Intelligent control and data analysis method and system for gas purification in adsorption tank

Abstract

The invention provides an intelligent control and data analysis method and system for gas purification in an adsorption tank. The method comprises the following steps: acquiring multi-dimensional environment data in an adsorption tank in real time to obtain a multi-dimensional environment data set; based on the multi-dimensional environment data set, utilizing a time sequence analysis and machine learning algorithm to predict the concentration change trend of gas components and the performance degradation condition of the adsorption material, and meanwhile, applying an anomaly detection algorithm to identify potential abnormal working conditions to obtain a comprehensive prediction result; according to the comprehensive prediction result, a multi-objective optimization algorithm and a fuzzy logic control theory are adopted, and optimized operation parameter setting is generated; the optimized operation parameter setting is applied to actual operation of the adsorption tank, the implementation effect of the adsorption tank is monitored, the actual operation effect is fed back to the prediction model, and a more accurate prediction model is generated. According to the technical scheme provided by the invention, the data analysis effect of gas purification in the adsorption tank can be improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of intelligent control and data analysis technology, and in particular to an intelligent control and data analysis method and system for gas purification in an adsorption tank. Background Art

[0002] Adsorption tanks are widely used in the fields of chemical industry, environmental protection, energy, etc. to remove harmful components in gas or recover valuable gas. In these applications, the operating status of the adsorption tank directly affects the effect of gas treatment and the overall performance of the system. In order to ensure the efficient, safe and stable operation of the adsorption tank, it is necessary to collect and analyze the multi-dimensional environmental data in the adsorption tank (such as gas component concentration, temperature, pressure, humidity and flow rate) in real time. These data not only help to predict the concentration trend of gas components and the performance attenuation of adsorption materials, but also can timely identify potential abnormal conditions, so as to take corresponding measures for adjustment and optimization.

[0003] At present, most adsorption tank systems have basic data acquisition functions and can monitor gas composition concentration, temperature, pressure, humidity, flow rate and other parameters in real time; existing data analysis methods mainly rely on traditional statistical analysis and simple prediction models, such as linear regression and moving average method. These methods can provide a certain degree of prediction and analysis, but the accuracy and robustness are limited; existing anomaly detection methods are usually based on threshold judgment and simple rule engines, which can identify some obvious abnormal situations, but for complex and changeable working conditions, their accuracy and timeliness need to be improved; existing parameter optimization methods mostly use single-objective optimization strategies, such as optimization control theory, but these methods often find it difficult to consider multiple optimization objectives at the same time, resulting in incomplete and ineffective optimization results.

[0004] Traditional statistical analysis and simple prediction models have difficulty capturing the inherent laws and dynamic changes of data when processing complex multi-dimensional environmental data, resulting in low prediction accuracy; existing anomaly detection methods mainly rely on fixed thresholds and rules. For complex and changeable working conditions, they cannot identify potential abnormal situations in a timely and accurate manner, and are prone to false alarms or omissions; existing parameter optimization methods mostly adopt single-objective optimization strategies, which make it difficult to consider multiple optimization objectives at the same time (such as minimizing the prediction error of gas component concentration trend, minimizing system operating costs, minimizing the frequency of abnormal operating condition warnings, and maximizing system stability), resulting in optimization results that are not comprehensive and effective; most of the existing solutions lack an effective closed-loop feedback mechanism, and cannot continuously optimize and adjust the prediction model according to the actual operating results, resulting in poor system adaptability and robustness. Summary of the invention

[0005] The embodiment of the present application provides an intelligent control and data analysis method and system for gas purification in an adsorption tank, so as to solve the problem of poor data analysis effect for the adsorption tank in the prior art.

[0006] In a first aspect, an embodiment of the present application provides an intelligent control and data analysis method for gas purification in an adsorption tank, comprising:

[0007] Collecting multi-dimensional environmental data in the adsorption tank in real time to obtain a multi-dimensional environmental data set, wherein the multi-dimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate;

[0008] Based on the multidimensional environmental data set, the concentration change trend of the gas components and the performance attenuation of the adsorption materials are predicted and processed using time series analysis and machine learning algorithms, and anomaly detection algorithms are applied to identify potential abnormal conditions to obtain comprehensive prediction results;

[0009] According to the comprehensive prediction results, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings;

[0010] The optimized operating parameter settings are applied to the actual operation of the adsorption tank, the implementation effect is monitored, and the actual operation effect is fed back to the prediction model to generate a more accurate prediction model.

[0011] In a second aspect, the embodiment of the present application provides an intelligent control and data analysis method for gas purification in an adsorption tank, comprising:

[0012] A collection module is used to collect multi-dimensional environmental data in the adsorption tank in real time to obtain a multi-dimensional environmental data set, wherein the multi-dimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate;

[0013] A processing module is used to predict the concentration change trend of gas components and the performance attenuation of adsorption materials based on the multi-dimensional environmental data set by using time series analysis and machine learning algorithms, and to identify potential abnormal conditions by using an anomaly detection algorithm to obtain a comprehensive prediction result;

[0014] A generation module, for intelligently adjusting the operating parameters of the adsorption tank according to the comprehensive prediction results by using a multi-objective optimization algorithm and fuzzy logic control theory to generate optimized operating parameter settings;

[0015] The monitoring module is used to apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor its implementation effect, and feed back the actual operation effect to the prediction model to generate a more accurate prediction model.

[0016] In a third aspect, an embodiment of the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an intelligent control and data analysis method for gas purification in an adsorption tank as described in the first aspect.

[0017] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program, which, when executed by a computer, implements an intelligent control and data analysis method for gas purification in an adsorption tank as described in the first aspect.

[0018] In an embodiment of the present application, multidimensional environmental data in the adsorption tank are collected in real time to obtain a multidimensional environmental data set, wherein the multidimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate; based on the multidimensional environmental data set, time series analysis and machine learning algorithms are used to predict the concentration change trend of the gas components and the performance attenuation of the adsorption material, and an anomaly detection algorithm is used to identify potential abnormal conditions to obtain a comprehensive prediction result; based on the comprehensive prediction result, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings; the optimized operating parameter settings are applied to the actual operation of the adsorption tank, the implementation effect is monitored, and the actual operation effect is fed back to the prediction model to generate a more accurate prediction model. By collecting the gas component concentration, temperature, pressure, humidity and flow rate in the adsorption tank in real time, the comprehensiveness and accuracy of the data are ensured, providing a reliable basis for subsequent prediction and optimization; using time series analysis and machine learning algorithms, the concentration change trend of the gas component and the performance attenuation of the adsorption material are predicted, which improves the accuracy and reliability of the prediction; using anomaly detection algorithms to identify potential abnormal conditions, timely discover and handle abnormal conditions, and avoid equipment failures or production accidents caused by abnormal conditions; using multi-objective optimization algorithms, comprehensively considering multiple optimization objectives (such as minimizing the prediction error of the gas component concentration change trend, minimizing the system operation cost, Minimize the frequency of abnormal operating condition warnings and maximize system stability), generate a preliminary optimization plan to ensure the optimal operation of the adsorption tank under multiple constraints; use fuzzy logic control theory to intelligently adjust the operating parameters of the adsorption tank, improve the system's adaptability and robustness, and ensure that it can still maintain efficient and stable operation under complex and changeable working conditions; apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor its implementation effect in real time, and ensure the effectiveness of the optimization plan; feed back the actual operation effect to the prediction model, and through continuous learning and optimization, generate a more accurate prediction model, form a closed-loop feedback mechanism, and achieve continuous improvement and optimization of the system. Through real-time monitoring and intelligent adjustment, the operating efficiency of the adsorption tank and the service life of the adsorption material are significantly improved; optimize operating parameters, reduce system operating costs, and improve economic benefits; timely identify and handle abnormal conditions, reduce equipment failures and production accidents, and improve the safety and reliability of the system; through a closed-loop feedback mechanism, continuously optimize the prediction model to improve the accuracy and reliability of the prediction; combine multi-objective optimization and fuzzy logic control to achieve intelligent management of the adsorption tank and improve the system's adaptability and robustness. Through the above measures, the operating performance and management level of the adsorption tank can be comprehensively improved to achieve efficient, economical, safe and reliable operation.

[0019] Fill missing values, process outliers and standardize data for multidimensional environmental data to generate high-quality data sets; use time series analysis technology to model the concentration trend of gas components and generate gas concentration trend predictions; use machine learning algorithms to predict the performance degradation of adsorption materials based on gas concentration trend predictions and generate adsorption material performance degradation predictions; implement anomaly detection algorithms to identify possible abnormal conditions in the system and generate abnormal operating condition warnings; combine high-quality data sets and various prediction results to conduct comprehensive analysis and generate comprehensive prediction results. By filling missing values, the integrity of data is ensured to avoid prediction bias caused by missing data; by processing outliers, the reliability of data is improved and prediction errors caused by outliers are reduced; by data standardization, the consistency and comparability of data are ensured and the training effect of the model is improved; by using time series analysis technology, the concentration change trend of gas components is modeled to improve the accuracy and reliability of prediction; by using machine learning algorithms, the performance attenuation of adsorption materials is predicted, which can capture complex nonlinear relationships and improve the accuracy of prediction; by implementing anomaly detection algorithms, potential abnormal conditions in the system can be identified in time, early warning can be given, and equipment failures or production accidents caused by abnormal working conditions can be avoided; by comprehensively analyzing the concentration change trend of gas components, the performance attenuation of adsorption materials and abnormal working condition warnings, comprehensive prediction results are generated to provide a scientific basis for the optimized operation of the system; based on the comprehensive prediction results, multi-objective optimization algorithms and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to improve the operating efficiency and stability of the system; the actual operating effect is fed back to the prediction model, and through continuous learning and optimization, a more accurate prediction model is generated to form a closed-loop feedback mechanism to achieve continuous improvement and optimization of the system.

[0020] Using the comprehensive prediction results, determine the target that needs to be optimized and generate the optimization target; based on the optimization target, select the multi-objective optimization algorithm, build the optimization model of the adsorption tank operating parameters, and generate a preliminary optimization plan; using fuzzy logic control theory, by defining fuzzy rules and membership functions, convert the preliminary optimization plan into an easier-to-execute operation instruction, generate fuzzy logic control instructions; according to the fuzzy logic control instructions, intelligently adjust the operation parameters of the adsorption tank to generate optimized operation parameter settings. Through the comprehensive prediction results, clarify the specific goals that need to be optimized to ensure the pertinence and effectiveness of the optimization work. Using the multi-objective optimization algorithm, comprehensively consider multiple optimization targets to generate a more comprehensive and accurate optimization plan; through fuzzy logic control theory, convert complex optimization results into easy-to-understand and easy-to-execute operation instructions to improve the convenience and accuracy of operation; intelligently adjust the operation parameters of the adsorption tank to improve the operating efficiency and stability of the system and extend the service life of the adsorption material; through continuous feedback and adjustment, form a closed-loop optimization mechanism to ensure that the system is always in the best operating state under different working conditions.

[0021] These and other aspects of the present application will become more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 paying any creative work.

[0023] Figure 1 A flow chart of an intelligent control and data analysis method for gas purification in an adsorption tank provided in an embodiment of the present application;

[0024] Figure 2 A schematic diagram of the structure of an intelligent control and data analysis system for gas purification in an adsorption tank provided in an embodiment of the present application;

[0025] Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0027] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0028] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0029] Figure 1 A flow chart of an intelligent control and data analysis method for gas purification in an adsorption tank is provided for the embodiment of the present application, such as Figure 1 As shown, the method includes:

[0030] 101. Collect multidimensional environmental data in the adsorption tank in real time to obtain a multidimensional environmental data set, wherein the multidimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate;

[0031] Obtain real-time environmental data inside the adsorption tank for subsequent analysis and processing; collect multi-dimensional environmental data inside the adsorption tank in real time, including but not limited to gas composition concentration, temperature, pressure, humidity and flow rate, etc.; use sensors and other monitoring equipment to transmit these data to the central processing system.

[0032] 102. Based on the multidimensional environmental data set, using time series analysis and machine learning algorithms, predict the concentration change trend of gas components and the performance attenuation of adsorption materials, and apply anomaly detection algorithms to identify potential abnormal conditions to obtain comprehensive prediction results;

[0033] By analyzing the collected data, the concentration trend of gas components and the performance degradation of adsorption materials are predicted, and potential abnormal conditions are identified; using time series analysis technology, the trend of gas component concentration changes over time is modeled to generate a prediction model; applying machine learning algorithms, based on historical data and current data, predict the performance degradation of adsorption materials under different conditions; using anomaly detection algorithms, possible abnormal conditions in the system are identified, and abnormal condition warnings are generated; application of data preprocessing, time series modeling, machine learning model training, and anomaly detection algorithms;

[0034] Optionally, the method in step 102 is based on the multidimensional environmental data set, using time series analysis and machine learning algorithms to predict and process the concentration change trend of gas components and the performance attenuation of adsorption materials, and applying anomaly detection algorithms to identify potential abnormal operating conditions to obtain comprehensive prediction results, including: using data preprocessing technology to fill missing values, process outliers, and standardize data on the time series data in the multidimensional environmental data set to obtain a high-quality data set; based on the high-quality data set, using time series analysis technology to model the change trend of gas component concentration over time to generate a gas concentration change trend prediction; based on the gas concentration change trend prediction, using machine learning algorithms to predict and process the performance attenuation of adsorption materials under different conditions to obtain an adsorption material performance attenuation prediction; based on the multidimensional environmental data under normal operating conditions, implementing anomaly detection algorithms to identify and process abnormal conditions that may exist in the system to generate abnormal operating condition warnings; using the high-quality data set and the gas concentration change trend prediction, comprehensively analyzing the gas component concentration change trend, the adsorption material performance attenuation, and the abnormal operating condition warning to generate a comprehensive prediction result;

[0035] Suppose we have an adsorption tank for purifying harmful gases in industrial waste gas; use sensors to collect data such as gas component concentration, temperature, pressure, humidity and flow rate in the adsorption tank in real time; the data is recorded once a minute and stored in the database; fill the missing values ​​of the data set with the values ​​of the previous moment; use the Z-score method to identify and process outliers, mark data points that exceed 3 standard deviations as outliers and remove them; perform Min-Max standardization on the data and convert all data to between 0 and 1; select the LSTM model to model the gas component concentration data; perform exploratory analysis on the data and find that the gas component concentration has obvious periodic changes; build an LSTM model and use the data from the past hour to predict the gas component concentration in the next 10 minutes; adjust the model parameters, optimize the prediction accuracy, and finally generate a gas concentration change trend prediction; build a feature set for the adsorption material performance attenuation prediction model, including the gas component concentration change trend prediction value, temperature, pressure, humidity and flow rate; select the random forest algorithm to learn the performance degradation of the adsorbent material under different conditions; use cross-validation technology to evaluate model performance, and optimize the model by adjusting the number and depth of trees; use the optimized model to predict the performance degradation of the adsorbent material under new or variable conditions, and generate a prediction of the performance degradation of the adsorbent material; use the multi-dimensional environmental data under normal operating conditions to generate a baseline model; select the IsolationForest algorithm to perform anomaly detection on the multi-dimensional environmental data collected in real time; combine domain knowledge to analyze the detected abnormal data points to determine whether they represent potential faults or abnormal conditions in the system; when abnormal conditions are detected, issue warning information in a timely manner and generate abnormal condition warnings; use high-quality data sets and gas concentration trend predictions to conduct a comprehensive analysis of gas component concentration trends, adsorbent material performance degradation, and abnormal condition warnings; generate comprehensive prediction results to provide a basis for subsequent intelligent adjustment and optimization;

[0036] Through the above steps, the system can effectively predict and manage the gas purification process in the adsorption tank, improving the purification efficiency and system stability.

[0037] Optionally, based on the high-quality data set, using time series analysis technology to model the change trend of the concentration of the gas component over time to generate a gas concentration change trend forecast, including:

[0038] Using statistical analysis methods, perform exploratory analysis on the gas component concentration data in the high-quality data set, identify the time dependence and periodicity characteristics of the data, and obtain data feature analysis results; based on the data feature analysis results, select a suitable time series analysis model, construct a prediction model for gas component concentration, and generate a preliminary prediction model; using the high-quality data set, train the preliminary prediction model, optimize the prediction accuracy by adjusting model parameters until a predetermined accuracy standard is reached, and obtain an optimized prediction model; based on the optimized prediction model, predict the future gas component concentration change trend and generate a gas concentration change trend prediction; using actual measurement values, verify the gas concentration change trend prediction, evaluate the accuracy and reliability of the prediction model, adjust and optimize the model, and generate a final gas concentration change trend prediction;

[0039] Suppose we have an adsorption tank for purifying sulfur dioxide (SO2) in industrial waste gas; use sensors to collect data such as gas component concentration (SO2), temperature, pressure, humidity and flow rate in the adsorption tank in real time; the data is recorded once a minute and stored in the database; fill the missing values ​​of the data set with the values ​​of the previous moment; use the Z-score method to identify and process outliers, mark data points that exceed 3 standard deviations as outliers and remove them; perform Min-Max standardization on the data, convert all data to between 0 and 1; use autocorrelation graphs and moving average graphs to conduct exploratory analysis on SO2 concentration data; find that SO2 concentration data has obvious daily periodicity characteristics, and the daily concentration changes show a certain pattern; further confirm the data characteristics through time series graphs and box plots; select Proph based on the data feature analysis results. et model, because it can handle data with periodic characteristics well; use high-quality data sets to build a preliminary Prophet model; use high-quality data sets to train the preliminary Prophet model; adjust the model's hyperparameters, such as the length and strength of the seasonal cycle, through cross-validation; until the model reaches the predetermined accuracy standard (for example, RMSE is less than 0.1ppm), generate an optimized prediction model; use the optimized Prophet model to predict future SO2 concentrations; generate a forecast of SO2 concentration trends within the next 24 hours; collect actual measurements and compare them with the forecast results; calculate MAE and RMSE to evaluate the accuracy and reliability of the model; based on the evaluation results, further adjust the model parameters, such as adding more seasonal factors; generate a final SO2 concentration trend forecast;

[0040] Through the above steps, the system can effectively predict the concentration change trend of gas components in the adsorption tank, providing a reliable basis for subsequent intelligent adjustment and optimization.

[0041] Optionally, the prediction of the performance attenuation of the adsorbent material under different conditions is performed using a machine learning algorithm based on the prediction of the gas concentration change trend to obtain a prediction of the performance attenuation of the adsorbent material, including: using the prediction of the gas concentration change trend in combination with multidimensional environmental data to construct a feature set of the adsorbent material performance attenuation prediction model to obtain a complete feature set; based on the complete feature set, using historical data, selecting a suitable machine learning algorithm to learn the performance attenuation of the adsorbent material under different conditions to generate a preliminary performance attenuation prediction model; using cross-validation technology to evaluate the preliminary performance attenuation prediction model, by adjusting model parameters and structure, optimizing the prediction performance of the model until satisfactory accuracy and stability are achieved, to obtain an optimized performance attenuation prediction model; based on the optimized performance attenuation prediction model, predicting the performance attenuation of the adsorbent material under new or variable conditions to generate a prediction of the performance attenuation of the adsorbent material; comparing and verifying the adsorbent material performance attenuation prediction with actual monitoring data, evaluating the applicability and reliability of the prediction model, and generating a final prediction of the adsorbent material performance attenuation;

[0042] Suppose we have an adsorption tank for purifying sulfur dioxide (SO2) in industrial waste gas, and hope to predict the performance attenuation of the adsorption material; use sensors to collect data such as gas component concentration (SO2), temperature, pressure, humidity and flow rate in the adsorption tank in real time; the data is recorded once a minute and stored in the database; the missing values ​​of the data set are filled with the values ​​of the previous moment; the Z-score method is used to identify and process outliers, and data points that exceed 3 standard deviations are marked as outliers and removed; the data is Min-Max standardized to convert all data to between 0 and 1; the predicted value of gas concentration change trend (such as the trend of SO2 concentration change in the next 24 hours) is combined with multidimensional environmental data (temperature, pressure, humidity and flow rate) to construct a feature set; a complete feature set is generated, including the predicted value of gas concentration change trend, temperature, pressure, humidity and flow rate; the random forest algorithm is selected as the machine learning model model; use historical data and a complete feature set to train a preliminary performance decay prediction model; use cross-validation techniques to evaluate the performance of the preliminary performance decay prediction model; optimize the model's prediction performance by adjusting model parameters (such as the number of trees in the random forest, the maximum depth) and structure; until the model reaches satisfactory accuracy and stability (for example, RMSE is less than 0.05), generate an optimized performance decay prediction model; use the optimized performance decay prediction model to predict the performance of the adsorbent material under new or variable conditions; generate performance decay predictions for the adsorbent material under different conditions, such as predicting the performance decay of the adsorbent material within the next 24 hours; collect actual monitoring data and compare it with the prediction results; calculate MAE and RMSE to evaluate the applicability and reliability of the model; based on the evaluation results, further adjust the model parameters, such as adding more features or using different machine learning algorithms; generate the final adsorbent material performance decay prediction;

[0043] Through the above steps, the system can effectively predict the performance attenuation of the adsorption material under different conditions, providing a scientific basis for the optimized operation and maintenance of the adsorption tank.

[0044] The present application considers that the formula is used to generate a prediction model for the performance decay of adsorbent materials under different conditions. Specifically, it is a comprehensive function trained with historical data and a complete feature set. The model combines a support vector machine, additional feature functions, and regularization terms to improve the accuracy and robustness of the prediction;

[0045] Optionally, based on the complete feature set, using historical data, selecting a suitable machine learning algorithm, learning and processing the performance attenuation of the adsorption material under different conditions, and generating a preliminary performance attenuation prediction model, including: using historical data and a complete feature set, selecting a suitable machine learning algorithm, and training to obtain model parameters:

[0046]

[0047] Among them, f(x) is a comprehensive function used to generate a prediction model of the performance decay of adsorption materials under different conditions, x = [C t ,T,P,H,F] is the input vector, which contains the predicted value C of the gas component concentration change trend t , temperature T, pressure P, humidity H and flow rate F; N is the number of support vectors; α i is the Lagrange multiplier; y i is the label of the i-th support vector; w i (z) is the adaptive weight of the i-th support vector; is the dynamic exponential term of the i-th support vector; K(z i ,x) is the kernel function; b is the bias term; M is the number of additional feature functions; β j is the weight of the additional feature function; g j (x) is the additional feature function; L is the number of regularization terms; γ k is the weight of the regularization term; h k (x) is the regularization term; i represents the index of the support vector, from 1 to N; j represents the index of the additional feature function, from 1 to M; k represents the index of the regularization term, from 1 to L;

[0048] For a new input vector x = [C t ,T,P,H,F], calculate the adaptive weight w of each support vector i (x) and the dynamic exponential term And the kernel function K(x i ,x):

[0049]

[0050] Among them, β is the weight adjustment parameter, is the average value of all support vector labels, λ is the distance adjustment parameter, ρ is the region distance adjustment parameter, σ is the similarity adjustment parameter, μ i is the mean of the i-th support vector, dist(x,x i ) is the input vector x and the support vector x i The distance, similarity(x,x i ) is the input vector x and the support vector x i similarity;

[0051]

[0052] Among them, α is the exponential adjustment parameter of the mean; γ is the exponential adjustment parameter of the standard deviation; δ is the exponential adjustment parameter of the variance; η is the exponential adjustment parameter of the skewness; φ is the exponential adjustment parameter of the kurtosis; ψ is the exponential adjustment parameter of the extract; mean(x i ) is the mean of the i-th support vector; std(x i ) is the standard deviation of the i-th support vector; var(x i ) is the variance of the i-th support vector; skew(x i ) is the skewness of the i-th support vector; kurt(x i ) is the kurtosis of the i-th support vector; entropy(x i ) is the entropy of the i-th support vector;

[0053] K(x i ,x)=exp(-γ||x i -x|| 2 )

[0054] Among them, γ is the parameter of the kernel function;

[0055] Calculate the additional characteristic function g j (x) and the regularization term h k (x):

[0056] g j (x) = exp(α j mean(x)+γ j std(x)+δ j var(x)+η j skew(x)+φ j

[0057] ·kurt(x)+ψ j ·entropy(x))

[0058] Among them, α j is the exponential adjustment parameter of the mean; γ j is the exponential adjustment parameter of the standard deviation; δ j is the exponential adjustment parameter of the variance; η j is the exponential adjustment parameter for skewness; φ j is the exponential adjustment parameter of kurtosis; ψ j is the exponential adjustment parameter of entropy;

[0059] h k (x) = exp(-λ k ||x-μ k || 2 )

[0060] Among them, λ kis the regularization parameter; μ k is the center of the regularization term;

[0061] Use the validation set data to evaluate the predictive performance of the model, adjust the model parameter settings based on the evaluation results, repeat the test and tuning, and generate a preliminary performance decay prediction model.

[0062] Suppose there is an adsorption tank used to purify sulfur dioxide (SO2) in industrial waste gas, and you want to predict the performance decay of the adsorption material:

[0063] The historical data set contains 1000 samples, each of which includes the predicted value C of the gas component concentration change trend. t , temperature T, pressure P, humidity H and flow rate F; Example of input vector x: x = [C t ,T,P,H,F]=[0.5,25,1013,50,2]; select support vector machine as the machine learning algorithm; use historical data and complete feature set to train the model and obtain model parameters; assume that the first support vector x1=[0.6,26,1014,51,2.1], label y1=1, mean μ1=[0.6,26,1014,51,2.1]; set the parameters to β=0.1, y - =0.5, λ=0.01, ρ=0.01, σ=0.01; calculate the distance

[0064] The calculated similarity is similarity(x,x1)=0.95; the calculated adaptive weight w1(x) is

[0065] Assume that the statistical characteristics of the first support vector are mean(x1)=0.6, standard deviation std(x1)=0.1, variance var(x1)=0.01, skew(x1)=0.05, kurt(x1)=3.0, entropy(x1)=0.5, and the parameters are set to α=0.1, γ=0.1, δ=0.1, η=0.1, φ=0.1, ψ=0.1; calculate the dynamic index term c ∧ 1(x) is c ∧ 1(x)=exp(0.1·0.6+0.1·0.1+0.1·0.01+0.1·0.05+0.1·3.0+0.1·0.5)≈1.65; Assuming the kernel function parameter γ=0.1, the kernel function K(x1,x) is calculated as

[0066] K(x1,x)=exp(-0.1·2.236 2)≈0.65; Assume that the statistical characteristics of the first additional feature function are mean mean(x)=0.5, standard deviation std(x)=0.1, variance var(x)=0.01, skewness skew(x)=0.05, kurtosis kurt(x)=3.0, entropy entropy(x)=0.5, and the parameters are set to α1=0.1, γ1=0.1, δ1=0.1, η1=0.1, φ1=0.1, ψ1=0.1; calculate the additional feature function g1(x) as g1(x)=exp(0.1·0.5+0.1·0.1+0.1·0.01+0.1·0.05+0.1·3.0+0.1·0.5)≈1.65; Assume that the regularization parameter λ k = 0.1, the center of the regularization term μ k =[0.5,25,1013,50,2], calculate the regularization term h1(x) as h1(x)=exp(-0.1·0)=1.0; assume that the number of support vectors N=1, the Lagrange multiplier α1=1, the bias term b=0, the number of additional feature functions M=1, the number of regularization terms L=1, the weight of the additional feature function β1=1, and the weight of the regularization term γ1=1; calculate the comprehensive function f(x) as f(x)=1·1·0.88·1.65·0.65+0+1·1.65+1·1.0≈3.47;

[0067] Through the above steps, the predicted value of the performance attenuation of the adsorption material under specific conditions is obtained. This predicted value can be used to guide the operation and maintenance of the adsorption tank and improve the efficiency and stability of the system.

[0068] Optionally, the multi-dimensional environmental data under normal operating conditions is used to implement an anomaly detection algorithm, identify and process abnormal conditions that may exist in the system, and generate abnormal operating condition warnings, including: using the multi-dimensional environmental data under normal operating conditions to perform statistical analysis and processing on the distribution characteristics of gas composition parameters to obtain a baseline model; based on the baseline model, selecting a suitable anomaly detection algorithm to perform anomaly detection processing on the multi-dimensional environmental data collected in real time to generate abnormal data points; using the abnormal data points, combined with domain knowledge, further analyzing and processing these abnormal data points to determine whether they represent potential faults or abnormal operating conditions in the system, and generating a preliminary abnormal operating condition report; formulating a corresponding warning strategy based on the preliminary abnormal operating condition report, and when an abnormal operating condition is detected, issuing warning information in a timely manner to generate an abnormal operating condition warning;

[0069] Suppose we have an adsorption tank for purifying sulfur dioxide (SO2) in industrial waste gas, and hope to identify abnormal conditions that may exist in the system; collect multidimensional environmental data of the system under normal operating conditions, including gas component concentration (SO2), temperature, pressure, humidity and flow rate; the data is recorded once a minute and stored in the database; perform statistical analysis on the multidimensional environmental data under normal operating conditions, and calculate the mean, standard deviation, maximum, minimum and other statistics of each parameter; generate a baseline model with a SO2 concentration mean of 0.5ppm, a temperature mean of 25℃, a pressure mean of 1013hPa, a humidity mean of 50%, a flow rate mean of 2m / s, and statistics such as the standard deviation of each parameter; select IsolationForest as the anomaly detection algorithm; use the baseline model and IsolationForest to perform anomaly detection on the multidimensional environmental data collected in real time; an example of real-time collected data is x=[0.8,30,1015,60,3]; use IsolationForest to perform anomaly detection on real-time data, generate abnormal data points, and the number of abnormal data points The abnormal data point is x1 = [0.8, 30, 1015, 60, 3], and the abnormal data point 2 is x2 = [0.7, 32, 1016, 65, 3.5]; Combined with domain knowledge, further analyze the abnormal data points; Determine whether these abnormal data points represent potential faults or abnormal conditions in the system; Generate a preliminary abnormal condition report, abnormal condition 1 is a significant increase in SO2 concentration, which may indicate adsorption material failure; abnormal condition 2 is a simultaneous increase in temperature and humidity, which may indicate equipment overheating or environmental changes; Based on the preliminary abnormal condition report, formulate The corresponding warning strategy is to set the SO2 concentration threshold to 0.7ppm, and issue a first-level warning when it exceeds the threshold; set the temperature threshold to 30℃, and issue a second-level warning when it exceeds the threshold; define the alarm level, such as the first-level warning notifies the operator, and the second-level warning notifies the engineer; when an abnormal operating condition is detected, the system automatically issues a warning message to notify relevant personnel to take necessary measures; the abnormal operating condition warning is generated as the first-level warning when the SO2 concentration exceeds 0.7ppm, and it is recommended to check the adsorption material; the second-level warning is when the temperature exceeds 30℃, and it is recommended to check the equipment cooling system;

[0070] Through the above steps, the system can effectively identify and warn of potential abnormal conditions to ensure the normal operation and safety of the adsorption tank.

[0071] 103. Based on the comprehensive prediction results, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings;

[0072] According to the comprehensive prediction results, optimize the operating parameters of the adsorption tank to improve the gas purification efficiency and system stability; determine the optimization target, select the appropriate multi-objective optimization algorithm, build the optimization model, and generate a preliminary optimization plan; use fuzzy logic control theory to convert the optimization results into specific operating instructions and generate fuzzy logic control instructions; define the optimization target, select the multi-objective optimization algorithm, build the optimization model, and generate a preliminary optimization plan; define fuzzy rules and membership functions, and convert the optimization results into specific operating instructions;

[0073] Optionally, in step 103, based on the comprehensive prediction result, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings, including:

[0074] Using the comprehensive prediction results, the targets to be optimized are determined and processed to obtain the optimization targets; based on the optimization targets, a multi-objective optimization algorithm is selected to construct an optimization model for the operating parameters of the adsorption tank to obtain a preliminary optimization scheme; using fuzzy logic control theory, the preliminary optimization scheme is processed, and by defining fuzzy rules and membership functions, the optimization results are converted into easier-to-execute operating instructions to generate fuzzy logic control instructions; according to the fuzzy logic control instructions, the operating parameters of the adsorption tank are intelligently adjusted to generate optimized operating parameter settings.

[0075] Assume that there is an adsorption tank for purifying sulfur dioxide in industrial waste gas, and it is hoped to improve the gas purification efficiency and system stability by optimizing the operating parameters; analyze the comprehensive prediction results and determine the optimization target. For example, the prediction results show that the SO2 concentration will increase in the next 24 hours, and the optimization target is to improve the gas purification efficiency; at the same time, the prediction shows that the performance of the adsorption material decays rapidly, and the optimization target also includes extending the life of the adsorption material; select NSGA-II as the multi-objective optimization algorithm; construct an optimization model and define the relationship between the operating parameters and the optimization target; the optimization model can be expressed as MaximizeE(T,P,F)and MaximizeL(T,P,F), E(T,P,F) represents the gas purification efficiency, L(T,P,F) represents the life of the adsorption material; use NSGA-II to solve the optimization model and generate a preliminary optimization plan, temperature T = 26℃, pressure P = 1014hPa, flow rate F = 2.1m / s; define fuzzy rules to describe the fuzzy relationship between operating parameters and optimization objectives, rule 1: if the temperature is high, reduce the flow rate, rule 2: if the pressure is low, increase the temperature; define membership functions to describe the degree to which the input variables belong to a fuzzy set, the membership functions of temperature are low temperature (0-20℃), medium temperature (20-30℃), high temperature (30-40℃), and the membership functions of pressure are low pressure (1000-1010hPa), medium pressure (1010-1020hPa ), high pressure (1020-1030hPa), the membership function of the flow rate is low speed (0-1.5m / s), medium speed (1.5-2.5m / s), high speed (2.5-3.5m / s); use the fuzzy logic controller to convert the preliminary optimization plan into specific operation instructions. If the current temperature is 26℃ (medium temperature), the pressure is 1014hPa (medium pressure), and the flow rate is 2.1m / s (medium speed), according to the fuzzy rules, rule 1 is that the temperature is high and the flow rate is reduced to 2.0m / s, and rule 2 is that the pressure is moderate and the temperature is kept unchanged; generate fuzzy logic control instructions: adjust the flow rate to 2.0m / s; according to the fuzzy logic control instructions, adjust the operating parameters of the adsorption tank; generate optimized operating parameters to set temperature T=26℃, pressure P=1014hPa, and flow rate F=2.1m / s;

[0076] Through the above steps, the system can intelligently adjust the operating parameters of the adsorption tank according to the comprehensive prediction results to improve the gas purification efficiency and system stability. This optimization method not only considers a single goal, but also takes into account multiple optimization goals to ensure that the system operates in the optimal state.

[0077] This application considers that the formula is used to construct a multi-objective optimization model for the operating parameters of the adsorption tank. Multiple optimization objectives are integrated through a comprehensive objective function, including minimizing the prediction error of the gas component concentration change trend, minimizing the system operation cost, minimizing the abnormal condition warning frequency, and maximizing the system stability. Each objective function has its own specific expression, which is combined into a comprehensive objective function by weighted sum;

[0078] Optionally, based on the optimization objective, a suitable multi-objective optimization algorithm is selected to construct an optimization model of the adsorption tank operating parameters to obtain a preliminary optimization scheme, including:

[0079] Integrate multiple optimization objectives through the comprehensive objective function f(x):

[0080]

[0081] Where x=[x1,x2,…,x n ] is the operating parameter vector, which contains the operating parameters of the adsorption; m is the number of objective functions; ω i is the weight of the i-th target, satisfying f i (x) is the i-th objective function; f i min is the minimum value of the i-th objective function; f i max is the maximum value of the i-th objective function; γ i (x) is the nonlinear adjustment parameter of the i-th objective function, which depends on the operating parameter vector; λ i is the distance adjustment parameter; is the optimal operating parameter vector of the i-th objective function; is the operating parameter vector x and the optimal operating parameter vector The distance; i represents the index of the weight, from 1 to m;

[0082] The prediction error of the gas component concentration trend is minimized as follows:

[0083]

[0084] The system operating cost is minimized to

[0085]

[0086] The frequency of abnormal condition warning is minimized to

[0087]

[0088] The maximum stability of the system is

[0089]

[0090] Where T is the number of time steps; is the predicted value of the gas component concentration; is the actual value of the gas component concentration; c i is the cost coefficient of the ith operating parameter; d ij is the interaction cost coefficient between the operation parameters; I(anomaly t ) is the abnormal condition indicator function at time t, which is 1 if there is an abnormality, otherwise 0; is the predicted value of system pressure; is the actual value of the system pressure; i represents the index of the cost coefficient, from 1 to n; t represents the index of the time step number, from 1 to T;

[0091] The values ​​of each objective function are calculated by the above formula, and these values ​​are substituted into the comprehensive objective function f(x), the operation parameter vector x is optimized to obtain the optimization result; according to the operation parameter vector x obtained by the optimization result, a preliminary optimization scheme is obtained;

[0092] Suppose there is an adsorption tank for purifying sulfur dioxide in industrial waste gas, and it is hoped to improve the gas purification efficiency and system stability by optimizing the operating parameters; the operating parameter vector x = [T, P, F], where T is temperature, P is pressure, and F is flow rate; the objective function is f1(x) to minimize the prediction error of the gas component concentration trend; f2(x) to minimize the system operating cost; f3(x) to minimize the abnormal condition warning frequency; f4(x) to maximize the system stability; weights ω1 = 0.3, ω2 = 0.2, ω3 = 0.2, ω4 = 0.3; nonlinear adjustment parameters γ1(x) = 1, γ2(x) = 1, γ3(x) = 1, γ4(x) = 1; distance adjustment parameters λ1 = 0.1, λ2 = 0.1, λ3 = 0.1, λ4 = 0.1; optimal operating parameter vector Assume that the current operating parameter x = [26, 1014, 2.1]; the gas component concentration trend prediction error is minimized as follows:

[0094] Assume T = 10, the predicted value and the actual value are Calculated Minimize system operating costs Assume c1 = 0.1, c2 = 0.1, c3 = 0.1, d 12 =0.01, d 13 =0.01, d 23 = 0.01; calculated f2(x) = 0.1·26 2 +0.1 1014 2+0.1 2.1 2 +0.01·26·1014+0.01·26·2.1+0.01·1014·2.1≈103000; the frequency of abnormal operating condition warning is minimized Assume that T = 10, the abnormal condition indicator functions are I (anomaly t )=[0,0,1,0,0,1,0,0,0,0]; calculated f3(x)=log(1+0)+log(1+0)+log(1+1)+log(1+0)+log(1+0)+log(1+1)+log(1+0)+log(1+0); maximize the stability of the system Assume T = 10, the predicted value and the actual value are Calculated

[0095]

[0096] Assume f1 min =0,f1 max =1,f2 min =0,f2 max =100000,f3 min =0,f3 max =2,f4 min =-1, f4 max =0, calculate distance for

[0097]

[0098] The calculation of the comprehensive objective function value is

[0099]

[0100] f(x)=0.3·0·exp(-0.147)+0.2·1.03·exp(-0.147)+0.2·0.693·exp(-0.147)+0.3·1·exp(-0.147)

[0101] f(x)=0+0.2·1.03·0.863+0.2·0.693·0.863+0.3·1·0.863

[0102] f(x)=0+0.180+0.121+0.259≈0.560

[0103] Through the above steps, we obtain the comprehensive objective function value f(x)≈0.560 of the operating parameter vector x=[26,1014,2.1]. This value can be used to guide the adjustment of the operating parameters of the adsorption tank to achieve multi-objective optimization.

[0104] Optionally, the preliminary optimization scheme is processed by using fuzzy logic control theory, and the optimization result is converted into an operation instruction that is easier to execute by defining fuzzy rules and membership functions, thereby generating fuzzy logic control instructions, including:

[0105] Utilizing the preliminary optimization scheme, the optimization results of the adsorption tank operating parameters are analyzed and processed to determine the key parameters that need to be converted into operating instructions and obtain a key parameter list; based on the key parameter list, fuzzy rules are defined to describe the nonlinear relationship and interaction between different parameters and generate a fuzzy rule set; membership functions are defined to convert specific numerical values ​​into membership degrees in fuzzy sets, each set corresponds to a membership function, indicating the degree to which different temperature values ​​belong to each set, and a membership function set is generated; utilizing the fuzzy rule set and membership function set, the specific numerical values ​​in the optimization results are converted into fuzzy logic control instructions and fuzzy logic control instructions are generated.

[0106] Assume that there is an adsorption tank used to purify sulfur dioxide in industrial waste gas, and it is hoped that the gas purification efficiency and system stability can be improved by optimizing the operating parameters;

[0107] Analyze the preliminary optimization plan and determine the key parameters that need to be converted into operating instructions; the operating parameters recommended by the preliminary optimization plan are temperature T = 26 ° C, pressure P = 1014hPa, and flow rate F = 2.1m / s; the key parameter list is temperature T, pressure P, and flow rate F; according to the key parameter list, define fuzzy rules to describe the nonlinear relationship and interaction between different parameters. Rule 1 is if the temperature is high, then reduce the flow rate, Rule 2 is if the pressure is low, then increase the temperature, and Rule 3 is if the flow rate is high, then reduce the pressure; the generated fuzzy rule set is, Rule 1 is if the temperature is high, then the flow rate is low, Rule 2 is that if the pressure is lower, the temperature is higher, and Rule 3 is that if the flow rate is higher, the pressure is lower; define the membership function, convert the specific value into the membership degree in the fuzzy set, the membership function of temperature TT is, low temperature (0-20℃) is a triangular membership function, medium temperature (20-30℃) is a trapezoidal membership function, and high temperature (30-40℃) is a triangular membership function; the membership function of pressure PP is, low pressure (1000-1010hPa) is a triangular membership function, medium pressure (1010-1020hPa) is a trapezoidal membership function, and high pressure (1020-1030hPa) is a triangular membership function. The membership function of the flow velocity FF is: low speed (0-1.5m / s) is a triangular membership function, medium speed (1.5-2.5m / s) is a trapezoidal membership function, and high speed (2.5-3.5m / s) is a triangular membership function; using the fuzzy rule set and the membership function set, the specific values ​​in the optimization result are converted into fuzzy logic control instructions. The current operating parameters are temperature T = 26°C, pressure P = 1014hPa, and flow velocity F = 2.1m / s; according to the membership function, the specific values ​​are converted into membership in the fuzzy set. The temperature T = 26°C belongs to medium temperature, and the pressure P = 1014hPa a belongs to medium pressure, and flow rate F=2.1m / s belongs to medium speed; the fuzzy rules applied are: Rule 1 is if the temperature is higher, the flow rate is lower. Since temperature T=26℃ belongs to medium temperature, this rule is not triggered; Rule 2 is if the pressure is lower, the temperature is higher. Since pressure P=1014hPa belongs to medium pressure, this rule is not triggered; Rule 3 is if the flow rate is higher, the pressure is lower. Since flow rate F=2.1m / s belongs to medium speed, this rule is not triggered; Combining the above rules, fuzzy logic control instructions are generated: keep the current operating parameters unchanged, that is, T=26℃, pressure P=1014hPa, flow rate F=2.1m / s.

[0108] Through the above steps, the system can generate specific operating instructions based on the preliminary optimization plan through fuzzy logic control theory to ensure that the adsorption tank operates in the optimal state. This method can not only handle complex nonlinear relationships, but also flexibly respond to different operating conditions and improve the stability and efficiency of the system.

[0109] 104. Apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor the implementation effect, and feed back the actual operation effect to the prediction model to generate a more accurate prediction model.

[0110] Apply the optimized operating parameters to actual operation, monitor the implementation effect, and feed back the actual operation effect to the prediction model to further optimize the model; apply the generated optimized operating parameter settings to the actual operation of the adsorption tank; monitor the operating status of the adsorption tank in real time and collect actual operation data; feed back the actual operation effect to the prediction model, evaluate the accuracy and reliability of the model, and adjust and optimize the model based on the feedback results; implement the optimized operating parameters, monitor the system operating status in real time, collect actual operation data, feed the data back to the prediction model, and adjust and optimize the model.

[0111] Figure 2 A schematic diagram of the structure of an intelligent control and data analysis system for gas purification in an adsorption tank is provided for the embodiment of the present application, such as Figure 2 As shown, the device comprises:

[0112] The acquisition module 21 is used to acquire multi-dimensional environmental data in the adsorption tank in real time to obtain a multi-dimensional environmental data set, wherein the multi-dimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate;

[0113] A processing module 22 is used to predict the concentration change trend of gas components and the performance attenuation of adsorption materials based on the multi-dimensional environmental data set by using time series analysis and machine learning algorithms, and to identify potential abnormal conditions by using an anomaly detection algorithm to obtain a comprehensive prediction result;

[0114] A generation module 23 is used to intelligently adjust the operating parameters of the adsorption tank according to the comprehensive prediction results by using a multi-objective optimization algorithm and fuzzy logic control theory to generate optimized operating parameter settings;

[0115] The monitoring module 24 is used to apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor the implementation effect, and feed back the actual operation effect to the prediction model to generate a more accurate prediction model.

[0116] Figure 2 The intelligent control and data analysis system for gas purification in an adsorption tank can be executed Figure 1The implementation principle and technical effect of the intelligent control and data analysis method for gas purification in an adsorption tank described in the embodiment shown will not be repeated. The specific way in which each module and unit performs operations in the intelligent control and data analysis system for gas purification in an adsorption tank in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0117] In one possible design, Figure 2 The intelligent control and data analysis system for gas purification in an adsorption tank of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0118] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0119] The processing component 32 is used to: collect multidimensional environmental data in the adsorption tank in real time to obtain a multidimensional environmental data set, wherein the multidimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate; based on the multidimensional environmental data set, use time series analysis and machine learning algorithms to predict the concentration change trend of the gas components and the performance attenuation of the adsorption material, and use anomaly detection algorithms to identify potential abnormal conditions to obtain comprehensive prediction results; based on the comprehensive prediction results, use multi-objective optimization algorithms and fuzzy logic control theory to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings; apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor its implementation effect, and feed back the actual operation effect to the prediction model to generate a more accurate prediction model.

[0120] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

[0121] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0122] Of course, the computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0123] The input / output interface provides an interface between the processing component and the peripheral interface module, which may be an output device, an input device, etc.

[0124] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0125] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0126] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The illustrated embodiment is an intelligent control and data analysis method for gas purification in an adsorption tank.

[0127] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0128] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0129] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent control and data analysis method for gas purification in an adsorption tank, characterized in that: include: Collecting multi-dimensional environmental data in the adsorption tank in real time to obtain a multi-dimensional environmental data set, wherein the multi-dimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate; Based on the multidimensional environmental data set, the concentration change trend of the gas components and the performance attenuation of the adsorption materials are predicted and processed using time series analysis and machine learning algorithms, and anomaly detection algorithms are applied to identify potential abnormal conditions to obtain comprehensive prediction results; According to the comprehensive prediction results, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings; The optimized operating parameter settings are applied to the actual operation of the adsorption tank, the implementation effect is monitored, and the actual operation effect is fed back to the prediction model to generate a more accurate prediction model.

2. The method according to claim 1, characterized in that Based on the multi-dimensional environmental data set, the concentration change trend of the gas components and the performance attenuation of the adsorption materials are predicted by using time series analysis and machine learning algorithms, and anomaly detection algorithms are applied to identify potential abnormal conditions to obtain comprehensive prediction results, including: Using data preprocessing technology, the time series data in the multidimensional environmental data set is filled with missing values, processed with outliers, and normalized to obtain a high-quality data set; Based on the high-quality data set, using time series analysis technology, modeling and processing the change trend of the concentration of gas components over time, and generating a gas concentration change trend forecast; According to the gas concentration change trend prediction, a machine learning algorithm is used to predict the performance attenuation of the adsorption material under different conditions to obtain a prediction of the performance attenuation of the adsorption material; Based on the multi-dimensional environmental data under normal operating conditions, anomaly detection algorithms are implemented to identify and process possible abnormal conditions in the system and generate abnormal operating condition warnings; By using the high-quality data set and the gas concentration change trend prediction, a comprehensive analysis is performed on the gas component concentration change trend, the adsorption material performance attenuation and the abnormal operating condition warning to generate a comprehensive prediction result.

3. The method according to claim 2, characterized in that Based on the high-quality data set, the time series analysis technology is used to model the change trend of the concentration of the gas component over time to generate a gas concentration change trend forecast, including: Using statistical analysis methods, an exploratory analysis is performed on the gas component concentration data in the high-quality data set to identify the time dependency and periodicity characteristics of the data and obtain data feature analysis results; Based on the data feature analysis results, a suitable time series analysis model is selected, a prediction model for gas component concentration is constructed and processed, and a preliminary prediction model is generated; Using the high-quality data set, the preliminary prediction model is trained and processed, and the prediction accuracy is optimized by adjusting the model parameters until a predetermined accuracy standard is reached, thereby obtaining an optimized prediction model; Based on the optimized prediction model, the future gas component concentration change trend is predicted and processed to generate a gas concentration change trend prediction; The actual measured values ​​are used to verify the gas concentration change trend prediction, evaluate the accuracy and reliability of the prediction model, adjust and optimize the model, and generate the final gas concentration change trend prediction.

4. The method according to claim 2, characterized in that: The method of predicting the performance attenuation of the adsorption material under different conditions using a machine learning algorithm based on the gas concentration change trend prediction to obtain the adsorption material performance attenuation prediction includes: By using the gas concentration change trend prediction and combining it with multi-dimensional environmental data, a feature set of an adsorption material performance attenuation prediction model is constructed and processed to obtain a complete feature set; Based on the complete feature set, using historical data, selecting a suitable machine learning algorithm, learning and processing the performance decay of the adsorption material under different conditions, and generating a preliminary performance decay prediction model; Using cross-validation technology, the preliminary performance degradation prediction model is evaluated and processed, and the prediction performance of the model is optimized by adjusting model parameters and structure until satisfactory accuracy and stability are achieved, thereby obtaining an optimized performance degradation prediction model; Based on the optimized performance decay prediction model, predict the performance decay of the adsorption material under new or variable conditions to generate a prediction of the performance decay of the adsorption material; Comparing and verifying the adsorption material performance decay prediction with actual monitoring data, evaluating the applicability and reliability of the prediction model, and generating a final adsorption material performance decay prediction; Wherein, based on the complete feature set, using historical data, selecting a suitable machine learning algorithm, learning and processing the performance attenuation of the adsorption material under different conditions, and generating a preliminary performance attenuation prediction model, including: Using historical data and a complete feature set, select an appropriate machine learning algorithm and train to obtain model parameters: Among them, f(x) is a comprehensive function used to generate a prediction model of the performance decay of adsorption materials under different conditions, x = [C t ,T,P,H,F] is the input vector, which contains the predicted value C of the gas component concentration change trend t , temperature T, pressure P, humidity H and flow rate F; N is the number of support vectors; α i is the Lagrange multiplier; y i is the label of the i-th support vector; w i (x) is the adaptive weight of the i-th support vector; is the dynamic exponential term of the i-th support vector; K(x i ,x) is the kernel function; b is the bias term; M is the number of additional feature functions; β j is the weight of the additional feature function; g j (x) is the additional feature function; L is the number of regularization terms; γ k is the weight of the regularization term; h k (x) is the regularization term; i represents the index of the support vector, from 1 to N; j represents the index of the additional feature function, from 1 to M; k represents the index of the regularization term, from 1 to L. For a new input vector x = [C t ,T,P,H,F], calculate the adaptive weight w of each support vector i (x) and the dynamic exponential term And the kernel function K(x i ,x): Among them, β is the weight adjustment parameter, is the average value of all support vector labels, λ is the distance adjustment parameter, ρ is the region distance adjustment parameter, σ is the similarity adjustment parameter, μ i is the mean of the i-th support vector, dist(x,x i ) is the input vector x and the support vector x i The distance, similarity(x,x i ) is the input vector x and the support vector x i similarity; Among them, α is the exponential adjustment parameter of the mean; γ is the exponential adjustment parameter of the standard deviation; δ is the exponential adjustment parameter of the variance; η is the exponential adjustment parameter of the skewness; φ is the exponential adjustment parameter of the kurtosis; ψ is the exponential adjustment parameter of the extract; mean(x i ) is the mean of the i-th support vector; std(x i ) is the standard deviation of the i-th support vector; var(x i ) is the variance of the i-th support vector; skew(x i ) is the skewness of the i-th support vector; kurt(x i ) is the kurtosis of the i-th support vector; entropy(x i ) is the entropy of the i-th support vector; K(x i ,x)=exp(-γ∥x i -x∥ 2 ) Among them, γ is the parameter of the kernel function; Calculate the additional characteristic function g j (x) and the regularization term h k (x): g j (z)=exp(α j ·mean(x)+γ j ·std(x)+δ j ·var(x)+η j ·skew(x)+φ j ·kurt(x)+ψ j ·entropy(x)) Among them, α j is the exponential adjustment parameter of the mean; γ j is the exponential adjustment parameter of the standard deviation; δ j is the exponential adjustment parameter of the variance; η j is the exponential adjustment parameter for skewness; φ j is the exponential adjustment parameter of kurtosis; ψ j is the exponential adjustment parameter of entropy; h k (x)=exp(-λ k ∥x-μ k ∥ 2 ) Among them, λ k is the regularization parameter; μ k is the center of the regularization term; Use the validation set data to evaluate the predictive performance of the model, adjust the model parameter settings based on the evaluation results, repeat the test and tuning, and generate a preliminary performance decay prediction model.

5. The method according to claim 2, characterized in that: The method implements an abnormality detection algorithm based on multi-dimensional environmental data under normal operating conditions, identifies and processes abnormal conditions that may exist in the system, and generates abnormal operating condition warnings, including: Using multi-dimensional environmental data under normal operating conditions, the distribution characteristics of gas composition parameters are statistically analyzed and processed to obtain a baseline model; Based on the baseline model, a suitable anomaly detection algorithm is selected to perform anomaly detection processing on the multi-dimensional environmental data collected in real time to generate anomaly data points; Using the abnormal data points, combined with domain knowledge, further analyzing and processing these abnormal data points to determine whether they represent potential faults or abnormal conditions in the system, and generating a preliminary abnormal condition report; According to the preliminary abnormal operating condition report, a corresponding early warning strategy is formulated, and when an abnormal operating condition is detected, early warning information is issued in a timely manner to generate an abnormal operating condition warning.

6. The method according to claim 1, characterized in that According to the comprehensive prediction results, a multi-objective optimization algorithm and fuzzy logic control theory are used to intelligently adjust the operating parameters of the adsorption tank to generate optimized operating parameter settings, including: Using the comprehensive prediction results, determining the target to be optimized to obtain the optimization target; Based on the optimization goal, a multi-objective optimization algorithm is selected to construct an optimization model of the adsorption tank operation parameters to obtain a preliminary optimization solution; Using fuzzy logic control theory, the preliminary optimization scheme is processed, and by defining fuzzy rules and membership functions, the optimization results are converted into easier-to-execute operation instructions to generate fuzzy logic control instructions; According to the fuzzy logic control instructions, the operating parameters of the adsorption tank are intelligently adjusted to generate optimized operating parameter settings; Wherein, based on the optimization objective, a suitable multi-objective optimization algorithm is selected to construct an optimization model of the adsorption tank operating parameters to obtain a preliminary optimization scheme, including: Integrate multiple optimization objectives through the comprehensive objective function f(x): Where x=[x1,x2,…,x n ] is the operating parameter vector, which contains the operating parameters of the adsorption; m is the number of objective functions; ω i is the weight of the ith objective, satisfying f i (x) is the i-th objective function; f i min is the minimum value of the i-th objective function; f i max is the maximum value of the i-th objective function; γ i (x) is the nonlinear adjustment parameter of the i-th objective function, which depends on the operating parameter vector; λ i is the distance adjustment parameter; is the optimal operating parameter vector of the i-th objective function; is the operating parameter vector x and the optimal operating parameter vector distance; Minimize the prediction error of gas component concentration trend: Minimize system operating costs: Minimize the frequency of abnormal condition warnings: Maximizing system stability: Where T is the number of time steps; is the predicted value of the gas component concentration; is the actual value of the gas component concentration; c i is the cost coefficient of the ith operating parameter; d ij is the interaction cost coefficient between the operation parameters; is the abnormal condition indicator function at time t, which is 1 if there is an abnormality, otherwise 0; is the predicted value of system pressure; is the actual value of the system pressure; The values ​​of each objective function are calculated by the above formula, and these values ​​are substituted into the comprehensive objective function f(x), the operating parameter vector x is optimized to obtain the optimization result; based on the operating parameter vector x obtained from the optimization result, a preliminary optimization plan is obtained.

7. The method according to claim 6, characterized in that The fuzzy logic control theory is used to process the preliminary optimization scheme, and the optimization results are converted into easier-to-execute operation instructions by defining fuzzy rules and membership functions to generate fuzzy logic control instructions, including: Utilizing the preliminary optimization scheme, analyzing and processing the optimization results of the adsorption tank operating parameters, determining the key parameters that need to be converted into operating instructions, and obtaining a key parameter list; Based on the key parameter list, fuzzy rules are defined to describe the nonlinear relationship and interaction between different parameters, and a fuzzy rule set is generated; Define membership functions to convert specific values ​​into membership in fuzzy sets. Each set corresponds to a membership function, which indicates the degree to which different temperature values ​​belong to each set, and generate a membership function set. The fuzzy rule set and the membership function set are used to convert specific numerical values ​​in the optimization result into fuzzy logic control instructions to generate fuzzy logic control instructions.

8. An intelligent control and data analysis system for gas purification in an adsorption tank, characterized in that: include: A collection module is used to collect multi-dimensional environmental data in the adsorption tank in real time to obtain a multi-dimensional environmental data set, wherein the multi-dimensional environmental data includes gas component concentration, temperature, pressure, humidity and flow rate; A processing module is used to predict the concentration change trend of gas components and the performance attenuation of adsorption materials based on the multi-dimensional environmental data set by using time series analysis and machine learning algorithms, and to identify potential abnormal conditions by using an anomaly detection algorithm to obtain a comprehensive prediction result; A generation module, for intelligently adjusting the operating parameters of the adsorption tank according to the comprehensive prediction results by using a multi-objective optimization algorithm and fuzzy logic control theory to generate optimized operating parameter settings; The monitoring module is used to apply the optimized operating parameter settings to the actual operation of the adsorption tank, monitor its implementation effect, and feed back the actual operation effect to the prediction model to generate a more accurate prediction model.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an intelligent control and data analysis method for gas purification in an adsorption tank as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, an intelligent control and data analysis method for gas purification in an adsorption tank as claimed in any one of claims 1 to 7 is implemented.

Images