A Wet Flue Gas Desulfurization Parameter Optimization Method Based on Data Processing

Abstract

The present invention relates to the technical field of data processing. More specifically, the present invention relates to a wet desulfurization parameter optimization method based on data processing, including: collecting the concentration values and preprocessed pH values of limestone slurry in the desulfurization system at each moment; taking the previous moment of any moment requiring concentration interpolation as the target moment, and calculating the concentration prominence of the target moment; obtaining the pH values corresponding to all concentration values within the time period to which the target moment belongs by using the Pearson correlation coefficient; calculating the data correlation of the target moment; calculating the data authenticity of the target moment; calculating the adaptive error factor of the moment requiring concentration interpolation; and performing interpolation on the moment requiring concentration interpolation by using the HTFE model to achieve the optimization of wet desulfurization parameters based on data processing. By calculating the concentration prominence and data correlation, the present invention can more accurately evaluate the authenticity of concentration data at each moment, reduce the influence of noise data on the interpolation result, and improve the interpolation accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing. More specifically, the present invention relates to a wet desulfurization parameter optimization method based on data processing. Background Art

[0002] Due to the current serious air pollution faced by the environment, among which sulfur dioxide not only causes acid rain but also is an important component of smog, posing a threat to public health and the ecological environment. Therefore, it becomes particularly important to reduce sulfur dioxide emissions and improve air quality through effective wet desulfurization technology. In the desulfurization system, the concentration of limestone slurry directly affects the rate and efficiency of the desulfurization reaction. An appropriate concentration can ensure sufficient reaction between sulfur dioxide and the slurry, thereby improving the desulfurization efficiency. The Proportional-Integral-Derivative (PID) control algorithm can optimize the desulfurization reaction process by precisely controlling the concentration of limestone slurry, ensuring that the concentration is maintained at the set value, thereby improving the desulfurization efficiency and system stability. However, during the transmission process of the concentration data of limestone slurry, due to problems such as communication line or network failures, some data points may be missing when the data is transmitted to the data recording system, that is, the concentration values at the corresponding moments of these data are also missing. There is currently a method for interpolating the missing limestone slurry concentration data, which can effectively process the missing data and provide more reliable and accurate concentration data.

[0003] In the 9th issue of the Journal of Computer Era in 2023, an article titled "Research on Time Series Prediction Method Based on Historical Trends and Prediction Errors" was published. In the article, aiming at problems such as complex training and poor transfer prediction ability of time series prediction methods, a time series adaptive prediction method based on historical trends and prediction errors (Historical Trends Forecast Errors, HTFE) was proposed. First, determine the range of the predicted value for the next moment according to the prediction error and the current value, and then determine the final predicted value in combination with the short-term historical trend. The obtained current predicted value is substituted into the next round of loop for continuous prediction, and the data prediction is realized through continuous "prediction - correction - prediction" loops. Finally, using time series data such as finance and wind power, a comparison was made among classical time series prediction algorithms such as LSTM, SVM, ARIMA, and MA in terms of prediction accuracy, transfer prediction ability, and operation speed.

[0004] During the process of collecting limestone slurry concentration data, due to the electromagnetic interference generated by surrounding equipment affecting the quality of sensor signals, there are noise data in the collected limestone slurry concentration data, and the noise data often has values close to some real limestone slurry concentration data. However, when the HTFE model interpolates missing data, it is based on the predicted values and actual values of the historical data before the data points at the missing moment. Then, if there is noise data in the historical data, the actual values of the historical data will not be accurate enough, which will further lead to inaccurate prediction errors obtained from the predicted values and actual values of the historical data, and subsequently reduce the accuracy of the initial predicted values and final predicted values of the data point values by the model. At the same time, when the HTFE model predicts and calculates the values of all data points that need interpolation, it will use the same error factor value, resulting in a decrease in the accuracy of the HTFE model when interpolating missing data, and ultimately affecting the control of limestone slurry concentration. Summary of the Invention

[0005] To solve the problem that when the HTFE model interpolates missing data, since the data of the previous moment used as a reference may be noise data, and the subsequent same-sized error factor is used, resulting in a decrease in the accuracy of the interpolation process and ultimately affecting the control of limestone slurry concentration, the present invention proposes a wet flue gas desulfurization parameter optimization method based on data processing. The method includes the following steps:

[0006] Collect the concentration values and preprocessed pH values of the limestone slurry at each moment in the desulfurization system;

[0007] Denote the previous moment of any moment that needs concentration interpolation as the target moment. Based on the concentration value and concentration predicted value of the target moment, as well as the absolute value of the concentration numerical skewness and the concentration mean of the time period to which the target moment belongs, calculate the concentration prominence degree of the target moment; use the Pearson correlation coefficient to obtain the pH values corresponding to all concentration values within the time period to which the target moment belongs. Based on the rankings of the concentration value and the corresponding pH value of the target moment in their respective time periods, the Pearson correlation coefficient of the concentration value and the concentration value corresponding pH value of the time period to which the target moment belongs, and the Pearson correlation coefficient of the remaining concentration values and the remaining concentration values corresponding pH values when the time period to which the target moment belongs does not include the concentration value of the target moment, calculate the data correlation of the target moment; based on the concentration prominence degree and data correlation of the target moment, calculate the data authenticity of the target moment; modify the preset error factor according to the data authenticity to obtain the adaptive error factor for the moment that needs concentration interpolation; according to the adaptive error factor, use the HTFE model to perform interpolation processing on the moment that needs concentration interpolation to obtain the interpolated concentration value of the moment that needs concentration interpolation, so as to realize the optimization of wet flue gas desulfurization parameters based on data processing.

[0008] By calculating the concentration prominence and data correlation, the authenticity of the concentration data at each moment can be evaluated more accurately, thereby reducing the influence of noise data on the interpolation result and improving the accuracy of interpolation; the error factor is dynamically corrected based on data authenticity, and the adaptive mechanism can adjust the interpolation strategy according to the changes in actual data, ensuring the interpolation accuracy under different conditions and making the interpolation process more flexible and reliable; the optimized concentration control can improve the operating stability of the wet flue gas desulfurization system and reduce system failures or efficiency reduction caused by concentration fluctuations.

[0009] Further, the preprocessed pH value is the pH value after interpolation processing using the simple moving average algorithm.

[0010] Further, the concentration prediction value is the prediction value obtained through the HTFE model.

[0011] Further, the concentration prominence satisfies the following relational expression:

[0012] ; where is the concentration prominence at the th target moment, is the concentration value at the th target moment, is the concentration mean of the time period to which the th target moment belongs, is the concentration prediction value at the th target moment, is the numerical skewness of the time period to which the th target moment belongs, is a hyperparameter, is the absolute value symbol.

[0013] By comparing the deviations of the actual concentration value, concentration mean, and prediction value, the abnormal degree of the data can be effectively captured, thereby more accurately evaluating the prominence of the concentration data. By introducing numerical skewness, and the absolute value of the numerical skewness provides information about the distribution characteristics of the concentration data, thus helping to identify abnormal fluctuations and making the evaluation of the concentration prominence more comprehensive.

[0014] Further, the data correlation satisfies the following relational expression:

[0015] ; where is the data correlation at the th target moment, and are respectively the rankings of the concentration value and the pH value corresponding to the concentration value at the th target moment in their respective time periods, is the The Pearson correlation coefficient between the remaining concentration values and the corresponding pH values of the remaining concentration values in the time period to which the target time belongs when the concentration value at the target time is not included is the Pearson correlation coefficient between the concentration values and the corresponding pH values of the time period to which the th target time belongs, is a hyperparameter, is the natural exponential function, is the absolute value symbol.

[0016] By considering the rankings of the concentration values and pH values, the changing trend between times can be better reflected, providing more accurate interpolation results; through normalization processing, the influence of outliers can be effectively suppressed, enhancing the model's resistance to noise; through the two dimensions of concentration and pH value, the interpolation not only depends on a single factor, but is more comprehensive and can capture potential interrelationships.

[0017] Furthermore, the data authenticity satisfies the following relational expression:

[0018] ; where is the data authenticity of the th target time, is the data correlation of the th target time, is the concentration prominence of the th target time, is the normalization function.

[0019] By combining the data correlation and the concentration prominence, the authenticity of the data can be effectively quantified, ensuring that when calculating the authenticity, not only the relationship between the data is reflected, but also the actual change situation of the data can be reflected.

[0020] Furthermore, the adaptive error factor satisfies the following relational expression:

[0021] ; where is the adaptive error factor of the th time requiring concentration interpolation, is the data authenticity of the th target time, is the preset error factor.

[0022] Further, the realization of the optimization of wet desulfurization parameters based on data processing includes: forming a complete concentration data sequence with the concentration values at each moment and the interpolated concentration values at the moments requiring concentration interpolation, and controlling the concentration of limestone slurry based on the complete concentration data sequence by using the PID control algorithm to complete the optimization of wet desulfurization parameters based on data processing.

[0023] Through the complete concentration data sequence, the concentration of limestone slurry can be controlled more accurately, ensuring the optimal reaction conditions during the desulfurization process, thereby improving the desulfurization efficiency and reducing sulfur dioxide emissions; by using the PID control algorithm, the slurry concentration can be dynamically adjusted according to the real-time concentration data to achieve rapid response and ensure the stability and efficiency of the system under different working conditions; by effectively controlling the desulfurization process, the emissions of harmful gases can be significantly reduced, promoting environmental protection and achieving a more sustainable production method.

[0024] The present invention has the following beneficial effects:

[0025] By analyzing the variation characteristics of the concentration data and pH value data of the limestone slurry in the desulfurization system, the authenticity of the data at each moment can be effectively evaluated, thereby realizing the adjustment of the adaptive error factor at the interpolation moment. The adjustment of the adaptive error factor can dynamically optimize the response of the model according to the actual situation, improving the flexibility and accuracy of data interpolation, thus enhancing the overall performance of the model, ultimately improving the desulfurization efficiency and stability, reducing the economic losses caused by operation errors or system failures, and reducing the overall operation cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of illustration and not limitation, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0027] Figure 1 is a step flowchart of a method for optimizing wet desulfurization parameters based on data processing according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0029] The core purpose of this solution is to interpolate the missing data points in the limestone slurry concentration data using the HTFE model before using the PID control algorithm to intelligently control the limestone slurry concentration data. During the process, the error factor at each interpolation moment will be adapted, mainly based on the analysis of the numerical change characteristics of the concentration data and pH value data of the limestone slurry at each moment.

[0030] The following will describe the specific implementation manners of the present invention in detail with reference to the accompanying drawings.

[0031] Please refer to Figure 1 , which shows a flowchart of the steps of a wet desulfurization parameter optimization method based on data processing provided by an embodiment of the present invention. The method includes the following steps:

[0032] S1: Collect the concentration values at each moment of the limestone slurry in the desulfurization system and the preprocessed pH value.

[0033] Collect the limestone slurry concentration data and the slurry pH value data in the desulfurization system through an on-line concentration sensor and a pH value sensor, and then use an analog-to-digital conversion device to digitally convert the collected data.

[0034] Implementers can set the collection duration and collection frequency according to the specific implementation situation. For example, the collection duration is 3h and the collection frequency is 2s / time; since there may be a situation where there are missing values at the first moment, in order to prevent the lack of historical data for reference in such a situation, pre-collection can be carried out first as historical data. For example, the pre-collection time is 1min.

[0035] It should be noted that since the present invention optimizes the parameters for the limestone slurry concentration data, and there may be data loss in the limestone slurry concentration data, the present invention will utilize the analysis of the change characteristics of the limestone slurry concentration data itself and its correlation with the pH value data at the corresponding same moment. Then, when analyzing the correlation between the two types of data, the pH value data may also be distorted or lost. However, since the pH value data plays an auxiliary reference role in this solution and considering the influence of the algorithm complexity, before analyzing the correlation between the two types of data, we can use the Simple Moving Average (SMA) algorithm to simply preprocess the pH value data to make the pH value data more comprehensive and accurate to ensure the subsequent analysis of the correlation between the two types of data.

[0036] Specifically, the preprocessed pH value is the pH value after interpolation processing using the simple moving average algorithm.

[0037] S2: Denote the previous moment of any moment requiring concentration interpolation as the target moment, and calculate the concentration prominence of the target moment.

[0038] It should be noted that by analyzing the concentration value characteristics of each target moment and the concentration value change characteristics of the time period to which it belongs, the data prominence of each target moment is obtained. When analyzing this indicator, the more prominent the concentration value of each target moment, the greater the difference between the concentration value of each target moment and the concentration prediction value, and the more discrete the numerical distribution of the concentration values in the time period to which each target moment belongs, the greater the prominence of this target moment can be explained.

[0039] Based on the concentration value and concentration prediction value of the target moment, as well as the absolute value of the concentration numerical skewness and the concentration mean of the time period to which the target moment belongs, calculate the concentration prominence of the target moment.

[0040] Specifically, the concentration prediction value is the prediction value obtained through the HTFE model.

[0041] Specifically, the concentration prominence satisfies the following relational expression:

[0042] ;

[0043] In the formula, is the concentration prominence of the th target moment, is the concentration value of the th target moment, is the concentration mean of the time period to which the th target moment belongs, is the concentration prediction value of the th target moment, is the numerical skewness of the time period to which the th target moment belongs, is a hyperparameter, is the absolute value symbol.

[0044] Implementers can set the number of moments included in the time period and the hyperparameter according to the specific implementation situation. For example, the time period includes 20 moments and the hyperparameter is 0.01. The existence of the hyperparameter is to prevent and from having a value of 0.

[0045] Among them, the larger it is, the more prominent the concentration value of the th target moment can be explained compared to the data in its time period, and the greater its data prominence will be; the larger it is, it indicates that the The more likely the concentration value at a target moment has mutated, the greater the likelihood that it belongs to potential noise data points, and the greater the corresponding prominence degree. Indicates the absolute magnitude of the numerical skewness of the values in the time period to which the target moment belongs. The larger this value, the more discrete the numerical distribution of the values in the time period to which the target moment belongs. Then the higher the credibility of a mutation existing at the target moment, and the greater the corresponding data prominence degree.

[0046] S3: Use the Pearson correlation coefficient to obtain the pH values corresponding to all concentration values within the time period to which the target moment belongs.

[0047] Since there is a time difference in the changes between the limestone slurry concentration data and the pH value data, it is necessary to align the two types of data on the time axis before analyzing the correlation between the limestone slurry concentration data and the pH value data. Specifically: Calculate the Pearson correlation coefficient of the limestone slurry concentration sequence and the pH value data sequence within the same time period. Then, move the pH value concentration data sequence as a whole backward by one moment each time to obtain the Pearson correlation coefficient of the pH value data sequence and the initial limestone slurry concentration data sequence, and so on, to obtain multiple groups of Pearson correlation coefficients. For example, 60 groups. Then select the two data sequences with the largest Pearson correlation coefficients, which are the pH values corresponding to the concentration values.

[0048] S4: Calculate the data correlation of the target moment.

[0049] It should be noted that based on scenario research, in the desulfurization system, limestone slurry is mainly used to neutralize acidic gases (such as sulfur dioxide). In this process, limestone (CaCO3) reacts with acidic substances to form calcium bicarbonate (Ca(HCO3)2). As the concentration of limestone slurry increases, the pH value in the reaction will also change, usually showing an increase in the pH value. Therefore, in this step, the correlation between the limestone slurry concentration data and the pH value data at each target moment will be analyzed.

[0050] Rank the concentration value at the target moment according to the numerical size within the time period to which the target moment belongs, and rank the pH value corresponding to the concentration value at the target moment according to the data size within its time period, respectively obtaining the rankings of the concentration value at the target moment and the corresponding pH value in their respective time periods. Based on the rankings of the concentration value at the target moment and the corresponding pH value in their respective time periods, the Pearson correlation coefficient of the concentration values and the pH values corresponding to the concentration values within the time period to which the target moment belongs, and the Pearson correlation coefficient of the remaining concentration values and the pH values corresponding to the remaining concentration values within the time period to which the target moment belongs when not including the concentration value at the target moment, calculate the data correlation of the target moment.

[0051] Specifically, the data correlation satisfies the following relational expression:

[0052] ;

[0053] In the formula, is the data correlation at the th target time, and are respectively the rankings of the concentration value and the pH value corresponding to the concentration value at the th target time in their respective time periods, is the Pearson correlation coefficient of the remaining concentration values and the pH values corresponding to the remaining concentration values when the concentration value of the time period to which the th target time belongs does not include the target time, is the Pearson correlation coefficient of the concentration value and the pH value corresponding to the concentration value of the time period to which the th target time belongs, is a hyperparameter, is the natural exponential function, is the normalization function, is the absolute value symbol.

[0054] Implementers can set the hyperparameter according to the specific implementation situation. For example, 0.01. The existence of the hyperparameter is to prevent the occurrence of the situation.

[0055] Among them, represents the difference in the rankings of the limestone slurry concentration data and the corresponding pH value data at the th target time in their respective time periods, that is, the difference in relative magnitudes. The larger this value is, the lower the correlation between the limestone slurry concentration data and the corresponding pH value data at the th target time will be; represents the normalized result of the difference in the correlation coefficients of all concentration values and the corresponding pH values before and after including the concentration value of the target time for the time period to which the target time belongs. The larger this value is, the greater the influence of the limestone slurry concentration data at the th target time on the Pearson correlation coefficient between the two data. Then, it can be further explained that the correlation between the limestone slurry concentration data and the corresponding pH value data at the th target time will be lower.

[0056] S5: Calculate the data authenticity of the target time.

[0057] It should be noted that the data authenticity at each target moment is obtained by combining the data prominence and data correlation at each target moment. Among them, the smaller the data correlation at each target moment, the greater the possibility that the limestone slurry concentration data at each target moment belongs to noise data, and the lower the authenticity.

[0058] Calculate the data authenticity at the target moment based on the concentration prominence and data correlation at the target moment.

[0059] Specifically, the data authenticity satisfies the following relational expression:

[0060] ;

[0061] In the formula, is the data authenticity at the th target moment, is the data correlation at the th target moment, is the concentration prominence at the th target moment, is the normalization function.

[0062] Among them, the smaller it is, the greater the possibility that the limestone slurry concentration data at the target moment belongs to noise data, and the lower its authenticity; the larger it is, it can also indicate that the possibility that the limestone slurry concentration data at the th target moment belongs to noise data is greater, and its authenticity is lower.

[0063] S6: Calculate the adaptive error factor at the moment requiring concentration interpolation.

[0064] It should be noted that in this step, it is necessary to adaptively adjust the preset error factor at each moment requiring concentration interpolation according to the data authenticity; among them, the lower the data authenticity at each target moment, the lower the error factor required to ensure that the error between the concentration value and the concentration prediction value at each target moment has a lower impact on the prediction result at this moment requiring concentration interpolation.

[0065] Modify the preset error factor according to the data authenticity to obtain the adaptive error factor at the moment requiring concentration interpolation.

[0066] Specifically, the adaptive error factor satisfies the following relational expression:

[0067] ;

[0068] In the formula, is the An adaptive error factor for the concentration interpolation moment, is the data authenticity of the th target moment, and is a preset error factor.

[0069] Implementers can set the error factor according to the specific implementation situation. For example, 0.9.

[0070] S7: Interpolate the concentration interpolation moment using the HTFE model to optimize the wet desulfurization parameters based on data processing.

[0071] Interpolate the concentration interpolation moment using the HTFE model according to the adaptive error factor to obtain the interpolated concentration value at the concentration interpolation moment, so as to optimize the wet desulfurization parameters based on data processing.

[0072] Implementers can set the number of historical moments and the value of the prediction interval error factor referred to by the HTFE model during interpolation prediction according to the specific implementation situation. For example, the number of historical moments is 20, and the prediction interval error factor is 0.95.

[0073] Specifically, the optimization of the wet desulfurization parameters based on data processing includes:

[0074] Form a complete concentration data sequence with the concentration values at each moment and the interpolated concentration value at the concentration interpolation moment. According to the complete concentration data sequence, use the PID control algorithm to control the concentration of the limestone slurry to complete the optimization of the wet desulfurization parameters based on data processing.

[0075] Implementers can set the proportional gain, integral time, and derivative time of the PID control algorithm according to the specific implementation situation. For example, the proportional gain is 0.6, and the integral time and derivative time are both 5.

[0076] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A wet desulfurization parameter optimization method based on data processing, characterized in that Including: Collecting the concentration values and the pH values after pretreatment at each moment of the limestone slurry in the desulfurization system; Denote the previous moment of any moment requiring concentration interpolation as the target moment; calculate the concentration prominence degree of the target moment based on the concentration value and concentration prediction value at the target moment, as well as the absolute value of the concentration numerical skewness and the concentration mean of the time period to which the target moment belongs, satisfying the relational expression: , is the concentration prominence degree of the th target moment, is the concentration value of the th target moment, is the concentration mean of the time period to which the th target moment belongs, is the concentration prediction value of the th target moment, is the numerical skewness of the time period to which the th target moment belongs, is a hyperparameter, is the absolute value symbol; Using the Pearson correlation coefficient to obtain the pH values corresponding to all concentration values within the time period to which the target moment belongs. Based on the rankings of the concentration value and the corresponding pH value at the target moment in their respective time periods, the Pearson correlation coefficient of the concentration value and the corresponding pH value in the time period to which the target moment belongs, and the Pearson correlation coefficient of the remaining concentration values and the corresponding pH values of the remaining concentration values when the concentration value at the target moment is not included in the time period to which the target moment belongs, calculating the data correlation of the target moment; Calculate the data authenticity at the target time based on the concentration prominence and data correlation at the target time, satisfying the relational expression: , is the data authenticity at the th target time, is the data correlation at the th target time, is the normalization function; According to the data authenticity, correcting the preset error factor to obtain the adaptive error factor at the moment when concentration interpolation is required; according to the adaptive error factor, using the HTFE model to perform interpolation processing on the moment when concentration interpolation is required to obtain the interpolated concentration value at the moment when concentration interpolation is required, so as to realize the optimization of wet desulfurization parameters based on data processing.

2. The wet desulfurization parameter optimization method based on data processing according to claim 1, characterized in that, The pH value after pretreatment is the pH value obtained by performing interpolation processing using the simple moving average algorithm.

3. A wet flue gas desulfurization parameter optimization method based on data processing according to claim 1, characterized in that The predicted concentration value is the predicted value obtained through the HTFE model.

4. The wet desulfurization parameter optimization method based on data processing according to claim 1, characterized in that, The data correlation satisfies the following relational expression: ； Wherein, is the data correlation of the th target time, and are respectively the rankings of the concentration value and the pH value corresponding to the concentration value of the th target time in their respective time periods, is the Pearson correlation coefficient of the remaining concentration values and the pH values corresponding to the remaining concentration values of the time period to which the th target time belongs when the concentration value of the target time is not included, is the Pearson correlation coefficient of the concentration value and the pH value corresponding to the concentration value of the time period to which the th target time belongs, is a hyperparameter, is the natural exponential function, is the normalization function, is the absolute value symbol.

5. A wet flue gas desulfurization parameter optimization method based on data processing according to claim 1, characterized in that, The adaptive error factor satisfies the following relational expression: ； In the formula, is the adaptive error factor at the th concentration interpolation moment, is the data authenticity at the th target moment, is the preset error factor.

6. A wet desulfurization parameter optimization method based on data processing according to claim 1, characterized in that The realization of the optimization of wet desulfurization parameters based on data processing includes: Combining the concentration values at each moment and the interpolated concentration values at the moments when concentration interpolation is required to form a complete concentration data sequence. According to the complete concentration data sequence, using the PID control algorithm to control the concentration of the limestone slurry, and completing the optimization of wet desulfurization parameters based on data processing.

Images