Optimization method of nitrogen oxide concentration prediction model based on supervised learning

Abstract

The present invention discloses a method for optimizing a nitrogen oxide concentration prediction model based on supervised learning, which specifically relates to the field of environmental protection, and comprises the following steps: step one, collecting a coal burning index project set according to a coal burning index, collecting a boiler operating index project set according to a boiler operating index, and collecting a boiler emission index project set according to a boiler emission index; step two, respectively processing and operating the coal burning index project set, the boiler operating index project set, and the boiler emission index project set; the present invention can use Spearman to respectively analyze the correlation between the time series data of a single item in the process of coal burning or boiler use and the nitrogen oxide concentration, screen out the index with a high correlation coefficient with the nitrogen oxide concentration, and use a random forest algorithm to construct a correlation model between the fuel index, the boiler operating index and the nitrogen oxide concentration, thereby strengthening the correlation between the multi-item data and the emission index, and improving the prediction accuracy of the model output result.

Description

Technical Field

[0001] The present invention relates to the field of environmental protection, and more specifically, to a method for optimizing a nitrogen oxide concentration prediction model based on supervised learning. Background Art

[0002] The random forest model is a supervised learning algorithm, which is mainly used to solve classification and regression problems. The random forest algorithm uses the idea of ​​ensemble learning to vote or average the prediction results of multiple trees, thereby improving the generalization ability of the model. Its basic principle is Bootstrap sampling, which independently builds n decision trees on random subsets of the original training data set. Then the decision results of n trees are voted or averaged to obtain regression or classification results. With the deepening of environmental protection awareness and the policy response of social sustainable development, the emission indicators of thermal power plants during operation, such as sulfur dioxide, nitrogen oxides, particulate matter, mercury and its compounds, are included in the key focus projects. Accurate prediction of nitrogen oxide concentration is crucial for environmental protection and equipment operation. Existing prediction models mostly use time series data of a single project during coal burning or boiler use to predict nitrogen oxide concentration. There are still certain deficiencies in prediction accuracy and error control. Therefore, a nitrogen oxide concentration prediction model optimization method based on supervised learning is proposed. Summary of the invention

[0003] To achieve the above object, the present invention provides the following technical solutions:

[0004] The method for optimizing the nitrogen oxide concentration prediction model based on supervised learning includes the following steps:

[0005] Step 1: Collect coal combustion index project sets according to coal combustion index, collect boiler operation index project sets according to boiler operation index, and collect boiler emission index project sets according to boiler emission index;

[0006] Step 2: respectively process the coal burning index project set, the boiler operation index project set, and the boiler emission index project set to obtain a coal burning index data set, a boiler operation index data set, and a boiler emission index data set;

[0007] Step 3: Screening the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set;

[0008] Step 4: Use the random forest model to construct a correlation model between the coal combustion index association set, the boiler operation index association set and the specific emission indicators in the boiler emission index data set;

[0009] Step 5: Obtain the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold range, the prediction model optimization is completed.

[0010] In a preferred embodiment, the specific emission index in the boiler emission index data set refers to nitrogen oxide concentration.

[0011] In a preferred embodiment, the items in the coal combustion index item set, the boiler operation index item set, and the boiler emission index item set are all collected according to the time dimension.

[0012] Processing the coal-fired index project set, boiler operation index project set, and boiler emission index project set separately means: performing unified unit conversion processing on the data collected from the coal-fired index project set, boiler operation index project set, and boiler emission index project set, respectively, so that the measurement units of the preset corresponding projects remain unified.

[0013] In a preferred embodiment, in step 3, the coal burning index data set and the boiler operation index data set are screened respectively to obtain the coal burning index association set and the boiler operation index association set, which refer to:

[0014] Step 1: Take the nitrogen oxide concentration in the boiler emission index data set as the analysis object, and then use Spearman to analyze the correlation between all indicators in the coal-fired index data set and the boiler operation index data set and the nitrogen oxide concentration; this is because through the normality test of the data, it is found that some data do not obey the normal distribution, so Spearman is used for correlation analysis;

[0015] Step 2: Set the correlation coefficient between each index in the coal combustion index data set and the nitrogen oxide concentration as Ai, and set the correlation coefficient between each index in the boiler operation index data set and the nitrogen oxide concentration as Bi;

[0016] Step 3: Compare Ai with the preset screening interval Compare, if , then the corresponding coal index data is retained, otherwise it is eliminated, and Bi is compared with the preset screening interval Compare, if , then the corresponding boiler operation index data will be retained, otherwise it will be eliminated;

[0017] Step 4: Summarize the retained data of the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set.

[0018] In a preferred embodiment, in step 5, the evaluation index preset for the association model in step 4 is obtained. When the evaluation index satisfies the preset evaluation threshold interval, the optimization of the prediction model is completed, which means:

[0019] Obtain the evaluation index Pi used to measure the difference between the predicted value and the true value, compare the value of the evaluation index Pi with the preset corresponding evaluation interval, and when the values ​​of all evaluation indicators Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed.

[0020] Technical effects and advantages of the present invention:

[0021] The present invention can use Spearman to perform correlation analysis on the time series data of a single item in the process of coal burning or boiler use and the nitrogen oxide concentration, and screen out indicators with a high correlation coefficient with the nitrogen oxide concentration, which can better reflect the correlation between fuel indicators, boiler operation indicators and emission indicators, strengthen the correlation between multi-item data and emission indicators, and use the random forest algorithm to construct a correlation model between fuel indicators, boiler operation indicators and nitrogen oxide concentration, thereby improving the prediction accuracy of the model output results.

[0022] The present invention can obtain an evaluation index Pi for measuring the difference between a predicted value and a true value, and compare the value of the evaluation index Pi with a preset corresponding evaluation interval. When the values ​​of all evaluation indexes Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed, and finally the prediction result of the prediction model is output, thereby ensuring the overall prediction accuracy and error control. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0024] Figure 1 This is a schematic diagram of the principle of the nitrogen oxide concentration prediction model optimization method based on supervised learning in the present invention.

[0025] Figure 2 This is a heat map of the Spearman correlation coefficient between the coal combustion index and the nitrogen oxide concentration in Example 3 of the present invention.

[0026] Figure 3 This is a schematic diagram of the construction process of the random forest algorithm in Example 3 of the present invention to construct a correlation model between fuel indicators, boiler operation indicators and nitrogen oxide concentration.

[0027] Figure 4 is the thermal dynamics of the Spearman correlation coefficient between the operating index and the nitrogen oxide concentration in Example 3 of the present invention. Figure 1 .

[0028] Figure 5 is the thermal dynamics of the Spearman correlation coefficient between the operating index and the nitrogen oxide concentration in Example 3 of the present invention. Figure 2 .

[0029] Figure 6is the thermal dynamics of the Spearman correlation coefficient between the operating index and the nitrogen oxide concentration in Example 3 of the present invention. Figure 3 .

[0030] Figure 7 is the thermal dynamics of the Spearman correlation coefficient between the operating index and the nitrogen oxide concentration in Example 3 of the present invention. Figure 4 .

[0031] Figure 8 is the thermal dynamics of the Spearman correlation coefficient between the operating index and the nitrogen oxide concentration in Example 3 of the present invention. Figure 5 .

[0032] Fig. 9 This is Table 1 in Example 3 of the present invention.

[0033] Fig.10 This is Table 2.1 in Example 3 of the present invention.

[0034] Fig.11 This is Table 2.2 in Example 3 of the present invention.

[0035] Fig.12 This is Table 3 in Example 3 of the present invention.

[0036] Fig.13 This is Table 4.1 in Example 3 of the present invention.

[0037] Fig.14 This is Table 4.2 in Example 3 of the present invention.

[0038] Fig.15 This is Table 4.3 in Example 3 of the present invention.

[0039] Fig.16 This is Table 5 in Example 3 of the present invention.

[0040] Fig.17 This is Table 6 in Example 3 of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Example 1

[0043] The method for optimizing the nitrogen oxide concentration prediction model based on supervised learning includes the following steps:

[0044] Step 1: Collect the following information based on the time dimension:

[0045] Collect coal burning index item set according to coal burning index;

[0046] Collect boiler operation index item set according to boiler operation index;

[0047] Collect boiler emission index project sets according to boiler emission index;

[0048] Step 2: respectively process the coal-fired index project set, the boiler operation index project set, and the boiler emission index project set to obtain a coal-fired index data set, a boiler operation index data set, and a boiler emission index data set; respectively process the coal-fired index project set, the boiler operation index project set, and the boiler emission index project set to refer to: respectively convert the data collected from the coal-fired index project set, the boiler operation index project set, and the boiler emission index project set into a unified unit so that the measurement units of the preset corresponding projects remain unified;

[0049] Step 3: filtering the coal-burning index data set and the boiler operation index data set respectively to obtain the coal-burning index association set and the boiler operation index association set; In step 3, filtering the coal-burning index data set and the boiler operation index data set respectively to obtain the coal-burning index association set and the boiler operation index association set refers to:

[0050] Step 1: Take the nitrogen oxide concentration in the boiler emission index data set as the analysis object, and then use Spearman to analyze the correlation between all indicators in the coal combustion index data set and the boiler operation index data set and the nitrogen oxide concentration;

[0051] Step 2: Set the correlation coefficient between each index in the coal combustion index data set and the nitrogen oxide concentration as Ai, and set the correlation coefficient between each index in the boiler operation index data set and the nitrogen oxide concentration as Bi;

[0052] Step 3: Compare Ai with the preset screening interval Compare, if , then the corresponding coal index data is retained, otherwise it is eliminated, and Bi is compared with the preset screening interval Compare, if , then the boiler operation index data corresponding to the item is retained, otherwise it is eliminated; i only indicates the item category and does not involve the specific meaning, which will not be repeated in the following text;

[0053] Step 4: Summarize the retained data of the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set;

[0054] Step 4: Use the random forest model to construct a correlation model between the coal combustion index association set, the boiler operation index association set and the specific emission index in the boiler emission index data set; the specific emission index in the boiler emission index data set refers to the nitrogen oxide concentration.

[0055] Step 5: Obtain the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold interval, the prediction model optimization is completed. In step 5, obtaining the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold interval, the prediction model optimization is completed, which means:

[0056] Obtain the evaluation index Pi used to measure the difference between the predicted value and the true value, compare the value of the evaluation index Pi with the preset corresponding evaluation interval, and when the values ​​of all evaluation indicators Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed.

[0057] Example 2

[0058] The method for optimizing the nitrogen oxide concentration prediction model based on supervised learning includes the following steps:

[0059] Step 1: Collect coal combustion index project sets according to coal combustion index, collect boiler operation index project sets according to boiler operation index, and collect boiler emission index project sets according to boiler emission index;

[0060] Step 2: respectively process the coal burning index project set, the boiler operation index project set, and the boiler emission index project set to obtain a coal burning index data set, a boiler operation index data set, and a boiler emission index data set;

[0061] Step 3: Screening the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set;

[0062] Step 4: Use the random forest model to construct a correlation model between the coal combustion index association set, the boiler operation index association set and the specific emission indicators in the boiler emission index data set;

[0063] Step 5: Obtain the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold range, the prediction model optimization is completed.

[0064] The boiler emission index data set includes data on emission indexes such as sulfur dioxide, nitrogen oxides, particulate matter, mercury and their compounds, which are used to reflect boiler efficiency. In this embodiment, the specific emission index refers to nitrogen oxide concentration.

[0065] The items in the coal-fired index project set, boiler operation index project set, and boiler emission index project set are all collected according to the time dimension. When using them, they can be matched separately according to the uniqueness of time to obtain the data of the coal-fired index project set, boiler operation index project set, and boiler emission index project set at the same time.

[0066] The processing operations for the coal-fired index project set, the boiler operation index project set, and the boiler emission index project set respectively refer to: performing unified unit conversion processing on the data collected from the coal-fired index project set, the boiler operation index project set, and the boiler emission index project set, so that the measurement units of the preset corresponding projects remain unified. The measurement units are selected in accordance with international standards. The unified unit conversion processing refers to converting the units into measurement units selected in accordance with international standards. The specific conversion method belongs to the existing technology well known to those skilled in the art and will not be elaborated on here.

[0067] In step 3, the coal burning index data set and the boiler operation index data set are screened respectively, and the coal burning index association set and the boiler operation index association set are obtained, which are:

[0068] Step 1: Take the nitrogen oxide concentration in the boiler emission index data set as the analysis object, and then use Spearman to analyze the correlation between all indicators in the coal combustion index data set and the boiler operation index data set and the nitrogen oxide concentration;

[0069] Step 2: Set the correlation coefficient between each index in the coal combustion index data set and the nitrogen oxide concentration as Ai, and set the correlation coefficient between each index in the boiler operation index data set and the nitrogen oxide concentration as Bi;

[0070] Step 3: Compare Ai with the preset screening interval Compare, if , then the corresponding coal index data is retained, otherwise it is eliminated, and Bi is compared with the preset screening interval Compare, if , then the corresponding boiler operation index data will be retained, otherwise it will be eliminated; , are preset comparison values, and , All are positive numbers;

[0071] Step 4: Summarize the retained data of the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set.

[0072] In step 5, the preset evaluation index of the association model in step 4 is obtained. When the evaluation index meets the preset evaluation threshold range, the optimization of the prediction model is completed, which means:

[0073] Obtain the evaluation index Pi used to measure the difference between the predicted value and the true value, compare the value of the evaluation index Pi with the preset corresponding evaluation interval, and when the values ​​of all evaluation indicators Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed.

[0074] Example 3

[0075] A nitrogen oxide concentration prediction model optimization method based on supervised learning, a nitrogen oxide concentration prediction model optimization method based on supervised learning, comprising the following steps:

[0076] Step 1: Collect coal combustion index project sets according to coal combustion index, collect boiler operation index project sets according to boiler operation index, and collect boiler emission index project sets according to boiler emission index;

[0077] Among the production and operation indicators of thermal power plants, the coal combustion indicators related to nitrogen oxide concentration are selected in this embodiment, including the total coal volume, the proportion of Shenhua coal, the proportion of high-calorific value and low-volatility coal, and the proportion of economic coal types, as shown in Table 1. Fig. 9 shown.

[0078] In this embodiment, according to the influence of boiler operation on nitrogen oxides in actual production operation, the total air volume of the boiler, the oxygen content of the furnace of #1 machine, the oxygen content of the exhaust gas of #1 machine, the exhaust gas temperature of #1 machine, the hot secondary air temperature at the outlet of A and B air preheaters, the secondary air flow rate of the lower layer on the left side before and after the furnace, the inlet temperature of the primary air heater on the A and B sides, the furnace pressure, etc. are selected as boiler operation indicators. The specific indicators are shown in Table 2: Table 2 is composed of Tables 2.1 and 2.2, such as Fig.10 , Fig.11 As shown;

[0079] Step 2: respectively process the coal burning index project set, the boiler operation index project set, and the boiler emission index project set to obtain a coal burning index data set, a boiler operation index data set, and a boiler emission index data set;

[0080] Step 3: Screening the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set;

[0081] Step 4: Use the random forest model to construct a correlation model between the coal combustion index association set, the boiler operation index association set and the specific emission indicators in the boiler emission index data set;

[0082] Step 5: Obtain the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold range, the prediction model optimization is completed.

[0083] The specific emission index in the boiler emission index data set refers to the nitrogen oxide concentration, that is, the NOx conversion value at the B side inlet, numbered U1_E0HSA21CQ102L.

[0084] The items in the coal combustion index project set, boiler operation index project set, and boiler emission index project set are all collected according to the time dimension.

[0085] Processing the coal-fired index project set, boiler operation index project set, and boiler emission index project set separately means: performing unified unit conversion processing on the data collected from the coal-fired index project set, boiler operation index project set, and boiler emission index project set, respectively, so that the measurement units of the preset corresponding projects remain unified.

[0086] In step 3, the coal burning index data set and the boiler operation index data set are screened respectively, and the coal burning index association set and the boiler operation index association set are obtained, which are:

[0087] Step 1: Take the nitrogen oxide concentration in the boiler emission index data set as the analysis object, and then use Spearman to analyze the correlation between all indicators in the coal combustion index data set and the boiler operation index data set and the nitrogen oxide concentration;

[0088] Step 2: Set the correlation coefficient between each index in the coal combustion index data set and the nitrogen oxide concentration as Ai, and set the correlation coefficient between each index in the boiler operation index data set and the nitrogen oxide concentration as Bi;

[0089] Step 3: Compare Ai with the preset screening interval Compare, if , then the corresponding coal index data is retained, otherwise it is eliminated, and Bi is compared with the preset screening interval Compare, if , then the corresponding boiler operation index data will be retained, otherwise it will be eliminated;

[0090] Step 4: Summarize the retained data of the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set.

[0091] Referring to Tables 1 and 2, the Spearman correlation coefficient analysis results of coal combustion index, boiler operation index and nitrogen oxide concentration are shown in Table 3. Fig.12 As shown; the results of the Spearman correlation coefficient analysis between the operating indicators and nitrogen oxides are shown in Table 4, which consists of Table 4.1, Table 4.2, and Table 4.3. Fig.13 , Fig.14 , Fig.15 As shown;

[0092] According to the analysis of Table 3 and Table 4, the total coal volume and the proportion of economic coal are positively correlated with the concentration of nitrogen oxides. Among them, the total coal volume is most correlated with the boiler efficiency, with a correlation coefficient of 0.300; the proportion of Shenhua coal and the proportion of high-heat and low-volatile coal are negatively correlated with the boiler efficiency. The secondary air flow rate of the middle layer on the left side of the furnace (U1_A08M028C) and the NOx conversion value of the B side inlet (U1_E0HSA21CQ102L) show the strongest positive correlation. The positive correlation indicators of other correlation coefficients are arranged from high to low as follows: the secondary air flow rate of the upper layer on the right side of the furnace (U1_A08M031C), the exhaust gas temperature of #1 machine ( PER1_TFLUEGAS), secondary air flow rate of OFA layer on the right side in front of the furnace (U1_A08M033C), outlet temperature of primary air heater on side A (U1_E1HLC10CT302), inlet temperature of primary air heater on side A (U1_E1HLC10CT301), inlet temperature of primary air heater on side B (U1_E2HLC10CT301), outlet temperature of primary air heater on side B (U1_E2HLC10CT302), total coal quantity (T_COAL41_TOTALL), and proportion of economic coal types (T_COAL41_JJ). The negative correlation between the NOx conversion value at the inlet of side B (U1_E0HSA21CQ102L) and the #1 unit_exhaust gas oxygen content (PER1_PERCOFLUEGAS) is the strongest, and the correlation coefficient is -0.1403. The correlation expressions of the coal-fired index association set, the boiler operation index association set and the nitrogen oxide concentration obtained after screening are as follows: Y=f(Xi), Xi refers to the index items in the coal-fired index association set and the boiler operation index association set respectively, and Y is the nitrogen oxide concentration. Y=f(Xi) is a function declaration, which represents the relationship between a variable xi and another variable y. The specific expression is constructed by the actual user. The model items selected in this embodiment are shown in Table 5. Table 5 is shown in Fig.16 shown.

[0093] Model construction: Use the random forest algorithm to build a correlation model between fuel indicators, boiler operation indicators and nitrogen oxide concentration. The specific construction process is as follows: Figure 3 As shown, the first 100 rows of data in the test set are selected to visualize the model fitting effect.

[0094] In step 5, the preset evaluation index of the association model in step 4 is obtained. When the evaluation index meets the preset evaluation threshold range, the optimization of the prediction model is completed, which means:

[0095] Obtain the evaluation index Pi used to measure the difference between the predicted value and the true value, and compare the value of the evaluation index Pi with the preset corresponding evaluation interval. When the values ​​of all the evaluation indicators Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed, and finally the prediction result of the prediction model can be output.

[0096] In this embodiment, the evaluation index P1 is RMSE, and the evaluation index P2 is MAPE, as shown in Table 6. Fig.17 As shown in the figure, RMSE=9.0878, indicating that the deviation between the predicted value and the true value of the RF model is small; MAPE=3.1539, indicating that the relative error of the RF model is small and the prediction accuracy is high. Therefore, the use of random forests in this study to construct a correlation model between fuel indicators, operation indicators and emission indicators is effective, and it needs to be further explained that:

[0097] If any of the evaluation indicators does not meet the preset evaluation threshold range, the prediction model optimization is not completed. In this case, the retained Ai and Bi can be , The corresponding indicator data items are deleted one at a time, and they can be deleted in order from small to large until all evaluation indicators meet the preset evaluation threshold range, and the prediction model optimization is completed.

[0098] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0099] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0100] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A nitrogen oxide concentration prediction model optimization method based on supervised learning, characterized in that: The following steps are involved: Step 1: Collect coal combustion index project sets according to coal combustion index, collect boiler operation index project sets according to boiler operation index, and collect boiler emission index project sets according to boiler emission index; Step 2: respectively process the coal burning index project set, the boiler operation index project set, and the boiler emission index project set to obtain a coal burning index data set, a boiler operation index data set, and a boiler emission index data set; Step 3: Perform screening operations on the coal burning index data set and the boiler operation index data set respectively to obtain a coal burning index association set and a boiler operation index association set; Step 4: Use the random forest model to construct a correlation model between the coal combustion index association set, the boiler operation index association set and the specific emission indicators in the boiler emission index data set; Step 5: Obtain the preset evaluation index of the association relationship model in step 4. When the evaluation index meets the preset evaluation threshold range, the prediction model optimization is completed; The specific emission indicators in the boiler emission indicator dataset refer to the nitrogen oxide concentration; The items in the coal combustion index project set, boiler operation index project set, and boiler emission index project set are all collected according to the time dimension; In step 3, the coal burning index data set and the boiler operation index data set are screened respectively, and the coal burning index association set and the boiler operation index association set are obtained, which are: Step 1: Take the nitrogen oxide concentration in the boiler emission index data set as the analysis object, and then use Spearman to analyze the correlation between all indicators in the coal combustion index data set and the boiler operation index data set and the nitrogen oxide concentration; Step 2: Set the correlation coefficient between each index in the coal combustion index data set and the nitrogen oxide concentration as Ai, and set the correlation coefficient between each index in the boiler operation index data set and the nitrogen oxide concentration as Bi; Step 3: Compare Ai with the preset screening interval Compare, if , then the corresponding coal index data is retained, otherwise it is eliminated, and Bi is compared with the preset screening interval Compare, if , then the corresponding boiler operation index data will be retained, otherwise it will be eliminated; Step 4: Summarize the retained data of the coal burning index data set and the boiler operation index data set to obtain a coal burning index association set and a boiler operation index association set.

2. The method for optimizing the nitrogen oxide concentration prediction model based on supervised learning according to claim 1, characterized in that: Processing the coal-fired index project set, boiler operation index project set, and boiler emission index project set separately means: performing unified unit conversion processing on the data collected from the coal-fired index project set, boiler operation index project set, and boiler emission index project set, respectively, so that the measurement units of the preset corresponding projects remain unified.

3. The method for optimizing the nitrogen oxide concentration prediction model based on supervised learning according to claim 2, characterized in that: In step 5, the preset evaluation index of the association model in step 4 is obtained. When the evaluation index meets the preset evaluation threshold range, the optimization of the prediction model is completed, which means: Obtain the evaluation index Pi used to measure the difference between the predicted value and the true value, compare the value of the evaluation index Pi with the preset corresponding evaluation interval, and when the values ​​of all evaluation indicators Pi fall within the preset corresponding evaluation interval, the prediction model optimization is completed.

Images