Energy-saving control system of mechanical ventilation dry type cooling tower

Abstract

The invention discloses an energy-saving control system for a mechanical ventilation dry type cooling tower, which belongs to the field of energy-saving control and comprises a sensing module, an edge computing gateway, a frequency converter and a cooling fan. The sensing module continuously collects and outputs cooling tower heat exchange state multi-source operation parameters including environment temperature, water return temperature, water supply temperature and circulating water flow data to the edge computing gateway in real time; based on an embedded adaptive learning model, the edge computing gateway performs combined operation including the cooling temperature difference sum on the multi-source operation parameters according to the input parameters and historical optimization experience, and generates a cooling fan operation frequency instruction corresponding to the current working condition; the frequency converter performs driving adjustment on the cooling fan according to the operation frequency instruction, and a water supply temperature sensor in the sensing module feeds back the adjusted water supply temperature to the edge computing gateway; the cooling tower has the beneficial effects that the water supply temperature of the cooling tower is accurately adjusted, energy is saved, and the abnormal self-protection function is achieved.

Description

Technical Field

[0001] This invention relates to the field of energy-saving technology, and more specifically, to an energy-saving control system for a mechanically ventilated dry cooling tower. Background Technology

[0002] In continuous industrial production processes such as power generation, chemical engineering, and metallurgy, mechanically ventilated dry cooling towers are widely used in circulating water cooling systems to efficiently dissipate heat generated during the process into the environment, ensuring that the circulating water temperature meets the requirements for equipment safety and stable process operation. These cooling towers typically use cooling fans to drive airflow through heat exchange coils, achieving forced convection heat exchange between the air and circulating cooling water. The cooling fan is the core power component of the dry cooling tower, accounting for over 60% of its total energy consumption, making it a key target for system energy-saving optimization. Against the backdrop of increasingly stringent "dual carbon" targets and industrial energy efficiency controls, reducing cooling fan energy consumption while ensuring cooling effect and stable water supply temperature has become a pressing technical problem for the industry.

[0003] The fan control methods of existing mechanically ventilated dry cooling towers have mainly evolved from fixed-frequency control to variable-frequency control. Traditional fixed-frequency control has a simple structure, but the fan operates at its rated condition for a long time, which cannot match the actual cooling load changes, resulting in significant energy waste under low load or low ambient temperature conditions. Subsequent start-stop control based on fixed thresholds or simple variable-frequency control, although reducing energy consumption to some extent, mostly rely on a single parameter (such as ambient temperature or return water temperature) for control, ignoring the coupling relationship between multiple factors such as ambient temperature, return water temperature, target supply water temperature, and cooling temperature difference, making it difficult to accurately reflect the true heat exchange state of the cooling tower.

[0004] Furthermore, existing control strategies mostly employ fixed parameters or linear adjustment methods, which lack adaptability and struggle to cope with rapid changes in operating conditions, seasonal transitions, and the degradation of equipment performance over time. Some systems also rely on remote centralized data processing, resulting in control response lag, further limiting energy-saving effects and operational stability. Therefore, there is an urgent need for an energy-saving control system for mechanically ventilated dry cooling towers that can integrate multi-parameter information, provide timely response, achieve high control precision, and possess energy-saving potential, in order to overcome the shortcomings of existing technologies in terms of energy consumption, control precision, and adaptability to operating conditions. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing an energy-saving control system for mechanically ventilated dry cooling towers. This system achieves precise and dynamic control of the cooling fans by deploying multi-parameter sensing modules, an edge computing gateway, and an adaptive learning model. Specific objectives include: 1. Reducing cooling fan energy consumption and improving the overall energy efficiency of the cooling tower; 2. Achieving precise control under multi-parameter coupling to ensure the water supply temperature remains stable within the required range; 3. Enhancing the system's adaptability to different operating conditions and equipment status changes; 4. Reducing data processing latency and improving control response speed.

[0006] Technical solution: An energy-saving control system for a mechanically ventilated dry cooling tower includes a sensing module, an edge computing gateway, a frequency converter, and a cooling fan. The sensing module, the edge computing gateway, the frequency converter, and the cooling fan form a continuous data acquisition link and frequency control link through industrial Ethernet and fieldbus. The sensing module continuously collects and outputs multi-source operating parameters of the cooling tower heat exchange status, including ambient temperature, return water temperature, supply water temperature and circulating water flow rate, to the edge computing gateway in real time. The edge computing gateway is based on an embedded adaptive learning model. It performs joint calculations on the multi-source operating parameters, including the cooling temperature difference, according to the input parameters and historical optimization experience, and generates a cooling fan operating frequency command corresponding to the current operating condition. The frequency converter drives and adjusts the cooling fan according to the operating frequency command, and the water supply temperature sensor in the sensing module feeds back the adjusted water supply temperature to the edge computing gateway, thus forming a closed-loop control system of parameter acquisition, model calculation, frequency execution and result feedback.

[0007] Preferably, the sensing module includes an ambient temperature sensor, a return water temperature sensor, a supply water temperature sensor, and a flow sensor. The ambient temperature sensor is installed at the air inlet of the cooling tower to collect the temperature of the ambient air entering the cooling tower. The return water temperature sensor is installed in the inlet pipe and located near the inlet of the heat exchange coil to collect the return water temperature before the circulating water enters the heat exchange coil. The water supply temperature sensor is installed in the water outlet pipe and located near the outlet of the heat exchange coil to collect the water supply temperature after heat exchange is completed. The flow sensor is installed in the water inlet pipe to collect real-time flow data of the circulating water.

[0008] Preferably, the ambient temperature sensor has a measurement range of -30℃ to 50℃ and an accuracy of ±0.5℃. The return water temperature sensor has a measurement range of 0℃~100℃ and an accuracy of ±0.2℃. The water supply temperature sensor has a measurement range of 0℃ to 100℃ and an accuracy of ±0.2℃. The measurement accuracy parameter is used as the input confidence interval constraint of the adaptive learning model.

[0009] Preferably, the circulating water flow data collected by the flow sensor is synchronized with the data collected by the return water temperature sensor and the supply water temperature sensor within the edge computing gateway. The cooling temperature difference is calculated based on the synchronized temperature data. The cooling temperature difference is defined as the temperature difference between the return water temperature and the supply water temperature. Cooling temperature difference = return water temperature - supply water temperature.

[0010] Preferably, the ambient temperature sensor, return water temperature sensor, supply water temperature sensor, and flow sensor in the sensing module all adopt a 4-20mA analog signal output method. The analog signal is quantized through the analog-to-digital conversion interface in the edge computing gateway before being used in model calculation.

[0011] Preferably, the edge computing gateway is installed inside the cooling tower control cabinet and uses an industrial-grade embedded processor to perform multi-parameter parallel acquisition, model calculation and frequency command generation operations. The overall processing delay from parameter acquisition to frequency command output is limited to no more than 100ms.

[0012] Preferably, the adaptive learning model embedded in the edge computing gateway includes an input layer, a parameter update layer, and an output layer. The input layer receives five input parameters: ambient temperature, return water temperature, required supply water temperature, cooling temperature difference, and circulating water flow rate. The ambient temperature, the return water temperature, the required supply water temperature, the cooling temperature difference, and the circulating water flow rate are combined into a parameter vector in a fixed order within the model to participate in joint calculations.

[0013] Preferably, the adaptive learning model is trained offline based on historical operating data of the cooling tower to obtain initial model parameters in the initial stage of the system. The historical operating data includes multi-source operating parameters under different operating conditions, cooling fan operating frequency, and corresponding water supply temperature deviation. During system operation, the adaptive learning model collects real-time running data at fixed time intervals and updates the model parameters online using the gradient descent algorithm.

[0014] Preferably, the edge computing gateway outputs a cooling fan operating frequency command to the frequency converter via an RS485 communication interface. The frequency converter uses vector control to drive the cooling fan. The output range of the operating frequency command is limited to 5Hz to 50Hz. Furthermore, when an abnormality is detected in the output of the sensing module or the operating frequency command, the frequency converter switches to the mains frequency operating state.

[0015] Compared with the prior art, the advantages of the present invention are as follows: (1) Compared with fixed frequency control, this system can reduce energy consumption by 30% to 40% by adjusting the fan speed as needed; compared with ordinary variable frequency control, the energy saving rate is further improved by 15% to 20% due to the adoption of multi-parameter coupled adaptive control.

[0016] (2) Through real-time feedback from multiple sensors and an adaptive learning model, the deviation of water supply temperature control can be reduced to ±0.5℃, avoiding "overcooling" or "undercooling" and improving the stability of the process system.

[0017] (3) The model can optimize parameters online and adapt to working conditions such as changes in ambient temperature, fluctuations in circulating water flow and equipment performance degradation, without the need for manual adjustment of control thresholds.

[0018] (4) The edge computing gateway processes data locally with a control delay of ≤100ms. Compared with remote monitoring center control, the response speed is improved by more than 50%, ensuring that the fan speed is adjusted in time when the operating conditions change.

[0019] (5) The system is equipped with a fault switching protection mechanism. When the sensor or gateway fails, it will automatically switch to power frequency operation to avoid the risk of cooling interruption. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall system template of the energy-saving control system for a mechanically ventilated dry cooling tower according to the present invention. Detailed Implementation

[0021] For examples, please refer to Figure 1 An energy-saving control system for a mechanically ventilated dry cooling tower includes a sensing module, an edge computing gateway, a frequency converter, and a cooling fan. The sensing module, edge computing gateway, frequency converter and cooling fan form a continuous data acquisition link and frequency control link through industrial Ethernet and fieldbus; Among them, the sensing module continuously collects and outputs multi-source operating parameters of the cooling tower heat exchange status, including ambient temperature, return water temperature, supply water temperature and circulating water flow data, to the edge computing gateway in real time; The edge computing gateway is based on an embedded adaptive learning model. It performs joint calculations on multi-source operating parameters, including cooling temperature difference, according to input parameters and historical optimization experience, and generates cooling fan operating frequency instructions corresponding to the current operating conditions. The frequency converter drives and adjusts the cooling fan according to the operating frequency command, and the water supply temperature sensor in the sensing module feeds back the adjusted water supply temperature to the edge computing gateway, thus forming a closed control system of parameter acquisition, model calculation, frequency execution and result feedback.

[0022] Specifically, the input layer receives the following operating parameters as model input variables: ambient temperature T1, return water temperature T2, demand water supply temperature T3, cooling temperature difference ΔT, and circulating water flow rate Q; where the demand water supply temperature T3 is the target water supply temperature set by the upper control system or process.

[0023] Specifically, before the system is put into operation, the adaptive learning model is trained offline based on the historical operating data of the cooling tower to obtain the initial model parameters. The historical operating data includes at least T1, T2, T4, Q, ΔT, cooling fan operating frequency and corresponding energy consumption data under different operating conditions. During the system operation, the adaptive learning model collects real-time operating data according to the preset time period and updates the model parameters online through the gradient descent algorithm.

[0024] Specifically, the output layer calculates and outputs the optimal operating frequency f of the cooling fan based on the input parameters T1, T2, T3, ΔT, and Q. The operating frequency f ranges from 5Hz to 50Hz. III. Inverter and Cooling Fan Control (Output Parameter Closed Loop): The edge computing gateway sends the optimal operating frequency f to the inverter via an RS485 communication interface. The inverter uses vector control to adjust the frequency and voltage of the output power supply according to the operating frequency f to drive the cooling fan. When the supply water temperature T4 is higher than the required supply water temperature T3, the adaptive learning model increases the operating frequency f accordingly; when the supply water temperature T4 is lower than the required supply water temperature T3, the adaptive learning model decreases the operating frequency f accordingly. The system has a protection mechanism: when an abnormal sensor output or an abnormal operating frequency f is detected, the inverter switches to mains frequency operation.

[0025] Specifically, the system workflow (with clear and traceable closed-loop parameters) Step S1: Collect ambient temperature T1, return water temperature T2, supply water temperature T4, and circulating water flow rate Q; Step S2: Calculate the cooling temperature difference ΔT and construct the input parameter set {T1, T2, T3, ΔT, Q}; Step S3: The adaptive learning model outputs the operating frequency f of the cooling fan; Step S4: The frequency converter drives the cooling fan to run according to the operating frequency f; Step S5: Feed back the water supply temperature T4 to the edge computing gateway to form a closed-loop regulation.

[0026] The sensing module includes an ambient temperature sensor, a return water temperature sensor, a supply water temperature sensor, and a flow sensor; Among them, the ambient temperature sensor is installed at the air inlet of the cooling tower to collect the temperature of the ambient air entering the cooling tower. The return water temperature sensor is installed on the inlet water pipe and is located near the inlet of the heat exchange coil to collect the return water temperature before the circulating water enters the heat exchange coil. The water supply temperature sensor is installed in the water outlet pipe and is located near the outlet of the heat exchange coil to collect the water supply temperature after heat exchange is completed. The flow sensor is installed in the inlet pipe to collect real-time flow data of the circulating water.

[0027] Ambient temperature sensing measurement range: -30℃ to 50℃, accuracy: ±0.5℃; The return water temperature sensor has a measurement range of 0℃ to 100℃ and an accuracy of ±0.2℃. The water supply temperature sensor has a measurement range of 0℃ to 100℃ and an accuracy of ±0.2℃. The measurement accuracy parameter is used as the input confidence interval constraint for the adaptive learning model.

[0028] The circulating water flow data collected by the flow sensor is synchronized with the data collected by the return water temperature sensor and the supply water temperature sensor within the edge computing gateway. The cooling temperature difference is calculated based on the synchronized temperature data. The cooling temperature difference is defined as the temperature difference between the return water temperature and the supply water temperature. Cooling temperature difference = return water temperature - supply water temperature.

[0029] Specifically, the edge computing gateway performs timestamp alignment processing on the sampled data from the flow sensor, return water temperature sensor, and supply water temperature sensor; The alignment method is as follows: using the edge computing gateway system clock as a reference, the most recent timestamp data collected by each sensor is interpolated or rounded so that ΔT calculation is based on synchronized data; The formula for calculating the cooling temperature difference ΔT is: ΔT = T2 - T4 Through the above time synchronization processing, the calculation error of ΔT is guaranteed to be ≤0.1℃, which can accurately reflect the heat exchange status of circulating water.

[0030] The quantized and synchronized T1, T2, T4, Q, and the calculated ΔT data are used as input parameters for the adaptive learning model. All data is updated in real time, ensuring that the edge computing gateway can generate cooling fan operating frequency instructions based on the latest operating conditions.

[0031] The ambient temperature sensor, return water temperature sensor, supply water temperature sensor, and flow sensor in the sensing module all use 4-20mA analog signal output. The analog signal is quantized through the analog-to-digital conversion interface in the edge computing gateway before being used in model calculation.

[0032] The edge computing gateway is installed inside the cooling tower control cabinet and uses an industrial-grade embedded processor to perform multi-parameter parallel acquisition, model calculation and frequency command generation operations. The overall processing latency from parameter acquisition to frequency command output is limited to no more than 100ms.

[0033] Specifically, the edge computing gateway is installed inside the cooling tower control cabinet and uses an industrial-grade embedded processor, such as an ARM Cortex-A72 quad-core processor with a main frequency of 1.8GHz. It is equipped with 4GB DDR4 memory to support real-time data caching, model calculation and multi-task scheduling; It supports multi-threaded / multi-tasking operating systems (such as real-time operating systems RTOS or LinuxRT), and can simultaneously handle tasks such as data acquisition, model calculation, and execution instruction output.

[0034] Multi-task acquisition-processing-output logic Edge computing gateways process data in parallel according to the following logic: Data acquisition thread: Receives the output signal from the sensor module in real time, with a sampling period of 1Hz, and performs quantization and filtering processing; Model calculation thread: The latest input parameters {T1,T2,T3,ΔT,Q} collected are fed into the adaptive learning model for joint calculation to generate the cooling fan operating frequency f; Output control thread: Sends the calculated frequency f to the inverter via RS485 interface, and simultaneously monitors the communication status and feedback data.

[0035] Multi-threaded parallel processing ensures that tasks do not block each other, improving system response speed and control precision.

[0036] Implementation method for overall processing latency ≤100ms Data acquisition delay: ADC conversion and signal preprocessing time ≤ 10ms; Model computation latency: The adaptive learning model requires ≤50ms to perform one forward computation with input parameters {T1,T2,T3,ΔT,Q}; Communication delay: The time required to send the operating frequency command to the inverter and receive the feedback signal via the RS485 interface is ≤30ms; Taking into account the time of each of the above stages, the overall delay from parameter acquisition to frequency command output is ≤100ms, which meets the requirements of real-time control.

[0037] With its multi-threaded architecture, the edge computing gateway can complete sensor data acquisition, model calculation, and output control within a single sampling period without blocking or waiting. It supports expanding the number of sensors and input parameters, while the system can still maintain an overall processing latency of ≤100ms; Equipped with an industrial-grade embedded processor and ample RAM, the system ensures long-term stable operation in cooling tower environments characterized by high temperature, high humidity, and vibration.

[0038] The adaptive learning model embedded in the edge computing gateway includes an input layer, a parameter update layer, and an output layer. The input layer receives five input parameters: ambient temperature, return water temperature, required supply water temperature, cooling temperature difference, and circulating water flow rate. Ambient temperature, return water temperature, required supply water temperature, cooling temperature difference, and circulating water flow rate are combined into a parameter vector in a fixed order within the model to participate in joint calculations.

[0039] Specifically, the input layer of the adaptive learning model receives five runtime parameters in a fixed order: Ambient temperature T1, return water temperature T2, required supply water temperature T3 (set by the upper control system or process), cooling temperature difference ΔT (ΔT=T2-T4, where T4 is the supply water temperature), and circulating water flow rate Q are all input parameters that are represented in vector form within the model, forming a five-dimensional vector X=[T1,T2,T3,ΔT,Q]ᵀ, and participating in joint calculations. Vectorization can be implemented using matrix operation libraries or SIMD instructions of embedded processors, thereby improving computational efficiency.

[0040] The adaptive learning model is trained offline based on historical operating data of the cooling tower in the initial stage of the system to obtain initial model parameters. Historical operating data includes multi-source operating parameters under different operating conditions, cooling fan operating frequency, and corresponding water supply temperature deviation. During system operation, the adaptive learning model collects real-time running data at fixed time intervals and updates the model parameters online using the gradient descent algorithm.

[0041] Specifically, in the initial stage of the system, offline training is carried out using historical operating data. The data volume includes at least 30 days of historical operating data, covering different loads, ambient temperatures and water supply demand conditions. Historical data includes: multi-source operating parameters {T1,T2,T3,ΔT,Q}, cooling fan operating frequency f, and corresponding water supply temperature deviation δT; The initial parameters and weights W0 of the model are obtained through offline training, which provides a foundation for subsequent online learning.

[0042] During system operation, the adaptive learning model collects real-time operational data at fixed time intervals (once every hour). Online parameter updates are performed using the gradient descent algorithm, with the gradient calculated based on the error between the current predicted output f_pred and the actual water supply temperature deviation δT. The gradient descent step size / learning rate α can be selected based on the fluctuation range of historical data and the system response requirements. For example, α can be set to an adjustable range of 0.01 to 0.1 to ensure convergence and avoid excessive jitter.

[0043] The edge computing gateway outputs the cooling fan operating frequency command to the frequency converter via the RS485 communication interface. The frequency converter uses vector control to drive the cooling fan. The output range of the operating frequency command is limited to 5Hz to 50Hz. Furthermore, when an abnormality is detected in the sensor module output or the operating frequency command, the inverter switches to the mains frequency operating state.

[0044] Specifically, in the edge computing gateway, anomaly detection logic is set up for the sensing module and frequency commands: The acquired temperature or flow signal is outside the measurement range or below the sensor's minimum output value; The sampled value changes by more than a set threshold (temperature ±5℃, flow rate ±10%) for N consecutive (e.g., 3) consecutive times. Signal disconnection or communication timeout (e.g., no data update for more than 200ms in RS485).

[0045] Frequency command error: Output frequency f exceeds the set range (5Hz~50Hz); Frequency change rate exceeds the safety threshold (e.g. ±10Hz / s). A significant difference from historical operating frequencies may lead to system oscillations.

[0046] When any of the above abnormal conditions are detected, the edge computing gateway sends an abnormal signal to the frequency converter. The inverter immediately stops receiving frequency commands from the adaptive learning model and switches to fixed power frequency operation (e.g., 50Hz). At the same time, abnormal events and timestamps are recorded at the gateway to facilitate subsequent diagnosis; This process corresponds to the following closed-loop process steps: S1, S2: When collecting operating parameters and calculating ΔT, if abnormal sensor data is found, the power frequency switching is immediately triggered. S3, S4: Before the adaptive model outputs the frequency, it performs anomaly detection to ensure that no abnormal frequency is sent. S5: Water supply temperature feedback monitoring. If the water supply temperature deviation continues to be abnormal, it can also trigger the power frequency switch.

[0047] Under power frequency operation, the system continuously monitors sensor and frequency command data; When the abnormal conditions are resolved and the following recovery conditions are met, the system can automatically resume adaptive frequency control: All sensor data remain stable for M (e.g. 5) consecutive sampling periods within the effective range; The model's calculated output frequency remains continuously and stably within a safe range. Historical data analysis did not reveal any sudden abnormal fluctuations.

[0048] Once the above conditions are met, the edge computing gateway resends the frequency command to the inverter, resuming the closed-loop regulation of steps S3 to S5.

[0049] The anomaly detection logic and power frequency switching are implemented in a multi-threaded manner within the edge computing gateway to ensure that data acquisition, calculation, output and anomaly handling do not interfere with each other. The RS485 communication interface ensures reliable command transmission, and abnormal triggering and recovery actions can be completed within 100ms, which meets the overall system latency requirements. Combined with the closed-loop process of steps S1 to S5, it can ensure that the system maintains basic cooling function when sensor or model malfunctions occur, thus avoiding impact on industrial processes.

[0050] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An energy-saving control system for a mechanically ventilated dry cooling tower, characterized in that, The energy-saving control system for a mechanically ventilated dry cooling tower includes a sensing module, an edge computing gateway, a frequency converter, and a cooling fan. The sensing module, the edge computing gateway, the frequency converter, and the cooling fan form a continuous data acquisition link and frequency control link through industrial Ethernet and fieldbus. The sensing module continuously collects and outputs multi-source operating parameters of the cooling tower heat exchange status, including ambient temperature, return water temperature, supply water temperature and circulating water flow rate, to the edge computing gateway in real time. The edge computing gateway is based on an embedded adaptive learning model. It performs joint calculations on the multi-source operating parameters, including the cooling temperature difference, according to the input parameters and historical optimization experience, and generates a cooling fan operating frequency command corresponding to the current operating condition. The frequency converter drives and adjusts the cooling fan according to the operating frequency command, and the water supply temperature sensor in the sensing module feeds back the adjusted water supply temperature to the edge computing gateway, thus forming a closed-loop control system of parameter acquisition, model calculation, frequency execution and result feedback.

2. The energy-saving control system for mechanically ventilated dry cooling towers according to claim 1, characterized in that, The sensing module includes an ambient temperature sensor, a return water temperature sensor, a supply water temperature sensor, and a flow sensor. The ambient temperature sensor is installed at the air inlet of the cooling tower to collect the temperature of the ambient air entering the cooling tower. The return water temperature sensor is installed in the inlet pipe and located near the inlet of the heat exchange coil to collect the return water temperature before the circulating water enters the heat exchange coil. The water supply temperature sensor is installed in the water outlet pipe and located near the outlet of the heat exchange coil to collect the water supply temperature after heat exchange is completed. The flow sensor is installed in the water inlet pipe to collect real-time flow data of the circulating water.

3. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 2, characterized in that, The ambient temperature sensor has a measurement range of -30℃ to 50℃ and an accuracy of ±0.5℃. The return water temperature sensor has a measurement range of 0℃~100℃ and an accuracy of ±0.2℃. The water supply temperature sensor has a measurement range of 0℃ to 100℃ and an accuracy of ±0.2℃. The measurement accuracy parameter is used as the input confidence interval constraint of the adaptive learning model.

4. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 2, characterized in that, The circulating water flow data collected by the flow sensor is synchronized with the data collected by the return water temperature sensor and the supply water temperature sensor within the edge computing gateway. The cooling temperature difference is calculated based on the synchronized temperature data. The cooling temperature difference is defined as the temperature difference between the return water temperature and the supply water temperature. Cooling temperature difference = return water temperature - supply water temperature.

5. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 1, characterized in that, The ambient temperature sensor, return water temperature sensor, supply water temperature sensor, and flow sensor in the sensing module all adopt a 4-20mA analog signal output method. The analog signal is quantized through the analog-to-digital conversion interface in the edge computing gateway before being used in model calculation.

6. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 1, characterized in that, The edge computing gateway is installed inside the cooling tower control cabinet and uses an industrial-grade embedded processor to perform multi-parameter parallel acquisition, model calculation and frequency command generation operations. The overall processing delay from parameter acquisition to frequency command output is limited to no more than 100ms.

7. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 6, characterized in that, The adaptive learning model embedded in the edge computing gateway includes an input layer, a parameter update layer, and an output layer. The input layer receives five input parameters: ambient temperature, return water temperature, required supply water temperature, cooling temperature difference, and circulating water flow rate. The ambient temperature, the return water temperature, the required supply water temperature, the cooling temperature difference, and the circulating water flow rate are combined into a parameter vector in a fixed order within the model to participate in joint calculations.

8. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 7, characterized in that, The adaptive learning model is trained offline based on historical operating data of the cooling tower in the initial stage of the system to obtain initial model parameters. The historical operating data includes multi-source operating parameters under different operating conditions, cooling fan operating frequency, and corresponding water supply temperature deviation. During system operation, the adaptive learning model collects real-time running data at fixed time intervals and updates the model parameters online using the gradient descent algorithm.

9. The energy-saving control system for a mechanically ventilated dry cooling tower according to claim 1, characterized in that, The edge computing gateway outputs the cooling fan operating frequency command to the frequency converter via the RS485 communication interface. The frequency converter uses vector control to drive the cooling fan. The output range of the operating frequency command is limited to 5Hz to 50Hz. Furthermore, when an abnormality is detected in the output of the sensing module or the operating frequency command, the frequency converter switches to the mains frequency operating state.

Images