Cascading type water-cooling heat dissipation control method and system and storage medium

Abstract

The invention provides a cascaded water-cooling heat dissipation control method and system and a storage medium, and the method comprises the steps: collecting a current temperature in real time, and calculating a current temperature change rate based on the current temperature and a historical temperature at a previous moment; inputting the current temperature and the current temperature change rate into a preset thermal dynamic model to predict a system temperature trend at a future moment; a feedforward comparison result of the system temperature trend and a preset temperature threshold value is determined, and a corresponding feedforward control instruction is generated; determining a feedback comparison result based on the current temperature, a preset starting temperature sequence and a preset stopping temperature sequence, and generating a corresponding feedback control instruction; the feedforward control instruction and the feedback control instruction are fused to generate the fan operation number; the accumulative running time of all the fans is obtained, and target fans to be started and stopped are determined based on the running number of the fans and the accumulative running time; and on the basis of the state of the target draught fan and the preset single-time minimum continuous operation duration, starting and stopping operation of the draught fan is executed. According to the invention, heat dissipation control can be accurately realized with low energy consumption.

Description

Technical Field

[0001] This application relates to the field of heat dissipation technology, and in particular to a cascaded water cooling heat dissipation control method, system and storage medium. Background Technology

[0002] In modern industrial production and daily life, the use of various electronic and mechanical devices is becoming increasingly widespread. These devices generate a large amount of heat during operation, and if this heat cannot be dissipated in a timely and effective manner, it will affect the normal operation of the equipment and may even lead to damage. Therefore, heat dissipation technology plays a crucial role in modern technology. Among them, water cooling systems are an effective heat dissipation method. They use the flow of water to carry away the heat generated by the equipment, thereby achieving the effect of heat dissipation.

[0003] Existing water cooling systems typically employ a fixed operating mode, meaning that all cooling fans operate at a constant speed throughout the entire process, regardless of changes in ambient temperature or equipment operating status. While this operating mode is simple and easy to implement, it often leads to energy waste in practical applications. Furthermore, the continuous operation of all fans can affect their lifespan and fails to achieve efficient and accurate heat dissipation control. Summary of the Invention

[0004] In order to achieve low-energy and accurate heat dissipation control, embodiments of this application provide a cascaded water-cooling heat dissipation control method, system, and storage medium.

[0005] In a first aspect, this embodiment provides a cascaded water-cooling heat dissipation control method, the method comprising: The current temperature of the water cooling system is collected in real time, and the rate of change of the current temperature at the current moment is calculated based on the current temperature and the historical temperature at the previous moment. The current temperature and the rate of change of the current temperature are input into a preset thermal dynamics model to predict the system temperature trend at future times. Determine the feedforward comparison result between the system temperature trend and the preset temperature threshold, and generate the corresponding feedforward control command; Based on the current temperature and the preset start temperature sequence and preset stop temperature sequence, the feedback comparison result is determined, and the corresponding feedback control command is generated; The number of operating fans is generated by integrating the feedforward control commands and feedback control commands. Obtain the cumulative operating time of each wind turbine, and determine the target wind turbines to be started or stopped based on the number of wind turbines in operation and the cumulative operating time; The start-up and shutdown operations of the wind turbine are performed based on the state of the target wind turbine and the preset minimum continuous running time.

[0006] In some embodiments, the real-time acquisition of the current temperature of the water-cooling system further includes: Obtain system initialization parameters, which include a preset start-up temperature sequence, a preset stop temperature sequence, a preset minimum continuous runtime, and a preset thermal dynamic model.

[0007] In some embodiments, the real-time acquisition of the current temperature of the water-cooling system includes: The topology and historical heat load distribution data of the water cooling system are obtained, and at least one temperature measuring node is determined based on the topology and historical heat load distribution data of the water cooling system. The temperature measuring node includes at least one of the following: the outlet of the main circulation loop, the cooling branch interface of the high heat density equipment, and the cold side inlet of the heat exchanger. Real-time acquisition of temperature data from temperature measurement nodes and acquisition of the overall thermal state of the water cooling system; determination of the correlation between each temperature measurement node and the overall thermal state. Obtain the historical temperature change rate of each temperature measurement node, and determine the weight coefficient of each temperature measurement node based on the historical temperature change rate and correlation of the corresponding temperature measurement node. The current temperature is obtained by weighting and fusing the temperature data from each temperature measurement node according to the corresponding weight coefficient.

[0008] In some embodiments, determining the feedforward comparison result between the system temperature trend and a preset temperature threshold includes: Based on the system temperature trend, multiple predicted temperature values ​​for a future preset time period are extracted, and the average value of the multiple predicted temperature values ​​is calculated to obtain the trend temperature mean. Calculate the difference between the trend temperature mean and the preset temperature threshold, and determine the difference as the feedforward deviation; Based on the feedforward deviation and the preset deviation-fan number mapping table, the number of fans to be added or removed for feedforward control is determined, and the number of fans to be added or removed is determined as the feedforward comparison result. The deviation-fan number mapping table is a quantitative relationship table pre-calibrated based on the system thermal inertia and the fan heat dissipation capacity.

[0009] In some embodiments, determining the feedback comparison result based on the current temperature and a preset start temperature sequence and a preset stop temperature sequence includes: The current temperature is compared with the various start-up thresholds in the preset start-up temperature sequence to determine the start-up level to which the current temperature belongs, and the number of additional fans to be started is determined according to the start-up level. The current temperature is compared with the stop thresholds at each level in the preset stop temperature sequence to determine the stop level to which the current temperature belongs, and the number of fans to be reduced or stopped is determined according to the stop level. The net change in the number of fans required for feedback control is determined based on the number of fans added and the number of fans stopped, and the net change in the number of fans is determined as the feedback comparison result.

[0010] In some embodiments, the process of generating the number of wind turbines by fusing the feedforward control commands and feedback control commands includes: Obtain the current operating status of the water cooling system and the number of working fans currently in operation at the current moment, and determine whether the current operating status is in the heating stage or the cooling stage; If the temperature is rising, add the number of fans that need to be added or stopped to the number of working fans to obtain the initial number of operating fans; If the cooling phase is underway, the net change in the number of fans is added to the number of working fans to obtain the initial number of operating fans; If the temperature is not rising or falling, the number of working fans will be determined as the initial number of operating fans; The initial number of operating wind turbines is constrained by upper and lower limits to obtain the total number of operating wind turbines.

[0011] In some embodiments, determining the target wind turbine to be started or stopped based on the wind turbine operating data and the cumulative operating time includes: Obtain the cumulative operating time of all wind turbines, and construct separate sets of operating wind turbines and shut-down wind turbines; If the number of operating fans is greater than the number of working fans, obtain the first fan difference between the total number of fans in the water cooling system and the number of working fans, and the second fan difference between the number of operating fans and the number of working fans, and determine the minimum value between the first fan difference and the second fan difference as the first target number of target fans. Select the first target number of wind turbines with the shortest cumulative operating time from the set of shut-down wind turbines as the target wind turbines; If the number of operating fans is less than the current number of working fans, obtain a third fan difference between the number of working fans and the number of operating fans, and determine the minimum value between the third fan difference and the number of working fans as the second target number of target fans; Select the second target number of wind turbines from the set of operating wind turbines that have the longest cumulative operating time and have not reached the preset minimum continuous operating time for a single run as the target wind turbines.

[0012] In some embodiments, the method further includes updating the cumulative operating time of the target wind turbine after each wind turbine start-up and shutdown operation.

[0013] Secondly, this embodiment provides a cascaded water-cooling heat dissipation control system, the system comprising: an acquisition module, a processing module, and a control module; wherein, The acquisition module is used to collect the current temperature of the water cooling system in real time; The processing module is used to calculate the rate of change of the current temperature at the current moment based on the current temperature and the historical temperature at the previous moment, and input the current temperature and the rate of change of the current temperature into a preset thermal dynamic model to predict the system temperature trend at future moments. The control module is used to determine the feedforward comparison result between the system temperature trend and the preset temperature threshold, and generate corresponding feedforward control instructions. It also determines the feedback comparison result based on the current temperature and the preset start temperature sequence and preset stop temperature sequence, and generates corresponding feedback control instructions. The processing module is also used to fuse the feedforward control command and the feedback control command to generate the number of wind turbines in operation, obtain the cumulative operating time of each wind turbine, and determine the target wind turbines to be started or stopped based on the number of wind turbines in operation and the cumulative operating time. The control module is also used to perform start-up and shutdown operations on the wind turbine based on the state of the target wind turbine and the preset minimum continuous running time.

[0014] Thirdly, this embodiment provides a computer-readable storage medium storing a computer program that can run on a processor, wherein the computer program, when executed by the processor, implements a cascaded water-cooling heat dissipation control method as described in the first aspect.

[0015] By employing the above method, this application first acquires the current temperature of the water-cooling system in real time, and calculates the rate of change of the current temperature based on the current temperature and the historical temperature of the previous moment. Then, the current temperature and the rate of change of the current temperature are input into a preset thermal dynamic model to predict the system temperature trend in the future. Next, the feedforward comparison result between the system temperature trend and a preset temperature threshold is determined, and a corresponding feedforward control command is generated; and a feedback comparison result is determined based on the current temperature and preset start-up and stop-down temperature sequences, and a corresponding feedback control command is generated. Next, the feedforward control command and the feedback control command are merged to generate the number of operating fans. Finally, the cumulative operating time of each fan is obtained, and the target fans to be started and stopped are determined based on the number of operating fans and the cumulative operating time. The start-up and stop operations of the fans are executed based on the status of the target fans and a preset minimum continuous operating time. This enables low-energy and accurate heat dissipation control. Attached Figure Description

[0016] Figure 1 This is a block diagram of a cascaded water cooling heat dissipation control method provided in this application.

[0017] Figure 2This is a block diagram of the method for real-time acquisition of the current temperature of a water-cooling system provided in this application.

[0018] Figure 3 This is a block diagram of the method provided in this application for determining the feedforward comparison result of the system temperature trend and the preset temperature threshold.

[0019] Figure 4 This is a flowchart of a method provided in this application for determining feedback comparison results based on the current temperature and preset start temperature sequence and preset stop temperature sequence.

[0020] Figure 5 This is a connection diagram of a cascaded water-cooled heat dissipation control system provided in this application. Detailed Implementation

[0021] To better understand the purpose, technical solutions, and advantages of this application, it has been described and illustrated below with reference to the accompanying drawings and embodiments. However, those skilled in the art should understand that this application can be implemented without these details. It will be apparent to those skilled in the art that various modifications can be made to the embodiments disclosed in this application, and the general principles defined in this application can be applied to other embodiments and application scenarios without departing from the principles and scope of this application. Therefore, this application is not limited to the illustrated embodiments, but is consistent with the broadest scope claimed in this application.

[0022] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0023] The specific application scenario for this application is a cascaded water cooling system. Figure 1 This is a block diagram of a cascaded water-cooling heat dissipation control method provided in this application. Figure 1 As shown, a cascaded water-cooling heat dissipation control method includes the following steps: Step S100: Collect the current temperature of the water cooling system in real time, and calculate the current temperature change rate based on the current temperature and the historical temperature at the previous moment.

[0024] This application describes the process from the control perspective. Before commencing heat dissipation control, the system requires initial configuration. This initialization process is executed by the control unit and aims to lay the foundation for subsequent predictive control, feedback judgment, and balanced scheduling. System initialization parameters include preset start-up temperature sequence, preset stop-down temperature sequence, preset minimum continuous runtime, and preset thermal dynamic model.

[0025] The preset start-up temperature sequence refers to a set of start-up thresholds Tstart1, Tstart2, ..., TstartN arranged from low to high temperature, where N is the total number of fans in the system. When the system temperature reaches a certain start-up threshold, the corresponding number of fans will be started. The preset stop-down temperature sequence refers to a set of stop thresholds Tstop1, Tstop2, ..., TstopN arranged from high to low temperature. When the system temperature drops to a certain stop-down threshold, the corresponding number of fans will be stopped. The preset start-up and preset stop-down temperature sequences are not simply evenly distributed, but are obtained based on the system's thermal characteristics testing and historical operating data statistics. Specifically, during the system design or commissioning phase, different heat load conditions are simulated, and the correspondence between system temperature and the minimum required number of fans is recorded. Combining the fan heat dissipation capacity curve and the system heat capacity characteristics, the temperature boundaries that can achieve smooth switching are determined while ensuring a heat dissipation safety margin. Finally, a non-equal interval temperature sequence is formed; for example, a denser threshold is set in the high-temperature zone to achieve more precise control under high-temperature conditions.

[0026] The preset minimum continuous operating time per cycle refers to the shortest time the fan must run after each startup, measured in seconds or minutes. Setting this preset minimum continuous operating time takes into account the fan's mechanical lifespan, avoiding lifespan reduction caused by repeated start-stop cycles due to current surges and mechanical stress. It also considers system thermal inertia, ensuring that the fan's start-stop actions have a noticeable impact on system temperature and preventing control oscillations caused by excessively rapid actions. Furthermore, it incorporates engineering experience values, typically set between 30 seconds and 5 minutes based on the fan motor's start-stop characteristics and the thermal management system's time constant, and fine-tuned through on-site commissioning.

[0027] The preset thermal dynamic model is a simplified mathematical model used to predict the future temperature trend of the system. In a preferred embodiment, a first-order inertial plus pure time-delay system is used as the model structure. The parameters in the preset thermal dynamic model can be obtained through system identification. Specifically, a step response test is first performed, i.e., after the system's initial commissioning or major overhaul, the number of operating fans is manually changed in a step manner, and the dynamic response curve of the system temperature is recorded. Then, data fitting is performed, i.e., parameter estimation algorithms such as the least squares method are used to fit the response curve, thereby identifying the model parameters. Finally, an online learning mechanism is implemented, i.e., during system operation, the model continuously compares the predicted temperature with the actual measured temperature. When the deviation continuously exceeds a set tolerance, a parameter fine-tuning process is automatically triggered, enabling the model to adapt to characteristic drift caused by scaling, flow rate changes, or environmental changes.

[0028] On the one hand, the preset start-up and stop-down temperature sequences provide reliable feedback control benchmarks, ensuring the basic safety and stability of the system. The preset thermal dynamic model gives the system the ability to anticipate future changes, realizing feedforward control. The combination of these two is the core of this application's ability to achieve proactive adjustment and rapid response. On the other hand, the introduction of a preset minimum continuous operating time reduces high-frequency vibrations of the fan from a control logic perspective, significantly reducing mechanical and electrical wear and tear on the fan and extending its overall service life.

[0029] After the system initialization is completed, the control terminal will collect the current temperature of the water cooling system in real time. Figure 2 This is a block diagram of the method for real-time acquisition of the current temperature of a water-cooling system provided in this application. Figure 2 As shown, real-time acquisition of the current temperature of the water cooling system includes the following steps: Step S101: Obtain the topology and historical heat load distribution data of the water cooling system, and determine at least one temperature measuring node based on the topology and historical heat load distribution data of the water cooling system. The temperature measuring node includes at least one of the following: the outlet of the main circulation loop, the cooling branch interface of the high heat density equipment, and the cold side inlet of the heat exchanger.

[0030] Step S102: Collect temperature data from temperature measurement nodes in real time and obtain the overall thermal state of the water cooling system, and determine the correlation between each temperature measurement node and the overall thermal state.

[0031] Step S103: Obtain the historical temperature change rate of each temperature measurement node, and determine the weight system of each temperature measurement node based on the historical temperature change rate and correlation degree of the corresponding temperature measurement node.

[0032] Step S104: The temperature data of each temperature measurement node are weighted and fused according to the corresponding weight coefficients to obtain the current temperature.

[0033] The topology of a water-cooling system refers to the physical connections and layout of components such as cooling pipes, water pumps, heat exchangers, and the equipment being cooled. For example, it can be a series loop or multiple parallel branches. Historical heat load distribution data refers to the heat generated by different areas or equipment during the system's past operation, which can be obtained through historical operating logs, equipment power data, or thermal imaging analysis.

[0034] The determination of temperature measurement nodes is not based on arbitrary placement of temperature sensors, but rather on strategic placement based on a deep understanding of the system. Specifically, the outlet of the main circulation loop best reflects the overall thermal state of the entire system after circulation. Cooling branch interfaces of high heat density equipment, such as the cooling inlets or outlets of core heat-generating components like CPUs, GPUs, and IGBTs, are also important. These points are hotspots of localized heat load, experiencing rapid temperature changes and being most sensitive to system thermal disturbances. The cold-side inlet of the heat exchanger reflects the effectiveness of the external cooling source, such as ambient air or a cooling tower, and is a key monitoring point for heat exchange between the system and the outside world.

[0035] Specifically, determining temperature measurement nodes based on the topology and historical heat load distribution data of a water-cooling system is a systematic and multi-step analysis and decision-making process. The specific implementation involves first performing topology analysis to identify key structural points. This includes identifying system confluence points, i.e., analyzing the topology to find the point where the coolant ultimately converges and flows to the heat exchanger or water tank in the entire cooling loop—the outlet of the main circulation loop. This point carries the total heat of all branches, and its temperature is the most direct and authoritative reflection of the overall system heat load, making it an essential node for global status monitoring. It also includes identifying high-heat-load branches, i.e., identifying cooling branches connecting known high-power devices in the topology, such as the core switch in a server rack or the cooling branches of IGBT modules in a converter. Placing monitoring points at the inlet or outlet of these branches allows for precise capture of the real-time heat dissipation of these critical devices. These points serve as outposts for dealing with localized hotspots and sudden heat loads. Finally, it includes identifying heat exchange interfaces, i.e., locating the main heat exchangers in the system and placing monitoring points at their cold-side inlets—where the external cooling medium enters. The temperature at this point directly affects the ultimate heat dissipation capacity of the entire water cooling system and is a key node for monitoring changes in the external cooling environment, such as a decrease in cooling tower efficiency or an increase in ambient temperature.

[0036] Then, historical heat load data analysis is performed to identify dynamic sensitive points. This includes constructing a spatiotemporal distribution map of the heat load, i.e., retrieving and analyzing historical heat load data, such as equipment power logs and historical temperature records, to draw a heat load distribution map of the system in both time and space dimensions. This heat load distribution map is used to identify persistently high-heat areas, areas with frequent load fluctuations, and historical overheating record points. Persistently high-heat areas refer to certain equipment or regions that operate at high power for extended periods, serving as stable heat sources for the system. Monitoring points are deployed in these areas for steady-state monitoring of the base load. Areas with frequent load fluctuations refer to certain equipment with drastic power changes and frequent start-stop cycles, such as server clusters whose power varies according to computing tasks. Monitoring points are deployed in these areas to capture rapid dynamic disturbances, providing early signals for predictive control. Historical overheating record points refer to areas where temperature alarms or faults have occurred in the past, which are then prioritized for enhanced reliability monitoring.

[0037] Finally, a decision is made by fusing topology and thermal data. This involves overlaying and comprehensively evaluating the results of the two previous steps. An ideal temperature sensing node typically meets one or more of the following conditions: it is located on a critical path in the topology, such as the main loop or a high-heat-flux branch; it corresponds to a high-load or high-fluctuation region in historical data; and its measurements are highly representative and predictive of the overall thermal state of the system. Through this fusion analysis, the final set of temperature sensing nodes constitutes a three-dimensional, intelligent temperature sensing network that can reflect the global steady state of the system, keenly perceive local dynamics, and take into account changes in external cooling conditions.

[0038] By analyzing CAD drawings, thermal simulation data, and historical operation and maintenance data, the locations of the aforementioned key nodes were comprehensively determined, and high-precision temperature sensors were installed. This ensured the comprehensiveness and representativeness of temperature acquisition and avoided control misjudgments caused by inaccurate or delayed single-point measurements.

[0039] Real-time acquisition of temperature data from temperature measurement nodes refers to periodically reading the temperature values ​​of each node using the sensors arranged as described above. The overall thermal state is a comprehensive indicator, defined as the weighted average of the temperatures of all temperature measurement nodes, the temperature of the main loop outlet, or the temperature corresponding to the total heat capacity of the system calculated using a simplified thermodynamic model. Correlation measures the degree of influence of node temperature changes on the overall thermal state of the system. Specifically, correlation can be calculated using a general calculation method, i.e., collecting the temperature data sequence of all nodes and the overall thermal state data sequence over a period of time, and calculating the Pearson correlation coefficient or mutual information entropy between the temperature of each node and the overall thermal state. A higher correlation coefficient, or greater mutual information, indicates a stronger correlation between the node and the overall state, and its weight should be greater. The correlation at the main loop outlet is usually the highest, while the correlation of a certain local branch may be relatively low.

[0040] The historical temperature change rate of each temperature measurement node can be obtained by calculating the temperature change rate within each sampling period using the temperature data from the existing temperature measurement nodes. The weighting coefficient is determined by two factors: correlation and historical temperature change rate. A weighted fusion formula can be used to determine the weighting coefficient of each temperature measurement node. For example, weighting coefficient = preset first harmonic coefficient * correlation + preset second harmonic coefficient * historical temperature change rate. This dynamic weighting adaptively focuses on which temperature measurement points are both important and active.

[0041] The temperature collected in real time from each node is multiplied by its corresponding weighting coefficient, and then summed to obtain a single current temperature for subsequent control. Specifically, the controller performs the above weighted summation operation in real time. This fused current temperature is no longer the original temperature of a certain physical point, but a synthetic virtual temperature that can more accurately and proactively reflect the overall thermal load status of the system.

[0042] This approach, through strategic placement of multiple key nodes, avoids control failures caused by single-point sensor malfunctions or unrepresentative locations, enhancing system robustness. Furthermore, by introducing correlation and historical rate of change to dynamically calculate weights, the fused current temperature not only reflects the static thermal state but also contains dynamic trend information of the system. This provides higher-quality and more forward-looking input data for subsequent prediction steps, indirectly strengthening the anticipatory adjustment capability of feedforward control. Additionally, it moves beyond a rigid view of all temperature measurement points, adaptively adjusting monitoring priorities based on changes in operating data and heat load distribution.

[0043] Once the current temperature is obtained, the previous current temperature is subtracted from it, i.e., the historical temperature at the previous moment, to obtain the temperature difference. Then, the temperature difference is divided by the time difference between the current time corresponding to the current temperature and the historical time corresponding to the historical temperature to obtain the current temperature change rate at the current moment.

[0044] Step S200: Input the current temperature and the current temperature change rate into the preset thermal dynamic model to predict the system temperature trend at future moments.

[0045] The above step S200 is to predict the system temperature in advance in the short term based on the current real-time state of the system, i.e. the current temperature and the dynamic trend, i.e. the current rate of temperature change, using a preset mathematical model that can characterize the thermal behavior of the system, thereby providing a decision basis for feedforward control and overcoming the inherent lag of traditional feedback control.

[0046] Specifically, the first step is to construct and identify the parameters of a preset thermal dynamic model. The construction and parameter identification of the preset thermal dynamic model can be completed by referring to the description of the preset thermal dynamic model in the initialization process above, which will not be repeated here.

[0047] Then, using the current temperature and its rate of change as input conditions for the model, the temperature at future moments is estimated. Specifically, the current temperature is used as the initial state for model prediction, and the current rate of change is used as a quantitative indicator of the thermal disturbance currently experienced by the system. This rate of change implies the degree and direction of the imbalance between the current heat load and heat dissipation capacity. In the offline time domain, the model is discretized, and the controller performs the following prediction calculation in each control cycle, assuming that the current number of operating fans remains unchanged in the future prediction time domain. Starting from the current temperature and using the trend reflected by the current rate of change as the initial excitation, the discretized thermal dynamic model iteratively calculates the system temperature value at each moment in the future prediction time domain, thus forming a complete future system temperature trend line. This combination of instantaneous rate of change and inertial model, where the rate of change provides the initial velocity of the system dynamics and the model simulates the system's inertia and delay, makes the prediction no longer a simple linear extrapolation. It can more accurately predict the temperature rise trajectory before reaching the peak or the temperature drop trajectory after the inflection point, thereby capturing potential risks that feedback control cannot detect.

[0048] Finally, a data sequence containing predicted temperatures at multiple future time points is output, and this trend is used for subsequent feedforward comparisons. Specifically, the controller determines the future temperature sequence calculated in the above steps as the system temperature trend. In this way, by outputting a complete, quantified future temperature trend, rather than a single predicted point, the control system can more comprehensively and robustly assess the future thermal state.

[0049] By combining the real-time rate of change with a simplified thermal dynamics model that features online adaptation, high-precision, adaptive short-term prediction of the system's future temperature trend is achieved. This mechanism enables the control system to "foresee the future," issuing control commands in advance when the temperature trend is clear but before reaching the action threshold, fundamentally solving the problem of response lag in traditional threshold control.

[0050] Step S300: Determine the feedforward comparison result between the system temperature trend and the preset temperature threshold, and generate the corresponding feedforward control command.

[0051] The above step S300 is to determine in advance whether the fan operation status needs to be adjusted to counteract the expected thermal disturbance based on the predicted future temperature trend before the system temperature actually reaches the action threshold of the feedback control, thereby overcoming the lag of pure feedback control. Figure 3 This is a block diagram of the method provided in this application for determining the feedforward comparison result between the system temperature trend and the preset temperature threshold. For example... Figure 3 As shown, determining the feedforward comparison result between the system temperature trend and the preset temperature threshold includes the following steps: Step S301: Extract multiple predicted temperature values ​​within a preset time period based on the system temperature trend, and calculate the average value of the multiple predicted temperature values ​​to obtain the trend temperature mean.

[0052] Step S302: Calculate the difference between the trend temperature mean and the preset temperature threshold, and determine the difference as the feedforward deviation value.

[0053] Step S303: Based on the feedforward deviation value and the preset deviation-to-fan number mapping table, determine the number of fans that need to be added or removed from the feedforward control, and determine the number of fans that need to be added or removed from the feedforward comparison result. The deviation-to-fan number mapping table is a quantitative relationship table that is pre-calibrated based on the system thermal inertia and the fan heat dissipation capacity.

[0054] The system temperature trend refers to a sequence or curve consisting of a series of predicted system temperatures over a future period, output by a pre-defined thermal dynamics model. The pre-defined future period is a pre-set time window length, the selection of which must consider the system's thermal inertia and the timeliness of the control response, typically covering the time required from the current moment until the control action produces a noticeable effect. Multiple predicted temperature values ​​refer to a series of discrete temperature predictions obtained within the pre-defined future period, calculated according to the model's calculation step size. The trend temperature mean is a single value obtained by calculating the arithmetic mean or other weighted average of the multiple predicted temperature values. It represents the expected average level of the system temperature over this future period, used to smooth short-term fluctuations and more stably reflect the trend.

[0055] Specifically, after obtaining the system temperature trend predicted in step S200, the controller first extracts the portion of data corresponding to a preset future time period from the sequence. Then, the controller calculates the arithmetic mean of these extracted temperature values ​​and determines the resulting arithmetic mean as the trend temperature average. Using the average rather than a single future time-based prediction effectively smooths out potential short-term fluctuations or minor errors in the model's prediction, making the feedforward judgment more robust and focusing more on reflecting a stable, medium-term temperature change trend, thus avoiding control command jitter caused by noise at a single prediction point.

[0056] The aforementioned preset temperature threshold is not a single fixed threshold as in the prior art, but rather a reference temperature related to the control system's objective. Preferably, this preset temperature threshold can be the next-level start-up threshold in a preset start-up temperature sequence, or the next-level stop threshold in a preset stop temperature sequence, representing the critical temperature the system might reach in the future without intervention. The feedforward deviation value is the algebraic difference between the trend temperature mean and the preset temperature threshold. This feedforward deviation value quantifies the degree of deviation between the expected future thermal state and the desired safety boundary. A positive value indicates expected overheating, requiring increased heat dissipation; a negative value indicates expected undercooling, allowing for reduced heat dissipation.

[0057] Specifically, the controller compares the calculated average trend temperature with a pre-selected preset temperature threshold and calculates the difference between the two. This difference is the feedforward bias. This feedforward bias is a signed scalar whose magnitude and direction directly indicate the magnitude and direction of the heat dissipation capacity adjustment required to offset the predicted future thermal disturbances.

[0058] The aforementioned pre-defined deviation-to-fan quantity mapping table is a lookup table calibrated beforehand through experiments, simulations, or theoretical calculations. It establishes a quantitative correspondence between the feedforward deviation and the number of fans needed to compensate for this deviation. This mapping relationship is non-linear, taking into account the system's thermal inertia and the heat dissipation capacity of each fan. The number of fans to be added or removed for feedforward control is obtained from the mapping table, suggesting an immediate increase or decrease in the number of operating fans. The feedforward comparison result is the final output of this step, i.e., a specific fan quantity adjustment result.

[0059] Specifically, the controller uses the feedforward deviation obtained in step S302 as input and queries a preset mapping table between deviation and fan quantity. This mapping table is typically established during system commissioning by applying known thermal disturbances under different heat load conditions, observing the number of fans required to start and stop to stabilize the system temperature near a threshold, recording the relationship between this number and the predicted deviation, and then performing curve fitting or piecewise linearization to create a table. For example, if the current feedforward deviation is greater than two degrees Celsius, the mapping is to add two fans; if the current feedforward deviation is no greater than two degrees Celsius but greater than one degree Celsius, the mapping is to add one fan; if the current feedforward deviation is no greater than one degree Celsius but greater than -1 degree Celsius, the mapping is that the number of fans remains unchanged; if the current feedforward deviation is no greater than -1 degree Celsius but greater than -2 degrees Celsius, the mapping is to stop one fan; if the current feedforward deviation is less than -2 degrees Celsius, the mapping is to stop two fans. The controller directly obtains the change in the number of fans corresponding to the current feedforward deviation by looking up the table or interpolating. The change in the number of wind turbines is the result of the feedforward comparison, which will serve as the core content of the feedforward control command.

[0060] Finally, the control unit generates corresponding feedforward control commands based on the obtained feedforward comparison results, which serve as a reference for the actuator's actions.

[0061] This approach uses trend temperature averages to smooth temperature fluctuations, quantifies future demand through feedforward deviation, and ultimately translates thermal demand into precise actuator (fan) commands via a pre-defined deviation-fan quantity mapping table, forming a complete and efficient feedforward control logic chain. On one hand, unlike feedback control which only responds passively after temperature exceeds limits, this solution anticipates trends and acts proactively, fundamentally eliminating system response lag and effectively preventing temperature overshoot or undershoot, resulting in high control precision. On the other hand, it can reduce heat dissipation capacity in advance—that is, when cooling is predicted, or only when necessary, when heating is predicted—avoiding the overcooling phenomenon common in traditional on / off control, thereby significantly reducing the total energy consumption of the fan system.

[0062] Step S400: Determine the feedback comparison result based on the current temperature and the preset start temperature sequence and preset stop temperature sequence, and generate the corresponding feedback control command.

[0063] Figure 4 This is a block diagram of the method provided in this application for determining feedback comparison results based on the current temperature and preset start temperature and preset stop temperature sequences. Figure 4 As shown, determining the feedback comparison result based on the current temperature and the preset start temperature sequence and preset stop temperature sequence includes the following steps: Step S401: Compare the current temperature with the various start-up thresholds in the preset start-up temperature sequence to determine the start-up level to which the current temperature belongs, and determine the number of additional fans to be turned on according to the start-up level.

[0064] Step S402: Compare the current temperature with the stop thresholds at each level in the preset stop temperature sequence to determine the stop level to which the current temperature belongs, and determine the number of fans to be stopped according to the stop level.

[0065] Step S403: Determine the net change in the number of fans required for feedback control based on the number of fans added and the number of fans stopped, and determine the net change in the number of fans as the feedback comparison result.

[0066] Step S401 above aims to determine, based on the current actual thermal state of the system, whether and how many additional fans are needed to enhance heat dissipation capacity. The aforementioned activation level refers to the range or grade of the current temperature within a preset activation temperature sequence. The number of additional fans to be activated is the difference between the total number of fans required to meet the heat dissipation requirements of the current activation level and the number of fans currently in operation.

[0067] Specifically, the process begins with comparison and grading. The controller collects and integrates the current temperature in real time and compares it step-by-step with the preset start-up temperature sequence stored in memory. The comparison logic is as follows: starting from the lowest start-up threshold Tstart1, it sequentially checks whether the current temperature has reached or exceeded a certain threshold until it finds the highest threshold that has been exceeded. The sequence number corresponding to this threshold is the start-up level to which the current temperature belongs. For example, suppose the preset start-up temperature sequence is [40,45,50,55]℃. If the current temperature is 47℃, it exceeds 45℃ (level 2) but does not reach 50℃ (level 3). Therefore, the start-up level to which the current temperature belongs is level 2. Then, the number of additional fans to be activated is determined. The start-up level to which the current temperature belongs directly indicates the total number of fans required for safe cooling at this temperature. Therefore, the number of additional fans to be activated = the start-up level to which the current temperature belongs - the number of fans currently running.

[0068] Step S402 above aims to determine, based on the current actual thermal state of the system, whether and by how much the heat dissipation capacity of the fans can be reduced to avoid overcooling. The aforementioned stop level refers to the range or level of the current temperature within a preset stop temperature sequence. The reduction in the number of fans to be stopped refers to the difference between the maximum number of fans allowed to operate by the system based on the current stop level and the number of fans currently operating.

[0069] Specifically, the process begins with comparison and grading. The controller compares the current temperature with a preset stop temperature sequence, level by level. The comparison logic typically starts with the highest stop threshold, Tstop1, and sequentially checks if the current temperature is below a certain threshold until the lowest current temperature is found that is still above its threshold. The sequence number corresponding to this threshold is the stop level to which the current temperature belongs. For example, suppose the preset stop temperature sequence is [52, 47, 42, 37]℃. If the current temperature is 45℃, it is below 47℃ (Level 2) but still above 42℃ (Level 3). Therefore, the stop level for the current temperature is Level 3. Next, the number of fans to be stopped is determined. The stop level indicates the maximum number of fans the system is allowed to operate at this temperature. Typically, the stop level is lower than the corresponding start level at the same temperature, creating hysteresis control. Therefore, the number of fans to be stopped = stop level - number of currently running fans. This result is usually 0 or negative, with a negative number indicating the number of fans that need to be stopped.

[0070] Step S403 above integrates and arbitrates the independent judgment results of steps S401 and S402 to generate a unified, clear, and conflict-free feedback control command. The net change in the number of fans refers to the net change in the number of operating fans that the feedback control loop ultimately determines to be made after comprehensive judgment; positive numbers indicate an increase, negative numbers indicate a decrease, and zero indicates no change. The feedback comparison result is the final output of step S403, representing a functionalized fan adjustment result.

[0071] Specifically, the system first performs logical arbitration, whereby the controller logically judges the number of additional fans and the number of fans to be shut down calculated in steps S401 and S402. If the number of additional fans is greater than zero, it indicates that the current temperature has reached a higher start-up level, posing a risk of overheating. In this case, the system should prioritize responding to the additional fan request to ensure safety, and the net change in the number of fans equals the number of additional fans. Requests to shut down fans are temporarily ignored at this point, as safe heat dissipation takes precedence over energy conservation. If the number of additional fans is 0 and the number of fans to be shut down is less than 0, it indicates that the current temperature has not triggered new start-up conditions but has decreased to a level where fan reduction is permissible. In this case, the system responds to the shutdown request to save energy, and the net change in the number of fans equals the number of fans to be shut down. If the number of additional fans is 0 and the number of fans to be shut down is 0, it indicates that the current number of operating fans meets both the start-up sequence requirements and the shutdown sequence limit, and the system is in a balanced state, with a net change in the number of fans equal to 0. Then the result is output, that is, the net change in the number of wind turbines calculated by the above arbitration logic is determined as the feedback comparison result.

[0072] This approach, by independently and asymmetrically handling start-up and stop conditions and prioritizing additional start-up requests, constructs an inherent and secure hysteresis control loop. This effectively avoids the "breathing effect"—the high-frequency oscillations between on-off and on-off states caused by small temperature fluctuations at the threshold point—significantly improving control stability and equipment lifespan, as seen in traditional single-threshold control. Furthermore, it decomposes the complex multi-level temperature threshold comparison problem into two orthogonal judgments: when to increase the start-up and when to decrease the stop-up. These judgments are then fused through explicit arbitration rules. This design ensures that under any temperature condition, the feedback control loop generates only one clear and unambiguous control command, effectively avoiding logical conflicts and command confusion, thus improving system reliability. Additionally, when the feedforward control and this feedback control command are fused, the feedback command acts as a safety anchor, ensuring that even with short-term deviations in the predictive model, the system does not deviate from the safety boundary based on actual temperature. This combination gives the entire control system both forward-looking rapid response capabilities and fundamental safety guarantees.

[0073] Step S500: The number of wind turbines in operation is generated by integrating the feedforward control command and the feedback control command.

[0074] Step S500 above demonstrates how to intelligently fuse the predicted and proactive feedforward control command with the safe and stable feedback control command based on the current measured temperature to generate the final number of operating wind turbines. The process of fusing the feedforward and feedback control commands to generate the number of operating wind turbines includes the following steps: Step S501: Obtain the current operating status of the water cooling system and the number of working fans currently in operation at the current moment, and determine whether the current operating status is in the heating stage or the cooling stage.

[0075] Step S502: If the temperature rises, add the number of fans that need to be added or stopped to the number of working fans to obtain the initial number of operating fans.

[0076] Step S503: If the cooling stage is underway, add the net change in the number of fans to the number of working fans to obtain the initial number of operating fans.

[0077] Step S504: If the temperature rises or falls, the number of working fans is determined as the initial number of operating fans.

[0078] Step S505: Apply upper and lower limit constraints to the initial number of operating fans to obtain the total number of operating fans.

[0079] The current operating status described above is a qualitative indicator characterizing the overall direction of heat change in the system, primarily based on the current rate of temperature change. The number of operating fans refers to the total number of fans currently in operation within the system.

[0080] Specifically, the controller reads the current temperature change rate calculated in step S100 above. Then, a positive threshold close to zero is set to avoid misjudgments caused by minor fluctuations when the temperature is nearly stable. The judgment logic is as follows: if the current temperature change rate is greater than the positive threshold, the system is determined to be in a heating phase, indicating that the system's heat generation is greater than its current heat dissipation, heat is accumulating, and the temperature is trending upwards. If the current temperature change rate is less than the negative threshold, the system is determined to be in a cooling phase, indicating that the system's heat dissipation is greater than its heat generation, heat is being carried away, and the temperature is trending downwards. If the current temperature change rate is between the negative and positive thresholds, the system is determined to be in a steady-state phase, meaning it is neither in a significant heating phase nor a significant cooling phase.

[0081] When the system is determined to be in the heating phase, the core of the control strategy is rapid response and prevention of overshoot. Therefore, the system prioritizes the use of feedforward control commands. The controller performs arithmetic calculations, i.e., the initial number of operating fans = the number of currently operating fans + the change in the number of fans in the feedforward control command.

[0082] When the system is determined to be in the cooling phase, the core of the control strategy is a smooth transition and avoiding oscillations. Overly aggressively following the predicted fan shutdown could lead to excessively rapid removal of cooling capacity, easily causing temperature rebounds and fan restarts should the heat load fluctuate unexpectedly. Therefore, the system prioritizes feedback control commands. The controller performs arithmetic calculations: the initial number of operating fans = the number of currently operating fans + the net change in the number of fans in the feedback control command.

[0083] When the system is in a steady state, it indicates that the heat load and heat dissipation capacity are basically balanced, and temperature changes are minimal. At this time, the controller determines that there is no need to immediately adjust the number of fans, so it initially sets the number of operating fans to be equal to the number of currently working fans, maintaining the existing operating state. This helps maintain system stability and reduces unnecessary actions.

[0084] The aforementioned upper and lower limit constraints refer to the minimum and maximum number of wind turbines that are physically allowed to operate in the system. The lower limit is usually 0, meaning all wind turbines are shut down, and the upper limit is the total number N of wind turbines installed in the system.

[0085] Specifically, the initial number of operating fans obtained through the aforementioned steps is a theoretical calculation value, which may exceed the system's physical capacity, such as a calculated value of -1 or greater than the total number of fans. The controller performs a limit operation on the initial number of operating fans. If the initial number of operating fans is less than zero, the number of operating fans is forcibly set to zero. If the initial number of operating fans is greater than the total number of fans, the number of operating fans is forcibly set to the total number of fans. If the initial number of operating fans is between zero and the total number of fans, then the number of operating fans equals the initial number of operating fans. This ensures that the final generated control commands are physically executable, prevents illegal commands, and improves the system's robustness and security.

[0086] In this way, during the heating phase, the system, guided by feedforward commands, achieves proactive control, significantly shortening the lag time in response to sudden increases in heat load, effectively suppressing temperature overshoot, and improving the dynamic performance of the control system. During the cooling phase, the system, guided by feedback commands, ensures that fan shutdown actions are based on the precise drop in actual temperature, avoiding premature or excessive fan shutdowns due to model prediction biases or short-term fluctuations, thus preventing control oscillations and ensuring the static stability and reliability of the system. On the other hand, it avoids blindly following predictions and excessively shutting down fans during the cooling trend, thereby reducing the number of fan restarts due to temperature rebounds. Simultaneously, the rapid response during heating avoids situations where prolonged high-power operation is forced due to response lag. These two factors combined reduce the overall energy consumption of the system. The reduced number of fan start-ups and shutdowns, especially unnecessary oscillatory start-ups and shutdowns, directly extends the mechanical and electrical life of the fans.

[0087] Step S600: Obtain the cumulative operating time of each wind turbine, and determine the target wind turbine to be started or stopped based on the number of wind turbines in operation and the cumulative operating time.

[0088] The system maintains a cumulative operating time counter in non-volatile memory for each wind turbine. The cumulative operating time of each wind turbine can be obtained by viewing each cumulative operating time counter. Then, the target wind turbines to be started or stopped are determined based on the number of operating wind turbines and their cumulative operating time. The process of determining the target wind turbines to be started or stopped based on the number of operating wind turbines and their cumulative operating time includes the following steps: Step S601: Obtain the cumulative running time of all wind turbines, and construct a set of running wind turbines and a set of shut-down wind turbines respectively.

[0089] Step S602: If the number of operating fans is greater than the number of working fans, obtain the first fan difference between the total number of fans in the water cooling system and the number of working fans, and the second fan difference between the number of operating fans and the number of working fans. Determine the minimum value between the first fan difference and the second fan difference as the first target number of target fans.

[0090] Step S603: Select the first target number of wind turbines with the shortest cumulative running time from the set of shut-down wind turbines as the target wind turbines.

[0091] Step S604: If the number of operating fans is less than the current number of working fans, obtain the third fan difference between the number of working fans and the number of operating fans, and determine the minimum value between the third fan difference and the number of working fans as the second target number of target fans.

[0092] Step S605: Select the second target number of wind turbines from the set of operating wind turbines that have the longest cumulative operating time and have not reached the preset minimum continuous operating time for a single run as target wind turbines.

[0093] The cumulative operating time mentioned above refers to the total operating time of each wind turbine since it was put into use or since the last maintenance reset. The set of operating wind turbines refers to the set of wind turbines currently in operation. The set of stopped wind turbines refers to the set of wind turbines currently in a stopped state.

[0094] Specifically, the controller reads the cumulative operating time data of each wind turbine from non-volatile memory. Simultaneously, based on the turbine's current operating status (running or stopped), it categorizes them into a set of running turbines and a set of stopped turbines, respectively. The construction of these two sets provides the data foundation for subsequent balanced control selection.

[0095] The first fan difference refers to the difference between the total number of available fans in the system and the number of fans currently in operation, representing the maximum number of fans that can be started. The second fan difference refers to the number of additional fans that need to be started, calculated based on control commands. The first target number refers to the actual number of fans that can be started, taking the smaller of the two differences to ensure that it does not exceed the system's physical limit.

[0096] Specifically, the actuator performs the following calculations: First fan difference = Total number of fans - Number of currently operating fans; Second fan difference = Number of operating fans - Number of currently operating fans; First target number = min(First fan difference, Second fan difference). This ensures that the system will not attempt to start more fans than actually available, enhancing the system's robustness.

[0097] The aforementioned shortest cumulative operating time refers to prioritizing the equipment with the shortest operating time among the shut-down fans to achieve wear equalization. The target fan refers to the one selected for this startup operation.

[0098] Specifically, the controller sorts the fans in the shutdown fan set in ascending order according to their cumulative running time, and selects the top target number of fans as the target for this startup. If multiple fans have the same running time, they can be further selected by equipment number or randomly to ensure fairness.

[0099] The aforementioned third fan difference refers to the difference between the current number of operating fans and the target number of operating fans, indicating the number of fans that need to be stopped. The second target number refers to the actual number of fans that can be stopped, which is the smaller value between the third fan difference and the current number of operating fans, to avoid illegal operations.

[0100] Specifically, the controller performs the following calculations: the third fan difference = the number of currently operating fans - the number of fans in operation; and the second target number = min(third fan difference, number of currently operating fans). This ensures that the system will not attempt to stop more than the number of currently running fans, preventing logical errors.

[0101] The aforementioned longest cumulative operating time refers to prioritizing the equipment with the longest operating time among the operating fans to achieve balanced wear. Not reaching the preset minimum continuous operating time per cycle ensures that the fan is not immediately shut down after startup, avoiding frequent start-stop cycles.

[0102] Specifically, the controller first filters out the fans in the set of operating fans whose single continuous operating time has exceeded the preset minimum continuous operating time, and then selects the second target number of fans with the longest cumulative operating time as the stop target. If none of the operating fans have reached the minimum operating time, the stop operation is not performed in this cycle, and the decision is made in the next control cycle.

[0103] This approach, on the one hand, prioritizes the start-up and shutdown of fans with the shortest or longest operating times, ensuring balanced use of all fans throughout their lifecycle and preventing excessive wear on some fans, thus significantly extending the overall lifespan of the fan group. On the other hand, combined with preset minimum continuous operating time limits, it effectively prevents high-frequency switching of fans under critical conditions, reducing electrical and mechanical shocks and improving the stability and reliability of system control. Furthermore, by using difference comparison and minimum value selection, it ensures that control commands are physically executable, avoiding logical errors such as starting non-existent fans or stopping non-operating fans, thereby enhancing the system's robustness.

[0104] Step S700: Perform start-up and shutdown operations on the wind turbine based on the status of the target wind turbine and the preset minimum continuous running time.

[0105] The execution of step S700 above is based on the target wind turbine list determined in the preceding steps and the system's preset minimum continuous operating time. The core of this step is to ensure that the wind turbine start-up and shutdown operations meet both real-time control requirements and comply with equipment protection rules, thereby avoiding mechanical and electrical stress caused by frequent switching. Specifically, it includes the following sub-steps.

[0106] First, the controller checks the status of the target wind turbine against the preset minimum continuous runtime constraint. Specifically, for each target wind turbine, the controller first queries its current status (running or stopped) and compares it with the preset minimum continuous runtime. For a target wind turbine to be started, the controller verifies whether the turbine is stopped and whether the interval between its last stop time and the current time exceeds the preset de-vibration time to prevent electrical shock. Simultaneously, it checks whether the turbine's cumulative running time has been recorded in non-volatile memory to ensure data persistence. For a target wind turbine to be stopped, the controller verifies whether the turbine is running and calculates its current continuous runtime, i.e., from the last start-up time to the current time. If this continuous runtime is less than the preset minimum continuous runtime, the stop operation is not performed this time; instead, the stop request is temporarily stored for re-evaluation in the next control cycle. This ensures that the wind turbine runs for a sufficient time after each start-up, avoiding frequent start-stops due to short-term fluctuations. If the target wind turbine does not meet the above constraints, such as the wind turbine to be stopped not reaching the minimum runtime, the controller records the abnormal state and triggers a delay handling mechanism. The system will recheck the status of the wind turbine in the next control cycle until the conditions are met. Simultaneously, the system will generate log events for subsequent maintenance analysis.

[0107] Then, the controller executes start / stop operations and monitors the results in real time. Specifically, the controller sends start / stop commands to the target fan's driver via a digital output module or communication bus. The commands include the fan's unique identifier, such as its ID number, and the action type, such as start or stop. After executing the command, the controller verifies in real time whether the fan's actual state matches the command through feedback signals, such as dry contact status, current detection, or driver status words. If a mismatch is detected, such as a command to start but the fan not running, the controller will retry the command, up to three times. If the retry fails, the controller will mark the fan as faulty and automatically remove it from the list of available fans to ensure the system continues to operate normally. For successfully executed start / stop operations, the controller immediately updates the fan's status flags; the started fan is set to running, and the stopped fan is set to stopped.

[0108] Next, the controller updates the cumulative running time and status records of the wind turbines. Specifically, the controller updates the cumulative running time each time a wind turbine starts or stops. For a started wind turbine, the start timestamp is recorded, and the continuous running timer for this period is initialized. For a stopped wind turbine, the duration of this continuous running is calculated and added to the total cumulative running time. The cumulative running time is stored in non-volatile memory to prevent loss in case of power failure. Simultaneously, the controller updates the set of running and stopped wind turbines to ensure that subsequent balanced control is based on the latest data. For example, a newly started wind turbine will be moved from the stopped set to the running set, and its cumulative running time will be used for subsequent priority start / stop decisions.

[0109] Finally, post-execution evaluation and adaptive adjustment are performed. Specifically, after completing the start-up and shutdown operations, the controller evaluates the overall system performance, comparing the deviation between the actual temperature change and the predicted trend. If the deviation continues to exceed the tolerance, a parameter fine-tuning process for the preset thermal dynamic model is triggered, such as updating the model parameters online using the least squares method. This allows the system to adapt to long-term changes, such as scaling or ambient temperature fluctuations. In addition, the system statistically analyzes the number of fan start-ups and shutdowns and the balance, i.e., the standard deviation of the cumulative operating time of all fans. If the balance exceeds a threshold, the controller will prioritize adjusting fans with large operating time deviations in subsequent control cycles to enhance the balancing effect.

[0110] This approach, by imposing a pre-set minimum continuous operating time, avoids repeated start-stop cycles of the wind turbines within a short period, reducing mechanical wear and electrical shocks. Combined with balancing strategies, such as prioritizing the start-stop of turbines with extreme cumulative operating times, it ensures a more even distribution of wear across the turbine group, thus extending overall lifespan. Furthermore, real-time monitoring of execution results and anomaly handling mechanisms prevent control failures caused by turbine malfunctions. Additionally, delaying the shutdown of turbines that have not reached the minimum operating time avoids unnecessary energy consumption fluctuations and temperature oscillations.

[0111] Preferably, the cumulative operating time of the target fan is updated after each fan start-up and shutdown operation. This real-time and accurate data update improves the accuracy and reliability of the balanced control, extends the fan life, and enhances the system's adaptability and robustness.

[0112] Figure 5 This is a connection diagram of a cascaded water-cooled heat dissipation control system provided in this application. Figure 5 As shown, a cascaded water-cooled heat dissipation control system includes: an acquisition module, a processing module, and a control module.

[0113] The system comprises several modules: an acquisition module for real-time temperature monitoring of the water-cooling system; a processing module for calculating the rate of change of the current temperature based on the historical temperature from the previous moment, inputting the current temperature and rate of change into a preset thermal dynamic model to predict the future system temperature trend; a control module for determining the feedforward comparison result between the system temperature trend and a preset temperature threshold, generating corresponding feedforward control commands, determining the feedback comparison result based on the current temperature and preset start and stop temperature sequences, and generating corresponding feedback control commands; a processing module for fusing the feedforward and feedback control commands to generate the number of operating fans, obtaining the cumulative operating time of each fan, and determining the target fans to be started or stopped based on the number of operating fans and the cumulative operating time; and a control module for executing the start and stop operations of the fans based on the status of the target fans and a preset minimum continuous operating time.

[0114] The other functions performed by the acquisition module, processing module, and control module, as well as the technical details of each function, are the same as or similar to the corresponding features in the cascaded water cooling heat dissipation control method described above, so they will not be repeated here.

[0115] This application also provides a computer storage medium storing a computer program that, when run on a computer, enables the computer to execute the steps in the cascaded water cooling heat dissipation control method described above.

[0116] It should be understood that although the steps in the flowcharts in the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise expressly stated herein, there is no strict order in which these steps are performed, and they may be performed in other orders.

[0117] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A cascade water cooling heat dissipation control method, characterized in that, The method includes: The current temperature of the water cooling system is collected in real time, and the rate of change of the current temperature at the current moment is calculated based on the current temperature and the historical temperature at the previous moment. The current temperature and the rate of change of the current temperature are input into a preset thermal dynamics model to predict the system temperature trend at future times. Determine the feedforward comparison result between the system temperature trend and the preset temperature threshold, and generate the corresponding feedforward control command; Based on the current temperature and the preset start temperature sequence and preset stop temperature sequence, the feedback comparison result is determined, and the corresponding feedback control command is generated; The number of operating fans is generated by integrating the feedforward control commands and feedback control commands. Obtain the cumulative operating time of each wind turbine, and determine the target wind turbines to be started or stopped based on the number of wind turbines in operation and the cumulative operating time; The start-up and shutdown operations of the wind turbine are performed based on the state of the target wind turbine and the preset minimum continuous running time.

2. The method of claim 1, wherein, Before the real-time acquisition of the current temperature of the water cooling system, the following is also included: Obtain system initialization parameters, which include a preset start-up temperature sequence, a preset stop temperature sequence, a preset minimum continuous runtime, and a preset thermal dynamic model.

3. The method of claim 1, wherein, The real-time temperature of the water cooling system includes: The topology and historical heat load distribution data of the water cooling system are obtained, and at least one temperature measuring node is determined based on the topology and historical heat load distribution data of the water cooling system. The temperature measuring node includes at least one of the following: the outlet of the main circulation loop, the cooling branch interface of the high heat density equipment, and the cold side inlet of the heat exchanger. Real-time acquisition of temperature data from temperature measurement nodes and acquisition of the overall thermal state of the water cooling system; determination of the correlation between each temperature measurement node and the overall thermal state. Obtain the historical temperature change rate of each temperature measurement node, and determine the weight coefficient of each temperature measurement node based on the historical temperature change rate and correlation of the corresponding temperature measurement node. The current temperature is obtained by weighting and fusing the temperature data from each temperature measurement node according to the corresponding weight coefficient.

4. The method of claim 2, wherein, The feedforward comparison result between the system temperature trend and the preset temperature threshold includes: Based on the system temperature trend, multiple predicted temperature values ​​for a future preset time period are extracted, and the average value of the multiple predicted temperature values ​​is calculated to obtain the trend temperature mean. Calculate the difference between the trend temperature mean and the preset temperature threshold, and determine the difference as the feedforward deviation; Based on the feedforward deviation and the preset deviation-fan number mapping table, the number of fans to be added or removed for feedforward control is determined, and the number of fans to be added or removed is determined as the feedforward comparison result. The deviation-fan number mapping table is a quantitative relationship table pre-calibrated based on the system thermal inertia and the fan heat dissipation capacity.

5. The method of claim 4, wherein, The step of determining the feedback comparison result based on the current temperature and the preset start temperature sequence and preset stop temperature sequence includes: The current temperature is compared with the various start-up thresholds in the preset start-up temperature sequence to determine the start-up level to which the current temperature belongs, and the number of additional fans to be started is determined according to the start-up level. The current temperature is compared with the stop thresholds at each level in the preset stop temperature sequence to determine the stop level to which the current temperature belongs, and the number of fans to be reduced or stopped is determined according to the stop level. The net change in the number of fans required for feedback control is determined based on the number of fans added and the number of fans stopped, and the net change in the number of fans is determined as the feedback comparison result.

6. The method of claim 5, wherein, The generation of the number of operating wind turbines by fusing the feedforward control commands and feedback control commands includes: Obtain the current operating status of the water cooling system and the number of working fans currently in operation at the current moment, and determine whether the current operating status is in the heating stage or the cooling stage; If the temperature is rising, add the number of fans that need to be added or stopped to the number of working fans to obtain the initial number of operating fans; If the cooling phase is underway, the net change in the number of fans is added to the number of working fans to obtain the initial number of operating fans; If the temperature is not rising or falling, the number of working fans will be determined as the initial number of operating fans; The initial number of operating wind turbines is constrained by upper and lower limits to obtain the total number of operating wind turbines.

7. The method of claim 2, wherein, The process of determining the target wind turbines to be started or stopped based on the wind turbine operating data and the cumulative operating time includes: Obtain the cumulative operating time of all wind turbines, and construct separate sets of operating wind turbines and shut-down wind turbines; If the number of operating fans is greater than the number of working fans, obtain the first fan difference between the total number of fans in the water cooling system and the number of working fans, and the second fan difference between the number of operating fans and the number of working fans, and determine the minimum value between the first fan difference and the second fan difference as the first target number of target fans. Select the first target number of wind turbines with the shortest cumulative operating time from the set of shut-down wind turbines as the target wind turbines; If the number of operating fans is less than the current number of working fans, obtain a third fan difference between the number of working fans and the number of operating fans, and determine the minimum value between the third fan difference and the number of working fans as the second target number of target fans; Select the second target number of wind turbines from the set of operating wind turbines that have the longest cumulative operating time and have not reached the preset minimum continuous operating time for a single run as the target wind turbines.

8. The method of claim 1, wherein, The method further includes updating the cumulative operating time of the target wind turbine after each wind turbine start-up and shutdown operation.

9. A cascade water-cooling heat dissipation control system, characterized in that, The system includes: an acquisition module, a processing module, and a control module; wherein, The acquisition module is used to collect the current temperature of the water cooling system in real time; The processing module is used to calculate the rate of change of the current temperature at the current moment based on the current temperature and the historical temperature at the previous moment, and input the current temperature and the rate of change of the current temperature into a preset thermal dynamic model to predict the system temperature trend at future moments. The control module is used to determine the feedforward comparison result between the system temperature trend and the preset temperature threshold, and generate corresponding feedforward control instructions. It also determines the feedback comparison result based on the current temperature and the preset start temperature sequence and preset stop temperature sequence, and generates corresponding feedback control instructions. The processing module is also used to fuse the feedforward control command and the feedback control command to generate the number of wind turbines in operation, obtain the cumulative operating time of each wind turbine, and determine the target wind turbines to be started or stopped based on the number of wind turbines in operation and the cumulative operating time. The control module is also used to perform start-up and shutdown operations on the wind turbine based on the state of the target wind turbine and the preset minimum continuous running time.

10. A computer readable storage medium having stored thereon a computer program, capable of running on a processor, characterized in that, When the computer program is executed by the processor, it implements a cascaded water-cooling heat dissipation control method as described in any one of claims 1 to 8.

Images