Dynamic energy-saving control method for ash conveying system of power plant

Abstract

The invention provides a dynamic energy-saving control method for a power plant ash conveying system, and the method comprises the following steps: S1, data collection and pseudo signal elimination: collecting multi-dimensional working condition data through a sensor group, and recognizing a pseudo signal based on a coupling criterion; s2, pipeline resistance coefficient self-calibration; s3, multi-bin pump coupling and decoupling are carried out; s4, carrying out pulsation suppression and optimal parameter calculation; s5, performing control execution; s6, performing closed-loop feedback optimization; according to the method, false signals are recognized and corrected through humidity, pressure and material level coupling criteria, a pipeline resistance coefficient self-calibration mechanism is combined, control deviation caused by signal distortion and parameter drift is effectively avoided, reliable data support is provided for control decision, the control precision is greatly improved, and the method is suitable for multi-bin pump coupling interference. Independent control is realized through cross-correlation analysis and a decoupling model; aiming at low-load airflow pulsation, a grading inhibition strategy is started based on the wind speed fluctuation rate, dynamic changes such as combustion working conditions and ash material characteristics are adapted, and the problem that a traditional control method is poor in adaptability is solved.

Description

Technical Field

[0001] This invention relates to an energy-saving control method, and more particularly to a dynamic energy-saving control method for a power plant ash conveying system, belonging to the field of energy-saving control technology for ash conveying. Background Technology

[0002] The ash conveying system in a power plant is a key auxiliary system for ensuring stable boiler operation and environmentally friendly emissions in coal-fired power plants. It uses pneumatic conveying to transport fly ash generated during boiler combustion to the ash silo. The accuracy and energy efficiency of the system's operation are directly related to the overall energy cost and equipment operational safety of the power plant. The mainstream control method for power plant ash conveying systems still relies on fixed parameter presets or manual experience adjustments. The core flaw of this method is its inability to adapt to the dynamic fluctuations of multi-dimensional operating conditions during ash conveying, leading to a disconnect between control parameters and actual needs. This results in a core contradiction between energy efficiency and stability: when the humidity and particle size of fly ash change with boiler combustion conditions, spurious signals can easily appear in the level and pressure signals collected by sensors. Existing control methods lack a mechanism for accurately identifying and correcting spurious signals based on the correlation characteristics of operating conditions, leading to distorted control decisions. Furthermore, long-term operation can cause ash and scale buildup on the inner walls of pipelines, resulting in a slow change in the pipeline resistance coefficient. The existing control methods do not perform real-time calibration of this coefficient, and still calculate the control values ​​such as conveying pressure and wind speed based on the initial pipeline parameters. This is prone to over-conveying at low loads or under-conveying at high loads. At the same time, the existing control methods only rely on general algorithms or fixed strategies to deal with problems such as air source pressure coupling interference when multiple compartment pumps are running in parallel and pipeline airflow pulsation under low load conditions. They lack targeted quantitative calculation and dynamic suppression schemes, which ultimately leads to high energy consumption per unit ash conveying in the ash conveying system. Moreover, safety risks such as pipe blockage and pipeline wear are difficult to control effectively, and cannot meet the power plant's dual requirements for energy saving and consumption reduction as well as safety and stability of the ash conveying system. To address the aforementioned technical issues, a dynamic energy-saving control method for power plant ash conveying systems is proposed. Summary of the Invention

[0003] In view of this, the present invention provides a dynamic energy-saving control method for a power plant ash conveying system to solve or alleviate the technical problems existing in the prior art, and at least provides a beneficial option.

[0004] The technical solution of this invention is implemented as follows: A dynamic energy-saving control method for a power plant ash conveying system, comprising the following steps: S1. Data Acquisition and False Signal Removal: Multi-dimensional working condition data is collected through a sensor array, false signals are identified based on coupling criteria, and the real working condition data is obtained after correction using the corresponding algorithm. S2. Pipeline resistance coefficient self-calibration: The ash conveying system is brought into an unloaded state. Based on the pipeline inner wall thickness signal and unloaded pressure difference in the real working condition data, the actual pipeline resistance coefficient is calculated by formula, and the pipeline parameter baseline of the control model is updated. S3. Multi-compartment pump coupling decoupling: The coupling interference strength is obtained through cross-correlation analysis. Based on this strength, the independent control compensation amount is calculated through the decoupling model to obtain the independent operating parameters after decoupling. S4. Pulsation Suppression and Optimal Parameter Calculation: Calculate the wind speed fluctuation rate based on the wind speed inside the pipeline. When the fluctuation rate reaches the set threshold, the pulsation suppression strategy is activated. Input the real operating condition data, the actual pipeline resistance coefficient and independent operating condition parameters into the adaptive optimization model, and output the optimal control parameter set. S5. Control Execution: Convert the optimal control parameter set into execution signals and transmit them to the ash conveying system's actuators. The actuators adjust their operating status according to the signals. S6. Closed-loop feedback optimization: Collect feedback data after the actuator has been running, calculate the deviation between the feedback index and the preset threshold, adjust the corresponding algorithm and model parameters according to the deviation, and update the optimal control parameter set.

[0005] Further preferably, the multi-dimensional operating condition data in step S1 includes boiler load signal, ash hopper level signal, ash hopper humidity signal, ash conveying pipeline inlet and outlet pressure signal, pipeline wind speed signal, pipeline inner wall thickness signal, and gas source pressure signal.

[0006] Further preferably, the sensor group in step S1 includes a load sensor, an ultrasonic level sensor, a humidity sensor, a differential pressure sensor, a wind speed sensor, an ultrasonic thickness sensor, and a pressure sensor. The signal conversion accuracy of all sensors is ≥0.1%FS, the data acquisition cycle is fifty milliseconds, and the data is transmitted to the data processing unit via industrial Ethernet.

[0007] More preferably, the coupling criterion in step S1 is the humidity-pressure-level coupling criterion. When the real-time humidity of the ash hopper is greater than the first humidity threshold and the level signal change amplitude in adjacent acquisition cycles is greater than the level change threshold, it is determined to be a wet ash adhesion pseudo signal. The tenth-order moving average algorithm is used for correction. The corrected level signal is calculated based on the average level signal of ten consecutive acquisition cycles. The correction window is five consecutive acquisition cycles. When the real-time humidity of the ash hopper is less than the second humidity threshold, the wear on the inner wall of the pipe is greater than the wear threshold, and the fluctuation amplitude of the current pressure signal and the average pressure of the recent ten acquisition cycles is greater than the pressure fluctuation threshold, it is determined to be a true signal of dry ash wear loss. Piecewise linear interpolation algorithm is used for compensation, and it is divided into three levels according to the pressure fluctuation amplitude. Each level corresponds to a compensation coefficient of 1.02, 1.05, and 1.08, respectively. The compensated pressure signal is the product of the original pressure signal and the corresponding compensation coefficient.

[0008] In a further preferred embodiment, the no-load state in step S2 is achieved by closing all silo pump feed valves and maintaining a constant fan speed. The calibration process is performed during a fixed low-load period each day, during which multiple sets of pipeline inlet and outlet pressure differences are continuously collected and the average value is taken as the measured pressure difference.

[0009] A further preferred embodiment is the formula for calculating the actual pipeline resistance coefficient in step S2: ;

[0010] in, This is the actual pipeline resistance coefficient. To measure the pressure difference between the inlet and outlet of the pipeline, This refers to the air density under standard operating conditions. To fix the wind speed of the fan, The effective length of the pipe, The calibrated pipe flow cross-sectional area is calculated based on the pipe inner wall thickness signal. It is the circular area obtained by subtracting twice the average wear thickness of the pipe inner wall from the nominal inner diameter of the pipe. The average wear thickness of the pipe inner wall is obtained by the difference between the pipe inner wall thickness signal and the original inner wall thickness of the pipe.

[0011] Further preferably, the coupling interference intensity in step S3 is obtained by performing a cross-correlation calculation on the air source pressure fluctuation curves of any two chamber pumps. The cross-correlation coefficient is calculated using the following formula: ;

[0012] in, This represents the cross-correlation coefficient between the air source pressure signals of the two silo pumps. For the first silo pump The gas source pressure signal at any given time. For the second silo pump The gas source pressure signal at any given time. , These are the average values ​​of the air source pressure signals from the two silo pumps, respectively. For signal acquisition duration, The delay time is defined as follows: cross-correlation coefficients greater than 0.7 indicate strong coupling interference, 0.3 to 0.7 indicate medium coupling interference, and less than 0.3 indicate weak coupling interference. Each level corresponds to a decoupling weight of 0.8, 0.5, and 0.2, respectively. The decoupling model is a multi-input multi-output decoupling model. The independent control compensation is calculated by recursive least squares method, and a convergence threshold is set during the iterative process.

[0013] A further preferred embodiment is the formula for calculating wind speed fluctuation rate in step S4: ; in, For wind speed fluctuation rate, , , These are the maximum, minimum, and average wind speeds within the data collection period. A full-pulse suppression strategy is activated when the wind speed fluctuation rate is greater than 15%, a half-suppression strategy is activated when it is between 10% and 15%, and no suppression strategy is activated when it is less than 10%. The full-pulse suppression strategy includes: three-segment frequency modulation of the fan inverter, with each segment increasing by a fixed value and each segment lasting 1 second; the silo pump discharge valve gradually opens fully over 3 seconds, with the opening linearly increasing to 100% over time; and the ash amount is calculated based on the ash hopper level signal, and the critical wind speed is dynamically adjusted, with a lower critical wind speed limit for lower ash amounts. The adaptive optimization model is constructed based on the pneumatic ash conveying mechanism model and deep reinforcement learning algorithm, and the input data is processed using conventional normalization. The pneumatic ash conveying mechanism model includes a pipeline resistance calculation sub-model and an ash conveying critical wind speed calculation sub-model: The formula for the pipeline resistance calculation sub-model is: ; in, This refers to the actual resistance loss of the pipeline under ash conveying conditions. The effective inner diameter of the pipe. This refers to the actual wind speed under ash conveying conditions. This refers to the actual pipe resistance coefficient after calibration in step S2. The effective length of the pipe, This refers to the air density under standard operating conditions. The formula for the sub-model of calculating the critical wind speed for ash conveying is:

[0014] in, The critical wind speed for ash conveying. This is a correction factor for the properties of ash materials. The average particle size of fly ash is... This refers to the bulk density of fly ash. This refers to the air density under standard operating conditions. The output optimal control parameter set includes the target conveying pressure, segmented frequency modulation parameters, and ash conveying cycle; the reward function formula for the deep reinforcement learning algorithm is: ; in, As a reward value, For energy consumption reduction rate, For wind speed fluctuation rate, The value for pipe blockage risk is set to 1 when the pipeline pressure change rate is greater than 0.1 MPa / s, otherwise it is reduced proportionally. The algorithm learning rate is set to 0.001, and the experience playback buffer capacity is set to 10,000 records.

[0015] More preferably, the execution signal in step S5 is a standard analog signal or a Modbus-RTU digital signal; the standard analog signal is a 4-20mA signal, and its current value has a linear correspondence with the control parameter value; The actuators include the fan frequency converter, the silo pump feed valve, the silo pump discharge valve, and the air source regulating valve. The response delay of the actuators is ≤200ms, and the parameter adjustment accuracy is ≤±2%.

[0016] Further preferably, the feedback data in step S6 includes actual conveying pressure, actual wind speed, ash conveying efficiency, and pipeline pressure change rate; the preset thresholds for the feedback indicators include spurious signal rejection accuracy, pipeline resistance coefficient calibration deviation, coupling interference cancellation rate, and pulsation fluctuation rate; the deviation value is calculated using the mean square error formula: ; in, This is the deviation value. For the first The actual value of the feedback indicator collected this time. Preset thresholds for feedback metrics. The sampling number is 1. When the deviation exceeds the allowable range, the threshold of the pseudo-signal elimination algorithm is finely adjusted by ±2%, the weight of the decoupling model is corrected by ±0.1, and the weight of the reinforcement learning reward function is adjusted by ±0.05. The update cycle is ten seconds. After each update, five sets of data are collected for verification. If the verification is successful, the updated parameters are maintained; if it is unsuccessful, they are readjusted.

[0017] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions: I. This invention identifies and corrects spurious signals through a coupling criterion of humidity, pressure, and material level. Combined with a pipeline resistance coefficient self-calibration mechanism, it effectively avoids control deviations caused by signal distortion and parameter drift, providing reliable data support for control decisions and significantly improving control accuracy. For multi-compartment pump coupling interference, it achieves independent control through cross-correlation analysis and decoupling models. For low-load airflow pulsation, it initiates a graded suppression strategy based on wind speed fluctuation rate, adapting to dynamic changes in combustion conditions, ash characteristics, etc., thus solving the problem of poor adaptability of traditional control methods. Second, this invention integrates the pneumatic ash conveying mechanism with deep reinforcement learning through an adaptive optimization model to output the optimal set of control parameters. It reduces excessive energy consumption during conveying by precisely regulating pressure and frequency, and avoids the risk of pipe blockage by dynamically adjusting the critical wind speed. This reduces energy consumption per unit of ash conveying while ensuring system stability. The closed-loop feedback optimization mechanism adjusts the algorithm and model parameters in real time according to the deviation value, continuously adapting to long-term changes such as pipe ash accumulation and equipment wear, reducing pipe wear and equipment failure, extending the service life of the ash conveying system, reducing operation and maintenance costs, and meeting the dual requirements of power plant energy conservation and safety.

[0018] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application 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 only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the dynamic energy-saving control method for power plant ash conveying system according to the present invention; Figure 2 This is a flowchart of the data acquisition and pseudo-signal removal process for the dynamic energy-saving control method of the power plant ash conveying system of the present invention; Figure 3 This is a flowchart illustrating the self-calibration process of the pipeline resistance coefficient in the dynamic energy-saving control method for power plant ash conveying systems of the present invention. Figure 4 This is a flowchart of the multi-compartment pump coupling and decoupling process for the dynamic energy-saving control method of the power plant ash conveying system of the present invention; Figure 5 This is a flowchart of the pulsation suppression and optimal parameter calculation method for the dynamic energy-saving control method of the power plant ash conveying system of the present invention. Figure 6 This is a flowchart illustrating the closed-loop feedback optimization process of the dynamic energy-saving control method for power plant ash conveying systems according to the present invention. Detailed Implementation

[0021] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0022] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] like Figure 1-6 As shown, this embodiment of the invention provides a dynamic energy-saving control method for a power plant ash conveying system, comprising the following steps: S1. Data Acquisition and False Signal Removal: Multi-dimensional operating condition data is acquired through a sensor array. False signals are identified based on coupling criteria, and corrected using corresponding algorithms to obtain the true operating condition data. The multi-dimensional operating condition data includes boiler load signal, ash hopper level signal, ash hopper humidity signal, ash conveying pipeline inlet and outlet pressure signal, pipeline wind speed signal, pipeline inner wall thickness signal, and gas source pressure signal. The sensor array includes a load sensor, ultrasonic level sensor, humidity sensor, differential pressure sensor, wind speed sensor, ultrasonic thickness sensor, and pressure sensor. The signal conversion accuracy of all sensors is ≥0.1%FS, the data acquisition cycle is fifty milliseconds, and the data is transmitted to the data processing unit via industrial Ethernet. The coupling criterion is the humidity-pressure-level coupling criterion. When the real-time humidity of the ash hopper is greater than the first humidity threshold and the level signal change amplitude of adjacent acquisition cycles is greater than the level change threshold, it is determined to be a wet ash adhesion pseudo signal. The tenth-order moving average algorithm is used for correction. The corrected level signal is calculated based on the average level signal of ten consecutive acquisition cycles. The correction window is five consecutive acquisition cycles. When the real-time humidity of the ash hopper is less than the second humidity threshold, the wear on the inner wall of the pipe is greater than the wear threshold, and the fluctuation amplitude of the current pressure signal and the average pressure of the recent ten acquisition cycles is greater than the pressure fluctuation threshold, it is determined to be a true signal of dry ash wear loss. Piecewise linear interpolation algorithm is used for compensation, and it is divided into three levels according to the pressure fluctuation amplitude. Each level corresponds to a compensation coefficient of 1.02, 1.05, and 1.08, respectively. The compensated pressure signal is the product of the original pressure signal and the corresponding compensation coefficient.

[0024] In one embodiment, a sensor array is deployed at key locations in the ash conveying system: a load sensor is installed at the boiler furnace outlet, an ultrasonic level sensor and a humidity sensor are installed at the top of the ash hopper, differential pressure sensors are installed at the inlet and outlet of the ash conveying pipeline, a wind speed sensor and an ultrasonic thickness sensor are installed in the middle of the pipeline, and a pressure sensor is installed in the gas source pipeline. Based on the humidity, pressure, and material level coupling criteria for identifying false signals, the following thresholds are set: a first humidity threshold of 60%RH and a second threshold of 30%RH; a material level change threshold of 1m between adjacent acquisition cycles; a pipe inner wall wear threshold of 0.5mm; and a pressure fluctuation threshold of 10% of the difference between the current pressure signal and the average pressure of the most recent ten acquisition cycles. When the real-time humidity in the ash hopper reaches 65%RH and the material level signal changes abruptly from 2.3m to 3.5m in an adjacent acquisition cycle, exceeding the threshold, it is identified as a false signal of wet ash adhesion. A tenth-order moving average algorithm is used for correction, with the material level signals of ten consecutive acquisition cycles as the calculation basis. The correction window was set to five consecutive acquisition cycles. After correction, the material level signal stabilized at 2.4m. When the real-time humidity of the ash hopper was 25%, the wear on the inner wall of the pipe was 0.6mm, and the fluctuation of the current pressure signal of 0.55MPa and the average pressure of the last ten acquisition cycles of 0.5MPa reached 10%, it was determined to be a false signal of dry ash wear loss. Piecewise linear interpolation algorithm was used for compensation. The compensation coefficient was selected according to the pressure fluctuation amplitude. The pressure fluctuation amplitude in this case belonged to the 10% level, corresponding to a compensation coefficient of 1.02. After compensation, the pressure signal was 0.56MPa. It was verified that the accuracy of false signal removal reached 92%.

[0025] S2. Pipeline Resistance Coefficient Self-Calibration: The ash conveying system is brought to an unloaded state. Based on the pipeline inner wall thickness signal and unloaded pressure difference from real operating data, the actual pipeline resistance coefficient is calculated using a formula, updating the pipeline parameter baseline of the control model. The unloaded state is achieved by closing all silo pump inlet valves and maintaining a constant fan speed. The calibration process is performed during fixed low-load periods each day, during which multiple sets of pipeline inlet and outlet pressure differences are continuously collected, and the average value is taken as the measured pressure difference. The formula for calculating the actual pipeline resistance coefficient is: ; in, This is the actual pipeline resistance coefficient. To measure the pressure difference between the inlet and outlet of the pipeline, This refers to the air density under standard operating conditions. To fix the wind speed of the fan, The effective length of the pipe, The calibrated pipe flow cross-sectional area is calculated based on the pipe inner wall thickness signal. It is the circular area obtained by subtracting twice the average wear thickness of the pipe inner wall from the nominal inner diameter of the pipe. The average wear thickness of the pipe inner wall is obtained by the difference between the pipe inner wall thickness signal and the original inner wall thickness of the pipe.

[0026] In one embodiment, the calibration process is performed during the low-load period from 2:00 to 2:10 AM daily, when the boiler load is maintained at 30% of the rated load, meeting the requirements for no-load calibration. By closing all silo pump feed valves and maintaining a constant fan speed of 12 m / s, the ash conveying system is brought into a no-load state. Within a ten-minute calibration window, 20 sets of pipeline inlet and outlet pressure difference data are continuously collected, which are 0.078 MPa, 0.082 MPa, 0.079 MPa, etc. The average value is taken to obtain the measured pressure difference ΔP as 0.08 MPa. Calculate the calibrated pipe flow cross-sectional area based on the pipe inner wall thickness signal. The original inner wall thickness of the pipeline was 0.01m, and the currently acquired inner wall thickness signal is 0.008m. Therefore, the average wear thickness of the pipeline inner wall is 0.002m. The calibrated pipeline flow cross-sectional area... for The calculated result is 0.0707 m²; the measured pressure difference will be... Air density under standard operating conditions Fan fixed wind speed Effective length of pipe and the cross-sectional area of ​​the calibrated pipeline Substituting into the actual pipeline resistance coefficient calculation formula, we can obtain the following: The calculated result is 0.028. Compared with the baseline value of 0.026 for the pipeline resistance coefficient in the current control model, the deviation is 7.7%, which exceeds the 5% update threshold. Therefore, the pipeline parameter baseline of the control model is immediately updated with the calculated actual pipeline resistance coefficient of 0.028.

[0027] S3. Multi-compartment pump coupling decoupling: The coupling interference strength is obtained through cross-correlation analysis. Based on this strength, the independent control compensation is calculated using a decoupling model to obtain the independent operating parameters after decoupling. The coupling interference strength is obtained by performing cross-correlation calculations on the air source pressure fluctuation curves of any two compartment pumps. The cross-correlation coefficient calculation formula is as follows: ; in, This represents the cross-correlation coefficient between the air source pressure signals of the two silo pumps. For the first silo pump The gas source pressure signal at any given time. For the second silo pump The gas source pressure signal at any given time. , These are the average values ​​of the air source pressure signals from the two silo pumps, respectively. For signal acquisition duration, The delay time is defined as follows: cross-correlation coefficients greater than 0.7 indicate strong coupling interference, 0.3 to 0.7 indicate medium coupling interference, and less than 0.3 indicate weak coupling interference. Each level corresponds to a decoupling weight of 0.8, 0.5, and 0.2, respectively. The decoupling model is a multi-input multi-output decoupling model. The independent control compensation is calculated by recursive least squares method, and a convergence threshold is set during the iterative process.

[0028] In one embodiment, the gas source pressure fluctuation curves of four silo pumps are collected, with a sampling frequency set to 20Hz and a signal acquisition duration of 10s; cross-correlation is performed on the gas source pressure fluctuation curves of any two silo pumps to calculate the cross-correlation coefficient. The first silo pump Gas source pressure signal at any time Second silo pump Gas source pressure signal at any time , As a delay time, values ​​within the range of 0-50ms were used to calculate the average value of the air source pressure signals from the two silo pumps. and Substituting the values ​​into the cross-correlation formula, the cross-correlation coefficient between pump 1 and pump 2 is 0.75, which is considered strong coupling interference, corresponding to a decoupling weight of 0.8; the cross-correlation coefficient between pump 3 and pump 4 is 0.45, which is considered medium coupling interference, corresponding to a decoupling weight of 0.5. The operating parameters and coupling interference intensity of each pump are input into a multi-input multi-output decoupling model. The independent control compensation is calculated iteratively using the recursive least squares method, with the iterative convergence threshold set as follows: After 30 iterations, the system reached a convergence state. For the strong coupling interference between silo pump 1 and silo pump 2, the calculated compensation amounts were +0.03MPa for air source pressure, +0.5m / s for wind speed, and +5s for ash conveying cycle. For the moderate coupling interference between silo pump 3 and silo pump 4, the calculated compensation amounts were +0.02MPa for air source pressure, +0.3m / s for wind speed, and +3s for ash conveying cycle. The coupling interference was offset by the compensation amounts, and the decoupled independent operating parameters were obtained.

[0029] S4. Pulsation Suppression and Optimal Parameter Calculation: Based on the wind speed within the pipeline, the wind speed fluctuation rate is calculated. When the fluctuation rate reaches a set threshold, a pulsation suppression strategy is activated. Real-world operating data, actual pipeline resistance coefficients, and independent operating parameters are input into the adaptive optimization model, which outputs the optimal control parameter set. The formula for calculating wind speed fluctuation rate is: ; in, For wind speed fluctuation rate, , , These are the maximum, minimum, and average wind speeds within the data collection period. A full-pulse suppression strategy is activated when the wind speed fluctuation rate is greater than 15%, a half-suppression strategy is activated when it is between 10% and 15%, and no suppression strategy is activated when it is less than 10%. The full-pulse suppression strategy includes: three-segment frequency modulation of the fan inverter, with each segment increasing by a fixed value and each segment lasting 1 second; the silo pump discharge valve gradually opens fully over 3 seconds, with the opening linearly increasing to 100% over time; and the ash amount is calculated based on the ash hopper level signal, and the critical wind speed is dynamically adjusted, with a lower critical wind speed limit for lower ash amounts. The adaptive optimization model is constructed based on the pneumatic ash conveying mechanism model and deep reinforcement learning algorithm, and the input data is processed using conventional normalization. The pneumatic ash conveying mechanism model includes a pipeline resistance calculation sub-model and an ash conveying critical wind speed calculation sub-model: the formula for the pipeline resistance calculation sub-model is as follows: ; in, This refers to the actual resistance loss of the pipeline under ash conveying conditions. The effective inner diameter of the pipe. This refers to the actual wind speed under ash conveying conditions. This refers to the actual pipe resistance coefficient after calibration in step S2. The effective length of the pipe, This refers to the air density under standard operating conditions. The formula for the sub-model of calculating the critical wind speed for ash conveying is: ; in, The critical wind speed for ash conveying. This is a correction factor for the properties of ash materials. The average particle size of fly ash is... This refers to the bulk density of fly ash. This refers to the air density under standard operating conditions. The output optimal control parameter set includes the target conveying pressure, segmented frequency modulation parameters, and ash conveying cycle; the reward function formula for the deep reinforcement learning algorithm is: ; in, As a reward value, For energy consumption reduction rate, For wind speed fluctuation rate, The value for pipe blockage risk is set to 1 when the pipeline pressure change rate is greater than 0.1 MPa / s, otherwise it is reduced proportionally. The algorithm learning rate is set to 0.001, and the experience playback buffer capacity is set to 10,000 records.

[0030] In one embodiment, wind speed fluctuation rate is calculated based on the wind speed signal inside the pipe. Wind speed data from 10 collection periods were selected, with the maximum wind speed among them. The minimum wind speed is 14 m / s. The average wind speed is 10 m / s. The value is 12 m / s. Substituting this into the wind speed fluctuation formula, we obtain... The percentage was 33.3%, exceeding the 15% threshold for initiating the full pulsation suppression strategy, thus triggering the full pulsation suppression strategy. The full-pulse suppression strategy is implemented as follows: the fan frequency converter adopts a three-stage segmented frequency regulation, with the initial frequency set at 35Hz, the transition frequency at 40Hz, and the target frequency at 45Hz. Each segment of frequency regulation lasts for 1 second to avoid sudden increases and decreases in wind speed; the silo pump discharge valve adopts a 3-second gradual full-opening mode, with the opening degree linearly increasing from 0% to 100%, increasing by 33.3% every second; the ash amount is calculated based on the ash hopper level signal. The current ash hopper level is 3m, corresponding to an ash amount of 8t / h, and the critical wind speed is dynamically adjusted to 10m / s; The actual operating data, the calibrated actual pipeline resistance coefficient of 0.028, and the decoupled independent operating parameters were input into the adaptive optimization model. This model is based on the pneumatic ash conveying mechanism model and a deep reinforcement learning algorithm. The input data underwent conventional normalization processing, normalizing each parameter to the [0,1] interval. The pipeline resistance calculation formula in the pneumatic ash conveying mechanism model was substituted into the numerical calculation to obtain the results. The calculated result is 0.37 MPa, where The effective inner diameter of the pipe is taken as 0.296m. The actual wind speed under ash conveying conditions is 12 m / s; the formula for calculating the critical wind speed for ash conveying is substituted into the numerical calculation to obtain... The calculated result is 9.8 m / s, where This is a correction factor for ash properties, adjusted to 1.1 based on the current ash hopper moisture content of 25%. The average particle size of fly ash is 0.1 mm; the deep reinforcement learning algorithm used is the DQN algorithm, the learning rate is set to 0.001, the experience replay buffer capacity is set to 10,000 records, and the reward value is calculated using the reward function formula. ,in For a 25% reduction in energy consumption, The current wind speed fluctuation rate is 33.3%. The risk value for pipe blockage is 0; the adaptive optimization model outputs the optimal set of control parameters, including the target conveying pressure of 0.45MPa, segmented frequency modulation parameters of 35Hz-40Hz-45Hz, and ash conveying cycle of 60s.

[0031] S5. Control Execution: Convert the optimal control parameter set into an execution signal and transmit it to the ash conveying system's actuator. The actuator adjusts its operating state according to the signal. The execution signal is a standard analog signal or a Modbus-RTU digital signal. The standard analog signal is a 4-20mA signal, and its current value has a linear correspondence with the control parameter value. The actuators include the fan frequency converter, the silo pump feed valve, the silo pump discharge valve, and the air source regulating valve. The response delay of the actuators is ≤200ms, and the parameter adjustment accuracy is ≤±2%.

[0032] In one embodiment, the optimal control parameter set is converted into an execution signal. The target conveying pressure of 0.45 MPa, the segmented frequency modulation parameters, and the ash conveying cycle of 60 s are converted into a 4-20 mA standard analog signal. The minimum control parameter value corresponds to 4 mA, and the maximum value corresponds to 20 mA. The target conveying pressure of 0.45 MPa corresponds to an analog signal current value of 12 mA. The execution signal is transmitted to the actuator of the ash conveying system. The response delay of the actuator is 150 ms, and the parameter adjustment accuracy reaches ±1.5%. After receiving the segmented frequency modulation signal, the fan frequency converter gradually adjusts the frequency in the order of 35 Hz-40 Hz-45 Hz. The silo pump discharge valve operates in a 3-second gradual full-open mode. The air source regulating valve adjusts its opening according to the target conveying pressure signal, so that the ash conveying system operates stably according to the optimal control parameter set. The actual conveying pressure detected after operation is 0.445 MPa, which meets the parameter adjustment accuracy requirements.

[0033] S6. Closed-Loop Feedback Optimization: Collect feedback data after the actuator has run, calculate the deviation between the feedback index and the preset threshold, adjust the corresponding algorithm and model parameters according to the deviation, and update the optimal control parameter set; the feedback data includes actual conveying pressure, actual wind speed, ash conveying efficiency, and pipeline pressure change rate; the preset thresholds for the feedback index include spurious signal rejection accuracy, pipeline resistance coefficient calibration deviation, coupling interference cancellation rate, and pulsation fluctuation rate; the deviation is calculated using the mean square error formula: ; in, This is the deviation value. For the first The actual value of the feedback indicator collected this time. Preset thresholds for feedback metrics. The sampling number is 1. When the deviation exceeds the allowable range, the threshold of the pseudo-signal elimination algorithm is finely adjusted by ±2%, the weight of the decoupling model is corrected by ±0.1, and the weight of the reinforcement learning reward function is adjusted by ±0.05. The update cycle is ten seconds. After each update, five sets of data are collected for verification. If the verification is successful, the updated parameters are maintained; if it is unsuccessful, they are readjusted.

[0034] In one embodiment, feedback data after the actuator operates is collected, including actual conveying pressure of 0.445 MPa, actual wind speed of 12.2 m / s, ash conveying efficiency of 95%, and pipeline pressure change rate of 0.03 MPa / s. Based on the feedback data, core feedback indicators are calculated: pseudo-signal rejection accuracy of 92%, pipeline resistance coefficient calibration deviation of 2.5%, coupling interference cancellation rate of 88%, and pulsation fluctuation rate of 12%. The preset thresholds for each feedback indicator are: pseudo-signal rejection accuracy ≥90%, pipeline resistance coefficient calibration deviation ≤3%, coupling interference cancellation rate ≥85%, and pulsation fluctuation rate ≤15%. Substitute the actual value of the feedback indicator and the preset threshold into the mean square error formula, where The accuracy of spurious signal removal was calculated based on 5 sampling times. Pipeline resistance coefficient calibration deviation Coupling interference cancellation rate pulsating volatility All deviations were within the allowable range; the update cycle was set to ten seconds, and five sets of data were collected for verification after each update. The verification results showed that all feedback indicators met the preset threshold requirements. The current algorithm and model parameters were maintained, and the closed-loop feedback optimization was completed to ensure the continuous and stable operation of the ash conveying system.

[0035] In operation, this invention: collects multi-dimensional operating condition data such as boiler load, ash hopper level, and pipeline resistance through multiple types of sensors; identifies false signals based on humidity, pressure, and level coupling criteria; outputs true operating condition data after correction by corresponding algorithms; during fixed low-load periods, the system is brought into an unloaded state; combining the pipeline inner wall thickness signal and the unloaded pressure difference, the actual pipeline resistance coefficient is calculated using a formula, and the baseline of the control model parameters is updated; cross-correlation analysis is performed on the pressure fluctuation curve of the silo pump air source to obtain the coupling interference intensity; compensation is calculated through a decoupling model to obtain independent operating condition parameters; wind speed fluctuation rate is calculated and corresponding suppression strategies are activated; the true operating condition data, resistance coefficient, and independent parameters are input into the adaptive optimization model, and the optimal control parameter set, such as the target delivery pressure, is output; the optimal parameters are converted into execution signals to drive the fan frequency converter, silo pump valves, and other actuators to adjust their operation; feedback data after execution is collected, the deviation from the preset threshold is calculated, the algorithm and model parameters are adjusted, and the optimal control parameters are updated to achieve dynamic iterative optimization.

[0036] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in the present invention, and these should all be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A dynamic energy-saving control method for a power plant ash conveying system, characterized in that, Includes the following steps: S1. Data Acquisition and False Signal Removal: Multi-dimensional working condition data is collected through a sensor array, false signals are identified based on coupling criteria, and the real working condition data is obtained after correction using the corresponding algorithm. S2. Pipeline resistance coefficient self-calibration: The ash conveying system is brought into an unloaded state. Based on the pipeline inner wall thickness signal and unloaded pressure difference in the real working condition data, the actual pipeline resistance coefficient is calculated by formula, and the pipeline parameter baseline of the control model is updated. S3. Multi-compartment pump coupling decoupling: The coupling interference strength is obtained through cross-correlation analysis. Based on this strength, the independent control compensation amount is calculated through the decoupling model to obtain the independent operating parameters after decoupling. S4. Pulsation Suppression and Optimal Parameter Calculation: Calculate the wind speed fluctuation rate based on the wind speed inside the pipeline. When the fluctuation rate reaches the set threshold, the pulsation suppression strategy is activated. Input the real operating condition data, the actual pipeline resistance coefficient and independent operating condition parameters into the adaptive optimization model, and output the optimal control parameter set. S5. Control Execution: Convert the optimal control parameter set into execution signals and transmit them to the ash conveying system's actuators. The actuators adjust their operating status according to the signals. S6. Closed-loop feedback optimization: Collect feedback data after the actuator has been running, calculate the deviation between the feedback index and the preset threshold, adjust the corresponding algorithm and model parameters according to the deviation, and update the optimal control parameter set.

2. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The multi-dimensional operating data in step S1 includes boiler load signal, ash hopper level signal, ash hopper humidity signal, ash conveying pipeline inlet and outlet pressure signal, pipeline wind speed signal, pipeline inner wall thickness signal, and gas source pressure signal.

3. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The sensor group in step S1 includes a load sensor, an ultrasonic level sensor, a humidity sensor, a differential pressure sensor, a wind speed sensor, an ultrasonic thickness sensor, and a pressure sensor. The signal conversion accuracy of all sensors is ≥0.1%FS, the data acquisition cycle is fifty milliseconds, and the data is transmitted to the data processing unit via industrial Ethernet.

4. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The coupling criterion in step S1 is the humidity-pressure-level coupling criterion. When the real-time humidity of the ash hopper is greater than the first humidity threshold and the level signal change amplitude of the adjacent acquisition cycle is greater than the level change threshold, it is determined to be a wet ash adhesion pseudo signal. The tenth-order moving average algorithm is used for correction. The corrected level signal is calculated based on the average level signal of ten consecutive acquisition cycles. The correction window is five consecutive acquisition cycles. When the real-time humidity of the ash hopper is less than the second humidity threshold, the wear on the inner wall of the pipe is greater than the wear threshold, and the fluctuation amplitude of the current pressure signal and the average pressure of the recent ten acquisition cycles is greater than the pressure fluctuation threshold, it is determined to be a true signal of dry ash wear loss. Piecewise linear interpolation algorithm is used for compensation, and it is divided into three levels according to the pressure fluctuation amplitude. Each level corresponds to a compensation coefficient of 1.02, 1.05, and 1.08, respectively. The compensated pressure signal is the product of the original pressure signal and the corresponding compensation coefficient.

5. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The no-load state in step S2 is achieved by closing all silo pump feed valves and maintaining a constant fan speed. The calibration process is performed during a fixed low-load period each day, during which multiple sets of pipeline inlet and outlet pressure differences are continuously collected and the average value is taken as the measured pressure difference.

6. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The formula for calculating the actual pipeline resistance coefficient in step S2 is as follows: in, This is the actual pipeline resistance coefficient. To measure the pressure difference between the inlet and outlet of the pipeline, This refers to the air density under standard operating conditions. To fix the wind speed of the fan, The effective length of the pipe, The calibrated pipe flow cross-sectional area is calculated based on the pipe inner wall thickness signal. It is the circular area obtained by subtracting twice the average wear thickness of the pipe inner wall from the nominal inner diameter of the pipe. The average wear thickness of the pipe inner wall is obtained by the difference between the pipe inner wall thickness signal and the original inner wall thickness of the pipe.

7. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The coupling interference intensity in step S3 is obtained by performing a cross-correlation calculation on the gas source pressure fluctuation curves of any two silo pumps. The cross-correlation coefficient is calculated using the following formula: in, This represents the cross-correlation coefficient between the air source pressure signals of the two silo pumps. For the first silo pump The gas source pressure signal at any given time. For the second silo pump The gas source pressure signal at any given time. , These are the average values ​​of the air source pressure signals from the two silo pumps, respectively. For signal acquisition duration, The delay time is defined as follows: cross-correlation coefficients greater than 0.7 indicate strong coupling interference, 0.3 to 0.7 indicate medium coupling interference, and less than 0.3 indicate weak coupling interference. Each level corresponds to a decoupling weight of 0.8, 0.5, and 0.2, respectively. The decoupling model is a multi-input multi-output decoupling model. The independent control compensation is calculated by recursive least squares method, and a convergence threshold is set during the iterative process.

8. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The formula for calculating the wind speed fluctuation rate in step S4 is: in, For wind speed fluctuation rate, , , These are the maximum, minimum, and average wind speeds within the data collection period. A full-pulse suppression strategy is activated when the wind speed fluctuation rate is greater than 15%, a half-suppression strategy is activated when it is between 10% and 15%, and no suppression strategy is activated when it is less than 10%. The full-pulse suppression strategy includes: three-segment frequency modulation of the fan inverter, with each segment increasing by a fixed value and each segment lasting 1 second; the silo pump discharge valve gradually opens fully over 3 seconds, with the opening linearly increasing to 100% over time; and the ash amount is calculated based on the ash hopper level signal, and the critical wind speed is dynamically adjusted, with a lower critical wind speed limit for lower ash amounts. The adaptive optimization model is constructed based on the pneumatic ash conveying mechanism model and deep reinforcement learning algorithm, and the input data is processed using conventional normalization. The pneumatic ash conveying mechanism model includes a pipeline resistance calculation sub-model and an ash conveying critical wind speed calculation sub-model: The formula for the pipeline resistance calculation sub-model is: in, This refers to the actual resistance loss of the pipeline under ash conveying conditions. The effective inner diameter of the pipe. This refers to the actual wind speed under ash conveying conditions. This refers to the actual pipe resistance coefficient after calibration in step S2. The effective length of the pipe, This refers to the air density under standard operating conditions. The formula for the sub-model of calculating the critical wind speed for ash conveying is: in, The critical wind speed for ash conveying. This is a correction factor for the properties of ash materials. The average particle size of fly ash is... This refers to the bulk density of fly ash. This refers to the air density under standard operating conditions. The output optimal control parameter set includes the target conveying pressure, segmented frequency modulation parameters, and ash conveying cycle; the reward function formula for the deep reinforcement learning algorithm is: in, As a reward value, For energy consumption reduction rate, For wind speed fluctuation rate, The value for pipe blockage risk is set to 1 when the pipeline pressure change rate is greater than 0.1 MPa / s, otherwise it is reduced proportionally. The algorithm learning rate is set to 0.001, and the experience playback buffer capacity is set to 10,000 records.

9. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The execution signal in step S5 is a standard analog signal or a Modbus-RTU digital signal; the standard analog signal is a 4-20mA signal, and its current value has a linear correspondence with the control parameter value. The actuators include the fan frequency converter, the silo pump feed valve, the silo pump discharge valve, and the air source regulating valve. The response delay of the actuators is ≤200ms, and the parameter adjustment accuracy is ≤±2%.

10. The dynamic energy-saving control method for a power plant ash conveying system according to claim 1, characterized in that, The feedback data in step S6 includes actual conveying pressure, actual wind speed, ash conveying efficiency, and pipeline pressure change rate; the preset thresholds for the feedback indicators include spurious signal rejection accuracy, pipeline resistance coefficient calibration deviation, coupling interference cancellation rate, and pulsation fluctuation rate; the deviation value is calculated using the mean square error formula: in, This is the deviation value. For the first The actual values ​​of the feedback indicators collected this time. Preset thresholds for feedback metrics. The sampling number is 1. When the deviation exceeds the allowable range, the threshold of the pseudo-signal elimination algorithm is finely adjusted by ±2%, the weight of the decoupling model is corrected by ±0.1, and the weight of the reinforcement learning reward function is adjusted by ±0.

05. The update cycle is ten seconds. After each update, five sets of data are collected for verification. If the verification is successful, the updated parameters are maintained; if it is unsuccessful, they are readjusted.

Images