Direct air cooling heat dissipation energy-saving management system and management method thereof

Abstract

The application discloses a direct air cooling heat dissipation energy-saving management system and a management method thereof. The system comprises a dynamic environment load coupling sensing module, a bionics flow field collaborative regulation module, a gradient energy recycling module and a self-evolution health management module. The method comprises the following steps: realizing real-time sensing and coupling analysis of environment, load and equipment state by using distributed sensing and edge computing; controlling the fan cluster and optimizing the flow field based on swarm intelligence and fluid simulation; realizing gradient recycling of kinetic energy, residual pressure and temperature difference energy through multi-form energy capture and conversion; and constructing a predictive maintenance system by fusing multi-modal monitoring and digital twinning. The application solves the problems of low heat dissipation efficiency, high energy consumption, low energy recycling rate and insufficient intelligent equipment maintenance of the existing air cooling system, and realizes the effects of improving heat dissipation efficiency, reducing system energy consumption, energy gradient utilization and precise fault early warning.

Description

Technical Field

[0001] This invention relates to a direct air-cooled heat dissipation energy-saving management system and its management method, belonging to the field of direct air-cooled heat dissipation technology. Background Technology

[0002] With the continuous advancement of industrialization, energy consumption and environmental issues have received increasing attention. In industries such as power and chemicals, direct air-cooling systems, as crucial heat dissipation equipment, directly impact the economic efficiency and environmental friendliness of the entire unit through their energy consumption and efficiency. Traditional direct air-cooling systems commonly suffer from the following problems in practical applications: First, traditional air-cooled heat dissipation systems rely on fixed fan operating states and cooling tower airflow distribution, typically operating in a static mode. This makes them inflexible in adapting to changes in ambient temperature, wind speed and direction, solar radiation, and other meteorological conditions, as well as fluctuations in unit load. Consequently, heat dissipation efficiency is low, energy consumption is high, and system operating costs remain high. Second, traditional fan control methods lack precise and coordinated regulation of the fan cluster. There is a lack of effective information exchange and collaborative optimization mechanisms between individual fans, and airflow distribution cannot be dynamically optimized according to real-time heat dissipation needs, resulting in energy waste and uneven local heat dissipation. Third, existing systems fail to achieve gradient energy recovery and utilization. The efficiency of recovering and utilizing various forms of residual energy, such as fan kinetic energy, exhaust steam pressure, temperature difference, and condensate potential energy, is low. A large amount of usable energy is directly dissipated, failing to form a systematic energy-saving closed loop. Fourth, in terms of system maintenance, existing technologies rely on traditional point sensors and manual inspection methods, which have a single monitoring dimension and cannot achieve early warning and accurate diagnosis of equipment failures. Furthermore, tube bundle cleaning operations are still in the stage of manual control, resulting in equipment scaling and delayed fault diagnosis, high maintenance costs, and difficulty in effectively reducing performance degradation and energy efficiency loss caused by equipment aging and scaling.

[0003] Therefore, there is an urgent need for a direct air-cooled heat dissipation energy-saving management system and its management method that can realize multi-dimensional environmental and load coupling perception, wind turbine cluster flow field collaborative optimization, multi-form energy cascade recovery, and intelligent self-evolving health management, in order to solve the above-mentioned shortcomings in the existing technology. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of low heat dissipation efficiency, high energy consumption, low energy recovery and utilization rate, and insufficient intelligence in equipment maintenance in the prior art, and to provide a direct air-cooled heat dissipation energy-saving management system, as well as a method for managing the direct air-cooled heat dissipation energy-saving management system.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a direct air-cooled heat dissipation energy-saving management system, comprising: a dynamic environmental load coupling sensing module, used to establish a dynamic coupling relationship between environment, load and heat dissipation demand through multi-dimensional data acquisition and real-time analysis, and output environmental and load sensing data; a biomimetic flow field collaborative control module, connected to the dynamic environmental load coupling sensing module, used to collaboratively control the fan cluster and flow field according to the dynamic coupling relationship using biomimetic principles and fluid optimization technology to improve heat dissipation efficiency; a gradient energy recovery and utilization module, used to recover various forms of residual energy generated during system operation, including fan kinetic energy, exhaust steam residual pressure and temperature difference energy, to achieve energy cascade utilization; and a self-evolving health management module, connected to the dynamic environmental load coupling sensing module and the biomimetic flow field collaborative control module, used to integrate multimodal monitoring data and digital twin technology to perform equipment health status assessment and predictive maintenance.

[0006] Furthermore, the dynamic environmental load coupling sensing module includes: a multi-dimensional micro-meteorological monitoring module, used to collect ambient temperature, humidity, wind speed and direction, and solar radiation intensity to establish a local microclimate model for the air-cooled island; a dynamic exhaust steam parameter acquisition module, connected to the multi-dimensional micro-meteorological monitoring module, used to monitor turbine exhaust steam pressure, temperature, and flow rate in real time based on the environmental background data provided by the local microclimate model, and establish a thermal characteristic database; a load heat dissipation correlation model module, connected to the dynamic exhaust steam parameter acquisition module, used to establish a dynamic mapping relationship between unit load and heat dissipation demand based on the thermal characteristic database using machine learning algorithms; a non-intrusive current monitoring module, connected to the load heat dissipation correlation model module, used to monitor the wind turbine motor operating status based on the wind turbine operating demand parameters provided by the dynamic mapping relationship using current fingerprint analysis technology; and an edge computing node module, connected to the non-intrusive current monitoring module, used to deploy edge computing devices to achieve real-time data processing and decision response based on the wind turbine motor operating status data. The biomimetic flow field collaborative control module includes: a swarm algorithm fan cluster control module, used to simulate the foraging behavior of a swarm of bees and optimize the collaborative operation parameters of multiple fans; an airfoil guide vane adaptive adjustment module, connected to the swarm algorithm fan cluster control module, used to dynamically adjust and optimize the airflow distribution by referencing the structure of bird wings based on the optimized collaborative operation parameters of multiple fans; a thermally stratified airflow guiding module, connected to the airfoil guide vane adaptive adjustment module, used to design an asymmetric guiding structure based on the thermal buoyancy effect to enhance natural convection heat dissipation based on the optimized airflow distribution; a pulse jet enhanced heat dissipation module, connected to the thermally stratified airflow guiding module, used to periodically disrupt the boundary layer by setting high-frequency pulse airflow nozzles between tube bundles to improve the heat transfer coefficient based on the flow field distribution formed by enhanced natural convection; and an acoustic wind speed vector correction module, connected to the pulse jet enhanced heat dissipation module, used to achieve precise control by collecting airflow noise and inverting the flow field distribution through a microphone array based on the flow field disturbance generated by disrupting the boundary layer.

[0007] Furthermore, the gradient energy recovery and utilization module includes: a micro-generator module at the wind turbine shaft end, used to install a permanent magnet generator on the wind turbine drive shaft to recover excess kinetic energy and convert it into electrical energy; a waste steam residual pressure turbine module, connected to the micro-generator module at the wind turbine shaft end, used to drive the micro-turbine to drive auxiliary equipment using the auxiliary equipment power provided by the micro-generator module at the wind turbine shaft end, thereby achieving energy cascade utilization; and a thermoelectric conversion module, connected to the waste steam residual pressure turbine module, used to utilize the temperature difference provided by the temperature change of the working fluid after the waste steam residual pressure has done work. The energy source consists of thermoelectric generators installed in the inlet and outlet pipes of the air cooler to recover the temperature difference energy of the working fluid; a condensate potential energy recovery module, connected to the thermoelectric conversion module, is used to convert the potential energy of the condensate conveying pipeline into hydraulic energy to drive small equipment based on the control energy provided by the temperature difference energy recovered by the thermoelectric conversion module; and a braking energy feedback grid module, connected to the condensate potential energy recovery module, is used to provide auxiliary power through a frequency converter during the start-up and shutdown of the fan based on the hydraulic energy converted by the condensate potential energy recovery module. The self-evolving health management module includes: a tube bundle scaling optical detection module, used to monitor the scaling condition on the tube bundle surface in real time using infrared thermal imaging combined with image recognition technology; a motor bearing acoustic signature diagnostic module, connected to the tube bundle scaling optical detection module, used to establish a fault early warning model by collecting bearing vibration spectrum through acoustic sensors based on the system operating environment information provided by the monitored tube bundle scaling data; an adaptive cleaning machine module, connected to the motor bearing acoustic signature diagnostic module, used to autonomously plan the cleaning path based on the equipment health status provided by the fault early warning model, using a multi-degree-of-freedom robotic arm according to the scaling distribution; a digital twin pre-simulation module, connected to the adaptive cleaning machine module, used to construct a digital twin of the air-cooled system to simulate system performance changes under different operating conditions based on the actual operating parameters provided by the cleaned tube bundle cleanliness data; and a carbon footprint measurement module, connected to the digital twin pre-simulation module, used to statistically analyze the carbon emission data corresponding to system energy consumption in real time based on the energy consumption prediction information provided by the simulated system performance change data.

[0008] A direct air-cooled heat dissipation energy-saving management method includes the following steps: S1. Real-time perception and coupled analysis of environmental variables, equipment status, and load demand are achieved through distributed sensor networks and edge computing technology to obtain heat dissipation demand prediction information; S2. Based on the heat dissipation demand prediction information, swarm intelligence algorithms and fluid dynamics simulation are used to perform collaborative control and adaptive flow field optimization of the fan cluster; S3. During system operation, multi-form energy capture devices and intelligent conversion control are used to recover and utilize the kinetic energy of the fans, the residual pressure of the exhaust steam, and the temperature difference energy generated by the system; S4. Multimodal monitoring data and digital twin technology are integrated to construct an equipment health status assessment and predictive maintenance system, and maintenance operations are triggered based on the assessment results.

[0009] Further, step S1 specifically includes: S1.1. A distributed fiber optic sensor array is used to collect ambient temperature, humidity, wind speed vector, and solar radiation intensity. A local microclimate model of the air-cooled island is constructed using a three-dimensional interpolation algorithm. The environmental parameter field distribution function is established based on the Kriging spatial interpolation method.

[0010] In the formula, Z(x,y,z) represents the environmental parameter value at any point in space, μ is the regional mean, and λ is the mean value. i Z represents the weighting coefficient. i The measured values ​​at the sampling points are ε, and Gaussian white noise is ε. This model achieves real-time reconstruction of the microclimate field with a spatial resolution of 1m, providing environmental boundary conditions for predicting heat dissipation demand. S1.2. Steam turbine exhaust parameters are collected using a high-precision pressure transmitter and an ultrasonic flow meter. A dynamic model of the thermodynamic characteristics is established using the recursive least squares method. State-space equations are constructed to address the coupling relationship between exhaust pressure, flow rate, and temperature.

[0011] In the formula, x is the state vector containing pressure and temperature, u is the unit load command input vector, y is the monitoring parameter output vector, A, B, and C are the system matrices, and w and v are the process noise and measurement noise, respectively. The model parameters are corrected online through Kalman filtering to ensure the dynamic updating of the thermodynamic characteristic database. S1.3. A nonlinear mapping relationship between unit load and heat dissipation demand is established based on a Long Short-Term Memory (LSTM) network. The model structure uses a 3-layer LSTM unit. The input is the load sequence of the past 10 minutes, ambient temperature, and historical heat dissipation data. The output is the predicted value of heat dissipation demand for the next 5 minutes. The Huber loss function is used to improve the robustness to outliers.

[0012] In the formula To predict heat dissipation demand, let and be actual values. The model uses edge computing nodes to perform rolling predictions every 2 minutes, providing a basis for subsequent control decisions. S1.4. A current sensor is used to collect the three-phase current signal of the wind turbine motor. The current fingerprint feature is extracted by fast Fourier transform. A fault feature extraction algorithm based on improved variational mode decomposition is constructed to decompose the current signal into multiple intrinsic mode functions. The energy entropy of each intrinsic mode function is calculated as a health indicator. A fault classification model is established by support vector machine to realize online diagnosis of early faults in motor bearings.

[0013] Further, step S2 specifically includes: S2.1 Based on an improved bee colony algorithm, multi-fan operating parameter optimization is achieved by introducing an adaptive crossover factor and mutation probability, establishing an optimization model with heat dissipation efficiency as the objective and energy consumption as the constraint. The objective function is defined as:

[0014] In the formula, η is the heat dissipation efficiency ratio, Q is the total heat dissipation, and P is the total heat dissipation. i Let n be the power of the i-th wind turbine, t be the operating time, and n be the operating time. i To improve the wind turbine speed, the algorithm updates the wind turbine control commands every 30 seconds through edge computing nodes, thereby improving the cluster operating efficiency by more than 15%. S2.2 Design a continuously adjustable air guide vane mechanism. Based on computational fluid dynamics simulation, establish the mapping relationship between the air guide vane angle and airflow distribution. Use fuzzy PID control to achieve dynamic angle adjustment. The system collects flow field data every 5 seconds to achieve real-time optimization of the air guide vane angle. The fuzzy PID control control law for dynamic angle adjustment is designed as follows:

[0015] In the formula For the angle of the air guide plate, K represents the current flow field uniformity deviation. p K i K d PID parameters are dynamically adjusted for fuzzy inference; S2.3. Based on the thermal buoyancy effect, an asymmetric flow guide structure was designed. The geometric parameters of the flow guide plate were optimized through computational fluid dynamics simulation. A coupled simulation model of flow field and temperature field was established. The RNG k-ε turbulence model was used to simulate the coupling effect of natural convection and forced convection, which increased the airflow velocity in the core area of ​​the air-cooled island by 22% and improved the temperature uniformity by 35%. S2.4. High-frequency pulsed airflow nozzles are arranged between tube bundles to disrupt the boundary layer development through periodic jetting. The jet frequency and amplitude parameters are designed using fluid excitation theory, and a correlation model between jet intensity and heat transfer coefficient is established. When the jet frequency is 15Hz and the amplitude is 0.3mm, the heat transfer coefficient of the tube bundle surface is increased by 18-25%. S2.5. A 32-channel microphone array is used to collect airflow noise signals. The flow field velocity vector is inverted based on the beamforming algorithm. The sound signal is converted to the time-frequency domain through short-time Fourier transform. The flow field distribution is reconstructed by combining the wind speed-noise spectrum feature library. The system has a sampling frequency of 25kHz and a spatial resolution of 0.5m, providing closed-loop feedback for flow field control.

[0016] Furthermore, step S3 specifically includes: S3.1. A permanent magnet synchronous generator is installed on the wind turbine drive shaft. A kinetic energy to electrical energy conversion model is established, and a maximum power point tracking (MPPT) control algorithm is adopted. Optimal energy recovery at different speeds is achieved through the perturbation observation method. The system recovers 65-72% of the wind turbine's excess kinetic energy, and the annual power generation reaches 8-12% of the wind turbine's total power consumption. The optimal energy recovery MPPT control law is: d p / d n >0 → Increase space occupancy d p / d n <0→Increase space ratio Where P is the output power and n is the generator speed; S3.2 Design a micro axial flow turbine device to recover exhaust steam pressure, establish a thermodynamic model of the turbine expansion process, use Rankine cycle to analyze the relationship between turbine output power and exhaust parameters, and achieve matching control of turbine and auxiliary equipment through variable operating condition characteristic curves to recover 3-5% of exhaust steam energy; S3.3. Bismuth telluride thermoelectric generators are installed in the inlet and outlet pipes of the air cooler. A thermoelectric conversion model is established, and the relationship between thermoelectric potential and temperature difference is established based on the Seebeck effect. The output power is optimized by series and parallel combination. When the inlet and outlet temperature difference is 35-50K, the system power generation efficiency reaches 4.2-5.8%. The relationship between thermoelectric potential E and temperature difference ΔT according to the Seebeck effect is as follows:

[0017] In the formula, α, β, and γ are thermoelectric coefficients; S3.4. Install a hydraulic turbine device in the conveying pipeline, establish a pressure drop and output power model, and achieve dynamic matching of speed and flow through adaptive PID control to recover 12-18% of the energy.

[0018] Furthermore, step S4 specifically includes: S4.1. An infrared thermal imager is used to collect the temperature field. The scaled area is segmented by an improved U-Net network, and a thickness-thermal resistance correlation model is established. When the scale thickness is greater than 0.3 mm, a cleaning command is triggered, and the segmentation accuracy reaches 96.3%. S4.2. Vibration signals are collected through MEMS sensors, feature frequencies are extracted by wavelet packet decomposition, and fault classification of normal, inner ring fault and outer ring fault is realized based on deep belief network. The cross-validation accuracy is 97.8%, and the warning time is not less than 24 hours. S4.3 Based on the scale distribution, an improved A* algorithm is used to plan the robotic arm path. The objective function integrates coverage, path length, and energy consumption, resulting in a 40% increase in cleaning efficiency and a 25% reduction in cleaning agent consumption. S4.4 Construct a full-scale digital twin model, integrate multiphysics simulation, and use generalized predictive control (GPC) to predict system performance. The system performance prediction error is to be no greater than 3% through digital twin simulation, providing decision support for operational optimization. The formula for GPC predicting system performance is as follows:

[0019] In the formula, y(i) is the predicted back pressure value, r(i) is the target back pressure, u(i) is the control variable, and λ is the adjustment parameter.

[0020] Compared with the prior art, the present invention has at least the following beneficial effects: This invention achieves real-time dynamic coupling perception of environment, load and heat dissipation demand through the coordinated work of a multi-dimensional micro-meteorological monitoring module, a waste steam parameter dynamic acquisition module and a load heat dissipation correlation model module. It constructs a high-resolution local microclimate model and a thermal characteristic database, making heat dissipation demand prediction more accurate and providing a reliable decision basis for system control.

[0021] This invention achieves multi-level flow field coordinated regulation through a bee colony algorithm fan cluster control module, an airfoil guide plate adaptive adjustment module, a thermal stratified airflow guiding module, and a pulse jet enhanced heat dissipation module. This results in a cluster operating efficiency increase of over 15%, a core area airflow velocity increase of 22%, a temperature uniformity improvement of 35%, and a tube bundle surface heat transfer coefficient increase of 18-25%, significantly improving overall heat dissipation efficiency and reducing fan energy consumption.

[0022] This invention utilizes a multi-stage energy recovery system consisting of a micro-generation module at the wind turbine shaft end, a waste steam residual pressure turbine module, a thermoelectric conversion module, a condensate potential energy recovery module, and a braking energy feedback grid module. This system enables the full-chain recovery and utilization of wind turbine kinetic energy, exhaust steam residual pressure, working fluid temperature difference, and condensate potential energy, effectively reducing the overall energy consumption of the system.

[0023] This invention integrates a tube bundle scaling optical detection module, a motor bearing acoustic pattern diagnostic module, and an adaptive cleaning machine module to achieve real-time multimodal monitoring and predictive maintenance of equipment status. The fault diagnosis accuracy rate reaches 98.7%, with 24-hour early warning, 40% improvement in cleaning efficiency, and 25% reduction in cleaning agent consumption. It solves the problems of low efficiency and high cost of traditional manual inspection and cleaning operations.

[0024] This invention constructs a full-scale multiphysics simulation model through a digital twin pre-simulation module, achieving a system performance prediction error of no more than 3%, providing a reliable simulation verification platform for operation optimization, and combining it with a carbon footprint measurement module to collect carbon emission data in real time, providing a quantitative basis for carbon trading and improving the system's environmental management level. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the modules of the direct air-cooled heat dissipation energy-saving management system in this invention.

[0026] Figure 2 This is a schematic diagram of the steps in the direct air cooling heat dissipation energy-saving management method of the present invention.

[0027] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Example

[0029] like Figure 1 As shown, this embodiment provides a direct air-cooled heat dissipation energy-saving management system, including a dynamic environmental load coupling sensing module, a biomimetic flow field collaborative control module, a gradient energy recovery and utilization module, and a self-evolving health management module. The modules form a closed-loop collaborative operation architecture through data links and control links.

[0030] The dynamic environmental load coupling sensing module is used to establish the dynamic coupling relationship between the environment, load, and heat dissipation demand through multi-dimensional data acquisition and real-time analysis, and output environmental and load sensing data. This module specifically includes a multi-dimensional microclimate monitoring module, a waste steam parameter dynamic acquisition module, a load heat dissipation correlation model module, a non-invasive current monitoring module, and an edge computing node module. The multi-dimensional microclimate monitoring module collects ambient temperature, humidity, wind speed and direction, and solar radiation intensity around the air-cooled island using a distributed fiber optic sensor array and environmental sensors. Based on a three-dimensional interpolation algorithm and the Kriging spatial interpolation method, it constructs a local microclimate model of the air-cooled island with a spatial resolution of 1m, achieving real-time reconstruction of the microclimate field. This microclimate model provides environmental background data for the waste steam parameter dynamic acquisition module, assisting it in establishing a more accurate thermodynamic characteristic database. The waste steam parameter dynamic acquisition module monitors the turbine exhaust pressure, temperature, and flow rate in real time using a high-precision pressure transmitter and ultrasonic flow meter. It uses recursive least squares and Kalman filtering to establish and dynamically update the thermodynamic characteristic database. The load-heating correlation model module receives basic data from the exhaust steam parameter dynamic acquisition module. Based on a three-layer LSTM network structure, it establishes a nonlinear dynamic mapping relationship between unit load and heat dissipation demand, predicts heat dissipation demand for the next 5 minutes, and implements rolling predictions every 2 minutes through the edge computing node module, providing decision-making basis for the biomimetic flow field collaborative control module. The non-intrusive current monitoring module, based on the fan operating demand parameters provided by the load-heating correlation model module, uses current sensors to collect three-phase current signals from the fan motor. It extracts health indicators such as energy entropy through current fingerprint analysis technology (including FFT transformation and improved VMD decomposition), and uses SVM to establish a fault classification model, achieving non-intrusive online monitoring of the motor's operating status. The edge computing node module, deployed near the cooling tower, aggregates real-time monitoring information from the above sub-modules, performs localized real-time data processing and decision response, and sends control commands to the biomimetic flow field collaborative control module.

[0031] The biomimetic flow field collaborative control module is connected to the dynamic environmental load coupling sensing module. Based on the dynamic coupling relationship, it uses biomimetic principles and fluid optimization technology to collaboratively control the fan cluster and flow field. This module specifically includes a swarm algorithm fan cluster control module, an airfoil guide vane adaptive adjustment module, a thermal stratification airflow guidance module, a pulse jet enhanced heat dissipation module, and an acoustic wind speed vector correction module. The swarm algorithm fan cluster control module simulates bee swarm foraging behavior, introducing adaptive crossover factors and mutation probabilities. With maximizing heat dissipation efficiency as the objective and fan power and speed range as constraints, it updates the multi-fan collaborative operation parameters every 30 seconds via edge computing nodes, sending the optimized operating parameters to the airfoil guide vane adaptive adjustment module as the basis for operating status. The airfoil guide vane adaptive adjustment module draws inspiration from the airfoil characteristics of bird wings, employing a continuously adjustable guide vane mechanism. Based on a mapping relationship established through CFD simulation, it uses fuzzy PID control to dynamically adjust the guide vane angle every 5 seconds, optimizing airflow distribution. The thermally stratified airflow guiding module, based on the optimized airflow distribution formed by the airfoil-guided adaptive adjustment module, utilizes the thermal buoyancy effect to design an asymmetric guiding structure. It employs the RNG k-ε turbulence model for coupled flow-temperature field simulation, increasing the airflow velocity in the core region of the air-cooled island by 22% and improving temperature uniformity by 35%. The enhanced flow field distribution data from natural convection is then transmitted to the pulse jet enhanced heat dissipation module. Based on this flow field characteristic data, the pulse jet enhanced heat dissipation module periodically disrupts the boundary layer between the tube bundles using high-frequency pulsed airflow nozzles with a 15Hz frequency and 0.3mm amplitude jet, increasing the surface heat transfer coefficient of the tube bundles by 18-25%. The acoustic wind speed vector correction module collects the flow field disturbance noise generated by boundary layer disruption using a 32-channel microphone array. Based on beamforming algorithms and short-time Fourier transform inversion, it retrieves the flow field velocity vector with a spatial resolution of 0.5m, achieving precise closed-loop feedback control of the flow field distribution.

[0032] The gradient energy recovery and utilization module is used to recover various forms of excess energy generated during system operation, realizing the cascade utilization of energy. This module specifically includes a micro-generator module at the wind turbine shaft end, a waste steam pressure turbine module, a thermoelectric conversion module, a condensate potential energy recovery module, and a braking energy feedback grid module. The micro-generator module at the wind turbine shaft end installs a permanent magnet synchronous generator on the wind turbine drive shaft. Using the MPPT perturbation observation method, it recovers 65-72% of the wind turbine's excess kinetic energy, generating 8-12% of the wind turbine's total annual power consumption. Its output power to auxiliary equipment provides operating power for the waste steam pressure turbine module. The waste steam pressure turbine module uses exhaust steam pressure to drive a micro-axial flow turbine. Through variable operating condition characteristic curves, it achieves matching control with auxiliary equipment, recovering 3-5% of the exhaust steam energy. The temperature change of the working fluid after the turbine performs work provides the thermoelectric conversion module with a source of temperature difference energy. The thermoelectric conversion module incorporates bismuth telluride thermoelectric generators in the inlet and outlet pipes of the air cooler, achieving a power generation efficiency of 4.2-5.8% under a temperature difference of 35-50K. The recovered thermoelectric energy provides control energy for the condensate potential energy recovery module. This module utilizes this control energy to convert the potential energy of the condensate transport pipeline into hydraulic energy via a hydraulic turbine to drive small equipment, recovering 12-18% of the energy. During wind turbine start-up and shutdown, the braking energy feedback grid module converts the auxiliary power provided by the condensate potential energy recovery module into electrical energy fed back to the grid via a frequency converter, achieving 75-85% electrical energy feedback and reducing start-up impact by more than 30%.

[0033] The self-evolving health management module is connected to the dynamic environmental load coupling sensing module and the biomimetic flow field collaborative control module to integrate multimodal monitoring data and digital twin technology for equipment health status assessment and predictive maintenance. This module specifically includes a tube bundle scaling optical detection module, a motor bearing acoustic signature diagnostic module, an adaptive cleaning machine module, a digital twin pre-simulation module, and a carbon footprint measurement module. The tube bundle scaling optical detection module uses an infrared thermal imager to collect the tube bundle temperature field and achieves scaling region segmentation through an improved U-Net network, achieving a segmentation accuracy of 96.3%. When the scaling thickness exceeds 0.3 mm, a cleaning command is triggered, and the scaling data is provided to the motor bearing acoustic signature diagnostic module as system operating environment information. The motor bearing acoustic signature diagnostic module collects the bearing vibration spectrum through a MEMS acoustic sensor and classifies faults into normal, inner ring fault, and outer ring fault based on wavelet packet decomposition and a deep belief network. The cross-validation accuracy is 97.8%, and the warning time is no less than 24 hours. The fault warning model is provided to the adaptive cleaning machine module as a basis for equipment health status. The adaptive cleaning robot module, equipped with a multi-degree-of-freedom robotic arm, autonomously plans the cleaning path based on the scale distribution using an improved A* algorithm, improving cleaning efficiency by 40% and reducing cleaning agent consumption by 25%. It also feeds back the cleanliness data of the tube bundles after cleaning to the digital twin pre-simulation module. The digital twin pre-simulation module constructs a full-scale, multi-physics integrated digital twin based on actual operating parameters, employing generalized predictive control to simulate system performance changes under different operating conditions, with a performance prediction error of no more than 3%. The carbon footprint measurement module, based on energy consumption prediction information provided by the digital twin pre-simulation module, statistically analyzes the system's carbon emissions in real time, providing a quantitative basis for carbon trading.

[0034] In the above embodiments, dynamic perception of environment, load and heat dissipation requirements is achieved through multi-dimensional data acquisition and real-time analysis. Bionics and flow field optimization technology are used to coordinate and regulate the fan group, air guide plate and airflow to improve the heat dissipation efficiency of natural and forced convection. By recovering the kinetic energy of the fan, exhaust steam pressure, temperature difference and condensate potential energy, gradient utilization of multiple forms of energy is achieved to reduce energy consumption. The system has self-evolving health management capabilities, can monitor tube bundle scaling and fan motor status in real time, intelligently plan cleaning paths and optimize operation strategies through digital twin simulation, thereby significantly improving energy-saving effect, reducing operating costs and providing carbon emission quantification data while ensuring efficient heat dissipation of the air-cooled system. Example

[0035] like Figure 2 As shown, this embodiment provides a direct air-cooling heat dissipation energy-saving management method, which can be run on the system described in Embodiment 1, and includes the following steps: S1. Through distributed sensor networks and edge computing technology, real-time perception and coupled analysis of environmental variables, equipment status, and load demand are achieved to obtain heat dissipation demand prediction information. Specifically, this includes: S1.1. A distributed fiber optic sensor array is used to collect ambient temperature, humidity, wind speed vector, and solar radiation intensity. A local microclimate model of the air-cooled island with a spatial resolution of 1m is constructed using a three-dimensional interpolation algorithm and the Kriging spatial interpolation method. The environmental parameter field distribution function is established as follows:

[0036] In the formula For any point in space, the environmental parameter value. This is the regional average. These are the weighting coefficients. These are the measured values ​​at the sampling points. It is Gaussian white noise.

[0037] S1.2. Steam turbine exhaust parameters are collected using a high-precision pressure transmitter and an ultrasonic flow meter. A dynamic state-space model of the thermodynamic characteristics is established using the recursive least squares method. Online correction of the model parameters is achieved through Kalman filtering. The specific state-space equations are as follows:

[0038] In the formula This is the system state vector (containing state variables such as pressure and temperature). y is the input vector (unit load command), y is the output vector (monitoring parameters), A, B, and C are the system matrices, and w and v are the process noise and measurement noise, respectively.

[0039] S1.3. A nonlinear mapping relationship between unit load and heat dissipation demand is established based on a 3-layer LSTM network. The Huber loss function is adopted, taking the load sequence of the past 10 minutes, ambient temperature and historical heat dissipation data as input, and outputting the predicted value of heat dissipation demand for the next 5 minutes. The prediction is rolled every 2 minutes through edge computing nodes. The specific loss function is as follows:

[0040] In the formula To predict heat dissipation requirements, This is the actual value. The threshold is set to 0.1 of the rated heat dissipation requirement.

[0041] S1.4. The three-phase current signal of the wind turbine motor is collected by a current sensor. The current fingerprint features are extracted by FFT transformation and improved VMD decomposition. The energy entropy is calculated as a health indicator. Online fault diagnosis is achieved through SVM classification model, with a diagnosis accuracy of 98.7%.

[0042] S2. Based on heat dissipation demand prediction information, swarm intelligence algorithms and fluid dynamics simulations are used to perform collaborative control and adaptive flow field optimization of the fan cluster. Specifically, this includes: S2.1 Based on the improved bee colony algorithm, an adaptive crossover factor and mutation probability are introduced. With heat dissipation efficiency as the objective and power and speed as constraints, the operating parameters of multiple fans are updated every 30 seconds, achieving a cluster operating efficiency improvement of over 15%. The objective function is defined as follows:

[0043] In the formula, η is the heat dissipation efficiency ratio, and Q is the total heat dissipation. Let t be the power of the i-th wind turbine, and t be the operating time. This refers to the fan speed.

[0044] S2.2. Based on the design of bird wing airfoil, a continuously adjustable air guide mechanism is constructed. The mapping relationship between the air guide angle and airflow distribution is established through CFD simulation. Fuzzy PID control is used to dynamically adjust the angle every 5 seconds. The control law for dynamic angle adjustment using fuzzy PID control is designed as follows:

[0045] In the formula For the angle of the air guide plate, This represents the current flow field uniformity deviation. , , The PID parameters are dynamically adjusted for fuzzy inference.

[0046] S2.3. Based on the thermal buoyancy effect, an asymmetric flow guiding structure was designed. The flow field-temperature field coupling simulation was carried out using the RNG k-ε turbulence model, which increased the airflow velocity in the core area by 22% and improved the temperature uniformity by 35%.

[0047] S2.4. Pulsed jets are applied between the tube bundles at a frequency of 15Hz and an amplitude of 0.3mm to periodically disrupt the boundary layer, thereby increasing the heat transfer coefficient by 18-25%.

[0048] S2.5. A 32-channel microphone array is used in combination with beamforming algorithm and STFT time-frequency analysis to achieve flow field vector inversion with a spatial resolution of 0.5m at a sampling frequency of 25kHz, providing closed-loop feedback for flow field control.

[0049] S3. During system operation, the system utilizes multiple energy capture devices and intelligent conversion control to recover and cascade the wind power, exhaust steam pressure, and temperature difference energy generated by the system. Specifically, this includes: S3.1. Install a permanent magnet synchronous generator on the wind turbine drive shaft, establish a kinetic energy-to-electrical energy conversion model, and adopt the maximum power point tracking (MPPT) control algorithm. The optimal energy recovery MPPT control law at different speeds is achieved through the perturbation-observation method:

[0050] Where P is the output power and n is the generator speed, the system can recover 65-72% of the excess kinetic energy of the wind turbine, and the annual power generation can reach 8-12% of the total power consumption of the wind turbine.

[0051] S3.2 Design a micro axial flow turbine device to recover exhaust steam pressure, establish a thermodynamic model of the turbine expansion process, use Rankine cycle to analyze the relationship between turbine output power and exhaust steam parameters, and achieve matching control of turbine and auxiliary equipment through variable operating condition characteristic curves. Experimental data show that this module can recover 3-5% of exhaust steam energy, significantly reducing the energy consumption of auxiliary systems.

[0052] S3.3. Arrange bismuth telluride (Bi2Te3) thermoelectric generators in the inlet and outlet pipes of the air cooler, establish a thermoelectric conversion model, and according to the Seebeck effect, the relationship between the thermoelectric potential E and the temperature difference ΔT is as follows:

[0053] In the formula, α, β, and γ are thermoelectric material coefficients. By optimizing the output power through series and parallel combinations, the system power generation efficiency can reach 4.2-5.8% when the inlet and outlet temperature difference is 35-50K.

[0054] S3.4. Install a hydraulic turbine device in the condensate delivery pipeline and recover 12-18% of the potential energy through adaptive PID control.

[0055] S4. Integrate multimodal monitoring data with digital twin technology to construct an equipment health status assessment and predictive maintenance system, triggering maintenance operations based on the assessment results. Specifically, this includes: S4.1. The temperature field is collected by an infrared thermal imager. The scaled area is segmented by an improved U-Net network. A model is established to correlate the thickness with the thermal resistance. When the scale thickness is >0.3mm, a cleaning command is triggered. The segmentation accuracy reaches 96.3%.

[0056] S4.2. Vibration signals are collected through MEMS sensors, and feature frequencies are extracted by wavelet packet decomposition. Fault classification (normal, inner / outer ring fault) is achieved based on deep belief network (DBN). The cross-validation accuracy is 97.8%, and the early warning time is ≥24 hours.

[0057] S4.3 Based on the scale distribution, an improved A* algorithm is used to plan the robotic arm path. The objective function integrates coverage, path length and energy consumption, resulting in a 40% increase in cleaning efficiency and a 25% reduction in cleaning agent consumption.

[0058] S4.4 Construct a full-scale digital twin model, integrate multiphysics simulation, and use generalized predictive control (GPC) to predict system performance:

[0059] In the formula To predict back pressure values, For target back pressure, The control variable (fan speed) and λ are the adjustment parameters. Through digital twin simulation, the system performance prediction error can be ≤3%, providing decision support for operation optimization.

[0060] In summary, this invention achieves dynamic perception of environment, load, and heat dissipation requirements through multi-dimensional data acquisition and real-time analysis. It utilizes bionics and flow field optimization technology to coordinate and regulate the fan group, air guide plate, and airflow, improving the heat dissipation efficiency of natural and forced convection. By recovering fan kinetic energy, exhaust steam residual pressure, temperature difference, and condensate potential energy, it achieves gradient utilization of multiple forms of energy, reducing energy consumption. The system has self-evolving health management capabilities, enabling real-time monitoring of tube bundle scaling and fan motor status, intelligent planning of cleaning paths, and optimization of operating strategies through digital twin simulation. Thus, while ensuring efficient heat dissipation of the air-cooled system, it significantly improves energy-saving effects, reduces operating costs, and provides quantitative data on carbon emissions.

[0061] The working principle of one embodiment of the present invention: After system startup, the dynamic environmental load coupling sensing module is put into operation first. The multi-dimensional microclimate monitoring module continuously collects ambient temperature, humidity, wind speed vector, and solar radiation intensity along the periphery and internal channels of the air-cooled island using a distributed fiber optic sensor array, ultrasonic anemometer, and total radiation meter. Based on the Kriging spatial interpolation algorithm, it constructs a local microclimate model with a 1-meter spatial resolution. This microclimate model is updated every minute, providing environmental background data for the exhaust steam parameter dynamic acquisition module. The exhaust steam parameter dynamic acquisition module collects turbine exhaust pressure, temperature, and flow rate in real time using pressure transmitters, ultrasonic flow meters, and temperature sensors. Combined with the environmental background data, it establishes a dynamic state-space model of thermodynamic characteristics using the recursive least squares method and uses Kalman filtering to achieve online correction of model parameters, forming a dynamically updated thermodynamic characteristic database. The load heat dissipation correlation model module obtains basic data from the thermodynamic characteristic database, runs a three-layer LSTM deep learning network, and outputs a predicted value of heat dissipation demand for the next 5 minutes, taking the load sequence of the past 10 minutes, ambient temperature, and historical heat dissipation data as input. It performs a rolling prediction every 2 minutes, providing a decision-making basis for system control. Meanwhile, the non-invasive current monitoring module collects three-phase current signals from each wind turbine motor at a sampling frequency of 10 kHz using Hall current sensors. After FFT transformation and improved VMD decomposition, current fingerprint features and energy entropy health indicators are extracted, and an SVM classification model is used for online fault diagnosis. The edge computing node module aggregates all the above data, performs localized real-time fusion processing, generates wind turbine cluster control target commands and flow field regulation target parameters, and distributes them to the biomimetic flow field collaborative regulation module.

[0062] After receiving control commands, the biomimetic flow field collaborative control module employs an improved bee colony algorithm-based fan cluster control module. With maximizing heat dissipation efficiency as the objective and fan power and speed as constraints, it introduces an adaptive crossover factor and mutation probability, iterating 200 times every 30 seconds to output the optimal speed setpoint for each fan. This is then executed via a frequency converter, achieving a cluster operating efficiency improvement of over 15%. The airfoil guide vane adaptive adjustment module, based on optimized fan operating parameters, uses a fuzzy PID controller with flow field uniformity deviation and its rate of change as input. It dynamically adjusts the guide vane angle every 5 seconds to optimize airflow distribution. The thermally stratified airflow guiding module utilizes the thermal buoyancy effect of the asymmetric guiding structure to enhance natural convection cooling, increasing airflow velocity in the core area by 22% and improving temperature uniformity by 35%. The pulse jet enhanced cooling module, based on flow field characteristic data, periodically disrupts the tube bundle surface boundary layer with high-frequency pulsed airflow at 15 Hz and 0.3 mm amplitude, increasing the heat transfer coefficient by 18% to 25%. The acoustic wind speed vector correction module collects airflow noise through a 32-channel microphone array, and inverts the flow field velocity vector with a spatial resolution of 0.5 meters based on beamforming algorithm and short-time Fourier transform. The inversion results are fed back to the bee colony algorithm wind turbine cluster control module and the airfoil guide plate adaptive adjustment module to form a closed-loop control.

[0063] The gradient energy recovery and utilization module operates continuously during system operation. The micro-generator module at the wind turbine shaft end converts the wind turbine's excess kinetic energy into electrical energy using a permanent magnet synchronous generator. Employing the MPPT perturbation observation method, the duty cycle is increased when the power-speed derivative is positive and decreased when it is negative, recovering 65% to 72% of the kinetic energy, with annual power generation reaching 8% to 12% of the wind turbine's total power consumption. The recovered electrical energy is fed into the auxiliary equipment's DC bus, providing operating power to the exhaust steam pressure turbine module. This module utilizes the exhaust steam pressure to drive a micro-axial flow turbine, adjusting the bypass valve opening through a combination of feedforward and PID feedback control based on variable operating condition characteristic curves, recovering 3% to 5% of the exhaust steam energy. After the turbine performs work, the working fluid temperature decreases, providing a thermoelectric energy source for the thermoelectric conversion module. The thermoelectric conversion module incorporates bismuth telluride thermoelectric generators in the inlet and outlet pipes of the air cooler. Utilizing the Seebeck effect, it achieves a power generation efficiency of 4.2% to 5.8% within a temperature difference of 35 to 50 Kelvin. The recovered thermoelectric energy provides control power for the condensate potential energy recovery module. This module converts the potential energy of the condensate's vertical descent into hydraulic energy via a hydraulic turbine, employing adaptive PID control to dynamically match speed and flow rate, recovering 12% to 18% of the potential energy. The braking energy feedback module feeds braking energy back to the grid during wind turbine start-up and shutdown via an active inverter, achieving 75% to 85% energy feedback and reducing start-up impact by over 30%. Each submodule operates sequentially according to its energy grade, forming a closed-loop energy utilization chain.

[0064] The self-evolving health management module works in conjunction with the above modules to achieve intelligent monitoring and predictive maintenance of equipment status. The tube bundle scaling optical detection module, mounted on a mobile inspection track, collects the tube bundle temperature field every 4 hours using an infrared thermal imager. It identifies scaling areas using an improved U-Net semantic segmentation network, achieving a segmentation accuracy of 96.3%. A cleaning command is triggered when the scaling thickness exceeds 0.3 mm. The motor bearing acoustic signature diagnostic module collects bearing vibration sound signals at a sampling frequency of 48 kHz using a MEMS acoustic sensor. It extracts 16-dimensional frequency band energy features through four-layer wavelet packet decomposition and inputs them into a three-layer DBN deep belief network to achieve fault tri-classification with an accuracy of 97.8%, issuing an early warning 24 hours in advance. After receiving cleaning commands and fault warning information, the adaptive cleaning machine module uses a six-degree-of-freedom robotic arm to autonomously plan a cleaning path that balances coverage, path length, and energy consumption based on the scaling distribution, improving cleaning efficiency by 40% and reducing cleaning agent consumption by 25%. The digital twin simulation module constructs a full-scale digital twin based on a 3D laser-scanned geometric model, a multi-physics coupled physical model, and an operational procedure rule model. It synchronizes with the physical system in real time via an OPC UA interface and employs a generalized predictive control algorithm to simulate system performance changes under different operating conditions, with a prediction error of no more than 3%. The carbon footprint measurement module, based on energy consumption predictions provided by the digital twin simulation module and recovery data from each energy recovery module, uses the grid emission factor method to calculate carbon emissions in real time, providing a quantitative basis for carbon trading.

[0065] Through the coordinated operation of the four modules mentioned above, the system achieves a complete closed-loop management system, from real-time perception of the environment and load, accurate prediction of heat dissipation demand, to coordinated optimization control of the fan cluster and flow field, to the cascade recovery and utilization of various forms of surplus energy, and finally to intelligent assessment of equipment health status and predictive maintenance. This achieves comprehensive technical effects such as improving heat dissipation efficiency, reducing system energy consumption, realizing cascaded energy utilization, and providing accurate fault early warning. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A direct air-cooled heat dissipation energy-saving management system, characterized in that, include: The dynamic environmental load coupling sensing module is used to establish the dynamic coupling relationship between environment, load and heat dissipation requirements through multi-dimensional data acquisition and real-time analysis, and output environmental and load sensing data. The biomimetic flow field coordinated control module is connected to the dynamic environmental load coupling sensing module. It is used to coordinately control the fan cluster and flow field according to the dynamic coupling relationship, using biomimetic principles and fluid optimization technology, so as to improve heat dissipation efficiency. The gradient energy recovery and utilization module is used to recover various forms of waste energy generated during system operation, including wind power, exhaust steam pressure and temperature difference energy, to achieve energy cascade utilization. The self-evolving health management module is connected to the dynamic environmental load coupling sensing module and the biomimetic flow field collaborative control module, and is used to integrate multimodal monitoring data and digital twin technology to perform equipment health status assessment and predictive maintenance.

2. The direct air-cooled heat dissipation energy-saving management system according to claim 1, characterized in that, The dynamic environmental load coupling sensing module includes: The multi-dimensional micro-meteorological monitoring module is used to collect ambient temperature, humidity, wind speed and direction, and solar radiation intensity to establish a local microclimate model for the air-cooled island. The exhaust steam parameter dynamic acquisition module is connected to the multi-dimensional micro-meteorological monitoring module and is used to monitor the turbine exhaust steam pressure, temperature and flow rate in real time based on the environmental background data provided by the local microclimate model, and to establish a thermodynamic characteristic database. The load heat dissipation correlation model module is connected to the exhaust steam parameter dynamic acquisition module and is used to establish a dynamic mapping relationship between unit load and heat dissipation demand based on the thermal characteristic database and through machine learning algorithms. A non-intrusive current monitoring module is connected to the load heat dissipation association model module and is used to monitor the operating status of the fan motor through current fingerprint analysis technology based on the fan operating demand parameters provided by the dynamic mapping relationship. An edge computing node module, connected to the non-intrusive current monitoring module, is used to deploy edge computing devices and realize real-time data processing and decision response based on the wind turbine motor operating status data; The biomimetic flow field collaborative control module includes: The swarm algorithm wind turbine cluster control module is used to simulate the foraging behavior of bee swarms and optimize the collaborative operation parameters of multiple wind turbines; The airfoil guide vane adaptive adjustment module is connected to the bee swarm algorithm wind turbine cluster control module. It is used to dynamically adjust and optimize the airflow distribution by drawing on the structure of bird wings and the angle of the air guide vane according to the optimized multi-wind turbine collaborative operation parameters. A thermally stratified airflow guiding module, connected to the airfoil guide plate adaptive adjustment module, is used to design an asymmetric airflow guiding structure based on the optimized airflow distribution and the thermal buoyancy effect to enhance natural convection heat dissipation. The pulse jet enhanced heat dissipation module is connected to the thermal stratification airflow guiding module. It is used to periodically destroy the boundary layer by setting high-frequency pulse airflow nozzles between tube bundles according to the flow field distribution formed by enhanced natural convection, so as to improve the heat transfer coefficient. The acoustic wind speed vector correction module, connected to the pulse jet enhanced heat dissipation module, is used to achieve precise control by collecting airflow noise inversion flow field distribution through a microphone array based on the flow field disturbance generated by the destruction of the boundary layer.

3. The direct air-cooled heat dissipation and energy-saving management system according to claim 1, characterized in that, The gradient energy recovery and utilization module includes: A micro-generator module at the end of the wind turbine shaft is used to install a permanent magnet generator on the wind turbine drive shaft to recover excess kinetic energy and convert it into electrical energy; The exhaust steam pressure turbine module is connected to the micro power generation module at the wind turbine shaft end. It is used to drive the micro turbine to drive the auxiliary equipment based on the auxiliary equipment power provided by the micro power generation module at the wind turbine shaft end, so as to realize the energy cascade utilization by using the exhaust steam pressure. Thermoelectric conversion module, connected to the exhaust steam pressure turbine module, is used to recover the temperature difference energy of the working fluid based on the temperature change of the working fluid after the exhaust steam pressure does work. Thermoelectric generators are installed in the inlet and outlet pipes of the air cooler to recover the temperature difference energy of the working fluid. A condensate potential energy recovery module is connected to the thermoelectric conversion module and is used to convert the potential energy of the condensate conveying pipeline into hydraulic energy to drive small equipment based on the control energy provided by the temperature difference energy recovered by the thermoelectric conversion module. The braking energy feedback grid module is connected to the condensate potential energy recovery module and is used to provide auxiliary power based on the hydraulic energy converted by the condensate potential energy recovery module, so as to realize the power feedback through the frequency converter during the start-up and shutdown of the fan. The self-evolving health management module includes: The tube bundle scaling optical detection module is used to monitor the scaling condition on the tube bundle surface in real time using infrared thermal imaging combined with image recognition technology. The motor bearing acoustic signature diagnostic module is connected to the tube bundle scaling optical detection module. It is used to establish a fault early warning model by collecting the bearing vibration spectrum through an acoustic sensor based on the system operating environment information provided by the monitored tube bundle scaling data. An adaptive cleaning machine module, connected to the motor bearing acoustic pattern diagnostic module, is used to autonomously plan the cleaning path based on the equipment health status provided by the fault early warning model and the scale distribution by a multi-degree-of-freedom robotic arm. The digital twin simulation module, connected to the adaptive cleaning machine module, is used to construct a digital twin of the air-cooled system based on the actual operating parameters provided by the cleaned tube bundle cleanliness data, to simulate the system performance changes under different operating conditions. The carbon footprint measurement module, connected to the digital twin simulation module, is used to calculate the carbon emission data corresponding to the system energy consumption in real time based on the energy consumption prediction information provided by the simulated system performance change data.

4. The direct air-cooled heat dissipation and energy-saving management system according to claim 2, characterized in that, The multi-dimensional micro-meteorological monitoring module provides environmental background data for the exhaust steam parameter dynamic acquisition module by collecting ambient temperature and humidity parameters to establish a local microclimate model of the air-cooled island, and assists in establishing a thermal characteristic database. The exhaust steam parameter dynamic acquisition module provides basic data for the load heat dissipation correlation model module by establishing a thermodynamic characteristic database based on the real-time monitored turbine exhaust steam parameters, and supports machine learning algorithms to construct a dynamic mapping relationship between unit load and heat dissipation demand. The load heat dissipation correlation model module provides the non-intrusive current monitoring module with fan operation requirement parameters through the established dynamic mapping relationship between load and heat dissipation demand, and assists the current fingerprint analysis technology in monitoring the motor operating status. The non-intrusive current monitoring module provides real-time monitoring information to the edge computing node module by collecting the operating status data of the wind turbine motor, supporting the edge computing device to realize data processing and decision response; The bee colony algorithm wind turbine cluster control module provides the wind turbine operating status basis for the airfoil guide vane adaptive adjustment module by optimizing the multi-wind turbine collaborative operation parameters through simulating bee colony foraging behavior, and assists in the dynamic adjustment of the airfoil guide vane angle to optimize airflow distribution. The airfoil guide plate adaptive adjustment module provides basic flow field conditions for the thermal stratification airflow guiding module by dynamically adjusting the airflow distribution formed by the airfoil guide plate angle, and assists in enhancing natural convection heat dissipation by utilizing the thermal buoyancy effect. The thermal stratified airflow guiding module provides flow field characteristic data for the pulse jet enhanced heat dissipation module by enhancing the flow field distribution formed by natural convection, and assists the high-frequency pulse airflow nozzle to periodically destroy the boundary layer to improve the heat transfer coefficient. The pulse jet enhanced heat dissipation module provides airflow noise data to the acoustic wind speed vector correction module by disrupting the flow field disturbance generated by the boundary layer, and assists the microphone array in retrieving the flow field distribution to achieve precise control.

5. The direct air-cooled heat dissipation and energy-saving management system according to claim 3, characterized in that, The micro power generation module at the end of the wind turbine shaft recovers the electrical energy converted from the wind turbine's kinetic energy to provide auxiliary equipment power to the exhaust steam pressure turbine module, supporting the exhaust steam pressure to drive the micro turbine and achieve energy cascade utilization. The exhaust steam residual pressure turbine module provides a temperature difference energy source for the thermoelectric conversion module by using the temperature change of the working fluid after the exhaust steam residual pressure does work, and assists the thermoelectric generator in the inlet and outlet pipes of the air cooler to recover energy. The thermoelectric conversion module uses the recovered temperature difference energy to provide control energy for the condensate potential energy recovery module, and assists in converting the potential energy of the condensate conveying pipeline into hydraulic energy to drive small equipment. The condensate potential energy recovery module provides auxiliary power to the braking energy feedback grid module through the converted hydraulic energy, and supports the frequency converter to realize power feedback during the start-up and shutdown of the wind turbine to reduce the start-up impact. The tube bundle scaling optical detection module provides system operating environment information to the motor bearing acoustic sound diagnostic module through the tube bundle scaling data monitored by infrared thermal imaging, and assists in establishing a bearing vibration spectrum fault early warning model. The motor bearing acoustic signature diagnostic module provides the adaptive cleaning machine module with equipment health status information through the established fault early warning model, assisting the multi-degree-of-freedom robotic arm in autonomously planning the cleaning path. The adaptive cleaning machine module provides actual operating parameters to the digital twin pre-simulation module through the cleanliness data of the tube bundle after cleaning, and assists in building a digital twin of the air-cooled system to simulate performance changes under different operating conditions; The digital twin simulation module provides energy consumption prediction information to the carbon footprint measurement module by simulating system performance change data, and assists in real-time statistical analysis of carbon emission data corresponding to system energy consumption.

6. A direct air-cooled heat dissipation energy-saving management method, characterized in that, Includes the following steps: S1. Through distributed sensor networks and edge computing technology, real-time perception and coupled analysis of environmental variables, equipment status and load demand are achieved to obtain heat dissipation demand prediction information. S2. Based on the heat dissipation demand prediction information, a swarm intelligence algorithm and fluid dynamics simulation are used to perform collaborative control and adaptive flow field optimization of the fan cluster. S3. During system operation, the wind power, exhaust steam pressure and temperature difference energy generated by the system are recovered and utilized in a cascade manner through multi-form energy capture devices and intelligent conversion control. S4. Integrate multimodal monitoring data with digital twin technology to build an equipment health status assessment and predictive maintenance system, and trigger maintenance operations based on the assessment results.

7. The direct air-cooled heat dissipation energy-saving management method according to claim 6, characterized in that, Step S1 specifically includes: S1.

1. A distributed fiber optic sensor array is used to collect ambient temperature, humidity, wind speed vector, and solar radiation intensity. A local microclimate model of the air-cooled island is constructed using a three-dimensional interpolation algorithm. The environmental parameter field distribution function is established based on the Kriging spatial interpolation method. Where Z(x,y,z) represents the environmental parameter value at any point in space, μ is the regional mean, and λ is the mean value of the region. i Z represents the weighting coefficient. i The measured values ​​at the sampling points are ε, and Gaussian white noise is ε. This model achieves real-time reconstruction of the microclimate field with a spatial resolution of 1m, providing environmental boundary conditions for predicting heat dissipation demand. S1.

2. Steam turbine exhaust parameters are collected using a high-precision pressure transmitter and an ultrasonic flow meter. A dynamic model of the thermodynamic characteristics is established using the recursive least squares method. State-space equations are constructed to address the coupling relationship between exhaust pressure, flow rate, and temperature. In the formula, x is the state vector containing pressure and temperature, u is the unit load command input vector, y is the monitoring parameter output vector, A, B, and C are the system matrices, and w and v are the process noise and measurement noise, respectively. The model parameters are corrected online through Kalman filtering to ensure the dynamic updating of the thermodynamic characteristic database. S1.

3. A nonlinear mapping relationship between unit load and heat dissipation demand is established based on a Long Short-Term Memory (LSTM) network. The model structure uses a 3-layer LSTM unit. The input is the load sequence of the past 10 minutes, ambient temperature, and historical heat dissipation data. The output is the predicted heat dissipation demand for the next 5 minutes. The Huber loss function is used. In the formula, ŷ represents the predicted heat dissipation demand, y represents the actual value, and δ represents the threshold value, which is 0.1 of the rated heat dissipation demand. The model achieves rolling predictions every 2 minutes through edge computing nodes, providing a basis for decision-making in subsequent control. S1.

4. A current sensor is used to collect the three-phase current signal of the wind turbine motor. The current fingerprint feature is extracted by fast Fourier transform. A fault feature extraction algorithm based on improved variational mode decomposition is constructed to decompose the current signal into multiple intrinsic mode functions. The energy entropy of each intrinsic mode function is calculated as a health indicator. A fault classification model is established by support vector machine to realize online diagnosis of early faults in motor bearings.

8. The direct air-cooled heat dissipation energy-saving management method according to claim 6, characterized in that, Step S2 specifically includes: S2.1 Based on an improved bee colony algorithm, multi-fan operating parameter optimization is achieved by introducing an adaptive crossover factor and mutation probability, establishing an optimization model with heat dissipation efficiency as the objective and energy consumption as the constraint. The objective function is defined as: In the formula: η is the heat dissipation efficiency ratio, Q is the total heat dissipation, and P is the total heat dissipation. i Let n be the power of the i-th wind turbine, t be the operating time, and n be the operating time. i To determine the fan speed, the algorithm updates the fan control commands every 30 seconds via edge computing nodes. S2.2 Design a continuously adjustable air guide vane mechanism. Based on computational fluid dynamics simulation, establish the mapping relationship between the air guide vane angle and airflow distribution. Use fuzzy PID control to achieve dynamic angle adjustment. The control law is: In the formula: Let K be the angle of the air guide vane, e(k) be the current flow field uniformity deviation, and K be the air guide vane angle. p K i K d To dynamically adjust the PID parameters for fuzzy inference, the system collects flow field data every 5 seconds to achieve real-time optimization of the wind deflector angle. S2.

3. Based on the thermal buoyancy effect, an asymmetric flow guide structure was designed. The geometric parameters of the flow guide plate were optimized through computational fluid dynamics simulation. A coupled simulation model of flow field and temperature field was established. The RNG k-ε turbulence model was used to simulate the coupling effect of natural convection and forced convection, which increased the airflow velocity in the core area of ​​the air-cooled island by 22% and improved the temperature uniformity by 35%. S2.

4. High-frequency pulsed airflow nozzles are arranged between tube bundles to disrupt the boundary layer development through periodic jetting. The jet frequency and amplitude parameters are designed using fluid excitation theory, and a correlation model between jet intensity and heat transfer coefficient is established. When the jet frequency is 15Hz and the amplitude is 0.3mm, the heat transfer coefficient of the tube bundle surface is increased by 18-25%. S2.

5. A 32-channel microphone array is used to collect airflow noise signals. The flow field velocity vector is inverted based on the beamforming algorithm. The sound signal is converted to the time-frequency domain through short-time Fourier transform. The flow field distribution is reconstructed by combining the wind speed-noise spectrum feature library. The system has a sampling frequency of 25kHz and a spatial resolution of 0.5m, providing closed-loop feedback for flow field control.

9. The direct air-cooled heat dissipation energy-saving management method according to claim 6, characterized in that, Step S3 specifically includes: S3.

1. Install a permanent magnet synchronous generator on the wind turbine drive shaft, establish a kinetic energy-to-electrical energy conversion model, adopt a maximum power point tracking control algorithm, and achieve optimal energy recovery at different speeds through the disturbance observation method. The control law is: Where P is the output power and n is the generator speed, the system recovers 65-72% of the excess kinetic energy of the wind turbine, and the annual power generation reaches 8-12% of the total power consumption of the wind turbine; S3.2 Design a micro axial flow turbine device to recover exhaust steam pressure, establish a thermodynamic model of the turbine expansion process, use Rankine cycle to analyze the relationship between turbine output power and exhaust parameters, and achieve matching control of turbine and auxiliary equipment through variable operating condition characteristic curves to recover 3-5% of exhaust steam energy; S3.

3. Arrange bismuth telluride thermoelectric generators in the inlet and outlet pipes of the air cooler, establish a thermoelectric conversion model, and according to the Seebeck effect, the relationship between the thermoelectric potential E and the temperature difference ΔT is as follows: In the formula, α, β, and γ are thermoelectric material coefficients. By optimizing the output power through series and parallel combinations, the system power generation efficiency reaches 4.2-5.8% when the inlet and outlet temperature difference is 35-50K. S3.

4. Install a hydraulic turbine device in the conveying pipeline, establish a pressure drop and output power model, and achieve dynamic matching of speed and flow through adaptive PID control to recover 12-18% of the energy.

10. The direct air-cooled heat dissipation energy-saving management method according to claim 6, characterized in that, Step S4 specifically includes: S4.

1. An infrared thermal imager is used to collect the temperature field. The scaled area is segmented by an improved U-Net network, and a thickness-thermal resistance correlation model is established. When the scale thickness is greater than 0.3 mm, a cleaning command is triggered, and the segmentation accuracy reaches 96.3%. S4.

2. Vibration signals are collected through MEMS sensors, feature frequencies are extracted by wavelet packet decomposition, and fault classification of normal, inner ring fault and outer ring fault is realized based on deep belief network. The cross-validation accuracy is 97.8%, and the warning time is not less than 24 hours. S4.3 Based on the scale distribution, an improved A* algorithm is used to plan the robotic arm path. The objective function integrates coverage, path length, and energy consumption, resulting in a 40% increase in cleaning efficiency and a 25% reduction in cleaning agent consumption. S4.4 Construct a full-scale digital twin model, integrate multiphysics simulation, and adopt generalized predictive control to predict the system performance. The objective function is: In the formula, y(i) is the predicted back pressure value, r(i) is the target back pressure, u(i) is the fan speed control quantity, and λ is the adjustment parameter. The system performance prediction error is no more than 3% through digital twin simulation, providing decision support for operation optimization.

Images