Intelligent control system for gradient input of cooling tower fan under greenhouse dynamic demand

Abstract

This invention discloses an intelligent control system for the gradient deployment of cooling tower fans under dynamic greenhouse demand. The system includes a multi-source data acquisition unit for collecting operational data from the greenhouse side and the cooling loop side. The operational data includes greenhouse internal temperature, greenhouse internal humidity, leaf surface temperature, light intensity, supplemental lighting equipment operation status, spraying or atomization status, ventilation execution status, circulating water flow rate, return water temperature, and environmental meteorological parameters. This invention relates to the field of greenhouse environmental control technology. This intelligent control system for the gradient deployment of cooling tower fans under dynamic greenhouse demand dynamically calculates the greenhouse's heat and humidity load requirements based on environmental data and a load propagation chain model. This ensures that the cooling tower system can flexibly adjust fan deployment according to changes in demand, thereby achieving precise cooling. Through fan candidate screening, benefit-cost calculation, and marginal net benefit analysis, the system can intelligently select the most suitable fan for deployment, avoiding energy waste.

Description

Technical Field

[0001] This invention relates to the field of greenhouse environmental control, and more specifically, to an intelligent control system for the gradient operation of cooling tower fans under dynamic greenhouse demand. Background Technology

[0002] With the development of greenhouse agriculture, precise control of the greenhouse environment has become crucial for improving crop growth efficiency. Various factors inside the greenhouse, such as temperature, humidity, and light intensity, significantly impact crop growth, thus requiring efficient cooling systems to maintain a suitable environment. However, in traditional cooling tower systems, fan operation is typically adjusted based on fixed conditions, failing to effectively adapt to the dynamic changes in the greenhouse environment, resulting in low cooling efficiency and energy waste.

[0003] Existing technologies generally suffer from the following problems: traditional fan control systems fail to flexibly adjust fan operation according to changes in greenhouse load, resulting in low fan operating efficiency and problems such as excessive energy consumption and unstable cooling effect; there is a lack of real-time monitoring and intelligent response mechanisms for dynamic changes in greenhouse environmental parameters, making it impossible to accurately predict cooling demand, which in turn affects the accuracy and timeliness of system regulation; in cooling towers with multiple fans, the overall cooling performance of the system is constrained due to interference and efficiency coordination issues between fans, and the system energy consumption is difficult to optimize.

[0004] Therefore, there is an urgent need for an intelligent cooling tower fan gradient control system that can accurately regulate the fans based on the dynamic needs of the greenhouse environment, and maximize system efficiency while meeting heat dissipation requirements and reducing energy waste. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent control system for the gradient operation of cooling tower fans under dynamic greenhouse demand. This system solves the problem that in traditional cooling tower systems, the operation of fans is usually adjusted according to fixed conditions, which fails to effectively adapt to the dynamic changes in the greenhouse environment, resulting in low cooling efficiency and energy waste, and failing to meet usage requirements.

[0006] This invention achieves the above objective through the following technical solution: an intelligent control system for the gradient operation of cooling tower fans under dynamic greenhouse demand, the system comprising: The multi-source data acquisition unit is used to collect operating data from the greenhouse side and the cooling circuit side. The operating data includes greenhouse internal temperature, greenhouse internal humidity, leaf surface temperature, light status, supplemental lighting equipment operating status, spraying or atomization status, ventilation execution status, circulating water flow rate, return water temperature, and environmental meteorological parameters. The load model construction unit is used to construct a load propagation chain model based on the operating data to transmit the greenhouse heat and humidity load to the cooling tower heat dissipation demand, and output the demand growth index and cooling gap advance index within a future preset time window according to the load propagation chain model. The wind turbine candidate generation unit is used to generate a set of candidate wind turbines to be put into operation when the demand growth index meets the pre-occupancy conditions. The revenue and cost calculation unit is used to calculate the additional cooling revenue and combined disturbance cost of each candidate fan after its commissioning, taking into account the current status of the fan combination already in operation, the relationship between adjacent units in the tower group, the ambient wind direction, and historical heat exchange response data. The marginal net revenue determination unit is used to determine the marginal net revenue value of each candidate fan based on the newly added cooling revenue value, the combined interference cost value, and the operating energy consumption cost, and to generate the target fan ranking result. The target fan activation unit is used to execute the pre-occupancy gradient activation of the target fans according to the target fan sorting result, and to continuously monitor the return water temperature change rate, outlet water stabilization time and unit energy consumption cooling benefit after activation. The model parameter correction unit is used to perform closed-loop correction on the load propagation chain model and marginal net revenue calculation parameters based on the actual response results after the target wind turbine is put into operation, so as to update the subsequent pre-occupancy timing and wind turbine rearrangement order.

[0007] Furthermore, the load model construction unit includes: The system includes a data preprocessing module, a model training module, a model validation module, a requirements output module, and an online deployment module. The data preprocessing module is used to perform anomaly removal, missing data completion, time alignment and normalization on the running data, and to form training set, validation set and test set; The model training module is used to construct a time-series prediction model consisting of a heat and humidity load generation sub-model, a load transmission sub-model, and a demand prediction sub-model, in order to extract the coupling relationship between changes in greenhouse heat and humidity load and changes in cooling tower heat dissipation demand. The model validation module is used to verify the usability of the model based on prediction error, good fit and stability metrics. The demand output module is used to output demand growth indicators and cooling gap lead time indicators. The online deployment module is used to deploy the accepted model to the control terminal to perform real-time demand prediction and online updates, and to provide a basis for pre-positioning judgment for the wind turbine candidate generation unit.

[0008] Furthermore, the wind turbine candidate generation unit includes the following steps: Based on the demand growth indicator reaching the pre-occupancy threshold and the cooling gap advance indicator meeting the minimum effective heat dissipation contribution condition of a single fan, a candidate set of fans that are in a standby state and whose hardware operating parameters meet the input requirements are selected. The hardware operating parameters include motor speed, bearing temperature, and fan vibration; The pre-occupancy threshold is determined based on a comprehensive assessment of historical cooling tower operating data, the sensitivity of greenhouse production to temperature and humidity fluctuations, and current environmental meteorological conditions. The candidate set generation process also includes verifying the electrical interlock status, maintenance lock status, and communication connectivity status, and outputting the fan availability identifier and priority detection order after the candidate set is generated.

[0009] Furthermore, the revenue-cost calculation unit includes the following steps: This is used to calculate the additional cooling revenue after the candidate fan is put into operation by combining the rated heat dissipation capacity of the candidate fan, the correction coefficient of environmental meteorological parameters, the heat dissipation synergy coefficient of the fan combination already put into operation, and the current actual heat dissipation capacity of the cooling tower. The environmental meteorological parameter correction coefficient is determined based on the ambient temperature, ambient humidity, ambient wind speed and solar radiation intensity, and the heat dissipation coordination coefficient is determined based on the spatial layout, installation spacing and air outlet influence relationship between the candidate wind turbine and the wind turbine already in operation. The newly added cooling benefit value is used to characterize the incremental contribution of the candidate fan to the effective heat dissipation capacity of the cooling tower after it is put into operation.

[0010] Furthermore, the revenue-cost calculation unit also includes: The costs of airflow recirculation interference, hot and humid wake re-intake interference, and local heat exchange short-circuit interference are calculated separately, and then weighted and summed based on their respective weights to obtain the combined interference cost. The weights of each interference cost are determined based on the cooling tower type, fan layout, windbreak facility arrangement and historical interference impact data, and are iteratively corrected based on the measured proportion of interference loss after the fan is put into operation. The combined interference cost is used to characterize the adverse effects of putting candidate fans into operation on the existing heat exchange efficiency. The airflow recirculation interference cost, the hot and humid wake re-intake interference cost, and the local heat exchange short-circuit interference cost correspond to aerodynamic interference, hot and humid superposition interference, and local flow distribution imbalance interference at different spatial locations, respectively.

[0011] Furthermore, the marginal net revenue determination unit includes the following steps: Based on the additional cooling revenue, combined disturbance cost, and operating energy consumption cost of the candidate fans, the marginal net revenue value of each candidate fan is determined, and they are sorted from largest to smallest according to the marginal net revenue value. When there are identical marginal net revenue values, a secondary screening is performed based on the wind turbine hardware operating loss coefficient to determine the target wind turbines for the next tier. The hardware operating loss coefficient is determined based on the fan running time, bearing wear degree, motor efficiency attenuation rate, and cumulative number of start-stop cycles; The operating energy cost is determined based on the rated operating power of the wind turbine and the unit electricity price.

[0012] Furthermore, the target wind turbine commissioning unit includes the following steps: The target wind turbine is put into operation in a gradual manner using a soft-start method, so that the operating frequency of the target wind turbine is gradually increased to the target frequency within a preset time, rather than starting it directly at full load. During the implementation process, the rate of change in return water temperature, the stabilization time of effluent water, and the cooling benefit per unit energy consumption are continuously monitored. When the rate of change of return water temperature exceeds the warning threshold, the gradient input rate is reduced. When the cooling benefit per unit energy consumption is lower than the energy efficiency threshold, a reordering or exit judgment is triggered to suppress system heat exchange disturbances and energy consumption increases caused by rapid load increase or inefficient operation. The target fan commissioning unit is also used to record the frequency ramp-up trajectory, actual cooling response trajectory, and corresponding timestamp information during the gradient commissioning period.

[0013] Furthermore, the model parameter correction unit includes the following steps: Based on the measured additional cooling revenue and the measured combined interference cost after the target wind turbine is put into operation, the interference cost weight, environmental meteorological parameter correction coefficient and heat dissipation synergy coefficient are iteratively corrected; The load propagation chain model is fine-tuned online based on incremental operating data after the fan is put into operation. At the same time, the time delay parameter in the model is corrected according to the deviation between the actual cooling gap occurrence time and the predicted time. The online fine-tuned model continues to be used for subsequent pre-positioning timing determination and wind turbine rearrangement order updates, and the model correction results are fed back to the load model construction unit and the marginal net revenue determination unit.

[0014] Furthermore, it also includes: The tiered input and gradient exit linkage control unit is used to execute the input of the next tier of wind turbines according to the target wind turbine ranking result when the demand growth index continues to rise and exceeds the warning threshold. When the demand growth indicator drops to the exit condition and the cooling gap advance indicator indicates that the current heat dissipation capacity is higher than the actual demand, the target fan is selected according to the reverse sorting result of the marginal net benefit value to perform gradient exit. In this system, only one fan is put into operation or taken out of operation at a single stage. The take-out process adopts a gradual frequency reduction control corresponding to the put-in process to maintain a stable heat exchange state in the cooling circuit and reduce temperature fluctuations and power surges caused by the simultaneous switching of multiple fans.

[0015] Furthermore, it also includes: The operation safety constraint unit is used to perform consistency verification on motor current, bearing temperature, fan vibration, communication status and inverter status before, during and after the target fan is put into operation. When any monitored parameter exceeds the corresponding safety boundary, a constraint blocking command is generated to stop the current candidate wind turbine from being put into operation or to trigger the degraded operation of the wind turbine that has been put into operation. The operational safety constraint unit is also used to feed back the safety boundary triggering result to the wind turbine candidate generation unit and the marginal net benefit determination unit, so as to remove wind turbines that do not meet the safety conditions from the candidate set or reduce their ranking priority, so that the gradient input control can simultaneously satisfy the heat dissipation benefit constraint, the interference cost constraint and the equipment operation safety constraint.

[0016] The beneficial effects of this invention are as follows: 1. Based on real-time collected environmental data and combined with the load propagation chain model, the heat and humidity load requirements of the greenhouse are dynamically calculated to ensure that the cooling tower system can flexibly adjust the fan input according to changes in demand, thereby achieving precise cooling and avoiding over-cooling or under-cooling.

[0017] 2. By screening candidate fans, calculating benefits and costs, and analyzing marginal net benefits, the system can intelligently select the most suitable fan for investment, avoid energy waste, and reduce energy consumption while ensuring cooling effect, thereby improving overall energy efficiency.

[0018] 3. The interference effect between fan combinations was considered. By calculating the combined interference cost, interference phenomena such as airflow backflow, hot and humid wake re-intake, and local heat exchange short circuit caused by improper fan operation were effectively reduced, thus optimizing the overall operational stability of the cooling tower.

[0019] 4. Through real-time monitoring and consistency verification of the running safety constraint unit, the system can perform safety checks on the wind turbine, ensure its safety during operation, avoid wind turbine failures or unstable factors, and ensure long-term stable operation of the system.

[0020] 5. Through online fine-tuning and model parameter correction, the load model and marginal net revenue calculation can be optimized in real time based on the response data after the wind turbine is put into operation, ensuring that the system continues to perform at its best under constantly changing environmental conditions. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating the overall system flow of the present invention. Figure 2 This is a flowchart illustrating the construction process of the load model for this invention. Figure 3 This is a flowchart of the wind turbine gradient input correction process of the present invention. Detailed Implementation

[0022] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0023] Example 1: Please see Figure 1-3 This invention provides a technical solution: an intelligent control system for the gradient operation of cooling tower fans under dynamic greenhouse demand, the system comprising: Multi-source data acquisition unit, load model construction unit, wind turbine candidate generation unit, revenue cost calculation unit, marginal net revenue determination unit, target wind turbine input unit, and model parameter correction unit; The multi-source data acquisition unit is used to collect multi-source operating data from the greenhouse side and the cooling circuit side. The multi-source operating data includes at least the greenhouse internal temperature and humidity, leaf surface temperature, light status, supplemental lighting equipment operating status, spraying or atomization status, ventilation execution status, circulating water flow rate, return water temperature and environmental meteorological parameters. Among them, multi-source operational data refers to data information from multiple different sources during the operation of the greenhouse and cooling circuit; greenhouse internal temperature and humidity, the temperature and humidity of the air inside the greenhouse, are important indicators reflecting the greenhouse environment and have a direct impact on crop growth; leaf surface temperature, the temperature of the plant leaf surface, is closely related to the physiological processes of crops such as photosynthesis and respiration, and can also reflect the thermal state of the environment in which the crops are located; light status, including light intensity and duration, etc., light is the main energy source for crops to carry out photosynthesis and plays a key role in crop growth and development; supplementary lighting equipment operation status, supplementary lighting equipment is used to provide additional light to the greenhouse when there is insufficient light, and its operation status, such as on / off and power, will affect the light conditions inside the greenhouse; spray or misting status, spray or misting systems are used to regulate the temperature inside the greenhouse. Humidity and temperature, achieved by spraying water mist into the air to cool and humidify, affect the greenhouse environment. Ventilation status, including ventilator openings and equipment power, significantly impacts the greenhouse environment. The flow rate of circulating water in the cooling loop affects the cooling tower's effectiveness and is a crucial parameter. Return water temperature reflects the temperature of the water returning from the greenhouse to the cooling tower, indicating the greenhouse's cooling condition and serving as an important indicator of cooling efficiency. Environmental meteorological parameters, including outdoor temperature, humidity, wind speed, and wind direction, influence heat exchange between the greenhouse and the external environment, thus affecting the greenhouse's internal heat and humidity load. The load model construction unit is used to build and complete the training and verification of the load propagation chain model of the transmission of greenhouse heat and humidity load to cooling tower heat dissipation demand based on multi-source operating data, and output the demand growth index and cooling gap lead time index within the future preset time window. The load propagation chain model, built based on multi-source operational data, describes how greenhouse heat and humidity loads are transmitted to cooling tower heat dissipation demand. This model analyzes the influence of various factors within the greenhouse on cooling tower heat dissipation demand, thus predicting future cooling demand. Training and validation involve using historical data to adjust the model's parameters, enabling it to better fit actual data. Validation uses independent datasets to evaluate the model's accuracy and reliability, ensuring it accurately outputs demand growth indicators and cooling gap lead time indicators in practical applications. The demand growth indicator reflects the growth trend of greenhouse cooling demand within a future preset time window, allowing for advance understanding of cooling demand changes and providing a basis for fan deployment decisions. The cooling gap lead time indicator represents the lead time between current cooling capacity and expected cooling demand within a future preset time window, used to determine whether fans need to be deployed in advance to meet cooling demand. The fan candidate generation unit is used to generate a set of fan candidates to be put into operation when the demand growth index reaches the preset pre-positioning conditions, rather than performing a compensatory full-scale additional operation after the cooling water temperature exceeds the limit. Among them, the demand growth indicator reaches the preset pre-positioning condition. This is a pre-set condition. When the demand growth indicator reaches this condition, it means that the future cooling demand has a significant growth trend, and it is necessary to consider putting fans in advance, rather than waiting until the cooling water temperature exceeds the limit before taking a compensatory full-scale additional investment approach. The candidate set of fans to be put into operation is the set of fans that may need to be put into operation when the demand growth indicator reaches the preset pre-positioning condition. These fans will be further determined through steps such as benefit-cost calculation to determine whether they will be put into operation. The benefit and cost calculation unit is used to calculate the additional cooling benefit value after the candidate fan is put into operation, as well as the combined disturbance cost value corresponding to airflow recirculation, hot and humid wake re-intake, and local heat exchange short circuit, for each candidate fan, in combination with the current status of the fan combination already in operation, the relationship between adjacent units in the tower group, the ambient wind direction and historical heat exchange response data. The criteria for calculating the cooling tower's performance include: 1) Current fan configuration status: This refers to the number, model, and operating power of the cooling tower fans currently in operation. Different configurations affect the effectiveness of adding new fans. 2) Adjacent unit relationships within the tower group: Cooling towers typically exist in groups, and adjacent towers interact in terms of airflow and heat / humidity. Considering these relationships allows for more accurate calculations of the effects of adding new fans. 3) Ambient wind direction: Wind direction influences the airflow around the cooling tower, affecting the airflow distribution and cooling effect after adding new fans. It also impacts airflow recirculation and the re-intake of hot and humid wakes. 4) Historical heat exchange response data: This data reflects the past heat exchange performance of cooling tower fans under different operating conditions, including the impact of adding new fans on cooling benefits and combination interference. This data can be used to calculate the increased cooling benefits and combination effects of adding current candidate fans. The combined interference cost is provided for reference; the additional cooling benefit value refers to the benefit value corresponding to the additional cooling effect that a candidate fan can bring to the cooling system after it is put into operation. It can usually be measured by indicators such as the increase in cooling capacity and the reduction in return water temperature; airflow recirculation, after the addition of a new fan, due to unreasonable airflow organization and other reasons, some airflow may flow back to the air inlet of the cooling tower, affecting the cooling effect; hot and humid wake re-intake, the hot and humid wake discharged from the cooling tower may be re-intaken by other cooling towers or new fans, resulting in a decrease in local heat exchange efficiency; local heat exchange short circuit, after the addition of a new fan, a local airflow short circuit phenomenon may be formed around the cooling tower, causing some air to be discharged without sufficient heat exchange, reducing the cooling effect; combined interference cost value, the comprehensive cost value of the negative impacts such as the decrease in cooling effect and the increase in energy consumption caused by phenomena such as airflow recirculation, hot and humid wake re-intake, and local heat exchange short circuit. The marginal net revenue determination unit is used to determine the marginal net revenue value of each candidate fan based on the new cooling revenue value and the combined disturbance cost value, and to determine the target fan for the next tier according to the size of the marginal net revenue value. Among them, the marginal net benefit value refers to the difference between the additional cooling benefit value and the combined disturbance cost value after a candidate fan is put into operation. This value reflects the overall net benefit of the cooling system after the candidate fan is put into operation. The target fans for the next tier are ranked according to the size of the marginal net benefit value of each candidate fan, and the candidate fan with the largest marginal net benefit value is selected as the target fans for the next tier to maximize the benefits of the cooling system. The target fan activation unit is used to execute the pre-occupancy gradient activation of the target fan and continuously monitor the return water temperature change rate, outlet water stabilization time and unit energy consumption cooling benefit after activation. Among these, the pre-positioning gradient deployment of target fans involves gradually putting them into operation in a predetermined order, rather than all at once. This gradient deployment approach better adapts to the dynamic changes in greenhouse cooling demand, avoiding over- or under-investment. The return water temperature change rate is the rate at which the return water temperature changes over time after the target fans are put into operation. Monitoring this rate allows for understanding the real-time impact of the fans on the cooling effect. The outlet water stabilization time is the time required for the cooling tower outlet water temperature to reach a stable state after the target fans are put into operation. This time reflects the response speed and stability of the cooling system after the fans are put into operation. The unit energy consumption cooling benefit refers to the cooling effect obtained for each unit of energy consumed, such as electricity. This is used to evaluate the energy utilization efficiency after the fans are put into operation and is an important indicator for measuring the economics of the cooling system. The model parameter correction unit is used to perform closed-loop correction of the load propagation chain model and marginal net revenue calculation parameters based on the actual response results, so as to update the subsequent pre-occupancy timing and wind turbine rearrangement order. Among them, the actual response results include data such as the rate of change of return water temperature, the stabilization time of outlet water, and the cooling benefit per unit energy consumption after the target fan is put into operation. These data reflect the real effect after the fan is put into operation. The closed-loop correction adjusts and optimizes the load propagation chain model and marginal net benefit calculation parameters based on the actual response results, so that the model can more accurately predict cooling demand and calculate marginal net benefit, thereby updating the subsequent pre-occupancy timing and fan rearrangement order, and improving the accuracy and reliability of the entire intelligent control system.

[0024] It should be noted that during use, comprehensive multi-source data acquisition can accurately grasp the status of the greenhouse and cooling circuit, providing a reliable basis for subsequent control. Load model construction can predict demand growth and cooling gaps in advance, enabling proactive control and avoiding reactive responses. Fan candidate generation prepares in advance when demand growth reaches the target, rather than waiting for water temperature to exceed the limit before putting them into operation, which can meet cooling needs in a timely manner and reduce the impact of temperature fluctuations on crops. Benefit-cost calculation comprehensively considers multiple factors, accurately assesses the benefits and disturbance costs of adding new fans, and ensures the rationality of investment. Marginal net benefit determination selects fans based on value to maximize benefits. Target fan gradient investment adapts to dynamic demand. Monitoring indicators can evaluate the effect. Model parameter correction forms a closed loop, continuously optimizing the model and parameters to improve the accuracy and reliability of the system. Overall, it realizes intelligent, efficient, and precise control of cooling tower fans, ensuring stable operation of the greenhouse.

[0025] In one embodiment, the load propagation chain model for the transmission of greenhouse heat and humidity load to cooling tower heat dissipation demand is a multi-link coupled deep learning model. The training, validation, and deployment steps of the model specifically include: Data preprocessing involves cleaning the collected multi-source operational data, removing outliers, missing values, and isolated points. Outliers are identified using... According to the criteria, missing values ​​are filled using linear interpolation. The cleaned data is normalized, mapping all feature parameters to... The normalization formula for the interval is:

[0026] in, This is the original data. This is the minimum value of the characteristic parameter. The maximum value of this feature parameter is given. The preprocessed data is divided into training set, validation set and test set in a ratio of 7:2:1. The training set is used for model parameter iteration, the validation set is used for hyperparameter tuning and overfitting suppression, and the test set is used for final model performance verification. The model construction, specifically the load propagation chain model, consists of a series of sub-models for heat and humidity load generation, load transmission, and demand prediction. The overall architecture employs a CNN-BiLSTM hybrid network: the CNN layer uses three one-dimensional convolutional kernels with kernel sizes of 3, 5, and 7, all with a stride of 1, to extract local spatiotemporal features from multi-source data; the BiLSTM layer has two hidden layers, each with 128 neurons, to capture the temporal dependencies of heat and humidity load transmission; and the output layer is a fully connected layer with the ReLU activation function, used to output demand growth indicators. and cooling gap advance indicator ; Model training was performed using the following hyperparameters: the optimizer was Adam, the initial learning rate was set to 0.001, cosine annealing was used to dynamically adjust the learning rate, the learning rate decay coefficient was 0.95, the batch size was set to 64, the number of training epochs was set to 100, and the early stopping strategy used the validation set mean squared error (MSE) as its monitoring metric. Training was terminated if the MSE on the validation set did not decrease for 10 consecutive epochs. The loss function used was the mean squared error loss (MSELoss), which measures the deviation between the model's predicted values ​​and the true values. The loss function formula is as follows:

[0027] in, This refers to the number of samples within a batch. These are the model's predicted values. The actual value; Model validation and evaluation involves inputting the test set into the trained model and using the mean absolute error. Root mean square error and coefficient of determination As a performance evaluation metric for the model, the formula for calculating the metric is:

[0028]

[0029]

[0030] in, The average of the true values; The model acceptance criteria are: on the test set , , If the target is not met, the hyperparameters are readjusted and training is performed. Model deployment involves lightweighting the trained and validated model, removing redundant neurons through model pruning, with a pruning rate set to [value missing]. The lightweight model is deployed to the intelligent control terminal of the cooling tower, supporting real-time online inference; After the model training is completed, the increase in sensible heat load of the greenhouse per unit time is calculated based on the heat and moisture balance equation. and latent heat load increment The formula is:

[0031]

[0032] in, The specific heat capacity of air at constant pressure. The air density inside the greenhouse, For the effective volume of the greenhouse, The rate of temperature change inside the greenhouse. The overall heat transfer coefficient of the greenhouse envelope, This refers to the real-time temperature inside the greenhouse. For real-time ambient temperature, The photothermal conversion coefficient, Real-time light intensity; This represents the rate of change in the humidity content of the air inside the greenhouse. The overall moisture transfer coefficient of the greenhouse enclosure structure, This refers to the real-time humidity level of the air inside the greenhouse. This refers to the real-time humidity level of the ambient air. The spray / atomization humidity exchange coefficient, Real-time water flow rate for spraying / atomization; Based on the heat transfer characteristics of the cooling circuit pipes, equipment heat exchange, and medium flow characteristics, the theoretical heat dissipation demand increment of the cooling tower is calculated. The formula is:

[0033] in, The overall heat loss coefficient of the cooling circuit is determined by the insulation performance of the pipes and the heat dissipation characteristics of the equipment. The overall heat exchange efficiency of the cooling circuit is determined by the heat transfer coefficient of the heat exchanger and the contact area of ​​the medium. The time delay factor for the transfer of heat and humidity load from the greenhouse side to the cooling tower side is determined by the circulating water flow rate, pipe length, and medium velocity. Based on the theoretical increase in heat dissipation demand Based on the actual heat dissipation capacity of the cooling tower Calculate the future preset time window Domestic demand growth indicators and cooling gap advance indicator The formula is:

[0034]

[0035] Future preset time window Duration The value is determined based on the historical average delay time of the greenhouse heat and humidity load transfer to the cooling tower, and is taken as [value missing]. ,and The range of values ​​is Demand growth indicators Cooling gap lead time index is used to characterize the growth trend and rate of cooling tower heat dissipation demand within a future time window. This is used to characterize the maximum difference between the theoretical heat dissipation demand of a cooling tower and its actual heat dissipation capacity within a future time window, enabling early prediction of the cooling demand gap.

[0036] This design employs a multi-link coupled deep learning model for the load propagation chain model. Multi-source data undergoes preprocessing such as cleaning and normalization to ensure data quality and provide a reliable foundation for model training. The CNN-BiLSTM hybrid architecture fully extracts local spatiotemporal features and temporal dependencies from the data, accurately predicting demand growth and cooling gaps. Rigorous model training, validation, and evaluation processes ensure model performance meets standards. Theoretical heat dissipation requirements are calculated based on thermal and humidity balance and cooling loop characteristics, making the index calculations more scientific and reasonable. This allows for early prediction of cooling demand gaps, providing accurate basis for the gradual deployment of wind turbines and improving the timeliness and accuracy of the cooling system's response.

[0037] In one embodiment, the preset pre-positioning condition is a demand growth indicator. Reaching the preset threshold And the cooling gap advance indicator satisfy ; Pre-occupancy threshold The determination rule is as follows: based on historical operating data of the cooling tower, the critical value of the maximum demand increase under the premise that the cooling water temperature does not exceed the limit is statistically calculated, and then corrected by combining the temperature and humidity sensitivity coefficient of greenhouse production. The more sensitive the greenhouse is to temperature and humidity fluctuations... The lower the value, The value range is 0.1-0.3; This represents the minimum effective heat dissipation contribution of a single fan under the current environment. This value is determined by multiple regression analysis based on the fan's rated performance, environmental meteorological parameters, and the cooling tower's heat transfer characteristics. Not less than the rated heat dissipation capacity of a single fan ; The generation of the candidate set of cooling tower fans to be put into operation requires hardware status detection of the fans. Fans currently in a standby state with hardware operating parameters such as motor speed, bearing temperature, and fan vibration meeting the conditions for commissioning are selected to form the candidate set.

[0038] in, The number of candidate wind turbines is determined to ensure their availability and operational safety.

[0039] This is the design. Based on historical data and greenhouse sensitivity coefficients, it is possible to anticipate the growth trend of cooling demand and avoid reactive compensation after water temperature exceeds limits. Considering fan performance and environmental factors, we ensure that the put-in fans can effectively dissipate heat. Fan hardware status detection ensures that candidate fans can be put into operation and operate safely. We guarantee the rationality and effectiveness of fan input from both demand and fan status perspectives, thereby improving the stability and reliability of the cooling system.

[0040] In one embodiment, the additional cooling benefit value after the candidate fan is put into operation is calculated. This refers to candidate wind turbines. The actual effective heat dissipation capacity improvement brought to the cooling tower system after its implementation is calculated as follows: Combined with candidate wind turbines Rated heat dissipation capacity Correction coefficients for current environmental meteorological parameters The heat dissipation synergy coefficient of the already put fan units The result is obtained through multi-parameter coupled calculation. The theoretical increase in heat dissipation capacity after the investment is taken as the additional cooling benefit value. The formula is:

[0041] in, The values ​​were obtained through multivariate nonlinear fitting of ambient temperature, humidity, wind speed, and solar radiation intensity. Environmental parameters are negatively correlated with the fan's heat dissipation capacity; the higher the ambient temperature and humidity, the lower the wind speed. The smaller the value; From the wind turbines already in operation and the candidate wind turbines The spatial layout and installation spacing are determined. If the candidate wind turbines and the existing wind turbines are arranged reasonably with no obvious airflow obstruction, they can form a heat dissipation synergy. It can be greater than 1, and vice versa; Environmental meteorological parameter correction coefficient The fitting model used was a random forest regression model. The training parameters of this model were: 100 decision trees, a maximum decision tree depth of 10, and a minimum number of splits of 2. After the model was trained, it was validated using a test set to determine the goodness of fit. .

[0042] This design adds cooling benefit value. By using multi-parameter coupled calculations, which comprehensively consider the rated heat dissipation capacity of the fan, the correction coefficient of environmental meteorological parameters, and the heat dissipation synergy coefficient, the actual effective heat dissipation capacity improvement after the candidate fan is put into operation can be accurately reflected. The random forest regression model fits the correction coefficient of environmental meteorological parameters with a high degree of fit, making the calculation results more accurate. This calculation method provides a scientific basis for evaluating the benefits of candidate fans, helps to select fans that can bring the greatest cooling benefits, and optimize the performance of the cooling system.

[0043] In one embodiment, candidate wind turbines are calculated. The combined interference cost after investment This refers to candidate wind turbines. The loss of heat dissipation capacity caused by the negative impact on the heat exchange efficiency of the already put into operation fan units after commissioning needs to comprehensively consider three core interference factors: airflow recirculation, re-intake of hot and humid wake, and local heat exchange short circuit. Specifically, these include: First, calculate the cost of airflow recirculation interference. The cost of hot and humid wake re-intake interference Local heat exchange short-circuit interference cost Then, based on the degree of influence of each disturbance factor on the heat exchange system, corresponding weight coefficients are assigned, and the combined disturbance cost is obtained by weighted summation. The formula is:

[0044] in, , , Let be the weighting coefficients for each interference cost, and The weighting coefficients were determined based on the cooling tower type, fan layout, and historical interference data. Initial weights were determined using the Analytic Hierarchy Process (AHP), and then iteratively corrected using the measured interference loss percentage after each fan commissioning. For crossflow cooling towers, airflow recirculation has the greatest impact, therefore... The hot and humid tail flow is then re-inhaled, and then... The local heat exchange short circuit is minimized, so take For counter-flow cooling towers, the re-intake of hot and humid wake has the greatest impact, and therefore... Airflow recirculation is the next best option. The local heat exchange short circuit is minimized, so take ; airflow backflow interference cost , The airflow recirculation coefficient was determined through multiple orthogonal tests based on the fan outlet angle and the layout of the cooling tower's windbreak facilities. The wind speed gradients for the calibration tests were set to 0.5 m / s, 1.0 m / s, 1.5 m / s, and 2.0 m / s, and the outlet angle gradients were set to 30°, 45°, 60°, and 90°. For real-time wind speed in the environment, For environmental wind direction and candidate wind turbines The smaller the angle between the air outlet and the air outlet, the more severe the airflow recirculation. The larger the value; Hot and wet wake re-intake interference cost , The wake re-intake coefficient is determined by the fan air volume and the outlet height. Candidate wind turbines The area overlapping with the hot and humid wake of the already-operated wind turbine. The larger the overlap area of ​​the wake, the more severe the re-intake of hot and humid wake. The larger the value; Local heat transfer short-circuit interference cost , The heat transfer short-circuit coefficient is determined by the water distribution method and packing distribution characteristics of the cooling tower. Candidate wind turbines Real-time flow rate of circulating water at the injection point. The average circulating water flow rate of the cooling tower is the total flow rate. The higher the local flow rate, the more severe the local heat exchange short-circuit phenomenon. The larger the value.

[0045] This design combines interference cost. By using weighted summation calculations, the costs of three types of disturbances—airflow recirculation, hot and humid wake re-intake, and local heat exchange short circuit—are calculated separately and assigned reasonable weights. This allows for a comprehensive and accurate assessment of the negative impact of candidate fan commissioning on the already commissioned fan combination. The weighting coefficients are determined based on the cooling tower type and historical data and are iteratively corrected to make the calculations more consistent with the actual situation. This calculation method helps to avoid a decrease in heat exchange efficiency due to fan commissioning and ensures the overall performance stability of the cooling system.

[0046] In one embodiment, candidate wind turbines are determined. marginal net income value This refers to the actual revenue generated by the candidate wind turbine after deducting the combined interference costs and its own operating energy consumption costs. The calculation formula is:

[0047] in, Candidate wind turbines Energy consumption cost per unit time

[0048] for Rated operating power, The local electricity price per unit; The ranking of marginal net revenue values ​​follows a descending order, and the wind turbine ranked first is selected as the target wind turbine for the next tier, thus ensuring that each wind turbine investment brings maximum net revenue to the system. If multiple wind turbines have the same marginal net revenue value and it is the maximum value, the wind turbine hardware operation loss coefficient is further introduced as a secondary screening criterion, and the wind turbine with the smallest hardware operation loss coefficient is selected as the target wind turbine for investment, thereby reducing the equipment loss cost of wind turbine long-term operation. The hardware operating loss coefficient of the wind turbine is determined by the wind turbine running time, bearing wear degree, and motor efficiency attenuation rate through the entropy weight method. The calculation steps of the entropy weight method are: normalize each index → ​​calculate the information entropy of the index → ​​calculate the index weight → weighted sum to obtain the hardware operating loss coefficient.

[0049] With this design, the marginal net benefit value By calculating and deducting interference costs and energy consumption costs, and comprehensively considering the additional cooling benefits, combined interference costs, and its own operating energy consumption costs, the actual benefits after the candidate fans are put into operation can be accurately reflected. The target fans are selected by ranking them according to marginal net benefits, ensuring that each investment brings the maximum net benefit. The fan hardware operating loss coefficient is introduced as a secondary screening basis to reduce equipment loss costs. This design helps to optimize fan investment decisions and improve the economic efficiency of the cooling system and the service life of the equipment.

[0050] In one embodiment, the target fan is pre-positioned and gradually engaged using a soft-start gradient engagement method. The operating frequency of the target fan is gradually increased according to a preset gradient engagement rate, rather than being directly engaged at full load. This avoids current surges and sudden changes in system heat transfer caused by full-load fan startup. The gradient engagement rate... satisfy:

[0051] in, The rated operating frequency of the fan. The preset time for soft start of the fan is determined by the characteristics of the fan motor and the stability requirements of the cooling tower system, and is generally 10-30 seconds. Simultaneously, during the operation of the fan, three core operating indicators need to be continuously monitored: return water temperature change rate, outlet water stabilization time, and cooling benefit per unit energy consumption. The formula used to quantify the energy efficiency level of wind turbines in operation is as follows:

[0052] in, After the target wind turbine is put into operation The rate of return water temperature drop over a period of time This refers to the real-time flow rate of the circulating water. The specific heat capacity of water, For the target wind turbine Average operating power over time The higher the value, the higher the energy efficiency level of the wind turbine. A warning threshold is set for the rate of change in return water temperature. This threshold is determined by the allowable rate of water temperature fluctuation in greenhouse production, and its value range is [value missing]. If the threshold is exceeded, the wind turbine gradient start-up rate will be reduced by a certain percentage. .

[0053] This design employs a soft-start gradient approach to engage the target fan and monitors key performance indicators. This avoids current surges and sudden changes in system heat exchange caused by full-load fan startup, protecting equipment and ensuring system stability. Continuous monitoring of return water temperature change rate, outlet water stabilization time, and cooling benefit per unit energy consumption allows for timely assessment of fan engagement effectiveness. Setting a return water temperature change rate warning threshold and adjusting the engagement rate as needed prevents excessive water temperature fluctuations from impacting greenhouse production and ensures the safe and stable operation of the cooling system.

[0054] In one embodiment, closed-loop correction of the marginal net revenue calculation parameters is performed based on the deviation between actual operating data and theoretical calculation data after the wind turbine is put into operation. This involves iterative optimization of the parameters and online fine-tuning of the load propagation chain model in conjunction with the cooling gap prediction deviation. Specifically, this includes: The marginal net revenue calculation parameters are adjusted based on the measured increase in cooling revenue after the actual target fan is put into operation. Combined interference cost with actual measurement Weighting coefficients for each interfering factor , , and environmental meteorological parameter correction coefficients Heat dissipation synergy coefficient The correction is performed iteratively, and the correction formula is as follows:

[0055]

[0056]

[0057] in, , , For the first Initial values ​​of parameters for the next iteration , , For the first The corrected parameter values ​​from each iteration are used to match the calculated parameters with the actual operating conditions through multiple iterations. The termination condition for iterative correction is the relative deviation between the theoretical calculated values ​​and the measured values. ; The load propagation chain model was fine-tuned online. Actual operating data after the wind turbines were put into operation were selected as incremental samples to fine-tune the load propagation chain model online. The fine-tuning hyperparameters were set as follows: the learning rate was reduced to 1 / 10 of the initial training learning rate, i.e., 0.0001; the batch size was set to 32; the number of training rounds was set to 20; and the loss function was still MSELoss. After the fine-tuning was completed, the model performance was re-verified to ensure that the model prediction accuracy met the acceptance criteria. Simultaneously, based on the deviation between the actual occurrence time of the cooling gap and the model's predicted time, the time delay coefficient in the load propagation chain model is adjusted. Linear correction is applied; the larger the deviation, the larger the correction magnitude. The correction formula is:

[0058] in, To correct the previous time delay factor, This is the corrected time delay factor. This is a correction factor, with a value ranging from 0.1 to 0.2. This represents the deviation between the actual occurrence time of the cooling gap and the predicted time. Through the above closed-loop correction and model fine-tuning, the model and parameters are dynamically updated, continuously optimizing the subsequent pre-occupancy timing determination and fan rearrangement sequence, ensuring the adaptability and accuracy of the control method.

[0059] This design involves closed-loop correction of the marginal net benefit calculation parameters and load propagation chain model. Based on the deviation between actual operating data and theoretical calculation data, iterative optimization of parameters and online fine-tuning of the model are performed to match the calculation parameters and model with the actual operating state, thereby improving prediction accuracy. The time delay coefficient is corrected according to the prediction deviation of the cooling gap to further optimize the model. The closed-loop correction enables dynamic updates of the model and parameters, continuously optimizing the pre-occupancy timing determination and fan rearrangement sequence, ensuring the adaptability and accuracy of the control method, and improving the performance of the cooling system.

[0060] In one embodiment, the system further includes a tiered start-up and gradient shutdown linkage control strategy for the cooling tower fans, specifically: After the cooling tower fans are put into operation, demand growth indicators will be continuously monitored in real time. and cooling gap advance indicator If demand growth indicators The number of cases continues to rise and exceeds the warning threshold. ( This indicates that the current fan assembly's heat dissipation capacity is still insufficient to meet future cooling demands. Warning threshold The determination rule is: based on the pre-occupancy threshold. Based on the benchmark, and combined with the maximum heat dissipation capacity margin of the cooling tower, ,in This is a margin correction factor, determined by the number of cooling tower fans and the heat dissipation capacity of a single fan, with a value ranging from 0.1 to 0.2. The maximum value shall not exceed 0.5; the next stage of wind turbines shall be invested in tiered manner according to the ranking results of marginal net income, so as to realize the incremental investment of wind turbines in tiers; If demand growth indicators Drops below zero and cooling gap advance indicator This indicates that the current fan combination's heat dissipation capacity exceeds the actual cooling demand, resulting in energy waste. At this point, based on the reverse ranking of the fans' marginal net benefits, the fan with the smallest marginal net benefit value is selected to perform a gradient exit operation. The exit process also adopts a soft-start gradient method, gradually reducing the fan's operating frequency until it stops. The frequency reduction rate is consistent with the input rate to avoid sudden changes in system heat exchange and achieve the gradient reduction and exit of the fans. The maximum number of tiers to be put into operation is determined by the total number of cooling tower fans. The number of fans put into operation in a single tier is 1, to avoid system disturbance caused by multiple fans being put into operation at the same time.

[0061] This design incorporates a tiered input and gradient output linkage control strategy, continuously monitoring demand growth indicators and cooling gap advance indicators. Based on changes in these indicators, it executes tiered incremental input or gradient reduction output of fans. The setting of early warning thresholds and reverse sorting results ensures that fan input and output decisions are scientific and reasonable. The soft-start gradient method avoids sudden changes in system heat exchange. Only one fan is put into operation per tier, reducing system disturbances. This strategy can dynamically adjust fan operation according to actual cooling demand, improving the energy efficiency of the cooling system and reducing energy waste.

[0062] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0063] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A smart control system for the gradient operation of cooling tower fans under dynamic demand in greenhouses, characterized in that, The system includes: The multi-source data acquisition unit is used to collect operating data from the greenhouse side and the cooling circuit side. The operating data includes greenhouse internal temperature, greenhouse internal humidity, leaf surface temperature, light status, supplemental lighting equipment operating status, spraying or atomization status, ventilation execution status, circulating water flow rate, return water temperature, and environmental meteorological parameters. The load model construction unit is used to construct a load propagation chain model based on the operating data to transmit the greenhouse heat and humidity load to the cooling tower heat dissipation demand, and output the demand growth index and cooling gap advance index within a future preset time window according to the load propagation chain model. The wind turbine candidate generation unit is used to generate a set of candidate wind turbines to be put into operation when the demand growth index meets the pre-occupancy conditions. The wind turbine candidate generation unit includes the following steps: Based on the demand growth indicator reaching the pre-occupancy threshold and the cooling gap advance indicator meeting the minimum effective heat dissipation contribution condition of a single fan, a candidate set of fans that are in a standby state and whose hardware operating parameters meet the input requirements are selected. The hardware operating parameters include motor speed, bearing temperature, and fan vibration; The pre-occupancy threshold is determined based on a comprehensive assessment of historical cooling tower operating data, the sensitivity of greenhouse production to temperature and humidity fluctuations, and current environmental meteorological conditions. The candidate set generation process also includes verifying the electrical interlock status, maintenance lock status, and communication connectivity status, and outputting the fan availability identifier and priority detection order after the candidate set is generated; The revenue and cost calculation unit is used to calculate the additional cooling revenue and combined disturbance cost of each candidate fan after its commissioning, taking into account the current status of the fan combination already in operation, the relationship between adjacent units in the tower group, the ambient wind direction, and historical heat exchange response data. The benefit-cost calculation unit further includes: The costs of airflow recirculation interference, hot and humid wake re-intake interference, and local heat exchange short-circuit interference are calculated separately, and then weighted and summed based on their respective weights to obtain the combined interference cost. The weights of each interference cost are determined based on the cooling tower type, fan layout, windbreak facility arrangement and historical interference impact data, and are iteratively corrected based on the measured proportion of interference loss after the fan is put into operation. The combined interference cost is used to characterize the adverse effects of putting candidate fans into operation on the existing heat exchange efficiency. The airflow recirculation interference cost, the hot and humid wake re-intake interference cost, and the local heat exchange short-circuit interference cost correspond to aerodynamic interference, hot and humid superposition interference, and local flow distribution imbalance interference at different spatial locations, respectively. The marginal net revenue determination unit is used to determine the marginal net revenue value of each candidate fan based on the newly added cooling revenue value, the combined interference cost value, and the operating energy consumption cost, and to generate the target fan ranking result. The target fan activation unit is used to execute the pre-occupancy gradient activation of the target fans according to the target fan sorting result, and to continuously monitor the return water temperature change rate, outlet water stabilization time and unit energy consumption cooling benefit after activation. The model parameter correction unit is used to perform closed-loop correction on the load propagation chain model and marginal net revenue calculation parameters based on the actual response results after the target wind turbine is put into operation, so as to update the subsequent pre-occupancy timing and wind turbine rearrangement order. The model parameter correction unit includes the following steps: Based on the measured additional cooling revenue and the measured combined interference cost after the target wind turbine is put into operation, the interference cost weight, environmental meteorological parameter correction coefficient and heat dissipation synergy coefficient are iteratively corrected; The load propagation chain model is fine-tuned online based on incremental operating data after the fan is put into operation. At the same time, the time delay parameter in the model is corrected according to the deviation between the actual cooling gap occurrence time and the predicted time. The online fine-tuned model continues to be used for subsequent pre-positioning timing determination and wind turbine rearrangement order updates, and the model correction results are fed back to the load model construction unit and the marginal net revenue determination unit.

2. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 1, characterized in that, The load model construction unit includes: The system includes a data preprocessing module, a model training module, a model validation module, a requirements output module, and an online deployment module. The data preprocessing module is used to perform anomaly removal, missing data completion, time alignment and normalization on the running data, and to form training set, validation set and test set; The model training module is used to construct a time-series prediction model consisting of a heat and humidity load generation sub-model, a load transmission sub-model, and a demand prediction sub-model, in order to extract the coupling relationship between changes in greenhouse heat and humidity load and changes in cooling tower heat dissipation demand. The model validation module is used to verify the usability of the model based on prediction error, good fit and stability metrics. The demand output module is used to output demand growth indicators and cooling gap lead time indicators. The online deployment module is used to deploy the accepted model to the control terminal to perform real-time demand prediction and online updates, and to provide a basis for pre-positioning judgment for the wind turbine candidate generation unit.

3. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 2, characterized in that, The revenue-cost calculation unit includes the following steps: This is used to calculate the additional cooling revenue after the candidate fan is put into operation by combining the rated heat dissipation capacity of the candidate fan, the correction coefficient of environmental meteorological parameters, the heat dissipation synergy coefficient of the fan combination already put into operation, and the current actual heat dissipation capacity of the cooling tower. The environmental meteorological parameter correction coefficient is determined based on the ambient temperature, ambient humidity, ambient wind speed and solar radiation intensity, and the heat dissipation coordination coefficient is determined based on the spatial layout, installation spacing and air outlet influence relationship between the candidate wind turbine and the wind turbine already in operation. The newly added cooling benefit value is used to characterize the incremental contribution of the candidate fan to the effective heat dissipation capacity of the cooling tower after it is put into operation.

4. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 3, characterized in that, The marginal net revenue determination unit includes the following steps: Based on the additional cooling revenue, combined disturbance cost, and operating energy consumption cost of the candidate fans, the marginal net revenue value of each candidate fan is determined, and they are sorted from largest to smallest according to the marginal net revenue value. When there are identical marginal net revenue values, a secondary screening is performed based on the wind turbine hardware operating loss coefficient to determine the target wind turbines for the next tier. The hardware operating loss coefficient is determined based on the fan running time, bearing wear degree, motor efficiency attenuation rate, and cumulative number of start-stop cycles; The operating energy cost is determined based on the rated operating power of the wind turbine and the unit electricity price.

5. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 1, characterized in that, The target wind turbine commissioning unit includes the following steps: The target wind turbine is put into operation in a gradual manner using a soft-start method, so that the operating frequency of the target wind turbine is gradually increased to the target frequency within a preset time, rather than starting it directly at full load. During the implementation process, the rate of change in return water temperature, the stabilization time of effluent water, and the cooling benefit per unit energy consumption are continuously monitored. When the rate of change of return water temperature exceeds the warning threshold, the gradient input rate is reduced. When the cooling benefit per unit energy consumption is lower than the energy efficiency threshold, a reordering or exit judgment is triggered to suppress system heat exchange disturbances and energy consumption increases caused by rapid load increase or inefficient operation. The target fan commissioning unit is also used to record the frequency ramp-up trajectory, actual cooling response trajectory, and corresponding timestamp information during the gradient commissioning period.

6. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 4, characterized in that, Also includes: The tiered input and gradient exit linkage control unit is used to execute the input of the next tier of wind turbines according to the target wind turbine ranking result when the demand growth index continues to rise and exceeds the warning threshold. When the demand growth indicator drops to the exit condition and the cooling gap advance indicator indicates that the current heat dissipation capacity is higher than the actual demand, the target fan is selected according to the reverse sorting result of the marginal net benefit value to perform gradient exit. In this system, only one fan is put into operation or taken out of operation at a single stage. The take-out process adopts a gradual frequency reduction control corresponding to the put-in process to maintain a stable heat exchange state in the cooling circuit and reduce temperature fluctuations and power surges caused by the simultaneous switching of multiple fans.

7. The intelligent control system for gradient operation of cooling tower fans under dynamic greenhouse demand as described in claim 5, characterized in that, Also includes: The operation safety constraint unit is used to perform consistency verification on motor current, bearing temperature, fan vibration, communication status and inverter status before, during and after the target fan is put into operation. When any monitored parameter exceeds the corresponding safety boundary, a constraint blocking command is generated to stop the current candidate wind turbine from being put into operation or to trigger the degraded operation of the wind turbine that has been put into operation. The operational safety constraint unit is also used to feed back the safety boundary triggering result to the wind turbine candidate generation unit and the marginal net benefit determination unit, so as to remove wind turbines that do not meet the safety conditions from the candidate set or reduce their ranking priority, so that the gradient input control can simultaneously satisfy the heat dissipation benefit constraint, the interference cost constraint and the equipment operation safety constraint.

Images