A Prediction Method for Process Parameters of Cement Rotary Kiln Based on the Collaboration of Mechanism and Data

Abstract

The present invention relates to the field of prediction of process parameters of a cement rotary kiln, and specifically discloses a method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data. According to the mechanism of firing cement clinker, the present invention selects the parameters related to production from the production data of the cement rotary kiln and records them as the process parameters of the cement rotary kiln; collects the sampling data of each time of the process parameters, and performs preprocessing such as missing value filling, outlier removal, mean filtering, time matching, and normalization to obtain a data set of the process parameters of the cement rotary kiln after preprocessing; selects the target process parameter, evaluates the correlation degree between other process parameters and the target process parameter, screens the key variables of the target process parameter, and combines with the LSTM model to train the prediction model of the target process parameter; constructs a prediction model by specifically selecting the variables with a large correlation degree with the target process parameter, so that the prediction model can accurately capture the change law of the target process parameter and improve the prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the field of prediction of process parameters of a cement rotary kiln, and relates to a method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data. Background Art

[0002] As a basic material in the construction industry, the stable and high-quality production of cement is crucial for ensuring the smooth progress of infrastructure construction. The cement rotary kiln is the core equipment in the cement production process, and its operating state directly affects the quality and output of cement clinker.

[0003] In the actual production process, the inside of a cement rotary kiln is an extremely complex physical and chemical reaction system, involving multiple processes such as fuel combustion, material transportation, heat exchange, and chemical reactions. These processes are intertwined, making the process parameters extremely vulnerable to the influence of various factors and prone to fluctuations. If the process parameters can be accurately predicted in advance, production personnel can adjust production operations in a timely manner according to the prediction results, preventing problems such as product quality decline and equipment failures caused by abnormal parameters, thereby effectively improving production efficiency and economic benefits.

[0004] Currently, there are obvious defects in the research and application of predicting the process parameters of a cement rotary kiln.

[0005] For example, the existing Chinese patent with the publication number CN117216498A discloses a prediction system and a prediction method for the operating conditions of a cement rotary kiln. The prediction method includes obtaining characteristic data representing the operating conditions of the cement rotary kiln, and the characteristic data includes training data and test data; constructing an operating condition prediction model, and the operating condition prediction model includes an LSTM model. Among them, the LSTM model includes a forgetting gate, an input gate, and an output gate, and at least one of the forgetting gate, the input gate, and the output gate exposes the cell information of the previous moment to the cell information of the current moment during operation; training the LSTM model with the training data to obtain a trained operating condition prediction model. Using the trained operating condition prediction model to predict the test data and output the prediction result. In this way, exposing the cell information of the previous moment to the cell information of the current moment improves the accuracy of the prediction result output by the operating condition prediction model.

[0006] For example, the existing Chinese patent with the publication number CN115392125A discloses a temperature prediction method for a cement rotary kiln, including the steps of: S1. Obtaining multiple groups of historical process parameter data at different time periods in the cement rotary kiln, the true value of the kiln head temperature, and the preset process parameter data when prediction is required; S2. jointly constructing an initial kiln head temperature prediction model through a part of the historical process parameter data, an integrated residual neural network, and a bidirectional new gated recurrent unit network; S3. inputting the remaining historical process parameter data and the true value of the kiln head temperature into the initial kiln head temperature prediction model for training to obtain a mature kiln head temperature prediction model; S4. inputting the preset process parameter data into the mature kiln head temperature prediction model for prediction to obtain the final predicted value of the kiln head temperature. This invention maximally utilizes the process parameter data, improves the prediction accuracy, reduces the phenomenon of prediction lag in the recurrent neural network, and meets the temperature prediction requirements of small-sample cement kilns.

[0007] When predicting the operating conditions or process parameters of the above-mentioned cement rotary kiln, fixed variables or all process parameters in the production of the cement rotary kiln are selected as input variables, and the correlation between various variables and process parameters in the production of the cement rotary kiln is not systematically analyzed based on the mechanism of firing cement clinker. As a result, variables with a large correlation with the process parameters cannot be targeted to construct a prediction model, leading to the prediction model being unable to accurately capture the change law of the process parameters, with low prediction accuracy, difficult to meet the requirements of high-precision prediction in actual production, and severely restricting the intelligent and refined development of the cement production process. Summary of the Invention

[0008] In view of the above problems, the present invention proposes a method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data. The specific technical solution is as follows: A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data includes the following steps: Step 1: Select parameters related to the production state from the temperature data, pressure data, flow-related data, equipment operation data, and gas composition data in the production of the cement rotary kiln according to the mechanism of firing cement clinker, and record them as the process parameters of the cement rotary kiln.

[0009] Step 2: Monitor the data of each sampling of the process parameters in the production of the cement rotary kiln, and perform preprocessing such as missing value filling, outlier removal, mean filtering, time matching, and normalization to obtain a data set of the process parameters of the cement rotary kiln after preprocessing.

[0010] Step 3: Select the target process parameter of the cement rotary kiln, and evaluate the correlation between other process parameters and the target process parameter based on the data set of the process parameters of the cement rotary kiln after preprocessing, and screen the key variables of the target process parameter.

[0011] Step 4: Use the data set of the target process parameters and their key variables as the data set for the target process parameter prediction model, and divide it into a training set and a test set. Input the training set into the LSTM model to train the target process parameter prediction model.

[0012] Step 5: Evaluate the accuracy of the target process parameter prediction model according to the test set of the target process parameter prediction model.

[0013] Compared with the prior art, the method for predicting the process parameters of a cement rotary kiln based on the collaboration of mechanism and data of the present invention has the following beneficial effects: 1. By preprocessing the data used to predict the target process parameters of the cement rotary kiln in sequence, including missing value filling, outlier removal, mean filtering, time matching, and normalization, the data can be better applied to the prediction of the process parameters of the cement rotary kiln, improving the accuracy of the prediction.

[0014] 2. The present invention initially selects the process parameters of the cement rotary kiln based on the mechanism of firing cement clinker, further evaluates the correlation between other process parameters and the target process parameters, screens the key variables of the target process parameters, and combines with the LSTM model to train the prediction model of the target process parameters. By specifically selecting variables with a large correlation with the target process parameters to construct the prediction model, the prediction model can accurately capture the change law of the target process parameters, improving the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0016] Figure 1 It is a schematic flow chart of the method of the present invention.

[0017] Figure 2 It is a schematic flow chart of the preprocessing of the process parameters of the cement rotary kiln of the present invention.

[0018] Figure 3 It is a schematic flow chart of screening the key variables of the target process parameters of the present invention.

[0019] Figure 4 It is a schematic flow chart of training the target process parameter prediction model based on the LSTM model of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to Figure 1 As shown, a method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data provided by the present invention includes the following steps: Step 1: Select parameters related to the production state from the temperature data, pressure data, flow-related data, equipment operation data, and gas composition data produced by the cement rotary kiln according to the mechanism of firing cement clinker, and record them as the process parameters of the cement rotary kiln.

[0022] As a preferred solution, the specific analysis process of Step 1 is: S1: Select parameters related to the production state from the various parameters of the temperature data produced by the cement rotary kiln according to the mechanism of firing cement clinker, including the temperature of the smoke chamber, the temperature of the kiln head hood, the temperature of the tertiary air, the temperature of the decomposition furnace, the temperature at the inlet of the electrostatic precipitator, the outlet temperatures of C5A and C5B and the temperature of the feed pipe, the temperature of the kiln shell surface, and the temperature of the clinker at the outlet of the cooler.

[0023] It should be noted that the mechanism of firing cement clinker is to mix raw materials such as limestone, clay, iron ore, and coal in a certain proportion, grind them finely, and undergo a series of complex physical and chemical changes at high temperatures to finally form cement clinker mainly composed of calcium silicate. Its main processes include: raw material crushing and pre-homogenization, raw meal grinding, raw meal homogenization, preheating and decomposition, clinker burning, clinker cooling, and waste gas treatment.

[0024] It should be noted that the temperature of the tertiary air refers to the temperature of the air that enters the tail of the rotary kiln from the grate cooler and then enters the decomposition furnace during the production process of the cement rotary kiln.

[0025] It should be noted that C5A and C5B represent two cyclones in the fifth-stage preheater. The outlet temperatures of C5A and C5B refer to the measured values of the gas temperatures at the outlets of the C5A cyclone and the C5B cyclone; the feed pipe temperatures of C5A and C5B refer to the temperatures of the materials in the feed pipes of the C5A cyclone and the C5B cyclone.

[0026] S2: Select parameters related to the production state from the various parameters of the pressure data produced by the cement rotary kiln, including the pressure of the smoke chamber, the pressure of the tertiary air, the negative pressure at the kiln head, the pressure at the inlet of the electrostatic precipitator, the pressure at the inlet of the high-temperature fan, the pressure at the outlet of the high-temperature fan, the negative pressure of coal injection at the kiln head and the kiln tail, the pressure at the kiln tail seal, the inlet and outlet pressures of C5A and C5B.

[0027] It should be noted that the tertiary air pressure refers to the air pressure when the hot air from the grate cooler is introduced into the precalciner through the tertiary air duct.

[0028] S3: Select the parameters related to the production status from the various parameters of the production flow-related data of the cement rotary kiln, including the raw meal feeding amount into the kiln, the coal feeding amounts at the kiln tail and the kiln head, the frequency and current of the kiln head exhaust fan, the current of the high-temperature fan, the cooling air flow rate, and the pulverized coal conveying gas flow rate.

[0029] S4: Select the parameters related to the production status from the various parameters of the production equipment operation data of the cement rotary kiln, including the main kiln current, the speed of the main kiln motor, the first-stage grate pressure of the grate cooler, the coal injection loads at the kiln tail and the kiln head, the temperature of the trunnion bearings of the rotary kiln, and the oil temperature and vibration value of the speed reducer.

[0030] S5: Select the parameters related to the production status from the various parameters of the production gas component data of the cement rotary kiln, including the nitrogen oxide concentration in the kiln, the oxygen content in the kiln, the CO concentration in the smoke chamber, the SO2 content in the kiln tail exhaust gas, and the content of reducing gases in the kiln.

[0031] S6: Record the parameters related to the production status of the cement rotary kiln in S1 - S5 as the process parameters of the cement rotary kiln, and summarize to obtain the set of process parameters of the cement rotary kiln.

[0032] It should be noted that when selecting the parameters related to the production status from the various parameters of the temperature data, pressure data, flow-related data, equipment operation data, and gas component data of the cement rotary kiln production, in addition to relying on the mechanism of firing cement clinker, it is also possible to rely on equipment characteristics, production experience, data analysis, etc.

[0033] Step 2: Monitor the data of each sampling of the process parameters during the production of the cement rotary kiln, and perform preprocessing such as missing value filling, outlier removal, mean filtering, time matching, and normalization to obtain the data set of the process parameters of the cement rotary kiln after preprocessing.

[0034] As a preferred solution, refer to Figure 2 As shown, the specific analysis process of Step 2 is: F1: Data acquisition: Set the time period for monitoring the production of the cement rotary kiln, and set the sampling time interval and the number of samplings.

[0035] Obtain the data of each sampling of the various process parameters of the cement rotary kiln during the production of the cement rotary kiln.

[0036] F2: Missing value filling: Judge whether there are missing data in each sampling of the various process parameters of the cement rotary kiln. If a certain process parameter has missing data in a certain sampling, use the mean value of the corresponding data of the adjacent previous sampling and the adjacent next sampling of this process parameter in this sampling as the data of this process parameter in this sampling and fill it.

[0037] It should be noted that there are various methods for data filling, and prediction filling can also be carried out through median filling, mode filling, filling based on similar samples, using machine learning algorithms, etc. According to the characteristics of the data and specific requirements, reasonably estimating and filling the missing values can reduce the impact of missing data on subsequent data analysis and modeling.

[0038] F3: Outlier removal: Remove the outliers in the sampling data of each process parameter of the cement rotary kiln according to the Pauta criterion.

[0039] F4: Mean filtering: Taking the sampling data of each process parameter of the cement rotary kiln as the center respectively, setting the neighborhood range, recording the average value of the data points within the neighborhood range as the neighborhood mean corresponding to the center point data, and replacing the center point data with the neighborhood mean, so as to obtain and replace the neighborhood mean corresponding to the sampling data of each process parameter of the cement rotary kiln.

[0040] It should be noted that for one-dimensional data, several data points before and after the current data point can be selected as the neighborhood.

[0041] F5: Time matching: Extract the residence time of the same batch of materials in each stage of firing cement clinker during the production of the cement rotary kiln stored in the database, combine it with the sampling time points corresponding to the sampling data of each process parameter of the cement rotary kiln, judge whether the sampling data of different process parameters of the cement rotary kiln are the data collected during the production of the same batch of materials and classify them, and then perform time alignment on the sampling data of each process parameter of the cement rotary kiln.

[0042] In a specific embodiment, the residence time of the same batch of materials in each stage of firing cement clinker during the production of the cement rotary kiln is shown in Table 1.

[0043] Table 1. Residence time of materials in each stage

[0044]

[0045] It should be noted that cement production is a continuous process. The residence time of the same batch of materials in different equipment is different, and various process parameters and laboratory test parameters will be generated at different stages. Therefore, time matching of each process parameter of the cement rotary kiln is required to be applied to the prediction of the process parameters of the cement rotary kiln. The laboratory test parameters that are not online detected also need to be time-matched with each process parameter.

[0046] It should be noted that the time points at which the process parameters of the cement rotary kiln are generated are different. In order to accurately predict the process parameters of the cement rotary kiln subsequently, it is necessary to reasonably correspond the relevant process parameters in time, that is, it is necessary to integrate all the relevant process parameters generated by the same batch of materials at different times and different positions in the order of the actual residence time of the materials in each equipment, so that they have an accurate corresponding relationship on the time axis, so that they can be used in subsequent analysis and prediction models.

[0047] F6: Normalization: Normalize the data of each sampling of the process parameters of the cement rotary kiln according to the normalization formula.

[0048] F7: Construct a data set of the process parameters of the preprocessed cement rotary kiln according to the data of each sampling of the process parameters of the preprocessed cement rotary kiln.

[0049] As an optimal solution, the specific analysis process of step F3 is as follows: F31: Denote the data of each sampling of the process parameters of the cement rotary kiln as , represents the number of the th process parameter, , represents the number of the th sampling, .

[0050] F32: Calculate the mean value of the process parameters of the cement rotary kiln by the calculation formula , where represents the number of samplings of the process parameter.

[0051] It should be noted that the number of samplings of the process parameter is the number of data of the process parameter.

[0052] F33: Calculate the deviation between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of its process parameters by the calculation formula .

[0053] F34: Calculate the standard deviation between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of its process parameters by the calculation formula .

[0054] F35: Analyze the anomaly index of the data of each sampling of the process parameters of the cement rotary kiln by the calculation formula . If the anomaly index of the data of a certain sampling of a certain process parameter of the cement rotary kiln is 1, it is determined that the data of this sampling of this process parameter is an outlier and is excluded.

[0055] As a preferred solution, the specific analysis process of step F6 is as follows: F61: By using the calculation formula obtain the variance between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of the process parameters .

[0056] F62: Substitute the mean value of the process parameters of the cement rotary kiln and the variance between the data of each sampling of the process parameters and the mean value of the process parameters into the calculation formula to obtain the data of each sampling of the process parameters of the cement rotary kiln after normalization.

[0057] In this embodiment, the present invention preprocesses the data for predicting the target process parameters of the cement rotary kiln by filling missing values, removing outliers, mean filtering, time matching, and normalization in sequence, so that the data can be better applied to the prediction of the process parameters of the cement rotary kiln and the prediction accuracy is improved.

[0058] Step three: Select the target process parameters of the cement rotary kiln, and evaluate the correlation degree between other process parameters and the target process parameters based on the data set of the process parameters of the cement rotary kiln after preprocessing, and screen the key variables of the target process parameters.

[0059] As a preferred solution, as shown in Figure 3 the specific analysis process of step three is as follows: D1: Determine the analysis sequence: Select the process parameter to be predicted from the process parameters of the cement rotary kiln, denote it as the target process parameter of the cement rotary kiln, and denote the other process parameters except the target process parameter as other process parameters.

[0060] It should be noted that the target process parameters of the cement rotary kiln are artificially selected according to actual application requirements. The target process parameters of the cement rotary kiln can be multiple, and the prediction principle is exactly the same. In a specific embodiment, the target process parameter is the temperature of the kiln head hood; in another specific embodiment, the target process parameters include the current of the kiln main body, the temperature of the kiln head hood, the temperature of the kiln tail flue gas chamber, the temperature of the tertiary air, the negative pressure of the kiln head, the negative pressure of the kiln tail, the carbon monoxide concentration, the oxygen concentration, and the nitrogen oxide concentration.

[0061] Based on the data set of the process parameters of the cement rotary kiln after preprocessing, screen the data set of the target process parameters and construct a mother sequence, obtain the data sets of other process parameters and construct sub-sequences corresponding to other process parameters, and denote them as each sub-sequence.

[0062] It should be noted that in the present invention, the grey relational analysis method is used to obtain the key variables of the target process parameters. Grey relational analysis is a multi-factor statistical analysis method used to determine the degree of association between factors so as to select key variables.

[0063] It should be noted that the mother sequence is also called the reference sequence, and the subsequence is also called the comparison sequence. The reference sequence is a variable sequence that reflects the behavioral characteristics of the system and is the target variable of interest to the researcher. The comparison sequence is a variable sequence of various influencing factors that are related to the reference sequence.

[0064] D2: Data preprocessing: Calculate the column means and normalize the data for the mother sequence and each subsequence.

[0065] It should be noted that the preprocessing of calculating the column means and normalizing the data for the mother sequence and each subsequence is to eliminate the influence of dimensions and orders of magnitude.

[0066] It should be noted that calculating the column means and normalizing the data in data preprocessing are existing and relatively mature methods and means, which will not be elaborated here.

[0067] D3: Calculate the correlation coefficients: Calculate the correlation coefficients of each subsequence and the mother sequence at each sampling according to the correlation coefficient formula , represents the number of the th subsequence, and also represents the number of other th process parameters, .

[0068] D4: Calculate the compensation amount of the correlation degree: Obtain the compensation amount of the correlation degree of each subsequence and the mother sequence through regression analysis method, and denote it as .

[0069] D5: Calculate the correlation degree: Analyze the correlation degree of each subsequence and the mother sequence through the calculation formula , where represents the correlation degree of the th subsequence and the mother sequence.

[0070] It should be noted that the correlation coefficient reflects the degree of correlation between the subsequence and the mother sequence at each sampling. However, in order to obtain a comprehensive correlation degree index, it is necessary to perform a weighted average on the correlation coefficients.

[0071] It should be noted that the compensation amount of the correlation degree obtained through regression analysis in the above correlation degree calculation formula is used to correct the correlation degree error caused by system deviation or non - linear relationship. For example, when there is a lag effect or non - linear dependence between the mother sequence and the subsequence, the regression residual can quantify such deviation; in the formula, the comprehensive average correlation coefficient and the compensation amount are used to evaluate the final correlation degree, comprehensively measuring the overall relevance between the subsequence and the mother sequence.

[0072] It should be noted that the above correlation calculation formula uses the mean value calculation, which can eliminate random interference and extract the overall correlation trend; using the regression analysis compensation amount can capture the systematic deviation ignored by the traditional correlation calculation; using the comprehensive formula of the mean value and the compensation amount can take into account both the local correlation strength and the global system deviation, and improve the accuracy of the correlation.

[0073] It should be noted that for a feasible simulation process, assume that the mother sequence of the target process parameter is , the corresponding subsequence of the first other process parameter is , and the corresponding subsequence of the second other process parameter is . The preset resolution coefficient is set manually at the initial stage of system operation. In this embodiment ; The correlation coefficient between and is: , and the mean value of the correlation coefficient is 0.96; The correlation coefficient between and is: , and the mean value of the correlation coefficient is 0.67; the compensation amount of the correlation degree between the corresponding subsequence of the first other process parameter and the mother sequence is

[0074] Table 2. Correlation calculation result table

[0075]

[0076] Through the above simulation calculation, the correlation degree between the subsequence and the mother sequence is greater than the correlation degree between the subsequence and the mother sequence, indicating that the subsequence has a stronger correlation with the mother sequence.

[0077] D6: Correlation degree ranking: Sort according to the correlation degree of each subsequence with the mother sequence from high to low to obtain the correlation degree ranking of other process parameters with the target process parameter.

[0078] It should be noted that the greater the correlation degree, the stronger the correlation between the subsequence and the mother sequence, and the more likely it is to be a key variable.

[0079] D7: Resolution coefficient verification: If the difference between the correlation degree of a certain subsequence with the mother sequence and the correlation degree of another subsequence with the mother sequence within the set ranking of the correlation degree ranking is less than the preset difference threshold, the resolution coefficient needs to be adjusted, and the alternative values of the resolution coefficient are extracted from the database for iteration.

[0080] In a specific embodiment, the set ranking is within the top ten.

[0081] In a specific embodiment, the alternative values of the discrimination coefficient are , .

[0082] It should be noted that the selection of the discrimination coefficient in the grey relational analysis is an important task before the analysis. Selecting a suitable discrimination coefficient is beneficial to better distinguish the correlation between the target process parameter and other process parameters. Therefore, in order to better select other process parameters that have a strong correlation with the target process parameter, and then modify the discrimination coefficient, so that the grey relational analysis can select process parameters with obvious mutual correlation relationships to better promote the prediction of the target process parameter.

[0083] In a specific embodiment, the target process parameter is the temperature of the kiln head hood. Based on , , The correlation degrees of some other process parameters with the temperature of the kiln head hood are calculated respectively for three discrimination coefficients. For details, refer to Tables 3, 4, and 5.

[0084] Table 3. Correlation degree results when the discrimination coefficient ρ = 0.5

[0085]

[0086] In Table 3, the top three in terms of the correlation degree with the temperature of the kiln head hood are the tertiary air temperature, the temperature of the smoke chamber, and the kiln head discharge frequency in sequence. However, the correlation degree values are all very close and there is no distinguishing effect.

[0087] Table 4. Correlation degree results when the discrimination coefficient ρ = 0.1

[0088]

[0089] Table 5. Correlation degree results when the discrimination coefficient ρ = 0.05

[0090]

[0091] Based on Tables 3, 4, and 5 comprehensively, when the discrimination coefficient is 0.5, other process parameters related to the temperature of the kiln head hood cannot be effectively screened out. When it is 0.1, the results after the correlation analysis are also not significantly distinguishable. When it is 0.05, the correlation relationships between other process parameters and the temperature of the kiln head hood can be clearly analyzed. Therefore, the discrimination coefficient is adjusted to 0.05 to better achieve grey relational discrimination.

[0092] D8: Recalculation of correlation degree: According to the adjusted discrimination coefficient, execute D3 - D6 to recalculate the correlation degrees of other process parameters with the target process parameter.

[0093] D9: Selection of key variables: Other process parameters with a correlation degree greater than or equal to a preset correlation degree threshold are recorded as the key variables of the target process parameter, and the key variables of the target process parameter are counted.

[0094] In a specific embodiment, the preset correlation degree threshold is 0.85.

[0095] It should be noted that according to the correlation degree sorting result, combined with the background and research purpose of the actual problem, the key variables are determined. Generally, other process parameters with a relatively high correlation degree are considered as key variables that have a greater impact on the target process parameter. However, when actually selecting, factors such as the operability and economic significance of the variables also need to be considered.

[0096] In a specific embodiment, the target process parameter is the temperature of the kiln head hood, and the key variables of the target process parameter are the temperature of the tertiary air, the feeding amount of raw meal into the kiln, the speed of the kiln main machine, the negative pressure of coal injection at the kiln head, the coal feeding amount at the kiln head, the temperature of the decomposition furnace, the concentration of nitrogen oxides in the kiln, the current of the kiln main machine, the pressure of the tertiary air, the negative pressure of the kiln head, and the oxygen content in the kiln.

[0097] As a preferred solution, the specific analysis process of step D3 is: through the calculation formula Analyze the correlation coefficients of each subsequence and the mother sequence at each sampling, where represents the correlation coefficient of the th subsequence and the mother sequence at the th sampling, represents the data of the mother sequence at the th sampling, represents the th subsequence at the th sampling, represents the preset resolution coefficient, , represents the two - level minimum difference, represents the two - level maximum difference.

[0098] As a preferred solution, the specific analysis process of step D4 is: taking the subsequence as the independent variable and the mother sequence as the dependent variable, establish a regression model between each subsequence and the mother sequence according to each subsequence and the mother sequence, use the least - squares method to analyze the parameters in the regression model, and calculate the determination coefficient of the regression model between each subsequence and the mother sequence.

[0099] Extract the relationship comparison table between the determination coefficient of the regression model stored in the database and the correlation degree compensation amount, and match to obtain the correlation degree compensation amount between each subsequence and the mother sequence.

[0100] It should be noted that the regression model includes but is not limited to linear regression, polynomial regression, etc.

[0101] It should be noted that the value of the coefficient of determination of the regression model ranges from 0 to 1. The closer it is to 1, the better the model fits the data, that is, the independent variable can better explain the changes in the dependent variable.

[0102] It should be noted that the method for obtaining the coefficient of determination of the regression model in regression analysis is a relatively mature technology available at present, and will not be elaborated here.

[0103] Step 4: Use the data set of the target process parameter and its key variables as the data set of the target process parameter prediction model, and divide it into a training set and a test set. Input the training set into the LSTM model to train the target process parameter prediction model.

[0104] As a preferred solution, refer to Figure 4 As shown, the specific analysis process of Step 4 is as follows: Use the data set of the target process parameter and its key variables as the data set of the target process parameter prediction model, and divide the data set according to the set ratio between the training set and the test set to obtain the training set and the test set of the target process parameter prediction model.

[0105] Use the target process parameter as the output and the key variables of the target process parameter as the input, and input the training set of the target process parameter prediction model into the LSTM time series prediction model to train the target process parameter prediction model.

[0106] In a specific embodiment, the data set is divided into a training set and a test set according to a ratio of 9:1.

[0107] It should be noted that LSTM is a long short-term memory network, which is a special type of recurrent neural network and is the result of improving the traditional recursive neural network. LSTM newly introduces the concepts of "gates" and "cell states". The gate structure can control the input and output of information, and the cell state can store the state for a long time. Moreover, LSTM can control the main cell state through the "gate" structure. In LSTM, the role of the gate is a way to selectively pass information, which is composed of a Sigmoid neural network and a matrix point-by-point multiplication operation. There are forget gates, input gates, and output gates in LSTM. These three gates control the flow of information and determine which information needs to be forgotten, which information needs to be added, and which information needs to be output. The main advantage of LSTM is that it can effectively process long sequence data. By introducing the gated mechanism and cell state, it can maintain information over a long time span and can better adapt to the time dynamic characteristics of the data center.

[0108] It should be noted that the production data of a cement rotary kiln is a type of time series data, which involves the changes of multiple parameters over time, such as temperature, pressure, material composition, etc. There are complex dependencies among these parameters, and this relationship may change over time. The LSTM neural network is a model particularly suitable for processing this kind of data with time dependence.

[0109] It should be noted that during the process of training the target process parameter prediction model, reasonably setting the hyperparameters of the LSTM time series prediction model can optimize the model performance. Among them, the hyperparameters include the number of LSTM layers, activation function, number of neurons, learning rate, etc. In a specific embodiment, the hyperparameter settings of the LSTM time series prediction model are shown in Table 6.

[0110] Table 6. Hyperparameter Settings Table

[0111]

[0112] Step Five: Evaluate the accuracy of the target process parameter prediction model according to the test set of the target process parameter prediction model.

[0113] As a preferred solution, the specific analysis process of Step Five is as follows: Substitute the data of each key variable of the target process parameter in each sampling of the test set into the target process parameter prediction model to obtain the predicted values of each sampling of the target process parameter in the test set, and record the data of each sampling of the target process parameter in the test set as the true values of each sampling of the target process parameter.

[0114] Record the difference between the predicted value and the true value of each sampling of the target process parameter in the test set as the error of each prediction of the target process parameter in the test set, and calculate the average value to obtain the mean prediction error of the target process parameter. Extract the relationship comparison table between the mean prediction error stored in the database and the model accuracy, and match to obtain the accuracy of the target process parameter prediction model.

[0115] It should be noted that when the accuracy of the target process parameter prediction model is lower than the required accuracy, incorporate the test set into the training set to obtain a new training set, and retrain the target process parameter prediction model based on the new training set. Optimize the target process parameter prediction model by expanding the dataset to improve the generalization ability of the model.

[0116] It should be noted that the target process parameter prediction model obtained by selecting the input quantity through correlation analysis, that is, through dynamic input quantity selection, has higher accuracy than the target process parameter prediction model obtained by selecting the input quantity without correlation analysis, that is, by fixed input quantity selection, and can better achieve the judgment of trends. In a specific embodiment, as shown in Table 7, the prediction effects of the dynamic input quantity training prediction model and the fixed input quantity training prediction model of some process parameters of the cement rotary kiln are compared from three indicators of the fitting metric R2, root mean square error, and mean absolute error of the regression model.

[0117] Table 7. Comparison table of the effects of some time series models

[0118]

[0119] In this embodiment, the present invention initially selects the process parameters of the cement rotary kiln based on the mechanism of firing cement clinker, further evaluates the correlation between other process parameters and the target process parameters, screens the key variables of the target process parameters, combines with the LSTM model, trains the prediction model of the target process parameters, and constructs the prediction model by specifically selecting the variables with a large correlation with the target process parameters, so that the prediction model can accurately capture the change law of the target process parameters and improve the prediction accuracy.

[0120] The above content is only an example and explanation of the concept of the present invention. Those skilled in the art of the present technology can make various modifications or supplements to the described specific embodiments or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all belong to the protection scope of the present invention.

Claims

1. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data, characterized in that, It includes the following steps: Step 1: Select the parameters related to the production status from the temperature data, pressure data, flow-related data, equipment operation data, and gas composition data produced by the cement rotary kiln according to the mechanism of firing cement clinker, and record them as the process parameters of the cement rotary kiln; Step 2: Monitor the data of each sampling of the process parameters during the production of the cement rotary kiln, and perform preprocessing such as missing value filling, outlier removal, mean filtering, time matching, and normalization to obtain the data set of the process parameters of the cement rotary kiln after preprocessing; Step 3: Select the target process parameters of the cement rotary kiln, and evaluate the correlation degree between other process parameters and the target process parameters according to the data set of the process parameters of the cement rotary kiln after preprocessing, and screen the key variables of the target process parameters, specifically as follows: D1: Determine the analysis sequence: Select the process parameter to be predicted from the various process parameters of the cement rotary kiln, record it as the target process parameter of the cement rotary kiln, and record the other process parameters except the target process parameter as other process parameters; According to the data set of the process parameters of the cement rotary kiln after preprocessing, screen the data set of the target process parameter and construct the mother sequence, obtain the data sets of other process parameters and construct the sub-sequences corresponding to other process parameters, and record them as each sub-sequence; D2: Data preprocessing: Calculate the mean value and normalize the data of the mother sequence and each sub-sequence; D3: Calculate the correlation coefficient: Calculate the correlation coefficient between each subsequence and the mother sequence for each sampling according to the correlation coefficient formula. , represents the number of the th sampling, , represents the number of samplings of the process parameter, represents the number of the th subsequence, and also represents the number of the th other process parameter, ; D4: Calculate the compensation amount of the correlation degree: Establish a regression model between each sub-sequence and the mother sequence with the sub-sequence as the independent variable and the mother sequence as the dependent variable, analyze the parameters in the regression model using the least squares method, and calculate the determination coefficient of the regression model between each sub-sequence and the mother sequence; Extract the relationship comparison table between the determination coefficient of the regression model stored in the database and the correlation degree compensation amount, and match to obtain the correlation degree compensation amount of each subsequence and the mother sequence ; D5: Calculate the correlation degree: Through the calculation formula Analyze the correlation degree between each subsequence and the mother sequence ; D6: Relevance sorting: Sort in the order from high to low according to to obtain rankings; D7: Discrimination coefficient verification: If the difference between the correlation degree between a certain sub-sequence and the mother sequence and the correlation degree between another sub-sequence and the mother sequence within the set ranking of the correlation degree ranking is less than the preset difference threshold, the discrimination coefficient needs to be adjusted, extract the alternative values of the discrimination coefficient from the database, and perform iteration; D8: Relevance recomputation: According to the adjusted discrimination coefficient, execute D3 - D6 for recomputation ; D9: Key variable selection: Record the other process parameters with the correlation degree greater than or equal to the preset correlation degree threshold as the key variables of the target process parameter, and count the key variables of the target process parameter; Step 4: Use the data set of the target process parameter and its key variables as the data set of the target process parameter prediction model, and divide it into a training set and a test set, input the training set into the LSTM model, and train the target process parameter prediction model; Step 5: Evaluate the accuracy of the target process parameter prediction model according to the test set of the target process parameter prediction model.

2. The method for predicting process parameters of a cement rotary kiln based on mechanism and data collaboration according to claim 1, wherein: The specific analysis process of the said Step 1 is: S1: Select the parameters related to the production status from the various parameters of the production temperature data of the cement rotary kiln according to the mechanism of firing cement clinker, including the temperature of the smoke chamber, the temperature of the kiln head hood, the temperature of the tertiary air, the temperature of the decomposition furnace, the temperature at the inlet of the electrostatic precipitator, the outlet temperatures of C5A and C5B and the temperature of the feeding pipe, the temperature of the kiln shell surface, and the temperature of the clinker at the outlet of the cooler; S2: Select the parameters related to the production status from the parameters of the cement rotary kiln production pressure data, including the pressure in the smoke chamber, the tertiary air pressure, the negative pressure at the kiln head, the pressure at the inlet of the electrostatic precipitator, the pressure at the inlet of the high-temperature fan, the pressure at the outlet of the high-temperature fan, the negative pressure of coal injection at the kiln head and the kiln tail, the pressure at the kiln tail seal, the inlet and outlet pressures of C5A and C5B; S3: Select the parameters related to the production status from the parameters of the cement rotary kiln production flow rate data, including the raw meal feeding rate into the kiln, the coal feeding rates at the kiln tail and the kiln head, the frequency and current of the kiln head exhaust fan, the current of the high-temperature fan, the cooling air flow rate, and the pulverized coal conveying gas flow rate; S4: Select the parameters related to the production status from the parameters of the cement rotary kiln production equipment operation data, including the main kiln current, the speed of the main kiln motor, the grate cooler first-stage grate pressure, the coal injection loads at the kiln tail and the kiln head, the temperature of the supporting roller bearings of the rotary kiln, and the oil temperature and vibration value of the reducer; S5: Select the parameters related to the production status from the parameters of the cement rotary kiln production gas composition data, including the concentration of nitrogen oxides in the kiln, the oxygen content in the kiln, the CO concentration in the smoke chamber, the SO2 content in the exhaust gas at the kiln tail, and the content of reducing gases in the kiln; S6: Denote the parameters related to the production status of the cement rotary kiln in S1 - S5 as the process parameters of the cement rotary kiln, and summarize to obtain the process parameter set of the cement rotary kiln.

3. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data according to claim 1, characterized in that: The specific analysis process of the second step is as follows: F1: Data acquisition: Set the time period for monitoring the production of the cement rotary kiln, and set the sampling time interval and the number of sampling times; Obtain the data of each sampling of each process parameter of the cement rotary kiln during the production of the cement rotary kiln; F2: Missing value filling: Judge whether there are missing data in each sampling of each process parameter of the cement rotary kiln. If there is missing data in a certain sampling of a certain process parameter, use the average value of the corresponding data in the adjacent previous sampling and the adjacent next sampling of this process parameter in this sampling as the data of this process parameter in this sampling and fill it; F3: Outlier rejection: Reject the outliers in the data of each sampling of each process parameter of the cement rotary kiln according to the Pauta criterion; F4: Mean filtering: Respectively take the data of each sampling of each process parameter of the cement rotary kiln as the center, set the neighborhood range, record the average value of the data points within the neighborhood range as the neighborhood mean corresponding to the center point data, and replace the center point data with the neighborhood mean, so as to obtain and replace the neighborhood mean corresponding to the data of each sampling of each process parameter of the cement rotary kiln; F5: Time matching: Extract the residence time of the same batch of materials in each stage of firing cement clinker during the production of the cement rotary kiln stored in the database, and combine the sampling time points corresponding to the data of each sampling of each process parameter of the cement rotary kiln to judge whether the sampling data of different process parameters of the cement rotary kiln are the data collected during the production of the same batch of materials and classify them, so as to align the time of the data of each sampling of each process parameter of the cement rotary kiln; F6: Normalization: Perform normalization processing on the data of each sampling of each process parameter of the cement rotary kiln according to the normalization formula; F7: Construct a data set of each process parameter of the pre-processed cement rotary kiln based on the data of each sampling of each process parameter of the pre-processed cement rotary kiln.

4. The method for predicting process parameters of a cement rotary kiln based on mechanism and data collaboration according to claim 3, characterized in that: The specific analysis process of step F3 is as follows: F31: Denote the data of each sampling of the process parameters of the cement rotary kiln as , indicating the number of the th process parameter, , indicating the number of the th sampling, ; F32: Calculate the mean of each process parameter of the cement rotary kiln through the calculation formula to obtain the mean value of each process parameter of the cement rotary kiln , where represents the number of sampling times of the process parameter; F33: Calculate through the formula to obtain the deviation between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of the process parameters ; F34: By calculation formula Obtain the standard deviation between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of its process parameters ; F35: Through the calculation formula Analyze the anomaly index of each sampling data of the process parameters of the cement rotary kiln , if the anomaly index of a certain sampling data of a certain process parameter of the cement rotary kiln is 1, it is determined that the sampling data of this process parameter is an outlier and is excluded.

5. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data according to claim 4, characterized in that: The specific analysis process of step F6 is as follows: F61: Calculate the variance between the data of each sampling of the process parameters of the cement rotary kiln and the mean value of the process parameters through the calculation formula and ; F62: The mean values of the process parameters of the cement rotary kiln and the variances between the sampled data of each process parameter and its mean value of the process parameter are substituted into the calculation formula to obtain the sampled data of each process parameter of the cement rotary kiln after homogenization.

6. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data according to claim 1, characterized in that: The specific analysis process of step D3 is as follows: By using the calculation formula Analyze the correlation coefficients between each subsequence and the mother sequence at each sampling, where represents the th subsequence and the correlation coefficient between the mother sequence at the th sampling, represents the data of the mother sequence at the th sampling, represents the th subsequence at the th sampling, represents the preset resolution coefficient, , represents the two-level minimum difference, represents the two-level maximum difference.

7. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data according to claim 1, characterized in that: The specific analysis process of step 4 is as follows: Use the data set of the target process parameter and its key variables as the data set of the target process parameter prediction model, divide the data set according to the set ratio between the training set and the test set, and obtain the training set and test set of the target process parameter prediction model; Use the target process parameter as the output and the key variables of the target process parameter as the input, and input the training set of the target process parameter prediction model into the LSTM time series prediction model to train the target process parameter prediction model.

8. A method for predicting process parameters of a cement rotary kiln based on the collaboration of mechanism and data according to claim 1, characterized in that: The specific analysis process of step 5 is as follows: Substitute the data of each sampling of the key variables of the target process parameter in the test set into the target process parameter prediction model to obtain the predicted values of each sampling of the target process parameter in the test set, and record the data of each sampling of the target process parameter in the test set as the true values of each sampling of the target process parameter; Record the difference between the predicted value and the true value of each sampling of the target process parameter in the test set as the error of each prediction of the target process parameter in the test set, calculate the average value, obtain the mean prediction error of the target process parameter, extract the relationship comparison table between the mean prediction error stored in the database and the model accuracy, and match to obtain the accuracy of the target process parameter prediction model.

Images