1-7.