An automated control system and method for induced draft fans based on DCS

Abstract

This invention relates to the field of induced draft fan automation control technology, and discloses an induced draft fan automation control system and method based on DCS. The system includes: an acquisition module for acquiring historical speed data, historical vibration data at several key locations, and historical control data of the induced draft fan; a first construction module for determining several historical resonance intervals and constructing a resonance interval analysis model; a second construction module for determining the correlation between historical boiler operating data and historical control parameters, and determining target control parameters by combining the first control parameter interval corresponding to each historical resonance interval, and constructing a control analysis model; and a control module for collecting real-time relevant operating data, real-time speed data, and real-time vibration data, and determining whether to generate control commands by combining the resonance interval analysis model and the control analysis model, so as to ensure normal boiler operation, reduce resonance risk, and improve operating efficiency.

Description

Technical Field

[0001] This application relates to the field of induced draft fan automation control technology, and in particular to an induced draft fan automation control system and method based on DCS. Background Technology

[0002] In the operation of thermal power plants, induced draft fans, as important auxiliary equipment, directly affect the safe and economical operation of the entire plant due to their operational stability and efficiency. Currently, induced draft fans in thermal power plants, especially dynamically adjustable axial flow fans, are prone to resonance problems across the entire speed range during operation. When the operating speed of the induced draft fan coincides with the resonant frequency, it will generate severe vibrations, which not only affect the service life of the equipment but may also cause safety accidents. At the same time, existing induced draft fan adjustments mostly rely solely on their own parameters without fully incorporating important boiler operating parameters, resulting in low operating efficiency and making it difficult to meet the requirements of efficient and stable operation of thermal power plants. Summary of the Invention

[0003] To address the aforementioned technical issues, this application provides a DCS-based automated control system and method for induced draft fans. By constructing a resonance interval analysis model and a control analysis model, the resonance interval is accurately determined. Furthermore, by combining important boiler parameters and induced draft fan operating efficiency for comprehensive analysis, reasonable control commands are generated to ensure normal boiler operation while reducing resonance risks and improving operating efficiency.

[0004] In some embodiments of this application, a DCS-based automated control system for induced draft fans is provided, comprising:

[0005] The acquisition module is used to acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data.

[0006] The first construction module is used to determine several historical resonance intervals based on historical rotational speed data, historical vibration data, and historical control data, and to construct a resonance interval analysis model.

[0007] The second construction module is used to determine the correlation between historical boiler operating data and historical control parameters, and, in conjunction with the first control parameter interval corresponding to each historical resonance interval, determine the target control parameters and construct a control analysis model.

[0008] The control module is used to determine whether to generate control commands based on the resonance interval analysis model and the control analysis model.

[0009] In some embodiments of this application, historical rotational speed data of the induced draft fan, historical vibration data of several key locations, and historical control data are obtained, including:

[0010] Determine the type of induced draft fan based on the current basic information of the induced draft fan;

[0011] Obtain several historical monitoring logs throughout the entire lifecycle of the current induced draft fan category, and extract historical initial speed data, historical initial vibration data, and historical initial control data from multiple identical historical monitoring nodes in each historical monitoring log;

[0012] The historical initial rotation speed data, historical initial vibration data, and historical initial control data at each historical monitoring node are preprocessed, including noise removal, missing value filling, and standardization.

[0013] Historical speed data, historical vibration data, and historical control data are obtained based on the preprocessed historical initial speed data, historical initial vibration data, and historical initial control data.

[0014] In some embodiments of this application, several historical resonance intervals are determined based on historical rotational speed data, historical vibration data, and historical control data, including:

[0015] Feature extraction is performed on the historical rotational speed data and historical vibration data at each identical historical monitoring node to obtain historical rotational speed features and historical vibration features.

[0016] Based on preset capture conditions, the historical rotation speed characteristics and historical vibration characteristics at multiple consecutive identical historical monitoring nodes are captured. The preset capture conditions include a first preset capture condition, a second preset capture condition, and a third preset capture condition.

[0017] Based on the capture results, several historical resonance intervals are determined, and each historical resonance interval is mapped with corresponding preset capture conditions and corresponding capture features.

[0018] The historical resonance intervals are classified according to the preset capture conditions and capture features to obtain several historical resonance intervals of the same category, and a group of historical resonance intervals for each category is constructed.

[0019] Analyze several historical resonance intervals in each category of historical resonance interval group to determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group.

[0020] In some embodiments of this application, determining a first resonance interval, a first data interval set, several second resonance intervals, and a corresponding second data interval set corresponding to a historical resonance interval group includes:

[0021] By comparing and analyzing multiple historical resonance intervals in the same historical resonance interval group, the difference in the first resonance interval between different historical resonance intervals is obtained.

[0022] Clusters corresponding to historical resonance interval groups are determined based on the differences in several first resonance intervals, and the clusters include several historical resonance intervals;

[0023] The first resonance interval is generated based on several historical resonance intervals in the cluster.

[0024] Obtain several historical operational data points and corresponding historical data intervals at the historical monitoring nodes corresponding to the historical resonance intervals in the clusters;

[0025] Generate the first data interval of the corresponding historical operation data based on multiple historical data intervals corresponding to the same historical operation data;

[0026] Each historical resonance interval that does not belong to a cluster within the same historical resonance interval group is designated as a second resonance interval.

[0027] Acquire several historical operating data points and corresponding historical data intervals at the historical monitoring nodes corresponding to each resonance interval in the same historical resonance interval group, and set the historical data interval as the second data interval of the corresponding historical operating data.

[0028] The first data interval of the same historical operating data is compared with the second data interval of the historical operating data corresponding to each second resonance interval to obtain the data interval difference.

[0029] A data interval difference matrix corresponding to the historical running data is constructed by using multiple data interval differences of the same historical running data. The data interval difference matrix includes several data interval differences sorted by size, and each data interval difference is mapped to a corresponding second resonance interval difference.

[0030] Determine whether there is a dependency relationship between the difference values ​​of several data intervals in the data interval difference matrix and the difference values ​​of several second resonance intervals that are mapped. If yes, calculate the dependency coefficient; if no, remove the corresponding historical running data.

[0031] Calculate the number of data interval differences in the same historical data interval difference matrix that exceed a preset data interval difference threshold;

[0032] The correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than the preset data interval difference threshold and the dependency coefficient.

[0033] Historical operational data with a correlation coefficient greater than a preset correlation coefficient threshold are identified as relevant operational data.

[0034] Construct a first data interval set based on a first data interval of several related data;

[0035] A set of second data intervals is constructed based on several second data intervals of several related data, and each second data interval is mapped to a corresponding second resonance interval.

[0036] In some embodiments of this application, the correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than a preset data interval difference threshold and the dependency coefficient, including:

[0037] The formula for calculating the correlation coefficient is:

[0038] ;

[0039] Where H is the correlation coefficient, n1 is the number of data interval differences in the data interval difference matrix that are greater than the preset data interval difference threshold, n2 is the total number of data interval differences in the data interval difference matrix, and ji is the difference of the i-th data interval. Let gi be the mean of the data interval differences in the data interval difference matrix, and gi be the difference of the i-th second resonance interval. It represents the mean of the second resonance interval difference in the data interval difference matrix.

[0040] In some embodiments of this application, a resonance interval analysis model is constructed, including:

[0041] Historical rotation speed data, historical vibration data, and the corresponding first data interval set are used as the first training input data, and the corresponding first resonance interval is used as the first training output data.

[0042] Historical rotation speed data, historical vibration data, and the corresponding second data interval set are used as the second training input data, and the corresponding second resonance interval is used as the second training output data.

[0043] The neural network is trained based on the first training input data, the first training output data, the second training input data, and the second training output data to obtain the resonance interval analysis model.

[0044] In some embodiments of this application, the correlation between historical boiler operating data and historical control parameters is determined, and the target control parameters are determined by combining the first control parameter interval corresponding to each historical resonance interval, and a control analysis model is constructed, including:

[0045] The load of the induced draft fan is set into a first preset load segment, a second preset load segment, a third preset load segment, and several preset load points in each preset load segment;

[0046] Extract historical load data, historical control data, and historical boiler operation data from historical monitoring logs;

[0047] Based on the correspondence between historical load data and preset load segments, a first set of analytical data for the first preset load segment, a second set of analytical data for the second preset load segment, and a third set of analytical data for the third preset load segment are constructed. Each set of analytical data includes several subsets of analytical data, and each subset of analytical data corresponds to a preset load point.

[0048] Correlation analysis is performed on each historical boiler operation data and historical control data in each subset of analysis data to obtain the correlation relationship between each historical boiler operation data and historical control data.

[0049] The correlation includes whether a correlation exists and the historical second influence coefficient;

[0050] Historical boiler operation data that are correlated and whose historical second influence coefficient is greater than the preset influence coefficient threshold are set as key parameters of historical boilers.

[0051] The target control parameters are determined based on the historical key boiler parameters and their corresponding relationships, and it is then determined whether the target control parameters are within the corresponding first control parameter range.

[0052] If not, calculate the target air volume based on boiler operating requirements and target control parameters;

[0053] Construct control parameter-airflow-efficiency curves for each subset of analytical data;

[0054] Map the target control parameters and target air volume to the corresponding control parameter-air volume-efficiency curve to obtain the target efficiency value;

[0055] Determine whether the target efficiency value is within the preset high efficiency range. If so, use the preset load point, historical boiler key parameters, and the corresponding target control parameters as a training parameter group. The preset load point and historical boiler key parameters are the training input parameters, and the corresponding target control parameters are the training output parameters.

[0056] Generate several sets of training parameters for each subset of the analysis data;

[0057] The control analysis model is obtained by training a neural network based on several sets of training parameters for each subset of the analysis data.

[0058] In some embodiments of this application, correlation analysis is performed on each historical boiler operation data and historical control data in each subset of the analysis data, including:

[0059] A single-factor correlation analysis was performed on each historical boiler operation data and historical control data in each subset of the analysis data to obtain the historical first influence coefficient of each historical boiler operation data on the historical control data.

[0060] Historical boiler operation data with historical influence coefficients greater than the preset influence coefficient threshold are selected from each subset of analysis data and used as input variables. The corresponding historical control data are used as output variables to generate variable training sets and variable test sets.

[0061] Establish a multiple linear regression equation;

[0062] The historical second influence coefficient of each historical boiler operation data point selected based on the multiple linear regression equation is calculated for historical control data.

[0063] In some embodiments of this application, determining whether to generate a control command based on a resonance interval analysis model and a control analysis model includes:

[0064] Collect real-time relevant operating data, real-time speed data, real-time vibration data, and real-time key boiler parameters;

[0065] Based on the resonance analysis model, real-time relevant operating data, real-time rotational speed data, and real-time vibration data are analyzed to determine the real-time resonance range;

[0066] The corresponding real-time first control parameter range is determined based on the real-time resonance range.

[0067] Extract the real-time load points from the real-time relevant operating data, and input the real-time load points and key real-time boiler parameters into the control analysis model to determine the real-time target control parameters;

[0068] Determine whether the real-time target control parameters are within the range of the first real-time control parameters. If not, generate control commands according to the real-time target control parameters.

[0069] If the real-time target control parameters are fine-tuned, and the efficiency value corresponding to the fine-tuned real-time target control parameters is within the preset high-efficiency range, control commands are generated according to the fine-tuned real-time target control parameters.

[0070] In some embodiments of this application, an automated control method for induced draft fans based on DCS is also included:

[0071] Acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data;

[0072] Based on historical rotational speed data, historical vibration data, and historical control data, several historical resonance intervals were determined, and a resonance interval analysis model was constructed.

[0073] The correlation between historical boiler operating data and historical control parameters is determined, and the target control parameters are determined by combining the first control parameter interval corresponding to each historical resonance interval, and a control analysis model is constructed.

[0074] Whether to generate a control command is determined based on the resonance interval analysis model and the control analysis model.

[0075] The advantages of the DCS-based automated control system and method for induced draft fans in this application compared with the prior art are as follows:

[0076] By constructing a resonance interval analysis model and a control analysis model, the resonance interval can be accurately determined. Combined with important boiler parameters and induced draft fan operating efficiency, a comprehensive analysis is conducted to generate reasonable control commands, ensuring normal boiler operation while reducing resonance risk and improving operating efficiency. Attached Figure Description

[0077] Figure 1 This is a schematic diagram of an automated control system for an induced draft fan based on a DCS in an embodiment of this application;

[0078] Figure 2 This is a flowchart illustrating an automated control method for an induced draft fan based on DCS, as described in an embodiment of this application. Detailed Implementation

[0079] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0080] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0081] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0082] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0083] like Figure 1 As shown in the figure, an automated control system for an induced draft fan based on a DCS according to an embodiment of this application includes:

[0084] The acquisition module is used to acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data.

[0085] The first construction module is used to determine several historical resonance intervals based on historical rotational speed data, historical vibration data, and historical control data, and to construct a resonance interval analysis model.

[0086] The second construction module is used to determine the correlation between historical boiler operating data and historical control parameters, and, in conjunction with the first control parameter interval corresponding to each historical resonance interval, determine the target control parameters and construct a control analysis model.

[0087] The control module is used to determine whether to generate control commands based on the resonance interval analysis model and the control analysis model.

[0088] In this embodiment, the historical vibration data includes vibration amplitude, vibration spectrum, and phase at multiple key locations.

[0089] In this embodiment, the key locations are the bearing housing on the guide fan, the casing near the impeller, the motor bearing, etc. Vibration sensors are installed at the key locations. The vibration signals collected by the vibration sensors are converted into vibration data, namely vibration amplitude, vibration spectrum, and phase.

[0090] In some embodiments of this application, historical rotational speed data of the induced draft fan, historical vibration data of several key locations, and historical control data are obtained, including:

[0091] Determine the type of induced draft fan based on the current basic information of the induced draft fan;

[0092] Obtain several historical monitoring logs throughout the entire lifecycle of the current induced draft fan category, and extract historical initial speed data, historical initial vibration data, and historical initial control data from multiple identical historical monitoring nodes in each historical monitoring log;

[0093] The historical initial rotation speed data, historical initial vibration data, and historical initial control data at each historical monitoring node are preprocessed, including noise removal, missing value filling, and standardization.

[0094] Historical speed data, historical vibration data, and historical control data are obtained based on the preprocessed historical initial speed data, historical initial vibration data, and historical initial control data.

[0095] In this embodiment, the basic information of the induced draft fan includes core structural parameters (such as impeller parameters, shaft parameters, and shell structure parameters that affect the natural frequency), material properties, component configuration, etc. The category to which the current induced draft fan belongs is determined by the basic information, i.e., the induced draft fan category, and several historical monitoring logs of the corresponding induced draft fan category throughout its entire life cycle are extracted to lay the foundation for subsequently determining the historical resonance interval and historical control strategy.

[0096] In some embodiments of this application, several historical resonance intervals are determined based on historical rotational speed data, historical vibration data, and historical control data, including:

[0097] Feature extraction is performed on the historical rotational speed data and historical vibration data at each identical historical monitoring node to obtain historical rotational speed features and historical vibration features.

[0098] Based on preset capture conditions, the historical rotation speed characteristics and historical vibration characteristics at multiple consecutive identical historical monitoring nodes are captured. The preset capture conditions include a first preset capture condition, a second preset capture condition, and a third preset capture condition.

[0099] Based on the capture results, several historical resonance intervals are determined, and each historical resonance interval is mapped with corresponding preset capture conditions and corresponding capture features.

[0100] The historical resonance intervals are classified according to the preset capture conditions and capture features to obtain several historical resonance intervals of the same category, and a group of historical resonance intervals for each category is constructed.

[0101] Analyze several historical resonance intervals in each category of historical resonance interval group to determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group.

[0102] In this embodiment, feature extraction of historical speed data and historical vibration data refers to extracting the change values ​​of historical speed data and historical vibration data. That is, historical speed features refer to historical speed adjustment values, historical speed abrupt change values, etc., and historical vibration features refer to historical vibration amplitude change values, historical vibration phase abrupt change values, etc.

[0103] In this embodiment, the first preset capture condition refers to the vibration amplitude at a certain speed point (or a small range) suddenly exceeding a preset threshold (e.g., more than 50% higher than the vibration amplitude of adjacent speeds) during the speed change process. The second preset capture condition refers to the significant increase in the amplitude ratio of "1x speed frequency" in the vibration spectrum at that speed (e.g., exceeding 60%), and the absence of other obvious fault frequencies (e.g., bearing or gear characteristic frequencies). The third preset capture condition is that the vibration amplitude shows a "sudden increase and decrease" trend as the speed "enters and exits the range" (e.g., when the speed increases from 1300 r / min to 1400 r / min, the vibration jumps from 0.3 mm / s to 1.2 mm / s, and then drops back to 0.4 mm / s when the speed increases to 1500 r / min).

[0104] In this embodiment, the capture feature corresponds to the corresponding preset capture condition. For example, when the first preset capture condition is met, the capture feature is the change in vibration amplitude during the rotational speed change process.

[0105] In this embodiment, the same historical resonance interval group refers to the historical resonance intervals subjected to the same vibration amplitude mutation value, spectrum ratio value, and other conditions. The first resonance interval refers to the standard resonance interval in the historical resonance interval group, that is, the resonance interval that is not affected by relevant operating data such as load conditions and equipment aging. The second resonance interval refers to the resonance interval after the standard resonance interval has deviated due to the influence of relevant operating data such as load conditions and equipment aging.

[0106] In this embodiment, by determining the rotational speed characteristics and vibration characteristics, it is determined whether the preset capture conditions are triggered, the historical resonance interval under the corresponding characteristics is determined, and a historical resonance interval group is constructed. The relevant operating data, the first data interval, and several second data intervals that affect the historical resonance interval are determined, laying the foundation for constructing an accurate resonance interval analysis model and improving the accuracy of subsequent resonance intervals.

[0107] In some embodiments of this application, determining a first resonance interval, a first data interval set, several second resonance intervals, and a corresponding second data interval set corresponding to a historical resonance interval group includes:

[0108] By comparing and analyzing multiple historical resonance intervals in the same historical resonance interval group, the difference in the first resonance interval between different historical resonance intervals is obtained.

[0109] Clusters corresponding to historical resonance interval groups are determined based on the differences in several first resonance intervals, and the clusters include several historical resonance intervals;

[0110] The first resonance interval is generated based on several historical resonance intervals in the cluster.

[0111] Obtain several historical operational data points and corresponding historical data intervals at the historical monitoring nodes corresponding to the historical resonance intervals in the clusters;

[0112] Generate the first data interval of the corresponding historical operation data based on multiple historical data intervals corresponding to the same historical operation data;

[0113] Each historical resonance interval that does not belong to a cluster within the same historical resonance interval group is designated as a second resonance interval.

[0114] Acquire several historical operating data points and corresponding historical data intervals at the historical monitoring nodes corresponding to each resonance interval in the same historical resonance interval group, and set the historical data interval as the second data interval of the corresponding historical operating data.

[0115] The first data interval of the same historical operating data is compared with the second data interval of the historical operating data corresponding to each second resonance interval to obtain the data interval difference.

[0116] A data interval difference matrix corresponding to the historical running data is constructed by using multiple data interval differences of the same historical running data. The data interval difference matrix includes several data interval differences sorted by size, and each data interval difference is mapped to a corresponding second resonance interval difference.

[0117] Determine whether there is a dependency relationship between the difference values ​​of several data intervals in the data interval difference matrix and the difference values ​​of several second resonance intervals that are mapped. If yes, calculate the dependency coefficient; if no, remove the corresponding historical running data.

[0118] Calculate the number of data interval differences in the same historical data interval difference matrix that exceed a preset data interval difference threshold;

[0119] The correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than the preset data interval difference threshold and the dependency coefficient.

[0120] Historical operational data with a correlation coefficient greater than a preset correlation coefficient threshold are identified as relevant operational data.

[0121] Construct a first data interval set based on a first data interval of several related data;

[0122] A set of second data intervals is constructed based on several second data intervals of several related data, and each second data interval is mapped to a corresponding second resonance interval.

[0123] In this embodiment, the resonance interval difference is a comprehensive difference obtained by quantifying the differences in interval length, interval center position, and overlap of different historical resonance intervals.

[0124] In this embodiment, the resonance interval difference value between any two historical resonance intervals in the cluster is lower than the preset resonance interval difference threshold. The preset data difference threshold can be set according to the average total resonance interval difference value in the same historical resonance interval group, for example, 1 / 2 of the average total interval difference value.

[0125] In this embodiment, the first resonance interval is obtained by averaging the endpoints of multiple historical resonance intervals in the cluster. The second resonance interval difference is calculated by comparing the resonance interval difference between the historical resonance intervals that do not belong to the cluster and the first resonance interval in the same historical resonance interval group.

[0126] In this embodiment, historical operating data includes historical load conditions, historical lifespan data, and historical fault data, etc. Historical lifespan data can be used to determine the degree of equipment aging.

[0127] In this embodiment, the first data interval refers to the interval union of several historical data intervals corresponding to historical monitoring nodes at the historical resonance interval in the cluster.

[0128] In this embodiment, the data interval difference matrix includes two columns, one of which is a number of data interval differences, and the other column is a number of second resonance interval differences between the second resonance interval and the first resonance interval mapped by the data interval differences.

[0129] In this embodiment, determining whether a dependency exists means plotting the data interval differences arranged in order in the data interval difference matrix and the mapped second resonance interval differences onto the same blank dot plot to obtain a scatter plot of the data interval differences and a scatter plot of the second resonance interval differences. If the scatter plots show a linear trend, a dependency exists; if they do not show a linear trend, a dependency does not exist. Historical running data of the data interval difference matrix is ​​then removed.

[0130] In this embodiment, the preset data interval difference threshold is set based on the statistical characteristics of the data interval difference in the data interval difference matrix. Specifically, it can be set to a certain multiple of the mean of the data interval difference, such as 1.5 times, to ensure that only data intervals that significantly deviate from the cluster are considered to have large differences.

[0131] In some embodiments of this application, the correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than a preset data interval difference threshold and the dependency coefficient, including:

[0132] The formula for calculating the correlation coefficient is:

[0133] ;

[0134] Where H is the correlation coefficient, n1 is the number of data interval differences in the data interval difference matrix that are greater than the preset data interval difference threshold, n2 is the total number of data interval differences in the data interval difference matrix, and ji is the difference of the i-th data interval. Let gi be the mean of the data interval differences in the data interval difference matrix, and gi be the difference of the i-th second resonance interval. It represents the mean of the second resonance interval difference in the data interval difference matrix.

[0135] In this embodiment,

[0136] The correlation coefficient is the coefficient of dependence. The range of the correlation coefficient is (0, 1). The closer it is to 1, the larger the correlation coefficient is, and vice versa.

[0137] In this embodiment, relevant operating data that affects the resonance interval are selected by calculating the correlation coefficient. Based on the first data interval and several second data intervals of the relevant operating data, the accurate first resonance interval and several second resonance intervals are determined, thereby improving the accuracy of the resonance interval judgment and avoiding the resonance interval frequency, providing a key basis for the subsequent construction of the control analysis model.

[0138] In some embodiments of this application, a resonance interval analysis model is constructed, including:

[0139] Historical rotation speed data, historical vibration data, and the corresponding first data interval set are used as the first training input data, and the corresponding first resonance interval is used as the first training output data.

[0140] Historical rotation speed data, historical vibration data, and the corresponding second data interval set are used as the second training input data, and the corresponding second resonance interval is used as the second training output data.

[0141] The neural network is trained based on the first training input data, the first training output data, the second training input data, and the second training output data to obtain the resonance interval analysis model.

[0142] In some embodiments of this application, the correlation between historical boiler operating data and historical control parameters is determined, and the target control parameters are determined by combining the first control parameter interval corresponding to each historical resonance interval, and a control analysis model is constructed, including:

[0143] The load of the induced draft fan is set into a first preset load segment, a second preset load segment, a third preset load segment, and several preset load points in each preset load segment;

[0144] Extract historical load data, historical control data, and historical boiler operation data from historical monitoring logs;

[0145] Based on the correspondence between historical load data and preset load segments, a first set of analytical data for the first preset load segment, a second set of analytical data for the second preset load segment, and a third set of analytical data for the third preset load segment are constructed. Each set of analytical data includes several subsets of analytical data, and each subset of analytical data corresponds to a preset load point.

[0146] Correlation analysis is performed on each historical boiler operation data and historical control data in each subset of analysis data to obtain the correlation relationship between each historical boiler operation data and historical control data.

[0147] The correlation includes whether a correlation exists and the historical second influence coefficient;

[0148] Historical boiler operation data that are correlated and whose historical second influence coefficient is greater than the preset influence coefficient threshold are set as key parameters of historical boilers.

[0149] The target control parameters are determined based on the historical key boiler parameters and their corresponding relationships, and it is then determined whether the target control parameters are within the corresponding first control parameter range.

[0150] If not, calculate the target air volume based on boiler operating requirements and target control parameters;

[0151] Construct control parameter-airflow-efficiency curves for each subset of analytical data;

[0152] Map the target control parameters and target air volume to the corresponding control parameter-air volume-efficiency curve to obtain the target efficiency value;

[0153] Determine whether the target efficiency value is within the preset high efficiency range. If so, use the preset load point, historical boiler key parameters, and the corresponding target control parameters as a training parameter group. The preset load point and historical boiler key parameters are the training input parameters, and the corresponding target control parameters are the training output parameters.

[0154] Generate several sets of training parameters for each subset of the analysis data;

[0155] The control analysis model is obtained by training a neural network based on several sets of training parameters for each subset of the analysis data.

[0156] In this embodiment, historical boiler operating data refers to important operating data that characterizes the boiler's operating status, such as steam pressure, temperature, flow rate, and furnace negative pressure. Historical boiler key parameters refer to data that are highly correlated with rotational speed or boiler opening degree.

[0157] In this embodiment, the historical control parameters include rotational speed and opening degree.

[0158] In this embodiment, the first preset load segment refers to 30%-50% low load, the second preset load segment refers to 50%-80% medium load, and the third preset load segment refers to 80%-100% high load. Each load segment is further subdivided into 5 preset load points. For example, the preset load points of the first preset load segment are 30, 35, 40, 45, and 50, thereby ensuring coverage of the entire operating range of the induced draft fan.

[0159] In this embodiment, the analysis dataset is constructed based on historical control data and historical boiler operation data at the corresponding preset load segment of historical load data, and the analysis data subset is constructed based on historical control data and historical boiler operation data at each of the same preset load points in the preset load segment of historical load data.

[0160] In this embodiment, a first control parameter interval is determined based on each historical resonance interval. The first control parameter interval refers to the speed interval with the same frequency as the resonance interval, and the first control parameter interval corresponding to the target control parameter is the first control parameter interval of the historical resonance interval determined based on the relevant operating data at the same historical node of the historical boiler key parameters.

[0161] In this embodiment, when the speed and resonant frequency are not in the first control parameter range, the resonance risk caused by the repetition of the speed and resonant frequency during the operation of the induced draft fan can be effectively avoided, thereby ensuring the safe and stable operation of the induced draft fan and related equipment and extending the service life of the equipment.

[0162] In this embodiment, the target control parameters are calculated based on historical load conditions, historical key boiler parameters, and their correlations. For example, when the boiler steam pressure decreases by 10 MPa and the steam flow rate increases or decreases by 50 t / h (corresponding to 65% of the rated load), the target induced draft fan speed is 1200 r / min and the opening degree is +10°, based on the parameter weighting coefficients in the correlation and the historical second influence coefficient.

[0163] In this embodiment, boiler operating requirements refer to boiler flue gas exhaust requirements and stable operation requirements. For example, based on boiler operating requirements and the boiler heat balance calculation formula, the required flue gas exhaust volume is calculated as follows: A decrease in boiler steam pressure of 10 MPa, a change in steam flow rate of 50 t / h (corresponding to 65% rated load), a target induced draft fan speed of 1200 r / min, and an opening angle of +10°. .

[0164] In this embodiment, the preset high efficiency range is 80%-100%.

[0165] In this embodiment, the control parameter-airflow-efficiency curve includes the speed-airflow-efficiency curve and the opening degree-airflow-efficiency curve. When the load is low, the opening degree-airflow-efficiency curve is used first, and when the load is medium to high, the speed-airflow-efficiency curve is used first.

[0166] In this embodiment, several historical speeds are selected from each preset load point in the historical monitoring log. For example, if the historical speed is 1200 r / min, the historical air volume and historical efficiency values ​​at historical adjustment openings from -30° to +25° are selected. With the opening as the horizontal axis, the air volume as the vertical axis, and the efficiency value as an auxiliary parameter (marked at the corresponding air volume), the opening-air volume-efficiency curves at different historical speeds are obtained, thereby constructing the opening-air volume-efficiency curve graph. The speed-air volume-efficiency curve graph is constructed in a similar way to the above method, and will not be described again here.

[0167] In this embodiment, by collecting and analyzing the historical control parameters of the induced draft fan at different load levels and historical boiler operating data, the changes in the induced draft fan control parameters and the operating efficiency of the induced draft fan are compared under different boiler operating data. This enables the boiler operating parameters to participate in the joint regulation of the fan, avoiding fan speed resonance while maximizing the fan operating efficiency at each load level.

[0168] In some embodiments of this application, correlation analysis is performed on each historical boiler operation data and historical control data in each subset of the analysis data, including:

[0169] A single-factor correlation analysis was performed on each historical boiler operation data and historical control data in each subset of the analysis data to obtain the historical first influence coefficient of each historical boiler operation data on the historical control data.

[0170] Historical boiler operation data with historical influence coefficients greater than the preset influence coefficient threshold are selected from each subset of analysis data and used as input variables. The corresponding historical control data are used as output variables to generate variable training sets and variable test sets.

[0171] Establish a multiple linear regression equation;

[0172] The historical second influence coefficient of each historical boiler operation data point selected based on the multiple linear regression equation is calculated for historical control data.

[0173] In this embodiment, the historical first influence coefficient refers to the analysis of the relationship between a single historical boiler operating data and historical control data while keeping other parameters constant. For example, when the steam flow rate increases from 300 t / h to 670 t / h, the changing trends of the induced draft fan speed and opening are recorded. It is found that for every 100 t / h increase in steam flow rate, the induced draft fan speed needs to be increased by an average of 120 r / min, and the opening needs to be increased by an average of 5°. The historical first influence coefficient is calculated based on the ratio of the increase in steam flow rate to the increase in speed and the increase in opening, respectively. When the increase in steam flow rate is smaller but the resulting increase in speed and opening is larger, the historical first influence coefficient is larger, and vice versa.

[0174] In this embodiment, the historical control data are the induced draft fan speed and the blade opening.

[0175] In this embodiment, the selected historical boiler operation data includes steam flow rate x1, furnace negative pressure x2, flue gas oxygen content x3, and steam pressure x4. These parameters are used as input variables, while induced draft fan speed y1 and blade opening y2 are used as output variables. Multiple variation samples are constructed as a variable training set, and regression equations are built. The speed regression equation is: y1 = a0 + a1x1 + a2x2 + a3x3 + a4x4, and the blade opening regression equation is: y2 = b0 + b1x1 + b2x2 + b3x3 + b4x4. a0 - b4 (i.e., the historical second influence coefficient) is calculated using the variable training set, and accuracy is checked based on the variable test set to improve the accuracy of the historical second influence coefficient, laying a data foundation for the subsequent construction of the control analysis model.

[0176] In this embodiment, the key parameters of the historical boiler are determined based on the historical second influence coefficient and the corresponding parameter weight coefficients are set, laying the foundation for the subsequent calculation of the target control parameters. The target control parameters refer to the speed or opening degree that meets the requirements of stable boiler operation and flue gas discharge.

[0177] In some embodiments of this application, determining whether to generate a control command based on a resonance interval analysis model and a control analysis model includes:

[0178] Collect real-time relevant operating data, real-time speed data, real-time vibration data, and real-time key boiler parameters;

[0179] Based on the resonance analysis model, real-time relevant operating data, real-time rotational speed data, and real-time vibration data are analyzed to determine the real-time resonance range;

[0180] The corresponding real-time first control parameter range is determined based on the real-time resonance range.

[0181] Extract the real-time load points from the real-time relevant operating data, and input the real-time load points and key real-time boiler parameters into the control analysis model to determine the real-time target control parameters;

[0182] Determine whether the real-time target control parameters are within the range of the first real-time control parameters. If not, generate control commands according to the real-time target control parameters.

[0183] If the real-time target control parameters are fine-tuned, and the efficiency value corresponding to the fine-tuned real-time target control parameters is within the preset high-efficiency range, control commands are generated according to the fine-tuned real-time target control parameters.

[0184] In this embodiment, the real-time first control parameter range refers to the speed range in which the automated control system avoids the real-time resonance frequency range, ensuring that the operating speed of the induced draft fan does not overlap with the resonance frequency, thus avoiding the risk of resonance.

[0185] In this embodiment, the real-time target control parameter refers to the control parameter that ensures the normal operation of the boiler and has high operating efficiency.

[0186] In some embodiments of this application, such as Figure 2 As shown, it also includes an automated control method for induced draft fans based on DCS:

[0187] Step S201: Obtain historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data;

[0188] Step S202: Based on historical rotational speed data, historical vibration data, and historical control data, determine several historical resonance intervals and construct a resonance interval analysis model;

[0189] Step S203: Determine the correlation between historical boiler operating data and historical control parameters, and combine the first control parameter interval corresponding to each historical resonance interval to determine the target control parameters and construct a control analysis model;

[0190] Step S204: Determine whether to generate control commands based on the resonance interval analysis model and the control analysis model.

[0191] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. An automated control system for induced draft fans based on DCS, characterized in that, include: The acquisition module is used to acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data. The first construction module is used to determine several historical resonance intervals based on historical rotational speed data, historical vibration data, and historical control data, and to construct a resonance interval analysis model. The second construction module is used to determine the correlation between historical boiler operating data and historical control parameters, and, in conjunction with the first control parameter interval corresponding to each historical resonance interval, determine the target control parameters and construct a control analysis model. The control module is used to determine whether to generate control commands based on the resonance interval analysis model and the control analysis model. Based on historical rotational speed data, historical vibration data, and historical control data, several historical resonance intervals were determined, including: Feature extraction is performed on the historical rotational speed data and historical vibration data at each identical historical monitoring node to obtain historical rotational speed features and historical vibration features. Based on preset capture conditions, the historical rotation speed characteristics and historical vibration characteristics at multiple consecutive identical historical monitoring nodes are captured. The preset capture conditions include a first preset capture condition, a second preset capture condition, and a third preset capture condition. Based on the capture results, several historical resonance intervals are determined, and each historical resonance interval is mapped with corresponding preset capture conditions and corresponding capture features. The historical resonance intervals are classified according to the preset capture conditions and capture features to obtain several historical resonance intervals of the same category, and a group of historical resonance intervals for each category is constructed. Analyze several historical resonance intervals in each category of historical resonance interval group to determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group. Determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group, including: By comparing and analyzing multiple historical resonance intervals in the same historical resonance interval group, the difference in the first resonance interval between different historical resonance intervals is obtained. Clusters corresponding to historical resonance interval groups are determined based on the differences in several first resonance intervals, and the clusters include several historical resonance intervals; The first resonance interval is generated based on several historical resonance intervals in the cluster. Obtain several historical operational data points and corresponding historical data intervals at the historical monitoring nodes corresponding to the historical resonance intervals in the clusters; Generate the first data interval of the corresponding historical operation data based on multiple historical data intervals corresponding to the same historical operation data; Each historical resonance interval that does not belong to a cluster within the same historical resonance interval group is designated as a second resonance interval. Acquire several historical operating data points and corresponding historical data intervals at the historical monitoring nodes corresponding to each resonance interval in the same historical resonance interval group, and set the historical data interval as the second data interval of the corresponding historical operating data. The first data interval of the same historical operating data is compared with the second data interval of the historical operating data corresponding to each second resonance interval to obtain the data interval difference. A data interval difference matrix corresponding to the historical running data is constructed by using multiple data interval differences of the same historical running data. The data interval difference matrix includes several data interval differences sorted by size, and each data interval difference is mapped to a corresponding second resonance interval difference. Determine whether there is a dependency relationship between the difference values ​​of several data intervals in the data interval difference matrix and the difference values ​​of several second resonance intervals that are mapped. If yes, calculate the dependency coefficient; if no, remove the corresponding historical running data. Calculate the number of data interval differences in the same historical data interval difference matrix that exceed a preset data interval difference threshold; The correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than the preset data interval difference threshold and the dependency coefficient. Historical operational data with a correlation coefficient greater than a preset correlation coefficient threshold are identified as relevant operational data. Construct a first data interval set based on a first data interval of several related data; A set of second data intervals is constructed based on several second data intervals of several related data, and each second data interval is mapped to a corresponding second resonance interval.

2. The DCS-based automated control system for induced draft fans as described in claim 1, characterized in that, Acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data, including: Determine the type of induced draft fan based on the current basic information of the induced draft fan; Obtain several historical monitoring logs throughout the entire lifecycle of the current induced draft fan category, and extract historical initial speed data, historical initial vibration data, and historical initial control data from multiple identical historical monitoring nodes in each historical monitoring log; The historical initial rotation speed data, historical initial vibration data, and historical initial control data at each historical monitoring node are preprocessed, including noise removal, missing value filling, and standardization. Historical speed data, historical vibration data, and historical control data are obtained based on the preprocessed historical initial speed data, historical initial vibration data, and historical initial control data.

3. The DCS-based automated control system for induced draft fans as described in claim 2, characterized in that, The correlation coefficient of the corresponding historical operational data is calculated based on the number of data interval differences exceeding a preset data interval difference threshold in the data interval difference matrix of the same historical operational data, and the dependency coefficient, including: The formula for calculating the correlation coefficient is: ； Where H is the correlation coefficient, n1 is the number of data interval differences in the data interval difference matrix that are greater than the preset data interval difference threshold, n2 is the total number of data interval differences in the data interval difference matrix, and ji is the difference of the i-th data interval. Let gi be the mean of the data interval differences in the data interval difference matrix, and gi be the difference of the i-th second resonance interval. It represents the mean of the second resonance interval difference in the data interval difference matrix.

4. The DCS-based automated control system for induced draft fans as described in claim 3, characterized in that, Constructing a resonance interval analysis model, including: Historical rotation speed data, historical vibration data, and the corresponding first data interval set are used as the first training input data, and the corresponding first resonance interval is used as the first training output data. Historical rotation speed data, historical vibration data, and the corresponding second data interval set are used as the second training input data, and the corresponding second resonance interval is used as the second training output data. The neural network is trained based on the first training input data, the first training output data, the second training input data, and the second training output data to obtain the resonance interval analysis model.

5. The DCS-based automated control system for induced draft fans as described in claim 4, characterized in that, The correlation between historical boiler operating data and historical control parameters is determined, and the target control parameters are determined by combining the first control parameter interval corresponding to each historical resonance interval. A control analysis model is then constructed, including: The load of the induced draft fan is set into a first preset load segment, a second preset load segment, a third preset load segment, and several preset load points in each preset load segment; Extract historical load data, historical control data, and historical boiler operation data from historical monitoring logs; Based on the correspondence between historical load data and preset load segments, a first set of analytical data for the first preset load segment, a second set of analytical data for the second preset load segment, and a third set of analytical data for the third preset load segment are constructed. Each set of analytical data includes several subsets of analytical data, and each subset of analytical data corresponds to a preset load point. Correlation analysis is performed on each historical boiler operation data and historical control data in each subset of analysis data to obtain the correlation relationship between each historical boiler operation data and historical control data. The correlation includes whether a correlation exists and the historical second influence coefficient; Historical boiler operation data that are correlated and whose historical second influence coefficient is greater than the preset influence coefficient threshold are set as key parameters of historical boilers. The target control parameters are determined based on the historical key boiler parameters and their corresponding relationships, and it is then determined whether the target control parameters are within the corresponding first control parameter range. If not, calculate the target air volume based on boiler operating requirements and target control parameters; Construct control parameter-airflow-efficiency curves for each subset of analytical data; Map the target control parameters and target air volume to the corresponding control parameter-air volume-efficiency curve to obtain the target efficiency value; Determine whether the target efficiency value is within the preset high efficiency range. If so, use the preset load point, historical boiler key parameters, and the corresponding target control parameters as a training parameter group. The preset load point and historical boiler key parameters are the training input parameters, and the corresponding target control parameters are the training output parameters. Generate several sets of training parameters for each subset of the analysis data; The control analysis model is obtained by training a neural network based on several sets of training parameters for each subset of the analysis data.

6. The DCS-based automated control system for induced draft fans as described in claim 5, characterized in that, Perform correlation analysis on each historical boiler operation data and historical control data in each subset of the analysis data, including: A single-factor correlation analysis was performed on each historical boiler operation data and historical control data in each subset of the analysis data to obtain the historical first influence coefficient of each historical boiler operation data on the historical control data. Historical boiler operation data with historical influence coefficients greater than the preset influence coefficient threshold are selected from each subset of analysis data and used as input variables. The corresponding historical control data are used as output variables to generate variable training sets and variable test sets. Establish a multiple linear regression equation; The historical second influence coefficient of each historical boiler operation data point selected based on the multiple linear regression equation is calculated for historical control data.

7. The DCS-based automated control system for induced draft fans as described in claim 6, characterized in that, The determination of whether to generate control commands is based on the resonance interval analysis model and the control analysis model, including: Collect real-time relevant operating data, real-time speed data, real-time vibration data, and real-time key boiler parameters; Based on the resonance analysis model, real-time relevant operating data, real-time rotational speed data, and real-time vibration data are analyzed to determine the real-time resonance range; The corresponding real-time first control parameter range is determined based on the real-time resonance range. Extract the real-time load points from the real-time relevant operating data, and input the real-time load points and key real-time boiler parameters into the control analysis model to determine the real-time target control parameters; Determine whether the real-time target control parameters are within the range of the first real-time control parameters. If not, generate control commands according to the real-time target control parameters. If the real-time target control parameters are fine-tuned, and the efficiency value corresponding to the fine-tuned real-time target control parameters is within the preset high-efficiency range, control commands are generated according to the fine-tuned real-time target control parameters.

8. An automated control method for induced draft fans based on DCS, characterized in that, include: Acquire historical speed data of the induced draft fan, historical vibration data of several key locations, and historical control data; Based on historical rotational speed data, historical vibration data, and historical control data, several historical resonance intervals were determined, and a resonance interval analysis model was constructed. The correlation between historical boiler operating data and historical control parameters is determined, and the target control parameters are determined by combining the first control parameter interval corresponding to each historical resonance interval, and a control analysis model is constructed. Determine whether to generate a control command based on the resonance interval analysis model and the control analysis model; Based on historical rotational speed data, historical vibration data, and historical control data, several historical resonance intervals were determined, including: Feature extraction is performed on the historical rotational speed data and historical vibration data at each identical historical monitoring node to obtain historical rotational speed features and historical vibration features. Based on preset capture conditions, the historical rotation speed characteristics and historical vibration characteristics at multiple consecutive identical historical monitoring nodes are captured. The preset capture conditions include a first preset capture condition, a second preset capture condition, and a third preset capture condition. Based on the capture results, several historical resonance intervals are determined, and each historical resonance interval is mapped with corresponding preset capture conditions and corresponding capture features. The historical resonance intervals are classified according to the preset capture conditions and capture features to obtain several historical resonance intervals of the same category, and a group of historical resonance intervals for each category is constructed. Analyze several historical resonance intervals in each category of historical resonance interval group to determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group. Determine the first resonance interval, the first data interval set, several second resonance intervals, and the corresponding second data interval set for the corresponding historical resonance interval group, including: By comparing and analyzing multiple historical resonance intervals in the same historical resonance interval group, the difference in the first resonance interval between different historical resonance intervals is obtained. Clusters corresponding to historical resonance interval groups are determined based on the differences in several first resonance intervals, and the clusters include several historical resonance intervals; The first resonance interval is generated based on several historical resonance intervals in the cluster. Obtain several historical operational data points and corresponding historical data intervals at the historical monitoring nodes corresponding to the historical resonance intervals in the clusters; Generate the first data interval of the corresponding historical operation data based on multiple historical data intervals corresponding to the same historical operation data; Each historical resonance interval that does not belong to a cluster within the same historical resonance interval group is designated as a second resonance interval. Acquire several historical operating data points and corresponding historical data intervals at the historical monitoring nodes corresponding to each resonance interval in the same historical resonance interval group, and set the historical data interval as the second data interval of the corresponding historical operating data. The first data interval of the same historical operating data is compared with the second data interval of the historical operating data corresponding to each second resonance interval to obtain the data interval difference. A data interval difference matrix corresponding to the historical running data is constructed by using multiple data interval differences of the same historical running data. The data interval difference matrix includes several data interval differences sorted by size, and each data interval difference is mapped to a corresponding second resonance interval difference. Determine whether there is a dependency relationship between the difference values ​​of several data intervals in the data interval difference matrix and the difference values ​​of several second resonance intervals that are mapped. If yes, calculate the dependency coefficient; if no, remove the corresponding historical running data. Calculate the number of data interval differences in the same historical data interval difference matrix that exceed a preset data interval difference threshold; The correlation coefficient of the corresponding historical running data is calculated based on the number of data interval differences in the data interval difference matrix of the same historical running data that are greater than the preset data interval difference threshold and the dependency coefficient. Historical operational data with a correlation coefficient greater than a preset correlation coefficient threshold are identified as relevant operational data. Construct a first data interval set based on a first data interval of several related data; A set of second data intervals is constructed based on several second data intervals of several related data, and each second data interval is mapped to a corresponding second resonance interval.

Images