Coal conveying control system and method for thermal power plant

Abstract

The invention relates to a coal conveying control system and method for a thermal power plant. The method comprises the following steps that S1, operation data of a coal conveying system are collected in real time through a preset multi-source sensor network; s2, constructing a time sequence prediction model based on an attention mechanism, and dynamically predicting the operation state of the coal conveying system; s3, executing real-time optimization calculation at the edge calculation node, and generating a control instruction set including the belt speed, the coal feeder opening degree and the transfer station cooperation strategy; s4, the control instruction is issued to an execution mechanism through a 5G-MEC network, and meanwhile key data is uploaded to a cloud end for model iteration updating; and S5, dynamically adjusting a safety protection threshold value based on an equipment vibration spectrum analysis result, and triggering a hierarchical protection mechanism when a coal blockage risk is detected. The method has the effects of effectively improving the performance of the coal conveying system and being green and energy-saving.

Description

Technical Field

[0001] The present application relates to the field of thermal power generation, and in particular to a coal transportation control system and method for a thermal power plant. Background Art

[0002] With the development of the power industry and the increasing demand for electricity, large-capacity and large-unit thermal power plants have gradually taken a dominant position in my country. The coal transportation system of thermal power plants is an important link to ensure the stable operation of power plants. Traditional coal transportation control systems mostly use relay control and manual operation, which have problems such as low automation, high failure rate, and low efficiency.

[0003] In the actual production of thermal power plants, constant-speed belt conveyors and screening and crushing equipment with fixed coal particle fineness are usually used to transport coal to boilers to ensure the reliability of electricity supply to users. Therefore, the coal conveyor belt load rate is usually low and the boiler operation efficiency is not high.

[0004] The Chinese patent application with the announcement number CN118938795A discloses a coal transportation control system and method for a thermal power plant, including a model prediction module for establishing an operating load prediction model, and obtaining the optimal value of the boiler operating load according to the operating load prediction model; a load calculation module for drawing an operating load difference curve, and determining the change in boiler coal consumption according to the operating load difference curve; a speed control module for determining the coal transportation volume of the coal transportation system, and controlling the operating speed of the coal transportation belt according to the coal transportation volume; a quality calculation module for calculating the boiler operating efficiency, and determining the coal quality parameters according to the boiler operating efficiency; a fineness control module for determining the coal particle parameters, setting the target coal particle fineness according to the coal particle parameters, and controlling the screening and crushing equipment according to the target coal particle fineness. This application predicts the optimal value of the boiler operating load by establishing an operating load prediction model, and then calculates the change in coal consumption by comparison, adjusts the coal transportation volume, and controls the belt speed. However, the load prediction model has a single data source and relies too much on historical data, and does not consider dynamic factors such as equipment aging and coal type changes. There are problems such as subjective threshold setting, failure to consider the influence of multiple factors, lack of theoretical support for the formula, and insufficient real-time and robustness. Forecasting deviations and adjustment lags are prone to occur, leading to increased coal consumption and higher operating costs.

[0005] In view of the above-mentioned related technologies, the present application proposes a coal transportation control system and method for a thermal power plant. Summary of the invention

[0006] In order to solve the above problems, the present application provides a coal transportation control system and method for a thermal power plant.

[0007] In the first aspect, the present application provides a method for controlling coal transportation in a thermal power plant, which adopts the following technical solution: A method for controlling coal transportation in a thermal power plant comprises the following steps: S1: Collect the operation data of the coal conveying system in real time through a pre-set multi-source sensing network. The operation data includes coal flow parameters, equipment status parameters, and environmental parameters. S2: Build a time series prediction model based on the attention mechanism to dynamically predict the operation status of the coal conveying system. S3: Perform real-time optimization calculations at the edge computing node to generate a control instruction set including belt speed, coal feeder opening, and transfer station coordination strategy. S4: Send the control instructions to the actuator through the 5G-MEC network, and at the same time upload the key data to the cloud for model iteration and update. S5: Dynamically adjust the safety protection threshold based on the equipment vibration spectrum analysis results, and trigger the hierarchical protection mechanism when the risk of coal blockage is detected.

[0008] Preferably, the multi-source sensing network in step S1 includes: An intersection detection array composed of a microwave moisture meter and a laser particle size analyzer arranged in the coal dropping pipe; A distributed vibration sensor network arranged along the coal conveying belt; An infrared thermal imager installed on the transfer station support; An acoustic emission sensor arranged on the bearing seat of the lower roller of the coal conveying belt.

[0009] Preferably, the intersection detection array adopts the following calibration method: Obtain the moisture content of the coal passing through the coal dropping pipe measured by the microwave moisture meter and the laser particle size analyzer in real time; Judge whether the deviation between the moisture content ω1 measured by the microwave moisture meter and the moisture content ω2 deduced by the laser particle size analyzer exceeds 5%; If it does not exceed, the calibration passes; If it exceeds, start a pre-set fusion algorithm based on the BP neural network for data compensation. The compensation formula is: ω = αω1+(1 - α)ω2; where ω is the moisture content after compensation; α = σ(ω1, ω2, ΔT), ΔT is the environmental temperature change amount, and σ is the sigmoid activation function.

[0010] Preferably, the optimization calculation in step S3 adopts the multi-objective particle swarm algorithm, and the objective function is: min(E)=k1·Pt + k2·Σ(1 / MTBF i ) + k3·ΔQ; Constraint conditions: C1:Q min ≤Qt≤0.9Q max ; C2:f vib ≤0.8f threshold; where Pt is the real-time power and k1 is the real-time power weight coefficient; MTBF i is the mean time between failures of the device, Σ(1 / MTBF i ) is the reliability score of the device. The larger the score value, the more likely the device is to fail. k2 is the reliability score of the device; ΔQ is the flow fluctuation amount, and k3 is the flow fluctuation coefficient; Qt is the coal flow rate, and f vib is the vibration frequency of the device, and f threshold is the vibration frequency safety threshold of the device.

[0011] Preferably, the data acquisition in step S1 includes the following preprocessing processes: Perform wavelet denoising on the vibration sensor signal; Perform spatial registration on the infrared thermal imaging data; Use the Kalman filter algorithm to fuse multi-source moisture content measurement values; Extract time-frequency domain features from the acoustic emission signal.

[0012] Preferably, the model iteration update in step S4 includes the following steps: Automatically generate a model performance evaluation report every week. The model performance evaluation report includes model accuracy, recall rate, and F1 value; Trigger incremental learning when the prediction error exceeds 2% for three consecutive days. When the prediction error exceeds 2% for three consecutive days, start the incremental learning mechanism, retain the top 10% important neurons in the neural network, and optimize the model to adapt to new data; Conduct cross-plant transfer learning every quarter, and transfer the data of other plants to this model for iterative learning; Establish a version rollback mechanism and save the model weights of the last five versions.

[0013] Preferably, the construction of the time series prediction model based on the attention mechanism for dynamically predicting the operating state of the coal conveying system specifically includes the following steps: S21. Data sorting and matrix construction: Slice the preprocessed multi-source sensing data according to the time window to construct an input matrix Xt∈RN×12, which contains 12-dimensional features, and the time step N takes a value of 60; S22. Network architecture construction: Build an improved LSTM-ATT network architecture, which includes a bidirectional LSTM layer with the number of hidden units set to 64 and a dropout rate of 0.2; a multi-head attention layer with the number of heads being 4 and the query vector dimension being 32; and a fully connected output layer, using LeakyReLU as the activation function; S23. Model parameter initialization: Initialize the model parameters using the transfer learning method. The source domain data used contains 200 groups of historical operating conditions; S24. Loss function definition: Define the loss function L = λ1MAE + λ2DTW, where λ1 takes the value of 0.7 and λ2 takes the value of 0.3; S25. Rolling prediction execution: By performing the rolling prediction operation, output the predicted value Q of the coal flow rate and the equipment health trend H within the next 5 minutes.

[0014] Preferably, the safety protection threshold is dynamically adjusted based on the equipment vibration spectrum analysis results. When the risk of coal blockage is detected, triggering the hierarchical protection mechanism specifically includes the following steps: S51. Spectrum transformation and feature extraction: Perform FFT transformation on the vibration signal and extract the energy distribution characteristics within the frequency band of 1 - 2000Hz; S52. Risk coefficient calculation: Calculate the coal blockage risk coefficient P = Σ(w i ·E i / E total ), where w i is the pre - training frequency band weight of the i - th frequency band within the frequency band of 1 - 2000Hz, set by the management personnel according to the historical training records, E i is the vibration energy of the i - th frequency band within the frequency band of 1 - 2000Hz; E total is the total vibration energy within the frequency band of 1 - 2000Hz; S53. Dynamic adjustment of the protection threshold: Dynamically adjust the protection threshold according to the coal blockage risk coefficient P: When P is less than 0.3, the protection threshold T is set to 1.2T0; when P is in the range of 0.3 (inclusive) to 0.5, T is equal to T0; when P is greater than or equal to 0.5, T is 0.8T0, and the T0 is the pre - set reference protection threshold; S54. Implementation of the hierarchical response strategy: Implement the pre - set hierarchical response strategy based on the coal blockage risk coefficient P; S55. Generation of the fault diagnosis report: Automatically generate a fault diagnosis report after the event, marking the abnormal frequency band characteristics and the disposal effectiveness score.

[0015] Preferably, the hierarchical response strategy includes: First - level response, when P is in the interval of [0.3, 0.5), control the upstream equipment to reduce the speed to 85%Vmax and activate the pre - set high - definition camera for inspection; Second - level response, when P is in the interval of [0.5, 0.7), start the pre - set air cannon coal - clearing device and push the AR maintenance guide to the smart terminal of the maintenance personnel at the same time; Third - level response, when P is greater than or equal to 0.7, perform reverse - order shutdown, the shutdown delay Δt = 1.2L / vcurrent, and trigger an audible and visual alarm.

[0016] In a second aspect, the present application provides a coal conveying control system for a thermal power plant, adopting the following technical solution: A coal conveying control system for a thermal power plant, comprising: A data acquisition module, configured to collect in real time the operation data of the coal conveying system through a pre-set multi-source sensing network, where the operation data includes coal flow parameters, equipment status parameters, and environmental parameters; A dynamic prediction module, configured to construct a time series prediction model based on an attention mechanism to dynamically predict the operation status of the coal conveying system; An edge optimization module, configured to perform real-time optimization calculations at an edge computing node to generate a control instruction set including belt speed, coal feeder opening, and transfer station cooperation strategy; A data sending module, configured to send the control instructions to an actuator through a 5G-MEC network, and at the same time upload key data to the cloud for model iteration and update; A risk regulation module, configured to dynamically adjust the safety protection threshold based on the analysis result of the equipment vibration spectrum, and trigger a hierarchical protection mechanism when a coal blockage risk is detected.

[0017] In summary, the present application includes at least one of the following beneficial technical effects: Through the deep integration of multi-source perception, dynamic optimization, and intelligent decision-making technologies, first, the operation data is collected through a multi-source sensing network, and the prediction model is constructed to perform dynamic prediction relying on comprehensive multi-source data, greatly improving the prediction accuracy; secondly, real-time cooperation with the multi-objective particle swarm optimization algorithm is carried out at the edge computing node for optimization calculations to generate an accurate control instruction set, and efficient communication and model iteration are realized through a 5G-MEC network, achieving accurate and efficient operation regulation and model optimization, and improving the operation efficiency of the coal conveying system; finally, the safety protection threshold is dynamically adjusted based on the analysis of the equipment vibration spectrum, and a hierarchical protection mechanism is triggered when a coal blockage risk is detected, effectively reducing the probability of equipment failure, ensuring the safe operation of the equipment, and achieving the effect of effectively improving the performance of the coal conveying system and green energy conservation; By analyzing the equipment vibration spectrum, the coal blockage risk coefficient can be accurately calculated, the protection threshold can be dynamically adjusted according to the risk coefficient, different risk levels can be reasonably matched, and the safety and adaptability of the coal conveying system under different working conditions can be improved; at the same time, a hierarchical response strategy is implemented according to the risk coefficient to solve the coal blockage risk before the coal blockage occurs, ensuring the normal operation of the coal conveying system, and achieving the effect of effectively improving the operation stability of the coal conveying system; Facing the complex and changeable environment and equipment operation conditions, the moisture content of coal measured by the microwave moisture meter and the laser particle size analyzer is compared in real time. When the measurement deviation exceeds the allowable range, the compensation mechanism is automatically triggered, and the fusion algorithm based on the BP neural network is started. Combining the change amount of the environmental temperature, the compensation coefficient α is calculated using the sigmoid activation function to compensate the moisture content, dynamically adjust the data, without frequent manual intervention, ensure the long-term stability and accuracy of the data, and maintain the efficient and stable operation of the coal conveying control system. Description of the Drawings

[0018] Figure 1 is the flowchart of a coal conveying control method in an embodiment of the present application; Figure 2 is the flowchart of the calibration method of the cross-detection array in an embodiment of the present application; Figure 3 is the flowchart of the method for preprocessing the data acquisition of the multi-source sensing network in an embodiment of the present application; Figure 4 is the flowchart of the method for dynamically predicting the operation state of the coal conveying system in an embodiment of the present application; Figure 5 is the flowchart of the method for iterative updating of the time series prediction model in an embodiment of the present application; Figure 6 is the flowchart of the method for triggering the hierarchical protection mechanism when the risk of coal blockage is detected in an embodiment of the present application; Figure 7 is the system block diagram of a coal conveying control system of a thermal power plant in an embodiment of the present application.

[0019] Description of the reference numerals: 1. Data acquisition module; 2. Dynamic prediction module; 3. Edge optimization module; 4. Data sending module; 5. Risk regulation module. Detailed Description of the Embodiment

[0020] The following will further describe the present application in detail Figures 1-7 with reference to the attached drawings.

[0021] The existing coal conveying system includes main conveying equipment such as coal yard equipment (such as bucket wheel stacker reclaimers, bulldozers, etc.), coal feeders, coal conveying belts, transfer stations, coal dropping pipes, and some auxiliary equipment such as crushers, vibrating screens, impurity removal equipment (such as magnetic separators, wood chip separators), electronic belt scales, and automatic sampling machines.

[0022] An embodiment of the present application discloses a coal conveying control method for a thermal power plant. Referring to Figure 1 , a coal conveying control method for a thermal power plant includes the following steps: S1: Real-time collect the operation data of the coal conveying system through a pre-set multi-source sensing network, and the operation data includes coal flow parameters, equipment status parameters, and environmental parameters; S2: Construct a time series prediction model based on the attention mechanism to dynamically predict the operating status of the coal conveying system; S3: Execute real-time optimization calculations at the edge computing node to generate a control instruction set including belt speed, coal feeder opening, and transfer station coordination strategy; in this example, the edge computing node uses an NVIDIA Jetson AGX Xavier edge controller, which is built-in with: a real-time data cleaning module (sliding window of 60 seconds, step size of 5 seconds); a multi-objective particle swarm optimization algorithm library (population size of 50, number of iterations of 100); a safety assessment engine (vibration spectrum resolution of 0.1 Hz); S4: Send the control instructions to the actuator through the 5G-MEC network, and at the same time upload the key data to the cloud for model iteration and update; the cloud platform can use a self-developed cloud platform or a commercially available industrial cloud platform; in addition, a digital twin system (built based on Unity 3D, geometric error < 2 mm), a model iteration module (supporting the TensorFlow 2.8 framework), and a cross-plant federated learning server (communication protocol MQTT + SSL) are deployed on the cloud platform; S5: Dynamically adjust the safety protection threshold based on the device vibration spectrum analysis results, and trigger a hierarchical protection mechanism when the coal blockage risk is detected. First, collect operation data through a multi-source sensor network, and construct a prediction model to perform dynamic prediction relying on comprehensive multi-source data, which greatly improves the prediction accuracy; second, cooperate with the multi-objective particle swarm algorithm in real time at the edge computing node for optimization calculation, generate accurate control instruction sets, and achieve efficient communication and model iteration through the 5G-MEC network, realizing accurate and efficient operation regulation and model optimization, and improving the operation efficiency of the coal conveying system; finally, dynamically adjust the safety protection threshold based on the device vibration spectrum analysis, and trigger a hierarchical protection mechanism when the coal blockage risk is detected, effectively reducing the probability of equipment failure, ensuring the safe operation of the equipment, and achieving the effect of effectively improving the performance of the coal conveying system and green energy conservation.

[0023] In the above step S1, the multi-source sensor network includes: A cross-detection array composed of a microwave moisture meter (model HYDRA Probe 2.0) and a laser particle size analyzer (MalvernPanalytical Insitec) set on the coal chute; A distributed vibration sensor network arranged along the coal conveying belt; in this embodiment, a three-axis vibration sensor (PCB 356A32, sampling rate of 10 kHz) is arranged every 20 m along the coal conveying belt, and multiple sampling nodes form a vibration network; An infrared thermal imager installed on the transfer station bracket; in this example, an infrared thermal imager (FLIR A700, resolution 640×480) is installed on the transfer station bracket to monitor the temperature of the belt joint, with a sampling interval of 10 seconds; in addition, a temperature and humidity composite sensor (Sensirion SHT45, accuracy ±1.5%RH) is deployed on the top of the coal conveying corridor; An acoustic emission sensor set on the bearing seat of the lower roller of the coal conveying belt. In this embodiment, an acoustic emission sensor (Physical Acoustics Nano30, frequency response range 100kHz - 1MHz) is installed on the roller bearing seat and fixed through a magnetic base. Through the setting of the multi-source sensing network, coal characteristic data can be accurately obtained, and the operating status, operating environment, and equipment health of each device in the coal conveying system can be monitored comprehensively, facilitating the dynamic prediction of the prediction model based on comprehensive multi-source data and helping to improve the prediction accuracy.

[0024] Refer to Figure 2 , the cross-detection array adopts the following calibration method: A1. Obtain the moisture content of coal passing through the coal chute measured by the microwave moisture meter and the laser particle size analyzer in real time; A2. Judge whether the deviation between the moisture content ω1 measured by the microwave moisture meter and the moisture content ω2 deduced by the laser particle size analyzer exceeds 5%; A3. If it does not exceed, the calibration passes; A4. If it exceeds, start the pre-set fusion algorithm based on the BP neural network for data compensation, and the compensation formula is: ω = αω1+(1 - α)ω2; Among them, ω is the moisture content after compensation; α = σ(ω1, ω2, ΔT), ΔT is the change in ambient temperature, and σ is the sigmoid activation function. Facing the complex and changeable environment and equipment operating conditions, compare the moisture content of coal measured by the microwave moisture meter and the laser particle size analyzer in real time. When the measurement deviation exceeds the allowable range, automatically trigger the compensation mechanism, start the fusion algorithm based on the BP neural network, combine the change in ambient temperature, calculate the compensation coefficient α using the sigmoid activation function, compensate the moisture content, dynamically adjust the data, without frequent manual intervention, ensure the long-term stability and accuracy of the data, and maintain the efficient and stable operation of the coal conveying control system.

[0025] Refer to Figure 3 , the data acquisition in step S1 includes the following preprocessing processes: B1. Wavelet denoising: Perform wavelet denoising on the vibration sensor signal; in this example, the db4 wavelet basis function is selected as the wavelet basis function, and the number of layers is 5 layers; B2. Spatial registration: Perform spatial registration on the infrared thermal imaging data; among them, the registration accuracy ≤ 0.5 pixels; B3. Moisture content calibration: The Kalman filter algorithm is used to fuse multi-source moisture content measurement values; B4. Band feature extraction: Time-frequency domain features of the acoustic emission signal are extracted; in this example, the energy ratio of the 1-3 MHz band is emphasized. Through the above preprocessing process, wavelet denoising is used to improve the signal quality, spatial registration with an accuracy of ≤0.5 pixels is performed on the infrared thermal imaging data to ensure the accuracy of the image, the Kalman filter algorithm is used to fuse multi-source moisture content measurement values, and the information of multiple data sources is comprehensively utilized to reduce measurement errors, significantly improve the accuracy of moisture content data; and the band containing rich equipment fault information is emphasized, which helps to accurately identify abnormal conditions inside the equipment, provides a key basis for the early diagnosis and prevention of equipment faults, and reduces the risk of sudden equipment failures.

[0026] Refer to Figure 4 , construct a time series prediction model based on the attention mechanism, and perform dynamic prediction on the operating state of the coal conveying system, which specifically includes the following steps: S21. Data sorting and matrix construction: The preprocessed multi-source sensing data is sliced according to a time window to construct an input matrix Xt∈RN×12, which contains 12-dimensional features, and the time step N takes a value of 60; the 12-dimensional features include coal flow parameters: instantaneous flow rate, moisture content, standard deviation of particle size distribution; equipment parameters: vibration RMS value, bearing temperature, motor current harmonic distortion rate; environmental parameters: corridor temperature, humidity gradient; S22. Network architecture construction: An improved LSTM-ATT network architecture is constructed. This architecture includes a bidirectional LSTM layer with the number of hidden units set to 64 and a dropout rate of 0.2; a multi-head attention layer with the number of heads being 4 and the query vector dimension being 32; and a fully connected output layer, using LeakyReLU as the activation function; S23. Model parameter initialization: The transfer learning method is used to initialize the model parameters, and the source domain data used contains 200 groups of historical operating conditions; S24. Loss function definition: Define the loss function L = λ1MAE + λ2DTW, where λ1 takes a value of 0.7 and λ2 takes a value of 0.3; S25. Rolling prediction execution: By performing rolling prediction operations, the predicted value Q of coal flow and the trend H of equipment health within the next 5 minutes are output. Through the above steps, 12-dimensional feature data in multiple dimensions such as coal flow, equipment, and environment are sliced by time windows to construct a matrix, fully mining the correlations between different types of data and providing rich information for accurate prediction; the improved LSTM-ATT network, with bidirectional LSTM capturing long-term dependencies, multi-head attention focusing on key features, and a fully connected layer combined with the LeakyReLU activation function, enhancing the model's prediction ability for complex operating states; on this basis, transfer learning is used to initialize the parameters, and with the help of 200 sets of historical working condition source domain data, the model training speed is accelerated, the training time cost is reduced, and it can adapt to new data faster; finally, the loss function combines MAE and DTW to more comprehensively and accurately measure the prediction error, improve the prediction accuracy, realize the rolling prediction to output the coal flow and equipment health within the next 5 minutes, provide timely and effective data support for operation regulation, and achieve the effect of effectively improving the model prediction accuracy.

[0027] The optimization calculation in the above step S3 uses the multi-objective particle swarm algorithm, and the objective function is: min(E)=k1·Pt + k2·Σ(1 / MTBFi) + k3·ΔQ; Constraint conditions: C1: Q min ≤ Qt ≤ 0.9Q max ; In this embodiment, Q min = 200t / h, Q max = 800t / h; C2: f vib ≤ 0.8f threshold ; In this embodiment, f threshold = 45Hz; Where Pt is the real-time power, k1 is the real-time power weight coefficient; MTBFi is the average trouble-free time of the equipment, Σ(1 / MTBF_i) is the equipment reliability score, and the larger the score value, the more likely it indicates that the equipment is prone to failure, k2 is the equipment reliability score; ΔQ is the flow fluctuation amount, k3 is the flow fluctuation coefficient; Qt is the coal flow rate, fvib is the vibration frequency of the equipment, and f threshold is the vibration frequency safety threshold of the equipment. In this embodiment, the weight coefficients k1 = 0.6, k2 = 0.3, k3 = 0.1 are determined by the analytic hierarchy process (AHP). Using the multi-objective particle swarm algorithm can quickly respond to the real-time data changes of the coal conveying system, without the need to transmit data to the cloud for processing, greatly shortening the data processing delay, quickly generating a control instruction set, and timely adjusting the belt speed, the opening of the coal feeder, and the collaborative strategy of the transfer station. It can seek a balance among reducing real-time power, improving equipment reliability, and reducing flow fluctuations, ensuring the real-time and efficient operation of the coal conveying system, and achieving the effect of effectively improving the performance of the coal conveying system and reducing energy consumption.

[0028] Referring to Figure 5 , the model iterative update in step S4 includes the following steps: C1. Generate a report weekly: Automatically generate a model performance evaluation report every week. The model performance evaluation report includes model accuracy, recall rate, and F1 value; C2. Monitor errors daily: Trigger incremental learning when the prediction error exceeds 2% for three consecutive days. When the prediction error exceeds 2% for three consecutive days, start the incremental learning mechanism, retain the top 10% important neurons in the neural network, and optimize the model to adapt to new data; C3. Conduct transfer learning quarterly: Conduct cross-plant transfer learning every quarter, and transfer the data of other plants to this model for iterative learning; specifically, feature distillation technology can be used to transfer valuable knowledge from other plants to this model to improve the generalization ability of the model; C4. Establish a version rollback mechanism: Establish a version rollback mechanism and save the model weights of the last five versions. Through the above steps, continuously monitor the time series prediction model, timely grasp the model status, trigger incremental learning when the prediction error exceeds 2% for three consecutive days, retain key neurons to optimize the model, enable it to quickly adapt to new data, improve prediction accuracy, and use feature distillation technology to transfer valuable knowledge from other plants to this model every quarter to improve the generalization ability of the model, achieving the effect of effectively improving the model performance and prediction accuracy.

[0029] Referring to Figure 6 , dynamically adjust the safety protection threshold based on the equipment vibration spectrum analysis results. When the risk of coal blockage is detected, triggering the hierarchical protection mechanism specifically includes the following steps: S51. Spectrum transformation and feature extraction: Perform FFT transformation on the vibration signal and extract the energy distribution characteristics in the frequency band of 1 - 2000Hz; in this embodiment, the frequency band division is: 1 - 100Hz, 100 - 500Hz, 500 - 2000Hz. The pre-trained frequency band weights w1, w2, w3 of the corresponding three frequency bands are obtained through training historical fault data using a logistic regression classifier, w1 = 0.2, w2 = 0.5, w3 = 0.3; S52. Calculate the risk coefficient: Calculate the coal blockage risk coefficient P = Σ(w i ·E i / E total ), where w i is the pre-trained frequency band weight of the i-th frequency band in the 1 - 2000Hz frequency band, set by the management personnel according to the historical training records, and E i is the vibration energy of the i-th frequency band in the 1 - 2000Hz frequency band; E total is the total vibration energy in the 1 - 2000Hz frequency band; S53. Dynamic adjustment of protection threshold: Dynamically adjust the protection threshold according to the coal blockage risk coefficient P. When P is less than 0.3, the protection threshold T is set to 1.2T0; when P is between 0.3 (inclusive) and 0.5, T is equal to T0; when P is greater than or equal to 0.5, T is 0.8T0, where T0 is the pre-set reference protection threshold. S54. Implementation of hierarchical response strategy: Implement the pre-set hierarchical response strategy based on the coal blockage risk coefficient P.

[0030] S55. Generation of fault diagnosis report: Automatically generate a fault diagnosis report after the event, marking the characteristics of the abnormal frequency band and the effectiveness score of the disposal. Through the above steps, by analyzing the vibration spectrum of the equipment, the coal blockage risk coefficient can be accurately calculated, the protection threshold can be dynamically adjusted according to the risk coefficient, different risk levels can be reasonably matched, and the safety and adaptability of the coal conveying system under different working conditions can be improved. At the same time, according to the risk coefficient, implement the hierarchical response strategy, solve the coal blockage risk before the coal blockage occurs, ensure the normal operation of the coal conveying system, and achieve the effect of effectively improving the operation stability of the coal conveying system.

[0031] In addition, automatically generate a fault diagnosis report after the event, marking the abnormal frequency band and the disposal score, which helps to summarize experience and continuously optimize the protection mechanism and fault handling process.

[0032] The above hierarchical response strategy includes: Level 1 response, when P is in the interval [0.3, 0.5), control the upstream equipment to reduce the speed to 85%Vmax and activate the pre-set AI high-definition camera (5 seconds / frame) for inspection; Level 2 response, when P is in the interval [0.5, 0.7), start the pre-set air cannon coal blockage clearing device and push the AR maintenance guide to the smart terminal of the maintenance personnel at the same time; Level 3 response, when P is greater than or equal to 0.7, execute reverse sequence shutdown, the shutdown delay Δt = 1.2L / vcurrent (L is the length of the coal conveying belt, vcurrent is the running speed of the coal conveying belt), and trigger an audible and visual alarm (≥90dB). By implementing the hierarchical response strategy, according to different intervals of the coal blockage risk coefficient P, implement differential measures. When the risk is low, preferentially adopt relatively mild means such as speed reduction and inspection to avoid excessive intervention; when the risk increases, take more radical measures, reasonably allocate resources, reduce unnecessary equipment losses and costs, and cooperate with each link to improve the efficiency and effect of fault response.

[0033] The embodiment of the present application also discloses a coal conveying control system for a thermal power plant. Refer to Figure 7 A coal conveying control system for a thermal power plant includes: A data acquisition module 1, which is used to collect the operation data of the coal conveying system in real time through a pre-set multi-source sensing network. The operation data includes coal flow parameters, equipment status parameters, and environmental parameters. The dynamic prediction module 2 is used to build a time series prediction model based on the attention mechanism to dynamically predict the operating state of the coal conveying system; The edge optimization module 3 is used to perform real-time optimization calculations at the edge computing node to generate a control instruction set including belt speed, coal feeder opening, and transfer station coordination strategy; The data sending module 4 is used to send the control instructions to the actuator through the 5G-MEC network, and at the same time upload the key data to the cloud for model iterative update; The risk regulation module 5 is used to dynamically adjust the safety protection threshold based on the analysis result of the equipment vibration spectrum, and trigger the hierarchical protection mechanism when the coal blockage risk is detected. First, the operation data is collected through the multi-source sensing network, and the prediction model is constructed to perform dynamic prediction relying on comprehensive multi-source data, which greatly improves the prediction accuracy; second, the multi-objective particle swarm algorithm is cooperated in real time at the edge computing node for optimization calculation to generate an accurate control instruction set, and efficient communication and model iteration are realized through the 5G-MEC network to achieve accurate and efficient operation regulation and model optimization, and improve the operation efficiency of the coal conveying system; finally, the safety protection threshold is dynamically adjusted based on the analysis of the equipment vibration spectrum, and the hierarchical protection mechanism is triggered when the coal blockage risk is detected, effectively reducing the probability of equipment failure, ensuring the safe operation of the equipment, and achieving the effect of effectively improving the performance of the coal conveying system and green energy conservation.

[0034] The embodiment of the present application also discloses a computer-readable storage medium, which stores a computer program that can be loaded and executed by a processor as in the above method. The computer-readable storage medium includes, for example: various media such as USB flash drives, external hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0035] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still, without conflict, make combinations, additions, deletions, or other adjustments to the features in the embodiments of the present invention according to the situation without creative efforts, so as to obtain different technical solutions that essentially do not deviate from the concept of the present invention, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. A method for controlling coal transportation in a thermal power plant, characterized in that: The following steps are involved: S1: real-time collection of coal handling system operation data through a pre-set multi-source sensor network, wherein the operation data includes coal flow parameters, equipment status parameters and environmental parameters; S2: Construct a time series prediction model based on the attention mechanism to dynamically predict the operation status of the coal transportation system; S3: Execute real-time optimization calculations at the edge computing node to generate a control instruction set including belt speed, coal feeder opening, and transfer station coordination strategy; S4: Control instructions are sent to the actuators through the 5G-MEC network, and key data are uploaded to the cloud for model iteration and update; S5: Dynamically adjust the safety protection threshold based on the equipment vibration spectrum analysis results, and trigger the hierarchical protection mechanism when the risk of coal blockage is detected.

2. A method for controlling coal transportation in a thermal power plant according to claim 1, characterized in that: The multi-source sensor network in step S1 includes: A cross detection array consisting of a microwave moisture meter and a laser particle size analyzer installed in the coal drop pipe; A distributed vibration sensor network arranged along the coal conveyor belt; Infrared thermal imager installed on the transfer station bracket; An acoustic emission sensor is installed on the bearing seat of the roller under the coal conveyor belt.

3. A method for controlling coal transportation in a thermal power plant according to claim 2, characterized in that: The cross-detection array uses the following calibration method: Real-time acquisition of moisture content of coal passing through the coal drop pipe measured by microwave moisture meter and laser particle size analyzer; Determine whether the deviation between the moisture content ω1 measured by the microwave moisture analyzer and the moisture content ω2 estimated by the laser particle size analyzer exceeds 5%; If it is not exceeded, the calibration is passed; If it exceeds, the preset fusion algorithm based on BP neural network is started to compensate the data. The compensation formula is: ω=αω1+(1-α)ω2; Among them, ω is the moisture content after compensation; α=σ(ω1,ω2,ΔT), ΔT is the change in ambient temperature, and σ is the sigmoid activation function.

4. A method for controlling coal transportation in a thermal power plant according to claim 1, characterized in that: The optimization calculation in step S3 adopts a multi-objective particle swarm algorithm, and the objective function is: min(E)=k1·Pt+k2·Σ(1 / MTBF i )+k3·ΔQ; Constraints: C1:Q min ≤Qt≤0.9Q max ; C2:f vib ≤0.8f threshold ; Where Pt is the real-time power, k1 is the real-time power weight coefficient; MTBF i is the mean time between failures of the equipment, Σ(1 / MTBF i ) is the reliability score of the equipment. The larger the score, the more likely the equipment is to fail. k2 is the reliability score of the equipment. ΔQ is the flow fluctuation, k3 is the flow fluctuation coefficient; Qt is the coal flow, f vib is the vibration frequency of the equipment, f threshold It is the vibration frequency safety threshold of the device.

5. A method for controlling coal transportation in a thermal power plant according to claim 1, characterized in that: The data collection in step S1 includes the following preprocessing process: Perform wavelet noise reduction on vibration sensor signals; Perform spatial registration on infrared thermography data; The Kalman filter algorithm is used to fuse the moisture content measurements from multiple sources; Extract time-frequency domain features of acoustic emission signals.

6. A method for controlling coal transportation in a thermal power plant according to claim 1, characterized in that: The model iterative update in step S4 includes the following steps: Automatically generate a model performance evaluation report every week, which includes model accuracy, recall rate, and F1 value; When the prediction error exceeds 2% for three consecutive days, incremental learning is triggered. When the prediction error exceeds 2% for three consecutive days, the incremental learning mechanism is started, the top 10% important neurons in the neural network are retained, and the model is optimized to adapt to new data; Carry out cross-plant migration learning every quarter, and migrate data from other plants to this model for iterative learning; Establish a version rollback mechanism and save the model weights of the last five versions.

7. A method for controlling coal transportation in a thermal power plant according to claim 5, characterized in that: The construction of a time series prediction model based on the attention mechanism to dynamically predict the operation status of the coal transportation system specifically includes the following steps: S21, data arrangement and matrix construction: the preprocessed multi-source sensor data is sliced ​​according to the time window to construct the input matrix Xt∈RN×12, which contains 12-dimensional features and the time step N is 60; S22. Network architecture construction: Build an improved LSTM-ATT network architecture, which includes a bidirectional LSTM layer with 64 hidden units and a dropout rate of 0.2; a multi-head attention layer with 4 heads and a query vector dimension of 32; and a fully connected output layer with LeakyReLU as the activation function; S23, model parameter initialization: use the transfer learning method to initialize the model parameters, and the source domain data used contains 200 sets of historical operating conditions; S24. Loss function definition: Define the loss function L=λ1MAE+λ2DTW, where λ1 is 0.7 and λ2 is 0.3; S25, rolling forecast execution: by executing the rolling forecast operation, the coal flow forecast value Q and the equipment health trend H within the next 5 minutes are output.

8. A method for controlling coal transportation in a thermal power plant according to claim 1, characterized in that: The method of dynamically adjusting the safety protection threshold based on the equipment vibration spectrum analysis result and triggering the hierarchical protection mechanism when the coal blocking risk is detected specifically includes the following steps: S51, spectrum transformation and feature extraction: perform FFT transformation on the vibration signal to extract the energy distribution characteristics in the frequency band of 1-2000Hz; S52, Risk coefficient calculation: Calculate the coal blocking risk coefficient P = Σ(w i ·E i / E total ), where w i is the pre-trained frequency band weight of the ith frequency band in the 1-2000 Hz frequency band, which is set by the management personnel based on historical training records. i is the vibration energy of the ith frequency band in the 1-2000Hz frequency band; E total It is the total vibration energy in the frequency band of 1-2000Hz; S53, dynamic adjustment of protection threshold: dynamically adjust the protection threshold according to the coal blocking risk coefficient P: when P is less than 0.3, the protection threshold T is set to 1.2T0; when P is between 0.3 (inclusive) and 0.5, T is equal to T0; when P is greater than or equal to 0.5, T is 0.8T0, and T0 is a preset benchmark protection threshold; S54, implementation of hierarchical response strategy: implementing the preset hierarchical response strategy according to the coal blocking risk factor P; S55. Fault diagnosis report generation: Automatically generate a fault diagnosis report after the event, marking the abnormal frequency band characteristics and the treatment effectiveness score.

9. A method for controlling coal transportation in a thermal power plant according to claim 8, characterized in that: The hierarchical response strategy includes: first-level response, when P is in the range of [0.3,0.5), the upstream equipment is controlled to slow down to 85% Vmax, and the pre-set high-definition camera is activated for inspection; second-level response, when P is in the range of [0.5,0.7), the pre-set air cannon clearing device is started, and AR maintenance instructions are pushed to the maintenance personnel's smart terminal; third-level response, when P is greater than or equal to 0.7, reverse shutdown is executed, the shutdown delay Δt=1.2L / vcurrent, and an audible and visual alarm is triggered.

10. A coal handling control system for a thermal power plant, characterized in that: include: A data acquisition module, used to collect the operation data of the coal handling system in real time through a pre-set multi-source sensor network, wherein the operation data includes coal flow parameters, equipment status parameters and environmental parameters; Dynamic prediction module, which is used to build a time series prediction model based on the attention mechanism and dynamically predict the operation status of the coal transportation system; The edge optimization module is used to perform real-time optimization calculations at the edge computing node and generate a control instruction set including belt speed, coal feeder opening, and transfer station coordination strategy; The data transmission module is used to send control instructions to the actuator through the 5G-MEC network, and upload key data to the cloud for model iteration and update; The risk control module is used to dynamically adjust the safety protection threshold based on the equipment vibration spectrum analysis results, and trigger the hierarchical protection mechanism when the risk of coal blockage is detected.

Images