A method and system for optimizing boiler operating parameters based on machine learning

Abstract

The present invention relates to the technical field of electronic digital data processing, and in particular to a method and system for optimizing boiler operating parameters based on machine learning. The method comprises obtaining operating parameters of the boiler during operation, steam pressure data and steam flow data in the boiler in each monitoring time period, determining an error compensation amount according to a neural network model, steam pressure data, steam flow data and operating parameters, analyzing the error compensation amount based on multiple types of data, thereby improving the accuracy of the error compensation amount and being more conducive to adapting to dynamic operating condition changes; determining a predicted boiler feed water amount according to the error compensation amount, performing pre-compensation using the error compensation amount, thereby improving the accuracy of predicting the boiler feed water amount; determining an optimization coefficient of the feed water amount according to the predicted boiler feed water amount, and optimizing the feed water amount of the operating parameters according to the optimization coefficient to obtain an optimized feed water amount, thereby improving the optimization effect of the feed water amount.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical digital data processing, and in particular to a method and system for optimizing boiler operating parameters based on machine learning. Background Art

[0002] In the field of modern automobile manufacturing, boilers are key heat energy supply equipment and are widely used in car body drying, metal heat treatment, workshop heating, etc. The operating efficiency of the boiler directly affects energy consumption, production costs and environmental emissions. Therefore, in order to improve its efficiency, stability and economy, the operating parameters of the boiler can be optimized.

[0003] Traditional boiler operating parameter optimization methods mainly rely on experience and rule-based control strategies. Although these methods have improved the operating efficiency of boilers to a certain extent, they often cannot adapt to complex and dynamic operating conditions, resulting in poor optimization effects. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a method and system for optimizing boiler operating parameters based on machine learning. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for optimizing boiler operating parameters based on machine learning, comprising:

[0006] In each monitoring period, obtain the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data;

[0007] Determining an error compensation amount according to the neural network model, the steam pressure data, the steam flow data, and the operating parameters;

[0008] Determining a predicted boiler feed water amount according to the error compensation amount;

[0009] According to the predicted boiler feed water volume, an optimization coefficient of the feed water volume is determined, and the feed water volume of the operating parameters is optimized according to the optimization coefficient to obtain the optimized feed water volume.

[0010] In one embodiment, determining the error compensation amount according to the neural network model, the steam pressure data, the steam flow data and the operating parameters includes:

[0011] Determining the pressure fluctuation factor in the boiler in each of the monitoring time periods according to the steam pressure data;

[0012] Determining the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, the steam flow data and the operating parameters;

[0013] Outputting health status according to the neural network model and the operating parameters;

[0014] An error compensation amount is determined according to the authenticity value and the health level.

[0015] In one embodiment, determining the pressure fluctuation factor in the boiler in each monitoring time period according to the steam pressure data includes:

[0016] Determining a monitoring fitting curve of the steam pressure data and a variance of the steam pressure data respectively according to the steam pressure data in each of the monitoring time periods;

[0017] Determine the maximum extreme value and the minimum extreme value in the monitoring fitting curve, and determine the extreme value difference between the maximum extreme value and the minimum extreme value;

[0018] A first product of the variance and the extreme value difference is determined, and a pressure fluctuation factor in the boiler in each of the monitoring time periods is determined according to the first product and a normalization function.

[0019] In one embodiment, determining the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, the steam flow data and the operating parameter includes:

[0020] Determining the load in each monitoring time period according to the steam flow data and the operating parameters;

[0021] determining a first standard deviation according to each of the loads, and determining a second standard deviation according to each of the pressure fluctuation factors;

[0022] determining a covariance according to each of the loads and each of the pressure fluctuation factors;

[0023] A second product of the first standard deviation and the second standard deviation is determined, and a authenticity value of the pressure fluctuation in the boiler is determined according to a ratio of the covariance to the second product.

[0024] In one implementation, determining the error compensation amount for the monitoring time period according to the authenticity value and the health level includes:

[0025] An error compensation amount for the monitoring time period is determined according to a ratio of the health degree to the authenticity value.

[0026] In one embodiment, determining the predicted boiler feed water amount according to the error compensation amount includes:

[0027] Determine, according to the operating parameters of each monitoring time period, a water supply change curve of each monitoring time period and a relevant parameter change curve corresponding to each relevant parameter related to the water supply in each monitoring time period;

[0028] Determine a prediction curve of boiler feed water volume according to the feed water volume variation curves of each monitoring time period and the related parameter variation curves;

[0029] The predicted boiler feed water volume is determined according to the error compensation amount and the prediction curve.

[0030] In one embodiment, determining the prediction curve of boiler feed water volume according to the feed water volume variation curves of each monitoring time period and each related parameter variation curve comprises:

[0031] Determine the mean square error between each of the water supply change curves and each of the corresponding related parameter change curves in each of the monitoring time periods, and determine the target mean square error between each of the water supply change curves and each of the related parameter change curves according to the mean square error corresponding to each of the monitoring time periods;

[0032] Determine the influence weight of each relevant parameter on the water supply according to each target mean square error;

[0033] By using each of the influence weights, a weighted sum is taken on the feed water quantity variation curves in the latest monitoring time period to determine a prediction curve for the boiler feed water quantity.

[0034] In one embodiment, determining the predicted boiler feed water amount according to the error compensation amount and the prediction curve includes:

[0035] Using the error compensation amount as the Kalman gain;

[0036] The prediction curve is smoothed by using the Kalman gain through a Kalman filter to determine the predicted boiler feed water volume for the next monitoring time period in the future.

[0037] In one embodiment, the step of determining an optimization coefficient of the feed water volume according to the predicted boiler feed water volume, and optimizing the feed water volume of the operating parameters according to the optimization coefficient to obtain the optimized feed water volume includes:

[0038] Determine the heat conversion efficiency of the boiler and the heat demand for the next monitoring period in the future;

[0039] Determining a predicted heat supply according to the predicted boiler feed water volume and the heat energy conversion efficiency;

[0040] Determining an optimization coefficient of water supply according to a ratio of the thermal energy demand to the predicted heating amount and a hyperbolic tangent function;

[0041] The optimized water supply is obtained according to the third product of the water supply of the operating parameters in the latest monitoring time period and the optimization coefficient.

[0042] In a second aspect, the embodiment of the present application provides a boiler operation parameter optimization system based on machine learning, including:

[0043] An acquisition module is used to acquire the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data in each monitoring time period;

[0044] An error module, used for determining an error compensation amount according to a neural network model, the steam pressure data, the steam flow data and the operating parameters;

[0045] A prediction module, used to determine a predicted boiler feed water amount according to the error compensation amount;

[0046] The optimization module is used to determine the optimization coefficient of the water supply according to the predicted boiler water supply, and optimize the water supply of the operating parameters according to the optimization coefficient to obtain the optimized water supply.

[0047] The present invention has the following beneficial effects:

[0048] By obtaining the operating parameters of the boiler during operation, the steam pressure data in the boiler and the steam flow data in each monitoring time period, determining the error compensation amount according to the neural network model, steam pressure data, steam flow data and operating parameters, and analyzing the error compensation amount based on various types of data, it is beneficial to improve the accuracy of the error compensation amount and to adapt to dynamic operating conditions. According to the error compensation amount, the predicted boiler feed water amount is determined, and the error compensation amount is used for pre-compensation, thereby improving the accuracy of the predicted boiler feed water amount. According to the predicted boiler feed water amount, the optimization coefficient of the feed water amount is determined, and the feed water amount of the operating parameters is optimized according to the optimization coefficient to obtain the optimized feed water amount, thereby improving the optimization effect of the feed water amount. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0050] Figure 1A schematic flow chart of the steps of a method for optimizing boiler operating parameters based on machine learning provided by one embodiment of the present invention;

[0051] Figure 2 A schematic diagram of steam pressure data in various monitoring time periods provided by an embodiment of the present invention;

[0052] Figure 3 A schematic diagram of pressure fluctuation factors in various monitoring time periods provided by an embodiment of the present invention;

[0053] Figure 4 A schematic diagram of loads in various monitoring time periods provided by an embodiment of the present invention;

[0054] Figure 5 A schematic diagram of a prediction curve of water supply before and after compensation provided by an embodiment of the present invention;

[0055] Figure 6 A structural block diagram of a boiler operating parameter optimization system based on machine learning provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a method and system for optimizing boiler operating parameters based on machine learning proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0057] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0058] It should be noted that the term “exemplary” in the embodiments of the present application refers to examples listed for the convenience of explanation, and other embodiments are not limited to the examples listed.

[0059] In the related art, the traditional boiler operating parameter optimization method mainly relies on experience and rule-based control strategies. Although it improves the operating efficiency of the boiler to a certain extent, it often cannot adapt to complex and dynamic operating conditions, resulting in limited optimization effects. In modern industrial environments, boilers need to maintain stable operation under changing operating conditions, which poses a challenge to traditional methods. The machine learning-based boiler operating parameter optimization method has emerged, which uses big data analysis to optimize and improve the overall performance of the boiler. However, in the related art, the accuracy of the machine learning model is highly dependent on data quality, and errors in the boiler operation process may affect the model's predictive ability. In addition, the machine learning model of the related art often performs well under specific operating conditions, but may not be able to adapt quickly when faced with changing operating conditions or new operating conditions, and the model's generalization ability is poor.

[0060] In the embodiment of the present application, by improving the generalization ability of the neural network model, the optimization of the operating parameters of the boiler is made more reasonable and accurate, and the processor can achieve a large closed loop of the heat network and boiler load linkage through more precise automatic adjustment.

[0061] The specific scheme of a boiler operating parameter optimization method and system based on machine learning provided by the present invention is described in detail below with reference to the accompanying drawings.

[0062] See also Figure 1 , which shows a flow chart of a boiler operating parameter optimization method based on machine learning provided by an embodiment of the present invention. The boiler operating parameter optimization method based on machine learning may at least include steps S100-S400:

[0063] S100. In each monitoring time period, obtain the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data.

[0064] S200, determining an error compensation amount according to a neural network model, steam pressure data, steam flow data and operating parameters.

[0065] S300: Determine the predicted boiler feed water volume according to the error compensation amount.

[0066] S400, determining an optimization coefficient of the water supply according to the predicted boiler water supply, and optimizing the water supply of the operating parameters according to the optimization coefficient to obtain an optimized water supply.

[0067] The technical solution of the embodiment of the present application obtains the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data in each monitoring time period, determines the error compensation amount according to the neural network model, the steam pressure data, the steam flow data, and the operating parameters, and analyzes the error compensation amount based on various types of data, which is beneficial to improving the accuracy of the error compensation amount and is more conducive to adapting to dynamic operating condition changes; determines the predicted boiler feed water amount according to the error compensation amount, and uses the error compensation amount for pre-compensation, thereby improving the accuracy of the predicted boiler feed water amount; determines the optimization coefficient of the feed water amount according to the predicted boiler feed water amount, and optimizes the feed water amount of the operating parameters according to the optimization coefficient to obtain the optimized feed water amount, thereby improving the optimization effect of the feed water amount.

[0068] In one embodiment, the length of the monitoring time period and the number of monitoring time periods are based on actual settings. For example, the length of the time period is 3 minutes. By installing various sensors and monitoring equipment or obtaining the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data from the boiler in each monitoring time period, these parameters and data can be transmitted to the processor in the control system such as PLC (programmable logic controller) for processing, and then the processor automatically and intelligently adjusts the boiler to achieve a large closed loop of heat network and boiler load linkage. Optionally, the operating parameters include but are not limited to feed water volume, inlet water temperature, boiler water temperature, combustion chamber temperature, steam temperature, exhaust temperature, fuel pressure, fuel flow, feed water flow, feed water temperature, inlet water temperature, fuel air flow, fuel composition, flue gas composition (such as CO 2 , CO, NOx, etc.), boiler switch status, pump and fan status, steam demand, production load, water quality, etc., or other parameters calculated based on the above parameters using existing methods, such as steam specific enthalpy, feed water specific enthalpy, boiler efficiency, etc.

[0069] It should be noted that during the actual operation of the boiler, since the steam pressure of the boiler usually fluctuates with the change of load, under normal circumstances, there is a positive correlation between the boiler load and the steam pressure in the boiler, and when the steam pressure in the boiler increases, the water level in the boiler may drop, and more water is converted into steam, indicating that the boiler is operating at a higher temperature and requires more precise water supply control. Therefore, steam pressure data is introduced for analysis in the embodiments of the present application.

[0070] In one implementation, step S200 includes steps S201-S203:

[0071] S201. Determine the pressure fluctuation factor in the boiler in each monitoring time period according to the steam pressure data.

[0072] First, according to the steam pressure data of each monitoring time period, the monitoring fitting curve of the steam pressure data is determined respectively, such as Figure 2 As shown, it is a monitoring fitting curve of the steam pressure data (boiler steam pressure) of each monitoring time period, and the variance T of the steam pressure data is determined based on the steam pressure data of each monitoring time period.

[0073] Secondly, determine the maximum extreme value U in the monitoring fitting curve max And the minimum extreme value U min , and determine the extreme value difference between the maximum extreme value and the minimum extreme value; (U max -U min ).

[0074] Furthermore, determine the first product of variance and extreme value difference T×(U max -U mi ), according to the first product T×(U max -U min ) and the normalization function norm to determine the pressure fluctuation factor F in the boiler during each monitoring period e :

[0075] F e =norm[T×(U max -U min )]

[0076] In the formula, F e It represents the pressure fluctuation factor (steam pressure fluctuation) in the boiler during the e-th monitoring period. Figure 3 As shown, the ordinate is the magnitude of the pressure fluctuation factor, T×(U max -U min ), the larger the first product is, the more drastic the fluctuation of the steam pressure in the boiler is during the monitoring period. The reason for the drastic fluctuation may be that the actual heating environment changes more complexly, or it may be due to potential abnormal conditions in the boiler operation system, and subsequent analysis is needed on the fluctuation of the steam pressure.

[0077] It should be noted that in the actual operation of the boiler, the change in steam pressure is positively correlated with the boiler load. However, a monitoring period cannot well represent the correlation between the change in steam pressure and boiler load. In multiple monitoring periods, each monitoring period may contain different pressure changes and load changes. The actual situation of steam pressure fluctuations can be further analyzed through the changes between different time periods.

[0078] S202. Determine the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, steam flow data and operating parameters.

[0079] First, the load of each monitoring period is determined based on the steam flow data and operating parameters. For example, based on the steam flow data, steam specific enthalpy and feed water specific enthalpy, the thermal efficiency of each monitoring period is calculated using the existing known methods, and the thermal efficiency is the load P e , represents the load of the e-th monitoring time period (or called real-time load), such as Figure 4 As shown, the vertical axis is the load value, in megawatts MW.

[0080] Secondly, the first standard deviation is determined according to each load, and the second standard deviation is determined according to each pressure fluctuation factor. For example, a first data set L is formed based on each load, a second data set M is formed according to each pressure fluctuation factor, and then the first standard deviation σ of the first data set L is determined. L and the second standard deviation σ of the second data set M M , when participating in subsequent calculations, σ L and σ M All of them are not 0, and when the value is 0, it is replaced by 0.1 to participate in the calculation.

[0081] Furthermore, according to each load and each pressure fluctuation factor, the covariance is determined, for example, the covariance Cov(L,M) of the first data set L and the second data set M is calculated.

[0082] Then, determine the second product of the first standard deviation and the second standard deviation σ L σ M , according to the ratio of the covariance to the second product, the authenticity value Y of the pressure fluctuation in the boiler is determined:

[0083]

[0084] In the formula, Y represents the authenticity of the steam pressure fluctuation, which characterizes the correlation coefficient between the pressure fluctuation factor and the load during the actual operation of the boiler. The larger the value, the stronger the correlation, which means that the fluctuation of the steam pressure during the actual operation is indeed caused by the change of the boiler operating load, that is, the higher the authenticity of the pressure fluctuation.

[0085] S203: Output the health status according to the neural network model and operating parameters.

[0086] It should be noted that there is a certain correlation between the steam pressure fluctuation and the boiler load. If there is a weak correlation between the change in steam pressure and the boiler load, it is considered that there is a certain error in the system or the boiler system itself may have abnormal operating conditions.

[0087] In the embodiment of the present application, multiple monitoring time periods are set in advance to collect historical operating parameters, steam pressure data, etc. to form a historical monitoring data set, and key features are selected from the historical monitoring data set, including but not limited to steam pressure data, steam temperature, feed water temperature, fuel flow, fuel air flow, exhaust temperature, flue gas composition and other features, and then the above key features are scored by manual scoring. The higher the score, the higher the health of the boiler during operation. Then, a neural network with a classification network structure of a five-layer fully connected neural network model, a loss function of a cross entropy function, and a training ratio of 7:3 set for the historical monitoring data set is used to train the neural network, and the input is the key features of the boiler during operation, and then the output is the health of the boiler during operation. After training, the trained neural network is obtained, which is recorded as a neural network model. Therefore, the above-mentioned key features in the operating parameters, steam pressure data, and steam flow data of step S100 are input into the neural network model, so as to output the health δ.

[0088] S204: Determine an error compensation amount according to the authenticity value and the health level.

[0089] Specifically, according to the ratio of the health degree δ to the authenticity value Y, the error compensation amount R of the monitoring period is determined e :

[0090]

[0091] In the formula, R represents the error compensation of the water supply in the latest (i.e. current) monitoring time period during the operation of the boiler, where Y represents the authenticity of the boiler steam pressure fluctuation; ε is a preset adjustment parameter with a value of 0.1, and the value can be adjusted by itself to prevent the denominator from being 0; δ represents the health of the latest (i.e. current) monitoring time period; the larger R is, the lower the authenticity of the actual steam pressure fluctuation, but the health of the boiler's operating system is higher, indicating that the "abnormal fluctuation" of the steam pressure is caused by the error in the actual boiler operation process, not the abnormality of the boiler operation system, that is, the higher the error compensation degree of the subsequent boiler feed water.

[0092] Then, based on the error compensation amount, compensation can be made in the process of predicting the water feed volume during the boiler operation process in the next monitoring time period in the future, so that the water feed volume of the boiler in the next monitoring time period in the future can be predicted more accurately, and the boiler's heating system can be pre-compensated.

[0093] In one implementation, step S300 includes steps S301-S303:

[0094] S301. Determine, according to the operating parameters of each monitoring time period, a water supply change curve of each monitoring time period and a related parameter change curve corresponding to each related parameter related to the water supply in each monitoring time period.

[0095] It should be noted that during the actual operation of the boiler to produce heat, the boiler's heat production will inevitably cause a loss of water in the boiler. It is very important to pre-compensate the boiler's water supply in advance according to the actual heat production demand, which can ensure that the boiler control system can adjust the boiler's operating status in real time to ensure that the boiler continues to effectively supply heat energy. During the operation of the heat energy supply boiler, the boiler's water supply is mainly related to factors such as production load, steam demand, boiler efficiency, and water inlet temperature. Therefore, the parameters related to the water supply include but are not limited to production load, steam demand, boiler efficiency, and water inlet temperature.

[0096] In the embodiment of the present application, the water supply change curve S of each monitoring time period is determined according to the operating parameters of each monitoring time period, and the relevant parameter change curve G corresponding to each relevant parameter related to the water supply in each monitoring time period is determined.

[0097] S302: Determine a prediction curve of boiler feed water volume according to feed water volume variation curves in each monitoring time period and various related parameter variation curves.

[0098] First, determine the mean square error between each water supply change curve S and each corresponding related parameter change curve G in each monitoring time period, and determine the target mean square error between each water supply change curve and each related parameter change curve based on the mean square error corresponding to each monitoring time period. For example, taking the inlet water temperature as an example, the mean square errors of each monitoring time period can be averaged to determine the final target mean square error when the related parameter is the inlet water temperature, or the minimum mean square error in the mean square errors can be determined as the final target mean square error when the related parameter is the inlet water temperature. Without specific limitation, the final target mean square error is recorded as MSE′(G,S). Therefore, based on the above principle, the target mean square error MSE′(G,S) between each water supply change curve and each related parameter change curve can be determined, which represents the target mean square error between the related parameter change curve of the Gth related parameter and the water volume change curve S.

[0099] Secondly, according to each target mean square error, determine the influence weight D of each relevant parameter on water supply. G :

[0100] D G =e -MSE′(G,S)

[0101] Where D GIt represents the influence weight of the Gth related parameter on the water supply, G represents the fitting curve of the operating parameters, in which, in the process of calculating the mean square error, the horizontal coordinate of the fitting curve can be kept unchanged, the vertical coordinate can be standardized, and the calculation can be performed with unified dimension); the smaller the value of MSE′(G,S), the greater the influence of the operating parameters on the water supply, that is, the higher the weight of the water supply curve, that is, D G The bigger.

[0102] Furthermore, the feed water change curve of the latest monitoring time period is weighted and summed using each influence weight to determine the prediction curve of the boiler feed water amount. For example, the first influence weight is multiplied by the feed water change curve of the latest monitoring time period to obtain the first weighted item, and the second influence weight is multiplied by the feed water change curve of the latest monitoring time period to obtain the second weighted item. Based on the same principle, multiple weighted items are obtained, and the multiple weighted items are added together to determine the prediction curve of the boiler feed water amount.

[0103] S303: Determine the predicted boiler feed water volume according to the error compensation amount and the prediction curve.

[0104] Specifically, the error compensation amount is used as the Kalman gain, and the Kalman filter is used to smooth the prediction curve obtained above using the Kalman gain, so as to determine the predicted boiler feed water volume in the next monitoring time period in the future.

[0105] In one implementation, step S400 includes steps S401-S404:

[0106] S401, determining the thermal energy conversion efficiency of the boiler and the thermal energy demand in the next monitoring period in the future.

[0107] Optionally, the thermal energy conversion efficiency of the boiler can be calculated based on the mass of water, the specific heat capacity of water, the calorific value of the fuel, the mass of the completely burned fuel, and the change in water temperature at the beginning and end of the monitoring time period, etc., using existing formulas. The thermal energy demand b for the next monitoring time period in the future can be calculated based on the mass of water, the specific heat capacity of water, and the change in water temperature at the beginning and end of the latest monitoring time period, and the existing formula will not be repeated here.

[0108] S402: Determine the predicted heat supply according to the predicted boiler feed water volume and heat energy conversion efficiency.

[0109] Optionally, the predicted heat supply b′ is determined based on the existing formula according to the predicted boiler feed water volume and thermal energy conversion efficiency obtained above, which will not be repeated; among them, the improvement point of the embodiment of the present application lies in how to determine the accurate predicted boiler feed water volume.

[0110] S403. Determine the optimization coefficient of water supply according to the ratio of thermal energy demand to predicted heating supply and the hyperbolic tangent function th.

[0111] Optionally, the optimization coefficient Z of water supply is:

[0112]

[0113] Among them, the larger the optimization coefficient Z of the water supply, the more it means that the predicted heat supply b′ may not be sufficient to provide the thermal energy demand b, and the preset heat supply b′ value is not 0. Therefore, it is necessary to pre-compensate the water supply of the boiler system in advance to ensure that it is not lower than the thermal energy demand in the production process.

[0114] S404. Obtain the optimized water supply according to the third product of the water supply of the operating parameter in the latest monitoring time period and the optimization coefficient.

[0115] Alternatively, if Figure 5 As shown, based on the product of the optimization coefficient Z of the feed water volume and the prediction curve of the boiler feed water volume (i.e., the prediction curve before compensation), the prediction curve of the compensated boiler feed water volume, i.e., the prediction curve after compensation, can be determined. Therefore, the third product of the feed water volume of the operating parameters in the latest monitoring time period and the optimization coefficient Z is used to obtain the final optimized feed water volume.

[0116] In the embodiment of the present application, the determined optimized water feed can ensure that during the actual operation of the boiler, the pre-compensation effect of the water feed of the boiler system is better, reducing the situation in which insufficient boiler water feed leads to low heat production efficiency during actual operation, thereby improving the stability and efficiency of boiler operation and better realizing a large closed loop of the linkage between the heat network and boiler load.

[0117] In the embodiment of the present application, the change between the steam pressure data in the furnace and the load is analyzed to verify the authenticity of the actual steam pressure change, and then the neural network model training method is combined to obtain the health of the boiler system, and finally the error compensation amount is calculated excluding the influence of boiler abnormality, and then the water supply during the operation of the boiler is predicted and compensated. The optimized water supply after compensation can be fed back to the processor, and then the boiler control system is pre-compensated. This improves the generalization ability of the neural network model to a certain extent, so as to quickly adapt to the ever-changing operating conditions and working conditions, and the pre-compensation operation of the boiler control system better realizes the large closed loop of the heat network and boiler load linkage.

[0118] Reference Figure 6 , shows a structural block diagram of a boiler operation parameter optimization system based on machine learning according to an embodiment of the present application, and the system may include:

[0119] An acquisition module is used to acquire the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data in each monitoring time period;

[0120] An error module, used for determining an error compensation amount according to a neural network model, steam pressure data, steam flow data and operating parameters;

[0121] A prediction module, used for determining a predicted boiler feed water amount according to an error compensation amount;

[0122] The optimization module is used to determine the optimization coefficient of the feed water quantity according to the predicted boiler feed water quantity, and optimize the feed water quantity of the operating parameters according to the optimization coefficient to obtain the optimized feed water quantity.

[0123] In the embodiment of the present application, the functions of each module in the system can be found in the corresponding description in the above method and will not be repeated here.

[0124] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0125] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for optimizing boiler operating parameters based on machine learning, characterized in that: The method comprises: In each monitoring period, obtain the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data; Determining an error compensation amount according to the neural network model, the steam pressure data, the steam flow data, and the operating parameters; Determining a predicted boiler feed water amount according to the error compensation amount; Determining an optimization coefficient of the water supply according to the predicted boiler water supply, and optimizing the water supply of the operating parameters according to the optimization coefficient to obtain an optimized water supply; Determining the error compensation amount according to the neural network model, the steam pressure data, the steam flow data and the operating parameters includes: Determining the pressure fluctuation factor in the boiler in each of the monitoring time periods according to the steam pressure data; Determining the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, the steam flow data and the operating parameters; Outputting health status according to the neural network model and the operating parameters; Determining an error compensation amount according to the authenticity value and the healthiness; Determining the predicted boiler feed water volume according to the error compensation amount includes: Determine, according to the operating parameters of each monitoring time period, a water supply amount change curve of each monitoring time period and a relevant parameter change curve corresponding to each relevant parameter related to the water supply amount in each monitoring time period; Determine a prediction curve of boiler feed water volume according to the feed water volume variation curves of each monitoring time period and the related parameter variation curves; The predicted boiler feed water volume is determined according to the error compensation amount and the prediction curve.

2. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: Determining the pressure fluctuation factor in the boiler in each monitoring time period according to the steam pressure data includes: Determining a monitoring fitting curve of the steam pressure data and a variance of the steam pressure data respectively according to the steam pressure data in each of the monitoring time periods; Determine the maximum extreme value and the minimum extreme value in the monitoring fitting curve, and determine the extreme value difference between the maximum extreme value and the minimum extreme value; A first product of the variance and the extreme value difference is determined, and a pressure fluctuation factor in the boiler in each of the monitoring time periods is determined according to the first product and a normalization function.

3. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: Determining the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, the steam flow data and the operating parameter includes: Determining the load in each monitoring time period according to the steam flow data and the operating parameters; determining a first standard deviation according to each of the loads, and determining a second standard deviation according to each of the pressure fluctuation factors; determining a covariance according to each of the loads and each of the pressure fluctuation factors; A second product of the first standard deviation and the second standard deviation is determined, and a authenticity value of the pressure fluctuation in the boiler is determined according to a ratio of the covariance to the second product.

4. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: The determining, according to the authenticity value and the health level, the error compensation amount for the monitoring time period comprises: An error compensation amount for the monitoring time period is determined according to a ratio of the health degree to the authenticity value.

5. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: Determining the prediction curve of boiler feed water volume according to the feed water volume change curves of each monitoring time period and each related parameter change curve comprises: Determine the mean square error between each of the water supply change curves and each of the corresponding related parameter change curves in each of the monitoring time periods, and determine the target mean square error between each of the water supply change curves and each of the related parameter change curves according to the mean square error corresponding to each of the monitoring time periods; Determine the influence weight of each relevant parameter on the water supply according to each target mean square error; By using each of the influence weights, a weighted sum is taken on the feed water quantity variation curves in the latest monitoring time period to determine a prediction curve for the boiler feed water quantity.

6. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: Determining the predicted boiler feed water volume according to the error compensation amount and the prediction curve includes: Using the error compensation amount as the Kalman gain; The prediction curve is smoothed by using the Kalman gain through a Kalman filter to determine the predicted boiler feed water volume for the next monitoring time period in the future.

7. The method for optimizing boiler operating parameters based on machine learning according to claim 1, characterized in that: Determining the optimization coefficient of the water supply according to the predicted boiler water supply, and optimizing the water supply of the operating parameters according to the optimization coefficient to obtain the optimized water supply includes: Determine the heat conversion efficiency of the boiler and the heat demand for the next monitoring period in the future; Determining a predicted heat supply according to the predicted boiler feed water volume and the heat energy conversion efficiency; Determining an optimization coefficient of water supply according to a ratio of the thermal energy demand to the predicted heating amount and a hyperbolic tangent function; The optimized water supply is obtained according to the third product of the water supply of the operating parameters in the latest monitoring time period and the optimization coefficient.

8. A boiler operating parameter optimization system based on machine learning, characterized in that: include: An acquisition module is used to acquire the operating parameters of the boiler during operation, the steam pressure data in the boiler, and the steam flow data in each monitoring time period; An error module, used for determining an error compensation amount according to a neural network model, the steam pressure data, the steam flow data and the operating parameters; Determining the error compensation amount according to the neural network model, the steam pressure data, the steam flow data and the operating parameters includes: Determining the pressure fluctuation factor in the boiler in each of the monitoring time periods according to the steam pressure data; Determining the authenticity value of the pressure fluctuation in the boiler according to the pressure fluctuation factor, the steam flow data and the operating parameters; Outputting health status according to the neural network model and the operating parameters; Determining an error compensation amount according to the authenticity value and the healthiness; A prediction module, used to determine a predicted boiler feed water amount according to the error compensation amount; Determining the predicted boiler feed water volume according to the error compensation amount includes: Determine, according to the operating parameters of each monitoring time period, a water supply amount change curve of each monitoring time period and a relevant parameter change curve corresponding to each relevant parameter related to the water supply amount in each monitoring time period; Determine a prediction curve of boiler feed water volume according to the feed water volume variation curves of each monitoring time period and the related parameter variation curves; Determining a predicted boiler feed water amount according to the error compensation amount and the prediction curve; The optimization module is used to determine the optimization coefficient of the water supply according to the predicted boiler water supply, and optimize the water supply of the operating parameters according to the optimization coefficient to obtain the optimized water supply.

Images