A method and system for intelligently improving the response speed of AGC in thermal power units
By quantifying the energy characteristics of coal and adaptively adjusting PID parameters, the problems of slow response speed and low control accuracy of AGC in thermal power units have been solved, achieving rapid response and stable control to grid load fluctuations.
Patent Information
- Application Number
- CN202611287083.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-24
- Publication Date
- 2026-11-10
AI Technical Summary
Existing thermal power units have slow AGC response speed and low control accuracy, making it difficult to adapt to grid load fluctuations. This is mainly due to the inability to perceive coal quality fluctuations in real time and the response lag and control conflicts caused by fixed PID parameters.
By acquiring standardized operating data of thermal power units, a soft measurement model is constructed to quantify the lower heating value of coal. A neural network model is used to identify the theoretical response delay and inertial time constant, and feedforward compensation and adaptive adjustment PID parameters are generated to achieve rapid tracking under dynamic operating conditions and optimal target finding under steady-state operating conditions.
It significantly improves the response speed and control accuracy of AGC, ensuring the stability and economy of the unit under grid load fluctuations, and solving the problem of response lag caused by the inability to detect coal quality fluctuations and fixed PID parameters in traditional control.
Smart Images

Figure CN123067083A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of AGC (Automatic Guided Vehicle) technology for thermal power units, and more specifically to a method and system for intelligently improving the response speed of AGC in thermal power units. Background Technology
[0002] The response speed and control accuracy of Automatic Generation Control (AGC) have become key indicators for measuring unit flexibility and grid support capabilities. However, in actual operation, frequent and significant fluctuations in the quality of coal fed into the furnace are a major bottleneck restricting AGC performance.
[0003] Traditional coal quality analysis relies on offline sampling and testing, resulting in data acquisition delays and making it unsuitable for real-time control. Furthermore, the coal feed control loop in a distributed control system (DCS) uses volume or mass as a benchmark, failing to consider real-time changes in coal energy density (BTU), leading to poor coal quality adaptability. When coal quality deteriorates and calorific value decreases, the actual input energy corresponding to the same coal feed command decreases, causing a slow rise in boiler main steam pressure and preventing the unit power from quickly tracking AGC commands, resulting in response lag. Conversely, when coal quality improves, overshoot is easily triggered, causing a decline in control quality.
[0004] Meanwhile, existing advanced control algorithms such as fuzzy control and predictive control, as well as traditional PID control, are mostly based on the design of coal quality or specific operating conditions for parameter tuning and strategy design. They lack the ability to make online adaptive adjustments based on real-time coal quality and energy characteristics, resulting in insufficient adaptability of response characteristics. Under complex operating conditions such as low load operation and variable coal quality, a single control mode cannot simultaneously ensure rapid response, operational stability, and fuel economy. This leads to slow AGC response speed, low control accuracy, and poor unit operation stability, making it difficult to adapt to the increasingly frequent operational demands of grid load fluctuations. Therefore, existing technologies have shortcomings. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for intelligently improving the AGC response speed of thermal power units. By deriving the received low calorific value of coal, the dynamic response characteristics of the unit are quantified. Under dynamic operating conditions, feedforward compensation and adaptive adjustment of PID parameters are generated, and under steady-state operating conditions, the current optimal operating target value is obtained. This method features fast AGC response speed and strong adaptability to operating conditions.
[0006] To achieve the above objectives, the present invention provides the following technical solution:
[0007] A method for intelligently improving the response speed of AGC (Automatic Guided Vehicle) in thermal power units includes:
[0008] Obtain standardized operating data of thermal power units, and obtain the lower heating value of coal based on the received basis based on the constructed soft measurement model;
[0009] Based on the received basis lower heating value of the coal and the pre-trained neural network model, the theoretical response delay and inertial time constant are obtained.
[0010] The rate of change of operating condition characteristics is calculated based on the standardized operating data to determine the current operating condition, which includes dynamic operating condition and steady-state operating condition.
[0011] If the current operating condition is a dynamic condition, switch to PID fast tracking mode: generate feedforward compensation based on the received low calorific value of the coal quality, and adaptively adjust the PID parameters using the theoretical response delay and the inertial time constant. The feedforward compensation is used as the input for PID control, and the adaptive adjustment of the PID parameters is used for parameter optimization of PID control.
[0012] If the current operating condition is a steady-state condition, switch to steady-state optimization mode: find the current optimal operating target value from historical operating data based on the standardized operating data.
[0013] As a further improvement of the present invention, the step of acquiring standardized operating data of thermal power units and obtaining the lower heating value of coal based on the constructed soft measurement model includes:
[0014] Obtain raw operating data of thermal power units and perform standardized preprocessing to obtain standardized operating data;
[0015] Based on the standardized operating data, the boiler thermal efficiency and total losses were calculated.
[0016] A system error correction coefficient is introduced to correct the mechanism formula of the soft measurement model;
[0017] Based on the standardized operating data, boiler thermal efficiency, and input mechanism formulas for various total losses, the received lower heating value of coal is obtained.
[0018] As a further improvement of the present invention, the introduction of a system error correction coefficient to correct the mechanism formula of the soft measurement model includes:
[0019] The optimal system error correction coefficient is obtained by using historical operating data and fitting it through machine learning.
[0020] The optimal system error correction coefficient is substituted into the mechanism formula of the soft measurement model.
[0021] As a further improvement of the present invention, based on the received-based lower heating value of the coal and a pre-trained neural network model, the theoretical response delay and inertial time constant are obtained, including:
[0022] Based on the received basis lower heating value of the coal and standardized operating data, the input characteristics are obtained by normalization.
[0023] The input features are input into a pre-trained neural network model to obtain predicted values;
[0024] The predicted values are denormalized to obtain the theoretical response delay and inertial time constant.
[0025] As a further improvement of the present invention, the current operating condition is determined by calculating the rate of change of operating condition characteristics based on the standardized operating data, including:
[0026] Based on the standardized operating data, the rate of change of operating conditions is calculated, wherein the rate of change of operating conditions includes the rate of change of AGC commands, the rate of change of coal quality energy, and the rate of change of main steam pressure.
[0027] The operating condition is determined based on the rate of change of the operating condition characteristics, and the current operating condition is obtained.
[0028] As a further improvement of the present invention, feedforward compensation is generated based on the received lower calorific value of the coal, including:
[0029] Based on the preset step determination rules, the effective AGC instruction change amount is calculated;
[0030] Calculate the BTU correction factor and feedforward coal feed rate based on the effective AGC command change amount;
[0031] Based on the feedforward coal feed rate and the pre-calibrated unit air-coal correlation table, the primary air adjustment amount and the secondary air adjustment amount are obtained.
[0032] The secondary air adjustment amount is corrected based on the standardized operating data to obtain the corrected secondary air adjustment amount.
[0033] As a further improvement of the present invention, the PID parameters are adaptively adjusted using the theoretical response delay and the inertial time constant, including:
[0034] Construct a correlation model between PID parameters and theoretical response delay and inertial time constant;
[0035] Substituting the theoretical response delay and the inertial time constant into the correlation model, the adjusted PID parameters are obtained.
[0036] As a further improvement of the present invention, it also includes: stable combustion control of the boiler under dynamic operating condition boundary conditions.
[0037] As a further improvement of the present invention, finding the current optimal operating target value from historical operating data based on the standardized operating data includes:
[0038] Based on the standardized operational data, a local wind and coal scheduling interval is constructed;
[0039] Based on the aforementioned local wind and coal dispatching interval, a steady-state high-efficiency sample set is selected from historical operating data;
[0040] Construct a multi-objective evaluation function;
[0041] The sample data with the smallest evaluation function value in the steady-state efficient sample set is selected to obtain the current optimal operating target value.
[0042] A smart system for improving the response speed of AGC (Automatic Guided Vehicle) in thermal power units, used to implement the aforementioned method for smartly improving the response speed of AGC in thermal power units, includes:
[0043] Coal quality characteristic soft measurement module: used to acquire standardized operating data of thermal power units and obtain the received basis lower heating value of coal based on the constructed soft measurement model;
[0044] Unit response lag measurement module: used to obtain the theoretical response delay and inertial time constant based on the received basis of the coal quality and the pre-trained neural network model;
[0045] Operating condition identification and mode switching module: used to calculate the rate of change of operating condition characteristics based on the standardized operating data and determine the current operating condition, wherein the operating condition includes dynamic operating condition and steady-state operating condition;
[0046] PID fast tracking control module: It generates feedforward compensation based on the received low calorific value of the coal and adaptively adjusts the PID parameters using the theoretical response delay and the inertial time constant. The feedforward compensation is used as the input for PID control, and the adaptive adjustment of the PID parameters is used for parameter optimization of PID control.
[0047] Steady-state optimization module: used to find the current optimal operating target value from historical operating data based on the standardized operating data.
[0048] The beneficial effects of this invention are as follows: By back-calculating the received basis of coal quality from operational data, coal quality fluctuations are transformed into controllable input variables. Then, based on coal quality information and a neural network model, the dynamic response characteristics of the unit are identified, and theoretical response delay and inertial time constant are output. Adaptive control strategies are derived for different operating conditions: Under dynamic conditions, feedforward compensation for coal feed rate and air volume is generated based on coal quality characteristics. Simultaneously, the identified theoretical response delay and inertial time constant are used to adaptively adjust PID parameters, achieving rapid AGC tracking and disturbance suppression. Under steady-state conditions, the optimal operating target value is quickly obtained based on historical operating data. This solves the problems of undetectable coal quality fluctuations and response lag caused by fixed PID parameters in traditional AGC control, as well as the control conflict between dynamic response and steady-state economy. While shortening load response delay and improving tracking accuracy, it ensures stable unit operation, effectively improving the unit's grid adaptability, operational stability, and economy. Attached Figure Description
[0049] Figure 1 This is a flowchart illustrating the steps of an intelligent method for improving the response speed of an AGC (Automatic Guided Vehicle) system in a thermal power unit according to the present invention.
[0050] Figure 2 This is a flowchart illustrating the steps of obtaining the received basis lower calorific value of coal based on the constructed soft measurement model of this invention.
[0051] Figure 3 A flowchart illustrating the steps involved in generating feedforward compensation for this invention;
[0052] Figure 4 This is a flowchart illustrating the steps involved in quickly finding the current optimal target value in this invention.
[0053] Figure 5 This is a schematic diagram of the structure of an intelligent system for improving the response speed of an AGC (Automatic Guided Vehicle) system for thermal power units according to the present invention. Detailed Implementation
[0054] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof.
[0055] The term "and / or" in the following text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.
[0056] Example 1:
[0057] like Figure 1As shown, this embodiment of the invention provides a method for intelligently improving the AGC response speed of thermal power units, including:
[0058] Obtain standardized operating data of thermal power units, and obtain the lower heating value of coal based on the received basis based on the constructed soft measurement model;
[0059] Based on the received lower heating value of coal and a pre-trained neural network model, the theoretical response delay and inertial time constant are obtained.
[0060] The rate of change of operating condition characteristics is calculated based on standardized operating data to determine the current operating condition, which includes dynamic operating condition and steady-state operating condition.
[0061] If the current operating condition is dynamic, switch to PID fast tracking mode: generate feedforward compensation based on the received low calorific value of coal, and adaptively adjust PID parameters using theoretical response delay and inertial time constant. Feedforward compensation is used as input for PID control, and adaptive adjustment of PID parameters is used for parameter optimization of PID control.
[0062] If the current operating condition is a steady state, switch to steady-state optimization mode: quickly find the current optimal operating target value from historical operating data based on standardized operating data.
[0063] Specifically, this method uses real-time operating data of thermal power units to back-calculate the received basis lower calorific value of coal, transforming the undetectable fluctuations in coal quality into measurable and controllable input variables. Then, based on coal quality information and neural networks, a dynamic response characteristic model of the unit is established, outputting theoretical response delay and inertial time constant. Operating conditions are identified through the rate of change of operating characteristics, and then adaptive control strategies are derived for different operating conditions: Under dynamic conditions, feedforward compensation for coal feed and air volume is generated based on coal quality characteristics. Simultaneously, the identified theoretical response delay and inertial time constant are used to adaptively adjust PID parameters, breaking the limitations of traditional fixed-parameter PID and achieving rapid tracking and disturbance suppression of AGC commands; Under steady-state conditions, a local air-coal scheduling interval is constructed based on current operating data, historical high-efficiency steady-state samples are screened, and the current optimal operating target value is found through a multi-objective evaluation function, balancing the unit's operating economy and stability.
[0064] This embodiment not only solves the key problems in traditional AGC control, such as the inability to detect coal quality fluctuations and the difficulty in adapting fixed PID parameters to changes in the dynamic characteristics of the unit, resulting in response lag, large tracking deviation, and pressure overshoot, but also solves the control objective conflict between rapid tracking in dynamic operating conditions and economic stability in steady-state operating conditions. While significantly shortening the unit load response delay and improving the tracking accuracy of AGC commands, it ensures reasonable combustion air distribution and stable main steam pressure and temperature, effectively improving the unit's adaptability to grid dispatch, operational stability, and economy.
[0065] Furthermore, this embodiment provides a step for obtaining standardized operating data of thermal power units and obtaining the received basis lower heating value of coal based on a constructed soft measurement model, including:
[0066] Obtain raw operating data of thermal power units and perform standardized preprocessing to obtain standardized operating data;
[0067] Based on standardized operating data, the boiler thermal efficiency and total losses were calculated.
[0068] A system error correction coefficient is introduced to correct the mechanism formula of the soft measurement model;
[0069] Based on standardized operating data, boiler thermal efficiency, and input mechanism formulas for various total losses, the received lower heating value of coal is obtained.
[0070] Specifically, the raw operating data of the thermal power unit is obtained, including: AGC commands, actual power generation, total coal feed, total air volume, main steam pressure, main steam temperature, and oxygen content. Among these, the AGC commands serve as the input reference for the unit's dynamic AGC response, quantifying the unit's tracking characteristics of grid load commands; actual power generation reflects the unit's actual output status and is the core output quantity used to extrapolate coal quality energy and quantify AGC response; total coal feed reflects the boiler fuel input, and combined with actual power generation, it can be used to inversely calculate coal quality energy density, serving as the core input for the soft measurement model; total air volume provides basic data for boiler efficiency calculation and combustion condition correction, assisting in improving coal quality soft measurement, and is the sum of primary and secondary air volumes. Primary air is used to feed pulverized coal into the furnace, while secondary air is used to supplement oxygen for combustion; main steam pressure reflects the energy conversion status of the boiler's steam-water system; main steam temperature is used to assist in correcting the energy quality characteristics of the main steam pressure; and oxygen content reflects the excess air coefficient corresponding to boiler combustion and is used to calculate boiler thermal efficiency. The data acquisition cycle is set uniformly, and the acquisition process adopts the industrial standard communication protocol to obtain the original data of the measurement points. The specific acquisition equipment is not limited. For example, setting the acquisition cycle to 0.5 seconds can meet the requirements of AGC real-time control, ensure data continuity and effectiveness, and control the computing load.
[0071] The collected raw operational data underwent standardized preprocessing, including: using 3D modeling... The criteria serve as the basic basis for data elimination, combined with the constraints of actual operating conditions of thermal power units, to avoid the erroneous elimination of valid data due to purely numerical criteria. When 3 When a value is initially diagnosed as an outlier according to the criteria, if it corresponds to the normal operating conditions of the unit, such as no mill tripping, no sudden damper adjustment, and no AGC command step change, it is determined to be a true outlier and deleted. If it corresponds to a sudden operating condition of the unit, such as a mill tripping causing a jump in the total coal feed, it is determined to be a false outlier, representing a fluctuation in operating conditions, and is retained. A linear interpolation method is used, employing three valid data points before and after the missing data as the interpolation benchmark to calculate the value of the missing data point and fill in a small number of missing data points. If data is missing for more than four consecutive acquisition cycles, a measurement point fault warning is triggered, and no filling is performed. The preprocessed raw data is used to generate standardized operating data, including standardized AGC commands. Actual power Total coal supply Total air volume Main steam pressure Main steam temperature Oxygen .
[0072] like Figure 2 As shown, the actual power generation in the standardized operating data Total coal supply Total air volume Main steam pressure Main steam temperature Oxygen The model is used for constructing a soft measurement model of calorific value. Based on the principles of energy conservation and material balance in thermal power unit boilers, it employs machine learning algorithms to correct system errors and extrapolates the received lower heating value of coal. The specific construction process is as follows:
[0073] Based on total air volume Main steam pressure Main steam temperature Oxygen The boiler thermal efficiency is calculated using a combination of the positive and negative balance methods as specified in the "Power Plant Boiler Performance Test Procedure". and various heat losses Specifically:
[0074] Based on oxygen content Calculate the excess air coefficient , , where 21 is the volume ratio of oxygen in the air;
[0075] Calculate flue gas heat loss: Flue gas heat loss is the loss caused by the heat carried away by the boiler flue gas. A commonly used empirical formula is: ,in, The average specific heat capacity of the exhaust gas. The exhaust gas temperature, The average specific heat capacity of the environment. For ambient temperature, The lower heating value of the base coal is the base calorific value. Used for unit conversion. The above parameters were obtained through data collection and analysis from boiler flue gas measuring points and environmental measuring points. These are commonly used parameters for calculating this loss, especially in relatively stable environments such as indoor spaces and boiler rooms. and Historical experience points can be used instead.
[0076] Calculating the heat loss from incomplete combustion: The heat loss from incomplete combustion is due to the heat carried away by unburned combustible gases in the flue gas, mainly CO. A commonly used empirical formula is... ,in, The percentage of carbon monoxide by volume in the flue gas is obtained from the flue gas monitoring instrument at the boiler flue gas measuring point. The empirical coefficient is determined based on the unit's combustion characteristics. Specifically, it is an empirical coefficient obtained by extracting steady-state historical operating sample data of the unit across the full load range based on the unit's design coal quality parameters, furnace structure dimensions, and burner arrangement. This coefficient is fixedly adapted to the unit's combustion characteristics and is used for dimensional conversion between combustible gas content, excess air coefficient, and heat loss ratio, remaining constant during online operation.
[0077] Calculating heat loss from incomplete mechanical combustion: Heat loss from incomplete mechanical combustion is the heat loss caused by some carbon in the fuel not participating in combustion and being discharged from the boiler with fly ash. A commonly used empirical formula is: ,in, The ash content is based on the received coal. This refers to the combustible content of fly ash. denoted as the calorific value of carbon. The above data were obtained through offline analysis of coal type and fly ash. For units with relatively stable coal types, historical empirical values can be used instead.
[0078] Calculate various heat losses and boiler thermal efficiency : , .
[0079] It should be noted that the above formulas are general engineering formulas for power plant boiler performance testing and efficiency calculation. This embodiment is only used for real-time calculation of boiler thermal efficiency and mechanism support for soft coal quality measurement, and does not involve any improvement to the existing standard formulas themselves.
[0080] Furthermore, this embodiment provides a step for introducing a system error correction coefficient to correct the mechanism formula of the soft measurement model, including:
[0081] The optimal system error correction coefficient is obtained by using historical operating data and fitting it through machine learning.
[0082] The optimal system error correction coefficient is substituted into the mechanism formula of the soft measurement model.
[0083] Furthermore, such as Figure 2 In the mechanism formula part of the modified soft sensor model, machine learning is used to fit the mechanism formula using historical operating data of the unit over the past three months. Specifically, a sample set is constructed based on the historical operating data. Each sample includes: total coal feed. Actual power Boiler thermal efficiency Total heat loss and offline acceptance to base low heat generation The fundamental formula for the original mechanism of boiler energy balance is: ,in, This is the conversion factor for boiler steam heat absorption, which can be calibrated under steady-state operating conditions using standard boiler thermal calculation methods. It represents the amount of boiler steam heat absorbed required to output a unit of electrical power, and is used to equivalently convert the heat absorbed by the boiler output steam into the corresponding heat of electrical power, thus unifying the energy dimensions on both sides of the formula. It should be noted that this basic formula is an uncorrected ideal theoretical formula and is not purely algebraically equivalent to subsequent corrected formulas. Because... The calculation is based on The obtained values, including thermal efficiency and heat loss, are both percentage-based relative units. However, the mechanistic model has inherent calculation biases. Therefore, a three-step extension and transformation is needed: unit unification, benchmark compensation, and error correction. First, multiply by 100 to unify the percentage and decimal operation units, and introduce... After completing the energy reference normalization compensation, a system error correction coefficient is introduced. The compensation model has inherent calculation biases, and the final formula is transformed into: Preferably, offline acceptance testing is used to measure the base low-grade heat generation. For the label value, the least squares method is used to correct the system error coefficient. By performing a fitting, the optimal system error correction coefficients are obtained. Therefore, the revised mechanism formula is: .
[0084] During the real-time simulation phase, based on the actual generated power collected in real time... Total coal supply and the boiler thermal efficiency calculated in real time. and various heat losses By substituting the modified mechanistic formula, the real-time coal quality received-based lower heating value can be derived. .
[0085] This embodiment collects and preprocesses operating data from thermal power units. Based on the traditional boiler energy conservation and thermal efficiency calculation theory, it introduces machine learning to correct system errors and constructs a soft measurement model of the received basis lower heating value of coal. This enables real-time online sensing of coal quality energy characteristics without offline testing or additional detection equipment. It effectively solves the problems of lagging coal quality detection and inaccurate data leading to AGC control inaccuracy in traditional methods. At the same time, outlier identification and missing value processing ensure data reliability, providing a reliable coal quality basis for subsequent identification of unit dynamic characteristics and intelligent control. This significantly improves the AGC's adaptability to coal quality fluctuations and control stability.
[0086] Furthermore, this embodiment provides a step for obtaining the theoretical response delay and inertial time constant based on the received lower heating value of coal and a pre-trained neural network model, including:
[0087] Input characteristics are obtained by normalization based on the received basis of coal quality and standardized operating data.
[0088] Input features are fed into a pre-trained neural network model to obtain predicted values;
[0089] The predicted values are denormalized to obtain the theoretical response delay and inertial time constant.
[0090] Based on a BP neural network model, the correlation between coal quality and energy characteristics and the dynamic response characteristics of the generating unit is fitted. This is done using AGC commands. Actual power and the output of the aforementioned soft measurement model The model uses 3-layer backpropagation (BP) neural network as input features. The input layer has 3 nodes, corresponding to 3 input parameters; the hidden layer has 12 nodes, using the ReLU activation function; and the output layer has 2 nodes, corresponding to the output parameters, including the theoretical response delay. and inertial time constant The activation function is a linear function.
[0091] For example, the training samples are historical operating data from the past three months, and each group of samples is labeled with a corresponding output label value. The theoretical response delay is... Annotate and extract the step change time of AGC commands. And the moment when the actual power output reaches 90% of the commanded target value. ,but For inertial time constant The process of tracking AGC commands with the actual generated power of the unit is fitted as a first-order inertial element. The least squares method is used to fit the data to obtain the corresponding data for each group of samples. ,in, For Laplace complex variables, the differential operation characteristics of the system's dynamic response are used to characterize the differential operation. It is a common differential operation symbol in the field of automatic control, corresponding to the first-order differential operation with respect to time in the time domain. To eliminate the influence of dimensional differences on the neural network, the input and output features need to be normalized, and the normalization parameters are saved for inverse normalization during the model inference stage. Gradient descent is used for training, with an initial learning rate of 0.01. During training, a learning rate decay strategy is adopted, i.e., the learning rate decays to 0.9 times its original value every 1000 iterations. The mean squared error loss function is used, with the formula: ,in, For the batch sample size, , The first The theoretical response delay label value and inertial time constant label value of each sample. , The output of the model is the first The predicted theoretical response delay and predicted inertial time constant for each sample.
[0092] During the real-time inference phase, AGC commands will be input. Actual power Soft measurement model output Normalization is performed to obtain the input features. ,in, , , Corresponding to the normalized , , The normalized input features are input into a pre-trained BP neural network model, and the normalized values are calculated through forward propagation. , Predicted value. Specifically, input features. Upon entering the input layer, each input node has connection weights with all 12 nodes in the hidden layer. Internally, the weighted sum of each node in the hidden layer is first calculated, then processed using the ReLU activation function to obtain the hidden layer output. (Hidden layer number...) The formula for calculating the weighted sum of the nodes is: ,in, For indexing hidden layer nodes, , For the input layer node index, , For the hidden layer The weighted sum of the nodes, For the input layer The node and the hidden layer The connection weight of each node For the hidden layer The bias term of each node, The input features are normalized; after ReLU activation, the hidden layer's... Output of each node: Ultimately, 12 hidden layer outputs are obtained. The 12 outputs of the hidden layer are used as inputs to the output layer. Weighted sum calculation of nodes: ,in, For output layer node index, , For the output layer The weighted sum of the nodes, For the hidden layer The node and the output layer The connection weight of each node For the output layer The bias term of the nth node; after linear activation, the output layer's nth node... The predicted output for each node: Finally, two normalized predicted values were obtained. , The connection weights and biases between input layer nodes and hidden layer nodes, and between hidden layer nodes and output layer nodes, are all determined during the training phase. Using the saved normalization parameters, the predicted values are denormalized to restore them to the theoretical response delay with physical meaning. and inertial time constant .
[0093] This embodiment constructs a three-layer BP neural network model, using coal quality received-based lower heating value, AGC commands, and actual power output as input features. It combines historical operating data from the past three months with labeling and model training. Through normalization and denormalization, dimensional differences are eliminated, enabling real-time prediction of theoretical response delay and inertial time constant. This effectively solves the problems of traditional methods being unable to dynamically identify unit response characteristics and quantify the impact of coal quality fluctuations on unit dynamic response. It captures the correlation between coal energy characteristics and unit dynamic response, providing a reliable basis for subsequent adaptive adjustment of PID parameters and rapid tracking of AGC commands. This significantly improves the adaptability of AGC control to unit dynamic changes and avoids problems such as control lag and overshoot caused by unknown unit response characteristics.
[0094] Furthermore, this embodiment provides a step for calculating the rate of change of operating condition characteristics based on standardized operating data and determining the current operating condition, including:
[0095] Based on standardized operating data, the rate of change of operating conditions is calculated, which includes the rate of change of AGC commands, the rate of change of coal quality energy, and the rate of change of main steam pressure.
[0096] The operating condition is determined by the rate of change of operating condition characteristics, and the current operating condition is obtained.
[0097] Specifically, in order to achieve operating condition identification and transform static operating data into dynamic change features, this embodiment calculates the rate of change of key parameters, with the calculation period being one data acquisition cycle. ,include:
[0098] The AGC command change rate reflects the degree of drastic change in load commands of thermal power units, and is used to distinguish between steady-state and dynamic operating conditions, calculating the AGC command at the current moment. Change in instruction from the previous AGC instruction AGC command change rate The rate of change of coal quality energy reflects the speed of fluctuations in coal quality energy. It is used to identify disturbances caused by sudden changes in coal quality and to calculate the received lower heating value of coal at the current moment. Change in coal energy quality compared to the previous moment's received lower heating value Coal quality energy change rate The main steam pressure change rate reflects the degree of fluctuation in main steam pressure and is used to determine whether the unit's current operating state is stable. It is used to calculate the main steam pressure at the current moment. Pressure change compared to the previous moment Main steam pressure change rate The above three rates of change constitute the rate of change of operating condition characteristics, providing a three-dimensional criterion for operating condition identification.
[0099] Based on the rate of change of operating conditions, the operating conditions are divided into steady-state conditions and dynamic conditions according to the pre-set rate of change thresholds. The rate of change thresholds include the command change threshold, the coal quality energy change threshold, and the main steam pressure change threshold. The rate of change thresholds are obtained by calibration through historical operating data of the unit. For example, the rate of change thresholds for conventional 300MW-600MW units are as follows: the command change threshold is 10MW / s, the coal quality energy change threshold is 100kJ / (kg•s), and the main steam pressure change threshold is 0.3MPa / s. The above data are example data and are not limited in this embodiment.
[0100] The steady-state condition determination rule is: the rate of change of AGC commands. ≤Instruction change threshold; Coal quality energy change rate ≤ Coal quality energy change threshold; Main steam pressure change rate ≤ Main steam pressure change threshold. A condition is considered steady-state only if all steady-state conditions are met simultaneously; otherwise, it is considered dynamic. Operating condition serves as the sole basis for switching between dual modes.
[0101] This embodiment transforms static operating data into dynamic operating condition characteristic change rates and constructs a three-dimensional change rate criterion, achieving accurate determination of steady-state and dynamic operating conditions. It effectively solves the problem that traditional operating condition identification relies on only a single parameter and cannot take into account load command changes and coal quality fluctuations. It can accurately capture dynamic operating conditions caused by AGC command step changes, coal quality changes, etc., and can also identify steady-state operating conditions of stable unit operation. It provides a unique and reliable judgment basis for subsequent dual-mode intelligent switching, avoids improper control strategy adaptation due to operating condition misjudgment, and ensures the rationality and stability of AGC control under different operating conditions.
[0102] Furthermore, this embodiment provides a step of switching to PID fast tracking mode if the current operating condition is dynamic: generating feedforward compensation based on the received low calorific value of coal, and adaptively adjusting PID parameters using feedforward compensation, theoretical response delay, and inertial time constant.
[0103] Feedforward compensation is used as input for PID control, and adaptive adjustment of PID parameters is used for parameter optimization of PID control. When the operating condition is dynamic, it automatically switches to PID fast tracking control to quickly track AGC commands, suppress pressure fluctuations, and eliminate disturbances. This ensures the unit's rapid response to load commands and avoids problems such as unstable combustion and parameter exceeding limits caused by improper adjustment of coal feed and air volume. It significantly improves the accuracy, speed, and stability of AGC control under dynamic conditions and further enhances the unit's ability to adapt to load scheduling fluctuations.
[0104] Furthermore, this embodiment provides a step for generating feedforward compensation based on the received calorific value of coal, including:
[0105] Based on the preset step determination rules, the effective AGC instruction change amount is calculated;
[0106] Calculate the BTU correction factor and feedforward coal feed rate based on the effective AGC command change.
[0107] Based on the feedforward coal feed rate and the pre-calibrated unit air-coal correlation table, the primary air adjustment amount and the secondary air adjustment amount are obtained.
[0108] The secondary air adjustment amount is corrected based on standardized operating data to obtain the corrected secondary air adjustment amount.
[0109] Specifically, such as Figure 3 As shown, As input for valid step judgment, invalid commands caused by minor fluctuations are eliminated, avoiding frequent compensation that could disrupt the unit's combustion conditions. Specifically, the valid step threshold is set according to the unit's AGC operating procedures and data acquisition accuracy; for example, a valid step judgment threshold can be set for conventional 300MW-600MW units. MW, specifically, can be adaptively adjusted according to the unit capacity; this embodiment does not limit this. The step determination rule is: if If it is, it is determined to be a valid AGC step change and is retained. The value is used as input for subsequent feedforward compensation calculations; if It was determined to be a minor fluctuation in the instruction, and the instruction was... Setting it to 0 will not trigger any compensation calculations or perform adaptive adjustments to the PID parameters, and the unit will operate stably with the current PID parameters.
[0110] Effective after step judgment The BTU correction factor, feedforward coal feed rate, and coordinated air volume adjustment are calculated using quantitative formulas. Specifically, this is based on the output of the soft sensor model. and Calculate the BTU correction factor Quantification and The energy density deviation provides a basis for correction of coal feed rate compensation, and the formula is: Then, reuse the calculation. Boiler thermal efficiency calculated in time Calculate the amount of coal fed forward. The formula is: ,in, and The signs are consistent: positive indicates increase, and negative indicates decrease; The base coal consumption rate of the unit is taken as the value corresponding to the design coal type or the average value based on historical operating data. It is a fixed calibration value, such as 0.28 kg / (kW·h) in this embodiment.
[0111] It is important to clarify that the total air volume of a thermal power unit is the sum of the primary air volume and the secondary air volume. The primary air is used to feed pulverized coal into the furnace, while the secondary air is used to supplement oxygen and aid combustion. The two are in a fixed ratio. When the coal feed rate changes, the primary / secondary air also needs to be adjusted proportionally. Otherwise, problems such as mill blockage and incomplete combustion of pulverized coal may occur. Therefore, the coal feed rate and air volume are adjusted in tandem.
[0112] Therefore, based on the obtained feedforward coal feed rate Based on the pre-calibrated unit air-fuel correlation table, the primary air adjustment amount is calculated. Secondary air adjustment amount This achieves coordinated compensation between coal feed and air volume, ensuring the rationality of furnace combustion air distribution after coal quality fluctuations and coal feed adjustments, and avoiding problems such as incomplete combustion or excessive air supply. Specifically, this includes:
[0113] The unit's air-coal correlation table is calibrated. Taking the unit's rated load operation as an example, based on historical operating data from the past three months, historical operating data under rated load and stable combustion conditions are selected, and the total coal feed rate of the selected historical operating data is calculated. The arithmetic mean of the values is used as the base coal feed rate for the unit. During the testing phase, the unit was stabilized at its rated load, and the coal feed rate was fixed at the unit's baseline coal feed rate. The primary air volume was then gradually adjusted. The primary air volume is taken as the benchmark primary air volume under the conditions of no clogging of the coal mill and stable coal feeding. Adjust the secondary air volume The secondary air volume under stable combustion conditions is taken as the benchmark secondary air volume. This allows us to calculate the sum of the baseline primary air volume and the baseline secondary air volume, thus obtaining the baseline total air volume. Furthermore, the above calibration process is repeated under different load ranges to form the benchmark coal feed rate, benchmark primary air volume, benchmark secondary air volume, and benchmark total air volume under different unit load ranges.
[0114] For example, under stable combustion conditions, the main steam pressure is stable at the design value ±0.5MPa and the oxygen content in the furnace is 3%-5%. The above data are example data and are not limited in this embodiment.
[0115] Based on the current unit load, retrieve the benchmark coal feed rate corresponding to the load range of the unit. Reference primary air volume Reference secondary air volume Calculate the primary air adjustment amount to match the coal feed rate adjustment. and secondary air adjustment amount The formula is: , .
[0116] This embodiment effectively eliminates minor fluctuations by performing step judgment on changes in AGC commands. Then, it calculates the BTU correction factor and feedforward coal feed rate based on the lower heating value of coal, and synchronously provides the primary and secondary air adjustment amounts in conjunction with the unit air-coal correlation table calibrated under load. This achieves coordinated feedforward regulation of coal feeding and air supply, which fully considers the impact of coal quality fluctuations on energy input and ensures that the air-coal ratio is always within a reasonable range. It can make adjustments when AGC commands change, effectively solving the problems of traditional feedforward control that do not consider coal quality differences, air-coal asynchrony easily leads to combustion deterioration, and large parameter fluctuations. It also avoids problems such as unstable combustion and parameter exceeding standards caused by improper adjustment of coal feed rate and air volume.
[0117] Furthermore, to reduce calibration errors in the unit's air-coal correlation table and to address unreasonable distribution caused by minor fluctuations in coal quality, real-time oxygen levels are used. Adjustment amount of secondary air Perform minor closed-loop corrections to ensure stable combustion. Set the oxygen content reference range. For example, the 3%-5% setting for stable combustion conditions, i.e. , The revised rule is: if oxygen content The problem was determined to be excessive air supply to the furnace. Perform a negative correction; if oxygen content The cause was determined to be oxygen deficiency in the furnace. Perform positive correction; if The air distribution was deemed reasonable and no adjustments were made. The revised secondary air adjustment amount is: ,in, For correction factor, , here The range is for example data, and this embodiment does not limit it.
[0118] This embodiment introduces real-time oxygen content to make closed-loop minor corrections to the secondary air adjustment amount. The secondary air is dynamically adjusted based on the comparison between the actual combustion oxygen content and the benchmark range, which effectively offsets the unreasonable air distribution caused by the calibration error of the unit's air-coal correlation table and the slight fluctuation of coal quality. This keeps the furnace in the optimal combustion state at all times, avoiding oxygen deficiency combustion or excessive air supply. It further improves the accuracy and adaptability of the air-coal ratio on the basis of the original feedforward compensation.
[0119] Furthermore, this embodiment provides a step for adaptively adjusting PID parameters using the theoretical response delay and the inertial time constant, including:
[0120] Construct a correlation model between PID parameters and theoretical response delay and inertial time constant;
[0121] Substituting the theoretical response delay and inertial time constant into the correlation model, the adjusted PID parameters are obtained.
[0122] To address the issues of slow response and overshoot that traditional fixed-parameter PID controllers often exhibit when coal quality changes, this study utilizes theoretical response time delay... and inertial time constant The PID parameters are adjusted to adapt to the dynamic characteristics of the unit. Specifically, the core objects of PID control here are the total coal feed, primary air volume, and secondary air volume of the thermal power unit. These three are controlled in a coordinated manner to ultimately achieve rapid tracking of AGC commands and stable unit operation. The parameter adaptive adjustment and control implementation process is as follows:
[0123] The PID controller employs a positional PID control algorithm. Its core control objective is to track the direction and magnitude of coal feed and airflow adjustments determined by the feedforward compensation strategy, while simultaneously suppressing main steam pressure fluctuations. This ensures that the actual power output rapidly approaches the AGC command and avoids instability phenomena such as overshoot and oscillation. Specifically, the total coal feed PID controller uses the feedforward coal feed rate... The input signal is used to adjust the coal feed rate, controlling the feeder speed to ensure rapid coal feeding. The corresponding adjustment amount; the primary air and secondary air PID controllers are respectively based on the primary air adjustment amount. Corrected secondary air adjustment amount As input, the output air volume adjustment signal controls the opening of the primary air damper and the secondary air damper, so as to achieve synchronous and coordinated matching of air volume and coal feed.
[0124] To achieve adaptive dynamic adjustment of parameters and break through the limitations of traditional fixed parameters, this method is based on theoretical response delay. and inertial time constant Establish PID parameters and , The correlation model, where the PID parameter is the proportional coefficient. Integral coefficient Differential coefficients The core logic of the association model is: It reflects the degree of lag in the unit's response. Reflecting the magnitude of the unit's inertia, both directly determine the adaptability of the PID parameters: when When the value increases, the response lag worsens, requiring an increase in the value. Increase To speed up the response time; when When the inertia of the unit increases, it needs to be reduced. , reduce To avoid oscillations caused by excessively rapid response, and at the same time increase To prevent overshoot caused by inertia. Based on the above correlation, the adaptive dynamic adjustment formula for parameters is: , , ,in, , and These are the baseline PID parameters under the rated operating conditions of the unit, obtained through calibration using historical operating data; , The baseline response delay and baseline inertia time constant are taken as the average values under steady-state operating conditions of the unit over the past three months. The above formula is only an example; the parameter adjustment coefficients can be optimized through machine learning fitting based on the characteristics of different units to further improve the adaptation accuracy. The adjusted parameters are obtained through PID controller parameter adjustment. , , The parameters are used as parameters for the PID controller to prioritize the tracking speed of AGC commands and quickly suppress disturbances caused by coal quality fluctuations and command step jumps.
[0125] For example, the baseline PID parameters under rated operating conditions of the unit are taken as follows: =1.2、 and The reference response delay and reference inertia time constant are taken as follows: , The above data are example data. The coefficients can be adjusted by fitting and optimizing parameters according to the characteristics of different units. This embodiment does not limit this.
[0126] This embodiment achieves real-time adaptive adjustment of the dynamic characteristics of randomized PID parameters by constructing a correlation model between PID parameters and theoretical response delay and inertia time constant. It can automatically match the optimal proportional, integral, and derivative parameters according to changes in coal quality, response lag, and inertia magnitude, solving the problems of slow response, overshoot, and oscillation that traditional fixed-parameter PID is prone to when coal quality fluctuates. At the same time, through coordinated PID regulation of total coal feed, primary air volume, and secondary air volume, and tracking of feedforward compensation commands, it improves the unit's rapid response capability to load commands and enhances the speed and stability of AGC control under dynamic operating conditions.
[0127] Furthermore, this embodiment also provides a step for stable combustion control of a boiler under dynamic operating condition boundary conditions.
[0128] Specifically, under dynamic operating conditions, when the unit is under boundary conditions, the boundary conditions are the near-low load range consisting of 30%-35% of the rated load. The 30% and 35% figures are example data; the specific values can be determined based on the unit model, boiler design parameters, and changes according to AGC instructions. The positive and negative values are used to determine the direction of load increase or decrease, with stable boiler combustion as the priority control objective.
[0129] when When this happens, it is necessary to respond to the AGC load increase command and execute the above-mentioned feedforward compensation calculation and PID parameter adaptive adjustment process normally; When responding to the AGC load reduction command, subsequent feedforward compensation adjustments are not executed. Instead, low-load stable combustion constraints are implemented, allowing the total coal feed and primary air volume to decrease to the preset stable combustion lower limit and remain unchanged. Unrestricted reduction is not permitted, thus ensuring stable boiler combustion during low-load load changes and avoiding combustion deterioration and furnace extinguishing caused by excessive adjustment or insufficient coal and air volume. Low-load stable combustion constraints include: when responding to the AGC load reduction command, forcing the total coal feed to be no less than the boiler's minimum stable combustion coal feed and the primary air volume to be no less than the minimum stable combustion primary air volume, maintaining the ratio of primary air volume to secondary air, and simultaneously maintaining the furnace oxygen content within the range required for low-load stable combustion to ensure complete combustion of pulverized coal; limiting the single adjustment step size of the coal feed, primary air volume, and secondary air volume during low-load load changes to avoid excessive adjustment actions that could disrupt the furnace combustion state.
[0130] For example, the above constraints can be based on the following data: the minimum stable coal feed rate and the minimum stable primary air volume of the boiler are based on the test calibration of the unit under boundary conditions, which are 30% of the rated load; the oxygen content in the furnace is maintained at 4%-6%; the single adjustment step of the coal feed rate does not exceed 0.5t / h, the single adjustment step of the primary air does not exceed 800m³ / h, and the single adjustment step of the secondary air does not exceed 600m³ / h. The above data are example data, based on the unit model and boiler design test calibration, and this embodiment does not limit them.
[0131] It should be noted that the above-mentioned low-load stable combustion control steps are only executed under dynamic operating conditions. They do not need to be considered under steady-state operating conditions because the unit load remains constant during steady-state operation. In addition, this scheme has used 30% of the rated load as the minimum stable combustion safety net value. Under steady-state operating conditions, the unit load and operating parameters will not be lower than this safety limit, and the operating data are all within a stable and reliable range. There is no risk of excessively low air and coal volume, combustion deterioration, or furnace flameout due to adjustment. Therefore, there is no need to implement additional low-load stable combustion constraints under steady-state operating conditions.
[0132] This embodiment sets up boiler stable combustion control logic near the low load boundary under dynamic operating conditions. While ensuring normal and rapid response of AGC when the load increases, it also imposes targeted constraints on the load reduction process. This effectively avoids safety risks such as furnace flameout and combustion deterioration caused by excessive reduction of air and coal volume in the low load range. Furthermore, by adjusting the step size smoothly, it ensures that the load change process is stable and controllable, significantly improving the safety, stability and anti-disturbance capability of the unit under boundary operating conditions.
[0133] Furthermore, this embodiment provides a step for switching to steady-state optimization mode if the current operating condition is a steady-state condition: quickly finding the current optimal operating target value from historical operating data based on standardized operating data, including:
[0134] Based on standardized operational data, a local wind and coal dispatching interval is constructed;
[0135] Based on the local wind and coal dispatch interval, a steady-state high-efficiency sample set was selected from historical operation data;
[0136] Construct a multi-objective evaluation function;
[0137] The optimal operating target value is obtained by taking the sample data with the smallest evaluation function value from the steady-state efficient sample set.
[0138] When the operating condition is steady-state, the unit enters the steady-state optimization mode. Under the current steady-state condition, no rapid disturbance compensation is performed. Instead, the optimal air-coal ratio is used as the core. Within a small range of current coal feed and air volume scheduling, historical high-efficiency steady-state data is used to find the optimal operating point under the current AGC command load. While maintaining stable actual power generation to track AGC commands, the unit achieves steady-state economic and stable operation.
[0139] like Figure 4 As shown, based on the current actual power output Determine the load range based on the current total coal feed rate. Current total air volume Construct a local wind and coal dispatching zone centered on the wind and coal system: , ,in, To allow for adjustment of the coal feed rate in increments, The allowable adjustment step size for the total air volume can be calibrated using historical steady-state operating data of the unit to ensure a safe adjustment range without significant disturbances. A steady-state high-efficiency sample set meeting the following conditions is selected from the historical steady-state operating data: within the current load range; main steam pressure stable near the preset value; oxygen content within the stable combustion range; and boiler thermal efficiency within the high range.
[0140] Within the aforementioned local wind and coal scheduling intervals, the wind and coal ratios of each sample in the steady-state high-efficiency sample set are traversed. Each sample corresponds to a set of historical operational data. , , Construct a multi-objective evaluation function, the formula of which is: ,in, For the first Evaluation value of each sample For the first Boiler thermal efficiency of a sample, For the first The main steam pressure of each sample Main steam pressure setpoint For the first Oxygen content of each sample Set the oxygen level to a set value. , , For the weighting coefficients, satisfying Calculate the evaluation function for each steady-state high-efficiency sample, select the set of historical operating data with the smallest evaluation function value, and use the corresponding coal feed rate and total air volume as the optimal operating target values for the current operating condition: This serves as the control target for the total coal feed and total air volume under steady-state operating conditions. By using closed-loop regulation to output control signals, the current total coal feed and total air volume are brought close to the optimal operating target value. This achieves efficient, economical, and stable operation while maintaining the actual power output of the unit in line with the current AGC command.
[0141] For example, take =0.5t / h, =300m³ / h, ensuring a safe adjustment range without significant disturbances. The stable combustion range for furnace oxygen content can be 3%-5%, the stable range for main steam pressure can be ±0.3MPa of the main steam pressure setpoint, and the boiler thermal efficiency should be within the top 30% of the efficiency range under the same load. The main steam pressure setpoint is given by the rated main steam pressure corresponding to the current load of the unit and is a fixed value given by the unit design or dispatch. The oxygen content setpoint is 4%. , , The above data are example data and can be calibrated according to the unit model and actual operation. This embodiment does not limit this.
[0142] This embodiment constructs a local air-coal ratio scheduling interval under steady-state conditions, selects historical high-efficiency steady-state samples, and uses a multi-objective evaluation function for optimization. It chooses to base the optimization on real historical operating data rather than complex online modeling or intelligent optimization algorithms. On the one hand, this avoids the problems of large computational load and long time consumption caused by online modeling. On the other hand, it can directly utilize the unit's verified high-efficiency and stable operating interval, ensuring that the optimization results are real, reliable, safe, and executable. It can quickly locate the optimal air-coal ratio operating target value under the current load, taking into account the stability of boiler thermal efficiency, oxygen content, and main steam pressure. Without affecting the AGC command tracking effect, it can achieve rapid unit response and improve the economy and reliability of unit operation under steady-state conditions.
[0143] Example 2:
[0144] like Figure 5 As shown, this embodiment of the invention provides an intelligent system for improving the AGC response speed of thermal power units, used to implement the above-mentioned intelligent method for improving the AGC response speed of thermal power units, including:
[0145] Coal quality characteristic soft measurement module: used to acquire standardized operating data of thermal power units and obtain the received basis lower heating value of coal based on the constructed soft measurement model;
[0146] Unit response lag measurement module: used to obtain the theoretical response delay and inertial time constant based on the received basis of coal calorific value and a pre-trained neural network model;
[0147] Operating condition identification and mode switching module: used to calculate the rate of change of operating condition characteristics based on standardized operating data and determine the current operating condition, which includes dynamic operating condition and steady-state operating condition;
[0148] PID fast tracking control module: It generates feedforward compensation based on the received low calorific value of coal and adaptively adjusts PID parameters using theoretical response delay and inertial time constant. The feedforward compensation is used as the input for PID control, and the adaptive adjustment of PID parameters is used for parameter optimization of PID control.
[0149] Steady-state optimization module: Used to quickly find the current optimal operating target value from historical operating data based on standardized operating data.
[0150] Specifically, the modules communicate with each other via industrial Ethernet, with the data transmission cycle consistent with the data acquisition cycle. The coal quality characteristic soft measurement module transmits the calculated lower heating value of the coal, along with standardized operating data, to the unit response lag measurement module, providing it with core input parameters. The theoretical response delay and inertial time constant output by the unit response lag measurement module are provided in real-time to the PID fast tracking control module for adaptive adjustment of PID parameters. The operating condition identification and mode switching module, as the core switching unit for control modes, establishes signal connections with both the PID fast tracking control module and the steady-state optimization module. Based on the operating condition determination results, it outputs corresponding mode switching signals, enabling selective activation of either the PID fast tracking control module or the steady-state optimization module, ensuring that the unit has an appropriate control module operating under both dynamic and steady-state conditions.
[0151] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.
[0152] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0153] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0154] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0155] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.
Claims
1. A method for intelligently improving the response speed of AGC in thermal power units, characterized in that, include: Obtain standardized operating data of thermal power units, and obtain the lower heating value of coal based on the received basis based on the constructed soft measurement model; Based on the received basis lower heating value of the coal and the pre-trained neural network model, the theoretical response delay and inertial time constant are obtained. The rate of change of operating condition characteristics is calculated based on the standardized operating data to determine the current operating condition, which includes dynamic operating condition and steady-state operating condition. If the current operating condition is a dynamic condition, switch to PID fast tracking mode: generate feedforward compensation based on the received low calorific value of the coal quality, and adaptively adjust the PID parameters using the theoretical response delay and the inertial time constant. The feedforward compensation is used as the input for PID control, and the adaptive adjustment of the PID parameters is used for parameter optimization of PID control. If the current operating condition is a steady-state condition, switch to steady-state optimization mode: find the current optimal operating target value from historical operating data based on the standardized operating data.
2. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, The acquisition of standardized operating data of thermal power units, and the determination of the lower heating value of coal based on the constructed soft measurement model, include: Obtain raw operating data of thermal power units and perform standardized preprocessing to obtain standardized operating data; Based on the standardized operating data, the boiler thermal efficiency and total losses were calculated. A system error correction coefficient is introduced to correct the mechanism formula of the soft measurement model; Based on the standardized operating data, boiler thermal efficiency, and input mechanism formulas for various total losses, the received lower heating value of coal is obtained.
3. The method for intelligently improving the AGC response speed of thermal power units according to claim 2, characterized in that, The mechanism formula for correcting the soft measurement model by introducing a system error correction coefficient includes: The optimal system error correction coefficient is obtained by using historical operating data and fitting it through machine learning. The optimal system error correction coefficient is substituted into the mechanism formula of the soft measurement model.
4. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, Based on the received basis lower heating value of the coal and a pre-trained neural network model, the theoretical response delay and inertial time constant are obtained, including: Based on the received basis lower heating value of the coal and standardized operating data, the input characteristics are obtained by normalization. The input features are input into a pre-trained neural network model to obtain predicted values; The predicted values are denormalized to obtain the theoretical response delay and inertial time constant.
5. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, The rate of change of operating condition characteristics is calculated based on the standardized operating data to determine the current operating condition, including: Based on the standardized operating data, the rate of change of operating conditions is calculated, wherein the rate of change of operating conditions includes the rate of change of AGC commands, the rate of change of coal quality energy, and the rate of change of main steam pressure. The operating condition is determined based on the rate of change of the operating condition characteristics, and the current operating condition is obtained.
6. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, Feedforward compensation is generated based on the received basis lower heating value of the coal, including: Based on the preset step determination rules, the effective AGC instruction change amount is calculated; Calculate the BTU correction factor and feedforward coal feed rate based on the effective AGC command change amount; Based on the feedforward coal feed rate and the pre-calibrated unit air-coal correlation table, the primary air adjustment amount and the secondary air adjustment amount are obtained. The secondary air adjustment amount is corrected based on the standardized operating data to obtain the corrected secondary air adjustment amount.
7. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, Adaptively adjusting PID parameters using the theoretical response delay and the inertial time constant includes: Construct a correlation model between PID parameters and theoretical response delay and inertial time constant; Substituting the theoretical response delay and the inertial time constant into the correlation model, the adjusted PID parameters are obtained.
8. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, Also includes: Stable combustion control of the boiler under dynamic operating boundary conditions.
9. The method for intelligently improving the AGC response speed of thermal power units according to claim 1, characterized in that, Finding the current optimal operating target value from historical operating data based on the standardized operating data includes: Based on the standardized operational data, a local wind and coal scheduling interval is constructed; Based on the aforementioned local wind and coal dispatching interval, a steady-state high-efficiency sample set is selected from historical operating data; Construct a multi-objective evaluation function; The sample data with the smallest evaluation function value in the steady-state efficient sample set is selected to obtain the current optimal operating target value.
10. A smart system for improving the response speed of AGC in thermal power units, used to implement the smart method for improving the response speed of AGC in thermal power units as described in any one of claims 1-9, characterized in that, include: Coal quality characteristic soft measurement module: used to acquire standardized operating data of thermal power units and obtain the received basis lower heating value of coal based on the constructed soft measurement model; Unit response lag measurement module: used to obtain the theoretical response delay and inertial time constant based on the received basis of the coal quality and the pre-trained neural network model; Operating condition identification and mode switching module: used to calculate the rate of change of operating condition characteristics based on the standardized operating data and determine the current operating condition, wherein the operating condition includes dynamic operating condition and steady-state operating condition; PID fast tracking control module: It generates feedforward compensation based on the received low calorific value of the coal and adaptively adjusts the PID parameters using the theoretical response delay and the inertial time constant. The feedforward compensation is used as the input for PID control, and the adaptive adjustment of the PID parameters is used for parameter optimization of PID control. Steady-state optimization module: used to find the current optimal operating target value from historical operating data based on the standardized operating data.
