Method and system for determining nonlinear flow characteristic curve of capacity air door of coal mill

Abstract

The invention provides a method and a system for determining a nonlinear flow characteristic curve of a capacity air door of a coal mill. The method comprises the following steps: constructing a unit power-fuel quantity model based on a neural network; through a field test, under the condition of univariate control, obtaining corresponding relation data between the active power of the unit and the opening degree of the target capacity air door; based on the unit power-fuel quantity model and the corresponding relation data of the unit active power and the target capacity air door opening degree, the relation between the target capacity air door opening degree and the air volume percentage is calculated; and performing data completion and inflection point identification on the relationship between the opening degree of the target capacity air door and the air volume percentage, and establishing an opening degree-flow percentage piecewise function of the target capacity air door so as to determine a nonlinear flow characteristic curve of the target capacity air door. According to the method, the nonlinear characteristic of the capacity air door can be accurately captured, a key basis is provided for coordinated control modeling and compensation of the thermal power generating unit, the main steam pressure stability and the AGC response capability of the unit are effectively improved, manual intervention is reduced, and the automatic power generation control efficiency is enhanced.

Description

Technical Field

[0001] This invention belongs to the field of coordinated control technology for thermal power units, and particularly relates to a method and system for determining the nonlinear flow characteristic curve of a coal mill capacity damper. Background Technology

[0002] In the coordinated control system of thermal power units, the capacity damper of the double-inlet, double-outlet ball mill (referred to as "double-outlet mill") plays a crucial role. It is not only the main means of regulating the output of the pulverizing system, but also a key control hub connecting the boiler load and the coal feed rate of the mill. Its control quality directly affects the stability of the main steam pressure and the unit's response speed to AGC commands, and is a core element in whether coordinated control can accurately achieve "boiler-turbine energy balance." However, the inherent nonlinear characteristics of the capacity damper's opening and actual ventilation volume (or corresponding fuel quantity) pose a significant challenge to control, making it difficult to accurately model and compensate for in the control system. The direct consequence is fluctuations in pulverizing output, leading to periodic oscillations or large deviations in main steam pressure, and delayed AGC response. In severe cases, the unit must exit the coordinated control mode and switch to manual intervention, weakening the unit's automatic power generation control capability. Therefore, obtaining the nonlinearity of the capacity damper is crucial. Summary of the Invention

[0003] This invention employs a combination of data mining and experimental methods to design a method for determining the nonlinear characteristics of capacity dampers in dual-inlet, dual-outlet coal mills. First, neural network modeling is used to obtain the relationship between unit power and fuel quantity. Then, experimental data pairs of power and single-damper opening are obtained to derive the relationship between airflow percentage and opening. Data is supplemented, and inflection points of the piecewise function are identified to determine the characteristics of a single damper. Finally, this process is repeated to sequentially obtain the piecewise function of opening-flow percentage for all coal mill capacity dampers. This method can accurately capture the nonlinear characteristics of capacity dampers, providing crucial data for coordinated control modeling and compensation of thermal power units. It effectively improves the stability of main steam pressure and the unit's AGC response capability, reduces manual intervention, and enhances the efficiency of automatic power generation control.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for determining the nonlinear flow characteristic curve of a coal mill capacity damper includes the following steps: A unit power-fuel quantity model based on neural networks is constructed, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Through field tests, under single-variable control conditions, the corresponding data of the active power of the unit and the target capacity damper opening were obtained; Based on the unit power-fuel model and the corresponding data of the unit active power and target capacity damper opening, the relationship between the target capacity damper opening and the air volume percentage is calculated. Data completion and inflection point identification are performed on the relationship between the target capacity damper opening and the air volume percentage, and the segmentation point is determined. Based on the segmentation point, a piecewise function of the target capacity damper opening-flow percentage is established to determine its nonlinear flow characteristic curve. Repeat the above steps to obtain the nonlinear flow characteristic curves of all coal mill capacity dampers in sequence.

[0005] Furthermore, the construction of the unit power-fuel model based on neural networks includes: Steady-state operating condition data were selected from historical data to construct a sample set of unit power and fuel input. The unit power-fuel quantity model is trained using a radial basis function neural network, with the unit active power as input and the amount of fuel fed into the furnace as output. The unit power-fuel quantity model is trained and optimized using training, validation, and test sets until the model error meets the preset accuracy requirements.

[0006] Furthermore, the selection criteria for the steady-state operating condition data are based on the stability of the unit power command, main steam pressure, and pulverizing system operating parameters.

[0007] Furthermore, the field test includes: Adjust the unit to rated load conditions and establish a stable baseline; Adjust only the target capacity damper opening, keep other parameters fixed, and record the unit active power corresponding to different opening ranges.

[0008] Furthermore, when adjusting the opening of the target capacity damper, different adjustment step sizes are used in different opening ranges of the damper.

[0009] Furthermore, the relationship between the opening degree of the target capacity damper and the percentage of air volume includes: The active power of the unit obtained from the field test is input into the unit power-fuel quantity model to obtain the theoretical fuel quantity; Using the theoretical fuel quantity when the target capacity damper is fully open as a benchmark, the percentage of air volume at each opening degree is calculated as the ratio of the theoretical fuel quantity to the benchmark value.

[0010] Furthermore, data completion and inflection point completion for the relationship between the target capacity damper opening and the percentage of air volume include: Interpolation is performed on discrete test data points to complete the data within the entire opening range, and the completed opening-airflow percentage data is obtained. Based on the completed opening-airflow percentage data, the characteristic inflection point is identified by analyzing its rate of change.

[0011] Furthermore, the inflection point of the recognition characteristic includes: Calculate the first and / or second derivatives of the completed opening-airflow percentage data, and preliminarily determine candidate inflection points based on the extreme points of the first derivative.

[0012] Further, determining the segmentation point based on the candidate inflection point includes: sorting the candidate inflection points in ascending order according to their corresponding opening values; and filtering the sorted candidate inflection points according to preset rules to determine the segmentation point used to construct the piecewise function.

[0013] On the other hand, the present invention provides a system for determining the nonlinear flow characteristic curve of a coal mill capacity damper, comprising: Model building module: It is used to build a unit power-fuel quantity model based on neural network, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Test control and data acquisition module: It is used to obtain the corresponding data of the active power of the unit and the damper opening of the target capacity through field test under single variable control conditions; Data conversion module: It is used to calculate the relationship between the target capacity damper opening and the percentage of air volume based on the power-fuel model of the unit and the corresponding data of the active power of the unit and the target capacity damper opening; Characteristic curve generation module: It is used to complete the data and identify the inflection point of the relationship between the opening degree of the target capacity damper and the percentage of air volume, and determine the segmentation point. Based on the segmentation point, it establishes a piecewise function of the opening degree-percentage of air volume of the target capacity damper to determine its nonlinear flow characteristic curve.

[0014] Compared with the prior art, the present invention has the following beneficial effects: Existing technologies mostly estimate the flow characteristics of capacity dampers based on linear assumptions or empirical formulas, which cannot adapt to the complex nonlinear relationships of capacity dampers, resulting in characteristic curve error rates generally exceeding 15%. This invention can accurately capture the nonlinear characteristics of the damper, and the generated nonlinear flow characteristic curve matches the actual operating conditions with a success rate of up to 92%, completely solving the problem of "linear fitting distortion" in traditional methods. This provides high-precision support for the coordinated control modeling and compensation of thermal power units. In terms of unit control performance, the piecewise function generated by this invention can be directly integrated into the coordinated control system to achieve real-time compensation for the nonlinear characteristics of the damper—reducing the main steam pressure fluctuation to within ±0.2MPa, improving stability, and avoiding boiler operation risks caused by pressure deviations; improving the response speed of AGC commands; and increasing the automation rate of damper adjustment to over 95%, reducing the average number of manual interventions per day to less than once, significantly reducing the labor intensity of maintenance personnel, and reducing the risk of single-unit operation caused by human error. Moreover, it has a wide range of applications, low engineering implementation costs, and is easy to promote on a large scale. It does not rely on specific equipment parameters and can be quickly adapted to dual-inlet dual-outlet coal mills of different capacities and coal qualities from 300MW to 1000MW. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart of the invention; Figure 2 This is a schematic diagram of the capacity damper opening-flow percentage characteristic curve of mill A; Figure 3 This is a schematic diagram of the capacity damper opening-flow percentage characteristic curve of mill B; Figure 4 This is a schematic diagram of the capacity damper opening-flow percentage characteristic curve of the C-mill. Figure 5 This is a schematic diagram of the characteristic curve of the damper opening and flow rate percentage of the D mill capacity. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] Example 1 The following description, in conjunction with the accompanying drawings, further illustrates this embodiment.

[0019] like Figure 1 As shown in the figure, this embodiment provides a method for determining the nonlinear flow characteristic curve of a coal mill capacity damper, including the following steps: A unit power-fuel quantity model based on neural networks is constructed, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Through field tests, under single-variable control conditions, the corresponding data of the active power of the unit and the target capacity damper opening were obtained; Based on the unit power-fuel model and the corresponding data of the unit active power and target capacity damper opening, the relationship between the target capacity damper opening and the air volume percentage is calculated. Data completion and inflection point identification are performed on the relationship between the target capacity damper opening and the air volume percentage, and the segmentation point is determined. Based on the segmentation point, a piecewise function of the target capacity damper opening-flow percentage is established to determine its nonlinear flow characteristic curve. Repeat the above steps to obtain the nonlinear flow characteristic curves of all coal mill capacity dampers in sequence.

[0020] This embodiment takes a mainstream four-mill, eight-unit capacity damper configuration as an example, denoted as mills A / B / C / D and A1 / 2, B1 / 2, C1 / 2, D1 / 2. It assumes that during the process of increasing unit load (active power), the mill start-up sequence is A, B, C, D. Generally, three mills are sufficient to meet the rated load requirement, with mill D as a backup. Specifically, it includes: Step 1. Use historical data mining methods to find the mapping relationship between unit power N and fuel quantity B: The nonlinear characteristics of the capacity damper are only measurable under steady-state conditions; therefore, steady-state data must first be screened from the historical database. The screening criteria are as follows: a. Stable power command: The unit's AGC command (or load setpoint) fluctuates by ≤±1% of the rated power within 10 minutes (e.g., for a 600MW unit, the fluctuation is ≤±6MW). b. Stable main steam pressure: The main steam pressure fluctuation of the boiler is less than 0.1MPa, avoiding interference of pressure fluctuations with fuel quantity regulation; c. Stable pulverizing system: The number of coal mills in operation remains unchanged, the coal feed rate fluctuation is ≤ ±2% of the design value, and the primary air pressure fluctuation is ≤ ±0.02 kPa; The operating condition that satisfies the above steady-state conditions with a continuous stable time of more than 10 minutes is selected as the qualified operating condition point, and the average value of the unit power N and the average value of the coal fed into the furnace B are taken. Extract NB sample pairs from steady-state data over the past two months, with the following requirements: a. Sample size: ≥300 groups, to ensure the generalization ability of the neural network training and avoid overfitting; b. Sample coverage: Power coverage of 50%~100% rated load of the unit (dual-outlet mills may stop operating under low load, so data below 50% load can be excluded); Data preprocessing: Take the average value of each sample over a 10-minute period. (t = 1-600 seconds, 1 data point per second), and the same applies to B; outliers are removed using the 3σ criterion (if a sample N or B deviates from its mean ± 3 times the standard deviation, it is determined to be an outlier and removed).

[0021] Training is conducted using the unit's active power N as input and the amount of coal fed into the furnace B as output; denoted as... .

[0022] a. Model selection: Considering that the "power-fuel quantity" relationship of thermal power units is affected by factors such as coal quality and coal mill efficiency, and has weak nonlinearity and time-varying characteristics, the radial basis function neural network (RBF-NN) adopts a local approximation strategy, which does not require backpropagation of all network parameters, thus improving training efficiency and achieving faster convergence. Moreover, it has local response characteristics, which has better predictive stability for input values ​​outside the sample distribution, and has smaller fitting error for nonlinear mapping relationships, making it more suitable for the actual operating rules of the unit.

[0023] b. Model training process: Data partitioning: The 300 NB samples were divided into a training set (210 sets), a validation set (60 sets), and a test set (30 sets) in a 7:2:1 ratio. The training set was used for learning model parameters; the validation set was used to monitor the risk of overfitting and dynamically adjust the model structure; and the test set was used to independently evaluate the generalization ability of the final model.

[0024] Parameter initialization: The K-means clustering algorithm is used to determine the centers of hidden layer neurons from the training set N values. To ensure the center covers the sample distribution range, the radial basis function width is initialized based on the sample standard deviation. This ensures that the basis functions provide a reasonable cover over the input space. The formula is as follows: .

[0025] In the formula, For cluster center c The maximum Euclidean distance (MW) between them. is the sequence number of the hidden layer neurons; k is the number of hidden layer neurons.

[0026] Weighting: The output layer weights are determined using the least squares method. The objective function is to minimize the mean squared error (MSE) of the training set.

[0027] Where m=210 is the number of samples in the training set; For sample serial number ( ); The actual fuel quantity for the j-th sample; Let be the theoretical fuel quantity of the model for the j-th sample.

[0028] Output layer weights; radial basis functions It is the output of the hidden layer neurons, commonly in the form of a Gaussian function: ; Let be the active power of the j-th sample; The center of the hidden layer neurons; The output represents the "distance" between the input of the j-th sample and the center of the hidden layer neuron after transformation by the radial basis function. b is the output layer bias term, used to compensate for the overall system bias; k is the number of hidden layer neurons.

[0029] The characteristic of radial basis functions is that they are only related to the distance of the input variable from a certain "center". The closer the distance (the smaller the absolute value of x), the greater the contribution, and the closer the function value is to 1, which means that the neuron has a strong response to the current input data. The farther the distance, the closer the function value is to 0, exhibiting the "local response" characteristic.

[0030] Model validation and optimization: After each training round, the MSE of the validation set is calculated. If the MSE of the validation set is > 0.5 t / h² (adjusted according to the allowable fuel control error requirements of the unit), the number of hidden layer neurons is increased (+5 each time), and steps 2-3 above are repeated for training until the MSE of the validation set meets the accuracy requirements. The model is evaluated using the test set (the MSE of the test set is required to be ≤ 0.6 t / h²), and the power-fuel ratio model is finally obtained. (1.1) In the formula, This represents the mapping of the "power-fuel quantity" relationship by the trained RBF neural network; B represents the fuel quantity (coal quantity fed into the furnace); N represents the active power of the unit.

[0031] Step 2. Use an experimental method to obtain the relationship between the power N and the opening of a damper of a certain capacity. Relationship Through field tests, under single-variable control conditions (changing only the target damper opening while keeping other parameters constant), the "capacity damper opening" was established. The correspondence between "→unit active power N" is used for subsequent conversion. The relationship provides experimental data.

[0032] Adjust to rated load condition: Objective: To enable the unit to reach its rated load through coordinated adjustment of multiple dampers, establish a rated load operating condition benchmark, and build a stable benchmark for subsequent single-variable tests.

[0033] Operation: Start mills A / B / C, adjust the capacity dampers of B1 / 2 and C1 / 2 to the fully open state (i.e., command μ=100%), and maintain the main steam pressure and primary air pressure at their rated values; gradually increase the unit load (active power) to the rated load and stabilize it by adjusting damper A1 / 2; record the specific opening degree of A1 / 2 at this time (assuming A1=85% and A2=82%), as a fixed benchmark for subsequent tests. Key parameter stabilization control includes: a. Main steam pressure: Controlled by the main regulating valve of the steam turbine, it is maintained at the rated value (e.g., 17MPa, deviation ≤ ±0.05MPa). b. Primary air pressure: Controlled by the guide vanes of the primary air fan, maintained at the design value (e.g., 9 kPa, deviation ≤ ±0.02 kPa). c. Coal feed rate: The coal feed rate of mills A / B / C is maintained at the design value (e.g., 50t / h, with a deviation ≤ ±1t / h) to avoid fluctuations in coal feed rate affecting power. Record the active power when the capacity damper C2 is 100% open: Under the established rated load condition baseline, following the principle of a single variable, only adjust the opening of damper C2, slowly resetting it from the current state (3.2.1) to the initial state to be adjusted (i.e., fully closed μ=0%, avoiding interference from uncertain initial opening), and then gradually adjust C2 to the fully open state (μ=100%); after the unit power has been running stably for 10 minutes, record the active power of the unit at this time. (Subscript 100 corresponds to 100% opening of capacity damper C2).

[0034] To accurately capture the nonlinearity of different opening ranges (flow rate is sensitive to changes at low openings and tends to saturate at high openings), different adjustment step sizes need to be set according to the opening range. The specific steps are as follows: a. 70%-100% high opening range: Maintain step size The target regulating damper C2 is gradually closed from 100%. When the opening reaches 97%, the primary air pressure and coal feed rate are maintained unchanged. After stabilizing for 10 minutes, the main steam pressure of the unit is restored to the rated pressure by adjusting the turbine main regulating damper. The opening of C2 and the corresponding active power are recorded at this time. Repeat the above steps, gradually decreasing C2 until... Record each group ( , ) data pairs.

[0035] b. 50%-70% medium opening range: maintain step size When C2 is reduced from 70% to 68%, after stabilizing for 10 minutes, adjust the main steam pressure to the rated pressure and record the C2 opening and corresponding active power at this time. Repeat the operation, gradually reducing C2 to... Record each group in this interval ( , ) data pairs.

[0036] c. 0%-50% low opening range: Maintain step size Reduce C2 from 50% to 49%, stabilize for 10 minutes, then adjust the main steam pressure to the rated value. Record the C2 opening and the corresponding active power at this time. Repeat the operation, gradually reducing C2 until C2 is completely turned off. Record each group Data pair.

[0037] After the above steps, by strictly following the corresponding step size to complete discrete sampling in each interval, the sets of sampling points in each interval are integrated to obtain several... Data pairs constitute the opening-power sample set of the C2 damper:

[0038] in .

[0039] Step 3. Obtain the relationship between the percentage of airflow and the opening of a damper of a certain capacity: Record each power group in step 2 As input, these are substituted into the RBF-NN neural network of equation (1.1) to infer the results for each group. Corresponding theoretical fuel quantity : (1.2) Due to the full opening of the capacity damper ( At this time, the ventilation volume is at its maximum, and the corresponding fuel consumption is also at its maximum. ),therefore Corresponding traffic percentage Defined as "the ratio of the theoretical fuel quantity at a certain opening degree to the theoretical fuel quantity at full opening", the formula is as follows: (1.3) Preliminary construction Sample set: Opening obtained from the previous steps Percentage of flow Preliminary results of C2 damper were obtained. Sample set:

[0040] Capacity damper The characteristic is usually a non-linear curve of "steep at low opening and gentle at high opening". It is necessary to complete the full opening range by data completion, identify the inflection point by adjacent flow difference analysis, and finally establish the piecewise function characteristic curve of air volume percentage and damper opening of C2 damper.

[0041] 1. Data completion: because The initial sample set is not based on 1% intervals. Only includes discrete opening degree The value needs to be filled in first. For data pairs within the range [0, 100], the completion method uses linear interpolation, as follows: a. Data sorting: By subscript Sort in ascending order to obtain the sorted sample set: b. If any integer degree Not here The subscript sequence needs to be padded. ,set up They are respectively The subscripts of the preceding and following terms, then Corresponding traffic percentage for: (1.4) Finally, we obtain the value that covers the entire opening. ;in .

[0042] 2. Take the first derivative , which represents the percentage change in flow rate when the opening increases by 1%, that is, the instantaneous rate of change of flow rate with the opening. ; 3. Take the second derivative , indicating concavity / convexity; if the second derivative is greater than 0, then it is a concave function. ; 4. Sort by size from largest to smallest, then select the top 13 largest. Sort the array in ascending order by its subscript index to obtain a sequence of 13 elements. As The inflection point. Taking the inflection point as the dividing point, establish a piecewise linear function, and the expression for each piecewise segment is the linear equation within that interval: that is... Describe this using the following piecewise function:

[0043] Thus, the characteristic curve of C2 is obtained.

[0044] Find the piecewise functions B1 and B2 1. Adjust to rated load condition: 2. Preparation for a single-variable experiment: 3. Keep A1 / 2 at the baseline opening, B2 fully open, and C1 / 2 fully open, adjust only B1, and record the active power when B1 is fully open. ; 4. Keep A1 / 2 at the reference opening, B1 fully closed, and C1 / 2 fully open, adjust only B2, and record the active power when B2 is fully open. ; 5. Curve Determination: Repeat steps 3.2.2-3.4 to obtain the characteristic curves of B1 and B2; Find the piecewise functions A1 and A2. 1. Adjust to rated load conditions: Start A / B / C mills and maintain the main steam pressure and primary air pressure at the rated values; adjust the capacity dampers of B1 / 2 and A1 / 2 to the fully open state (μ=100%), and ensure that the unit load (active power) is maintained at the rated load conditions by adjusting the damper of C1 / 2.

[0045] 2. Repeat the above steps to obtain the characteristic curves of A1 and A2; Find the piecewise functions D1 and D2 Repeat the above steps to obtain the characteristic curves D1 and D2; based on the above steps, the nonlinear characteristic curve function of the entire wear capacity damper can be obtained.

[0046] like Figure 2-5 The figures shown are schematic diagrams of the nonlinear characteristic curve functions finally obtained for dampers A1, A2, B1, B2, C1, C2, D1, and D2.

[0047] Based on the above embodiments, this embodiment can accurately capture the nonlinear characteristics of the capacity damper, and the generated nonlinear flow characteristic curve has a consistency of up to 92% with the actual operating conditions, completely solving the problem of "linear fitting distortion" in traditional methods, and providing high-precision support for the coordinated control modeling and compensation of thermal power units.

[0048] Example 2 This embodiment provides a system for determining the nonlinear flow characteristic curve of a coal mill capacity damper, including: Model building module: It is used to build a unit power-fuel quantity model based on neural network, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Test control and data acquisition module: It is used to obtain the corresponding data of the active power of the unit and the damper opening of the target capacity through field test under single variable control conditions; Data conversion module: It is used to calculate the relationship between the target capacity damper opening and the percentage of air volume based on the power-fuel model of the unit and the corresponding data of the active power of the unit and the target capacity damper opening; Characteristic curve generation module: It is used to complete the data and identify the inflection point of the relationship between the opening degree of the target capacity damper and the percentage of air volume, and determine the segmentation point. Based on the segmentation point, it establishes a piecewise function of the opening degree-percentage of air volume of the target capacity damper to determine its nonlinear flow characteristic curve.

[0049] It should be understood that any parts not described in detail in this specification belong to the prior art.

[0050] It should be understood that the above description of the preferred embodiments is quite detailed, but this should not be construed as limiting the scope of protection of this invention. It is neither necessary nor possible to exhaustively describe all possible implementations. Those skilled in the art, guided by this invention, can make substitutions or modifications without departing from the scope of the claims, all of which fall within the scope of protection of this invention. The scope of protection of this invention should be determined by the appended claims.

Claims

1. A method for determining the nonlinear flow characteristic curve of a coal mill capacity damper, characterized in that, Includes the following steps: A unit power-fuel quantity model based on neural networks is constructed, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Through field tests, under single-variable control conditions, data on the relationship between the unit's active power and the target capacity damper opening were obtained. Based on the unit power-fuel model and the corresponding data of the unit active power and target capacity damper opening, the relationship between the target capacity damper opening and the air volume percentage is calculated. Data completion and inflection point identification are performed on the relationship between the target capacity damper opening and the air volume percentage, and the segmentation point is determined. Based on the segmentation point, a piecewise function of the target capacity damper opening-flow percentage is established to determine its nonlinear flow characteristic curve. Repeat the above steps to obtain the nonlinear flow characteristic curves of all coal mill capacity dampers in sequence.

2. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 1, characterized in that, The construction of the unit power-fuel model based on neural networks includes: Steady-state operating condition data were selected from historical data to construct a sample set of unit power and fuel input. The unit power-fuel quantity model is trained using a radial basis function neural network, with the unit active power as input and the amount of fuel fed into the furnace as output. The unit power-fuel quantity model is trained and optimized using training, validation, and test sets until the model error meets the preset accuracy requirements.

3. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 1, characterized in that, The selection criteria for the steady-state operating condition data are based on the stability of the unit power command, main steam pressure, and pulverizing system operating parameters.

4. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 1, characterized in that, The field tests included: Adjust the unit to rated load conditions and establish a stable baseline; Adjust only the target capacity damper opening, keep other parameters fixed, and record the unit active power corresponding to different opening ranges.

5. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 4, characterized in that, When adjusting the opening of the target capacity damper, different adjustment step sizes are used in different opening ranges of the damper.

6. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 1, characterized in that, The relationship between the opening degree of the damper used to calculate the target capacity and the percentage of air volume includes: The active power of the unit obtained from the field test is input into the unit power-fuel quantity model to obtain the theoretical fuel quantity; Using the theoretical fuel quantity when the target capacity damper is fully open as a benchmark, the percentage of air volume at each opening degree is calculated as the ratio of the theoretical fuel quantity to the benchmark value.

7. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 1, characterized in that, Data completion and inflection point completion for the relationship between the target capacity damper opening and the percentage of air volume include: Interpolation is performed on discrete test data points to complete the data within the entire opening range, and the completed opening-airflow percentage data is obtained. Based on the completed opening-airflow percentage data, the characteristic inflection point is identified by analyzing its rate of change.

8. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 7, characterized in that, The inflection point of the recognition characteristic includes: Calculate the first and / or second derivatives of the completed opening-airflow percentage data, and preliminarily determine candidate inflection points based on the extreme points of the first derivative.

9. The method for determining the nonlinear flow characteristic curve of a coal mill capacity damper according to claim 8, characterized in that, Determining segmentation points based on candidate inflection points includes: sorting the candidate inflection points in ascending order according to their corresponding opening values; and filtering the sorted candidate inflection points according to preset rules to determine the segmentation points used to construct the piecewise function.

10. A system for determining the nonlinear flow characteristic curve of a coal mill capacity damper, characterized in that, include: Model building module: It is used to build a unit power-fuel quantity model based on neural network, which is used to map the relationship between the unit's active power and the amount of fuel fed into the furnace; Test control and data acquisition module: It is used to obtain the corresponding data of the active power of the unit and the damper opening of the target capacity through field test under single variable control conditions; Data conversion module: It is used to calculate the relationship between the target capacity damper opening and the percentage of air volume based on the power-fuel model of the unit and the corresponding data of the active power of the unit and the target capacity damper opening; Characteristic curve generation module: It is used to complete the data and identify the inflection point of the relationship between the opening degree of the target capacity damper and the percentage of air volume, and determine the segmentation point. Based on the segmentation point, it establishes a piecewise function of the opening degree-percentage of air volume of the target capacity damper to determine its nonlinear flow characteristic curve. The system for determining the nonlinear flow characteristic curve of a coal mill capacity damper is used to perform the steps in the method for determining the nonlinear flow characteristic curve of a coal mill capacity damper as described in any one of claims 1-9.

Images