Data center cooling system dynamic control method adaptive to equipment aging

Abstract

The invention relates to the technical field of data center energy control, and discloses a data center cooling system dynamic control method adaptive to equipment aging, and the method comprises the steps: firstly constructing an energy consumption model for controlling a data center cooling system based on equipment parameters and an energy conservation principle; in order to solve the problem of performance degradation possibly caused by long-term operation of equipment, the method monitors operation parameters of key equipment such as a cooling tower, a heat exchanger, a water chilling unit and an air conditioner terminal in real time; according to the method, an equipment energy consumption model is iterated in a rolling manner by using a long-short term memory (LSTM) network model in deep learning, so that the energy consumption model for control is dynamically updated, the prediction precision is improved, and finally efficient model prediction control is realized. The method is not only suitable for the design and construction of a new data center, but also suitable for the transformation and upgrading of an existing data center, and has wide application prospects and practical values.

Description

Technical Field

[0001] The present invention relates to the technical field of data center energy control, and in particular to a dynamic control method for a data center cooling system adapting to equipment aging. Background Art

[0002] As the core of information processing in modern society, data centers are responsible for the storage, processing and transmission of massive amounts of data. With the rapid development of technologies such as cloud computing, big data, and the Internet of Things, the scale and complexity of data centers are increasing, and the demand for energy is also growing. Traditional data center energy management methods often use static or extensive control strategies, which are difficult to adapt to the dynamic changes in equipment and energy demand in data centers, resulting in low energy efficiency and high operating costs.

[0003] In the operation of data centers, the energy consumption of key components such as servers, storage devices, network equipment, and cooling systems accounts for the vast majority of total energy consumption. The energy consumption of these devices is not only related to the performance of the equipment itself, but also affected by multiple factors such as operating load, ambient temperature, humidity, etc. In addition, the energy demand of data centers has obvious temporal and spatial fluctuations, and the energy demand in different time periods and different regions may vary significantly.

[0004] In order to improve the energy efficiency of data centers, reduce operating costs, and achieve green and sustainable development, there is an urgent need for a method that can perform adaptive dynamic control based on the aging status of equipment and the real-time energy demand of data centers. This control strategy should be able to monitor the operating status of data centers in real time, accurately predict the changing trends of equipment and energy demand, and dynamically adjust equipment operating parameters and energy supply strategies to achieve optimal allocation and utilization of energy.

[0005] Based on the above needs, the present invention proposes a dynamic control method for a data center cooling system that adapts to equipment aging. This method constructs a prediction model of equipment aging status and energy demand, and combines advanced control algorithms to achieve refined management of various types of equipment in the data center and dynamic optimization of energy supply. Through the application of this method, the energy utilization efficiency of the data center can be significantly improved, energy consumption can be reduced, operating costs can be reduced, and the operational stability and reliability of the data center can be improved. Summary of the invention

[0006] The present invention aims to solve the problems of inaccurate prediction and high energy consumption caused by equipment aging in traditional cooling systems of data centers, and proposes a dynamic control method for data center cooling systems that adapts to equipment aging. This method combines the control energy consumption model with the long short-term memory network (LSTM) model, and uses the historical operation data of cooling towers, water pumps, heat exchangers, and chillers to train the model to predict the aging degree of these devices. According to the prediction results, the energy consumption system is fed back, and the equipment parameters are adjusted through the control equation for adaptive control. At the same time, the uncertainty factors in the system operation are considered to achieve accurate equipment control, thereby improving energy efficiency.

[0007] In the specific implementation, the control energy consumption model of the data center cooling system is first constructed based on equipment parameters and the principle of energy conservation. This model is used to describe the energy consumption of the system under different working conditions and provide a basis for subsequent control strategies. In order to address the performance degradation problem that may occur in the long-term operation of the equipment, the method monitors the operating parameters of key equipment such as cooling towers, heat exchangers, chillers and air-conditioning terminals in real time. When it is detected that the predicted value of the control energy consumption model deviates from the actual operating value by a certain threshold, the long short-term memory network (LSTM) model in deep learning is used to update the equipment energy consumption model in a rolling iterative manner. The LSTM model can capture the long-term dependencies in time series data and improve the accuracy of predicting equipment aging trends by learning and analyzing operating data.

[0008] The present invention is not only applicable to the design and construction of new data centers, but also to the transformation and upgrading of existing data centers. For new data centers, the control method of the present invention can be integrated into the design stage to optimize the layout and configuration of the cooling system and improve the overall energy efficiency; for existing data centers, the control system software and hardware facilities can be upgraded and the method of the present invention can be applied to carry out energy-saving transformation. Whether it is a new construction or transformation project, the present invention can significantly improve the energy utilization rate of the data center and reduce operating costs, and has broad application prospects and practical value.

[0009] In order to achieve the technical purpose, the present invention adopts the following technical solutions to achieve it.

[0010] The dynamic control method of a data center cooling system adapted to equipment aging proposed by the present invention comprises the following steps: S1. Build a control energy consumption model based on the data center data. A control energy consumption model is constructed based on the control equation of the data center. The model mainly includes four parts: cold load calculation, air conditioning terminal module calculation, cold source module calculation and energy consumption module calculation; The cooling load calculation mainly includes the heat transfer of the computer room enclosure structure, the heat dissipation of the server, the heat dissipation of the PDU and UPS, and the heat dissipation of the lighting and dehumidification equipment, so as to calculate the total cooling load Qhp ; The air conditioning terminal module calculates the cooling capacity Q of the air conditioning terminal using the ε-NTU method crah ; The cold source module calculates the cooling capacity Q of the cooling tower ctw , cooling capacity of chiller Q chiller and the COP of the chiller, and determine the operating mode of the equipment; the energy consumption module calculates the number of equipment turned on based on the equipment parameters and operating parameters; S2. Obtain key parameter information of the device. Real-time collection of key equipment operation data, cooling tower cooling water outlet temperature T cw_out , Cooling tower return water temperature T cw_in 、Air volume of cooling tower m cta and cooling tower water flow m ctw ; Cold water supply temperature T ch_in , cold water return temperature T ch_out , air supply temperature at the air conditioning terminal Tcrah_in, air return temperature at the air conditioning terminal T crah_out ; Collect equipment operation data once an hour, and collect data for a set period of time for training and testing deep learning models; S3: Equipment performance evaluation, The cooling efficiency is used as the evaluation index of the cooling tower performance. The cooling efficiency E is calculated using the following formula, where Ts represents the wet bulb temperature at that time: , The efficiency calculation formula of the cold water pump and the cooling water pump is as follows: , Where ρ is the density of water in kg / m3, g is the acceleration of gravity in m / s2, Q is the water flow in m3 / s, H is the pump head in m, and P represents the shaft power in kilowatts (kW); The heat exchanger efficiency calculation formula is as follows: , in It refers to the real-time water flow through the heat exchanger, in kg / s, mmax refers to the maximum water flow through the heat exchanger, in kg / s, CP refers to the specific heat capacity of water, in J / (kg•℃), delta_T refers to the theoretical maximum temperature difference of mmax through the heat exchanger, in ℃; Use COP to judge the performance of the chiller. The COP calculation formula is as follows: , in Refers to the cooling capacity of the chiller, in kW. Refers to the electrical energy consumed by the chiller, in kW.

[0011] Preferably, the key parameter information of the device acquired in step S2 is used as a data set of a long short-term memory network (LSTM) model for training and testing the model, specifically comprising the following steps: S2.1. Data preprocessing, The collected data is preprocessed to remove abnormal data caused by external force majeure factors (such as equipment maintenance, equipment damage, etc.), and then the maximum and minimum normalization method is used to normalize the data. The return water temperature and the outlet water temperature are selected as important features of the cooling tower for normalization. The inlet water temperature and the outlet water temperature of the heat exchanger are also normalized. The standardized data is converted into a time series format suitable for LSTM model input; this step requires the definition of the time step (time_step), that is, how many time points each input sample contains; the maximum and minimum normalization formula is as follows: , in is the original value of the ith data point, and are the minimum and maximum values ​​of the data set; S2.2, Model training, Since the equipment needs to be dynamically and adaptively adjusted according to the operating data, the long short-term memory network (LSTM) in deep learning can be used to achieve prediction. The specific LSTM unit formulas are shown in formulas 6 to 11. The historical operating parameter data set is divided into a training set and a validation set in a ratio of 7:3. The mean_squared_error is imported from the sklearn.metrics library in Python as an evaluation indicator of the model quality. , , , , , , in is the sigmoid activation function, It's the Gate of Forgetfulness. is the input gate, is the candidate state, is the cell state, is the output gate; According to step S3, we can know that: the key parameters of cooling tower operation are obtained: the return water temperature T of the cooling tower cw_in , cooling water flow m ctwand the cooling tower outlet water temperature T cw_out , The model contains two LSTM layers and one fully connected layer. To prevent overfitting, a Dropout layer is added after the LSTM layer. The Dropout layer randomly discards some neurons during the training process, thereby enhancing the generalization ability of the model. After the model structure is built, the model needs to be compiled. When compiling, the loss function, optimizer, and evaluation index need to be specified. The loss function selects the mean square error (MSE). The formula is as follows: , in is the true value, is the predicted value, N is the number of samples, and the optimizer used is the Adam optimizer. The Adam optimizer combines the advantages of the gradient descent method and the momentum method. Its update rules are as follows: , in is the current parameter, is the learning rate, which can be set according to needs. is the first-order moment estimate of the gradient, is the second moment estimate of the gradient, It is an extremely small number set to prevent the denominator from being 0. The evaluation index uses the commonly used mean square error as an indicator to evaluate performance.

[0012] Preferably, the device parameter prediction and model update strategy includes the following specific steps: The equipment is predicted in real time based on the constructed LSTM model; the cooling water outlet temperature T of the cooling tower in the past n hours is used cw_out , cooling water return temperature of cooling tower T cw_in , cold water supply temperature T ch_in , cold water return temperature T ch_out Perform data preprocessing, standardize the data, convert the data into time series, and then input it into the constructed LSTM model to predict the efficiency of the equipment; By real-time monitoring of the cooling tower's outlet and return water temperatures, cold water supply and outlet temperatures, pump head and chiller's power consumption key operating parameters, combined with the prediction model, the system can dynamically calculate the equipment's instant efficiency; this process not only promotes the continuous optimization of the model and control strategy, but also ensures that the equipment can operate in the best state; when the difference between the actual measured value and the predicted value exceeds the preset threshold, the system will automatically adjust the relevant equipment parameters until it returns to the standard operating conditions; and an advanced rolling update mechanism is used to continuously improve the drive equipment model; the update formula is as follows: , , , in is the predicted energy consumption of the chiller, is the temperature difference of the liquid passing through the chiller, in °C, is the mass of liquid flowing into the chiller, in kg / s, and is the weight trained by the model, Bias, is the predicted energy consumption of the fan, where , and is the weight parameter obtained through model training; b is the bias; is the heat transfer efficiency of the heat exchanger, α1 and α2 are the convection heat transfer coefficients of the fluids on both sides, δ is the plate thickness in m, and λ is the thermal conductivity of the plate material. and Usually refers to the convection heat transfer area of ​​the fluid on both sides, unit m 2 , and The weight parameters and bias are obtained through model training; This mechanism is based on the collected real-time operating parameters of the equipment, uses a deep learning algorithm to retrain the original model structure, and adjusts it according to the law of conservation of energy; this not only improves the accuracy of the model's prediction of equipment energy consumption, but also provides more accurate data support for comprehensive energy consumption analysis; the entire process forms a closed-loop control system, which not only ensures efficient energy saving, but also greatly improves the overall operation management level.

[0013] Preferably, the system is adaptively adjusted and controlled according to the results predicted by the model, and the steps are as follows: When the operating time is less than the set time and the equipment model does not trigger an update, continue the adaptive dynamic control; otherwise, update the aging parameters of the cooling tower, water pump and heat exchanger, re-collect the operating data of the cooling tower and other equipment and re-train the model.

[0014] Compared with the related art, the dynamic control method of the data center cooling system adapted to equipment aging proposed by the present invention has the following beneficial effects: Based on the data set obtained by the energy consumption model of the data center, the present invention combines the long short-term memory network method in deep learning to predict the performance of the equipment and update the aging factor. By real-time monitoring of the equipment operation status, the present invention can collect data feedback in a timely manner and adjust the equipment parameters according to these data. This real-time adjustment ensures that the equipment always operates in the best working conditions, thereby optimizing the overall energy consumption and energy utilization efficiency of the data center. By combining the data center energy consumption model with LSTM technology, the present invention not only improves the accuracy of equipment performance prediction, but also optimizes equipment aging management, providing an innovative solution for energy consumption optimization and equipment health management of the data center. DETAILED DESCRIPTION

[0015] In order to make it easier for those skilled in the art to grasp and perform the present invention, we will explain the method steps of the present invention in detail. Please note that these examples are only used to explain the present invention and do not limit the scope of protection of the present invention. In addition, it should be understood that after reading and understanding the details of the present invention, the technician can make various modifications or adjustments on the basis of the present invention, and these equivalent modifications or adjustments also belong to the scope of protection claimed in this application. Example

[0016] The dynamic control method of a data center cooling system adapted to equipment aging proposed by the present invention comprises the following steps: S1. Build a control energy consumption model based on the data center data. A control energy consumption model is constructed based on the control equation of the data center. The model mainly includes four parts: cold load calculation, air conditioning terminal module calculation, cold source module calculation and energy consumption module calculation; The cooling load calculation mainly includes the heat transfer of the computer room enclosure structure, the heat dissipation of the server, the heat dissipation of the PDU and UPS, and the heat dissipation of the lighting and dehumidification equipment, so as to calculate the total cooling load Q hp ; The air conditioning terminal module calculates the cooling capacity Q of the air conditioning terminal using the ε-NTU method crah ; The cold source module calculates the cooling capacity Q of the cooling tower ctw , cooling capacity of chiller Q chiller and the COP of the chiller, and determine the operating mode of the equipment; the energy consumption module calculates the number of equipment turned on based on the equipment parameters and operating parameters; S2. Obtain key parameter information of the device. Real-time collection of key equipment operation data, cooling tower cooling water outlet temperature T cw_out , Cooling tower return water temperature T cw_in 、Air volume of cooling tower m cta and cooling tower water flow m ctw ; Cold water supply temperature Tch_in , cold water return temperature T ch_out , air supply temperature at the air conditioning terminal Tcrah_in, air return temperature at the air conditioning terminal T crah_out ; Collect equipment operation data once an hour, and collect data for a set period of time for training and testing deep learning models; S3: Equipment performance evaluation, The cooling efficiency is used as the evaluation index of the cooling tower performance. The cooling efficiency E is calculated using the following formula, where Ts represents the wet bulb temperature at that time: , The efficiency calculation formula of the cold water pump and the cooling water pump is as follows: , Where ρ is the density of water in kg / m3, g is the acceleration of gravity in m / s2, Q is the water flow in m3 / s, H is the pump head in m, and P represents the shaft power in kilowatts (kW); The heat exchanger efficiency calculation formula is as follows: , in It refers to the real-time water flow through the heat exchanger, in kg / s, mmax refers to the maximum water flow through the heat exchanger, in kg / s, CP refers to the specific heat capacity of water, in J / (kg•℃), delta_T refers to the theoretical maximum temperature difference of mmax through the heat exchanger, in ℃; Use COP to judge the performance of the chiller. The COP calculation formula is as follows: , in Refers to the cooling capacity of the chiller, in kW. Refers to the electrical energy consumed by the chiller, in kW.

[0017] Furthermore, the key parameter information of the device obtained in step S2 is used as a data set of a long short-term memory network (LSTM) model for training and testing the model, which specifically includes the following steps: S2.1. Data preprocessing, The collected data is preprocessed to remove abnormal data caused by external force majeure factors (such as equipment maintenance, equipment damage, etc.), and then the maximum and minimum normalization method is used to normalize the data. The return water temperature and the outlet water temperature are selected as important features of the cooling tower for normalization. The inlet water temperature and the outlet water temperature of the heat exchanger are also normalized. The standardized data is converted into a time series format suitable for LSTM model input; this step requires the definition of the time step (time_step), that is, how many time points each input sample contains; the maximum and minimum normalization formula is as follows: , in is the original value of the ith data point, and are the minimum and maximum values ​​of the data set; S2.2, Model training, Since the equipment needs to be dynamically and adaptively adjusted according to the operating data, the long short-term memory network (LSTM) in deep learning can be used to achieve prediction. The specific LSTM unit formulas are shown in formulas 6 to 11. The historical operating parameter data set is divided into a training set and a validation set in a ratio of 7:3. The mean_squared_error is imported from the sklearn.metrics library in Python as an evaluation indicator of the model quality. , , , , , , in is the sigmoid activation function, It's the Gate of Forgetfulness. is the input gate, is the candidate state, is the cell state, is the output gate; According to step S3, we can know that: the key parameters of cooling tower operation are obtained: the return water temperature T of the cooling tower cw_in , cooling water flow m ctw and the cooling tower outlet water temperature T cw_out , The model contains two LSTM layers and one fully connected layer. To prevent overfitting, a Dropout layer is added after the LSTM layer. The Dropout layer randomly discards some neurons during the training process, thereby enhancing the generalization ability of the model. After the model structure is built, the model needs to be compiled. When compiling, the loss function, optimizer, and evaluation index need to be specified. The loss function selects the mean square error (MSE). The formula is as follows: , in is the true value, is the predicted value, N is the number of samples, and the optimizer used is the Adam optimizer. The Adam optimizer combines the advantages of the gradient descent method and the momentum method. Its update rules are as follows: , in is the current parameter, is the learning rate, which can be set according to needs. is the first-order moment estimate of the gradient, is the second moment estimate of the gradient, It is an extremely small number set to prevent the denominator from being 0. The evaluation index uses the commonly used mean square error as an indicator to evaluate performance.

[0018] Furthermore, the device parameter prediction and model update strategy, the specific steps are as follows: The equipment is predicted in real time based on the constructed LSTM model; the cooling water outlet temperature T of the cooling tower in the past n hours is used cw_out , cooling water return temperature of cooling tower T cw_in , cold water supply temperature T ch_in , cold water return temperature T ch_out Perform data preprocessing, standardize the data, convert the data into time series, and then input it into the constructed LSTM model to predict the efficiency of the equipment; By real-time monitoring of the cooling tower's outlet and return water temperatures, cold water supply and outlet temperatures, pump head and chiller's power consumption key operating parameters, combined with the prediction model, the system can dynamically calculate the equipment's instant efficiency; this process not only promotes the continuous optimization of the model and control strategy, but also ensures that the equipment can operate in the best state; when the difference between the actual measured value and the predicted value exceeds the preset threshold, the system will automatically adjust the relevant equipment parameters until it returns to the standard operating conditions; and an advanced rolling update mechanism is used to continuously improve the drive equipment model; the update formula is as follows: , , , in is the predicted energy consumption of the chiller, is the temperature difference of the liquid passing through the chiller, in °C, is the mass of liquid flowing into the chiller, in kg / s, and is the weight trained by the model, Bias, is the predicted energy consumption of the fan, where , and is the weight parameter obtained through model training; b is the bias; is the heat transfer efficiency of the heat exchanger, α1 and α2 are the convection heat transfer coefficients of the fluids on both sides, δ is the plate thickness in m, and λ is the thermal conductivity of the plate material. and Usually refers to the convection heat transfer area of ​​the fluid on both sides, unit m 2 , and The weight parameters and bias are obtained through model training; This mechanism is based on the collected real-time operating parameters of the equipment, uses a deep learning algorithm to retrain the original model structure, and adjusts it according to the law of conservation of energy; this not only improves the accuracy of the model's prediction of equipment energy consumption, but also provides more accurate data support for comprehensive energy consumption analysis; the entire process forms a closed-loop control system, which not only ensures efficient energy saving, but also greatly improves the overall operation management level.

[0019] Furthermore, the system is adaptively adjusted and controlled according to the results predicted by the model, and the steps are as follows: When the operating time is less than the set time and the equipment model does not trigger an update, continue the adaptive dynamic control; otherwise, update the aging parameters of the cooling tower, water pump and heat exchanger, re-collect the operating data of the cooling tower and other equipment and re-train the model.

[0020] Compared with the related art, the dynamic control method of the data center cooling system adapted to equipment aging proposed by the present invention has the following beneficial effects: Based on the data set obtained by the energy consumption model of the data center, the present invention combines the long short-term memory network method in deep learning to predict the performance of the equipment and update the aging factor. By real-time monitoring of the equipment operation status, the present invention can collect data feedback in a timely manner and adjust the equipment parameters according to these data. This real-time adjustment ensures that the equipment always operates in the best working conditions, thereby optimizing the overall energy consumption and energy utilization efficiency of the data center. By combining the data center energy consumption model with LSTM technology, the present invention not only improves the accuracy of equipment performance prediction, but also optimizes equipment aging management, providing an innovative solution for energy consumption optimization and equipment health management of the data center.

[0021] The present invention is not only applicable to the design and construction of new data centers, but also to the transformation and upgrading of existing data centers. For new data centers, the control method of the present invention can be integrated into the design stage to optimize the layout and configuration of the cooling system and improve the overall energy efficiency; for existing data centers, the control system software and hardware facilities can be upgraded and the method of the present invention can be applied to carry out energy-saving transformation. Whether it is a new construction or transformation project, the present invention can significantly improve the energy utilization rate of the data center and reduce operating costs, and has broad application prospects and practical value.

[0022] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A dynamic control method for a data center cooling system adapted to equipment aging, characterized in that: The following steps are involved: S1. Build a control energy consumption model based on the data center data. According to the control equation of the data center, a control energy consumption model is constructed. The model consists of four parts: cold load calculation, air conditioning terminal module calculation, cold source module calculation and energy consumption module calculation; The cooling load calculation includes the heat transfer of the computer room enclosure structure, the heat dissipation of the server, the heat dissipation of the PDU and UPS, and the heat dissipation of the lighting and dehumidification equipment, so as to calculate the total cooling load Q hp ; The air conditioning terminal module calculates the cooling capacity Q of the air conditioning terminal using the ε-NTU method crah ; The cold source module calculates the cooling capacity Q of the cooling tower ctw , cooling capacity of chiller Q chiller and the COP of the chiller, and determine the operating mode of the equipment; the energy consumption module calculates the number of equipment turned on based on the equipment parameters and operating parameters; S2. Obtain key parameter information of the device. Real-time collection of key equipment operation data, cooling tower cooling water outlet temperature T cw_out , Cooling tower return water temperature T cw_in 、Air volume of cooling tower m cta and cooling tower water flow m ctw ; Cold water supply temperature T ch_in , cold water return temperature T ch_out , air supply temperature at the air conditioning terminal Tcrah_in, air return temperature at the air conditioning terminal T crah_out ; Collect equipment operation data once an hour, and collect data for a set period of time for training and testing deep learning models; S3: Equipment performance evaluation, The cooling efficiency is used as the evaluation index of the cooling tower performance. The cooling efficiency E is calculated using the following formula, where Ts represents the wet bulb temperature at that time: ， The efficiency calculation formula of the cold water pump and the cooling water pump is as follows: ， Where ρ is the density of water in kg / m3, g is the acceleration of gravity in m / s2, Q is the water flow in m3 / s, H is the pump head in m, and P represents the shaft power in kilowatts (kW); The heat exchanger efficiency calculation formula is as follows: ， in It refers to the real-time water flow through the heat exchanger, in kg / s, mmax refers to the maximum water flow through the heat exchanger, in kg / s, CP refers to the specific heat capacity of water, in J / (kg•℃), delta_T refers to the theoretical maximum temperature difference of mmax through the heat exchanger, in ℃; Use COP to judge the performance of the chiller. The COP calculation formula is as follows: ， in Refers to the cooling capacity of the chiller, in kW. Refers to the electrical energy consumed by the chiller, in kW.

2. The method for dynamically controlling a data center cooling system that adapts to equipment aging according to claim 1, characterized in that: The key parameter information of the device obtained in step S2 is used as a data set of a long short-term memory network (LSTM) model for training and testing the model, and specifically includes the following steps: S2.

1. Data preprocessing, The collected data is preprocessed to remove abnormal data caused by external force majeure factors such as equipment maintenance and equipment damage. Then the maximum and minimum normalization method is used to normalize the data. The return water temperature and the outlet water temperature are selected as important features of the cooling tower for normalization. The inlet water temperature and the outlet water temperature of the heat exchanger are also normalized. The standardized data is converted into a time series format suitable for LSTM model input. This step requires defining the time step (time_step), that is, how many time points each input sample contains. The maximum and minimum normalization formula is as follows: ， in is the original value of the ith data point, and are the minimum and maximum values ​​of the data set; S2.2, Model training, Since the equipment needs to be dynamically and adaptively adjusted according to the operating data, the long short-term memory network (LSTM) in deep learning can be used to achieve prediction. The specific LSTM unit formulas are shown in formulas 6 to 11. The historical operating parameter data set is divided into a training set and a validation set in a ratio of 7:

3. The mean_squared_error is imported from the sklearn.metrics library in Python as an evaluation indicator of the model quality. ， ， ， ， ， ， in is the sigmoid activation function, It's the Gate of Forgetfulness. is the input gate, is the candidate state, is the cell state, is the output gate; According to step S3, we can know that: the key parameters of cooling tower operation are obtained: the return water temperature T of the cooling tower cw_in , cooling water flow m ctw and the cooling tower outlet water temperature T cw_out , The model contains two LSTM layers and one fully connected layer. To prevent overfitting, a Dropout layer is added after the LSTM layer. The Dropout layer randomly discards some neurons during the training process, thereby enhancing the generalization ability of the model. After the model structure is built, the model needs to be compiled. When compiling, the loss function, optimizer, and evaluation index need to be specified. The loss function selects the mean square error (MSE). The formula is as follows: ， in is the true value, is the predicted value, N is the number of samples, and the optimizer used is the Adam optimizer. The Adam optimizer combines the advantages of the gradient descent method and the momentum method. Its update rules are as follows: ， in is the current parameter, is the learning rate, which can be set according to needs. is the first moment estimate of the gradient, is the second moment estimate of the gradient, It is an extremely small number set to prevent the denominator from being 0. The evaluation index uses the commonly used mean square error as an indicator to evaluate performance.

3. The method for dynamically controlling a data center cooling system that adapts to equipment aging according to claim 2, characterized in that: The specific steps of device parameter prediction and model update strategy are as follows: The equipment is predicted in real time based on the constructed LSTM model; the cooling water outlet temperature T of the cooling tower in the past n hours is used cw_out , cooling water return temperature of cooling tower T cw_in , cold water supply temperature T ch_in , cold water return temperature T ch_out Perform data preprocessing, standardize the data, convert the data into time series, and then input it into the constructed LSTM model to predict the efficiency of the equipment; By real-time monitoring of the cooling tower's outlet and return water temperatures, cold water supply and outlet temperatures, pump head and chiller's power consumption key operating parameters, combined with the prediction model, the system can dynamically calculate the equipment's instant efficiency; this process not only promotes the continuous optimization of the model and control strategy, but also ensures that the equipment can operate in the best state; when the difference between the actual measured value and the predicted value exceeds the preset threshold, the system will automatically adjust the relevant equipment parameters until it returns to the standard operating conditions; and an advanced rolling update mechanism is used to continuously improve the drive equipment model; the update formula is as follows: ， ， ， in is the predicted energy consumption of the chiller, is the temperature difference of the liquid passing through the chiller, in °C, is the mass of liquid flowing into the chiller, in kg / s, and is the weight trained by the model, Bias, is the predicted energy consumption of the fan, where , and is the weight parameter obtained through model training; b is the bias; is the heat transfer efficiency of the heat exchanger, α1 and α2 are the convection heat transfer coefficients of the fluids on both sides, δ is the plate thickness in m, and λ is the thermal conductivity of the plate material. and Usually refers to the convection heat transfer area of ​​the fluid on both sides, unit m 2 , and The weight parameters and bias are obtained through model training; This mechanism uses a deep learning algorithm to retrain the original model structure based on the collected real-time operating parameters of the equipment and makes adjustments according to the law of conservation of energy.

4. The method for dynamically controlling a data center cooling system that adapts to equipment aging according to claim 3, characterized in that: According to the results predicted by the model, the system is adaptively adjusted and controlled. The steps are as follows: When the operating time is less than the set time and the equipment model does not trigger an update, continue the adaptive dynamic control; otherwise, update the aging parameters of the cooling tower, water pump and heat exchanger, re-collect the operating data of the cooling tower and other equipment and re-train the model.

Images