A method and system for measuring catalyst carbon content in a methanol to olefins process

Abstract

The present application provides a method and system for measuring the carbon content of catalysts in a methanol to olefins process, wherein the reaction data is input into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value. The carbon content prediction model is established based on a gradient boosting regression algorithm, and is trained using historical reaction data and corresponding catalyst carbon content true values ​​as training data. There is no need to sample the catalyst, and dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit is achieved. By monitoring the data changes of relevant production variables, online and accurate measurement of the catalyst carbon content of the DMTO unit is achieved, which is used to evaluate catalyst activity and process progress, and assist in DMTO production decision-making.

Description

Technical Field

[0001] The present application relates to the technical field of key process parameter measurement in the coal chemical industry, and specifically to a method and system for measuring catalyst carbon content in a methanol to olefins process. Background Art

[0002] The service life and reaction activity of a catalyst are directly related to its economic benefits, and carbon deposition is one of the main factors affecting this. Since carbon deposition is unavoidable, the industry typically uses carbon burn-off to partially restore catalyst activity. Hydrocarbon pool theory suggests that in the early stages of a chemical reaction, reactants are consumed to form polymethylbenzenes (PMBs) on active sites (a carbon deposition process), which serve as catalytic centers for the reaction. Therefore, there is an optimal carbon content for the catalyst used in DMTO production plants, and varying the carbon content can be used to increase product yield and regulate product distribution.

[0003] Catalyst carbon determination can be divided into direct measurement methods and indirect measurement methods. The former involves sampling the catalyst in the reactor and performing offline detection, while the latter uses acoustic, optical, and electrical methods to obtain the relationship between the catalyst carbon content and changes in its physical properties and achieves online determination through modeling. Huang Jujun et al. used gas chromatography to determine the amount of catalyst carbon deposits and used a graphite-alumina mixture combustion standard curve to determine the amount of carbon deposits on ethylbenzene dehydrogenation catalysts. This method has high accuracy but takes a long time to detect. Zhang Feng et al. first measured the color characteristics of the catalyst online, then sampled the catalyst and measured the amount of carbon deposits offline to establish a standard curve between the catalyst color characteristics and the amount of carbon deposits, thus achieving rapid measurement of the amount of catalyst carbon deposits. Li Xin, Li Zhi et al. established an online carbon determination measurement system for catalysts in a moving bed unit by connecting a char measurement reactor to a flue gas analyzer, achieving online carbon determination measurement of laboratory catalysts. However, catalyst sampling is still required.

[0004] In the indirect measurement method, Sun Ziqiang et al. used the PLS-BP algorithm to estimate the catalyst carbon deposit content in the continuous catalytic reforming reactor, but the accuracy was not high; Meng Shuanghe et al. determined the catalyst carbon content-capacitance value relationship curve equation based on multiple standard samples with different carbon contents, used the ECT sensor to measure the capacitance value between the electrodes online, substituted it into the capacitance value and catalyst carbon content fitting curve equation, thereby realizing the online rapid detection of catalyst carbon deposits under high temperature conditions; Yang Yongrong et al. used an acoustic signal detection device to receive the acoustic emission signal generated by the catalyst impacting the wall of the reactor or regenerator, used a filtering algorithm to obtain the catalyst carbon deposit signal characteristic value, and established a catalyst carbon deposit prediction model through the least squares support vector machine method, but because the acoustic wave detection technology is not sensitive to the catalyst carbon deposit amount, it only remains in laboratory research; Huang Dexian et al. first used the regenerator mechanism analysis dynamic model to preliminarily estimate the molar flow rate of each main component of the regenerated flue gas and the heat capacity of the regenerated flue gas, and then estimated the CO content of the regenerated flue gas by the incinerator heat balance equation, and used a mechanism-based Analysis and dynamic model observation technology have realized the dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit; Cheng Youwei et al. took out the solid particles of the catalyst sample through the online sampling system, and used a high-temperature resistant optical fiber sensor to guide light into the surface of the catalyst sample particles to obtain the spectral signal of the catalyst sample during the reaction process. The L, a and b values ​​of the reflected light chromaticity values ​​were used to estimate the carbon deposition data of the catalyst sample from the model, realizing the online detection of the catalyst carbon deposition in the methanol to olefins process; Wang Xilei et al. obtained the original production data of the reaction-regeneration system in the fluidized catalytic cracking production process, selected auxiliary variables using the FCC mechanism analysis method, and extracted high-order features through deep learning to realize the real-time measurement of catalyst carbon deposition in fluidized catalytic cracking; in order to further improve the measurement accuracy and generalization ability of the catalyst carbon content soft measurement model, Wang Xilei conducted a mechanism analysis of the FCC reaction-regeneration system based on the empirical model, established a reaction-regeneration system mechanism model based on the pseudo-component theory, and realized soft measurement hybrid model modeling through a series combination method.

[0005] As a novel pathway for synthesizing light olefins from methanol, DMTO (dimethyl olefins or DMTO) holds significant practical significance for balancing the supply and demand of light olefins, reducing China's dependence on crude oil imports, and promoting national energy security. As the core of modern coal-to-olefin (CTO) plants and a leading achievement in my country's coal-based olefin industrialization, 14 DMTO units were operational nationwide as of 2020. However, due to limited production experience and data accumulation, only a small number of soft-sensing studies on diene yields in the DMTO process have been conducted. However, these studies, with average relative errors exceeding 10%, have limited industrial application. To date, no soft-sensing method for determining the carbon content of the SAPO-34 zeolite catalyst in DMTO plants has been developed. Summary of the Invention

[0006] In order to monitor the operating status of the DMTO unit in the coal chemical process, ensure the stable operation of the unit, and realize real-time diagnosis and optimization control of the unit, the present application provides a catalyst carbon content measurement method and system in the methanol to olefins process. By monitoring the data changes of relevant production variables, the online accurate measurement of the catalyst carbon content of the DMTO unit is realized, which is used to evaluate the catalyst activity and process progress, and assist in DMTO production decision-making.

[0007] To solve the above technical problems, this application provides the following technical solutions:

[0008] In a first aspect, the present application provides a method for measuring carbon content of a catalyst in a methanol to olefins process, comprising:

[0009] Obtain multiple reaction data in the methanol to olefins process;

[0010] The reaction data is input into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value;

[0011] The carbon content prediction model is established based on a gradient boosting regression algorithm and is trained using historical reaction data and the corresponding catalyst carbon content true values ​​as training data.

[0012] Furthermore, the training steps of the carbon quantity prediction model include:

[0013] Taking each of the historical reaction data as a first input and the corresponding catalyst carbon determination true value as a second input, the first input and the second input are inputted into a weak learner, the weak learner outputting a predicted value corresponding to the current first input and a predicted difference value, wherein the predicted difference value is the difference between the predicted value corresponding to the current second input and the first input;

[0014] Perform an iterative operation, using the prediction difference data of the previous weak learner as the current second input, inputting the first input and the current second input together into the current weak learner, the current weak learner outputting another prediction difference data, and serving as the current second input of the next weak learner, until all learners are traversed;

[0015] The predicted value corresponding to the first input of each weak learner is input to the strong learner, and the strong learner outputs the catalyst carbon value.

[0016] Furthermore, the method for measuring the carbon content of the catalyst in the methanol to olefins process further comprises:

[0017] The reaction data is preprocessed; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

[0018] Furthermore, the preprocessing of the reaction data includes:

[0019] Cleaning the reaction data to delete abnormal reaction data and missing reaction data;

[0020] determining a leading variable and an auxiliary variable in the reaction data;

[0021] The reaction data were normalized.

[0022] In a second aspect, the present application provides a catalyst carbon measurement system in a methanol to olefins process, comprising:

[0023] Data acquisition module: acquires multiple reaction data in the methanol to olefins process;

[0024] Carbon content prediction module: inputs the reaction data into a carbon content prediction model, which outputs a catalyst carbon content, and adjusts the catalyst content in the methanol to olefins process according to the catalyst carbon content;

[0025] Model training module: The carbon content prediction model is established based on the gradient boosting regression algorithm, and is trained using historical reaction data and the corresponding catalyst carbon content true value as training data.

[0026] Furthermore, the model training module includes:

[0027] First input unit: each of the historical reaction data is used as a first input, the corresponding catalyst carbon value is used as a second input, the first input and the second input are inputted into a weak learner, the weak learner outputs a predicted value corresponding to the current first input and a predicted difference value, the predicted difference value being the difference between the predicted value corresponding to the current second input and the first input;

[0028] Iteration unit: performs an iterative operation, uses the prediction difference data of the previous weak learner as the current second input, inputs the first input and the current second input into the current weak learner, and the current weak learner outputs another prediction difference data as the current second input of the next weak learner, until all learners are traversed;

[0029] Strong learner unit: The predicted value corresponding to the first input of each weak learner is input into the strong learner, and the strong learner outputs the catalyst carbon value.

[0030] Furthermore, the catalyst carbon measurement system in the methanol to olefins process further includes:

[0031] Data preprocessing module: preprocesses the reaction data; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

[0032] Furthermore, the data preprocessing module includes:

[0033] Data cleaning unit: cleaning the reaction data, deleting abnormal reaction data and missing reaction data;

[0034] Variable selection unit: determining the leading variables and auxiliary variables in the reaction data;

[0035] Data normalization unit: performs normalization processing on the reaction data.

[0036] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for measuring catalyst carbon content in the methanol to olefins process is implemented.

[0037] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for measuring carbon content of a catalyst in a methanol-to-olefins process.

[0038] It can be seen from the above technical solution that the present application provides a method and system for measuring the carbon content of catalysts in a methanol to olefins process, in which the reaction data is input into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value. The carbon content prediction model is established based on a gradient boosting regression algorithm, and is trained using historical reaction data and corresponding catalyst carbon content true values ​​as training data. There is no need to sample the catalyst, and dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit is achieved. By monitoring the data changes of relevant production variables, online and accurate measurement of the carbon content of the catalyst in the DMTO unit is achieved, which is used to evaluate the catalyst activity and process progress, and to assist in DMTO production decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0040] Figure 1It is a flow chart of a method for measuring catalyst carbon content in a methanol to olefins process according to an embodiment of the present application.

[0041] Figure 2 It is a schematic diagram of the methanol to olefins process system in the embodiment of the present application.

[0042] Figure 3 It is a schematic diagram of the training process of the fixed carbon amount prediction model in the embodiment of the present application.

[0043] Figure 4 It is a graph showing the predicted value and the actual value of the fixed carbon amount prediction model in the embodiment of the present application.

[0044] Figure 5 It is the relative error histogram (full sample) of the GBR fixed carbon amount prediction model in the embodiment of this application.

[0045] Figure 6 It is a schematic diagram of the preprocessing process of reaction data in the embodiment of this application.

[0046] Figure 7 It is a structural schematic diagram of the catalyst carbon measurement system in the methanol to olefins process in the embodiment of the present application.

[0047] Figure 8 It is a structural diagram of the model training module of the fixed carbon amount prediction model in the embodiment of the present application.

[0048] Figure 9 It is a schematic diagram of the iterative training process of the learner in the embodiment of the present application.

[0049] Figure 10 Schematic diagram of the structure of the data preprocessing module for reaction data in an embodiment of the present application.

[0050] Figure 11 It is a structural diagram of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0051] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0052] Efficient and stable catalysts can convert methanol to ethylene and propylene, resulting in high methanol conversion rates and excellent light olefin selectivity. However, the highly exothermic methanol conversion process can experience adiabatic temperature rises of up to 250°C, rapidly deactivating the catalyst due to coke deposition. To ensure process stability, the DMTO production process often utilizes a fluidized bed reactor-regenerator configuration to burn off catalyst coke. Furthermore, because light olefin selectivity is influenced by catalyst shape selectivity, the product composition of the DMTO process can be controlled in real time by using catalyst carbon titration.

[0053] Currently, industrial methods for measuring catalyst carbon deposition primarily rely on offline sampling and analysis and mechanistic model estimation. The sampling and analysis method involves extracting catalysts at different reaction times and from different locations in the reactor, and determining the catalyst carbon content through methods such as thermogravimetric analysis. The model estimation method establishes a mechanistic model by analyzing the reactor structure, reaction conditions, reaction time, and catalyst properties, and calculates the catalyst carbon deposition based on parameters such as the catalyst's reaction residence time. The sampling process affects reaction stability and equipment safety, and the analysis is time-consuming, making it unsuitable for real-time diagnosis and optimized control of industrial processes. Furthermore, the mechanistic model suffers from relatively large errors, making it difficult to meet the precision requirements of industrial production.

[0054] In order to monitor the operating status of the DMTO unit in the coal chemical process, ensure its stable operation, and realize real-time diagnosis and optimization control of the unit, this technology realizes online and accurate measurement of the carbon content of the DMTO unit catalyst by monitoring the data changes of relevant production variables. It is used to evaluate catalyst activity and process progress, and assist in DMTO production decision-making.

[0055] Based on this, the present application provides an embodiment of a method for measuring the carbon content of a catalyst in a methanol to olefins process, see Figure 1 ,include:

[0056] Step S100: Acquire multiple reaction data in a methanol to olefins process;

[0057] Step S200: inputting the reaction data into a carbon content prediction model, the carbon content prediction model outputting a catalyst carbon content value, and then adjusting the catalyst content in the methanol to olefins process according to the catalyst carbon content value;

[0058] The carbon content prediction model is established based on a gradient boosting regression algorithm and is trained using historical reaction data and the corresponding catalyst carbon content true values ​​as training data.

[0059] It can be understood that catalyst carbon content refers to the mass percentage of organic matter present on the catalyst surface and in the pores. The DMTO process mainly includes three systems: reaction-regeneration system, quenching-stripping system, and heat recovery system. Figure 2The reaction-regeneration system continuously burns off coke deposited on the catalyst and is considered the core of the DMTO unit. Table 1 lists the reaction data that influence the catalyst carbon content. This data is input into the carbon content prediction model to predict the carbon content of the regenerated catalyst. The model then outputs the catalyst carbon content, which is used to evaluate catalyst activity and process progress, assisting in DMTO production decision-making.

[0060]

[0061]

[0062] Table 1

[0063] The fixed carbon volume prediction model uses a gradient boosting regression algorithm as its parameter prediction algorithm, taking into account both high accuracy and real-time performance. Real production data is used for model training and evaluation. Gradient Boosting Regression (GBR): Gradient Boosting trains newly added weak classifiers based on the negative gradient of the current model's loss function. These trained weak classifiers are then added to the existing model in an additive manner. This is an ensemble learning regression technique. In practice, a new model is generated based on the gradient descent of the previous model's loss function and then added to the previous model. Performance is primarily limited by the loss function, weak classifiers, and additive model.

[0064] As can be seen from the above description, the method for measuring the carbon content of the catalyst in the methanol to olefins process provided in the embodiment of the present application uses a gradient boosting regression algorithm to establish a carbon content prediction model, and the reflection data during the reaction process is input into the carbon content prediction model. There is no need to sample the catalyst, thereby realizing dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit.

[0065] In one embodiment of the method for measuring the carbon content of a catalyst in a methanol to olefins process provided in the present application, a preferred method for training a carbon content prediction model is provided, see Figure 3 The training steps of the carbon quantity prediction model include:

[0066] Step S001: taking each of the historical reaction data as the first input and the corresponding catalyst carbon value as the second input, inputting the first input and the second input together into a weak learner, wherein the weak learner outputs a predicted value corresponding to the current first input and a predicted difference value, wherein the predicted difference value is the difference between the predicted values ​​corresponding to the current second input and the first input;

[0067] Step S002: performing an iterative operation, using the prediction difference data of the previous weak learner as the current second input, inputting the first input and the current second input together into the current weak learner, and the current weak learner outputting another prediction difference data as the current second input of the next weak learner, until all learners are traversed;

[0068] Step S003: inputting the predicted value corresponding to the first input of each weak learner into the strong learner, and the strong learner outputs the catalyst carbon value.

[0069] It is understandable that the basic idea of ​​the gradient boosting regression algorithm is to generate multiple weak learners in series. The goal of each weak learner is to fit the negative gradient of the loss function of the previous cumulative model, so that the cumulative model loss after adding the weak learner decreases in the direction of the negative gradient. In the 1≤m≤M steps of gradient boosting, a model F with relatively weak prediction accuracy is first found. m(x) , the difference between the predicted value and the actual y value, that is, the residual: R (x) =yF m(x) , the gradient boosting algorithm does not change F (x) model, but by adding the estimator h (x) Build a new model, F m+1(x) =F m(x) +h (x) , so the residual becomes R (x) =yF m(x) , becomes significantly smaller. (x) The best makes: F m+1(x) =F m(x) +h (x) =y, that is, h(x)=yF m(x) . Gradient boosting algorithm h (x) With yF m(x) Phase fitting, it is observed that the reduction of residual is actually the loss function of square error 1 / 2[yF (x) ] 2 The negative gradient direction.

[0070] A plurality of reaction data and the corresponding true carbon determination are input into a weak learner, and the weak learner outputs the fitting results of the plurality of reaction data and the true carbon determination, and the fitting results are the predicted values ​​of the carbon determination; the difference between the true carbon determination and the predicted values ​​of the carbon determination and the plurality of reaction data are used as the input of the next weak learner, and the weak learner outputs the fitting result, and the fitting result is the predicted value of the carbon determination; and so on, the difference between the true carbon determination and all the fitting results is used as the input of the next weak learner until all the weak learners are trained; the fitting results output by each weak learner are input into a strong learner, and after the strong learner assigns a weight to each weak learner, all the fitting results are weighted to finally obtain the predicted value of the catalyst carbon determination.

[0071] In some specific embodiments, the training step of the carbon content prediction model further includes: passing the test set into the trained model to predict the carbon content of the spent catalyst to obtain a predicted value of the carbon content of the spent catalyst.

[0072] For the trained carbon quantity prediction model, this application uses mean squared error (MSE), mean absolute percentage error (MAPE), goodness of fit (Coefficient of determination, R 2 ) measures the accuracy of the prediction estimate, and the computing time T spent on each sample from preprocessing to generating the predicted value is used to measure real-time performance.

[0073] Assume y i is the actual value, y i ' is the predicted value, is the actual value average, and N is the number of samples.

[0074] ① Mean Square Error (MSE)

[0075] This indicator calculates the mean of the sum of squares of the errors between the fitted data and the original data corresponding to the sample points. The smaller the value, the better the fitting effect. The calculation formula is as follows:

[0076]

[0077] ② Mean Percent Error (MAPE)

[0078] The range of this indicator is 0 to +∞. A MAPE of 0 indicates a perfect model, while a MAPE greater than 100% indicates a poor model. The calculation formula is as follows:

[0079]

[0080] ③ Goodness of fit (R 2 )

[0081] This indicator reflects the degree to which the predicted value explains the actual value, and the larger the better. The calculation formula is as follows:

[0082]

[0083] ④Calculation time T

[0084] This metric measures the time it takes for new data to be imported into the model program, preprocessed, and then generated by the model. It reflects whether the model can process new data online in real time. In this model, the difference between the time t1 when the test set samples are imported and the time t2 when the model completes all predictions is calculated to obtain the total processing and prediction time. This is then divided by the number of samples to obtain the sample average prediction time T. The calculation formula is as follows:

[0085]

[0086] Compare the actual carbon value of the catalyst to be produced in the test set with the model prediction value, and calculate the MSE, MAPE, and R 2 The average relative error between the carbon determination results of DMTO catalyst by this technology and the offline measurement results is only 1.36%, and the goodness of fit is as high as 0.93. The specific results are shown in Table 2.

[0087] Model MSE MAPE <![CDATA[R 2 ]]> T / ms Model_GBR 0.01790 0.01356 0.92718 1.6952 Model_SVR 0.06532 0.02475 0.73421 1.5445 Model_PLS 0.06372 0.02594 0.74071 1.5687

[0088] Table 2 Regression indicators of the carbon quantity prediction model

[0089] As shown in Table 2, the carbon quantity prediction model based on gradient boosting regression is better than the other two traditional soft sensor models, support vector machine regression model SVR and partial least squares regression model PLS in terms of MSE, MAPE, R 2 All three metrics have improved significantly, with only a slight lag in computational time. Furthermore, the computational time for all three models is in the millisecond range, which is highly valuable for real-time guidance of optimized control and production in actual production.

[0090] Select the first 40 true values ​​and predicted values ​​of the test set to make a line graph, see Figure 4 , make relative error histogram of carbon quantity prediction model, see Figure 5 .

[0091] As shown in Table 2, GBR is better than the other two traditional soft sensor models, support vector machine regression model SVR and partial least squares regression model PLS in MSE, MAPE, R 2 All three metrics have improved significantly, with only a slight lag in computational time. Furthermore, the computational time for all three models is in the millisecond range, which is highly valuable for real-time guidance of optimized control and production in actual production.

[0092] like Figure 5 First, for new samples, the model's predicted points show consistent trends with the true values. Second, the model's predicted values ​​do not appear to be too high for the true values ​​and too low for the predicted values, or vice versa. Finally, it can be seen that the model can accurately predict high values ​​and low values. In summary, the DMTO carbon soft sensing model based on gradient boosting regression demonstrates excellent predictive power and practicality.

[0093] The total number of samples is 2480, of which 1737 samples have a relative error range of [-0.01, 0.01], accounting for about 70.04%; 2131 samples have a relative error range of [-0.015, 0.015], accounting for about 85.93%; 2340 samples have a relative error range of [-0.02, 0.02], accounting for about 94.35%; and 2479 samples have a relative error range of [-0.05, 0.05], accounting for about 100%.

[0094] In one embodiment of the method for measuring carbon content of a catalyst in a methanol-to-olefins process provided in the present application, a preferred method for processing reaction data is provided, wherein the method for measuring carbon content of a catalyst in a methanol-to-olefins process further comprises:

[0095] The reaction data is preprocessed; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

[0096] It is understandable that not every reaction data collected during the methanol to olefins process can be used directly. The reaction data needs to be processed uniformly before the catalyst carbon content can be predicted. Before the carbon content prediction model is trained, the historical reaction data also needs to be preprocessed.

[0097] In one embodiment of the method for measuring the carbon content of the catalyst in the methanol to olefins process provided in this application, a preferred method for processing reaction data is provided, see Figure 6 , the preprocessing of the reaction data includes:

[0098] Step S011: performing data cleaning on the reaction data, deleting abnormal reaction data and missing reaction data;

[0099] Step S012: determining the leading variable and the auxiliary variable in the reaction data;

[0100] Step S013: normalizing the reaction data.

[0101] As can be understood, data preprocessing includes four main steps: data cleaning, variable selection, data normalization, and data set partitioning. Data cleaning involves removing samples containing outliers and missing values ​​from the actual recorded production data. This model uses the "3σ principle" to filter out and delete outliers and missing samples. The 3σ principle, also known as the Laida criterion, assumes that a set of test data contains only random errors. The standard deviation is calculated and processed, and an interval is determined based on a certain probability. Errors exceeding this interval are considered non-random but rather gross errors, and data containing such errors should be eliminated. Variable selection involves determining the auxiliary and dominant variables of the soft sensor model. Auxiliary variables are selected based on availability, lack of time lag, and controllability. In this model, easily measurable data such as reaction temperature, reaction pressure, external dilution steam quantity, and methanol feed quantity (Table 1) are selected as auxiliary variables, while the carbon content of the spent catalyst is selected as the dominant variable. Data normalization involves dimensionlessly converting original production data, which has variables of varying magnitudes and units, into uniform size ranges. This model uses the MaxAbsSclar method. Splitting the test set involves dividing the dataset into a training set and a test set. The training set is used to train the model, and the test set is used to test the model's predictive performance on new samples.

[0102] From the above description, it can be seen that the present application provides a method for measuring the carbon content of catalysts in a methanol to olefins process, in which the reaction data is input into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value. The carbon content prediction model is established based on a gradient boosting regression algorithm, and is trained using historical reaction data and corresponding catalyst carbon content true values ​​as training data. There is no need to sample the catalyst, and dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit is achieved. By monitoring the data changes of relevant production variables, online and accurate measurement of the catalyst carbon content of the DMTO unit is achieved, which is used to evaluate catalyst activity and process progress, and assist in DMTO production decision-making.

[0103] From the software level, in order to solve the problem of relatively large prediction error of carbon content, an embodiment of the catalyst carbon content measurement system in the methanol to olefins process provided in this application is shown in FIG. Figure 7 ,include:

[0104] Data acquisition module: acquires multiple reaction data in the methanol to olefins process;

[0105] Carbon content prediction module: inputs the reaction data into a carbon content prediction model, which outputs a catalyst carbon content, and adjusts the catalyst content in the methanol to olefins process according to the catalyst carbon content;

[0106] Model training module: The carbon content prediction model is established based on the gradient boosting regression algorithm, and is trained using historical reaction data and the corresponding catalyst carbon content true value as training data.

[0107] It can be understood that the reaction data that affect the carbon content of the catalyst are shown in Table 1. The data acquisition module obtains the relevant reaction data in Table 1, and inputs the reaction data in Table 1 into the carbon content prediction model. The carbon content prediction module predicts the carbon content of the catalyst to be produced. The carbon content prediction model outputs the catalyst carbon content, and then evaluates the catalyst activity and process progress to assist DMTO production decision-making.

[0108] The fixed carbon volume prediction model uses a gradient boosting regression algorithm as its parameter prediction algorithm, taking into account both high accuracy and real-time performance. Real production data is used for model training and evaluation. Gradient Boosting Regression (GBR): Gradient Boosting trains newly added weak classifiers based on the negative gradient of the current model's loss function. These trained weak classifiers are then added to the existing model in an additive manner. This is an ensemble learning regression technique. In practice, a new model is generated based on the gradient descent of the previous model's loss function and then added to the previous model. Performance is primarily limited by the loss function, weak classifiers, and additive model.

[0109] The main parameters of the carbon content prediction model are: 600 individual learners, the loss function is defined as the mean square error, and the model learning rate is set to 0.1. Model input parameters: See Table 1. Model output parameters: catalyst carbon content.

[0110] As can be seen from the above description, the catalyst carbon content measurement system in the methanol to olefins process provided in the embodiment of the present application uses a gradient boosting regression algorithm to establish a carbon content prediction model, and inputs the reflection data during the reaction process into the carbon content prediction model. There is no need to sample the catalyst, thereby realizing dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit.

[0111] An embodiment of the catalyst carbon content measurement system in the methanol to olefins process provided in this application provides an optimal method for training a carbon content prediction model, see Figure 8 , the model training module includes:

[0112] First input unit: each of the historical reaction data is used as a first input, the corresponding catalyst carbon value is used as a second input, the first input and the second input are inputted into a weak learner, the weak learner outputs a predicted value corresponding to the current first input and a predicted difference value, the predicted difference value being the difference between the predicted value corresponding to the current second input and the first input;

[0113] Iteration unit: performs an iterative operation, uses the prediction difference data of the previous weak learner as the current second input, inputs the first input and the current second input into the current weak learner, and the current weak learner outputs another prediction difference data as the current second input of the next weak learner, until all learners are traversed;

[0114] Strong learner unit: The predicted value corresponding to the first input of each weak learner is input into the strong learner, and the strong learner outputs the catalyst carbon value.

[0115] It is understandable that the basic idea of ​​the gradient boosting regression algorithm is to generate multiple weak learners in series. The goal of each weak learner is to fit the negative gradient of the loss function of the previous cumulative model, so that the cumulative model loss after adding the weak learner decreases in the direction of the negative gradient. In the 1≤m≤M steps of gradient boosting, a model F with relatively weak prediction accuracy is first found. m(x) , the difference between the predicted value and the actual y value, that is, the residual: R (x) =yF m(x) , the gradient boosting algorithm does not change F (x) model, but by adding the estimator h (x) Build a new model, F m+1(x) =F m(x) +h (x) , so the residual becomes R (x) =yF m(x) , becomes significantly smaller. (x) The best makes: F m+1(x) =F m(x) +h (x) =y, that is, h(x)=yF m(x) . Gradient boosting algorithm h (x) With yF m(x) Phase fitting, it is observed that the reduction of residual is actually the loss function of square error 1 / 2[yF (x) ] 2 The negative gradient direction.

[0116] The learner training process can be found in Figure 9The first input unit inputs multiple reaction data and the corresponding true carbon content into a weak learner, and the weak learner outputs the fitting results of the multiple reaction data and the true carbon content, and the fitting result is the predicted value of the carbon content; the iterative unit uses the difference between the true carbon content and the predicted value of the carbon content and the multiple reaction data as the input of the next weak learner, and the weak learner outputs the fitting result, which is the predicted value of the carbon content; and so on, the difference between the true carbon content and all the fitting results is used as the input of the next weak learner until all the weak learners are trained; the strong learner unit inputs the fitting result output by each weak learner into a strong learner, and the strong learner assigns a weight to each weak learner, and then weights all the fitting results to finally obtain the predicted value of the catalyst carbon content.

[0117] In one embodiment of the catalyst carbon measurement system in the methanol to olefins process provided in this application, a preferred method for processing reaction data is provided. Figure 7 The catalyst carbon measurement system in the methanol to olefins process also includes:

[0118] Data preprocessing module: preprocesses the reaction data; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

[0119] It is understandable that not every reaction data collected during the methanol to olefins process can be used directly. The data preprocessing module needs to uniformly process the reaction data before predicting the catalyst carbon content. Before training the carbon content prediction model, the historical reaction data also needs to be preprocessed.

[0120] In one embodiment of the catalyst carbon measurement system in the methanol to olefins process provided in this application, a preferred method for processing reaction data is provided. Figure 10 , the data preprocessing module includes:

[0121] Data cleaning unit: cleaning the reaction data, deleting abnormal reaction data and missing reaction data;

[0122] Variable selection unit: determining the leading variables and auxiliary variables in the reaction data;

[0123] Data normalization unit: performs normalization processing on the reaction data.

[0124] As can be understood, data preprocessing includes four main steps: data cleaning, variable selection, data normalization, and data set partitioning. Data cleaning involves removing samples containing outliers and missing values ​​from the actual recorded production data. The data cleaning unit uses the "3σ principle" to filter out and delete outliers and missing samples. The 3σ principle, also known as the Laida criterion, assumes that a set of test data contains only random errors. The standard deviation is calculated and processed, and an interval is determined based on a certain probability. Errors exceeding this interval are considered gross errors, not random errors, and data containing such errors should be eliminated. Variable selection involves determining the auxiliary and dominant variables of the soft sensor model. Auxiliary variables are selected based on availability, lack of time lag, and controllability. In the variable selection unit, easily measurable data such as reaction temperature, reaction pressure, external dilution steam quantity, and methanol feed quantity (Table 1) are selected as auxiliary variables, and the carbon content of the spent catalyst is selected as the dominant variable. Data normalization involves dimensionlessly converting original production data, which has variables of varying magnitudes and units, into uniform size ranges. The MaxAbsSclar method is used for data normalization. Finally, test set partitioning involves dividing the dataset into a training set and a test set. The training set is used to train the model, and the test set is used to test the model's predictive performance on new samples.

[0125] From the above description, it can be seen that the present application provides a catalyst carbon content measurement system in a methanol to olefins process, which inputs the reaction data into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and then the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value. The carbon content prediction model is established based on a gradient boosting regression algorithm, and is trained using historical reaction data and the corresponding catalyst carbon content true value as training data. There is no need to sample the catalyst, and dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit is achieved. By monitoring the data changes of relevant production variables, online and accurate measurement of the catalyst carbon content of the DMTO unit is achieved, which is used to evaluate the catalyst activity and process progress, and assist in DMTO production decision-making.

[0126] From a hardware perspective, in order to solve the existing problem of catalyst carbon content prediction, the present application provides an embodiment of an electronic device for implementing all or part of the content of the catalyst carbon content prediction method. The electronic device specifically includes the following content:

[0127] Figure 11 Schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Figure 11 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that the Figure 11is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0128] In one embodiment, the catalyst carbon content prediction function may be integrated into the central processing unit. The central processing unit may be configured to perform the following control:

[0129] Step S100: Acquire multiple reaction data in a methanol to olefins process;

[0130] Step S200: inputting the reaction data into a carbon content prediction model, the carbon content prediction model outputting a catalyst carbon content value, and then adjusting the catalyst content in the methanol to olefins process according to the catalyst carbon content value;

[0131] The carbon content prediction model is established based on a gradient boosting regression algorithm and is trained using historical reaction data and the corresponding catalyst carbon content true values ​​as training data.

[0132] It can be understood that catalyst carbon content refers to the mass percentage of organic matter present on the catalyst surface and in the pores. The DMTO process mainly includes three systems: reaction-regeneration system, quenching-stripping system, and heat recovery system. Figure 2 The reaction-regeneration system continuously burns off coke deposited on the catalyst and is considered the core of the DMTO unit. Table 1 lists the reaction data that influence the catalyst carbon content. This data is input into the carbon content prediction model to predict the carbon content of the regenerated catalyst. The model then outputs the catalyst carbon content, which is used to evaluate catalyst activity and process progress, assisting in DMTO production decision-making.

[0133] <![CDATA[x1]]> External dilution steam volume <![CDATA[x 12 ]]> Regenerator dense phase storage <![CDATA[x2]]> Amount of protective steam in the reaction <![CDATA[x 13 ]]> Reaction-rotor inlet linear speed <![CDATA[x3]]> Waiting stripping section (upper part) <![CDATA[x 14 ]]> Regeneration first rotation inlet linear speed <![CDATA[x4]]> Waiting stripping section (lower part) <![CDATA[x 15 ]]> Air volume entering regenerator <![CDATA[x5]]> Methanol feed rate <![CDATA[x 16 ]]> Main air discharge volume <![CDATA[x6]]> Steam injection rate <![CDATA[x 17 ]]> Main air vent valve position <![CDATA[x7]]> Reaction temperature <![CDATA[x 18 ]]> Regeneration slide valve position <![CDATA[x8]]> Reaction pressure <![CDATA[x 19 ]]> Regenerator burn total air volume <![CDATA[x9]]> Regeneration temperature <![CDATA[x 20 ]]> Double-acting slide valve A position <![CDATA[x 10 ]]> Regeneration pressure <![CDATA[x 21 ]]> Double-acting spool valve position B <![CDATA[x 11 ]]> Reactor dense phase storage <![CDATA[x 22 ]]> Coke yield

[0134] Table 1

[0135] The fixed carbon volume prediction model uses a gradient boosting regression algorithm as its parameter prediction algorithm, taking into account both high accuracy and real-time performance. Real production data is used for model training and evaluation. Gradient Boosting Regression (GBR): Gradient Boosting trains newly added weak classifiers based on the negative gradient of the current model's loss function. These trained weak classifiers are then added to the existing model in an additive manner. This is an ensemble learning regression technique. In practice, a new model is generated based on the gradient descent of the previous model's loss function and then added to the previous model. Performance is primarily limited by the loss function, weak classifiers, and additive model.

[0136] As can be seen from the above description, the method for measuring the carbon content of the catalyst in the methanol to olefins process provided in the embodiment of the present application uses a gradient boosting regression algorithm to establish a carbon content prediction model, and the reflection data during the reaction process is input into the carbon content prediction model. There is no need to sample the catalyst, thereby realizing dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit.

[0137] In another embodiment, the catalyst carbon determination measurement system can be configured separately from the central processing unit 9100. For example, the catalyst carbon determination measurement system can be configured as a chip connected to the central processing unit 9100, and the catalyst carbon determination measurement prediction function can be realized through the control of the central processing unit.

[0138] like Figure 11 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Figure 11 In addition, the electronic device 9600 may also include all components shown in Figure 11 For components not shown, reference may be made to the prior art.

[0139] like Figure 11 As shown, the central processing unit 9100 is sometimes also referred to as a controller or operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0140] Memory 9140 can be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It can store the aforementioned failure-related information and also store programs that execute the relevant information. The CPU 9100 can execute the programs stored in memory 9140 to implement information storage or processing.

[0141] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 may be, for example, a keypad or touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display objects such as images and text. The display may be, for example, an LCD display, but is not limited thereto.

[0142] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), or a SIM card. Alternatively, it may be a memory that retains information even when power is off, can be selectively erased, and is provided with more data. Examples of such memory are sometimes referred to as EPROMs. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 by the central processing unit 9100.

[0143] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various driver programs for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0144] The communication module 9110 is a transmitter / receiver 9110 that transmits and receives signals via an antenna 9111. The communication module (transmitter / receiver) 9110 is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as in a conventional mobile communication terminal.

[0145] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all steps of the catalyst carbon content prediction method in the above embodiments. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, all steps of the catalyst carbon content prediction method in the above embodiments are implemented, wherein the execution subject is a server or a client. For example, when the processor executes the computer program, the following steps are implemented:

[0146] Step S100: Acquire multiple reaction data in a methanol to olefins process;

[0147] Step S200: inputting the reaction data into a carbon content prediction model, the carbon content prediction model outputting a catalyst carbon content value, and then adjusting the catalyst content in the methanol to olefins process according to the catalyst carbon content value;

[0148] The carbon content prediction model is established based on a gradient boosting regression algorithm and is trained using historical reaction data and the corresponding catalyst carbon content true values ​​as training data.

[0149] It can be understood that catalyst carbon content refers to the mass percentage of organic matter present on the catalyst surface and in the pores. The DMTO process mainly includes three systems: reaction-regeneration system, quenching-stripping system, and heat recovery system. Figure 2 The reaction-regeneration system continuously burns off coke deposited on the catalyst and is considered the core of the DMTO unit. Table 1 lists the reaction data that influence the catalyst carbon content. This data is input into the carbon content prediction model to predict the carbon content of the regenerated catalyst. The model then outputs the catalyst carbon content, which is used to evaluate catalyst activity and process progress, assisting in DMTO production decision-making.

[0150]

[0151]

[0152] Table 1

[0153] The fixed carbon volume prediction model uses a gradient boosting regression algorithm as its parameter prediction algorithm, taking into account both high accuracy and real-time performance. Real production data is used for model training and evaluation. Gradient Boosting Regression (GBR): Gradient Boosting trains newly added weak classifiers based on the negative gradient of the current model's loss function. These trained weak classifiers are then added to the existing model in an additive manner. This is an ensemble learning regression technique. In practice, a new model is generated based on the gradient descent of the previous model's loss function and then added to the previous model. Performance is primarily limited by the loss function, weak classifiers, and additive model.

[0154] As can be seen from the above description, the method for measuring the carbon content of the catalyst in the methanol to olefins process provided in the embodiment of the present application uses a gradient boosting regression algorithm to establish a carbon content prediction model, and the reflection data during the reaction process is input into the carbon content prediction model. There is no need to sample the catalyst, thereby realizing dynamic soft measurement of the carbon content of the catalyst in the regenerator of the catalytic cracking unit.

[0155] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatus, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0156] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0159] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A method for measuring the carbon content of a catalyst in a methanol to olefins process, characterized in that: include: Obtain multiple reaction data in the methanol to olefins process; The reaction data is input into a carbon content prediction model, and the carbon content prediction model outputs a catalyst carbon content value, and the catalyst content in the methanol to olefins process can be adjusted according to the catalyst carbon content value; The carbon content prediction model is established based on a gradient boosting regression algorithm and is trained using historical reaction data and the corresponding catalyst carbon content real values ​​as training data. The training steps of the carbon quantity prediction model include: Taking each of the historical reaction data as a first input and the corresponding catalyst carbon determination true value as a second input, the first input and the second input are inputted into a weak learner, the weak learner outputting a predicted value corresponding to the current first input and a predicted difference value, wherein the predicted difference value is the difference between the predicted value corresponding to the current second input and the first input; Perform an iterative operation, using the prediction difference data of the previous weak learner as the current second input, inputting the first input and the current second input together into the current weak learner, the current weak learner outputting another prediction difference data, and serving as the current second input of the next weak learner, until all learners are traversed; Inputting the predicted value corresponding to the first input of each weak learner into the strong learner, and the strong learner outputs the catalyst carbon value; The reaction data include: external dilution steam volume, internal protection steam volume, stripping section to be generated, methanol feed volume, steam injection rate, reaction temperature, reaction pressure, regeneration temperature, regeneration pressure, reactor dense phase storage volume, regenerator dense phase storage volume, reaction rotary inlet linear speed, regeneration rotary inlet linear speed, air volume entering regenerator, main air vent volume, main air vent valve position, regeneration slide valve position, total regenerator charring air volume, two valve positions of double-acting slide valve and coking rate.

2. The method for measuring the carbon content of a catalyst in a methanol to olefins process according to claim 1, wherein: The method for measuring the carbon content of the catalyst in the methanol to olefins process further comprises: The reaction data is preprocessed; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

3. The method for measuring the carbon content of a catalyst in a methanol to olefins process according to claim 2, wherein: The preprocessing of the reaction data includes: Cleaning the reaction data to delete abnormal reaction data and missing reaction data; determining a leading variable and an auxiliary variable in the reaction data; The reaction data were normalized.

4. A catalyst carbon measurement system in a methanol to olefins process, characterized in that: include: Data acquisition module: acquires multiple reaction data in the methanol to olefins process; Carbon content prediction module: inputs the reaction data into a carbon content prediction model, which outputs a catalyst carbon content, and adjusts the catalyst content in the methanol to olefins process according to the catalyst carbon content; Model training module: The carbon content prediction model is established based on the gradient boosting regression algorithm, and is trained using historical reaction data and the corresponding catalyst carbon content real value as training data; The model training module includes: First input unit: each of the historical reaction data is used as a first input, the corresponding catalyst carbon value is used as a second input, the first input and the second input are inputted into a weak learner, the weak learner outputs a predicted value corresponding to the current first input and a predicted difference value, the predicted difference value being the difference between the predicted value corresponding to the current second input and the first input; Iteration unit: performs an iterative operation, uses the prediction difference data of the previous weak learner as the current second input, inputs the first input and the current second input into the current weak learner, and the current weak learner outputs another prediction difference data as the current second input of the next weak learner, until all learners are traversed; Strong learner unit: inputs the predicted value corresponding to the first input of each weak learner into the strong learner, and the strong learner outputs the catalyst carbon value; The reaction data include: external dilution steam volume, internal protection steam volume, stripping section to be generated, methanol feed volume, steam injection rate, reaction temperature, reaction pressure, regeneration temperature, regeneration pressure, reactor dense phase storage volume, regenerator dense phase storage volume, reaction rotary inlet linear speed, regeneration rotary inlet linear speed, air volume entering regenerator, main air vent volume, main air vent valve position, regeneration slide valve position, total regenerator charring air volume, two valve positions of double-acting slide valve and coking rate.

5. The catalyst carbon measurement system in the methanol to olefins process according to claim 4, characterized in that: The catalyst carbon measurement system in the methanol to olefins process also includes: Data preprocessing module: preprocesses the reaction data; wherein, inputting the reaction data into the fixed carbon amount prediction model includes: inputting the preprocessed reaction data into the fixed carbon amount prediction model.

6. The catalyst carbon measurement system in the methanol to olefins process according to claim 5, characterized in that: The data preprocessing module includes: Data cleaning unit: cleaning the reaction data, deleting abnormal reaction data and missing reaction data; Variable selection unit: determining the leading variables and auxiliary variables in the reaction data; Data normalization unit: performs normalization processing on the reaction data.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the catalyst carbon measurement method in the methanol to olefins process according to any one of claims 1 to 3 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for measuring catalyst carbon content in a methanol to olefins process according to any one of claims 1 to 3 is implemented.

Images