Thermal power plant adaptive model modeling method, device and equipment and storage medium

Abstract

The invention relates to the technical field of thermal power generation modeling, and particularly discloses a thermal power plant adaptive model modeling method, device and equipment and a storage medium, and the method comprises the steps: obtaining multiple key control parameters of a target unit; optimizing the key control parameters to obtain target control parameters corresponding to the target unit; obtaining an optimization target of the target unit; dividing the target unit into a plurality of target working conditions based on the target control parameters; determining a plurality of influence characteristic variables under each target working condition; and performing [1, 2] interval normalization processing on a plurality of influence characteristic variables corresponding to each target working condition, performing high-order modeling and verification, and obtaining a thermal power plant adaptive model passing verification. According to the scheme, working condition division is performed through adaptive optimization targets, target unit models matched with different optimization targets can be obtained, and the efficiency of the thermal power plant adaptive model is improved. The method has self-adaptability to the thermal power plant with a multi-target unit, the model adaptation capability is improved, and the model precision is improved by optimizing key control parameters.

Description

Technical Field

[0001] The present invention belongs to the technical field of thermal power generation modeling, and in particular relates to a method, device, equipment and storage medium for adaptive modeling of a thermal power plant. Background Art

[0002] A thermal power plant, also known as a thermal power plant, uses combustible materials (such as coal) as fuel to produce electricity. The basic production process is as follows: The combustion of the fuel heats water to generate steam, converting the fuel's chemical energy into thermal energy. The steam pressure then drives a turbine, converting the thermal energy into mechanical energy. The turbine then drives a generator, converting the mechanical energy into electrical energy.

[0003] In the field of thermal power generation, efficient and stable unit operation is crucial to energy utilization and power supply. As the power industry's demand for energy conservation, emission reduction, and refined management increases, establishing models that adapt to unit characteristics and optimize production processes and control strategies has become a key direction for industry development. Summary of the Invention

[0004] In view of the shortcomings of the prior art, according to one aspect of the present application, a method for modeling a thermal power plant adaptive model is disclosed, the method comprising: Obtain multiple key control parameters of the target unit; Optimizing the key control parameters to obtain target control parameters corresponding to the target unit; Obtaining an optimization target for the target unit; Dividing the target unit into operating conditions based on the target control parameter to obtain multiple target operating conditions that meet the optimization target; Determining a plurality of influencing characteristic variables of the optimization target under each target operating condition, wherein no coupling relationship exists between the plurality of influencing characteristic variables; Perform 1-2 interval normalization on multiple characteristic variables corresponding to each target working condition; The normalized characteristic variables are subjected to high-order modeling and then verified to obtain the verified adaptive model of the thermal power plant; wherein the adaptive model of the thermal power plant is characterized based on the following formula: ; Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Characterized by 、 Interaction terms 、 Order coefficient; is a constant term.

[0005] In some embodiments, the optimizing the key control parameters to obtain the target control parameters corresponding to the target unit includes: Determining shutdown events, bad point events and bottleneck events among the key control parameters; Acquire shutdown non-steady-state data, bad pixel non-steady-state data, and bottleneck non-steady-state data respectively caused by the shutdown event, the bad pixel event, and the bottleneck event; Obtaining missing parameters of the key control parameters; Filling the missing parameters and deleting the shutdown non-steady-state data, bad point non-steady-state data and bottleneck non-steady-state data from the key control parameters to obtain intermediate control parameters; De-noising is performed on the intermediate control parameters to obtain the target control parameters.

[0006] In some embodiments, the acquiring of the shutdown non-steady-state data, the bad pixel non-steady-state data, and the bottleneck non-steady-state data caused by the shutdown event, the bad pixel event, and the bottleneck event, respectively, includes: Get the downtime period corresponding to the downtime event; Determining the shutdown-related key control parameters corresponding to the shutdown time period as shutdown non-steady-state data; Determining the time point of the bad pixel event; Determine the bad point associated key control parameters corresponding to the target time range at the time point as bad point non-steady-state data; Determining the interruption point of the bottleneck event based on the push-pull principle; Determine past-related key parameters of a preset past time period of the interruption point and future-related key parameters of a preset future time period of the interruption point; The past-related key parameters and the future-related key parameters are determined as the bottleneck non-steady-state data.

[0007] In some embodiments, dividing the target unit operating conditions based on the target control parameter to obtain multiple target operating conditions that meet the optimization target includes: Obtaining a target-oriented parameter of the optimization target in the target control parameter; The target unit is divided into operating conditions based on the target-oriented parameters.

[0008] In some embodiments, determining a plurality of influencing characteristic variables of the optimization target under each target operating condition includes: Determining necessary characteristic variables of the optimization target under the target operating conditions based on the process mechanism; Determining the marginal characteristic variables of the optimization target under the target working condition based on the Pearson phase relationship formula; Determining the influencing characteristic variables based on the necessary characteristic variables and the marginal characteristic variables; Wherein, the Pearson correlation coefficient formula is: ; In the formula, the numerator is used to measure the covariance between the feature variable X and the optimization index Y (or the covariance between feature X1 and feature X2, reflecting the direction and strength of linear association; The denominator is the product of the standard deviations of X and Y (or feature X1 and feature X2), which is used to normalize the result so that it is in the range [-1, 1].

[0009] In some embodiments, if there is a strong coupling relationship between the multiple influencing feature variables, the method further includes: determining a mutually coupled coupling characteristic variable group from a plurality of the influencing characteristic variables; Based on the Pearson correlation coefficient formula, determining reasonable coupling characteristic variables and redundant coupling characteristic variables in the coupling characteristic variable group; The redundant coupling characteristic variables are deleted from the coupling characteristic variable group to achieve decoupling of the coupling characteristic variable group.

[0010] In some embodiments, the method performs high-order modeling and post-validation on the normalized feature variables, including: Performing residual analysis, main effect analysis, and interaction effect analysis on the adaptive model of the thermal power plant; When the residual analysis, the main effect analysis, and the interaction effect analysis all meet their corresponding threshold conditions, it is determined that the verification is passed, and the verified thermal power plant adaptive model is obtained.

[0011] According to another aspect of the present application, a thermal power plant adaptive modeling device is also disclosed, the device comprising: Key control parameter acquisition module, used to obtain multiple key control parameters of the target unit; a target control parameter determination module, configured to optimize the key control parameters to obtain target control parameters corresponding to the target unit; An optimization target acquisition module, used to obtain the optimization target of the target unit; an operating condition division module, configured to divide the operating conditions of the target unit based on the target control parameter to obtain a plurality of target operating conditions that meet the optimization target; an influencing characteristic variable determination module, configured to determine a plurality of influencing characteristic variables of the optimization target under each target operating condition, wherein no coupling relationship exists between the plurality of influencing characteristic variables; A normalization processing module is used to perform normalization processing on multiple characteristic variables corresponding to each target working condition in the interval [1,2]; The model determination module is used to perform high-order modeling and verification on the normalized characteristic variables to obtain the verified adaptive model of the thermal power plant; wherein the adaptive model of the thermal power plant is characterized based on the following formula: ; Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Characterized by 、 Interaction terms 、 Order coefficient; is a constant term.

[0012] According to another aspect of the present application, an electronic device is also disclosed, which includes a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes each step of the adaptive modeling method for a thermal power plant as described in any of the above items.

[0013] According to another aspect of the present application, a computer-readable storage medium is also disclosed, on which instructions are stored, characterized in that when the instructions are executed by a processor, each step of the adaptive modeling method for a thermal power plant as described in any of the above items is implemented.

[0014] The present invention includes but is not limited to the following beneficial effects: (1) This scheme divides the working conditions by adapting the optimization target, and can obtain the target unit model that matches different optimization targets. It is adaptive to the thermal power plant with multiple target units, improves the model adaptability, and improves the model accuracy by optimizing the key control parameters. (2) The influencing characteristic variables in this scheme are influencing characteristic variables with no coupling relationship, which can avoid mutual interference between variables, reduce modeling errors, and allow the model to more accurately capture the independent and synergistic influence of each variable on the optimization target, thereby improving the accuracy of model prediction and control. (3) By normalizing the influencing characteristic variables in the interval [1, 2], the problem of causing the opposite optimization result can be avoided. (4) This scheme can identify shutdowns, bad points, bottleneck events and corresponding non-steady-state data, and specifically eliminate these interfering non-steady-state data, greatly improving the purity of the modeling data, and laying a high-quality data foundation for subsequent model construction. By obtaining and filling in the missing parameters, it can avoid insufficient model training or deviation due to missing data, ensure the integrity of the modeling data in the time and parameter dimensions, and enable the model to fully learn the unit operation characteristics. After a series of processing such as non-steady-state data, gap filling, and denoising, the intermediate control parameters and target control parameters can more truly reflect the operating status of the unit. Inputting such data into the training model can reduce interference factors and make the model output more in line with reality. In addition, high-quality and clean data can make the model training process more stable, reduce model oscillation or overfitting problems caused by abnormal data, and allow the model to maintain good performance under different working conditions, thereby improving the model's generalization ability and long-term operation reliability. (5) This scheme first determines the necessary characteristic variables through the process mechanism to ensure that the characteristics conform to physical laws and avoid the data-driven model from selecting features without reason. Then, the Pearson correlation coefficient is used to screen the marginal characteristic variables, quantify the linear correlation between the variables and the optimization target, and supplement the weakly correlated but valuable features not covered by the mechanism, so that the feature set is more complete and more in line with the actual working conditions, thereby improving the model accuracy. Furthermore, the Pearson coefficient can identify the characteristic variables with strong coupling effects, avoid the redundant information between features interfering with model training, and improve the model accuracy. (6) After decoupling the strongly coupled characteristic variables, this scheme deletes the redundant characteristic variables to avoid the prediction bias caused by repeated learning of the same influence and improve the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for describing the embodiments or the prior art.

[0016] Figure 1 This is a flow chart of a method for modeling an adaptive model of a thermal power plant according to an embodiment of the present application; Figure 2 This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application; Figure 3 This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application; Figure 4 This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application; Figure 5 This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application; Figure 6 This is a structural block diagram of a thermal power plant adaptive modeling device according to an embodiment of the present application; Figure 7 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention; Figure 8 This is a comparison chart of the predicted value and the actual value of the model established in Example 1; Figure 9 is the residual analysis diagram in Example 1; Figure 10 This is the main effect analysis diagram in Example 1; Figure 11 is the interaction effect analysis diagram in Example 1; Figure 12 This is a comparison chart of the predicted value and the actual value of the model established in Example 2; Figure 13 This is the residual analysis diagram in Example 2; Figure 14 This is the main effect analysis diagram in Example 2; Figure 15 This is the interaction effect analysis diagram in Example 2. DETAILED DESCRIPTION

[0017] An embodiment of the present invention provides a method for modeling an adaptive model of a thermal power plant, the method comprising obtaining multiple key control parameters of a target unit; optimizing the key control parameters to obtain target control parameters corresponding to the target unit; obtaining an optimization target for the target unit; dividing the target unit into operating conditions based on the target control parameters to obtain multiple target operating conditions that meet the optimization target; determining multiple influencing characteristic variables of the optimization target under each target operating condition, wherein no coupling relationship exists between the multiple influencing characteristic variables; normalizing the multiple characteristic variables corresponding to each target operating condition to a [1, 2] interval; and performing high-order modeling and subsequent verification on the normalized influencing characteristic variables to obtain a verified adaptive model of the thermal power plant. This solution divides the operating conditions according to the adaptive optimization target to obtain a target unit model that matches different optimization targets. This solution is adaptive to thermal power plants with multiple target units, improves the model's adaptability, and improves model accuracy by optimizing key control parameters.

[0018] The terms "first," "second," "third," "fourth," and so on (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that shown or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, product, or apparatus.

[0019] For ease of understanding, the specific process of the embodiment of the present invention is described below. Specifically, Figure 1 This is a flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application. Figure 1 As shown, the following steps are included: S100: Acquire multiple key control parameters of the target unit.

[0020] Specifically, the power generation company's plant-level real-time monitoring system (SIS) can be used to retrieve key control parameters for the target unit's air, smoke, steam, and pulverizing subsystems within a target timeframe, such as one year or six months. Field reports can then be used to obtain incoming coal quality analysis data for that timeframe. The target unit can be a specific generating unit within a thermal power plant requiring real-time optimization modeling and analysis, such as a 660MW supercritical generating unit. Key control parameters (CCPs) refer to core parameters retrieved from the thermal power plant's real-time monitoring system (SIS) that reflect the unit's operating status. These parameters include, but are not limited to, total coal feed, primary air flow, secondary air flow, furnace negative pressure, furnace oxygen content, air preheater outlet oxygen content, flue gas temperature, and induced draft fan current for the air and smoke subsystem; feedwater flow, feedwater temperature, main steam flow, main steam pressure, main steam temperature, reheat steam temperature, reheat steam pressure, water-to-coal ratio, and superheat for the steam-water subsystem; operating parameters related to pulverized coal preparation for the pulverized coal preparation subsystem; and turbine-related parameters such as turbine load (active power), maximum turbine steam inlet, condenser vacuum, circulating water feed temperature, and circulating water pump operating status. Specifically, the time granularity for these CCPs is 1 minute, while coal quality test data is collected once per shift (8 hours). First, align the timestamps and granularity of the control parameters and test data to achieve data merging.

[0021] S102: Optimize key control parameters to obtain target control parameters corresponding to the target unit.

[0022] Specifically, the optimization processing of key control parameters includes but is not limited to data cleaning and denoising processing, wherein the data processing is used to clean out some abnormal data, and the abnormal data may include but is not limited to duplicate data.

[0023] S104: Obtain the optimization target of the target unit.

[0024] The optimization objectives of the target unit may include but are not limited to combustion optimization and cold end optimization.

[0025] S106 , dividing the target unit into operating conditions based on the target control parameters to obtain multiple target operating conditions that meet the optimization target.

[0026] Specifically, different optimization objectives correspond to different operating conditions. Combustion optimization improves boiler efficiency (steam production per ton of coal, steam production per ton of standard coal) by adjusting air volume (primary air / secondary air). Cold-end optimization improves net power generation (power generation per ton of steam) by adjusting turbine backpressure (condenser vacuum). Net power generation is the difference between turbine active power and circulating water pump power consumption. In one example, combustion optimization can be divided into operating conditions based on the actual unit load (active power), a target control parameter. Depending on the amount of data, four to eight operating conditions can be divided. For example, dividing four operating conditions uses the first quartile, median, and third quartile of the active power dataset as the dividing boundaries. In addition to dividing operating conditions based on actual unit load, cold-end optimization also depends on external environmental conditions, such as circulating water feed temperature. Furthermore, since most power plants use fixed-speed or dual-speed pumps with high and low speed switching, the operating states of the circulating water pumps are discrete. Therefore, under the condition of sufficient data volume, it can be divided into 4×4=16 operating conditions according to the main steam flow rate and circulating water feed temperature of the unit, and finally the optimal start and stop state of the circulating water pump under each operating condition can be found.

[0027] In another example, the target control parameters can be first derived from the target optimization objective. The target unit's operating conditions can then be classified based on these target control parameters. These target control parameters are directly related to the optimization objective and accurately reflect the direction and degree of impact of the target unit's operating state on the objective. For example, to optimize power generation efficiency, parameters directly influencing energy conversion, such as unit load, main steam pressure, and coal quality parameters (calorific value, volatile matter, etc.), are selected as target control parameters. Furthermore, based on the different combinations and variations of these parameters, the target unit's operating state can be classified into several operating conditions. For example, complex operating conditions can be categorized by unit load range (low load, medium load, high load), coal quality type (high-quality coal, low-quality coal), or main steam pressure range, allowing the model to optimize for different operating conditions and improve adaptability.

[0028] S108. Determine multiple influencing characteristic variables of the optimization target under each target operating condition.

[0029] Among them, there is no coupling relationship between multiple influencing characteristic variables. Specifically, for the boiler combustion system modeling, the optimization target y can be selected from the main steam flow, gas production per ton of coal, steam production per ton of standard coal and boiler efficiency. The dependent variable can be screened among multiple influencing characteristic variables x, such as the total coal feed, water flow, water temperature, coal received base low calorific value, flue gas temperature, induced draft fan current, primary air volume, secondary air volume, furnace negative pressure, furnace oxygen content, air preheater outlet oxygen content, water-coal ratio, superheat, etc. The influencing characteristic variables usually selected are related to the amount of data and should not be too many. For example, for each working condition with more than 1,000 sets of data, 3-6 influencing characteristic variables can be selected. In one example, such as Figure 2 As shown, it is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the application. The flow chart is an exemplary description of the multiple influencing characteristic variables of the optimization target under each target operating condition in step S108, such as Figure 2 As shown, the following steps are included: S200: Based on the process mechanism, determine the necessary characteristic variables of the optimization target under the target working conditions.

[0030] The process mechanism refers to the physical and chemical laws governing the thermal power production system and the operating principles of the equipment. When selecting influencing characteristic variables, the first step is to determine the necessary characteristic variables for the optimization objective under the target operating conditions. For example, in the combustion optimization process, characteristics representing air volume and boiler load are necessary characteristic variables.

[0031] S202. Based on the Pearson phase relationship formula, determine the marginal characteristic variables of the optimization target under the target working condition.

[0032] Specifically, some features that cannot be directly determined, such as water-coal ratio and furnace negative pressure, can be temporarily retained. Based on the Pearson correlation formula, the retained features x are correlated with the optimization target y, and the Pearson correlation coefficient r is calculated. Focus on the correlation coefficient between the features x and y temporarily retained in the previous step, and retain features with |r|>0.4. The formula is as follows: ; In the formula, the numerator is used to measure the covariance between the feature variable X and the optimization index Y (or the covariance between feature X1 and feature X2, reflecting the direction and strength of linear association; The denominator is the product of the standard deviations of X and Y (or feature X1 and feature X2), which is used to normalize the result so that it is in the range [-1, 1].

[0033] S204: Determine the influencing characteristic variables based on the necessary characteristic variables and the marginal characteristic variables.

[0034] Specifically, necessary characteristic variables and marginal characteristic variables are used as influencing characteristic variables.

[0035] It is understandable that multiple features x and y are still retained by the above method, and then multiple influencing feature variables can be matched with multiple optimization objectives through permutations and combinations. However, the currently retained influencing feature variable features may show strong coupling between features and between features and objectives, such as strong coupling between coal feed rate and water feed rate, water feed rate and main steam flow rate, boiler efficiency and exhaust gas temperature, which need to be decoupled before modeling.

[0036] Specifically, in this example, Figure 3 As shown in the figure, another flow chart of the adaptive modeling method of thermal power plant is given, see Figure 3 , including the following steps: S300: Determine a mutually coupled coupling characteristic variable group from a plurality of influencing characteristic variables.

[0037] S302. Determine reasonable coupling characteristic variables and redundant coupling characteristic variables in the coupling characteristic variable group based on the Pearson correlation coefficient formula.

[0038] For example, based on the Pearson correlation coefficient formula, the coal feed and water feed are calculated to be r=0.85, which indicates strong coupling, and the coal feed and optimization target (such as boiler efficiency) are r=0.7, and the water feed and optimization target (boiler efficiency) are r=0.6; it can be determined that the coal feed is a reasonable coupling characteristic variable, and the water feed is a redundant coupling characteristic variable.

[0039] S304: Deleting redundant coupling characteristic variables from the coupling characteristic variable group to achieve decoupling of the coupling characteristic variable group.

[0040] Specifically, steps S302-S304 enable decoupling of strongly coupled feature variables. This means retaining one feature while deleting features that are strongly coupled to the target. Otherwise, the impact of other features on the target will be significantly reduced or misjudged. This step also significantly reduces the number of permutations and combinations, reducing the modeling workload. For steam turbine modeling, the modeling objective y can be selected from unit active power and power generation per ton of steam. The dependent variables can be selected from features x, such as main steam flow, temperature, and pressure; reheat steam temperature and pressure; condenser vacuum, and circulating water feed temperature. The subsequent feature screening method is similar to combustion optimization.

[0041] It can be understood that the influencing characteristic variables in this solution are uncoupled influencing characteristic variables, which can avoid mutual interference between variables, reduce modeling errors, and allow the model to more accurately capture the independent and synergistic effects of each variable on the optimization target, thereby improving the accuracy of model prediction and control. Furthermore, this example solution first determines the necessary characteristic variables through the process mechanism to ensure that the characteristics conform to physical laws and avoid the nonsensical feature selection of the data-driven model. It then uses the Pearson correlation coefficient to screen marginal characteristic variables, quantify the linear correlation between variables and optimization targets, and supplement weakly correlated but valuable features not covered by the mechanism, making the feature set more complete and more in line with actual working conditions, thereby improving model accuracy. Furthermore, the Pearson coefficient can identify strongly coupled influencing characteristic variables, avoid redundant information between features interfering with model training, and improve model accuracy.

[0042] S110 , performing normalization processing on the [1, 2] interval for multiple characteristic variables corresponding to each target operating condition.

[0043] Specifically, to ensure that different feature variables have the same weight in the model and avoid the situation where "large numbers eat up small numbers", each feature variable needs to be "normalized" before model training. The "normalization" method is as follows: ; Where, is a dimensioned characteristic predictor variable; for The maximum value of an array; for The minimum value of the array.

[0044] Among them, the maximum and minimum values ​​can be defined by the process engineer according to the actual process (when the historical data contains extreme conditions, the upper limit of the historical data can be used as the upper limit, and the lower limit as the lower limit); to ensure the accuracy of the model, under the normal operating conditions of the factory, the upper and lower limits of the actual characteristic variables should not exceed this range; this range should not be set too large, so as not to affect the effect of the characteristic variable in the model. In the above way, all characteristic variables are converted to values ​​between 1 and 2. The reason why the characteristic variables are not converted to the traditional normalization strategy between [0,1] is mainly because in actual operation, there will be conditions where the characteristic variables are less than their lower limit. At this time, the normalized result of the characteristic variable will be less than 0, which makes the effect of the characteristic variable in the model opposite, resulting in the problem of opposite optimization results. Specifically, by normalizing the influencing characteristic variables in the interval [1,2], the problem of opposite optimization results can be avoided.

[0045] S112. Perform high-order modeling and verification on the normalized influencing characteristic variables to obtain a verified adaptive model of the thermal power plant.

[0046] Specifically, the adaptive model of a thermal power plant is characterized based on the following formula: ; Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Characterized by 、 Interaction terms 、 Order coefficient; is a constant term.

[0047] According to the above formula, list the response y and the predictor variable , , ...... of This method aims to extract modeling information from the input and output data of the system, and use the main factors that affect the output information to build a mathematical relationship between independent variables. If the optimization target y is the steam production per ton of coal after feature screening, the influencing characteristic variable x is: total primary air volume , total secondary air volume , Main water flow , water-coal ratio ,when =1, the first-order polynomial ,when =2, the second-order polynomial .

[0048] Power plant logistics typically have clear cause-and-effect relationships, such as the relationship between total coal feed and steam pressure, and between wind temperature and combustion efficiency. Therefore, the present invention constructs independent models by sampling first-order polynomials for the primary responses to plant output and various constraints. This model retains cross terms in high-dimensional representations and removes square terms that have no practical physical meaning, thus avoiding the risk of overfitting and maintaining interpretability of process mechanisms.

[0049] Specifically, after feature screening, multiple combinations of optimization objectives y and multiple influencing characteristic variables x are obtained, so that multiple polynomial models can be constructed. Past experience usually selects the model with the best accuracy performance based on the evaluation indicators of the regression model, such as mean square error, mean absolute error, etc. However, in actual industrial production, the accuracy of the model is important, but more important is the adjustment direction and amplitude of the characteristic variables. If the coefficients of the characteristic variables in the model have opposite signs, the actual adjustment direction is wrong, which will lead to a decrease in unit efficiency. In this application, the polynomial model is verified by performing residual analysis, main effect analysis and interaction effect analysis on the adaptive model of the thermal power plant; when the residual analysis, main effect analysis and interaction effect analysis all meet their corresponding threshold conditions, it is determined that the verification is passed, and a verified thermal power plant adaptive model is obtained.

[0050] (1) For residual analysis: After multidimensional characterization, the actual value and fitted value of each set of optimization targets y can be obtained. The residuals of the actual value and the fitted value are calculated as follows: Before performing residual analysis, the normal probability plot of the residual, the residual frequency histogram, the residual and fitted value plot and the residual sequence plot are mainly made. When performing model residual analysis, the main analysis is in three aspects: The residuals of the model should roughly conform to the normal distribution: in the normal probability plot, ideally, the data points should be distributed approximately in a straight line along the diagonal line, indicating that the residuals meet the normality assumption. In the frequency histogram, check whether it is a symmetrical bell-shaped distribution.

[0051] Residual and fitted value distribution analysis: examine whether the residuals maintain equal variance, that is, the residuals are randomly scattered around 0, without systematic changes, and should not be funnel-shaped or trumpet-shaped.

[0052] Residual sequence plot analysis: examine whether the points in the scatter plot fluctuate randomly and irregularly up and down the horizontal axis.

[0053] (2) For main effect analysis: The main effect refers to the average effect of different levels of a single independent variable (factor) on the dependent variable in a multifactor experiment or study, without considering the interaction effects of other variables.

[0054] The specific implementation method is as follows: after the model training is completed, it is necessary to keep other characteristic variables unchanged. The normalized data mean can be taken, and only a single independent variable is changed. The result of substituting it into the model equation is observed to see the effect of its change on the response. For example, when analyzing the effect of coal calorific value on steam production per ton of coal, it is necessary to exclude the influence of variables such as boiler load and air volume on the system before observing. Therefore, for variables other than coal calorific value, the average value of the normalized historical data is taken and brought into the model, and the time series data of coal calorific value normalization is brought into the model to observe the target trend. This trend should be consistent with the experience of engineers, such as the positive correlation between steam production per ton of coal and coal calorific value, the positive correlation between steam production per ton of coal and boiler load, and the positive correlation between air volume and outlet oxygen volume. Since the present invention uses a 1st-order model, a main effect analysis is performed on each characteristic variable, and the degree of influence of each feature on the target can be judged by the slope of the corresponding straight line, and the relationship between the change of a single physical quantity and the response is observed to see whether it conforms to the process logic and the range of change in actual production. The feature screening combination and its training model that conforms to the experience of engineers are retained.

[0055] (3) For interaction effect analysis: Interaction effect refers to the situation where two or more parameters act together and their combined effect on the result is not equal to the simple superposition of the individual effects of each parameter. For example, the combination of total coal feed and total air volume during boiler combustion will have a nonlinear effect on combustion efficiency, that is, is the introduced interaction term.

[0056] The calculation method of interaction effect is as follows: If the characteristic variable is selected , need to calculate and The interaction effect of and The characteristic variables remain unchanged, and the normalized data mean is also taken. First, Take the maximum value (normalized to 2), Take the normalized average value of historical data and bring it into the model to observe The effect of the change on the optimization objective y; then Take the minimum value (normalized to 1), Take the normalized average value of historical data and bring it into the model to observe The effect of the change on the optimization target y, and vice versa, Take the maximum or minimum value and observe The impact of the change on the target y. By constructing a lower triangle interaction effect analysis diagram, we traverse all parameter combinations, identify the impact of key interaction pairs, and retain the feature screening combination and its training model that conforms to the engineer's experience.

[0057] In one example, Figure 4This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application. This flow chart is an exemplary description of optimizing the key control parameters in step S102 to obtain the target control parameters corresponding to the target unit. For details, see Figure 2 , including the following steps: S400: Determine shutdown events, bad point events, and bottleneck events among key control parameters.

[0058] Specifically, a shutdown event refers to the transition from an "operating state" to a "shutdown state" for a unit, including planned shutdowns (such as maintenance and peak-shaving shutdowns) and unplanned shutdowns (such as fault trips and protective trips). A bad point event refers to an "irrational data jump" caused by sensor failure or transient equipment anomalies. A bottleneck event refers to an "operational bottleneck" caused by equipment hardware limitations or operational strategies, preventing the unit from continuing to optimize its objectives (such as efficiency or load), manifesting as "parameters adjusted to their limits but with no response."

[0059] S402: Obtain shutdown non-steady-state data, bad pixel non-steady-state data, and bottleneck non-steady-state data caused by the shutdown event, bad pixel event, and bottleneck event, respectively.

[0060] Specifically, Figure 5 This is another flow chart of the adaptive modeling method for a thermal power plant according to an embodiment of the present application. This flow chart is an exemplary description of step S402 of obtaining shutdown non-steady-state data, bad point non-steady-state data, and bottleneck non-steady-state data caused by shutdown events, bad point events, and bottleneck events, respectively. For details, see Figure 5 , including the following steps: S500: Obtain a downtime period corresponding to a downtime event.

[0061] In one example, a thermal power plant's plant-level monitoring information system (SIS) and distributed control system (DCS) can be used to capture status signals from real-time databases related to shutdown equipment (such as boilers and turbines). When the equipment's operating status changes from "running" to "shutdown" (e.g., turbine speed drops to 0, boiler main burners are all shut down, and other such notable status changes), the timestamp at that time is recorded as the shutdown start point. When the equipment resumes operation and enters a stable operating state (such as when the speed reaches rated and the boiler resumes stable steam production), the recovery timestamp is recorded as the shutdown end point. The shutdown start and end points are used to determine the shutdown period.

[0062] In yet another feasible solution, the downtime period may be determined in conjunction with the log.

[0063] S502: Determine the shutdown-related key control parameters corresponding to the shutdown time period as shutdown non-steady-state data.

[0064] Specific shutdown-related key control parameters refer to parameters that are directly affected by the shutdown action and are strongly related to the shutdown process. Examples include equipment status parameters, process parameters, and auxiliary system parameters. Equipment status parameters may include turbine speed, generator active power, and boiler total coal feed rate; process parameters may include main steam pressure and reheat steam temperature; and auxiliary system parameters may include circulating water pump frequency and induced draft fan current.

[0065] S504: Determine the moment of the bad pixel event.

[0066] Specifically, the time point when the sensor or the like fails may be determined as the time point of the bad point event.

[0067] S506: Determine the bad pixel-associated key control parameters corresponding to the target time range at the time point as bad pixel non-steady-state data.

[0068] It's understandable that thermal power plant parameter fluctuations have "inertia," or sensor failures can be "gradual" (e.g., a response delay precedes drift), or a fault can trigger a "chain fluctuation" (e.g., false coal feed data leading to abnormal air volume regulation). Therefore, starting with the moment of the fault event, we can extend the target time forward or backward, and identify the corresponding key parameters associated with the fault within that time range as the fault-point non-steady-state data. These key parameters and control parameters associated with the fault can refer to the physical mechanism, control logic, and device relationships of the sensor or hardware device involved in the fault event.

[0069] S508: Determine the interruption point of the bottleneck event based on the push-pull rule.

[0070] S510 determines the past-related key parameters of the preset past time period of the interruption point and the future-related key parameters of the preset future time period of the interruption point.

[0071] S512: Determine the past associated key parameters and the future associated key parameters as bottleneck non-steady-state data.

[0072] Understandably, in addition to deleting non-steady-state data from extended downtimes and bad pixel data caused by market fluctuations, raw material supply, and equipment failures, production lines often experience operational bottlenecks, which limit production capacity. An effective method for identifying production bottlenecks is the "Pull-Push" approach: Before a bottleneck, upstream processes produce according to a planned or fixed pace, "pushing" materials downstream. If upstream capacity exceeds the bottleneck, continued production push results in a buildup of intermediate products before the bottleneck. After the bottleneck, downstream processes have higher capacity than the bottleneck, but because the bottleneck limits material supply, they are forced to wait (pull-based production). A bottleneck can cause a disruption in the raw material supply of a production line. If sufficient data is available, the point of the disruption can be directly identified. Key parameters associated with the disruption in both the past and future time periods are identified as non-steady-state data and deleted. For power generation companies, the point of total coal supply interruption is identified, and non-steady-state data surrounding that point is searched for. All rows with the corresponding timestamp are deleted.

[0073] The preset past time period may refer to the period before the interruption point occurs, and the preset future time period may refer to the period after the interruption point occurs. Past key-related parameters may refer to process parameters, equipment status parameters, and operational control parameters that are directly or indirectly associated with the interruption point (e.g., key time nodes such as equipment failure and data anomalies) within a past historical period and can reflect, influence, or be affected by the interruption point. Future key-related parameters may refer to process parameters, equipment status parameters, and operational control parameters that are directly or indirectly associated with the interruption point (e.g., key time nodes such as equipment failure and data anomalies) within a future period and can reflect, influence, or be affected by the interruption point.

[0074] It is understandable that this example solution can identify shutdowns, bad points, bottleneck events and corresponding non-steady-state data, and specifically eliminate these interfering non-steady-state data, greatly improving the purity of modeling data and laying a high-quality data foundation for subsequent model construction; by obtaining and filling in missing parameters, it avoids insufficient model training or deviations due to missing data, ensures the integrity of modeling data in time and parameter dimensions, and enables the model to comprehensively learn the operating characteristics of the unit; through a series of processing such as deleting non-steady-state data, filling in gaps, and denoising, the intermediate control parameters and target control parameters can more truly reflect the operating status of the unit. Inputting such data into the training model can reduce interference factors and make the model output more in line with reality. In addition, high-quality and clean data can make the model training process more stable, reduce model oscillation or overfitting problems caused by abnormal data, and enable the model to maintain good performance under different working conditions, thereby improving the model's generalization ability and long-term operation reliability.

[0075] S404: Obtain missing parameters of key control parameters.

[0076] S406 , completing the missing parameters and deleting the shutdown non-steady-state data, the bad point non-steady-state data and the bottleneck non-steady-state data from the key control parameters to obtain the intermediate control parameters.

[0077] S408: De-noising the intermediate control parameters to obtain target control parameters.

[0078] Furthermore, to ensure the data integrity of the data integrity system, when the collected data contains a large number of intermittent gaps, data filling methods are needed to complete the data. If the difference between the breakpoint data is not large, the data of the previous moment can be used for data filling. If the difference between the data before and after the breakpoint is large, the average of the data before and after the breakpoint can be used for data filling.

[0079] To prevent noise from affecting the data results, digital filtering was required to remove the noise and improve the data's representativeness. Because the data sampling period was one minute, a large amount of data was deleted during data cleaning. This resulted in discontinuities in the existing data. By adding a column to the merged table, rounding the original timestamps to 30 minutes, and applying a 30-minute average filter to all parameters, we were able to achieve this.

[0080] For ease of understanding, this application is exemplified with reference to a 660MW supercritical thermal power generating unit: Example 1: A boiler combustion system model was established for a 660MW supercritical thermal power generator unit. First, the power plant's real-time monitoring system (SIS) was used to retrieve key control parameters for the target unit's coordinated control, air / smoke, steam / water, and pulverizing subsystems, as well as the turbine extraction, high-pressure heater, low-pressure heater, and cold terminal systems for the target period (e.g., July 1, 2023, to June 30, 2024). The data was collected at a 1-minute granularity, totaling approximately 530,000 sets of data. On-site boiler coal quality test results, slag test results, and fly ash combustible test results were also obtained. The test data was collected once per shift (8 hours), totaling approximately 1,100 sets of data.

[0081] First, the control system and test system data are aligned according to the timestamp to achieve data merging. Secondly, according to the data processing flow, the shutdown non-steady-state data, bad point non-steady-state data and bottleneck non-steady-state data are deleted. Further data filling and 30-minute data filtering are performed, and 12,600 sets of processed data are obtained for subsequent modeling.

[0082] Modeling of the boiler combustion system of a 660MW supercritical power plant: The boiler combustion system data is based on the unit power generation load P gen As the basis for the division of working conditions, there are “Pgen <262MW", "262<=P gen <287MW", "287<=P gen <304MW", "304<=P gen <360MW", "360<=P gen <422MW",、、"422<=P gen <495MW", "495<=P gen <547MW" and " Pgen There are 8 operating conditions in total, with approximately 1,800 sets of data for each operating condition.

[0083] Multiple optimization objectives are available for modeling, including main steam flow, steam production per ton of coal, steam production per ton of standard coal, boiler efficiency, and coal consumption per standard coal power generation. The main steam flow is directly obtained through data processing; steam production per ton of coal = main steam flow / total coal consumption; steam production per ton of standard coal = steam production per ton of coal × 29.307 / Qnet,ar, where Qnet,ar is the received lower calorific value of the incoming coal, in MJ / kg. Boiler efficiency is calculated using the counterbalance method; coal consumption per standard coal power generation = total coal consumption × Qnet,ar / 29.307 / power generation.

[0084] According to the feature screening, 6 influencing characteristic variables are obtained for permutation and combination: total primary air volume, total secondary air volume, total coal feed volume, main feed water flow, water-coal ratio, superheat and low calorific value of coal entering the furnace. Since the total coal feed volume is highly coupled with the main feed water flow, the two should not be selected at the same time; the main feed water flow and the target main steam flow are highly coupled. Therefore, when the main steam flow is the target, it is best not to select the total coal feed volume and the main feed water flow as characteristic variables. Each combination has no less than 4 variables at most, and a multidimensional characterization model for each characteristic variable and target combination is constructed. Taking the operating condition of "Pgen>=547MW", the target is the steam production per ton of coal, and the characteristic variables are total primary air volume, total secondary air volume, main feed water flow, water-coal ratio and Qnet,ar as an example, a comparison chart of the predicted value and the actual value of the model is established as shown below. Figure 8 shown.

[0085] The established model was then analyzed step by step based on the process-oriented verification principle, where the residual analysis diagram is shown in the figure below. Figure 9 As shown, Figure 9 Figure a is the residual normal probability plot, b is the residual frequency histogram, c is the residual vs. fitted value plot, and d is the residual sequence plot. The normal probability plot and histogram show that the model residuals should roughly conform to a normal distribution. The fitted value distribution diagram shows that the residuals maintain roughly equal variances and lack a funnel or trumpet shape. The residual sequence plot shows that the points fluctuate irregularly above and below the horizontal axis, indicating that the model has passed residual analysis verification.

[0086] Further main effect analysis of the model is carried out, and the main effect analysis diagram is as follows Figure 10 As shown in the figure, the greater the main feed water flow, the greater the boiler load, and the more steam is produced per ton of coal, which is consistent with the physical law and production experience that efficiency increases with increasing load. The low calorific value of the coal entering the furnace is Q net,ar The positive correlation with steam production per ton of coal also conforms to the physical law that higher calorific value results in higher steam production. Without considering other variables that may have an uncertain impact on the trend, the model is considered to have passed the main effects analysis validation. If this model is used for optimization, the main effects analysis indicates that reducing the primary and secondary air volumes can increase steam production per ton of coal, thereby improving furnace efficiency. Figure 10 Among them, e is the main effect analysis diagram of total primary air volume, f is the main effect analysis diagram of total secondary air volume, g is the main effect analysis diagram of main feed water flow, h is the main effect analysis diagram of water-coal ratio, and i is the main effect analysis diagram of low calorific value of coal entering the furnace.

[0087] Further interaction effect analysis is carried out, and the interaction effect analysis diagram is as follows Figure 11 As shown in the figure, taking the interactive effect of water-coal ratio and total secondary air volume as an example, under high load, when the water-coal ratio takes the maximum value, the total secondary air volume can increase the steam production per ton of coal by reducing the total secondary air volume; when the water-coal ratio takes the minimum value, the opposite is true, which is consistent with production experience; the synergistic effect of main feed water flow and primary air volume is weak, and the water-coal ratio and Q net,ar The synergistic effect is weak. The interaction effect analysis diagram does not clearly violate physical laws and production experience, and it is believed that the model has been verified by the interaction effect analysis. In summary, the feasibility of the model is proven.

[0088] Example 2: In the application of 660MW supercritical thermal power generating unit, the boiler combustion system model and the steam turbine system model are established. The steam turbine modeling does not consider the realization of cold end optimization for the time being, and takes the main steam flow as the F vapor Based on the working conditions, there are 8 working conditions in total. F vapor Divided into: F vapor < 731.24t / h", "731.24 <=F vapor < 814.6t / h", "814.6 <=F vapor < 868.38t / h","868.38 <=F vapor < 1031.1t / h", "1031.1<= F vapor <=1219.0t / h", "1219.0<= F vapor < = 1426.7t / h", "1426.7<= F vapor <= 1578.8t / h" and " F vapor > = 1578.8t / h".

[0089] Steam Turbine Modeling Objectives y The unit active power and ton steam power generation can be selected, and 6 characteristic variables are obtained according to the characteristic screening for permutation and combination: main steam flow, main steam pressure, main steam temperature, high temperature reheat steam temperature, high temperature reheat steam pressure, and condenser vacuum. Due to the high coupling between the main steam flow and the unit active power, the main steam flow is not selected as the characteristic variable when the unit active power is the target. Each combination has no less than 4 variables, and a multidimensional standard model of each characteristic variable and target combination is constructed. Taking the operating condition as "731.24 <=F vapor < 814.6t / h", the target is the power generation per ton of steam, the characteristic variables are main steam flow, main steam pressure, main steam temperature, high-temperature reheat steam temperature, and condenser vacuum, and the comparison chart between the predicted value and the actual value of the model is as follows Figure 12 shown.

[0090] The established model was then analyzed step by step based on the process-oriented verification principle. The residual analysis diagram is shown in the figure below. Figure 13 As shown, Figure 13 In the figure, A is the residual normal probability plot, B is the residual frequency histogram, C is the residual and fitted value plot, and D is the residual sequence plot.

[0091] It can be seen from the normal probability plot and histogram that the residuals of the model should roughly conform to the normal distribution, and the fitted value distribution diagram shows that the coal residuals roughly maintain equal variance and no funnel or trumpet shape. From the residual sequence diagram, it can be seen that the points fluctuate irregularly above and below the horizontal axis, indicating that the model has passed the residual analysis verification.

[0092] Further, the main effect analysis diagram is performed, and the main effect analysis diagram is as follows Figure 14As shown in the figure, E is the main effect analysis diagram for the main steam flow rate, F is the main effect analysis diagram for the main steam header pressure, G is the main effect analysis diagram for the main steam header steam temperature, H is the main effect analysis diagram for the high-temperature reheat steam temperature, and I is the main effect analysis diagram for the condenser vacuum value. The power generation per ton of steam increases with increasing main steam pressure and temperature, as well as high-temperature reheat steam temperature, and decreases with decreasing condenser back pressure, which conforms to the physical laws of steam turbine power generation and indicates that the model has passed the main effect analysis.

[0093] Furthermore, the interaction effect analysis is carried out, and the interaction effect analysis diagram is as follows Figure 15 As shown in the figure, taking main steam temperature and pressure as examples, ideally, increasing both temperature and pressure simultaneously improves power generation efficiency. However, in real systems, the effects of temperature and pressure changes on power generation are nonlinear and interactive. When the main steam temperature is at its maximum, changes in main steam pressure have little impact on the target. However, when the main steam temperature is at its minimum, power generation per ton of steam increases with increasing main steam pressure. This phenomenon is essentially due to the differential sensitivity of steam thermodynamic properties in different temperature ranges. At high temperatures, the marginal contribution of pressure to enthalpy decreases, while at low temperatures, increasing pressure effectively improves steam quality and work capacity. There is no significant synergistic effect between main steam flow and condenser vacuum, between main steam temperature and high-temperature reheat steam temperature, or between main steam temperature and condenser vacuum. The interaction effect analysis diagram does not clearly violate physical laws or production experience, and the model is considered to have been validated by the interaction effect analysis. In summary, the feasibility of the model is demonstrated.

[0094] Further, Figure 6 The structural block diagram of the modeling device for the adaptive model of a thermal power plant is shown in the figure. Figure 6 As shown, the device includes: Key control parameter acquisition module, used to obtain multiple key control parameters of the target unit; The target control parameter determination module is used to optimize the key control parameters and obtain the target control parameters corresponding to the target unit; Optimization target acquisition module, used to obtain the optimization target of the target unit; The operating condition division module is used to divide the operating conditions of the target unit based on the target control parameters to obtain multiple target operating conditions that meet the optimization target; An influencing characteristic variable determination module is used to determine multiple influencing characteristic variables of the optimization target under each target working condition, wherein there is no coupling relationship between the multiple influencing characteristic variables; A normalization processing module is used to perform normalization processing on multiple characteristic variables corresponding to each target working condition in the interval [1,2]; The model determination module is used to perform high-order modeling and verification on the normalized characteristic variables to obtain a verified adaptive model of the thermal power plant. The adaptive model of the thermal power plant is characterized by the following formula: ; Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Characterized by 、 Interaction terms 、 Order coefficient; is a constant term.

[0095] The application introduction of the relevant modules of the device in this example can refer to the relevant introduction of the above method principles, which will not be repeated here.

[0096] The scheme divides the working conditions by adapting the optimization objectives, and can obtain target unit models that match different optimization objectives. It is adaptive to thermal power plants with multiple target units, improves the model adaptability, and improves the model accuracy by optimizing key control parameters. Furthermore, the influencing characteristic variables in the scheme are uncoupled influencing characteristic variables, which can avoid mutual interference between variables, reduce modeling errors, and allow the model to more accurately capture the independent and synergistic effects of each variable on the optimization objectives, thereby improving the accuracy of model prediction and control. By normalizing the influencing characteristic variables in the interval [1, 2], the problem of opposite optimization results can be avoided. Furthermore, the scheme can identify shutdowns, bad points, bottleneck events and corresponding non-steady-state data, and specifically eliminate these interfering non-steady-state data, greatly improving the purity of modeling data, and laying a high-quality data foundation for subsequent model construction. By obtaining and filling in missing parameters, insufficient model training or deviation due to missing data is avoided, and the integrity of modeling data in time and parameter dimensions is ensured, so that the model can fully learn the unit operation characteristics. After a series of processing such as vacancy elimination and denoising, the intermediate control parameters and target control parameters can more truly reflect the operating status of the unit. Inputting such data into the training model can reduce interference factors and make the model output more in line with reality. In addition, high-quality and clean data can make the model training process more stable, reduce model oscillation or overfitting problems caused by abnormal data, and enable the model to maintain good performance under different operating conditions, thereby improving the model's generalization ability and long-term operation reliability. Furthermore, the scheme first determines the necessary characteristic variables through the process mechanism to ensure that the characteristics conform to physical laws and avoid the data-driven model from selecting features without reason. Then, the Pearson correlation coefficient is used to screen marginal characteristic variables, quantify the linear correlation between variables and optimization targets, and supplement weakly correlated but valuable features not covered by the mechanism, making the feature set more complete and more in line with actual operating conditions, thereby improving model accuracy. Furthermore, the Pearson coefficient can identify characteristic variables with strong coupling effects, avoid redundant information between features interfering with model training, and improve model accuracy. After decoupling the strongly coupled characteristic variables, redundant characteristic variables are deleted to avoid prediction bias caused by repeated learning of the same influence, thereby improving model accuracy.

[0097] above Figure 6 The apparatus in the embodiment of the present invention is described in detail from the perspective of modular functional entities, and the electronic device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0098] Figure 7: This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device 700 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 710 (for example, one or more processors) and a memory 720, and one or more storage media 730 (for example, one or more mass storage devices) for storing application programs 733 or data 732. The memory 720 and the storage medium 730 may be temporary storage or permanent storage. The program stored in the storage medium 730 may include one or more modules (not shown in the figure), each of which may include a series of instruction operations on the electronic device 700. Furthermore, the processor 710 may be configured to communicate with the storage medium 730 to execute the series of instruction operations in the storage medium 730 on the electronic device 700.

[0099] The electronic device 600 may further include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input and output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 7 The illustrated electronic device structure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0100] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the method for modeling a thermal power plant adaptive model.

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

[0102] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for building an adaptive model of a thermal power plant, characterized in that: The method comprises Obtain multiple key control parameters of the target unit; Optimizing the key control parameters to obtain target control parameters corresponding to the target unit; Obtaining an optimization target for the target unit; Dividing the target unit into operating conditions based on the target control parameter to obtain multiple target operating conditions that meet the optimization target; Determining a plurality of influencing characteristic variables of the optimization target under each target operating condition, wherein no coupling relationship exists between the plurality of influencing characteristic variables; Normalize multiple characteristic variables corresponding to each target working condition to the interval [1,2]; The normalized influencing characteristic variables are subjected to high-order modeling and then verified to obtain the verified adaptive model of the thermal power plant; wherein the adaptive model of the thermal power plant is characterized based on the following formula: ； Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Features 、 Interaction terms 、 Order coefficient; is a constant term.

2. The adaptive modeling method for a thermal power plant according to claim 1, characterized in that: The optimizing process of the key control parameters to obtain the target control parameters corresponding to the target unit includes: Determining shutdown events, bad point events and bottleneck events among the key control parameters; Acquire shutdown non-steady-state data, bad pixel non-steady-state data, and bottleneck non-steady-state data respectively caused by the shutdown event, the bad pixel event, and the bottleneck event; Obtaining missing parameters of the key control parameters; Filling the missing parameters and deleting the shutdown non-steady-state data, bad point non-steady-state data and bottleneck non-steady-state data from the key control parameters to obtain intermediate control parameters; De-noising is performed on the intermediate control parameters to obtain the target control parameters.

3. The adaptive modeling method for a thermal power plant according to claim 2, characterized in that: The acquiring of the shutdown non-steady-state data, the bad pixel non-steady-state data, and the bottleneck non-steady-state data respectively caused by the shutdown event, the bad pixel event, and the bottleneck event includes: Get the downtime period corresponding to the downtime event; Determining the shutdown-related key control parameters corresponding to the shutdown time period as shutdown non-steady-state data; Determining the time point of the bad pixel event; Determine the bad point associated key control parameters corresponding to the target time range at the time point as bad point non-steady-state data; Determining the interruption point of the bottleneck event based on the push-pull principle; Determine past-related key parameters of a preset past time period of the interruption point and future-related key parameters of a preset future time period of the interruption point; The past-related key parameters and the future-related key parameters are determined as the bottleneck non-steady-state data.

4. The adaptive modeling method for a thermal power plant according to claim 1, characterized in that: The dividing the operating conditions of the target unit based on the target control parameter to obtain multiple target operating conditions that meet the optimization target includes: Obtaining a target-oriented parameter of the optimization target in the target control parameter; The target unit is divided into operating conditions based on the target-oriented parameters.

5. The adaptive modeling method for a thermal power plant according to claim 1, characterized in that: Determining the multiple influencing characteristic variables of the optimization target under each target operating condition includes: Determining necessary characteristic variables of the optimization target under the target operating conditions based on the process mechanism; Determining the marginal characteristic variables of the optimization target under the target working condition based on the Pearson phase relationship formula; Determining the influencing characteristic variables based on the necessary characteristic variables and the marginal characteristic variables; Wherein, the Pearson correlation coefficient formula is: ； In the formula, the numerator is used to measure the covariance between the feature variable X and the optimization index Y (or the covariance between feature X1 and feature X2, reflecting the direction and strength of linear association; The denominator is the product of the standard deviations of X and Y (or feature X1 and feature X2), which is used to normalize the result so that it is in the range [-1, 1].

6. The method for building an adaptive model of a thermal power plant according to claim 1, characterized in that: If there is a strong coupling relationship between the multiple influencing characteristic variables, the method further includes: determining a mutually coupled coupling characteristic variable group from a plurality of the influencing characteristic variables; Based on the Pearson correlation coefficient formula, determining reasonable coupling characteristic variables and redundant coupling characteristic variables in the coupling characteristic variable group; The redundant coupling characteristic variables are deleted from the coupling characteristic variable group to achieve decoupling of the coupling characteristic variable group.

7. The method for building an adaptive model of a thermal power plant according to claim 1, characterized in that: The method performs high-order modeling on the normalized feature variables and then verifies the results, including: Performing residual analysis, main effect analysis, and interaction effect analysis on the adaptive model of the thermal power plant; When the residual analysis, the main effect analysis, and the interaction effect analysis all meet their corresponding threshold conditions, it is determined that the verification is passed, and the verified thermal power plant adaptive model is obtained.

8. A thermal power plant adaptive modeling device, characterized in that: The device comprises: Key control parameter acquisition module, used to obtain multiple key control parameters of the target unit; a target control parameter determination module, configured to optimize the key control parameters to obtain target control parameters corresponding to the target unit; An optimization target acquisition module, used to obtain the optimization target of the target unit; an operating condition division module, configured to divide the operating conditions of the target unit based on the target control parameter to obtain a plurality of target operating conditions that meet the optimization target; an influencing characteristic variable determination module, configured to determine a plurality of influencing characteristic variables of the optimization target under each target operating condition, wherein no coupling relationship exists between the plurality of influencing characteristic variables; A normalization processing module is used to perform normalization processing on multiple characteristic variables corresponding to each target working condition in the interval [1,2]; The model determination module is used to perform high-order modeling and verification on the normalized characteristic variables to obtain the verified adaptive model of the thermal power plant; wherein the adaptive model of the thermal power plant is characterized based on the following formula: ； Where, To optimize the indicators; The total number of characteristic variables corresponding to an optimization indicator; is the highest order polynomial; For a single feature of Order coefficient; Features 、 Interaction terms 、 Order coefficient; is a constant term.

9. An electronic device, characterized in that: The electronic device includes a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes each step of the adaptive modeling method for a thermal power plant according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the various steps of the method for modeling a thermal power plant adaptive model according to any one of claims 1 to 7 are implemented.

Images