Control method and device for intelligent equipment cooling tower fan group, and equipment

Abstract

The application relates to the technical field of intelligent devices, and discloses a control method and device for a cooling tower fan group of an intelligent device and equipment. The control method comprises the following steps: collecting operation parameters of a water chiller unit, operation parameters of the cooling tower fan group and outdoor environment temperature and humidity; based on the collected operation parameters and the outdoor environment temperature and humidity, taking the number of opened cooling tower fans and the operation frequency of a single fan as coupled optimization variables, constructing a global energy consumption cost function with the minimum total power consumption of the intelligent device as a target, and setting corresponding operation constraints; based on the global energy consumption cost function and the operation constraints, performing global optimization on the coupled optimization variables to obtain optimal numbers of opened cooling tower fans and optimal operation frequencies that meet the operation constraints; and controlling the cooling tower fan group to operate at the optimal numbers of opened cooling tower fans and the optimal operation frequencies. The method can realize global energy consumption optimization of the water chiller unit and the cooling tower fan system in the whole working condition range.

Description

Technical Field

[0001] This application relates to the field of intelligent device technology, such as a control method, apparatus, and device for a cooling tower fan group of intelligent devices. Background Technology

[0002] With the increasing demands for energy consumption in HVAC systems from large public buildings, data centers, and industrial production, chiller units and cooling tower systems, as core energy-consuming units in HVAC systems, directly impact the overall system's energy consumption level through their operational efficiency. In actual operation, the cooling tower fan group is responsible for providing cooling water to the chiller units. The matching degree between the number of operating fans, their operating frequency, the chiller unit load, and outdoor environmental conditions plays a crucial role in the system's total energy consumption and operational stability.

[0003] While existing technologies attempt to optimize the operating frequency or number of fans in operation, they generally suffer from the following problems: First, the optimization objective is singular, focusing only on minimizing fan energy consumption or controlling cooling water temperature, failing to achieve optimal overall system energy consumption. Second, the target cooling water outlet temperature often uses a fixed setpoint, leading to either excessive or insufficient heat dissipation capacity under certain operating conditions, limiting energy-saving potential. Third, optimization calculations rely on directly solving complex global energy consumption models, resulting in large computational loads and long processing times, making it difficult to meet real-time control requirements and limiting engineering applicability.

[0004] Existing cooling tower fan group control methods cannot achieve optimal global energy consumption for the chiller unit and cooling tower fan system across all operating conditions, and cannot meet the high-efficiency, energy-saving, stable, and reliable operation requirements of modern HVAC systems.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.

[0007] This disclosure provides a control method, apparatus, and device for a cooling tower fan group of intelligent equipment, so as to achieve optimal global energy consumption of the chiller unit and cooling tower fan system across all operating conditions.

[0008] In some embodiments, the method includes: collecting operating parameters of the chiller unit, operating parameters of the cooling tower fan group, and outdoor ambient temperature and humidity; based on the collected operating parameters and outdoor ambient temperature and humidity, constructing a global energy consumption cost function with the goal of minimizing the total power consumption of the intelligent device, using the number of cooling tower fans in operation and the operating frequency of a single fan as coupled optimization variables, and setting corresponding operating constraints; based on the global energy consumption cost function and the operating constraints, performing global optimization on the coupled optimization variables to obtain the optimal number of cooling tower fans in operation and the optimal operating frequency that satisfy the operating constraints; and controlling the cooling tower fan group to operate at the optimal number of fans in operation and the optimal operating frequency.

[0009] In some embodiments, the apparatus includes a processor and a memory storing program instructions, the processor being configured to, when executing the program instructions, perform the aforementioned control method for a cooling tower fan group for a smart device.

[0010] In some embodiments, the device includes: a device body; and a control device for a cooling tower fan group for intelligent devices, as described above, which is installed on the device body.

[0011] The control method, apparatus, and equipment for cooling tower fan groups of intelligent devices provided in this disclosure can achieve the following technical effects:

[0012] By using the number of operating fans and the operating frequency of each fan as coupled optimization variables, and combining data such as the chiller units, cooling tower fan groups, and outdoor ambient temperature and humidity, a global energy consumption cost function is constructed and operational constraints are set. Optimal control parameters are obtained through global optimization, enabling precise control of the cooling tower fan group. This achieves global energy consumption optimization for the chiller units and cooling tower fan system, meeting the requirements of efficient, energy-saving, stable, and reliable operation of HVAC systems.

[0013] The above general description and the description below are exemplary and illustrative only and are not intended to limit this application. Attached Figure Description

[0014] One or more embodiments are illustrated by way of example with reference to the accompanying drawings. These illustrations and drawings do not constitute a limitation on the embodiments. Elements having the same reference numerals in the drawings are shown as similar elements. The drawings are not to be scaled. And wherein:

[0015] Figure 1 This is a schematic diagram of a control method for a cooling tower fan group for intelligent devices provided in an embodiment of this disclosure;

[0016] Figure 2 This is a schematic diagram illustrating the method provided in this embodiment of the present disclosure for constructing a global energy consumption cost function with the goal of minimizing the total power consumption of the smart device;

[0017] Figure 3 This is a schematic diagram of another control method for a cooling tower fan group for intelligent devices provided in an embodiment of this disclosure;

[0018] Figure 4 This is a schematic diagram illustrating the global optimization of coupling optimization variables in the method provided in this embodiment of the disclosure;

[0019] Figure 5 This is a schematic diagram illustrating how a trained radial basis function surrogate model assists a multi-objective genetic algorithm in global optimization in the method provided in this embodiment of the disclosure.

[0020] Figure 6 This is a schematic diagram of a control device for a cooling tower fan group of intelligent devices provided in an embodiment of this disclosure. Detailed Implementation

[0021] To provide a more detailed understanding of the features and technical content of the embodiments of this disclosure, the implementation of the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this disclosure. In the following technical description, for ease of explanation, several details are used to provide a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be simplified in their depiction to simplify the drawings.

[0022] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0023] Unless otherwise stated, the term "multiple" means two or more.

[0024] In this embodiment of the disclosure, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.

[0025] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0026] The term "correspondence" can refer to an association or binding relationship. The correspondence between A and B means that there is an association or binding relationship between A and B.

[0027] In this embodiment of the disclosure, the intelligent device is a heating, ventilation and air conditioning (HVAC) system, which includes a chiller unit, a cooling water circulation loop, a cooling tower fan group, terminal air conditioning equipment, and a control system. The various parts work together to complete the cooling and heat dissipation cycle.

[0028] Combination Figure 1 As shown in the figure, this disclosure provides a control method for a cooling tower fan group of intelligent devices, including:

[0029] S101, the sensor collects the operating parameters of the chiller unit, the operating parameters of the cooling tower fan group, and the outdoor ambient temperature and humidity.

[0030] S102, based on the collected operating parameters and outdoor ambient temperature and humidity, the processor constructs a global energy consumption cost function with the goal of minimizing the total power consumption of the intelligent device, using the number of cooling tower fans in operation and the operating frequency of a single fan as coupled optimization variables, and sets corresponding operating constraints.

[0031] S103, the processor performs global optimization on coupled optimization variables based on the global energy consumption cost function and operating constraints, and obtains the optimal number of cooling tower fans to be turned on and the optimal operating frequency that meets the operating constraints.

[0032] S104, the processor controls the cooling tower fan group to operate at the optimal number of fans and the optimal operating frequency.

[0033] The number of cooling tower fans in operation and the operating frequency of a single fan have a non-linear superposition relationship. The operating constraints include heat dissipation requirements, fan frequency, and the number of fans in operation.

[0034] Here, the operating parameters of the chiller unit reflect the current operating load and heat exchange status of the chiller unit; the operating parameters of the cooling tower fan group are used to define the adjustable range and safe operating boundaries of the fans; outdoor ambient temperature and humidity are used to calculate the outdoor wet-bulb temperature, characterizing the heat dissipation capacity of the current environment. The collected data provides comprehensive and accurate basic data for the subsequent construction of the global energy consumption cost function, the setting of operating constraints, and optimization calculations, avoiding optimization deviations caused by missing parameters. The operating parameters of the chiller unit include, but are not limited to, chiller unit load rate, cooling water circulation flow rate, cooling water inlet temperature, and cooling water outlet temperature. The operating parameters of the cooling tower fan group include, but are not limited to, fan rated parameters, minimum operating frequency, and maximum operating frequency. Outdoor ambient temperature and humidity include outdoor dry-bulb temperature and relative humidity.

[0035] In existing technologies, a single variable adjustment method is often used, adjusting only the fan frequency or the number of fans in operation, leading to local optima rather than global optima. This disclosure uses the number of cooling tower fans in operation and the operating frequency of a single fan as coupled optimization variables, breaking the limitations of single adjustment and enabling them to work synergistically. The global energy consumption cost function comprehensively considers the power consumption of the chiller unit and the power consumption of the cooling tower fan group, achieving unified control of the system's global power consumption. Specifically, the number of cooling tower fans in operation and the operating frequency of a single fan have a non-linear superposition relationship; the number of fans in operation... k With single unit operating frequency f The total power consumption of the cooling tower fan group is determined by coupling the parallel characteristic curves of the fans. P fan Operational constraints include heat dissipation demand constraints, fan frequency constraints, and fan quantity constraints. Among these, the heat dissipation demand constraint ensures that the actual heat dissipation of the cooling tower meets the heat dissipation requirements of the chiller unit, preventing chiller unit efficiency degradation and shutdown due to insufficient heat dissipation. The fan frequency and quantity constraints ensure that the fans operate within a safe and reasonable range, reducing equipment wear caused by frequent fan starts and stops and excessive frequency fluctuations. This extends the service life of the fans and chiller unit, ensuring stable system operation.

[0036] Based on the global energy consumption cost function and operational constraints, a global optimization of the coupled optimization variables is performed. Under the premise of satisfying all operational constraints, the optimal solution that minimizes the global power consumption of the system is searched, yielding the optimal number of cooling tower fans and the optimal operating frequency. This global optimization effectively avoids the energy waste caused by experience-based and single-variable adjustments in traditional control, ensuring that the operating parameters of each fan group are the optimal choice under the current operating conditions, achieving coordinated and efficient operation of the cooling tower fan group and the chiller unit. The optimal control parameters obtained through optimization are converted into actual control commands to drive the cooling tower fan group to operate precisely, enabling the entire HVAC system to operate stably under the current conditions with the minimum total power consumption.

[0037] The control method for cooling tower fan groups of intelligent devices provided in this disclosure uses the number of fans in operation and the operating frequency of a single fan as coupled optimization variables. Combined with data such as the chiller unit, the cooling tower fan group, and outdoor ambient temperature and humidity, a global energy consumption cost function is constructed and operating constraints are set. Optimal control parameters are obtained through global optimization, enabling precise operation of the cooling tower fan group. This effectively avoids the energy waste caused by traditional single-variable adjustment and experience-based adjustment.

[0038] Optionally, in step S101, the sensor collects the operating parameters of the chiller unit, the operating parameters of the cooling tower fan group, and the outdoor ambient temperature and humidity, including:

[0039] The sensors collect data on the chiller unit's load rate, cooling water circulation flow rate, cooling water inlet temperature, cooling water outlet temperature, the rated parameters and minimum operating frequency of the cooling tower fan, as well as the outdoor dry-bulb temperature and relative humidity.

[0040] Here, the core operating parameters of the chiller unit include the chiller unit's load rate, cooling water circulation flow rate, cooling water inlet temperature, and cooling water outlet temperature. The chiller unit's load rate reflects the current cooling load and is the basis for calculating the chiller unit's power consumption and determining the system's heat dissipation requirements. The cooling water circulation flow rate characterizes the circulation speed of the cooling water between the chiller unit's condenser and the cooling tower, affecting heat exchange efficiency and the system's total energy consumption. The cooling water inlet temperature and cooling water outlet temperature are key parameters for calculating the total heat that the chiller unit needs to dissipate and for judging the cooling tower's heat dissipation effect. Furthermore, the difference between these two temperatures reflects the actual heat dissipation of the cooling water, providing accurate data support for setting subsequent heat dissipation constraints and avoiding insufficient heat dissipation or energy waste due to inaccurate heat exchange data.

[0041] The operating parameters of the cooling tower fan group include the rated parameters and minimum operating frequency of the cooling tower fans. The rated parameters, including rated power, rated frequency, and rated airflow, define the maximum operating capacity and design limits of the fans, providing an upper limit reference for adjusting the fan operating frequency. The minimum operating frequency constrains the minimum operating threshold of the fans, preventing insufficient airflow and reduced heat dissipation due to excessively low operating frequencies. It also prevents problems such as vibration, excessive noise, and accelerated motor losses during low-frequency operation, ensuring the safe and stable operation of the fan group. Outdoor dry-bulb temperature and relative humidity are used to obtain the outdoor wet-bulb temperature, accurately characterizing the current outdoor environmental heat dissipation conditions. This provides environmental parameters for calculating the power consumption of the chiller unit and the fan group in the global energy consumption cost function, ensuring the accuracy of global energy consumption modeling.

[0042] In practical applications, temperature and humidity sensors are deployed at the air inlet of the cooling tower to collect dry-bulb temperature and relative humidity in real time, and the outdoor wet-bulb temperature is calculated in real time using the Psychrometric formula. T wb An atmospheric pressure sensor is also configured to correct for the impact of air density on fan performance. Temperature sensors and electromagnetic flow meters are deployed in the cooling water supply and return pipelines to monitor the inlet water temperature in real time. T in Outlet water temperature T out and circulating water volume G w Each cooling tower fan is equipped with an independent frequency converter, and the operating frequency of each fan is read in real time via communication. f i Motor current I iIt can monitor power factor and fault status, and can also send frequency setpoints to the frequency converter. f set And start / stop commands. After filtering and noise reduction, sensor data can be used as input variables for optimization algorithms, forming the sensing basis for closed-loop control.

[0043] Combination Figure 2 Optionally, in step S102, the processor constructs a global energy consumption cost function aimed at minimizing the total power consumption of the smart device, including:

[0044] S121, the processor constructs the power consumption of the chiller unit by performing polynomial fitting on the condensing temperature and the total heat that the chiller unit needs to dissipate.

[0045] S122, the processor calculates the total power consumption of the cooling tower fan group based on the cubic law of fan speed, motor efficiency, and inverter efficiency.

[0046] S123, the processor uses the sum of the power consumption of the chiller unit and the total power consumption of the cooling tower fan group as a global energy consumption cost function with the goal of minimizing the total power consumption of intelligent devices.

[0047] Here, by establishing separate power consumption models for the chiller unit and the cooling tower fan group, these two models are coupled to form a global optimization objective aimed at minimizing the total system power consumption, achieving a unified representation and optimization of the overall energy consumption of the HVAC system. The condensing temperature reflects the heat exchange effect on the condenser side of the chiller unit, and its magnitude is related to the cooling tower's heat dissipation capacity and outdoor environmental conditions. The total heat that the chiller unit needs to dissipate represents the unit's heat dissipation demand under the current load. By collecting condensing temperature and heat dissipation data under different operating conditions, a multinomial fitting method is used to establish a power consumption model for the chiller unit. This chiller unit power consumption model can accurately reflect the nonlinear characteristics of the chiller unit's power consumption changing with operating conditions, avoiding calculation errors caused by simplified models, and providing an accurate chiller energy consumption basis for global energy consumption optimization.

[0048] The power consumption of the wind turbines is cubically related to their rotational speed. Simultaneously considering the loss characteristics of motor efficiency and inverter efficiency at different frequencies, this model more accurately reflects the actual energy consumption of the wind turbine group under varying numbers and frequencies. Motor efficiency and inverter efficiency are introduced to correct the relationship between wind turbine power consumption and rotational speed. The wind turbine group power consumption model established in this way takes into account both physical mechanisms and actual operating characteristics, making the power consumption calculation more closely resemble engineering practice. Finally, the sum of the power consumption of the chiller units and the total power consumption of the cooling tower wind turbine group is used as the global energy consumption cost function with the goal of minimizing the total power consumption of intelligent devices.

[0049] In this embodiment, the model is constructed based on actual operating mechanisms, the parameters are easy to obtain, and the computational efficiency is high, which can meet the needs of real-time optimization control of cooling tower fan groups. By coupling the energy consumption of the chiller unit and the cooling tower fan group into a unified objective function, true system-level optimization is achieved, rather than localized equipment energy saving, resulting in a more significant overall energy-saving effect.

[0050] In detail, based on the total power consumption of smart devices Minimize the cost function J The total power is determined by the chiller unit's power consumption. Total power consumption of cooling tower fan group It consists of two parts:

[0051] .

[0052] Among them, the power consumption of the chiller unit It is the condensation temperature. A strongly nonlinear function. Condensation temperature. With cooling tower outlet water temperature There is a linear correlation , The condenser heat exchange temperature difference is used. Polynomial fitting is employed to obtain the chiller unit's power consumption. :

[0053] .

[0054] in, It was obtained based on multiple regression analysis of historical operating data. This represents the total heat that the system actually needs to remove. The model demonstrates the ability to reduce the cooling tower outlet water temperature. It can reduce the power consumption of chiller units However, this will increase the power consumption of the wind turbine.

[0055] The total power consumption model for the wind turbine group considers the characteristics of parallel operation of the turbines and the efficiency curve of the frequency converter. For the activated... Assuming uniform airflow distribution, the frequency of a single typhoon is... The fan shaft power follows a cubic law, and is corrected by introducing motor efficiency and inverter efficiency. Among these, motor efficiency... for:

[0056] .

[0057] in, The rated efficiency of the motor, k 1 represents the attenuation coefficient in the high-efficiency region. The efficiency benchmark is set at a frequency of 30Hz. k 2 represents the droop factor in the low-frequency region, the minimum operating frequency. f minIt is 25Hz.

[0058] Inverter efficiency Represented as:

[0059] .

[0060] in, To maximize the efficiency of the frequency converter, μ This represents the low-frequency attenuation coefficient.

[0061] Fan shaft power for:

[0062] .

[0063] in, For real-time air density, Standard air density, This refers to the rated shaft power of the fan. For the open... k Wind turbine shaft power Summing up yields the total power consumption of the cooling tower fan group. .

[0064] Optionally, in step S121, the total heat to be discharged by the chiller unit is obtained through the following method:

[0065] The processor calculates the actual heat dissipation of the cooling water based on the collected data of the chiller's cooling water circulation flow rate, cooling water inlet temperature, cooling water outlet temperature, and the water heat dissipation formula.

[0066] The processor uses the actual heat dissipation of the cooling water as the total heat that the chiller unit needs to remove.

[0067] Here, the cooling water circulation flow rate reflects the amount of water participating in heat exchange per unit time, and the difference between the cooling water inlet temperature and the outlet temperature reflects the amount of heat carried away by the cooling water during circulation. Then, the actual heat dissipation of the cooling water is calculated using the isobaric specific heat capacity of water. Q reject The calculation formula is: Q reject = C p × G w ×( T in -T out ).in, T in This refers to the inlet temperature of the cooling water. T out This refers to the outlet temperature of the cooling water. G w This refers to the cooling water circulation flow rate.C p This refers to the specific heat capacity of water at constant pressure. The condensation heat generated by the chiller during operation is entirely transferred to the cooling water. The cooling water, through circulation, carries this heat to the cooling tower and dissipates it outdoors. Therefore, the actual heat dissipation carried away by the cooling water during circulation is consistent with the total heat that the chiller needs to remove during operation. The calculated actual heat dissipation of the cooling water... Q reject The total heat that the chiller unit needs to remove Q load .

[0068] For example, the specific heat capacity of water at constant pressure C p =4.186kJ / (kg ℃), cooling water circulation flow rate G w =100m 3 / h, cooling water inlet temperature T in =35℃, outlet water temperature T out =30℃, then the actual heat dissipation of cooling water Q reject =2093000kJ / h=581.4kW.

[0069] Optionally, in step S102, the processor sets corresponding operating constraints, including:

[0070] The processor sets heat dissipation requirement constraints to ensure that the actual heat dissipation of the cooling tower is greater than or equal to the sum of the total heat to be dissipated by the chiller unit and the preset heat dissipation safety margin; and sets fan frequency constraints to ensure that the operating frequency of a single fan is between the minimum operating frequency and the rated frequency; and sets constraints on the number of fans that can be turned on to ensure that the number of fans turned on is a positive integer between 1 and the total number of fans.

[0071] Here, operational constraints include heat dissipation demand constraints, fan frequency constraints, and constraints on the number of fans in operation. These three constraints work together and are indispensable, jointly ensuring the safe and efficient operation of the system. The cooling tower dissipates the condensation heat generated by the chiller unit to the outside. If the actual heat dissipation of the cooling tower is less than the total heat that the chiller unit needs to remove, it will lead to increased condensing pressure, decreased cooling efficiency, and even system failure and shutdown. Therefore, setting heat dissipation demand constraints ensures the normal operation of the system. Among these constraints, a preset heat dissipation safety margin is included. Q safety This is a preset value, which can be a preset percentage of the total heat that the chiller unit needs to dissipate. For example, a preset heat dissipation safety margin. Q safety = Q reject ×8%.

[0072] The operating frequency of a fan determines its airflow, power consumption, and operational stability. If the fan's operating frequency is lower than the minimum operating frequency... f min This can lead to insufficient airflow from the fan and a decrease in the cooling tower's heat dissipation capacity. Simultaneously, the fan is prone to vibration, excessive noise, and increased motor wear due to idling, shortening its lifespan. If the fan's operating frequency exceeds its rated frequency... f max This can lead to overload operation of the wind turbine, increased losses in the motor and frequency converter, and even equipment overload failures, posing safety hazards. Therefore, wind turbine frequency constraints... f min < f < f max This establishes a safety boundary for the operating frequency of a single wind turbine, ensuring stable operation within its design range. It avoids both insufficient heat dissipation and equipment wear associated with low-frequency operation, and mitigates the safety risks of high-frequency operation, while simultaneously balancing wind turbine efficiency and energy consumption control. For example, f min =25Hz, f max =50Hz.

[0073] Cooling tower fan groups typically consist of multiple fans connected in parallel. The number of fans operating affects the total airflow and heat dissipation capacity of the cooling tower, as well as the total power consumption of the fan group. If the number of fans operating is zero, the cooling tower has no heat dissipation capacity, leading to a rapid shutdown of the chiller unit. If the number of fans operating exceeds the total number of fans, it constitutes an invalid control that cannot be implemented. If the number of fans operating is a non-positive integer, it does not conform to the actual equipment control logic. Therefore, the constraint on the number of fans operating clearly defines the legal range of the number of fans operating, ensuring that the optimized number of fans operating is engineering-feasible and avoiding invalid control or implementation failures.

[0074] Combination Figure 3 As shown, this disclosure provides another control method for a cooling tower fan group of intelligent devices, including:

[0075] S201, the sensor collects the operating parameters of the chiller unit, the operating parameters of the cooling tower fan group, and the outdoor ambient temperature and humidity.

[0076] S202, based on the collected operating parameters and outdoor ambient temperature and humidity, the processor constructs a global energy consumption cost function with the goal of minimizing the total power consumption of the intelligent device, using the number of cooling tower fans in operation, the operating frequency of a single fan, and the dynamic target outlet water temperature as coupled optimization variables, and sets corresponding operating constraints.

[0077] S203, the processor performs global optimization on coupled optimization variables based on the global energy consumption cost function and operating constraints, and obtains the optimal number of cooling tower fans to be turned on, the optimal operating frequency and the optimal target outlet water temperature that meet the operating constraints.

[0078] S204, the processor controls the cooling tower fan group to operate at the optimal number of fans and the optimal operating frequency.

[0079] In this embodiment of the disclosure, a dynamic target outlet water temperature is introduced. T setopt The number of cooling tower fans in operation k Single unit operating frequency f The ternary coupled optimization variables are, i.e., the optimization variables are x =( k, f, T setopt This further expands the optimization space and achieves a deeper level of optimization of the system's total energy consumption. In the previous embodiment of binary coupled optimization variables, the target outlet temperature of the cooling water was usually set to a fixed value, failing to achieve coordinated optimization according to changes in operating conditions. However, in actual operation, the target outlet temperature of the cooling water directly affects the distribution relationship between the power consumption of the chiller unit and the power consumption of the cooling tower fan. When the target outlet temperature is too low, the fan power consumption increases, but the chiller unit's condensing pressure decreases and its power consumption decreases; when the target outlet temperature is too high, the fan power consumption decreases, but the chiller unit's power consumption increases. Therefore, this embodiment introduces the dynamic target outlet temperature as an independent optimization variable, which, together with the number of cooling tower fans in operation and the operating frequency of a single fan, constitutes a ternary coupled optimization variable group.

[0080] The optimal solution obtained through optimization simultaneously includes the optimal target outlet water temperature, the optimal number of fans to be activated, and the optimal operating frequency of a single fan. These three factors are matched and optimized synergistically to minimize the total power consumption of the system while satisfying heat dissipation constraints. The optimal target outlet water temperature is a dynamically changing value, updated in real time according to the chiller load, outdoor ambient temperature and humidity, and other operating conditions. It is not a fixed setpoint, thus enabling the system to maintain a globally optimal operating state across the entire operating range, further tapping the system's energy-saving potential and achieving control effects superior to traditional binary optimization schemes.

[0081] Optionally, in step S202, the dynamic target outlet water temperature is determined in the following way:

[0082] The processor uses the chiller load rate and outdoor wet-bulb temperature under the current operating conditions as input parameters to fit the corresponding best approximation.

[0083] The processor uses the sum of the outdoor wet-bulb temperature and the best approximation as the dynamic target outlet water temperature.

[0084] As mentioned earlier, the outdoor wet-bulb temperature characterizes the heat dissipation capacity of the outdoor environment and affects the heat exchange efficiency of the cooling tower. The lower the outdoor wet-bulb temperature, the stronger the heat dissipation capacity of the cooling tower, and the lower the achievable cooling water outlet temperature. Conversely, the higher the outdoor wet-bulb temperature, the weaker the heat dissipation capacity of the cooling tower, and the cooling water outlet temperature needs to be increased accordingly.

[0085] The optimal approximation degree refers to the reasonable difference between the cooling water outlet temperature and the outdoor wet-bulb temperature that the cooling tower can achieve under the current load rate and outdoor wet-bulb temperature conditions, balancing heat dissipation requirements and energy consumption. By collecting operating data under different conditions (different load rates and different outdoor wet-bulb temperatures), a mapping relationship between the chiller unit load rate, outdoor wet-bulb temperature, and optimal approximation degree is established using polynomial fitting or machine learning fitting. This ensures that the optimal approximation degree accurately matches the current operating conditions, avoiding unreasonable water temperature settings caused by a fixed approximation degree. For example, when the chiller unit load rate is high and the outdoor wet-bulb temperature is low, the system's heat dissipation demand is high and the environment's heat dissipation capacity is strong; the optimal approximation degree can be appropriately reduced to make the dynamic target outlet water temperature closer to the outdoor wet-bulb temperature, satisfying heat dissipation requirements while reducing fan power consumption. When the load rate is low and the outdoor wet-bulb temperature is high, the optimal approximation degree can be appropriately increased to avoid energy waste caused by excessive fan operation.

[0086] The minimum achievable outlet water temperature of a cooling tower is theoretically infinitely close to the outdoor wet-bulb temperature. Therefore, the dynamic target outlet water temperature needs to be based on the outdoor wet-bulb temperature, with a reasonable optimal approximation factor added. This ensures that the cooling tower can achieve the target water temperature, avoiding setting it too low and causing the fan to operate ineffectively, while also minimizing the target water temperature to reduce the power consumption of the chiller unit.

[0087] For example, outdoor wet-bulb temperature T wb Cooling water outlet temperature T out Compared with outdoor wet-bulb temperature T wb The difference is called the approximation degree. .definition The best approximation at any given time is Load factor and wet-bulb temperature T wb The nonlinear function is:

[0088] .

[0089] in, These are the system characteristic coefficients obtained by fitting historical operating data. From these, the dynamic target outlet water temperature is obtained. .

[0090] Combination Figure 4Optionally, in steps S103 and S203, the processor performs global optimization on the coupled optimization variables based on the global energy consumption cost function and operating constraints, including:

[0091] The S310 processor uses a trained radial basis function surrogate model to replace the global energy cost function.

[0092] The S320 processor utilizes a trained radial basis function surrogate model to assist a multi-objective genetic algorithm in global optimization.

[0093] Here, a radial basis function (RBF) surrogate model is used to replace the global energy cost function, and a multi-objective genetic algorithm is combined to complete the optimization. This solves the technical problems of low computational efficiency and poor real-time performance when directly using the global energy cost function, i.e., the real physical model, for optimization in existing technologies, which cannot meet the real-time control requirements of wind turbine groups, thus ensuring that global optimization is both accurate and efficient.

[0094] As discussed above, the global energy cost function (i.e., the true physical model) is calculated using complex thermodynamic formulas and fitting models. Its calculation requires multiple parameters, including cooling water circulation flow rate, inlet and outlet water temperatures, and condensation temperature. While accurate, it suffers from high computational complexity and time consumption. Directly using this function for global optimization necessitates performing a complete physical model calculation for each set of candidate optimization variables, resulting in extremely slow optimization speeds that cannot meet the real-time control requirements of cooling tower fan groups (real-time control requires optimization to be completed within seconds or tens of seconds). In contrast, the radial basis function (RBF) surrogate model is a high-precision, fast-response approximation model. After training with a large amount of pre-collected operating data, it can quickly output a predicted total system power consumption value highly consistent with the true physical model, and its calculation speed is several orders of magnitude faster than the true physical model. Therefore, this embodiment uses a trained radial basis function surrogate model to replace the global energy cost function. This significantly improves the computational efficiency of the optimization process while ensuring the accuracy of power consumption prediction, ensuring that global optimization can respond to changes in operating conditions in real time and providing support for the real-time control of the fan group.

[0095] Specifically, the system energy consumption function is approximated using the RBFN (Radial Basis Function Network) model, and a mapping relationship between input variables and system energy consumption is established using historical operating data and real-time sampled data.

[0096] .

[0097] This model can predict the total energy consumption of the system, thus replacing the highly complex physical model in the genetic algorithm's fitness calculation. The surrogate model uses a radial basis function network prediction function in the form of:

[0098] .

[0099] in, w j These are the weighting coefficients. c j Centered on the basis functions These are radial basis functions.

[0100] It should be noted that the radial basis function neural network model, referred to as the radial basis function surrogate model below, is not independent of the global energy cost function. Instead, it uses the calculation result of the global energy cost function as the training label and is an approximate model established through data fitting. The prediction accuracy of this surrogate model can meet the optimization requirements, ensuring that the optimization results are consistent with the results obtained by directly using the real physical model, while completely solving the problem of insufficient real-time performance.

[0101] Multi-objective genetic algorithms (MAGs) are intelligent algorithms suitable for handling multi-constraint, multi-variable global optimization problems. They can efficiently search for the global optimum within a complex optimization space, avoiding getting trapped in local optima. By using a radial basis function (RBF) surrogate model to assist the MAG in optimization, during the iterative evaluation process, each set of candidate optimization variables does not need to be substituted into complex global energy cost function calculations. Only the trained RDF surrogate model needs to be input to quickly obtain the predicted total power consumption of the system and the deviation between the cooling water outlet temperature and the target outlet temperature. Based on these rapidly obtained evaluation results, the MAG completes iterative steps such as individual sorting, selection, crossover, and mutation, gradually approaching the global optimum. The combination of the RDF surrogate model and the MAG leverages the advantages of the RDF surrogate model—fast computation and high accuracy—solving the real-time optimization problem, while also utilizing the strong global search capability and adaptability to multiple constraints and variables of the MAG, ensuring the global optimum of the optimization result. This ensures that the entire optimization process is efficient and accurate.

[0102] In this embodiment, the radial basis function surrogate model uses the calculation result of the global energy cost function as the training label. After training with a large amount of operating data, the prediction accuracy can be highly consistent with the real physical model, avoiding the deviation of the optimization result caused by the insufficient accuracy of the approximate model, and ensuring that the optimal solution searched by the multi-objective genetic algorithm is found. The multi-objective genetic algorithm can handle multivariate coupled optimization problems, and can flexibly adapt to operating constraints, avoiding getting trapped in local optima, and ensuring that a global optimal solution can be found even under complex operating conditions.

[0103] Optionally, S310, the trained radial basis function surrogate model is obtained in the following way:

[0104] Based on collected historical and real-time operating data, the processor constructs an initial training sample set that includes the number of fans in operation, the operating frequency of a single fan, the outdoor wet-bulb temperature, the chiller unit load rate, the cooling water circulation flow rate, and the corresponding total power consumption of real intelligent devices.

[0105] The processor initializes the basis function centers and weight coefficients of the radial basis function network.

[0106] The processor uses the initial training sample set to train the initialized radial basis function surrogate model, thus obtaining the trained radial basis function surrogate model.

[0107] Here, a sample set is constructed by collecting actual operational data, and the network is trained using this sample set to obtain a high-precision, high-generalization radial basis function surrogate model. The trained model is used to quickly replace the computationally complex global energy consumption cost function in global optimization, significantly improving optimization efficiency while ensuring energy consumption prediction accuracy.

[0108] Specifically, when the intelligent device is a HVAC system, historical operating data of the HVAC system under different seasons, load rates, and outdoor environmental conditions are acquired, along with the latest operating parameters collected under the current operating conditions, to construct a sample set covering the entire operating range. Each sample uses the number of fans in operation, the frequency of operation of a single fan, the outdoor wet-bulb temperature, the chiller load rate, and the cooling water circulation flow rate as input features, and the total power consumption of the actual intelligent device calculated through a global energy consumption cost function as the output label, thereby enabling the surrogate model to accurately learn the nonlinear mapping relationship between the input variables and the total system energy consumption.

[0109] Before model training begins, the structural parameters of the radial basis function network are initialized, including the center position and width of the basis functions, as well as the weight coefficients of the network output layer. Initialization methods can include random initialization, cluster initialization, or initialization based on sample distribution, so that the basis functions can initially cover the sample space, providing an initial structure for subsequent iterative optimization and avoiding training from getting trapped in local minima.

[0110] A radial basis function surrogate model is trained using an initial training sample set. Through iterative training, the center, width, and weight coefficients of the basis functions are continuously adjusted, gradually approximating the true energy consumption value to obtain the trained model. For example, when the prediction error converges to within a preset threshold, training is complete, and the trained radial basis function surrogate model is obtained. The trained model can quickly and accurately predict the total system energy consumption corresponding to any set of optimization variables, thus replacing the global energy cost function in the multi-objective genetic algorithm optimization process and achieving efficient global optimization.

[0111] Combination Figure 5 Optionally, in step S320, the processor uses the trained radial basis function surrogate model to assist the multi-objective genetic algorithm in global optimization, including:

[0112] S321, the processor uses the trained radial basis function surrogate model to evaluate individuals in the initial population in order to select superior individuals.

[0113] S322, the processor performs crossover and mutation on superior individuals to generate a new generation of candidate solutions.

[0114] S323, the processor, based on the new generation of candidate solution population, determines whether the Pareto optimal front obtained in the current iteration and the new generation of candidate solution population satisfy the convergence condition.

[0115] S324 If not satisfied, the processor selects the key solution with the minimum predicted energy consumption and the maximum uncertainty, calculates the total energy consumption using the global energy consumption cost function, and updates the key solution and the corresponding total energy consumption to the sample set of the radial basis function surrogate model to retrain the radial basis function surrogate model and continue iterating.

[0116] If S325 is satisfied, the processor outputs the Pareto optimal solution set.

[0117] Here, the complete iterative process of optimization includes initial population evaluation, generation of the next generation population, convergence judgment, and branch processing, ensuring that global optimization is both efficient and accurate. Simultaneously, it achieves dynamic updates of the radial basis function surrogate model, further improving subsequent optimization accuracy. The initial population of the multi-objective genetic algorithm consists of candidate control schemes randomly generated from multiple sets of coupled optimization variables, with each individual in the population corresponding to a specific set of candidate optimization variables. The optimization variables corresponding to each individual are input into the trained radial basis function surrogate model, which quickly outputs the predicted total system power consumption value for that individual. Using the predicted total power consumption value as the basis for individual evaluation, individuals with lower total power consumption and meeting the operational constraints are selected as superior individuals. These superior individuals are used to provide high-quality parent individuals for subsequent crossover and mutation operations, ensuring the optimization potential of the next generation population.

[0118] After performing crossover and mutation operations on superior individuals, all generated offspring individuals are aggregated to form a new generation of candidate solutions. The size of the new generation population remains consistent with the initial population to ensure the stability and continuity of the optimization process. The crossover operation involves randomly selecting two sets of individuals from the selected superior individuals as parents, and then performing crossover substitution on the corresponding coupled optimization variables of the parent individuals to generate two sets of offspring individuals, allowing the offspring individuals to inherit the superior characteristics of their parents. The mutation operation involves randomly making small modifications to the offspring individuals generated by the crossover operation to prevent the population from getting trapped in local optima, increasing the diversity of the population, and ensuring that the optimization process can search a wider optimization space.

[0119] Then, by comparing the Pareto optimal front of the current iteration with the overall characteristics of the next generation of candidate solutions, it is determined whether the optimization process has stabilized and whether further iterations are needed. The Pareto optimal front refers to the set of all superior solutions in the current iteration that cannot be dominated by other solutions. The optimization schemes corresponding to the individuals in this set, under the premise of satisfying the operational constraints, cannot further reduce the total power consumption of the system by adjusting the optimization variables, and simultaneously satisfy the water temperature deviation requirement; these are the optimal solutions for the current iteration. Convergence conditions are used to determine whether the optimization process has reached stability and whether a global optimum has been found.

[0120] When the optimization fails to meet the convergence condition, it indicates that the Pareto optimal frontier of the current iteration still has room for optimization, and the prediction accuracy of the radial basis function surrogate model can be further improved. At this point, two types of key solutions are selected: first, the individual with the minimum predicted energy consumption, which is the candidate solution with the greatest optimization potential; second, the individual with the highest prediction uncertainty, i.e., the candidate solution where the surrogate model's prediction error may be large. These two types of key solutions are substituted into the global energy cost function to calculate the accurate total system energy consumption. These key solutions (corresponding optimization variables) and the corresponding total energy consumption data are then updated to the training sample set of the radial basis function surrogate model. The surrogate model is retrained based on the updated sample set to improve the model's prediction accuracy. This entire iterative process is repeated until the convergence condition is met, ensuring both optimization speed and improved optimization accuracy through dynamic model updates. The updated training sample set... Initial training sample set , x i For individuals, Let i be the total energy consumption of the i-th individual; x new For new individuals, This represents the total energy consumption corresponding to the new individual.

[0121] When the Pareto optimal front obtained in the current iteration and the new generation of candidate solutions meet the convergence condition, it indicates that the optimization process has stabilized and the global optimal solution set under the current operating conditions, i.e., the Pareto optimal solution set, has been obtained. This solution set contains multiple sets of optimally coupled optimization variables, each of which satisfies the operating constraints and cannot further reduce the total power consumption of the system.

[0122] Optionally, in step S321, the initial population is obtained in the following way:

[0123] The processor determines the preset size of the initial population and randomly generates multiple sets of candidate control schemes, including the number of cooling tower fans that are turned on and the operating frequency of a single fan, to form the initial population.

[0124] In this context, all individuals in the initial population satisfy the preset operational constraints; each individual in the initial population corresponds to a set of data combinations of the number of blowers started and the operating frequency of a single blower; if the coupled optimization variables also include the dynamic target effluent temperature, then each individual corresponds to a set of data combinations of the number of blowers started, the operating frequency of a single blower, and the dynamic target effluent temperature.

[0125] Here, the initial population size is a fixed value set in advance based on the actual engineering situation, optimization accuracy requirements, and processor computing power. An excessively large initial population size can increase the controller's computational load and reduce the optimization speed, while an excessively small size can lead to insufficient population diversity and cause the optimization to get stuck in local optima. For example, the initial population size can be set to 50 to 200 groups. Candidate control schemes are randomly generated combinations of the number of cooling tower fans in operation and their operating frequency, or combinations of the number of cooling tower fans in operation, their operating frequency, and the dynamic target outlet water temperature. Each combination corresponds to a fan group operation mode. Multiple randomly generated combinations are aggregated to form the initial population.

[0126] All individuals in the initial population satisfy the preset operational constraints. That is, all randomly generated candidate control schemes must be constrained in advance to eliminate invalid combinations that do not meet the operational constraints (such as the fan frequency being lower than the minimum operating frequency, the number of fans being turned on exceeding the total number of fans, etc.), to ensure that each individual in the initial population is feasible for engineering purposes and to avoid invalid individuals consuming computing resources and affecting optimization efficiency.

[0127] Optionally, in step S321, the processor evaluates individuals in the initial population using the trained radial basis function surrogate model to select superior individuals, including:

[0128] The processor outputs the bi-objective function value for each individual based on the trained radial basis function surrogate model; the bi-objective function includes the objective function of minimizing total energy consumption and the objective function of minimizing the deviation between the actual outlet water temperature and the target outlet water temperature.

[0129] The processor performs a non-dominated sort on all individuals, divides the individuals into different superiority and inferiority levels, and calculates the crowding distance of each individual within the same level; among them, the individuals in the first level are Pareto optimal individuals with no other individuals to dominate them.

[0130] The processor selects a preset number of individuals as superior individuals according to the rules of hierarchical priority and the crowding distance of individuals at the same level from large to small.

[0131] Here, the dual objective functions include a total energy consumption minimization objective function and a deviation between the actual and target cooling water outlet temperatures. For each individual in the initial population, the trained radial basis function surrogate model synchronously outputs two mutually cooperating and constraining objective function values. The smaller the total energy consumption minimization objective function value predicted by the radial basis function surrogate model, the better the energy efficiency of the candidate solution. The smaller the deviation between the actual and target cooling water outlet temperatures predicted by the radial basis function surrogate model, the higher the water temperature control accuracy of the candidate solution, better meeting the system's heat dissipation requirements and ensuring stable operation of the chiller unit. Specifically, the target outlet temperature is a fixed target temperature in binary optimization and a dynamic target outlet temperature in ternary optimization. The optimization objective function can be expressed as: .in, Let the total energy consumption objective function be... Indicates the outlet temperature of the cooling water With the target outlet water temperature The deviation between them is a system stability indicator.

[0132] All individuals are hierarchically divided and crowding distances are calculated. Non-dominated ranking is a method for selecting superior individuals in multi-objective optimization. If an individual's two objective function values ​​are both no worse than another individual's, and at least one objective function value is better than the first individual's, then the first individual dominates the second individual. Through non-dominated ranking, all individuals in the initial population are divided into different hierarchical levels of superiority and inferiority. The higher the level, the better the individual's overall performance. Individuals in the first level are Pareto optimal individuals with no other individuals to dominate them; that is, each individual in this level is the individual with the best overall performance in the current population, and no other individual can simultaneously outperform individuals in the first level in both minimizing total energy consumption and minimizing water temperature deviation. The higher the level, the worse the individual's overall performance, and there is a possibility of being dominated by individuals in higher levels. At the same time, to avoid over-concentration of individuals within the same level and to ensure population diversity, the crowding distance of each individual within the same level needs to be calculated. The larger the crowding distance, the stronger the individual's uniqueness within that level, and the lower the possibility of being replaced by other individuals, thus better preserving population diversity and avoiding getting trapped in local optima in subsequent optimization. Among them, the crowding distance is... d i for:

[0133] .

[0134] in, d i This represents the crowding distance of the i-th individual, used to measure the local density of that individual in the target space. When d i A larger value indicates fewer solutions near the individual, which helps maintain a uniform distribution of the solution set in the objective space. M represents the number of objective functions in the optimization problem. f m Let m represent the m-th objective function. When calculating the crowding distance, the population needs to be sorted along each objective function dimension, and then the difference in objective functions between adjacent individuals is calculated. f m ( i+1 Let represent the objective function value corresponding to the (i+1)th individual after sorting the m-th objective function. f m ( i-1 ) represents the objective function value of the (i-1)th individual after sorting. These represent the maximum and minimum values ​​of the m-th objective function in the current population, respectively. This term is used to normalize the difference in objective functions to eliminate the influence of differences in the dimensions of different objective functions on the calculation of congestion distance. In this embodiment, there are two objective functions: the total energy consumption minimization objective function mentioned above and the objective function for minimizing the deviation between the actual and target outlet water temperatures.

[0135] When selecting superior individuals, priority is given to those at the top of the tier to ensure the best overall performance. Individuals with good energy efficiency and small water temperature deviations are prioritized for retention. Within the same tier, individuals with large crowding distances are prioritized to preserve population diversity and provide abundant parent individuals for subsequent crossover and mutation operations, preventing the optimization process from getting stuck in local optima. The preset number is a fixed value set in advance based on the initial population size and optimization accuracy requirements. For example, if the initial population size is 100 groups, the preset selection number is 50 groups.

[0136] Optionally, in step S322, the processor performs crossover and mutation on superior individuals to generate a new generation of candidate solution population, including:

[0137] The processor randomly selects two sets of individuals from the selected superior individuals as parents, and cross-replaces the parameters corresponding to the parent individuals to generate two sets of offspring individuals.

[0138] The processor performs random, small-amplitude adjustments on the offspring individuals generated by crossover; wherein the adjustment amplitude is less than a first threshold and the parameters of the offspring individuals after adjustment still satisfy the operating constraints.

[0139] The processor aggregates all offspring individuals generated by crossover and mutation to form a new generation of candidate solution population; the size of the new generation of candidate solution population remains the same as the initial population.

[0140] Here, cross-substitution involves replacing the corresponding optimization parameters of the two sets of parent individuals. This integrates the advantageous parameters of the parent individuals, achieving the inheritance of superior characteristics and enhancing the optimization potential of the next generation. If it's a binary coupled variable optimization, where the optimization variables are the number of fans in operation and the operating frequency per fan, then the number of fans in operation and the operating frequency per fan in the two sets of parent individuals are cross-substituted to generate two sets of offspring individuals. The parameter combinations of the offspring individuals inherit the superior parameter characteristics of the parents. If it's a ternary coupled variable optimization, where the optimization variables also include the dynamic target outlet water temperature, then the number of fans in operation, the operating frequency per fan, and the dynamic target outlet water temperature are all cross-substituted simultaneously for the two sets of parent individuals, generating two sets of offspring individuals. This ensures that the offspring individuals simultaneously inherit the superior characteristics of the parents in energy consumption and water temperature control.

[0141] Mutation operations are used to avoid getting stuck in local optima. By randomly and slightly modifying the offspring individuals generated through crossover, the diversity of the population is increased. This ensures that the optimization process can explore a wider optimization space and avoids the limitations caused by over-concentration of individuals in the population. The first threshold is a fixed value set according to the actual engineering requirements and parameter adjustment precision needs. For example, the fan frequency correction threshold is ≤2Hz, the dynamic target outlet water temperature correction threshold is ≤0.5℃, and the fan operation number correction threshold is 0. Since the number of fans is a positive integer, small modifications do not change the number of fans. The correction magnitude is strictly controlled within the first threshold to ensure that the parameters of the offspring individuals do not fluctuate significantly and retain the excellent characteristics of the parent generation. Simultaneously, after modification, the offspring individuals need to be constrained and verified to ensure that the modified parameters still meet the aforementioned operational constraints, namely, heat dissipation requirement constraints, fan frequency constraints, and fan operation number constraints. Invalid individuals that do not meet the constraints after modification are removed to ensure that all offspring individuals are feasible for engineering purposes.

[0142] Repeat the parent selection, crossover, and mutation operations described above to continuously generate effective offspring individuals until the number of offspring individuals reaches the preset size of the initial population. All effective offspring individuals are then aggregated to form a new generation of candidate solutions. In this way, the aggregated new generation of candidate solutions inherits the energy efficiency and water temperature stability of the superior parent individuals, while also increasing population diversity through mutation.

[0143] Optionally, in step S323, the convergence conditions include:

[0144] Compared to the Pareto optimal frontier of the previous iteration, the average change in the biobjective function value of the current iteration is less than the second threshold, and this condition is satisfied for multiple consecutive iterations. And / or,

[0145] The improvement in the biobjective function values ​​of the new generation of candidate solutions is less than the third threshold. And / or,

[0146] The algorithm has reached the preset maximum number of iterations.

[0147] Here, by setting multi-dimensional and quantifiable convergence conditions, the algorithm balances optimization accuracy, computational efficiency, and engineering practicality. This ensures that the algorithm terminates iterations promptly when it finds a stable and reliable global optimal solution, avoiding ineffective iterations that waste computational resources. Simultaneously, it guarantees sufficient convergence of the optimization results, meeting the requirements for real-time optimization control of cooling tower fan groups. The Pareto optimal front reflects the set of individuals with the best overall performance in the current iteration. By comparing the Pareto optimal front of the current iteration with that of the previous iteration, the average change of the objective function values ​​for minimizing total energy consumption and minimizing the deviation between the actual and target cooling water outlet temperatures is calculated. When this average change is less than a preset second threshold, and this state is maintained for multiple consecutive iterations, it indicates that the Pareto optimal front has stabilized, and the optimization objective function value no longer changes significantly. Therefore, it is determined that the algorithm has found a stable set of optimal solutions.

[0148] The improvement in the dual objective function values ​​of the new generation of candidate solutions is less than the third threshold, which is used to judge the optimization effect from the perspective of the entire population. If the improvement in both total energy consumption and water temperature deviation for all individuals in the new generation of candidate solutions is less than the preset third threshold, it indicates that further iteration is difficult to optimize the control scheme, the algorithm search space has been fully traversed, the optimization effect tends to saturate, and the convergence requirement is met. To ensure the real-time performance of the control system and avoid the algorithm iterating indefinitely due to difficulty in convergence, a maximum number of iterations is set as a fallback convergence condition. When the number of algorithm iterations reaches the preset maximum number of iterations, the iteration is forcibly terminated regardless of whether the aforementioned convergence condition is met, to ensure that the optimization process is completed within the specified time and to meet the response speed requirements of real-time control of the cooling tower fan group.

[0149] The above convergence conditions can be satisfied individually or in combination. A comprehensive judgment is made based on preset logic to achieve efficient and stable iterative termination while ensuring optimization accuracy.

[0150] Optionally, after the processor outputs the Pareto optimal solution set, S325 also includes:

[0151] The processor selects the individual with the minimum total energy consumption and the deviation between the actual outlet water temperature and the target outlet water temperature from the Pareto optimal solution set. The parameters corresponding to this individual are used as the optimal number of units to be turned on and the optimal operating frequency.

[0152] When the coupling optimization variables also include dynamic target outlet water temperature, the processor synchronously obtains the optimal target outlet water temperature.

[0153] Here, all individuals in the Pareto optimal solution set have satisfied the aforementioned operational constraints, including heat dissipation demand constraints, fan frequency constraints, and fan operating number constraints, and are all engineering-feasible candidate solutions. Based on this, to further determine the set of control parameters with optimal overall performance, this embodiment prioritizes minimizing total energy consumption while ensuring that the deviation between the actual cooling water outlet temperature and the target outlet temperature is within an allowable range, avoiding excessive temperature deviation from affecting the chiller's heat exchange effect and operational stability. Specifically, the individual with the minimum total energy consumption and whose actual cooling water outlet temperature deviation does not exceed a preset deviation threshold is selected from the Pareto optimal solution set. The corresponding fan operating number and single-fan operating frequency are then used as the optimal operating number and optimal operating frequency.

[0154] In a ternary optimization scenario where the coupled optimization variables also include the dynamic target outlet water temperature, after selecting the individual with the minimum total energy consumption and satisfactory water temperature deviation, the optimal target outlet water temperature corresponding to that individual is obtained simultaneously. This yields a set of mutually matched and synergistically optimal control parameters, including the optimal number of units activated, the optimal operating frequency, and the optimal target outlet water temperature. This enables the system to minimize the total power consumption of the intelligent devices while meeting heat dissipation requirements, ensuring safe equipment operation, and maintaining stable water temperature.

[0155] Optionally, in steps S104 and S204, the processor controls the cooling tower fan group to operate at the optimal number of fans in operation and the optimal operating frequency, including:

[0156] When the current number of operating fans and their operating frequency deviate from the number of fans in operation and the optimal operating frequency, the processor adjusts the fan operating frequency with a preset step size.

[0157] After the processor reaches the preset threshold after adjusting the operating frequency, it adjusts the number of fans to be turned on through a delayed start-stop mechanism.

[0158] The processor gradually brings the wind turbine's operating state closer to the optimal operating state over multiple control cycles.

[0159] Here, through the coordinated control of step-by-step frequency adjustment and unit number delay adjustment, equipment shocks and system fluctuations caused by sudden parameter changes are avoided, ensuring that the wind turbine group smoothly and efficiently transitions to the optimal operating state, while taking into account both equipment safety and system operation stability.

[0160] The current operating status of the fans is compared with the optimal status to determine if there is any deviation. If the current number of operating fans and the operating frequency are completely consistent with the optimal parameters, the current operating status is maintained. If a deviation exists, the fan operating frequency is adjusted first, rather than directly adjusting the number of fans in operation. The frequency is adjusted gradually with a preset step size to avoid problems such as sudden changes in fan airflow, motor impact, and excessive fluctuations in system water temperature caused by abrupt frequency changes. This ensures stable fan operation while gradually bringing the cooling water outlet temperature closer to the target outlet temperature, balancing system stability and adjustment efficiency.

[0161] After the adjusted operating frequency reaches a preset threshold, the number of fans in operation is adjusted through a delayed start-stop mechanism. The preset threshold is a critical frequency value set based on the rated parameters of the fan equipment and the system's heat dissipation requirements. For example, a minimum operating frequency of 5Hz and a rated frequency of 50Hz correspond to a preset threshold of 5Hz or 48Hz. If the current operating frequency has been adjusted to the lower limit of the preset threshold (e.g., 5Hz) but has not yet reached the optimal operating frequency, and the current number of fans in operation exceeds the optimal number, it indicates that the number of fans in operation needs to be reduced. In this case, the delayed start-stop mechanism is activated, and after a preset delay (e.g., 30-60 seconds), the excess fans are shut down. This avoids insufficient heat dissipation and sudden changes in water temperature caused by a sudden reduction in the number of fans. If the current operating frequency has been adjusted to the upper limit of the preset threshold (e.g., 48Hz) but has not yet reached the optimal operating frequency, and the current number of fans in operation is less than the optimal number, it indicates that the number of fans in operation needs to be increased. In this case, the delayed start-stop mechanism is activated, and after a preset delay, the newly added fans are started. This avoids motor overload and a sudden increase in system energy consumption caused by a sudden increase in the number of fans. The delayed start-stop mechanism provides a buffer time for the system, allowing the cooling water temperature and system heat dissipation status to adapt to changes after frequency adjustment, avoiding system fluctuations caused by the superposition of unit number adjustment and frequency adjustment, and ensuring equipment safety and system stability.

[0162] The wind turbine operating state is gradually brought closer to the optimal operating state over multiple control cycles. The adjustment of the wind turbine operating state is not completed all at once, but rather through iterative adjustments over multiple control cycles. Within each control cycle, the frequency is first adjusted by a preset step size to determine if a preset threshold has been reached. If it has, a delayed start-stop mechanism is activated to adjust the number of turbines; if not, the frequency adjustment continues. This process is repeated until the current number of operating turbines and operating frequency are completely consistent with the optimal parameters, achieving a smooth approach to the optimal operating state of the wind turbine group. In this way, the adjustment process is smooth and controllable, avoiding equipment and system shocks caused by sudden parameter changes, and enabling rapid adaptation to optimal control parameters.

[0163] Combination Figure 6As shown, this disclosure provides a control device 100 for a cooling tower fan group of an intelligent device, including a processor 101 and a memory 102. Optionally, the device may further include a communication interface 103 and a bus 104. The processor 101, communication interface 103, and memory 102 can communicate with each other via the bus 104. The communication interface 103 can be used for information transmission. The processor 100 can call logical instructions in the memory 102 to execute the control method for a cooling tower fan group of an intelligent device according to the above embodiments.

[0164] Furthermore, the logical instructions in the aforementioned memory 102 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0165] The memory 102, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this disclosure. The processor 101 executes functional applications and data processing by running the program instructions / modules stored in the memory 102, thereby implementing the control method for the cooling tower fan group of intelligent devices in the above embodiments.

[0166] The memory 102 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 102 may include high-speed random access memory and may also include non-volatile memory.

[0167] This disclosure provides a device, including: a device body, and the aforementioned control device 100 for a cooling tower fan group of an intelligent device. The cooling tower fan group collaborative control device 100 for the intelligent device is installed on the device body. The installation relationship described herein is not limited to placement inside the device, but also includes installation connections with other components of the device, including but not limited to physical connections, electrical connections, or signal transmission connections. Those skilled in the art will understand that the control device for a cooling tower fan group of an intelligent device can be adapted to feasible device bodies to achieve other feasible embodiments. The device may be a heating, ventilation, and air conditioning (HVAC) system.

[0168] This disclosure provides a computer-readable storage medium storing computer-executable instructions configured to execute the above-described control method for cooling tower fan groups of intelligent devices.

[0169] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0170] The technical solutions of this disclosure can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this disclosure. The aforementioned storage medium can be a non-transitory storage medium, including: a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and other media capable of storing program code; it can also be a transient storage medium.

[0171] The foregoing description and accompanying drawings fully illustrate embodiments of this disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terminology used in this application is for describing embodiments only and is not intended to limit the claims. As used in the description of embodiments and claims, the singular forms “a,” “an,” and “the” are intended to equally include the plural forms unless the context clearly indicates otherwise. Similarly, the term “and / or” as used in this application means including one or more of the associated listed items and all possible combinations thereof. Additionally, when used in this application, the term "comprise" and its variations "comprises" and / or "comprising" refer to the presence of stated features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof. Without further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes said element. In this document, each embodiment may focus on the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, the relevant parts can be referred to the description of the method section.

[0172] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this disclosure. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0173] The methods and products (including but not limited to devices and equipment) disclosed in the embodiments herein can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may be merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to implement this embodiment according to actual needs. In addition, the functional units in the embodiments of this disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0174] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

Claims

1. A control method for a cooling tower fan group of intelligent equipment, characterized in that, include: Collect operating parameters of the chiller unit, operating parameters of the cooling tower fan group, and outdoor ambient temperature and humidity; Based on the collected operating parameters and outdoor ambient temperature and humidity, a global energy consumption cost function is constructed with the number of cooling tower fans in operation and the operating frequency of each fan as coupled optimization variables. The function aims to minimize the total power consumption of the intelligent devices, and corresponding operating constraints are set. Among these, the number of cooling tower fans in operation... k With single unit operating frequency f The relationship is a non-linear superposition, and the operating constraints include heat dissipation requirements, fan frequency, and the number of fans in operation; the total power consumption of the intelligent devices is considered. Minimize the global energy cost function J for ， Among them, total power Including chiller unit power consumption Total power consumption of cooling tower fan group Power consumption of chiller units It is the condensation temperature. A strongly nonlinear function; condensation temperature With cooling tower outlet water temperature There is a linear correlation , The condenser heat exchange temperature difference is used; polynomial fitting is employed to obtain the power consumption of the chiller unit. , in, It was obtained based on multiple regression analysis of historical operating data. This represents the total heat that the chiller unit needs to remove. Based on the global energy consumption cost function and operational constraints, global optimization is performed on the coupled optimization variables to obtain the optimal number of cooling tower fans to be turned on and the optimal operating frequency that satisfies the operational constraints. Control the cooling tower fan group to operate at the optimal number of units and the optimal operating frequency.

2. The method according to claim 1, characterized in that, The collection of operating parameters of the chiller unit, the cooling tower fan group, and the outdoor ambient temperature and humidity includes: Collect data on the chiller unit's load rate, cooling water circulation flow rate, cooling water inlet temperature, cooling water outlet temperature, cooling tower fan's rated parameters and minimum operating frequency, as well as outdoor dry-bulb temperature and relative humidity.

3. The method according to claim 1, characterized in that, The construction of the global energy consumption cost function aimed at minimizing the total power consumption of smart devices includes: The power consumption of the chiller unit is constructed by performing a polynomial fitting on the condensing temperature and the total heat to be dissipated by the chiller unit. Based on the cubic law of fan speed, motor efficiency, and frequency converter efficiency, the total power consumption of the cooling tower fan group is calculated. The sum of the power consumption of the chiller unit and the total power consumption of the cooling tower fan group is used as the global energy consumption cost function with the goal of minimizing the total power consumption of intelligent devices.

4. The method according to claim 3, characterized in that, The total heat that the chiller unit needs to remove is obtained through the following methods: Based on the collected data on the cooling water circulation flow rate, cooling water inlet temperature, cooling water outlet temperature of the chiller unit, and the heat dissipation formula of water, the actual heat dissipation of the cooling water is calculated. The actual heat dissipation of the cooling water is taken as the total heat that the chiller unit needs to remove.

5. The method according to claim 1, characterized in that, Set the corresponding runtime constraints, including: Set heat dissipation demand constraints so that the actual heat dissipation of the cooling tower is greater than or equal to the sum of the total heat that the chiller unit needs to remove and the preset heat dissipation safety margin; and, Set fan frequency constraints to ensure that the operating frequency of a single fan is between the minimum operating frequency and the rated frequency; and, Set a constraint on the number of fans to be activated, so that the number of fans to be activated is a positive integer between 1 and the total number of fans.

6. The method according to claim 1, characterized in that, Before performing global optimization on the coupling optimization variables, the following is also included: The dynamic target outlet water temperature has been added as an optimization variable. Global optimization is performed on the coupled optimization variables consisting of dynamic target outlet water temperature, number of cooling tower fans in operation, and operating frequency of a single fan.

7. The method according to claim 6, characterized in that, The dynamic target outlet water temperature is determined in the following way: Using the chiller unit load rate and outdoor wet-bulb temperature under the current operating conditions as input parameters, the corresponding best approximation is obtained by fitting. The sum of the outdoor wet-bulb temperature and the optimal approximation is used as the dynamic target outlet water temperature.

8. The method according to any one of claims 1 to 7, characterized in that, Global optimization of coupling optimization variables includes: The global energy cost function is replaced by a trained radial basis function surrogate model. The trained radial basis function surrogate model is used to assist the multi-objective genetic algorithm in global optimization.

9. The method according to claim 8, characterized in that, The trained radial basis function surrogate model is obtained in the following way: Collect historical and real-time operating data to construct an initial training sample set that includes the number of fans in operation, the operating frequency of a single fan, the outdoor wet-bulb temperature, the chiller unit load rate, the cooling water circulation flow rate, and the corresponding total power consumption of real intelligent devices. Initialize the basis function centers and weight coefficients of the radial basis function network; The initial radial basis function surrogate model is trained using the initial training sample set to obtain the trained radial basis function surrogate model.

10. The method according to claim 8, characterized in that, The trained radial basis function surrogate model is used to assist a multi-objective genetic algorithm in global optimization, including: The trained radial basis function surrogate model is used to evaluate individuals in the initial population in order to select superior individuals; Superior individuals are cross-crossed and mutated to generate a new generation of candidate solutions. Based on the new generation of candidate solution population, determine whether the Pareto optimal front obtained in the current iteration and the new generation of candidate solution population satisfy the convergence condition. If the conditions are not met, the key solution with the minimum predicted energy consumption and the maximum uncertainty is selected. The total energy consumption is calculated using the global energy consumption cost function. The key solution and the corresponding total energy consumption are then updated to the sample set of the radial basis function surrogate model to retrain the radial basis function surrogate model and continue iterating. If satisfied, output the Pareto optimal solution set.

11. The method according to claim 10, characterized in that, The initial population was obtained through the following methods: The initial population size is determined, and multiple candidate control schemes containing the number of cooling tower fans turned on and the operating frequency of a single fan are randomly generated to form the initial population. In this context, all individuals in the initial population satisfy the preset operational constraints; each individual in the initial population corresponds to a set of data combinations of the number of blowers started and the operating frequency of a single blower; if the coupled optimization variables also include the dynamic target effluent temperature, then each individual corresponds to a set of data combinations of the number of blowers started, the operating frequency of a single blower, and the dynamic target effluent temperature.

12. The method according to claim 10, characterized in that, The trained radial basis function surrogate model is used to evaluate individuals in the initial population to select superior individuals, including: The trained radial basis function surrogate model outputs the bi-objective function value for each individual; the bi-objective function includes the objective function of minimizing total energy consumption and the objective function of minimizing the deviation between the actual outlet water temperature and the target outlet water temperature. All individuals are non-dominated and ranked, and then divided into different levels of superiority and inferiority. The crowding distance of each individual in the same level is calculated. The first level of individuals is the Pareto optimal individual with no other individuals to dominate them. Based on the principle of prioritizing hierarchical levels and selecting individuals from those with the largest to the smallest crowding distance within the same level, a predetermined number of individuals are selected as superior individuals.

13. The method according to claim 10, characterized in that, Superior individuals are crossovered and mutated to generate a new generation of candidate solutions, including: Two groups of individuals are randomly selected from the selected superior individuals as parents. The parameters corresponding to the parent individuals are cross-substituted to generate two groups of offspring individuals. The offspring individuals generated by crossover are randomly modified with small magnitudes; wherein the modification magnitude is less than the first threshold, and the parameters of the offspring individuals after modification still meet the operational constraints. All offspring individuals generated by crossover and mutation are aggregated to form a new generation of candidate solution population; the size of the new generation of candidate solution population remains the same as the initial population.

14. The method according to claim 10, characterized in that, Convergence conditions include: Compared to the Pareto optimal frontier of the previous iteration, the average change in the biobjective function value of the current iteration is less than the second threshold, and this condition is satisfied for multiple consecutive iterations; and / or, The improvement in the biobjective function values ​​of the new generation of candidate solutions is less than the third threshold; The algorithm has reached the preset maximum number of iterations.

15. The method according to claim 10, characterized in that, Also includes: Select the individual with the minimum total energy consumption and the deviation between the actual outlet water temperature and the target outlet water temperature from the Pareto optimal solution set. Use the parameters corresponding to this individual as the optimal number of units to be turned on and the optimal operating frequency. When the coupled optimization variables also include dynamic target effluent temperature, the optimal target effluent temperature is obtained simultaneously.

16. The method according to any one of claims 1 to 7, characterized in that, Controlling the cooling tower fan group to operate at the optimal number of fans and the optimal operating frequency includes: If the current number of operating fans and the operating frequency deviate from the number of fans in operation and the optimal operating frequency, respectively, the fan operating frequency is adjusted by a preset step size. After the adjusted operating frequency reaches the preset threshold, the number of fans to be turned on is adjusted through a delayed start-stop mechanism; The fan's operating state is gradually brought closer to the optimal operating state over multiple control cycles.

17. A control device for a cooling tower fan group of intelligent equipment, comprising a processor and a memory storing program instructions, characterized in that, The processor is configured to execute, when running the program instructions, the control method for a cooling tower fan group for a smart device as described in any one of claims 1 to 16.

18. A device, characterized in that, include: Equipment body; The control device for a cooling tower fan group of a smart device as described in claim 17 is installed on the device body.

Images