A temperature control method for smelting flue gas acid making

Abstract

The application provides a kind of metallurgical flue gas acid making temperature control method, it is related to metallurgical flue gas acid making technical field.Process data set in smelter flue gas acid making process is collected first;The collected process data is preprocessed and denoised, and the dimensional influence between each process parameter is eliminated by normalization;Characteristic analysis is carried out on the pretreated process data, Spearman correlation coefficient is calculated, and ICA method is used for dimension reduction, and main influence parameters are selected;The selected main parameter sequence is analyzed, the main parameter sequence trend is drawn, and the SWAB method is used for division and corresponds to the actual working condition respectively;VAR model under different working conditions is constructed, control optimization rules of SO2 fan guide vane opening under different working conditions in metallurgical flue gas acid making process are formed according to model prediction results, and finally the advanced control of process temperature is realized.The application can provide a solution strategy for the temperature optimization control method of metallurgical flue gas acid making production process, so as to reduce energy consumption.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mineral and metallurgical off-gas acid making, and particularly relates to a mineral and metallurgical off-gas acid making temperature control method. BACKGROUND

[0002] Sulfuric acid is one of important basic chemical raw materials and plays an important role in industrial production. In recent years, the mineral and metallurgical off-gas acid making industry in China develops rapidly and forms a strong production capacity. Lead sulfide, zinc sulfide and other metal sulfides are main raw materials for non-ferrous metal smelting. In the smelting process, industrial waste gas containing high concentration of SO2 is generated. Direct emission of the industrial waste gas will cause great pollution to the atmosphere and is a waste of resources. When the SO2 content in the industrial waste gas reaches 3.5%, the smelting off-gas acid making method can be used to fully utilize the part of sulfur resources. This acid making method not only reduces the waste of resources and the pollution to the atmosphere, but also increases the income of the metal smelting plant.

[0003] At present, the smelting off-gas acid making mainly adopts a two-conversion and two-absorption process. The conversion section is an important link in the process flow of the off-gas acid making. In the off-gas acid making process, stable temperature control of each layer of the converter is crucial to improve the acid making efficiency. The opening degree of the SO2 fan guide vane, the oxygen-sulfur ratio and other factors are closely related to the inlet temperature of each layer of the converter. However, the current temperature control mainly relies on the experience of the operator, and the opening degree of the SO2 fan is manually adjusted or conventionally controlled according to the actual working condition on site. When the input off-gas state changes, the adjustment is often not timely due to the control lag, and the SO2 fan is a high energy consumption equipment. Therefore, a large amount of electric energy is wasted.

[0004] In view of the current difficulties in temperature control in the field of mineral and metallurgical off-gas acid making, how to provide an effective industrial big data analysis and advanced control method to reduce energy consumption based on the massive process data sequences generated in the off-gas acid making process is a problem to be solved by the person skilled in the art.

[0005] Therefore, a new technical scheme is needed. SUMMARY

[0006] Therefore, the present application provides a mineral and metallurgical off-gas acid making temperature control method. The existing process data sequences of the off-gas acid making process of a smelting plant are taken as input, data preprocessing, denoising and normalization are performed, data analysis is carried out, main auxiliary variables are screened through feature correlation analysis, a multivariate vector autoregressive (VAR) model is constructed, and an advanced control strategy of the opening degree of the SO2 fan guide vane on the temperature is realized, so that the energy saving and efficiency improving goals are achieved.

[0007] The present application provides the following technical scheme: a mineral and metallurgical off-gas acid making temperature control method, comprising

[0008] S1: Collecting process data in a smelter gas acid-making process, the data at least including the inlet temperature of each layer of the converter, the inlet pressure of each layer of the converter, the SO2 fan inlet guide vane valve opening, the opening of the flue gas supplemental air regulating valve and the SO2 fan outlet guide vane opening, etc.

[0009] S2: Denoising and normalizing the collected process data, so that the data is quantized in [0, 1], and the dimensional influence between the process parameters is eliminated.

[0010] S3: Feature analysis is performed on the processed process data set, mainly including correlation analysis to obtain the correlation degree between the data sequences, and then independent component analysis is performed to reduce the dimension of the data to determine the principal components, so as to facilitate the construction of the subsequent control optimization model.

[0011] S4: For the selected principal component sequences, including the SO2 fan guide vane opening and the inlet temperature of each layer of the converter, data analysis test is performed, the data correlation trend is drawn by using tools or methods, the key features of the sequences are found, and then the sliding window bottom-up method is used to segment the data set, and the segmentation results correspond to different actual working conditions.

[0012] S5: A multivariate vector autoregressive model is constructed in combination with the selected main variables and the segmentation results of the sliding window bottom-up method processing, and the control rule of the SO2 fan guide vane opening in the smelter gas acid-making process under different working conditions is realized according to the model prediction results.

[0013] Optionally, in S2, the collected process data set is preprocessed to exclude unrealistic data and normalized.

[0014] Optionally, in step S3, first, the correlation coefficient method is used to calculate the correlation between the process data sets, the key variables with an absolute value greater than 0.75 are selected, the selection of auxiliary variables is realized, and the Spearman correlation coefficient method is used to calculate the formula as follows.

[0015]

[0016] Wherein: d i represents the difference between the bit values of the i-th data pair; n represents the total number of observation samples, i.e. the number of collected smelter acid-making process data.

[0017] Optionally, in S3, after calculating the correlation coefficient, the correlation degree between the main manipulated variables and the controlled variables can be obtained, and then the independent component analysis method is used again to reduce the dimension, which can effectively screen out the principal components with the highest correlation degree with the manipulated variables and the controlled variables as the input and output variables of the subsequent multivariate vector autoregressive model.

[0018] Optionally, in the S4, data analysis is performed on the selected principal component sequence, a tool or method is used to draw a data correlation trend, key features of the sequence are found, and then a sliding window bottom-up method is used to segment the data set, and the segmentation results correspond to different working conditions of the smelter off-gas acid making.

[0019] Optionally, in the S5, a multivariate vector autoregressive model is constructed for each segment to realize an advanced control strategy for the temperature, and then automatic adjustment rules for the SO2 fan guide vane opening are formed according to the model prediction results.

[0020] The S5 includes the following steps:

[0021] S501: Draw a sequence correlation graph of multiple process data, calculate a correlation coefficient, and view a linear correlation degree.

[0022] S502: Perform a stationarity test on the original data of the multiple process data, and the multivariate vector autoregressive model requires that all process data are co-integrated, that is, if one of the multiple process data is not stationary, all of them need to be differentiated until they are stationary.

[0023] S503: Perform a co-integration test on the inlet temperature of each layer of the converter of the smelter off-gas acid making process and the influencing factor SO2 fan guide vane opening.

[0024] S504: Divide the process data set into a training set and a test set, and the ratio of the training set to the test set can be 7:3.

[0025] S505: Select the order P of the multivariate vector autoregressive model and train the model.

[0026] S506: The fitting result of the autocorrelation model usually needs to check the correlation of the process data sequence, and for the sequence correlation, a Durbin Watson test can be used.

[0027]

[0028] The value is between 0 and 4, and the closer to 2, the more it indicates that there is no sequence correlation; close to 0 indicates that there is positive sequence correlation; and close to 4 indicates that there is negative sequence correlation.

[0029] Step S507: Use the multivariate vector autoregressive model to predict the time series process data, check and evaluate the prediction results, and a calculation index can be used for evaluation and analysis.

[0030] Compared with the prior art, the above at least one technical solution adopted by the embodiments of the present specification can achieve at least the following beneficial effects:

[0031] 1. The present application uses the industrial big data set of the mineral smelting flue gas acid making process, deeply mines the multi-dimensional association information between the process data, especially the information between the SO2 fan guide vane opening and the converter related temperature, as the basis for forming the temperature advanced control strategy of the mineral smelting flue gas acid making production process, and produces the effect of energy saving and efficiency reduction.

[0032] 2. The present application processes the mineral smelting flue gas acid making process data set in combination with ICA, SWAB and VAR algorithm, this scheme can be popularized to the process data analysis of some non-linear, multi-dimensional and unstable industrial processes, expands the breadth of application of mathematical statistics method in the industrial field, and provides a technical solution idea. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0034] Figure 1 is the implementation step flow chart of a mineral smelting flue gas acid temperature control method in the embodiments of the present application;

[0035] Figure 2 is the comparison chart of the data sequence before and after segmentation of a certain process data sequence in step S4 in the embodiments of the present application using SWAB algorithm;

[0036] Figure 3 is the general step chart of establishing a prediction model using VAR algorithm. DETAILED DESCRIPTION

[0037] The embodiments of the present application will be described in detail below with reference to the drawings.

[0038] The embodiments of the present application will be described in detail below with reference to the drawings.

[0039] It is to be understood that the embodiments described herein are illustrative only and the scope of the application should not be deemed limited thereto based on their description in the attached claims. As will be apparent to those of ordinary skill in the art, the aspects described herein can be implemented in a wide variety of forms and that any particular structure and / or function described herein is merely illustrative. An aspect described herein can be implemented alone or in combination with any other aspect(s). Further, any aspect described herein can be implemented with respect to any other structure instead of the structure described.

[0040] It is also to be understood that the diagrams provided in the following description are merely schematic and that actual implementations can differ from those of the figures in their configuration, number, disposition, and / or proportions of the various components, and that individual components can be implemented in a different manner from that shown in the figures.

[0041] In addition, in the following description, numerous specific details are provided, such as examples of components and / or methods, to provide a thorough understanding of examples. One skilled in the relevant art will recognize, however, that the application can be practiced without one or more of the specific details, or with other methods, components, etc.

[0042] Sulfuric acid is one of the important basic chemical raw materials, plays a pivotal role in industrial production. In recent years, China's mining and smelting gas acid industry has developed rapidly, and a strong scale of production capacity has been formed. Lead sulfide, zinc sulfide and other metal sulfides are the main raw materials for non-ferrous metal smelting. In the smelting process, industrial waste gas containing high concentration of SO2 will be directly discharged, which will cause great pollution to the atmosphere, and is a waste of resources. When the SO2 content in industrial waste gas reaches 3.5%, smelting flue gas acid making method can be used to make full use of this part of sulfur resources. This acid making method not only reduces the waste of resources and pollution to the atmosphere, but also increases the income of metal smelting plants.

[0043] At present, the smelting plant flue gas acid making mainly adopts two conversion and two absorption process, and the conversion section is an important link in the process flow of flue gas acid making. In the process of flue gas acid making, stable temperature control of each layer of the converter is crucial to improve the acid making efficiency. The opening degree of SO2 fan guide vane and the oxygen-sulfur ratio are closely related to the inlet temperature of each layer of the converter. However, the current temperature control mainly relies on the experience of operators, and the SO2 fan opening is manually adjusted or controlled according to the actual working condition of the site. When the input flue gas state changes, the control lag adjustment is often not timely, and the SO2 fan is a high energy consumption equipment, so a large amount of electric energy will be wasted.

[0044] Addressing the current challenges in temperature control during sulfuric acid production from mining and metallurgical flue gas, and considering the massive amounts of process data generated during the process, how to provide an effective method based on industrial big data analysis and advanced control to reduce energy consumption is a pressing issue for those skilled in the art. Therefore, this invention proposes a data processing method for sulfuric acid production from mining and metallurgical flue gas and an advanced temperature control method based on a multivariate vector autoregression (VAR) model.

[0045] like Figure 1 As shown, the present invention provides a method for controlling the temperature of sulfuric acid production from mining and metallurgical flue gas, which mainly includes the following steps:

[0046] Step S1: Collect process data set of flue gas acid production process in a smelter. The process data set includes at least the inlet temperature of each layer of the converter, the inlet pressure of each layer of the converter, the opening of the SO2 blower inlet guide vane valve, the opening of the flue gas make-up air regulating valve, and the opening of the SO2 blower outlet guide vane valve.

[0047] Step S2: Perform data preprocessing and noise reduction on the collected process dataset, and normalize the data to make it within [0, 1], so as to eliminate the influence of dimensions between various indicators;

[0048] Step S3: Feature extraction is performed on the processed data sequence, mainly including correlation analysis to obtain the correlation between each data sequence, and then the independent component analysis (ICA) algorithm is used to reduce the dimensionality of the data sequence to determine the principal components, which facilitates the construction of the control optimization model in the future.

[0049] Step S4: Perform data analysis on several selected principal component sequences, including the opening degree of SO2 fan guide vane valve and the inlet temperature of each layer of converter. Conduct data analysis tests, use tools or methods to plot data correlation trends, discover key features of data sequences, and then use the sliding window bottom-up (SWAB) algorithm to segment the data sequence set. The segmentation results correspond to different working conditions of sulfuric acid production from mining and metallurgical flue gas.

[0050] Figure 2 This is a comparison chart showing the process data sequence before and after step S4 in this embodiment of the invention, where the data sequence is segmented using the SWAB algorithm.

[0051] like Figure 2 As shown, in step S4, data analysis is performed on several selected principal component sequences, and the data sequence trend is plotted using tools or methods to discover the key features of the data sequence. Then, the sliding window bottom-up (SWAB) algorithm is used to segment the data sequence, and the segmentation results correspond to different working conditions of sulfuric acid production from mining and metallurgical flue gas.

[0052] Step S5: Based on the selected main variables and the SWAB processing results, a multivariate vector autoregressive (VAR) prediction model is constructed in segments, and the advanced control of SO2 fan guide vane opening on temperature under different working conditions in the smelter off-gas acid-making process is realized according to the prediction results, so as to realize the transformation of the current control method, and achieve the goal of reducing the energy consumption of the fan and saving resources.

[0053] In S2, the collected process data set is preprocessed to exclude false data and implement normalization processing.

[0054] In step S3, the correlation coefficient method is first used to calculate the correlation between each process data, and the key variables with an absolute value greater than 0.75 are selected as the first choice for subsequent processing. For example, the Spearman correlation coefficient method is suitable for correlation analysis between ordered variables, and its rank correlation coefficient is calculated based on the ordering of two variables (rather than the original numerical value). Therefore, it is more suitable for data analysis in this example, and the difference between the actual numerical values has no direct impact on the calculation results.

[0055] The complete calculation formula is

[0056]

[0057] Wherein:

[0058] d i represents the difference between the ranks of the ith data pair;

[0059] n represents the total number of observation samples, i.e. the number of collected smelter off-gas acid-making process data;

[0060] The simplified formula is

[0061]

[0062] Wherein:

[0063] d i represents the difference between the ranks of the ith data pair;

[0064] n represents the total number of observation samples, i.e. the number of collected smelter off-gas acid-making process data.

[0065] In step S3, after calculating the correlation coefficient, the correlation between the main manipulated variables and the controlled variables can be obtained, and then the independent component analysis (ICA) method is used again to reduce the dimension, which can effectively screen out the principal components with the highest correlation with the manipulated variables (SO2 fan guide vane opening) and the controlled variables (converter first-stage inlet temperature) as the input and output variables of the subsequent multivariate vector autoregressive model.

[0066] The basic principle of ICA algorithm is as follows:

[0067] For the obtained sample data, where x (i) ∈R n , the ICA algorithm assumes that these data are linear mixed sequences. That is, the real data source is s (i) ∈R n , and x is obtained after passing through a mixing matrix A, and is expressed as

[0068] x=As

[0069] It analyzes the source data s through a set of sample data, and if W is the demixing matrix, then W=A -1 . The update rule of the matrix W is

[0070]

[0071] When the algorithm terminates, the estimate of W is calculated, and then it can be used to restore the independent components

[0072] s (i) =Wx (i)

[0073] Here, the SWAB (Sliding Window and Bottom-up) algorithm retains a cache area with a size of W. The cache area is initialized so that there is enough data to create 5-6 segments. The processing algorithm is applied to the sequence data of the cache area to obtain the leftmost segment. The sequence data corresponding to the reported segment is deleted from the cache area, and then more sequence data is read, and the amount of data read depends on the structure of the input data. The processing algorithm is applied again, and the process of applying the bottom-up processing to the buffer is repeated as data arrives, the leftmost segment is reported, and the next "best" sub-sequence is read. A certain time sequence data sequence segmented by the SWAB method is shown in, for example, Figure 2

[0074] The step S5 segments to build a multivariate vector autoregressive (VAR) prediction model, and then realizes the advanced control strategy of the SO2 fan guide vane opening degree on the temperature according to the model prediction result.

[0075] As a comparison, the traditional time series model, such as ARIMA, ARIMA-GARCH, etc., only analyzes the change of the data sequence itself, and the form of the model is

[0076] y t =β1·y t-1 +β2·y t-2 +…

[0077] Where y t-i is called the own lag term. ​

[0078] In addition to analyzing the impact of its own lagged terms, the VAR forecasting model also analyzes the impact of the lagged terms of other related factors on future forecast values. The model takes the form of...

[0079] y t =β1·y t-1 +α1·x t-1 +β2·y t-2 +α2·x t-2 +…

[0080] Where x t-j It refers to the lagged terms of other factors.

[0081] Therefore, the VAR model can be viewed as a comprehensive model combining the advantages of multiple linear regression (allowing the inclusion of multiple factors) and time series models (allowing the analysis of the impact of lagged terms). The main steps in constructing a multiple vector autoregressive (VAR) prediction model are as follows:

[0082] like Figure 3 As shown, the VAR algorithm is used to build a predictive model for the process dataset of a mining and metallurgical flue gas sulfuric acid production plant:

[0083] Step S501: Draw a sequence correlation diagram of the process dataset, calculate the correlation coefficient, and check the linear correlation. Some factor correlation data are shown in Table 1.

[0084] Table 1

[0085] TI_D0273_AV TI_D0274_AV TI_D0275_AV Factor M... TI_D0273_AV 1 0.872795751 0.509112171 …… TI_D0274_AV 0.872795751 1 0.490058115 …… TI_D0275_AV 0.509112171 0.490058115 1 …… Factor M… …… …… …… 1

[0086] Step S502: Perform a stationarity test on the original data of multiple factors. The VAR model requires that all factor data be cointegrated of the same order. That is, if the data of one factor is not stationary, all factors need to be differencing until they are stationary.

[0087] Step S503: Corresponding variable Y and influence factor X i To perform a cointegration test, in this embodiment, variable Y corresponds to the inlet temperature of each layer of the converter, and the influencing factor X... i This corresponds to the opening degree of the guide vane valve of the SO2 fan and its related process parameters. If a factor fails the cointegration test, it needs to be removed.

[0088] Step S504: Divide the dataset into a training set and a test set. In this embodiment, the ratio of the training set to the test set is 7:3.

[0089] Step S505: Select the order P of the VAR model and train the model;

[0090] Step S506: Use the DW test to check for serial correlation. For the fitting results of autocorrelation models, it is usually necessary to check for serial correlation. Serial correlation refers to the presence of correlation in the model's residuals. If this occurs, it indicates that the model is missing some information, and the fitting result is not optimal.

[0091] For sequence correlation, the Durbin-Watson (DW) test is usually used, as shown in the following relationship.

[0092]

[0093] This value ranges from 0 to 4. The closer it is to 2, the less serial correlation there is; close to 0 indicates positive serial correlation; and close to 4 indicates negative serial correlation.

[0094] Step S507: Use the VAR prediction model to predict time series data, verify and evaluate the prediction results, and use calculated indicators to participate in the evaluation analysis.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0096] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0097] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A temperature control method for smelter off-gas acid production, characterized by: The method comprises the following steps: S1: collecting process data in a smelter off-gas acid-making process, the data at least including inlet temperature of each layer of the converter, inlet pressure of each layer of the converter, SO2 fan inlet guide vane opening, off-gas air regulating valve opening and SO2 fan outlet guide vane opening; S2: denoising and normalizing the collected process data, so that the data is quantified in [0, 1] and the dimensional influence between the process parameters is eliminated; S3: extracting features from the processed process data set, including correlation analysis to obtain the correlation between each data sequence, and then performing independent component analysis to reduce the dimension of the data to determine the principal components, so as to facilitate the construction of a subsequent control optimization model; S4: for a plurality of selected principal component sequences, including SO2 fan guide vane opening and inlet temperature of each layer of the converter, performing data analysis and testing, using tools or methods to draw data correlation trends, finding key features of the data sequence, and then using a sliding window top-down method to segment the data set, and the segmentation results correspond to different actual working conditions respectively; S5: combining the selected main variables and the segmentation results of the sliding window top-down method to construct a multivariate vector autoregressive model, and realizing the control rule of the SO2 fan guide vane opening in the smelter off-gas acid-making process under different working conditions according to the model prediction result.

2. The temperature control method for smelter acid gas according to claim 1, characterized in that: In S2, the collected process data set is preprocessed to exclude false data and normalized.

3. The temperature control method for smelter acid gas according to claim 2, characterized in that: In S3, the correlation between each process data set is calculated by selecting the correlation coefficient method, and the key variables with an absolute value greater than 0.75 are selected to realize the selection of auxiliary variable set. The Spearman correlation coefficient method is used to calculate the formula as follows: where: d i represents the difference between the bit values of the ith data pair; n represents the total number of observation samples, i.e., the number of collected mineral and metallurgical acid-making process data.

4. The smelter gas sulphuric acid plant temperature control method according to claim 3, characterised in that: In S3, after calculating the correlation coefficient, the correlation between a plurality of manipulated variables related to the main manipulated variable and the controlled variable is obtained. Then, the independent component analysis method is used again to reduce the dimension to effectively screen out the principal components with the highest correlation degree with the manipulated variables and the controlled variables as the input and output variables of the subsequent multivariate vector autoregressive model.

5. The temperature control method for smelter acid gas according to claim 4, characterized in that: In S5, the multivariate vector autoregressive model is constructed in segments to realize the advanced control strategy of temperature, and then the automatic adjustment rule of the SO2 fan guide vane opening is formed according to the model prediction result. S5 comprises the following steps: S501: drawing a sequence correlation graph of a plurality of process data, calculating the correlation coefficient, and checking the linear correlation degree; S502: performing stationarity test on the original data of the plurality of process data. The multivariate vector autoregressive model requires that all process data are co-integrated, that is, if one of the plurality of process data is not stationary, all of them need to be differentiated until they are stationary; S503: performing co-integration test on the inlet temperature of each layer of the converter and the SO2 fan guide vane opening in the smelter off-gas acid-making process; S504: dividing the process data set into a training set and a test set, and the ratio of the training set to the test set is 7:3; S505: selecting the order P of the multivariate vector autoregressive model and training the model; S506: For the fitting results of the autocorrelation model, the correlation of the process data sequence needs to be checked. For the sequence correlation, the Durbin Watson test is used: The value is between 0 and 4, the closer to 2, the more it means that there is no sequence correlation; close to 0, it is considered that there is positive sequence correlation; and close to 4, it is considered that there is negative sequence correlation; Step S507: Use a multivariate vector autoregressive model to predict the time series process data, test and evaluate the prediction results, and use the calculation index to participate in the evaluation analysis.

Images