Steam turbine, regenerative system and control method of regenerative system

Abstract

The invention provides a steam turbine, a regenerative system and a control method thereof, and the method comprises the steps: collecting data through a multi-mode sensor such as a steam pressure sensor, and constructing a digital twin model containing a thermodynamic equation and equipment aging parameters through the fusion of a space-time alignment algorithm; model deviation is predicted and dynamically calibrated by means of an LSTM network, a near-end strategy optimization algorithm is adopted to solve a Pareto optimal solution of heat recovery efficiency maximization and corrosion rate minimization in a digital twin environment, and a control instruction is generated. And when the corrosion prediction risk exceeds a threshold value, a control strategy is dynamically adjusted, and health-efficiency cooperative control is realized. Meanwhile, the invention further relates to a corresponding collaborative optimization device and corrosion monitoring equipment, and a multi-modal data fusion module is deployed at an edge computing node to realize efficient data processing and analysis. The thermodynamic cycle efficiency of the steam turbine regenerative system is effectively improved, the corrosion risk is reduced, the service life of equipment is prolonged, and remarkable economic and technical benefits are achieved.

Description

Technical Field

[0001] The present invention relates to the field of steam turbines, and in particular to a steam turbine, a heat recovery system and a control method thereof. Background Art

[0002] As the core link of thermal power generation, the efficiency and reliability of the steam turbine heat recovery system directly affect the energy utilization rate and equipment life. The prior art (CN215860369U) discloses a steam turbine heat recovery system (such as Figure 2 and Figure 3 ), which heats steam in a boiler to drive a steam turbine to generate electricity, and the steam is transported to a condensing device through a gas push fan and a one-way valve in the gas pipeline to recover heat. The system has the following defects: 1. Insufficient efficiency optimization: Relying on fixed mechanical structures (such as fan speed and one-way valve opening), it cannot adapt to dynamic working conditions, resulting in low heat recovery efficiency; 2. Lack of health management: There is no corrosion monitoring. Conventional technical means can only rely on a single sensor or manual inspection, and it is impossible to predict pipeline corrosion risks in real time.

[0003] 3. Lack of efficiency optimization: There is no efficiency optimization. Conventional technical means can only rely on separate digital twin modeling to achieve efficiency optimization, and it cannot be coordinated with corrosion monitoring.

[0004] The existing technology does not involve technical means such as digital twin modeling, multimodal sensor fusion or reinforcement learning control, and cannot meet the intelligent needs under complex working conditions. The present invention solves the above problems by introducing digital twin dynamic optimization and multimodal sensing technology, achieving the dual goals of improving thermal cycle efficiency and extending equipment life. Summary of the invention

[0005] The embodiments of the present invention provide a steam turbine, a heat recovery system and a control method thereof, which aims to address the problems that current technologies involving digital twin modeling, multimodal sensor fusion or reinforcement learning control cannot meet the intelligent requirements under complex working conditions.

[0006] The core technology of this invention is mainly to achieve coordinated control of improving the thermal cycle efficiency of the turbine heat recovery system and predicting equipment corrosion through digital twin dynamic optimization and multi-modal sensor data fusion.

[0007] In a first aspect, the present invention provides a method for controlling a steam turbine heat recovery system, the method comprising the following steps: S00. Build a digital twin model: Integrate thermodynamic equations with equipment aging parameters to establish a virtual model of steam flow, condensation heat transfer and corrosion evolution. Equipment aging parameters include sponge pad porosity attenuation coefficient and pipeline roughness growth model. S10, Multimodal data acquisition: Real-time acquisition of steam pressure, condensate flow, pipeline inner wall image, acoustic signal and water quality data through synchronously deployed sensor arrays; S20, data fusion and dynamic calibration: Use the spatiotemporal alignment algorithm to align the timestamps of multi-source data, predict the model deviation through the LSTM network and update the parameters of the digital twin model; S30, Reinforcement Learning Optimization: Run a multi-objective reinforcement learning agent in a digital twin environment to solve the Pareto optimal solution of maximizing heat recovery efficiency and minimizing corrosion rate, and generate control instructions; S40, Health-efficiency collaborative control: When the predicted corrosion risk exceeds the threshold, the control strategy is dynamically adjusted, and the digital twin model is optimized through virtual-reality interactive closed-loop feedback.

[0008] Furthermore, in step S10, the sensors used for multimodal data collection include: Steam pressure sensor, used to monitor the steam pressure in the gas pipeline in real time; A condensate flow meter, used to measure the condensate flow rate of the liquid in the condensate output pipeline; Fan speed encoder, used to detect the speed of the gas push fan; High temperature resistant endoscope camera, used to obtain images of the inner wall of curved gas flow pipes; Wideband acoustic sensor, used to collect acoustic signals in the frequency range of 10Hz-200kHz in the gas activity chamber; Water quality sensor, used to detect the Cl content of the liquid in the condensate output pipeline - Concentration and pH.

[0009] Furthermore, in step S00, the thermodynamic equations of the digital twin model include the continuity equation, the momentum equation and the energy equation, and the equipment aging parameters are obtained by fitting the historical operation data.

[0010] Furthermore, in step S20, the data fusion step includes: The time series of high-frequency acoustic signals are reconstructed using cubic spline interpolation; The attention mechanism is used to weight the fusion of corrosion area ratio, acoustic energy peak and Cl - Concentration characteristics.

[0011] Furthermore, in step S30, the reinforcement learning agent adopts the proximal strategy optimization algorithm, and the reward function R is: R=0.7×efficiency improvement rate-0.3×corrosion rate+0.1×(1-control action frequency).

[0012] Furthermore, in step S10, the high temperature resistant endoscope probe is integrated with a micro camera and a MEMS microphone, and can rotate 360° to scan the inner wall of the curved gas circulation pipe.

[0013] Furthermore, in step S40, the health-efficiency control adjusts the control strategy when any of the following conditions is triggered: Corrosion rate ≥ 0.1mm / year; The virtual model predicted that the thermal resistance of a certain section of pipe would increase by 15% over the baseline value.

[0014] Furthermore, in step S10, multimodal data acquisition uses IEEE1588v2 protocol to achieve multi-source data timestamp alignment, and the synchronization accuracy is ≤10μs.

[0015] In a second aspect, the present invention provides a steam turbine heat recovery system, including a control end, and the control end is equipped with the above-mentioned steam turbine heat recovery system control method.

[0016] In a third aspect, the present invention provides a steam turbine, comprising the above-mentioned steam turbine heat recovery system.

[0017] The main contributions and innovations of the present invention are as follows: 1. Significant efficiency improvement Dynamic adaptive adjustment: Existing technologies rely on fixed mechanical structures and are difficult to adapt to changes in working conditions. The present invention uses multimodal sensors to collect data such as steam pressure and condensate flow in real time, and dynamically adjusts the speed of the gas push fan and the frequency of the condensate pump through a digital twin model and reinforcement learning algorithm, significantly improving the heat recovery efficiency. For example, when the steam pressure changes suddenly, the system parameters can be quickly optimized to maintain efficient operation.

[0018] Precise thermal cycle control: The digital twin model integrates thermodynamic equations and equipment aging parameters to accurately simulate steam flow and condensation heat transfer processes, optimize gas residence time and heat exchange efficiency, and significantly improve energy utilization compared to existing technologies.

[0019] 2. Health management is real-time and accurate Multimodal fusion prediction: The existing corrosion monitoring methods are single and lagging. The present invention uses a multimodal sensor array composed of an endoscopic camera, an acoustic sensor and a water quality sensor to collect pipeline inner wall images, acoustic signals and water quality data in real time. Through advanced image processing and machine learning algorithms, the corrosion risk can be accurately predicted in advance, and the false alarm rate of corrosion is significantly reduced.

[0020] Proactive maintenance decision-making: Once the corrosion rate is predicted to exceed the safety threshold, the system automatically triggers the health-efficiency collaborative optimizer to proactively adjust the control strategy, such as reducing the flow rate of high-risk pipelines to avoid further damage to equipment and reduce maintenance costs.

[0021] 3. Strong system synergy Cross-level collaboration mechanism: In the prior art, efficiency optimization and health management are independent of each other. The present invention realizes the deep collaboration of digital twin optimization and corrosion prediction modules, shares sensor data, and extends the life of equipment while ensuring heat recovery efficiency. For example, when it is predicted that the corrosion risk of a certain curved pipe section increases, the system reduces the steam flow rate in that area while increasing the load in other areas to compensate for the efficiency loss, achieving a dynamic balance between the dual goals of "efficiency-life".

[0022] 4. High prediction and control accuracy Digital twin dynamic calibration: The LSTM network is used to predict model deviations in real time and update digital twin model parameters, allowing the virtual model to more accurately reflect the state of the physical system. The prediction error of the digital twin model is ≤2.5%, which is much lower than traditional methods, providing a reliable basis for optimized control.

[0023] Reinforcement learning multi-objective optimization: The proximal strategy optimization algorithm is used to solve the Pareto optimal solution, and the heat recovery efficiency and corrosion rate are comprehensively considered. The generated control instructions are more scientific and reasonable, improving the overall performance of the system.

[0024] The details of one or more embodiments of the invention are set forth in the following drawings and description so that other features, objects, and advantages of the invention are more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 is a flow chart of a method for controlling a steam turbine heat recovery system according to an embodiment of the present invention; Figure 2 It is a structural diagram of the prior art; Figure 3 It is a structural diagram of a condensing device of the prior art; In the figure, 1. boiler; 2. turbine body; 3. generator; 4. gas transmission pipeline; 5. gas push fan; 6. one-way valve; 7. condensing device body; 71. condensing device end cover; 72. condensation protection outer cylinder; 73. gas activity chamber; 74. middle gas condensation pipeline; 75. curved gas circulation pipeline; 76. sponge pad; 8. condensate liquid delivery pipeline; 9. condensate liquid output pipeline; 10. gas circulation pipeline; 11. condensate device; 12. deaerator; 13. clean water tank; 14. water pump; 15. condensate tank. DETAILED DESCRIPTION

[0026] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of this specification. Instead, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0027] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

[0028] like Figure 2-3 As shown, the prior art (CN215860369U) discloses a steam turbine heat recovery system, the core structure of which includes: the boiler 1 is heated to generate steam to drive the steam turbine body 2 and the generator 3, and the steam is delivered to the condensing device body 7 through the gas push fan 5 and the one-way valve 6 in the gas pipeline 4. The condensed liquid is recovered to the condensate tank 15 through the condensate liquid delivery pipeline 8 and the condensate liquid output pipeline 9, and the remaining gas enters the condensate device 11, the deaerator 12 and the water pump 14 through the gas circulation pipeline 10 for recycling. The system realizes the basic heat recovery function through mechanical structures such as the one-way valve to prevent backflow and the sponge pad 76 to assist condensation. The specific structure can be found in the content disclosed in CN215860369U, which will not be repeated here.

[0029] The existing technology relies on fixed mechanical structures such as fixed speed of the gas push fan and preset opening of the one-way valve, which cannot adapt to dynamic working conditions such as sudden changes in steam pressure and fluctuations in condensate flow. For example, when the turbine load changes and causes steam flow fluctuations, the gas push fan 5 of the gas pipeline 4 cannot dynamically adjust the speed, resulting in a decrease in condensation efficiency. The measured heat recovery efficiency is only 65%-70%. In addition, the curved gas circulation pipeline 75 and the middle gas condensation pipeline 74 of the condensing device body 7 do not have an integrated flow regulation mechanism, making it difficult to optimize the gas residence time and heat exchange efficiency. And there is also a lack of water quality monitoring and other means.

[0030] Based on this, the present invention improves on the existing technology and introduces digital twin dynamic optimization and multimodal perception technology to solve the problems existing in the existing technology.

[0031] Embodiment 1 The present invention aims to propose a control method for a steam turbine heat recovery system, introduce digital twin dynamic optimization and multimodal sensing technology, and solve the core problems of the prior art: The digital twin model maps the state of the physical system in real time and predicts the deviation of thermodynamic parameters (such as heat transfer coefficient decay); A multimodal sensor array (including endoscopic cameras, acoustic sensors, and water quality sensors) simultaneously collects efficiency and health data; The reinforcement learning agent solves the Pareto optimal solution of efficiency-corrosion rate in the digital twin environment and generates collaborative control instructions to achieve the dual goals of "efficiency improvement" and "life extension".

[0032] Specifically, the embodiment of the present invention provides a method for controlling a steam turbine heat recovery system. Specifically, referring to Figure 1 , the method comprising: S00. Build a digital twin model: Integrate thermodynamic equations with equipment aging parameters to establish a virtual model of steam flow, condensation heat transfer and corrosion evolution. Equipment aging parameters include sponge pad porosity attenuation coefficient and pipeline roughness growth model. In this embodiment, the digital twin model includes: 1. Digital Twin Layer: 1.1 Virtual model construction: 1.1.1 Thermodynamic model: A three-dimensional unsteady flow and heat transfer model is established based on ANSYS Fluent, including the following equations: Continuity equation: ; The continuity equation is the embodiment of the law of conservation of mass in fluid mechanics. Among them, ρ represents the density of the fluid, and u is the velocity vector of the fluid. is the divergence operator, It represents the divergence of the product of density and velocity. This equation shows that in a stable flow fluid system, the mass of fluid flowing into a control volume per unit time is equal to the mass of fluid flowing out of the control volume, that is, the mass of the fluid will not be created or disappeared in the system.

[0033] Momentum equation: ; The momentum equation is the application of Newton's second law in fluid mechanics. The left side of the equation represents the convective rate of change of fluid momentum, where ρuu is a second-order tensor representing momentum flux. The meanings of the terms on the right side of the equation are as follows: : Pressure gradient force, p is the pressure of the fluid, and the negative sign indicates that the direction of the pressure gradient force is opposite to the direction of increasing pressure, that is, the fluid flows from the high-pressure area to the low-pressure area.

[0034] : Viscous force, μ is the dynamic viscosity of the fluid, which describes the stress generated inside the fluid due to viscosity.

[0035] ρg: gravity, g is the gravitational acceleration vector.

[0036] This equation describes that the change in fluid momentum is the result of the combined effects of pressure gradient force, viscosity force and gravity.

[0037] Energy equation: ; The energy equation is the expression of the law of conservation of energy in fluid mechanics. The left side of the equation represents the convective transfer rate of fluid enthalpy, where h is the specific enthalpy of the fluid. The terms on the right side of the equation have the following meanings: : Heat conduction term, k is the thermal conductivity of the fluid, T is the temperature of the fluid, this term describes the heat conduction due to temperature gradient.

[0038] S_h: Heat source term, which represents other heat sources or heat sinks in the system, such as heat generated by chemical reactions.

[0039] The equation describes the change in fluid energy as a result of the combined effects of enthalpy convection transfer, heat conduction, and other heat sources.

[0040] Embedded device aging parameters: Sponge pad 76 porosity decay model: ε(t)=ε0×exp(-0.01t) (t is the number of operating hours); this formula describes the decay of the porosity of the sponge pad over time. Among them, ε(t) is the porosity of the sponge pad at time t, ε0 is the initial porosity, and t is the number of operating hours. The exponential function exp(-0.01t) means that the porosity decays exponentially with time, and the decay coefficient is 0.01.

[0041] Pipeline inner wall roughness growth model: k s (t) = k s0 +0.002t (mm). This formula describes the growth of the pipe inner wall roughness over time. s (t) is the roughness of the inner wall of the pipe at time t, k s0 is the initial roughness, and t is the number of operating hours. The linear function 0.002t means that the roughness increases linearly with time, and the growth rate is 0.002 mm / hour.

[0042] 1.1.2 Corrosion prediction model: The improved MaskR-CNN model (ResNet50 backbone network, Focal Loss to handle sample imbalance) is used to segment the corroded areas in the endoscopic images; for example: MaskR-CNN improvements: Backbone network: ResNet50 (pre-trained on COCO dataset); Loss function: Focal Loss (α=0.25, γ=2) + Dice Loss; Input image size: 640×480, corrosion area marking accuracy ±0.1mm; The acoustic signal is transformed by wavelet packet to extract the energy characteristics of the 150-200kHz frequency band, and then input into the SVM classifier (RBF kernel function) to determine the corrosion type (pitting corrosion, uniform corrosion). For example: SVM kernel function: RBF, C=10, γ=0.1; Feature extraction: energy proportion of 150-200kHz frequency band, decomposed by 5 layers of wavelet packets.

[0043] 1.2 Dynamic calibration module: LSTM network structure: input layer (10 sensor features) → bidirectional LSTM layer (128 neurons) → fully connected layer (output model bias correction coefficient); Training data: historical operating data of a power plant (3 years, 100,000 samples in total).

[0044] 2. Decision-making level: 2.1 Reinforcement Learning Control Agent: The proximal strategy optimization (PPO) algorithm is used, and the state space includes 15 parameters such as steam pressure, condensate flow rate, and corrosion area ratio; Action space: gas push fan 5 speed (0-100%), condensate pump 14 frequency (0-50Hz); Reward function: R = 0.7 × efficiency improvement rate - 0.3 × corrosion rate + 0.1 × (1-control action frequency), where efficiency improvement rate = (current efficiency - benchmark efficiency) / benchmark efficiency, and the corrosion rate unit is mm / year. 0.7 × efficiency improvement rate represents the reward for improving heat recovery efficiency, with a weight of 0.7. The higher the efficiency improvement rate, the greater the reward. -0.3 × corrosion rate represents the penalty for corrosion rate, with a weight of -0.3. The higher the corrosion rate, the greater the penalty. 0.1 × (1-control action frequency) represents the reward for control action frequency, with a weight of 0.1. The lower the control action frequency, the greater the reward, which encourages the agent to minimize unnecessary control actions.

[0045] 2.2 Health-Efficiency Co-Optimizer: Trigger conditions: Corrosion rate ≥ 0.1mm / year; The digital twin model predicted that the thermal resistance of a certain section of the pipeline increased by 15% over the baseline value; Control strategy: Reduce the steam flow rate of the corresponding pipeline by 10%-20%, and increase the load in other areas to compensate for efficiency losses.

[0046] S10, Multimodal data acquisition: Real-time acquisition of steam pressure, condensate flow, pipeline inner wall image, acoustic signal and water quality data through synchronously deployed sensor arrays; In this embodiment, the multimodal sensor array: Efficiency Optimization Sensor Group: Steam pressure sensor (such as model: Rosemount 3051S, accuracy ±0.075%FS, can be installed at the upstream flange of the gas pipeline 4, and collect pressure values ​​every 500ms); Condensate flow meter (such as model: Cologne Promag53, range 0-50m 3 / h, can be installed at the elbow of the condensate output pipeline 9, and adopts the electromagnetic induction principle to measure the flow); Gas push fan 5 speed encoder (such as an incremental encoder (OMRON E6B2-CWZ6C), with a resolution of 1024 lines / rev, integrated into the gas push fan 5 motor shaft through a coupling).

[0047] Health Monitoring Sensor Group: Pipeline endoscope camera (such as customized high-temperature resistant fiber optic probe, model: LumaScope360, 5 million pixels, installed at the entrance of the curved gas circulation pipeline 75); or 316L stainless steel material, temperature resistant to 300°C, built-in micro stepper motor (step angle 1.8°) to drive 360° rotation scanning; integrated MEMS microphone (model: KnowlesSPM0408UD5H) and LED lighting module (wavelength 450nm, light intensity 5000lux).

[0048] Wideband acoustic sensor (model: PCB377B10, frequency response 10Hz-200kHz, can be pasted on the inner wall of the gas activity chamber 73 by high temperature resistant epoxy resin, sampling rate 200kHz); Water quality sensor (model: HachLDO101, detects Cl - Concentration (range 0-1000ppm, accuracy ±2%) and pH value (range 0-14, accuracy ±0.01), installed at the outlet of condensate output pipe 9).

[0049] The above sensors are for reference only and can be selected or customized according to actual conditions. The installation method of the sensors can be installed according to the needs or the method required by the instructions. This is a conventional technical means. Therefore, there is no limitation on the model and installation method of the sensors.

[0050] Preferably, multimodal data acquisition uses IEEE 1588v2 (PTP) protocol to achieve multi-source data timestamp alignment, with a synchronization accuracy of ≤10μs; data acquisition frequency: 1kHz for pressure / flow sensor, 1Hz for endoscopic image, and 200kHz for acoustic signal.

[0051] S20, data fusion and dynamic calibration: Use the spatiotemporal alignment algorithm to align the timestamps of multi-source data, predict the model deviation through the LSTM network and update the parameters of the digital twin model; In this embodiment, the specific method of data fusion and dynamic calibration is as follows: 1. Multimodal data fusion method: 1.1 Spatiotemporal alignment algorithm: The acoustic signal (200kHz) was timestamped to match the time scale of the endoscopic image (1Hz) and the sensor data (1kHz); Cubic spline interpolation is used to reconstruct the time series of high-frequency signals. For example, the acoustic signal (200kHz) is downsampled to 1kHz through cubic spline interpolation. The endoscope image (1Hz) and the sensor data (1kHz) are matched by timestamps, and linear interpolation is used to fill in missing values.

[0052] 1.2 Feature fusion method: The corrosion area ratio (image segmentation result), acoustic energy peak (wavelet packet transform), and Cl⁻ concentration (water quality sensor) are weightedly fused through the attention mechanism:

[0053] The weights w_i are learned through training data. This formula is used for feature fusion in multimodal data fusion. i is the i-th feature, such as corrosion area ratio, acoustic energy peak, Cl - concentration, etc., where i takes values ​​from 1 to 3.

[0054] α i is the weight of the i-th feature, calculated through the attention mechanism. i is a learnable weight parameter, exp(w i ) is an exponential function, the denominator It is the sum of the exponential weights of all features, which is used for normalization so that the sum of all weights is 1. The final fusion feature fusion_feature is the weighted sum of each feature, and the weight reflects the importance of each feature in the fusion.

[0055] 2. Virtual-reality interactive closed-loop mechanism: After the physical system executes the control instructions, real-time feedback data (such as efficiency changes and new corrosion point coordinates) is used to update the digital twin model through the API interface; Model iteration cycle: LSTM prediction and parameter correction are performed every 5 minutes.

[0056] Among them, the parameters of the LSTM model are as follows: Input layer: 10 sensor features (pressure, flow, speed, etc.); Bidirectional LSTM layer: 128 neurons, dropout rate 0.2; Output layer: model deviation correction factor (such as heat transfer coefficient correction factor); Training data: 3 years of historical operating data of a power plant (100,000 samples), MSE loss function, Adam optimizer (lr=0.001).

[0057] S30, Reinforcement Learning Optimization: Run a multi-objective reinforcement learning agent in a digital twin environment to solve the Pareto optimal solution of maximizing heat recovery efficiency and minimizing corrosion rate, and generate control instructions; In this embodiment, for example, the parameters optimized by reinforcement learning are as follows: 1.PPO algorithm parameters: Discount factor γ=0.99, gae_lambda=0.95; Batch size 2048, update step 10; Network structure: 2 fully connected layers (64→32 neurons), ReLU activation function; 2. Reward function:

[0058] Among them, η t is the current heat recovery efficiency, η0 is the benchmark efficiency (65%), r t is the corrosion rate (mm / year), and Δa is the number of control actions within 10 minutes.

[0059] S40, Health-efficiency collaborative control: When the predicted corrosion risk exceeds the threshold, dynamically adjust the control strategy (such as reducing the flow rate of high-risk pipelines), and optimize the digital twin model through virtual-reality interactive closed-loop feedback.

[0060] In this embodiment, the execution mechanism is as follows: Gas push fan 5: Model ABB ACS880-01-072A-3, variable frequency speed regulation range 0-1500rpm; Condensate pump 14: Grundfos CRN5-10 centrifugal pump, frequency adjustment range 0-50Hz, controlled by frequency converter (Siemens MM440).

[0061] In this embodiment, data fusion and dynamic calibration are implemented through edge computing nodes, which play the following core roles in the present invention: 1. Multimodal data preprocessing and fusion Process high-frequency data in real time: The integrated GPU-accelerated image processing unit and acoustic signal preprocessing unit can process the 5-megapixel image (1Hz) of the endoscope camera and the 200kHz signal of the broadband acoustic sensor in real time, reconstruct the time series through cubic spline interpolation, and achieve spatiotemporal alignment of multi-source data.

[0062] Feature extraction and dimensionality reduction: An improved MaskR-CNN model is deployed to segment the corrosion area and extract the area ratio features. Wavelet packet transform is used to extract the energy features of the 150-200kHz frequency band from the acoustic signal to reduce data redundancy.

[0063] 2. Model reasoning and decision support Localized intelligent analysis: It is equipped with an LSTM network to predict digital twin model deviations (such as heat transfer coefficient correction) and run an SVM classifier to determine the corrosion type (pitting corrosion / uniform corrosion) without relying on cloud servers, reducing latency.

[0064] Reinforcement Learning Policy Generation: The Proximal Policy Optimization (PPO) algorithm is executed to solve the Pareto optimal solution of efficiency-corrosion rate in the digital twin environment and generate control instructions (such as fan speed and pump frequency) with a response time of ≤200ms.

[0065] 3. Edge intelligence and cloud collaboration Reduce the load on the cloud: More than 90% of real-time data processing (such as corrosion prediction and control instruction generation) is completed on the edge side, and only key decision results (such as corrosion risk level and control parameters) are uploaded to the cloud through the 5G module (Huawei ME909s-821), reducing network bandwidth pressure.

[0066] Offline emergency processing: When the network is interrupted, the system continuously executes health monitoring and control strategies based on the locally stored 1TB SSD historical data and pre-trained models to ensure system reliability.

[0067] 4. Hardware integration and environment adaptation Industrial-grade deployment capabilities: It uses NVIDIA Jetson AGX Xavier edge computing nodes, supports a wide operating temperature range of -40°C to 70°C, and meets the harsh working conditions such as high temperature and vibration of the turbine heat recovery system.

[0068] Modular expansion interface: Expand AI accelerator cards (such as NVIDIA T4) through the PCIe interface to support future model iteration upgrades without changing the hardware platform.

[0069] 5. Virtual-Real Interaction Closed-Loop Core Real-time feedback correction: The actual effect data after the physical system executes the control instructions (such as efficiency changes, new corrosion point coordinates) is updated to the digital twin model in real time through the API interface, driving the iteration of model parameters (once every 5 minutes).

[0070] 6. Health-efficiency synergy trigger: When the predicted corrosion risk exceeds the threshold, the control signal is directly output to the actuator (such as a regulating valve) through the GPIO interface, achieving millisecond-level response without relying on a central controller.

[0071] Preferably, the edge computing node can be configured as follows: Hardware configuration: NVIDIA Jetson AGX Xavier, 32GB DDR4 memory, 1TB SSD storage, equipped with 5G communication module (Huawei ME909s-821); Software environment: Ubuntu 18.04 operating system, CUDA 11.4, cuDNN 8.2, TensorRT 8.0; Integrated modules: GPU-accelerated image processing unit (based on the TensorRT-optimized Mask R-CNN inference engine); Acoustic signal preprocessing unit (wavelet packet transform library PyWavelets).

[0072] In order to verify the experimental effect of the present invention, the following test scenarios are provided: A 100MW steam turbine heat recovery system in a power plant, with a test period of 6 months; Comparison schemes: existing technology (CN215860369U), single digital twin optimization (without corrosion prediction), and single corrosion prediction (without efficiency optimization).

[0073] The experimental results are shown in the following table:

[0074] Among them, statistical significance: the t-test was used for verification, and the efficiency improvement result was p<0.01, which was statistically significant.

[0075] Improved heat recovery efficiency (η): The difference between the heat recovery efficiency of the optimized system and the traditional system. Formula:

[0076] (Q is the heat recovery amount, unit: kJ / s) Corrosion false positive rate (FPR): The proportion of normal samples that are misjudged as corrosion by the model to the total normal samples. Formula:

[0077] Maintenance cost reduction (C): cost savings due to reduced maintenance times under the optimization strategy, formula:

[0078] Digital twin prediction error (ε): the root mean square error (RMSE) between the model prediction value and the true value, formula:

[0079] Technical term explanation SIMPLE algorithm: A semi-implicit method for solving the pressure-coupled equations (used for solving thermodynamic equations of digital twin models), which decouples velocity and pressure through the pressure correction equation.

[0080] Dice Loss: A loss function used for image segmentation that measures the overlap between the predicted area and the true area.

[0081] Embodiment 2 Based on the same concept, the present invention also proposes a steam turbine heat recovery system, including a control end, and the control end is equipped with a steam turbine heat recovery system control method of embodiment 1.

[0082] In this embodiment, the structure of the steam turbine heat recovery system in the prior art is not completely improved, and only sensors and edge computing nodes and their corresponding devices are added.

[0083] Embodiment 3 This embodiment also provides a steam turbine, comprising the steam turbine heat recovery system of the above-mentioned embodiment 2.

[0084] In general, various embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects of the invention may be implemented in hardware, while other aspects may be implemented in firmware or software executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flow charts, or using some other graphical representation, it should be understood that, as non-limiting examples, the boxes, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0085] Embodiments of the present invention can be implemented by computer software, which is executable by a data processor of a mobile device, such as in a processor entity, or implemented by hardware, or implemented by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets and / or macros can be stored in any device readable data storage medium, and they include program instructions for performing specific tasks. Computer program products can include one or more computer executable components configured to perform embodiments when the program is running. One or more computer executable components can be at least one software code or a part thereof. In addition, at this point, it should be noted that any box of the logic flow in the figure can represent a program step, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. Software can be stored in physical media such as memory chips or storage blocks implemented in processors, magnetic media such as hard disks or floppy disks, and optical media such as, for example, DVDs and data variants thereof, CDs. Physical media are non-transient media.

[0086] Those skilled in the art should understand that the technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A steam turbine heat recovery system control method, characterized in that: The following steps are involved: S00. Build a digital twin model: integrate thermodynamic equations with equipment aging parameters to establish a virtual model of steam flow, condensation heat transfer and corrosion evolution. The equipment aging parameters include the sponge pad porosity attenuation coefficient and the pipeline roughness growth model. S10, Multimodal data acquisition: Real-time acquisition of steam pressure, condensate flow, pipeline inner wall image, acoustic signal and water quality data through synchronously deployed sensor arrays; S20, data fusion and dynamic calibration: Use the spatiotemporal alignment algorithm to align the timestamps of multi-source data, predict the model deviation through the LSTM network and update the parameters of the digital twin model; S30, Reinforcement Learning Optimization: Run a multi-objective reinforcement learning agent in a digital twin environment to solve the Pareto optimal solution of maximizing heat recovery efficiency and minimizing corrosion rate, and generate control instructions; S40, Health-efficiency collaborative control: When the predicted corrosion risk exceeds the threshold, the control strategy is dynamically adjusted, and the digital twin model is optimized through virtual-reality interactive closed-loop feedback.

2. A steam turbine heat recovery system control method according to claim 1, characterized in that: In step S10, the sensors used for the multimodal data collection include: Steam pressure sensor, used to monitor the steam pressure in the gas pipeline in real time; A condensate flow meter, used to measure the condensate flow rate of the liquid in the condensate output pipeline; Fan speed encoder, used to detect the speed of the gas push fan; High temperature resistant endoscope camera, used to obtain images of the inner wall of curved gas flow pipes; Wideband acoustic sensor, used to collect acoustic signals in the frequency range of 10Hz-200kHz in the gas activity chamber; Water quality sensor, used to detect the Cl content of the liquid in the condensate output pipeline - Concentration and pH.

3. A steam turbine heat recovery system control method according to claim 1, characterized in that: In step S00, the thermodynamic equations of the digital twin model include the continuity equation, the momentum equation and the energy equation, and the equipment aging parameters are obtained by fitting the historical operation data.

4. A steam turbine heat recovery system control method according to claim 1, characterized in that: In step S20, the data fusion step includes: The time series of high-frequency acoustic signals are reconstructed using cubic spline interpolation; The attention mechanism is used to weight the fusion of corrosion area ratio, acoustic energy peak and Cl - Concentration characteristics.

5. A steam turbine heat recovery system control method according to claim 1, characterized in that: In step S30, the reinforcement learning agent adopts a proximal strategy optimization algorithm, and the reward function R is: R=0.7×efficiency improvement rate-0.3×corrosion rate+0.1×(1-control action frequency).

6. A steam turbine heat recovery system control method according to claim 2, characterized in that: In step S10, the high temperature resistant endoscope probe is integrated with a micro camera and a MEMS microphone, and can rotate 360 ​​degrees to scan the inner wall of the curved gas circulation pipeline.

7. A steam turbine heat recovery system control method according to claim 1, characterized in that: In step S40, the health-efficiency control adjusts the control strategy when any of the following conditions is triggered: Corrosion rate ≥ 0.1mm / year; The virtual model predicted that the thermal resistance of a certain section of pipe would increase by 15% over the baseline value.

8. A steam turbine heat recovery system control method according to any one of claims 1 to 7, characterized in that: In step S10, multimodal data acquisition uses the IEEE1588v2 protocol to achieve multi-source data timestamp alignment, with a synchronization accuracy of ≤10μs.

9. A steam turbine heat recovery system, comprising a control end, characterized in that: The control end is equipped with a steam turbine heat recovery system control method as described in any one of claims 1-8.

10. A steam turbine, characterized in that: It includes a steam turbine heat recovery system as described in claim 9.

Images