Device and method for selecting optimal boiler combustion model
By selecting and updating the best model in the boiler combustion model, the problems of low combustion efficiency and polluted exhaust in coal thermal power plants are solved, and more efficient combustion and environmentally friendly combustion optimization are achieved.
Patent Information
- Application Number
- CN202010411511.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-07-11
- Filing Date
- 2020-05-15
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2040-05-15
AI Technical Summary
The combustion efficiency of coal thermal power plants is low and produces polluting exhaust, resulting in increased combustion costs.
Design an optimal boiler combustion model selection device to improve combustion efficiency and reduce contaminated exhaust by selecting the best model among multiple boiler combustion models and updating the stored combustion model with the processor. The device includes memory and processors that use data analysis and model selection algorithms to select and update the best combustion model.
It improves the combustion efficiency of boilers, reduces polluting exhaust, and reduces the operating costs of power generation.
Smart Images

Figure CN112215387B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an optimal boiler combustion model selection device and method, and more specifically, the present invention relates to an optimal boiler combustion model selection device and method for selecting the optimal boiler combustion model from multiple boiler combustion models for boiler combustion to update a memory storing the boiler combustion model. Background Art
[0002] The boiler of a coal-fired power plant uses the exothermic reaction generated when coal is burned to heat water and generate steam for power generation. At this time, polluting exhaust gases such as nitrogen oxides are generated. The cost of treating the exhaust gases increases when the amount of polluting exhaust gases generated is large. Furthermore, if the coal combustion is incomplete, the combustion efficiency decreases, resulting in increased power generation / operation costs. Therefore, it is necessary to develop a boiler combustion optimization device that can both improve combustion efficiency and reduce polluting exhaust gases. Summary of the invention
[0003] The present invention aims to solve the above-mentioned problems. The purpose of the present invention is to provide an optimal boiler combustion model selection device and method, which selects the best boiler combustion model from multiple boiler combustion models for boiler combustion to update the stored boiler combustion model.
[0004] In addition to the technical issues of the present invention mentioned above, a person having ordinary knowledge in the field to which the present invention belongs can clearly understand other features and advantages of the present invention from the following description or from its description and explanation.
[0005] In order to achieve the above-mentioned purpose, an optimal boiler combustion model selection device for selecting an optimal boiler combustion model for boiler combustion in an embodiment of the present invention comprises: a memory storing a pre-derived first boiler combustion model and a pre-derived second boiler combustion model; and a processor configured to select an optimal boiler combustion model from the first boiler combustion model and the second boiler combustion model; the processor updates the second boiler combustion model stored in the memory according to the characteristics of the selected optimal boiler combustion model.
[0006] Furthermore, in order to achieve the aforementioned purpose, the best boiler combustion model selection method for selecting the best boiler combustion model for boiler combustion in an embodiment of the present invention may include the following steps: storing a pre-derived first boiler combustion model and a pre-derived second boiler combustion model in a memory; and a processor selecting the best boiler combustion model from the first boiler combustion model and the second boiler combustion model, and updating the second boiler combustion model stored in the memory according to the characteristics of the selected best boiler combustion model.
[0007] The optimal boiler combustion model selection device and method according to the embodiment of the present invention can select the optimal boiler combustion model from multiple boiler combustion models.
[0008] Furthermore, the optimal boiler combustion model selection device and method can automatically update the model stored in the memory with the selected optimal boiler combustion model.
[0009] In addition, more features and advantages of the present invention can be grasped through the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Figure 1 It is a diagram illustrating a boiler combustion optimization device for optimizing boiler combustion according to an embodiment of the present invention.
[0011] Figure 2 It is a diagram showing the structure of the optimal boiler combustion model selection device according to the embodiment of the present invention.
[0012] Figure 3 1 is a diagram showing the updating of the memory unit when the first boiler combustion model is selected as the optimal boiler combustion model according to the embodiment of the present invention.
[0013] Figure 4 1 is a diagram showing the updating of the memory unit when the second boiler combustion model is selected as the optimal boiler combustion model according to the embodiment of the present invention.
[0014] Figure 5 1 is a diagram showing the updating of the memory unit when the third boiler combustion model according to the embodiment of the present invention is selected as the optimal boiler combustion model.
[0015] Figure 6 1 is a diagram showing a method for selecting an optimal boiler combustion model according to an embodiment of the present invention. DETAILED DESCRIPTION
[0016] In order to enable those with general knowledge in the technical field to which the present invention belongs to easily implement the embodiments of the present invention, the following is a detailed description with reference to the accompanying drawings. However, the present invention can be implemented in various forms and is therefore not limited to the embodiments described herein.
[0017] Figure 1 FIG. 1 is a diagram illustrating a boiler combustion optimization system for optimizing boiler combustion according to an embodiment of the present invention.
[0018] See also Figure 1 The boiler combustion optimization system may include a combustion management unit 100, an automatic adjustment management unit (model / controller) 200, a data preprocessing unit 300, a data analysis unit 400, a combustion model design algorithm unit 500, a combustion controller design algorithm unit 600, an optimal model / controller selection unit 700, and a combustion optimization algorithm unit 800.
[0019] The boiler of a coal-fired power plant generates steam for power generation by heating water using the exothermic reaction generated when burning coal. At this time, polluting exhaust gases such as nitrogen oxides are generated, and the cost of treating the exhaust gases increases when the amount of polluting exhaust gases generated is large. Furthermore, if the coal combustion is incomplete, the combustion efficiency decreases, resulting in an increase in power generation / operation costs. Therefore, the following is a boiler combustion optimization system that can both improve combustion efficiency and reduce polluting exhaust gases.
[0020] First, the combustion management unit 100 can perform comprehensive management of combustion optimization. Specifically, the combustion management unit 100 can determine whether to perform combustion optimization based on tag data information measured or calculated in real time. Moreover, when the combustion management unit 100 is used in a power plant, it can determine whether to perform combustion optimization based on DCS values such as power generation output, output change status, whether soot blowers are started, and boiler logic actions.
[0021] Moreover, the automatic adjustment management unit (model / controller) 200 can manage the combustion model and the combustion controller. Specifically, the automatic adjustment management unit (model / controller) 200 can determine whether to adjust the combustion model and the combustion controller based on information such as label data information measured or calculated in real time, whether the combustion optimization is implemented, and whether the model and the controller are learned and adjusted. Here, the automatic adjustment management unit (model / controller) 200 executes the combustion model design algorithm unit 500 and the combustion controller design algorithm unit 600 when it is determined that the combustion model and the combustion controller are adjusted, and does not execute the combustion model design algorithm unit 500 and the combustion controller design algorithm unit 600 when it is determined that the combustion model and the combustion controller are not adjusted.
[0022] Moreover, the data preprocessing unit 300 can perform data preprocessing using the real-time measured or calculated tag data information and the stored tag data information. Specifically, the data preprocessing unit 300 can perform signal processing, normal data processing (based on basic knowledge or based on data) and outlier processing, etc. In this way, the data preprocessing unit 300 can remove noise in the data or remove data that has a bad influence when generating a combustion model or designing a controller.
[0023] Furthermore, the data analysis unit 400 can perform the following sampling: only the important data of the normal state required for the combustion model and controller design can be selected through data pattern analysis. Furthermore, the data analysis unit 400 can select the input variables that have a significant impact on the model output variables for the design of the combustion model through methods such as correlation analysis.
[0024] Furthermore, the combustion model design algorithm unit 500 may be an algorithm for designing a combustion model, which is one of the most important factors in optimizing boiler combustion. The combustion model design algorithm unit 500 can design a combustion model that can predict important combustion variables, i.e., power generation output, combustion state including steam and exhaust temperature, exhaust gas composition, oxygen residual after combustion, etc., based on inputs such as fuel input amount, air input amount, air temperature, water input amount, and air temperature.
[0025] Furthermore, the combustion controller design algorithm unit 600 can be an algorithm for designing a combustion controller, which is one of the most important factors in optimizing boiler combustion. The combustion controller design algorithm unit 600 enables the combustion controller designed using the combustion model to generate an optimal target value for optimal combustion control.
[0026] Furthermore, the optimal model / controller selection unit 700 can select the optimal combustion model and combustion controller from a plurality of previously created combustion models and combustion controllers based on the result of analyzing the current data.
[0027] Furthermore, the combustion optimization algorithm unit 800 can calculate the optimal target value for combustion optimization after inputting the current data into the optimal combustion model and the combustion controller. Furthermore, the combustion optimization algorithm unit 800 calculates the optimal control target value or related auxiliary value using the current DCS target value and the manual.
[0028] The optimal boiler combustion model selection device of the embodiment of the present invention is equivalent to the optimal model / controller selection unit 700, which can select the optimal model from multiple combustion models and update the selected optimal model to the memory.
[0029] Figure 2 It is a diagram showing the structure of the optimal boiler combustion model selection device according to the embodiment of the present invention.
[0030] See also Figure 2 The optimal boiler combustion model selection device 700 according to the embodiment of the present invention may include a first memory unit 710 , a second memory unit 720 , a model generation unit 730 , a model selection unit 740 and a control unit 750 .
[0031] The first memory unit 710 and the second memory unit 720 each refer to a field capable of storing data, which may each refer to a memory device. According to an embodiment, the first memory unit 710 and the second memory unit 720 may be implemented as one memory device or as separate memory devices.
[0032] The model generation unit 730, the model selection unit 740, and the control unit 750 can be implemented by at least one processor. According to an embodiment, the model generation unit 730, the model selection unit 740, and the control unit 750 can be implemented by one processor, but the present invention is not limited thereto, and at least two of the model generation unit 730, the model selection unit 740, and the control unit 750 can be implemented by one processor.
[0033] In the following, the functions and actions of the model generation unit 730, the model selection unit 740, and the control unit 750 described in this specification should be interpreted as referring to the functions and actions performed by the processor that implements it. For example, the functions and actions of the model generation unit 730 refer to the functions and actions performed by the processor that implements the model generation unit 730.
[0034] According to an embodiment, at least one processor may execute a program storing commands that instruct functions and actions of the model generation unit 730 , the model selection unit 740 , and the control unit 750 .
[0035] The first memory unit 710 can store the first boiler combustion model 10. Here, the first boiler combustion model 10 is a combustion model commonly used in existing boiler combustion, and can be a pre-set combustion model. For example, it can be a combustion model whose performance has been verified through long-term experiments and operations. The first boiler combustion model 10 can be pre-stored in the first memory unit 710 in advance (for example, when designing the device), and can be manually (manually) changed or deleted by the administrator, but cannot be automatically (automatically) changed or deleted. For example, access (Access) to the first memory 710 can be restricted. Moreover, multiple first boiler combustion models 10 can be stored in the first memory unit 710.
[0036] The second memory unit 720 can store the second boiler combustion model 20 derived in advance. Here, the second memory unit 720 can store the best boiler combustion model 20 previously selected by learning. Here, the first boiler combustion model 10 stored in the first memory unit 710 can be a model verified by long-term experiments, and the second boiler combustion model 20 stored in the second memory unit 720 can be a model selected by learning. That is, compared with the first boiler combustion model 10, the second boiler combustion model 20 can be the latest derived model. For example, the first boiler combustion model 10 cannot be automatically changed by the best boiler combustion model selection device 700, but the second boiler combustion model 20 can be automatically changed by the best boiler combustion model selection device 700. In addition, the second boiler combustion model 20 can be distinguished by components (for example, carbon components, etc.) or by operating conditions, and the second memory unit 720 can store data sets (sets) of the second boiler combustion models 20 distinguished by components or by operating conditions. The second memory unit 720 can store multiple data sets, and the multiple data sets can each store multiple second boiler combustion models 20.
[0037] Furthermore, the second combustion models 20 stored in the second memory unit 720 may be stored in order according to the order in which they are generated. For example, the second memory unit 720 may have a hierarchical structure, and the second memory unit 720 may store the boiler combustion models in the order of their derivation from the upper level to the lower level. As an example, the first derived second boiler combustion model may be stored in the uppermost level (e.g., the first level) of the second memory unit 720, and the second boiler combustion model derived immediately thereafter may be stored in the second upper level (e.g., the second level) after the uppermost level of the second memory unit 720.
[0038] The model generation unit 730 may output the third boiler combustion model 30 for boiler combustion to the model selection unit 740 described later. Figure 1 The combustion model design algorithm unit 500 generates and outputs the third boiler combustion model 30. Alternatively, the model generation unit 730 can generate the third boiler combustion model 30 by itself and output the generated third boiler combustion model 30.
[0039] Here, the model generation unit 730 can be Figure 1The third boiler combustion model 30 is generated in the same manner as the combustion model design algorithm unit 500. For example, the model generation unit 730 can design a combustion model that can predict important combustion variables, namely, power generation output, combustion state including steam and exhaust temperature, exhaust gas composition, and oxygen residual amount after combustion based on inputs such as fuel input amount, air input amount, air temperature, water input amount, and air temperature.
[0040] The third boiler combustion model 30 can be stored in the third memory unit. The third memory unit can be composed of a volatile memory such as a cache memory. In this case, the third boiler combustion model 30 can be temporarily stored in the third memory unit. In addition, the third boiler combustion model 30 can be stored in the second memory unit 720. In this case, the third boiler combustion model 30 can be deleted as described later.
[0041] The model selection unit 740 can select the best combustion model from among the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30. That is, the best boiler combustion model can be one of the first boiler combustion models 10, one of the second boiler combustion models 20, or the third boiler combustion model 30. Here, in the case where the third boiler combustion model 30 is not generated, the model generation unit 730 can also select the best combustion model from among the first boiler combustion model 10 and the second boiler combustion model 20.
[0042] Specifically, the model selection unit 740 can generate verification data for model verification after analyzing the pattern of the data output by the boiler. Here, the data output by the boiler can be the latest data, which is the data output by the boiler most recently. According to an embodiment of the present invention, the latest data is not limited to the data output most recently, and can also include data before the data output most recently. For example, in the case where the data output most recently is the Nth data, the latest data can represent the Nth, N-1th, N-2th, and N-3th data. Here, the present invention does not limit the number of the latest data, and it can also be pre-set. The model selection unit 740 can generate verification data for model verification through the latest data and data on basic characteristics, thereby selecting a boiler combustion model that can reflect the latest operating data of the power plant (or boiler).
[0043] The model selection unit 740 can compare the output values of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30 with the verification data, and select the boiler combustion model with the smallest error with the verification data as the optimal boiler combustion model according to the comparison result. Here, the verification data may include input data and output data, and the input data of the verification data may be input as the input data of each boiler combustion model 10, 20, 30.
[0044] The model selection unit 740 may represent the error between the output value of each of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30 and the verification data as one value for each model. As an example, the model selection unit 740 may calculate the average value of the square value of the difference between the output value of each model and the verification data and represent it as one error. Furthermore, the model selection unit 740 may calculate the average value of the square value of the difference between the output value of each model and the verification data and calculate the square root of the average value and represent it as one error.
[0045] Furthermore, the model selection unit 740 may calculate the sum of square values of the differences between the output values of each model and the verification data or the sum or average value of the absolute values of the differences between the output values of each model and the verification data and represent the differences as an error.
[0046] The model selection unit 740 may select the boiler combustion model having the smallest error as the optimal boiler combustion model.
[0047] Moreover, a plurality of boiler combustion models can be verified (or evaluated) using a test data set including a test input value / test output value, and the best boiler combustion model can be selected from the plurality of boiler combustion models based on the verification result. That is, the test input value can be input into the plurality of boiler combustion models and the operation result for the test input value can be calculated from each boiler combustion model. Moreover, the operation result calculated by each boiler combustion model can be compared with the test output value and the boiler combustion model can be verified based on the error thereof. Here, the model selection unit 740 can select the boiler combustion model with the smallest error as the best boiler combustion model based on the result of comparing the operation result calculated by each boiler combustion model with the test output value. In addition, the boiler combustion model can be verified and selected by many methods.
[0048] The control unit 750 may update the second memory unit 120 according to the characteristics of the optimal boiler combustion model selected by the model selection unit 740 .
[0049] Specifically, the control unit 750 may update the second memory unit 720 according to which of the first boiler combustion model 10 , the second boiler combustion model 20 , and the third boiler combustion model 30 the optimal boiler combustion model selected by the model selection unit 740 is the same.
[0050] If the best boiler combustion model selected by the model selection unit 740 is the first boiler combustion model 10, the control unit 750 may not update the second memory unit 720. Here, the control unit 750 does not store the third boiler combustion model 30 in the second memory unit 720 but deletes it, so that the capacity of the second memory unit 720 can be maintained at a certain amount. Here, when the best boiler combustion model is selected in a state where the third boiler combustion model 30 is not generated, the action of storing or deleting the third boiler combustion model 30 by the control unit 750 can be omitted.
[0051] Furthermore, since the first boiler combustion model 10 is a model verified after a long period of testing and the third boiler combustion model 30 is a model generated by learning, when the first boiler combustion model 10 is selected as the optimal boiler combustion model, it is not necessary to store the third boiler combustion model 30 that has not been verified. Therefore, by deleting the third boiler combustion model 30 and using the first boiler combustion model 10 stored in the first memory unit 710 as the optimal boiler combustion model, the boiler combustion can be optimized.
[0052] Furthermore, when the optimal boiler combustion model selected by the model selection unit 740 is the second boiler combustion model 20, the control unit 750 can store the optimal boiler combustion model as the most recently derived boiler combustion model in the second memory unit 720. That is, when the optimal boiler combustion model is the second boiler combustion model 20, the control unit 750 can configure (or store) the optimal boiler combustion model in the lowest level of the second memory unit 720, thereby improving the performance of the second boiler combustion model 20 stored in the second memory unit 720. Here, the control unit 750 can delete the third boiler combustion model 30, and by deleting the third boiler combustion model 30, the capacity of the second memory unit 720 can be maintained at a certain amount. Here, when the optimal boiler combustion model is selected in a state where the third boiler combustion model 30 is not generated, the operation of the control unit 750 to store or delete the third boiler combustion model 30 can be omitted.
[0053] Furthermore, when the best boiler combustion model selected by the model selection unit 740 is the third boiler combustion model 30, the control unit 750 may delete the boiler combustion model derived first among the second boiler combustion models 20 stored in the second memory unit 720 and store the best boiler combustion model as the most recently derived boiler combustion model in the second memory unit 720. That is, when the best boiler combustion model is the third boiler combustion model 30, the control unit 750 may delete the second boiler combustion model derived first among the second boiler combustion models 20 stored in the second memory unit 720 and store the third boiler combustion model 30 selected as the best boiler combustion model in the second memory unit 720. That is, the second boiler combustion model derived first may be updated to the third boiler combustion model 30 selected most recently. At this time, the control unit 750 may store the third boiler combustion model 30 in the lowest level of the second memory unit 720.
[0054] In this way, the capacity of the second memory unit 720 can be maintained at a certain amount. According to an embodiment of the present invention, the first derived second boiler combustion model 20 is deleted first, so the newly derived optimal boiler combustion model is stored as the latest derived boiler combustion model, thereby preventing it from being deleted first. That is, the boiler combustion model is selected by comparing with the latest data of the boiler, so it can be the model most similar to the latest state of the boiler. Therefore, the selected optimal boiler combustion model is stored in the lowest level of the second memory unit 720, and then the newly selected optimal boiler combustion model can be stored in the lowest level of the second memory unit 720. Repeating this step can store the model similar to the latest data of the boiler in the lower level of the second memory unit 720. On the contrary, the second boiler combustion model 20 stored in the upper level of the second memory unit 720 is not selected as the optimal boiler combustion model and cannot be updated to the lower level of the second memory unit 720. This can indicate that it is not similar to the latest state of the boiler. Therefore, when the third boiler combustion model 30 is selected as the optimal combustion model, the second boiler combustion model at the top level of the second memory unit 720 is deleted in order to store the third boiler combustion model 30 in the second memory unit 720 .
[0055] Furthermore, the second boiler combustion model 20 stored in the second memory unit 720 may not be deleted according to the capacity of the second memory unit 720. For example, when the capacity of the second memory unit 720 is lower than the allowable value and the third boiler combustion model 30 is selected as the optimal boiler combustion model, the control unit 750 can store the third boiler combustion model 30 without deleting the second boiler combustion model 20. At this time, the third boiler combustion model 30 can be stored at the lowest end of the second memory unit 720 as the most recently derived boiler combustion model.
[0056] Figure 3 1 is a diagram showing the updating of the memory unit when the first boiler combustion model is selected as the optimal boiler combustion model according to the embodiment of the present invention.
[0057] See also Figure 3 The first memory unit 710 can store a first boiler combustion model (Seed Model1-SeedModelN) 10 , and the second memory unit 720 can store a second boiler combustion model (Fruit Model1-FruitModelN) 20 .
[0058] The model selection unit 740 can select the best boiler combustion model from the first boiler combustion model 10 stored in the first memory unit 710, the second boiler combustion model 20 stored in the second memory unit 720, and the third boiler combustion model 30 output by the model generation unit 730. Here, the model selection unit 740 compares the latest data extracted from the boiler and the output values of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30 with the verification data, and according to the comparison result, the boiler combustion model with the smallest error with the verification data can be selected as the best boiler combustion model.
[0059] When the best boiler combustion model 12 selected by the model selection unit 740 is the first boiler combustion model 10 , the control unit 750 may not update the first memory unit 710 and the second memory unit 720 .
[0060] Figure 4 1 is a diagram showing the updating of the memory unit when the second boiler combustion model is selected as the optimal boiler combustion model according to the embodiment of the present invention.
[0061] See also Figure 4 The first memory unit 710 can store a first boiler combustion model (SM1-SMN) 10, and the second memory unit 720 can store a second boiler combustion model (FM1-FMN) 20.
[0062] The model selection unit 740 can select the best boiler combustion model from the first boiler combustion model 10 stored in the first memory unit 710, the second boiler combustion model 20 stored in the second memory unit 720, and the third boiler combustion model 30 output by the model generation unit 730. Here, the model selection unit 740 compares the latest data extracted from the boiler and the output values of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30 with the verification data, and according to the comparison result, the boiler combustion model with the smallest error with the verification data can be selected as the best boiler combustion model.
[0063] In the case where the best boiler combustion model 22 selected by the model selection unit 740 is the second boiler combustion model 20, the control unit 750 can store the best boiler combustion model as the most recently derived boiler combustion model in the second memory unit. Specifically, the second memory unit 720 has a hierarchical structure, and the earlier the second boiler combustion model is derived, the higher the hierarchy can be stored. In the case where the best boiler combustion model 22 selected is the second boiler combustion model 20, the control unit 750 can store the second boiler combustion model 22 selected as the best boiler combustion model in the lowest hierarchy of the second memory unit 720. At this time, the control unit 750 can delete the third boiler combustion model 30 output by the model generation unit 730.
[0064] Figure 5 1 is a diagram showing the updating of the memory unit when the third boiler combustion model according to the embodiment of the present invention is selected as the optimal boiler combustion model.
[0065] See also Figure 5 The first memory unit 710 can store a first boiler combustion model (SM1-SMN) 10, and the second memory unit 720 can store a second boiler combustion model (FM1-FMN) 20.
[0066] The model selection unit 740 can select the best boiler combustion model from the first boiler combustion model 10 stored in the first memory unit 710, the second boiler combustion model 20 stored in the second memory unit 720, and the third boiler combustion model 30 output by the model generation unit 730. Here, the model selection unit 740 compares the latest data extracted from the boiler and the output values of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30 with the verification data, and according to the comparison result, the boiler combustion model with the smallest error with the verification data can be selected as the best boiler combustion model.
[0067] When the optimal boiler combustion model selected by the model selection unit 740 is the third boiler combustion model 30 , the control unit 750 replaces the second boiler combustion model derived first among the second boiler combustion models with the optimal boiler combustion model, thereby updating the second memory unit 720 .
[0068] Here, the second memory unit 720 has a hierarchical structure, and the first derived second boiler combustion model can be stored in the upper layer. In the case where the best boiler combustion model is the third boiler combustion model 30, the control unit 750 can delete the first derived second boiler combustion model among the second boiler combustion models stored in the second memory unit 720 and store the third boiler combustion model 30 selected as the best boiler combustion model in the lowest layer of the second memory unit 720.
[0069] Figure 6 1 is a diagram showing a method for selecting an optimal boiler combustion model according to an embodiment of the present invention.
[0070] See also Figure 6 , the first memory unit can store the first boiler combustion model (S10). The first memory unit 710 can store the first boiler combustion model 10 derived in advance, and the first boiler combustion model 10 can be a combustion model previously commonly used for boiler combustion. Here, the first boiler combustion model 10 has a hierarchical structure, and the first boiler combustion model derived earlier can be stored in the upper level.
[0071] The second memory unit can store the second boiler combustion model (S20). The second memory unit 720 can store the pre-derived second boiler combustion model 20, and the second boiler combustion model 20 can be a combustion model pre-selected as the optimal boiler combustion model. Here, the second boiler combustion model 20 has a hierarchical structure, and the first derived second boiler combustion model can be stored in the upper level.
[0072] The model generation unit may output a third boiler combustion model (S30). Here, the third boiler combustion model is not generated every time, so the action of step 30 may be omitted.
[0073] The model selection unit may select an optimal boiler combustion model among the first boiler combustion model, the second boiler combustion model, and the third boiler combustion model ( S40 ).
[0074] The model selection unit 740 can compare the latest data extracted from the boiler and the verification data reflecting the basic characteristics with the output values of the first boiler combustion model 10, the second boiler combustion model 20, and the third boiler combustion model 30. Based on the comparison result, the model selection unit 740 can select the boiler combustion model with the smallest error with the verification data as the optimal boiler combustion model. Here, in the case where the operation of step S30 is omitted, the model selection unit 740 can select the optimal boiler combustion model from the first boiler combustion model 10 and the second boiler combustion model 20.
[0075] The control unit may update the second memory unit according to the characteristics of the optimal boiler combustion model selected by the model selection unit (S50).
[0076] Specifically, the control unit 750 may update the second boiler combustion model stored in the second memory unit 720 according to whether the boiler combustion model selected by the model selection unit 740 is the first boiler combustion model 10 , the second boiler combustion model 20 , or the third boiler combustion model 30 .
[0077] That is, in a case where the optimal boiler combustion model selected by the model selection unit 740 is the first boiler combustion model 10 , the control unit 750 may delete the third boiler combustion model 30 and not update the second memory unit 720 .
[0078] Furthermore, when the optimal boiler combustion model selected by the model selection unit 740 is the second boiler combustion model 20, the control unit 750 can update the optimal boiler combustion model as the most recently derived boiler combustion model to the second memory unit 720. That is, when the optimal boiler combustion model is the second boiler combustion model 20, the control unit 750 can store the second boiler combustion model selected as the optimal boiler combustion model in the lowest level of the second memory unit 720 to update the second memory unit 720, and can delete the third boiler combustion model 30.
[0079] Furthermore, when the best boiler combustion model selected by the model selection unit 740 is the third boiler combustion model 30, the control unit 750 may delete the first derived boiler combustion model in the second boiler combustion model 20 stored in the second memory unit 720 and store the best boiler combustion model as the latest derived boiler combustion model in the second memory unit 720. That is, when the best boiler combustion model is the third boiler combustion model 30, the control unit 750 may delete the first derived second boiler combustion model in the second boiler combustion model 20 stored in the second memory unit 720 and store the third boiler combustion model 30 selected as the best boiler combustion model in the second memory unit 720. At this time, the control unit 750 may store the third boiler combustion model 30 in the lowest level of the second memory unit 720.
[0080] As described above, according to an embodiment of the present invention, an optimal boiler combustion model selection device and method can be implemented, which can update the memory unit storing the boiler combustion model after selecting the optimal boiler combustion model from multiple boiler combustion models for boiler combustion.
[0081] Without changing the technical concept and essential features of the present invention, a person skilled in the art of the present invention may implement the present invention into other specific embodiments. Therefore, the embodiments described above should be understood as illustrative rather than limiting in all aspects. The above detailed description cannot be used to limit the present invention, and the scope of the present invention should be defined by the claims. The scope of the present invention should be interpreted as including all modifications and variations derived from the meaning, scope and equivalent concepts of the claims.
[0082]
Explanation of symbols
[0083] 700: Optimal boiler combustion model selection device
[0084] 710: First memory unit
[0085] 720: Second memory unit
[0086] 730: Model generation unit
[0087] 740: Model selection unit
[0088] 750: Control unit
Claims
1. An optimal boiler combustion model selection device, used to select the optimal boiler combustion model for boiler combustion, characterized in that: include: A memory having a hierarchical structure and comprising: a first memory unit, wherein the first memory unit is configured to store a pre-derived first boiler combustion model, wherein the first boiler combustion model is a model verified by experiments, and the first boiler combustion model cannot be automatically changed; and a second memory unit configured to store a pre-derived second boiler combustion model, the second boiler combustion model being a model selected through learning and capable of being automatically changed; and a processor configured to generate a third boiler combustion model for the boiler combustion, and select an optimal boiler combustion model from among the first boiler combustion model, the second boiler combustion model, and the third boiler combustion model, wherein the third boiler combustion model predicts combustion variables based on inputs of a fuel input amount, an air input amount, an air temperature, a water input amount, and an air temperature, wherein the combustion variables include power generation output, a combustion state including temperatures of steam and exhaust gas, exhaust gas composition, and an oxygen residual amount after combustion; The processor updates the second memory unit according to the characteristics of the selected optimal boiler combustion model, wherein the processor updates the second memory unit according to which of the selected optimal boiler combustion model and the first boiler combustion model, the second boiler combustion model and the third boiler combustion model is the same, and When the selected optimal boiler combustion model is the third boiler combustion model, the processor replaces the first derived boiler combustion model among the second boiler combustion models stored in the second memory unit with the optimal boiler combustion model to update the second memory unit.
2. The optimal boiler combustion model selection device according to claim 1, characterized in that: The processor generates verification data for model verification using latest data measured in the boiler and data on basic characteristics.
3. The optimal boiler combustion model selection device according to claim 2, characterized in that: The processor compares verification data generated for model verification with output values of the first boiler combustion model, the second boiler combustion model, and the third boiler combustion model, and selects the boiler combustion model having the smallest error with the latest data as the optimal boiler combustion model.
4. The optimal boiler combustion model selection device according to claim 1, characterized in that: The processor calculates the error between the output value of each model and the verification data using at least one of the following calculation formulas: A first calculation formula that calculates the average of the squared values of the differences between the output values of each model and the validation data and expresses them as an error; A second calculation formula for calculating an average value of the squared value of the difference between the output value of each model and the verification data, and calculating the square root of the average value to express it as an error; A third calculation formula in which the sum of the squared values of the differences between the output values of each model and the validation data is calculated and expressed as an error; A fourth calculation formula for calculating the sum of the absolute values of the differences between the output values of each model and the verification data and expressing the sum as an error; and A fifth calculation formula is used to calculate the average value of the absolute value of the difference between the output value of each model and the verification data and express it as one error.
5. The optimal boiler combustion model selection device according to claim 1, characterized in that: In a case where the selected optimal boiler combustion model is the first boiler combustion model, the processor does not update the second memory unit.
6. The optimal boiler combustion model selection device according to claim 1, characterized in that: In a case where the selected optimal boiler combustion model is the second boiler combustion model, the processor stores the optimal boiler combustion model as the most recently derived boiler combustion model in the second memory unit.
7. The optimal boiler combustion model selection device according to claim 1, characterized in that: When the optimal boiler combustion model is the third boiler combustion model, the processor deletes the first derived second boiler combustion model among the second boiler combustion models stored in the second memory unit and stores the optimal boiler combustion model as the latest derived boiler combustion model in the second memory unit.
8. The optimal boiler combustion model selection device according to claim 1, characterized in that: The second memory unit has a hierarchical structure and includes a memory space for storing the boiler combustion models in order from a higher level to a lower level in the order in which they were derived.
9. A method for selecting an optimal boiler combustion model, for selecting an optimal boiler combustion model for boiler combustion, characterized in that: The following steps are involved: storing a pre-derived first boiler combustion model in a first memory unit of a memory, wherein the first boiler combustion model is a model verified by experiments and the first boiler combustion model cannot be automatically changed, wherein the memory has a hierarchical structure; storing a pre-derived second boiler combustion model in a second memory unit of the memory, wherein the second boiler combustion model is a model selected through learning and the second boiler combustion model can be automatically changed; The processor generates a third boiler combustion model for the boiler combustion, and selects the best boiler combustion model from among the first boiler combustion model, the second boiler combustion model and the third boiler combustion model, wherein the third boiler combustion model predicts combustion variables based on inputs of fuel input amount, air input amount, air temperature, water input amount and air temperature, and the combustion variables include power generation output, combustion state including steam and exhaust temperature, exhaust gas composition, and oxygen residual after combustion; and The processor updates the second memory unit according to the characteristics of the selected optimal boiler combustion model, wherein, in the step of updating the second memory unit, the processor updates the second boiler combustion model stored in the second memory unit according to which of the selected optimal boiler combustion model and the first boiler combustion model, the second boiler combustion model and the third boiler combustion model is the same; When the selected optimal boiler combustion model is the third boiler combustion model, the processor replaces the first derived boiler combustion model among the second boiler combustion models stored in the second memory unit with the optimal boiler combustion model to update the second memory unit.
10. The optimal boiler combustion model selection method according to claim 9, characterized in that: In the step of selecting the best boiler combustion model, The processor generates verification data for model verification using latest data measured in the boiler and data on basic characteristics.
11. The optimal boiler combustion model selection method according to claim 9, characterized in that: In the step of selecting the best boiler combustion model, The processor compares verification data generated for model verification with output values of the first boiler combustion model, the second boiler combustion model, and the third boiler combustion model, and selects the boiler combustion model having the smallest error with the latest data as the optimal boiler combustion model.
12. The optimal boiler combustion model selection method according to claim 9, characterized in that: In the step of selecting the optimal boiler combustion model, the error between the output value of each model and the verification data is calculated using at least one of the following calculation formulas: A first calculation formula that calculates the average of the squared values of the differences between the output values of each model and the validation data and expresses them as an error; A second calculation formula for calculating an average value of the squared value of the difference between the output value of each model and the verification data, and calculating the square root of the average value to express it as an error; A third calculation formula in which the sum of the squared values of the differences between the output values of each model and the validation data is calculated and expressed as an error; A fourth calculation formula for calculating the sum of the absolute values of the differences between the output values of each model and the verification data and expressing the sum as an error; and A fifth calculation formula is used to calculate the average value of the absolute value of the difference between the output value of each model and the verification data and express it as one error.
13. The optimal boiler combustion model selection method according to claim 9, characterized in that: In the updating step of the second memory unit, In a case where the selected optimal boiler combustion model is the first boiler combustion model, the processor does not update the second memory unit.
14. The optimal boiler combustion model selection method according to claim 9, characterized in that: In the updating step of the second memory unit, When the optimal boiler combustion model is the second boiler combustion model, the processor stores the optimal boiler combustion model as the most recently derived boiler combustion model in the second memory unit.
Citation Information
Patent Citations
Combustion optimization method and system thereof
CN105444201A
Power station boiler combustion modeling and optimization method facing mass high-dimensional data
CN108038306A
Replacing operation method and device after model update and business server
CN108573355A
A model iteration updating method and device
CN109902827A
HIERARCHICAL STORAGE FOR LSM-BASED NoSQL STORES
US20150058568A1
