Ash deposition rate prediction method, ash blowing control method, boiler system and related device

Abstract

The invention discloses an ash deposition rate prediction method, an ash blowing control method, a boiler system and a related device, and relates to the field of boilers, and the method comprises the steps: calling a first neural network and a second neural network which are jointly trained in advance, and predicting an ash deposition rate value at a designated position at a designated ash deposition time based on the actual operation data of a coal-fired boiler. When the combined model formed by the first neural network and the second neural network is trained, the wall surface ash accumulation distribution trend information and the real ash accumulation information of the coal-fired boiler are fused, so that the combined model trained based on the multi-source fusion data has relatively high prediction precision and generalization performance. On the basis, the ash deposition rate value of any position on the wall surface of the water-cooled wall at any ash deposition time can be accurately predicted by calling the joint model, and on the basis, the ash deposition amount of each position on the wall surface of the water-cooled wall is determined through accumulation, so that the purpose of accurately predicting the overall ash deposition distribution of the wall surface of the water-cooled wall is achieved. And a basis is provided for realizing a precise soot blowing task.

Description

Technical Field

[0001] The present application relates to the technical field of boilers, and in particular, to a method for predicting ash fouling rate, a sootblowing control method, a boiler system, and related devices. Background Art

[0002] Ash fouling specifically refers to the behavior that fly ash generated by coal combustion adheres to the boiler heating surface to form an ash layer. The ash fouling problem will seriously affect the heat transfer efficiency during the operation of coal-fired boilers, threaten the safe operation of boilers, and thus become a pain point problem for coal-fired boilers.

[0003] By controlling multiple furnace sootblowers to perform timed and uniform sootblowing, the ash fouling problem on the boiler heating surface can be solved to a certain extent. Limited by factors such as the boiler structure, the ash fouling distribution on the heating surface is extremely uneven. The method of timed and uniform sootblowing will lead to under-sootblowing in severely fouled areas, making it difficult to effectively improve the boiler heat transfer efficiency; it will also lead to over-sootblowing in slightly fouled areas, increasing the risk of tube wall damage and reducing the service life of the heating surface, which is not conducive to the safe operation of coal-fired boilers.

[0004] Therefore, how to know the ash fouling conditions at each position on the water wall surface to control the sootblower for precise purging has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, the present application provides a method for predicting ash fouling rate, a sootblowing control method, a boiler system, and related devices to determine the ash fouling conditions at each position on the water wall surface, so as to achieve the precise control task of the sootblower.

[0006] The specific solutions are as follows:

[0007] The first aspect of the present application provides a method for predicting ash fouling rate, including:

[0008] Obtain the actual operation data of the coal-fired boiler;

[0009] Call the pre-jointly trained first neural network and second neural network, input the actual operation data into the called first neural network, and input the first data of the wall ash fouling rate output by the first neural network, the actual operation data, and the actual ash fouling time at a specified position into the called second neural network; the specified position is the position on the water wall of the coal-fired boiler corresponding to the sootblowing range of a preset sootblower, and the preset sootblower is one of the multiple sootblowers arranged on the water wall;

[0010] Obtain the ash fouling rate value at the specified position from the second data of the wall ash fouling rate output by the second neural network as the ash fouling rate prediction value of the actual ash fouling time at the specified position;

[0011] Among them, the training database of the first neural network includes the operation data of the coal-fired boiler labeled with simulation calculation data, and the simulation calculation data includes the ash fouling rate values at various positions on the water wall calculated based on computational fluid dynamics; the training database of the second neural network includes the operation data of the coal-fired boiler labeled with measurement calculation data and the ash fouling time at preset measurement positions on the water wall, and the measurement calculation data includes the ash fouling rate values at the preset measurement positions calculated based on the change rate of measurement values, and the measurement values include the heat flux density measurement values or temperature measurement values at the preset measurement positions.

[0012] The second aspect of the present application provides a sootblowing control method, which is applied to a coal-fired boiler. A plurality of sootblowers are arranged on the water wall of the coal-fired boiler, and the sootblowing ranges of different sootblowers are different; the method includes:

[0013] Determine a preset sootblower, and use the positions on the water wall corresponding to the sootblowing range of the preset sootblower as designated positions; the preset sootblower is one of the plurality of sootblowers;

[0014] After the last sootblowing in the sootblowing range of the preset sootblower, calculate the current cumulative ash fouling time as the actual ash fouling time;

[0015] Obtain the ash fouling rate prediction value of the actual ash fouling time at the designated position, and the ash fouling rate prediction value is determined by using the ash fouling rate prediction method described in the first aspect above;

[0016] Calculate the ash fouling amount in the sootblowing range of the preset sootblower according to the actual ash fouling time and the ash fouling rate prediction value;

[0017] When the ash fouling amount exceeds a preset ash fouling amount threshold, control the preset sootblower to perform a sootblowing operation for a preset sootblowing duration to remove the ash fouling in the sootblowing range of the preset sootblower;

[0018] When the ash fouling amount does not exceed the ash fouling amount threshold, return to execute the step of calculating the current cumulative ash fouling time.

[0019] The third aspect of the present application provides an ash fouling rate prediction device, including:

[0020] A data acquisition unit for acquiring the actual operation data of the coal-fired boiler;

[0021] A rate prediction unit, configured to call a pre-jointly trained first neural network and a second neural network, input the actual operation data into the called first neural network, and input the first data of the fouling rate of the wall area output by the first neural network, the actual operation data, and the actual fouling time at a specified position into the called second neural network; the specified position is the position on the water-cooled wall of the coal-fired boiler corresponding to the soot blowing range of a preset soot blower, and the preset soot blower is one of a plurality of soot blowers arranged on the water-cooled wall; wherein, the training database of the first neural network includes the operation data of the coal-fired boiler marked with simulated calculation data, and the simulated calculation data includes the fouling rate values at various positions on the water-cooled wall calculated based on computational fluid dynamics; the training database of the second neural network includes the operation data of the coal-fired boiler marked with measured calculation data and the fouling time at a preset measurement position on the water-cooled wall, and the measured calculation data includes the fouling rate values at the preset measurement position calculated based on the change rate of the measured values, and the measured values include the heat flux density measured value or the temperature measured value at the preset measurement position;

[0022] The rate prediction unit is further configured to obtain the fouling rate value at the specified position from the second data of the fouling rate of the wall area output by the second neural network as the fouling rate prediction value of the actual fouling time at the specified position.

[0023] A fourth aspect of the present application provides a soot blowing control device applied to a coal-fired boiler. A plurality of soot blowers are arranged on the water-cooled wall of the coal-fired boiler, and the soot blowing ranges of different soot blowers are different; the device includes:

[0024] A control object determination unit, configured to determine a preset soot blower and use the position on the water-cooled wall corresponding to the soot blowing range of the preset soot blower as the specified position; the preset soot blower is one of the plurality of soot blowers;

[0025] A fouling monitoring unit, configured to calculate the current cumulative fouling time as the actual fouling time after the last soot cleaning in the soot blowing range of the preset soot blower; obtain the fouling rate prediction value of the actual fouling time at the specified position, and the fouling rate prediction value is determined by using the fouling rate prediction method described in the first aspect above; calculate the fouling amount in the soot blowing range of the preset soot blower according to the actual fouling time and the fouling rate prediction value; and return to execute the step of calculating the current cumulative fouling time when the fouling amount does not exceed a preset fouling amount threshold;

[0026] A soot blowing control unit, configured to control the preset soot blower to perform a soot blowing operation for a preset soot blowing duration to remove the fouling in the soot blowing range of the preset soot blower when the fouling amount exceeds the fouling amount threshold.

[0027] The fifth aspect of the present application provides a boiler system, including: a coal-fired boiler, a soot blower controller, and a boiler control system of the coal-fired boiler;

[0028] Wherein, a plurality of soot blowers are arranged on the water-cooled wall of the coal-fired boiler, and the soot blowing ranges of different soot blowers are different; the boiler control system is communicatively connected to the plurality of soot blowers;

[0029] The soot blower controller is communicatively connected to the boiler control system, and the soot blower controller is configured to:

[0030] Take the position on the water-cooled wall corresponding to the soot blowing range of a preset soot blower as a designated position; the preset soot blower is one of the plurality of soot blowers;

[0031] After the soot on the soot blowing range of the preset soot blower is cleared last time, calculate the current accumulated soot time as the actual soot accumulation time;

[0032] Obtain a predicted soot accumulation rate value of the actual soot accumulation time at the designated position, and the predicted soot accumulation rate value is determined by using the soot accumulation rate prediction method described in the first aspect above;

[0033] Calculate the amount of accumulated soot within the soot blowing range of the preset soot blower according to the actual soot accumulation time and the predicted soot accumulation rate value;

[0034] When the amount of accumulated soot exceeds a preset accumulated soot amount threshold, control the preset soot blower to perform a soot blowing operation for a preset soot blowing duration to remove the accumulated soot within the soot blowing range of the preset soot blower; wherein, controlling the preset soot blower to perform a soot blowing operation for a preset soot blowing duration includes: outputting a control instruction to the boiler control system, so that the boiler control system outputs a start instruction to the preset soot blower and outputs a stop instruction to the preset soot blower after the preset soot blowing duration;

[0035] When the amount of accumulated soot does not exceed the accumulated soot amount threshold, return to execute the step of calculating the current accumulated soot time.

[0036] With the above technical solution, the present application calls the pre-jointly trained first neural network and second neural network to predict the ash deposition rate value at a specified ash deposition time at a specified position based on the actual operation data of the coal-fired boiler. Since the training database of the first neural network consists of the operation data of the coal-fired boiler marked with simulated calculation data, and the simulated calculation data includes the ash deposition rate values at various positions on the water wall calculated based on computational fluid dynamics, where there is detailed wall area ash deposition rate distribution information. Based on this, by inputting the data output by the first neural network into the second neural network, it provides ash deposition distribution trend information for subsequent ash deposition rate prediction; on this basis, the training database of the second neural network consists of the operation data of the coal-fired boiler marked with measurement calculation data and the ash deposition time at preset measurement positions on the water wall. The measurement calculation data includes the ash deposition rate values at the preset measurement positions calculated based on the change rate of the measurement values, which can reflect the actual ash deposition situation of the coal-fired boiler over time. That is to say, when the joint model composed of the first neural network and the second neural network is trained, it integrates the ash deposition distribution trend information on the water wall and the real ash deposition information of the coal-fired boiler, so that the joint model trained based on the integrated data not only has high prediction accuracy but also has high generalization performance. Based on this, the present application calls the above joint model to accurately predict the ash deposition rate value at any position on the water wall at any ash deposition time. On this basis, by cumulatively determining the ash deposition amount at each position on the water wall, the purpose of accurately predicting the overall ash deposition distribution on the water wall is achieved, providing a basis for realizing the accurate sootblowing task. In addition, the present application is implemented based on the pre-jointly trained first neural network and second neural network. The trained neural network model has a high response ability during prediction, thereby ensuring the real-time performance of the ash deposition rate prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of illustrating the preferred embodiments and are not considered to be a limitation of the present application. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0038] Figure 1 It is a schematic flow chart of a method for predicting ash deposition rate provided by the present application;

[0039] Figure 2 It shows a schematic structural diagram of a coal-fired boiler;

[0040] Figure 3 It exemplifies a schematic diagram of the ash deposition rate distribution of a coal-fired boiler under a load of 244 MW;

[0041] Figure 4 It shows a hardware structure block diagram of an electronic device;

[0042] Figure 5 It is a schematic flowchart of a soot blowing control method shown according to an embodiment of the present application;

[0043] Figure 6 It exemplifies a schematic diagram of the communication connection between the soot blowing controller and the boiler control system. Specific embodiments

[0044] The embodiments of the present application will be described below in conjunction with the accompanying drawings in the embodiments of the present application. The terms used in the embodiments of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application. Those of ordinary skill in the art will know that with the development of technology and the emergence of new scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0045] The applicant of this case has found through research that due to the high-temperature flow field in the furnace and the boiler structure during boiler operation, it is difficult to directly determine the ash accumulation situation in the boiler through observation. For this reason, the present application provides an ash accumulation rate prediction scheme. By using a neural network model trained with training data that combines the ash accumulation distribution trend information on the water-cooled wall and the actual ash accumulation information of the coal-fired boiler, the ash accumulation rate on the water-cooled wall is predicted to determine the ash accumulation situation on the water-cooled wall of the boiler, providing a necessary basis for realizing the precise control of the furnace soot blower. On the basis of the above ash accumulation rate prediction scheme, the present application also provides a soot blowing control scheme and a boiler system to obtain the ash accumulation situation of the boiler, especially the ash accumulation situation within the soot blowing range of the furnace soot blower, and control the furnace soot blower accordingly to achieve the precise soot blowing task.

[0046] Figure 1 It is a schematic flowchart of an ash accumulation rate prediction method shown according to an embodiment of the present application. This method can be applied to a coal-fired boiler. Combining Figure 1 As shown, this method may include the following steps:

[0047] Step S101, obtain the actual operation data of the coal-fired boiler.

[0048] Step S102, call the first neural network and the second neural network that are jointly trained in advance, input the actual operation data into the called first neural network, and input the first data of the wall ash accumulation rate output by the first neural network, the actual operation data, and the actual ash accumulation time at a specified position into the called second neural network.

[0049] Among them, the output of the first neural network is the input of the second neural network. The first data of the wall dust accumulation rate output by the first neural network (referred to as the first data for short) and the second data of the wall dust accumulation rate output by the second neural network (referred to as the second data for short) may include: the dust accumulation rate values ​​at various positions on the wall surface of the water-cooled wall. The difference is that when the second neural network obtains the second data, it takes into account the characteristics of the dust accumulation rate changing with the dust accumulation time, so that compared with the first data, the second data can better reflect the actual dust accumulation situation.

[0050] The designated position is a position on the water-cooled wall of the coal-fired boiler corresponding to the soot blowing range of the preset soot blower, and the preset soot blower is one of the multiple soot blowers arranged on the water-cooled wall; the actual ash accumulation time at the designated position may refer to the accumulated ash accumulation time after the last soot cleaning at the designated position; exemplarily, if the coal-fired boiler is always in operation after the last soot blowing by the preset soot blower, the time difference between the current moment and the end moment of the last soot blowing by the preset soot blower may be calculated as the actual ash accumulation time at the designated position, so as to determine the ash accumulation situation at the designated position at the current moment.

[0051] Next, the training data of the first neural network and the second neural network are explained.

[0052] The training database of the first neural network may include the operation data of the coal-fired boiler annotated with simulation calculation data, and the simulation calculation data may include the ash accumulation rate values ​​at various positions on the water-cooled wall calculated based on computational fluid dynamics (CFD), that is, the simulation calculation data is the theoretical data of the wall ash accumulation rate calculated based on the structure and mechanism of the coal-fired boiler by numerical simulation calculation, which can reflect the wall ash accumulation rate of the boiler system under an ideal steady state operating according to a certain set of operation data, and has detailed wall ash accumulation rate distribution information. In other words, by means of the pre-trained first neural network, the mapping relationship between the boiler operation parameters and the ash accumulation rate values ​​at various coordinate positions on the water-cooled wall surface can be obtained. By inputting the data output by the first neural network into the second neural network, the ash accumulation distribution trend information can be provided for the subsequent ash accumulation rate prediction, making up for the deficiency of small amount of measured data.

[0053] The training database of the second neural network may include the operation data of the coal-fired boiler labeled with measurement calculation data and the ash deposition time at the preset measurement positions on the water wall. The measurement calculation data includes the ash deposition rate value at the preset measurement position that can be calculated based on the change rate of the measurement value. Since the change rates of the measurement values calculated based on the measurement values collected at different times are different, and different collection times of the measurement values correspond to different ash deposition times, the measurement calculation data can reflect the actual ash deposition situation of the coal-fired boiler changing with time. The measurement value may include the heat flux density measurement value or the temperature measurement value at the preset measurement position. Among them, based on the wall heat transfer model and the ash deposition thermal resistance model, the change rate of the heat flux density (i.e., the heat flow per unit area) or the temperature can be converted into the ash deposition rate. The above models can be expressed as Q = R×ΔT. In the formula, Q represents the heat flow, ΔT represents the heat transfer temperature difference, and R represents the thermal resistance. R = δ÷λ. In the formula, δ represents the ash deposition thickness, and λ represents the thermal conductivity. It should be noted that assuming that the temperature inside the boiler remains unchanged, the thicker the ash deposition, the greater the thermal resistance, the lower the temperature on the outside of the boiler, and the smaller the heat flux density. Based on this, the actual ash deposition rate can be deduced by measuring the change trend of the temperature on the outer wall of the boiler.

[0054] Since the first neural network and the second neural network are jointly trained, the first neural network and the second neural network can form a joint model. For each deep neural network in the joint model, the network can be composed of an input layer, an output layer, and hidden layers. Each layer can include several neurons. The weights and biases stored in the neurons of the output layer and the hidden layers of the network jointly implement the mapping relationship between the input data and the output data of the network. And, the output layer of the first neural network is connected to the input layer of the second neural network.

[0055] Optionally, the mean square error loss function can be used to optimize the model parameters during the training process of the joint model. The above loss function can be expressed as: ; where , , in the formula represents the loss value of the joint model, represents the loss value of the first neural network, represents the loss value of the second neural network, represents the number of label data of the first neural network, is the number of label data of the second neural network, is the i-th sample label of the first neural network, is the j-th sample label of the second neural network, represents the output data of the first neural network corresponding to the sample label of the first neural network, represents the sample label of the second neural network The output data of the corresponding second neural network.

[0056] Step S103: Obtain the ash fouling rate value at the specified position from the second data of the wall ash fouling rate output by the second neural network, and use it as the ash fouling rate prediction value of the actual ash fouling time at the specified position.

[0057] Based on the above, this embodiment calls the pre-jointly trained first neural network and second neural network to predict the ash fouling rate value at a specified position at a specified ash fouling time based on the actual operation data of the coal-fired boiler. Since the joint model composed of the first neural network and the second neural network incorporates the information on the ash fouling distribution trend of the water-cooled wall and the real ash fouling information of the coal-fired boiler during training, the joint model trained based on multi-source fusion data not only has high prediction accuracy but also has high generalization performance. Based on this, the present application can accurately predict the ash fouling rate value at any position on the water-cooled wall at any ash fouling time by calling the above joint model. On this basis, by cumulatively determining the ash fouling amount at each position on the water-cooled wall, the purpose of accurately predicting the overall ash fouling distribution on the water-cooled wall is achieved, providing a basis for realizing the precise soot blowing task.

[0058] In addition, compared with the prediction method of obtaining the ash fouling rate on the heating surface by modeling the physical processes of pulverized coal combustion, ash particle movement and deposition in the boiler and then solving the calculation, this embodiment realizes the ash fouling rate prediction task based on the pre-jointly trained first neural network and second neural network, without the need for complex numerical calculations. The trained neural network model has a high response ability during prediction, thus ensuring the real-time performance of the ash fouling rate prediction. Therefore, in the face of frequently changing boiler operating conditions, this embodiment can quickly determine the corresponding ash fouling rate data following the changes in the operation data, meeting the need for real-time prediction of ash fouling distribution. And because the real ash fouling information is combined during the training of the joint model, the problem of limited calculation accuracy caused by factors such as physical assumptions and iterative errors in the numerical calculation process is solved to a certain extent.

[0059] Compared with the scheme of monitoring the wall ash fouling rate based on measuring instruments covering the entire water-cooled wall area, this embodiment does not need to install measuring instruments at each position on the water-cooled wall, saving the implementation cost of the prediction task.

[0060] In one or more embodiments provided by the present application, the operation data may include: the load value, coal feeding amount, water feeding amount, air distribution ratio, and steam parameters of the coal-fired boiler.

[0061] Optionally, the measuring instruments for collecting measurement values may be arranged on the backfire side of the water-cooled wall and surround the furnace once. In addition, 1 to 2 layers may be arranged at the upper soot blower. Exemplarily,Figure 2 A structural schematic diagram of a coal-fired boiler is shown, in which the installation positions of measuring instruments and soot blowers are shown. The installation position of the measuring instrument (i.e., the preset measurement position) can be located at the 1 / 2 position of the soot blowing radius of the soot blower.

[0062] In one or more embodiments provided by the present application, the configuration process of the training database of the second neural network may include the following steps S201-S206:

[0063] Step S201: Obtain the measurement values at multiple measurement times after the last ash cleaning at the preset measurement position as measurement data.

[0064] A measuring instrument is provided at the preset measurement position. The measuring instrument can be a heat flux density meter or a thermocouple. Optionally, the measuring instrument can also be other instruments capable of measuring heat flux density or temperature, which is not limited in the present application.

[0065] Step S202: Obtain the operation data of the coal-fired boiler at each of the multiple measurement times.

[0066] A set of data obtained in steps S201-S202 can be expressed as: {measurement value at the preset measurement position at measurement time t i and the operation data of the coal-fired boiler at measurement time t i . In a possible implementation, after ash cleaning at the preset measurement position, historical data can be extracted from the historical database of the boiler control system (such as a distributed control system DCS) of the coal-fired boiler at a preset time interval (such as 1 minute), specifically including the operation data and measurement data of the coal-fired boiler.

[0067] Step S203: According to the obtained operation data, screen the measurement data that meets the stable operating condition from the measurement data, and determine the operation data that can characterize the stable operating condition corresponding to the screened measurement data as the operation data corresponding to the screened measurement data.

[0068] The stable operating condition may include: within a measurement period composed of at least two measurement times, the operation data is the same or the degree of change is less than a preset limit. Through the above screening, the situation that the training data is inaccurate caused by the fluctuation of the operation data can be avoided to a certain extent.

[0069] Step S204: Group the measurement data corresponding to the same operation data in the screened measurement data, and calculate the change rate of the measurement value according to each group of measurement data respectively.

[0070] Step S205: Determine the ash deposition rate value at the preset measurement position corresponding to the measured value according to the calculated change rate of the measured value, and determine the ash deposition time and operation data corresponding to each ash deposition rate value.

[0071] Step S206: Configure the training database of the second neural network.

[0072] The training database of the second neural network can also be referred to as the ash deposition rate database based on actual measured values, which may include: the calculated ash deposition rate values at the preset measurement positions and the ash deposition time and operation data corresponding to each ash deposition rate value.

[0073] Configure the training data of the second neural network based on the above solution, so that the trained second neural network can utilize the actual change trend of the measured values provided by the measuring instrument to reflect the change trend of the wall ash deposition rate, and thus has the ability to output the wall ash deposition rate data changing with time.

[0074] In one or more embodiments provided in the present application, the configuration process of the training database of the first neural network may include:

[0075] Step S301: Respectively use each group of operation data in the training database of the second neural network as the boundary conditions for numerical simulation calculation, and call the pre-established boiler ash deposition rate calculation model to perform numerical simulation calculation to obtain the simulation calculation data corresponding to each group of operation data.

[0076] It should be noted that the operation data in the training database of the above-mentioned second neural network refers to the operation data that can represent stable operation conditions; being used as boundary conditions may refer to being used as input variables or limiting conditions. For example, the inlet flow rate, inlet pressure, etc. can be used as boundary conditions. Among them, the boiler ash deposition rate calculation model is established based on the boiler structure and boiler mechanism of the coal-fired boiler.

[0077] In a possible implementation, the operation conditions can be batch selected for numerical simulation calculation based on the boiler parameter change characteristics involved in the training database of the second neural network, where the parameter changes of the selected operation conditions can cover the operation parameter change range corresponding to the measurement data.

[0078] By means of the above step S301, it can be ensured that the data from different sources correspond to the unified operation range of the coal-fired boiler, providing a basis for subsequent joint training.

[0079] Step S302: Configure the training database of the first neural network.

[0080] Among them, the training database of the first neural network includes: the calculated simulation calculation data and the operation data corresponding to each simulation calculation data.

[0081] Configure the training data of the first neural network based on the above solution, so that the first neural network can learn the ash fouling distribution trend information under different working conditions during the training process. Since there is an implicit correlation between the ash fouling rate data from different sources, the data obtained through simulation calculation and the data obtained through measurement calculation can be fused. Therefore, taking the first data that can characterize the wall ash fouling distribution information output by the first neural network and other data as the input of the second neural network, the ash fouling distribution information based on numerical simulation can be transmitted to the second neural network in the form of features, providing a rich data basis for subsequent ash fouling rate prediction, thus making up for the shortcoming of the insufficient measurement data range and enhancing the generalization ability of the second neural network while providing trend information.

[0082] In one or more embodiments provided by the present application, the establishment process of the boiler ash fouling rate calculation model may include the following steps:

[0083] Step S401: Establish the geometric model of the coal-fired boiler.

[0084] Exemplarily, the geometric model can be established according to the elevation and dimensions of the coal-fired boiler drawings; in practical applications, modeling tools such as the 3D visualization entity simulation software inventor and the 3D modeling software solidworks can be used.

[0085] Step S402: Perform mesh division on the geometric model.

[0086] It should be noted that when performing mesh division on the geometric model, the geometric model can be divided into different volumes according to different boundary conditions, and then grids are generated for each volume. When setting the grid density, the size of the sootblower sootblowing range for a single sootblower can be considered to avoid the problem of low prediction accuracy caused by too large grids, and at the same time, it is necessary to ensure that the number of grids is not too large to avoid excessive subsequent calculation amount.

[0087] Step S403: Configure mechanism models for each of the divided grids respectively to obtain the calculation networks corresponding to each grid, and form the boiler ash fouling rate calculation model.

[0088] Among them, the mechanism model includes at least one of a turbulent flow model, a pulverized coal combustion model, a particle motion model, and a wall deposition model. Exemplarily, in practical applications, a simulation calculation software fluent can be used to establish a boiler ash deposition rate calculation model. When calculating the temperature of a certain grid, a heat transfer model is configured for this grid. When calculating the velocity of a certain grid, a turbulence model is configured for this grid. When combustion needs to be considered, a chemical reaction model is configured. The configured calculation model can be used for numerical calculation on the geometric model with the generated grid.

[0089] In one or more embodiments provided by the present application, when the specified position belongs to the preset measurement position (i.e., a measuring instrument is arranged within the soot blowing range of the preset soot blower), the method may further include:

[0090] Step S104, obtaining a measurement value corresponding to the actual ash deposition time at the specified position at the measurement moment.

[0091] Step S105, determining the actual ash deposition rate value at the specified position according to the change rate of the measurement value calculated from the obtained measurement value.

[0092] Step S106, correcting the joint model composed of the first neural network and the second neural network according to the actual operation data, the actual ash deposition time at the specified position, and the actual ash deposition rate value at the specified position.

[0093] When the training database of the first neural network does not include the actual operation data, the loss function used in the model correction stage can be determined based on the loss of the above-mentioned second neural network, so as to save the time required for the numerical simulation calculation process and improve the model correction efficiency.

[0094] By the above steps, the first neural network and the second neural network that are jointly trained in advance can be retrained, which helps to further improve the accuracy of ash deposition rate prediction.

[0095] The trained joint model described in this embodiment can output the wall ash fouling rate information based on the input boiler operation parameters. Since in the joint model, the mapping relationship between the operation data and the wall ash fouling rate is determined by the first neural network, and the characteristics of the ash fouling rate changing with time are determined by the second neural network. After the boiler operation information is given, by inputting the output of the first neural network into the second neural network, when the mapping relationship between the boiler operation parameters and the wall ash fouling rate is determined by the second neural network, the ash fouling distribution trend information in the numerical simulation data and the value information of the measurement and calculation data can be considered simultaneously, and finally the purpose of data fusion can be achieved. Based on this, the trained joint model can quickly obtain detailed and highly accurate ash fouling rate information, so as to provide guidance for the soot blower control and optimize the soot blowing method of the coal-fired boiler.

[0096] Based on the above content, when the coal-fired boiler is a 350 MW supercritical wall-fired opposed fired boiler, Figure 3 The schematic diagram of the ash fouling rate distribution of the coal-fired boiler under a load of 244 MW is shown, in which the ash fouling rate values at different positions are shown, providing a basis for realizing the precise control of the soot blower.

[0097] The ash fouling rate prediction device provided by the embodiment of the present application will be described below. The ash fouling rate prediction device described below can be correspondingly referred to the ash fouling rate prediction method described above.

[0098] The ash fouling rate prediction device provided by the embodiment of the present application may include:

[0099] A data acquisition unit, configured to acquire the actual operation data of the coal-fired boiler;

[0100] A rate prediction unit, configured to call a pre-jointly trained first neural network and a second neural network, input the actual operation data into the called first neural network, and input the first data of the fouling rate of the wall area output by the first neural network, the actual operation data, and the actual fouling time at a specified position into the called second neural network; the specified position is a position on the water-cooled wall of the coal-fired boiler corresponding to the blowing range of a preset soot blower, and the preset soot blower is one of multiple soot blowers arranged on the water-cooled wall; wherein, the training database of the first neural network includes the operation data of the coal-fired boiler marked with simulation calculation data, and the simulation calculation data includes the fouling rate values at various positions on the water-cooled wall calculated based on computational fluid dynamics; the training database of the second neural network includes the operation data of the coal-fired boiler marked with measurement calculation data and the fouling time at a preset measurement position on the water-cooled wall, and the measurement calculation data includes the fouling rate values at the preset measurement position calculated based on the change rate of the measurement values, and the measurement values include the heat flux density measurement values or temperature measurement values at the preset measurement position;

[0101] The rate prediction unit is further configured to obtain the fouling rate value at the specified position from the second data of the fouling rate of the wall area output by the second neural network as the fouling rate prediction value of the actual fouling time at the specified position.

[0102] In one or more embodiments provided by the present application, the device may further include a database configuration unit, configured to configure the training database of the second neural network. The process of the database configuration unit configuring the training database of the second neural network may include:

[0103] Obtain the measurement values at multiple measurement moments after the last soot cleaning at the preset measurement position as measurement data;

[0104] Obtain the operation data of the coal-fired boiler at each of the multiple measurement moments;

[0105] According to the obtained operation data, screen the measurement data that meets the stable operating condition requirements from the measurement data, and determine the operation data that can characterize the stable operating condition corresponding to the screened measurement data as the operation data corresponding to the screened measurement data;

[0106] Group the measurement data corresponding to the same operation data in the screened measurement data, and calculate the change rate of the measurement values respectively according to each group of measurement data;

[0107] Determine the fouling rate value at the preset measurement position corresponding to the measurement value according to the calculated change rate of the measurement value, and determine the fouling time and operation data corresponding to each fouling rate value;

[0108] Configure the training database of the second neural network, where the training database of the second neural network includes: the calculated ash fouling rate values at the preset measurement positions, and the ash fouling time and operation data corresponding to each ash fouling rate value respectively.

[0109] In one or more embodiments provided by the present application, the database configuration unit can also be used to configure the training database of the first neural network. The process of the database configuration unit configuring the training database of the first neural network can include:

[0110] Respectively use each group of operation data in the training database of the second neural network as the boundary conditions for numerical simulation calculation, and call the pre-established boiler ash fouling rate calculation model to perform numerical simulation calculation to obtain the simulation calculation data corresponding to each group of operation data; wherein, the boiler ash fouling rate calculation model is established based on the boiler structure and boiler mechanism of the coal-fired boiler.

[0111] Configure the training database of the first neural network, where the training database of the first neural network includes: the calculated simulation calculation data and the operation data corresponding to each simulation calculation data respectively.

[0112] In one or more embodiments provided by the present application, the establishment process of the boiler ash fouling rate calculation model can include:

[0113] Establish the geometric model of the coal-fired boiler;

[0114] Perform mesh division on the geometric model;

[0115] Respectively configure mechanism models for each of the divided meshes to obtain calculation networks corresponding to each mesh, and form a boiler ash fouling rate calculation model; the mechanism model includes at least one of a turbulent flow model, a pulverized coal combustion model, a particle operation model, and a wall deposition model.

[0116] In one or more embodiments provided by the present application, the operation data can include: the load value, coal feeding amount, water feeding amount, air distribution ratio, and steam parameters of the coal-fired boiler.

[0117] In one or more embodiments provided by the present application, the device can also include a model training unit for jointly training the first neural network and the second neural network using the training database of the first neural network and the training database of the second neural network.

[0118] In one or more embodiments provided by the present application, the model training unit can also be used for:

[0119] When the specified position belongs to the preset measurement position, obtain the measurement value corresponding to the actual ash accumulation time at the specified position at the measurement moment;

[0120] Determine the actual ash accumulation rate value at the specified position according to the change rate of the measurement value calculated from the obtained measurement value;

[0121] Modify the joint model composed of the first neural network and the second neural network according to the actual operation data, the actual ash accumulation time at the specified position, and the actual ash accumulation rate value at the specified position.

[0122] The ash accumulation rate prediction device provided by the embodiments of the present application can be applied to electronic devices, such as terminals with data processing capabilities: mobile phones, computers, etc. Optionally, Figure 4 The hardware structure block diagram of the electronic device is shown. Refer to Figure 4 , the hardware structure of the electronic device may include: at least one processor 1, at least one communication interface 2, at least one memory 3, and at least one communication bus 4;

[0123] In the embodiments of the present application, the number of the processor 1, the communication interface 2, the memory 3, and the communication bus 4 is at least one, and the processor 1, the communication interface 2, and the memory 3 complete mutual communication through the communication bus 4;

[0124] The processor 1 may be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention, etc.;

[0125] The memory 3 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory;

[0126] Among them, the memory is used to store a computer program, and the processor is used to execute the computer program so that the electronic device can implement any of the above ash accumulation rate prediction methods.

[0127] In the embodiments of the present application, a storage medium is also provided. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the ash accumulation rate prediction methods provided by the embodiments of the present application.

[0128] An embodiment of the present application also provides a computer program product including computer-readable instructions. When the computer-readable instructions run on an electronic device, the electronic device can implement any of the ash accumulation rate prediction methods provided in the embodiments of the present application.

[0129] Figure 5 FIG. 4 is a schematic flowchart of a sootblowing control method shown according to an embodiment of the present application. This method can be applied to a coal-fired boiler. A plurality of sootblowers are arranged on the water-cooled wall of the coal-fired boiler, and the sootblowing ranges of different sootblowers are different. With reference to Figure 5 as shown, this method may include the following steps:

[0130] Step S501: Determine a preset sootblower, and use the position on the water-cooled wall corresponding to the sootblowing range of the preset sootblower as a designated position.

[0131] Wherein, the preset sootblower is one of the plurality of sootblowers.

[0132] Step S502: After sootblowing on the sootblowing range of the preset sootblower for the last time, calculate the current accumulated ash time as the actual ash accumulation time.

[0133] Step S503: Obtain the ash accumulation rate prediction value of the actual ash accumulation time at the designated position.

[0134] Wherein, the ash accumulation rate prediction value can be determined using any of the above ash accumulation rate prediction methods.

[0135] Step S504: Calculate the ash accumulation amount within the sootblowing range of the preset sootblower based on the actual ash accumulation time and the ash accumulation rate prediction value.

[0136] Step S505: Determine whether the ash accumulation amount exceeds a preset ash accumulation amount threshold. If so, execute Step S506; otherwise, execute Step S507.

[0137] Step S506: Control the preset sootblower to perform a sootblowing operation for a preset sootblowing duration to remove the ash within the sootblowing range of the preset sootblower.

[0138] Wherein, the sootblowing capacity of the sootblowing operation for the preset sootblowing duration can match the ash accumulation capacity at the designated position after the last sootblowing, so as to just completely achieve the sootblowing task through the sootblowing operation.

[0139] Step S507: Calculate the current accumulated ash time as the actual ash accumulation time, and return to execute Step S503.

[0140] In a possible implementation, to reduce the data processing pressure caused by frequent calculations, before performing step S507, a preset duration can be waited, that is, after the soot blowing range of the preset soot blower was last cleaned, step S507 can be performed according to a preset time step.

[0141] The above solution uses the accumulation of the ash fouling rate value predicted by the joint model and time to determine the ash fouling situation on the water wall surface, especially at the position corresponding to the soot blowing range of the soot blower. When the ash fouling at the water wall position corresponding to any soot blower reaches the limit value, soot blowing recommendation is made, achieving the purpose of controlling the soot blower to blow soot as needed, solving the differentiated soot blowing tasks of different soot blowers, and avoiding the situations of under-soot blowing or over-soot blowing, thereby realizing the precise treatment of the boiler ash fouling problem.

[0142] In a possible implementation, based on the soot blowing cycle of the preset soot blower, the ash fouling condition of the water wall corresponding to the soot blowing range of the preset soot blower can be calculated within each soot blowing cycle, and based on the predicted ash fouling rate value that changes with time obtained within a single soot blowing cycle, the ash fouling amount of the soot blower within a single soot blowing cycle can be determined, and the ash fouling amount threshold corresponding to the soot blower can be set. Since the soot blowing operation performed by the soot blower matches the ash fouling amount, when the boiler operation data remains unchanged and the boiler operates stably, subsequently, the soot blower can be periodically controlled according to the above-mentioned soot blowing cycle to perform the above-mentioned soot blowing operation to save computing resources.

[0143] The soot blowing control device provided in the embodiments of the present application will be described below. The soot blowing control device described below can be correspondingly referred to the soot blowing control method described above.

[0144] The soot blowing control device provided in the embodiments of the present application can be applied to a coal-fired boiler, and a plurality of soot blowers are arranged on the water wall of the coal-fired boiler, and the soot blowing ranges of different soot blowers are different; the device can include:

[0145] A control object determination unit, configured to determine a preset soot blower, and use the position on the water wall corresponding to the soot blowing range of the preset soot blower as a designated position; the preset soot blower is one of the plurality of soot blowers;

[0146] An ash fouling monitoring unit, configured to calculate the current accumulated ash fouling time as the actual ash fouling time after the soot blowing range of the preset soot blower was last cleaned; obtain the predicted ash fouling rate value of the actual ash fouling time at the designated position, and the predicted ash fouling rate value is determined by any of the above-mentioned ash fouling rate prediction methods; calculate the ash fouling amount within the soot blowing range of the preset soot blower according to the actual ash fouling time and the predicted ash fouling rate value; and return to execute the step of calculating the current accumulated ash fouling time when the ash fouling amount does not exceed the preset ash fouling amount threshold.

[0147] The soot blowing control unit is configured to control the preset soot blower to perform a soot blowing operation for a preset soot blowing duration when the soot accumulation amount exceeds the soot accumulation amount threshold, so as to remove the soot accumulation within the soot blowing range of the preset soot blower.

[0148] The soot blowing control device provided by the embodiments of the present application can be applied to a soot blowing controller, specifically referring to a terminal with data processing capabilities, and its hardware structure can refer to Figure 4 as shown. The soot blowing controller may include at least one processor and a memory connected to the processor, where:

[0149] The memory is used to store computer programs;

[0150] The processor is configured to execute the computer programs so that the soot blowing controller implements the above-mentioned soot blowing control method.

[0151] In the embodiments of the present application, a storage medium is further provided. The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the soot blowing control method provided by the embodiments of the present application.

[0152] In the embodiments of the present application, a computer program product is further provided, including computer-readable instructions. When the computer-readable instructions run on an electronic device, the electronic device can implement the soot blowing control method provided by the embodiments of the present application.

[0153] The embodiments of the present application further provide a boiler system, which may include: a coal-fired boiler, a soot blowing controller, and a boiler control system of the coal-fired boiler.

[0154] Wherein, a plurality of soot blowers are arranged on the water-cooled wall of the coal-fired boiler, and the soot blowing ranges of different soot blowers are different. Exemplarily, the above-mentioned soot blower may be a short telescopic soot blower, which is arranged at the water-cooled wall of the boiler and has a short time to enter the furnace for soot blowing.

[0155] The boiler control system is communicatively connected to the plurality of soot blowers. Through this connection, the boiler control system can output start / stop signals to the soot blowers to control the soot blowers to perform soot blowing / stop soot blowing operations. Exemplarily, the boiler control system may refer to a distributed control system (DCS), in which a soot blowing control system, that is, a control system for the soot blower, may be provided, which can be used to implement the original sequence control logic of the soot blower and functions such as step skipping, suspension, and reset. In addition, the soot blowing control system can also be used to monitor safety parameters such as the temperature of the soot blowing pipeline and the soot blowing pressure to ensure that steam soot blowing is carried out within a normal range.

[0156] The soot blowing controller is communicatively connected to the boiler control system, and the soot blowing controller can be used for:

[0157] Take the position on the water-cooled wall corresponding to the soot blowing range of the preset soot blower as the designated position; the preset soot blower is one of the multiple soot blowers; after soot cleaning in the soot blowing range of the preset soot blower once, calculate the current cumulative soot accumulation time as the actual soot accumulation time; obtain the predicted soot accumulation rate of the actual soot accumulation time at the designated position, and the predicted soot accumulation rate is determined by using the above-mentioned soot accumulation rate prediction method; calculate the soot accumulation amount in the soot blowing range of the preset soot blower according to the actual soot accumulation time and the predicted soot accumulation rate; when the soot accumulation amount exceeds the preset soot accumulation amount threshold, control the preset soot blower to perform a soot blowing operation for a preset soot blowing duration to remove the soot accumulation in the soot blowing range of the preset soot blower; wherein, the control of the preset soot blower to perform a soot blowing operation for a preset soot blowing duration includes: outputting a control instruction to the boiler control system, so that the boiler control system outputs a start instruction to the preset soot blower and outputs a stop instruction to the preset soot blower after the preset soot blowing duration; and, when the soot accumulation amount does not exceed the soot accumulation amount threshold, return to execute the step of calculating the current cumulative soot accumulation time.

[0158] In a possible implementation, the soot blowing controller can be an industrial control computer, arranged in the power plant electronic room or the engineer station. As Figure 6 shown, the soot blowing controller can communicate bidirectionally with the boiler control system (such as DCS), extract measurement data and operation data from the DCS, and output soot blowing control instructions, wall soot accumulation information, etc. to the DCS after data processing.

[0159] In addition, in the DCS, there can be selection boxes corresponding to each soot blower for the operator to perform the selection operation of the soot blower. Based on this, after manually selecting the soot blowing optimization recommendation, the soot blowing control system can control the soot blower according to the instructions issued by the soot blowing controller.

[0160] In one or more embodiments provided by the present application, the boiler system may further include a measuring instrument, and the measuring instrument may be a heat flux density meter or a thermocouple, or other measuring instruments capable of measuring heat flux density or temperature, which is not limited in the present application.

[0161] On the above basis, the soot blowing controller can also be used for:

[0162] Obtain a measurement value corresponding to the actual ash accumulation time at the specified position from the measuring instrument disposed at the specified position; determine the actual ash accumulation rate value at the specified position according to the change rate of the measurement value calculated from the obtained measurement value; correct the combined model composed of the first neural network and the second neural network according to the actual operation data, the actual ash accumulation time at the specified position, and the actual ash accumulation rate value at the specified position.

[0163] By controlling the installation accuracy so that the measuring instrument is installed within the purging range of the soot blower, the obvious change in the parameter value measured by the measuring instrument can reflect the ash accumulation change before and after soot blowing, providing data support for obtaining the soot blowing effect and correcting the combined model.

[0164] In addition, other descriptions of the soot blower controller, measuring instrument, etc., such as the way the soot blower controller obtains the predicted ash accumulation rate value and the setting method of the measuring instrument, can be referred to the above description and will not be elaborated here.

[0165] Based on the above content, the boiler system provided by this application can quickly predict the ash accumulation information on the wall surface, providing detailed prediction information with high prediction accuracy for the control of the soot blower. Accordingly, the soot blowers set at different positions are controlled for soot blowing to match the ash accumulation amount within their purging ranges, realizing the optimized operation task of the soot blower.

[0166] Finally, it should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0167] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0168] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting dust accumulation rate, characterized in that: The method includes: Obtain actual operating data of coal-fired boilers; Calling a first neural network and a second neural network that are jointly trained in advance, inputting the actual operation data into the called first neural network, and inputting the first data of the wall dust accumulation rate output by the first neural network, the actual operation data, and the actual dust accumulation time at a specified position into the called second neural network; the specified position is a position on the water-cooled wall of the coal-fired boiler corresponding to the soot blowing range of a preset soot blower, and the preset soot blower is one of a plurality of soot blowers arranged on the water-cooled wall; Obtaining the dust accumulation rate value at the designated position from the second data of the wall dust accumulation rate output by the second neural network as the dust accumulation rate prediction value of the actual dust accumulation time at the designated position; Among them, the training database of the first neural network includes the operating data of the coal-fired boiler marked with simulation calculation data, and the simulation calculation data include the ash accumulation rate values ​​at various positions on the water-cooled wall calculated based on computational fluid dynamics; the training database of the second neural network includes the operating data of the coal-fired boiler marked with measurement calculation data and the ash accumulation time at preset measurement positions on the water-cooled wall, the measurement calculation data include the ash accumulation rate values ​​at the preset measurement positions calculated based on the rate of change of the measurement values, and the measurement values ​​include heat flux density measurement values ​​or temperature measurement values ​​at the preset measurement positions.

2. The dust accumulation rate prediction method according to claim 1, characterized in that: The configuration process of the training database of the second neural network includes: Obtaining measurement values ​​at multiple measurement moments after the last dust cleaning at the preset measurement position as measurement data; Acquire the operation data of the coal-fired boiler at each of the multiple measurement times; According to the acquired operating data, the measurement data satisfying the stable working condition is screened from the measurement data, and the operating data capable of representing the stable working condition corresponding to the screened measurement data is determined as the operating data corresponding to the screened measurement data; The measurement data corresponding to the same operation data in the screened measurement data are grouped together, and the change rate of the measurement value is calculated based on each group of measurement data; Determine the dust accumulation rate value at the preset measurement position corresponding to the measurement value according to the calculated change rate of the measurement value, and determine the dust accumulation time and operation data corresponding to each dust accumulation rate value; A training database for the second neural network is configured, wherein the training database for the second neural network includes: the calculated dust accumulation rate values ​​at the preset measurement positions, and the dust accumulation time and operation data corresponding to each dust accumulation rate value.

3. The dust accumulation rate prediction method according to claim 2, characterized in that: The configuration process of the training database of the first neural network includes: Using each group of operating data in the training database of the second neural network as the boundary conditions of the numerical simulation calculation, calling the pre-established boiler ash accumulation rate calculation model to perform numerical simulation calculation, and obtaining the simulation calculation data corresponding to each group of operating data; Configuring a training database for the first neural network, wherein the training database for the first neural network includes: the calculated simulation calculation data and the operation data corresponding to each simulation calculation data; Wherein, the boiler ash deposition rate calculation model is established based on the boiler structure and boiler mechanism of the coal-fired boiler.

4. The dust accumulation rate prediction method according to claim 3, characterized in that: The process of establishing the boiler ash deposition rate calculation model includes: Establishing a geometric model of the coal-fired boiler; Meshing the geometric model; A mechanism model is configured for each grid obtained by division, and a calculation network corresponding to each grid is obtained to form a boiler ash deposition rate calculation model; the mechanism model includes at least one of a turbulent flow model, a pulverized coal combustion model, a particle operation model and a wall deposition model.

5. The dust accumulation rate prediction method according to any one of claims 1 to 4, characterized in that: The operation data include: the load value, coal supply, water supply, air distribution ratio and steam parameters of the coal-fired boiler.

6. The dust accumulation rate prediction method according to any one of claims 1 to 4, characterized in that: In the case where the designated position belongs to the preset measurement position, the method further includes: Obtaining a measurement value corresponding to the actual dust accumulation time at the specified position at the measurement time; Determining an actual dust accumulation rate value at the designated location according to a change rate of the measured value calculated from the acquired measured value; The joint model composed of the first neural network and the second neural network is modified according to the actual operation data, the actual dust accumulation time at the designated location and the actual dust accumulation rate value at the designated location.

7. A sootblowing control method, characterized in that: Applied to a coal-fired boiler, a plurality of sootblowers are arranged on the water-cooled wall of the coal-fired boiler, and the sootblowers have different sootblowers with different sootblowers. The method comprises: Determine a preset sootblower, and use a position on the water-cooled wall corresponding to a sootblowing range of the preset sootblower as a designated position; the preset sootblower is one of the multiple sootblowers; After the soot blowing range of the preset soot blower was last cleaned, the current accumulated soot accumulation time is calculated as the actual soot accumulation time; Obtaining a predicted value of dust accumulation rate of the actual dust accumulation time at the designated position, wherein the predicted value of dust accumulation rate is determined using the dust accumulation rate prediction method according to any one of claims 1 to 6; Calculating the amount of dust accumulation within the soot blowing range of the preset soot blower according to the actual dust accumulation time and the predicted value of the dust accumulation rate; When the amount of accumulated dust exceeds a preset threshold value of accumulated dust, controlling the preset soot blower to perform a soot blowing operation for a preset soot blowing time to clear the accumulated dust within the soot blowing range of the preset soot blower; When the dust accumulation amount does not exceed the dust accumulation amount threshold, the process returns to the step of calculating the current dust accumulation cumulative time.

8. A dust accumulation rate prediction device, characterized in that: include: A data acquisition unit, used to acquire actual operation data of the coal-fired boiler; A rate prediction unit, used to call a first neural network and a second neural network that are jointly trained in advance, input the actual operation data into the called first neural network, and input the first data of the wall ash accumulation rate output by the first neural network, the actual operation data, and the actual ash accumulation time at a specified position into the called second neural network; the specified position is a position on the water-cooled wall of the coal-fired boiler corresponding to the soot blowing range of a preset soot blower, and the preset soot blower is one of a plurality of soot blowers arranged on the water-cooled wall; wherein the training database of the first neural network includes the operation data of the coal-fired boiler annotated with simulation calculation data, and the simulation calculation data includes the ash accumulation rate values ​​at various positions on the water-cooled wall calculated based on computational fluid dynamics; the training database of the second neural network includes the operation data of the coal-fired boiler annotated with measurement calculation data and the ash accumulation time at a preset measurement position on the water-cooled wall, the measurement calculation data includes the ash accumulation rate value at the preset measurement position calculated based on the rate of change of the measurement value, and the measurement value includes the heat flux density measurement value or the temperature measurement value at the preset measurement position; The rate prediction unit is also used to obtain the dust accumulation rate value at the specified position from the second data of the wall dust accumulation rate output by the second neural network as the dust accumulation rate prediction value of the actual dust accumulation time at the specified position.

9. A sootblowing control device, characterized in that: Applied to a coal-fired boiler, a plurality of sootblowers are arranged on the water-cooled wall of the coal-fired boiler, and the sootblowers have different sootblowers with different sootblow ranges; the device comprises: a control object determination unit, configured to determine a preset sootblower, and use a position on the water-cooled wall corresponding to a sootblowing range of the preset sootblower as a designated position; the preset sootblower is one of the plurality of sootblowers; A dust accumulation monitoring unit is used to calculate the current dust accumulation cumulative time after the last dust cleaning in the dust blowing range of the preset dust blower as the actual dust accumulation time; obtain the dust accumulation rate prediction value of the actual dust accumulation time at the specified position, and the dust accumulation rate prediction value is determined using the dust accumulation rate prediction method described in any one of claims 1 to 6; calculate the dust accumulation amount in the dust blowing range of the preset dust blower according to the actual dust accumulation time and the dust accumulation rate prediction value; and when the dust accumulation amount does not exceed a preset dust accumulation amount threshold, return to the step of calculating the current dust accumulation cumulative time; A sootblowing control unit is used to control the preset sootblower to perform a sootblowing operation for a preset sootblowing time when the soot accumulation exceeds the soot accumulation threshold, so as to clear the soot accumulation within the sootblowing range of the preset sootblower.

10. A boiler system, characterized in that: include: A coal-fired boiler, a sootblowing controller and a boiler control system of the coal-fired boiler; Wherein, a plurality of sootblowers are arranged on the water-cooled wall of the coal-fired boiler, and the sootblowers of different sootblowers have different sootblow ranges; the boiler control system is in communication connection with the plurality of sootblowers; The sootblowing controller is in communication with the boiler control system, and the sootblowing controller is used to: A position on the water-cooled wall corresponding to the sootblowing range of a preset sootblower is used as a designated position; the preset sootblower is one of the multiple sootblowers; After the soot blowing range of the preset soot blower was last cleaned, the current accumulated soot accumulation time is calculated as the actual soot accumulation time; Obtaining a predicted value of dust accumulation rate of the actual dust accumulation time at the designated position, wherein the predicted value of dust accumulation rate is determined using the dust accumulation rate prediction method according to any one of claims 1 to 5; Calculating the amount of dust accumulation within the soot blowing range of the preset soot blower according to the actual dust accumulation time and the predicted value of the dust accumulation rate; When the ash accumulation amount exceeds the preset ash accumulation amount threshold, the preset soot blower is controlled to perform a soot blowing operation for a preset soot blowing time to clear the ash accumulation within the soot blowing range of the preset soot blower; wherein, the step of controlling the preset soot blower to perform a soot blowing operation for a preset soot blowing time comprises: outputting a control instruction to the boiler control system so that the boiler control system outputs a start instruction to the preset soot blower, and outputs a stop instruction to the preset soot blower after the preset soot blowing time; When the dust accumulation amount does not exceed the dust accumulation amount threshold, the process returns to the step of calculating the current dust accumulation cumulative time.

11. The boiler system according to claim 10, characterized in that: The boiler system further comprises a measuring instrument, which is a heat flux density meter or a thermocouple; The sootblowing controller is also used for: Obtaining, from the measuring instrument disposed at the designated position, a measurement value corresponding to the actual dust accumulation time at the designated position at the measuring time; Determining an actual dust accumulation rate value at the designated location according to a change rate of the measured value calculated from the acquired measured value; The joint model composed of the first neural network and the second neural network is modified according to the actual operation data, the actual dust accumulation time at the designated location and the actual dust accumulation rate value at the designated location.

Images