Data center combined cooling heating and power system and control method thereof

Abstract

This invention provides a combined cooling and heating system for data centers and its control method. By integrating historical data sensing, multi-step prediction, fuzzy decision-making, and real-time control, it achieves intelligent, efficient, and economical regulation of the data center's heating and cooling environment. The system possesses strong predictive and adaptive capabilities, enabling it to respond in advance to load and environmental changes, smoothly switch operating modes, and significantly improve system stability and energy efficiency. In particular, by introducing a heat recovery mechanism, it achieves the cascade utilization of waste heat, further reducing dependence on external energy sources and carbon emissions. Furthermore, the method combining the LSTM prediction model and the fuzzy logic decision-making model used in this invention effectively solves the control challenges of large time delays and nonlinearities in thermal systems, demonstrating good engineering applicability and promotional value.

Description

Technical Field

[0001] This invention relates to the field of data center air conditioning technology, and in particular to a data center combined cooling and heating system and its control method. Background Technology

[0002] As the core of digital infrastructure, data centers face increasingly prominent energy consumption issues, especially under extreme climatic conditions, where traditional single-cooling systems struggle to meet year-round operational needs. In frigid regions, heating is required in winter to maintain a stable environment in the server room, while efficient cooling is needed in summer to dissipate heat generated by equipment. Therefore, a combined cooling and heating system capable of addressing both heating and cooling requirements is urgently needed. Most existing data center air conditioning systems only have cooling capabilities, and the few systems with heating functions mostly rely on traditional methods such as electric heating or boilers, resulting in low energy efficiency, high carbon emissions, and a lack of dynamic coordination with IT loads and the external environment.

[0003] Currently, some data centers have attempted to introduce combined cooling, heating, and power (CCHP) systems, using equipment such as gas turbines to achieve combined power, heat, and cooling. However, many problems still exist: First, most systems rely on preset thresholds for mode switching, lacking the ability to predict future load and environmental changes, resulting in delayed response and low energy efficiency. Second, existing control strategies are mostly based on real-time feedback, failing to fully utilize historical data for learning and optimization, thus limiting the system's intelligence level. Third, there is a lack of a coordinated scheduling mechanism for multiple modes (such as mechanical cooling, free cooling, and heat recovery heating), making it difficult to achieve optimal energy efficiency while ensuring accurate temperature control in the data center. In addition, traditional methods perform poorly when dealing with nonlinear, large-delay thermal processes, easily causing system oscillations or energy waste.

[0004] Therefore, there is an urgent need in this field for a data center environment control system that can be based on data-driven prediction, support intelligent switching of multiple modes, and take into account both combined cooling and heating and energy efficiency optimization, in order to address the shortcomings of existing technologies in terms of adaptability, intelligence, and economy. Summary of the Invention

[0005] The purpose of this invention is to provide a data center combined cooling and heating system and its control method to solve the problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] This invention provides a control method for a data center combined cooling and heating system, comprising:

[0008] S1. Collect historical operational and environmental data from the data center, and preprocess the collected data to obtain standardized time series data;

[0009] S2. Based on the standardized time series data, construct a load forecasting model for predicting IT equipment load and a mode decision model for determining the operating mode of the decision system;

[0010] S3. The load forecasting model is used to predict the power consumption of IT equipment in the future period, and the prediction results are combined with environmental data and input into the mode decision model to output the optimal operating mode instruction.

[0011] S4. Based on the optimal operating mode instruction, generate corresponding equipment control signals to jointly control each execution device in the combined cooling and heating system; wherein, the combined cooling and heating system includes mechanical refrigeration mode, free refrigeration mode, direct heating mode and heat recovery heating mode.

[0012] Preferably, in step S1, the historical operating data includes IT equipment power consumption, return air temperature, supply air temperature, compressor power, water pump frequency, cooling tower outlet water temperature, and valve opening; the environmental data includes outdoor dry-bulb temperature, wet-bulb temperature, humidity, wind speed, and solar radiation intensity.

[0013] Preferably, in step S1, the preprocessing of the collected data includes outlier removal, filtering and smoothing, timestamp alignment, and data normalization; wherein, outlier removal adopts the Raida criterion, filtering and smoothing adopts the weighted moving average method, and normalization adopts the minimum-maximum normalization method.

[0014] Preferably, in step S2, the load forecasting model is a long short-term memory network model, whose input features include historical IT load, outdoor temperature, and time periodic encoding, and whose output is the future load forecasting model. IT load forecast values ​​at each time point;

[0015] The cell state update formula for the LSTM model is as follows:

[0016] ;

[0017] in, Output for the forget gate. For input gate output, This is the hidden state from the previous moment. For the current input, and These are the model parameters.

[0018] Preferably, in step S2, the mode decision model is a fuzzy logic control system, with the input being the predicted IT load and outdoor temperature, and the output being the system operating mode;

[0019] The fuzzy logic control system includes a fuzzification interface, a fuzzy rule base, an inference engine, and a defuzzification interface;

[0020] The fuzzy rule base contains multiple IF-THEN rules, which are used to determine the operating mode based on the fuzzy linguistic values ​​of the input variables.

[0021] Preferably, the fuzzy rule base includes:

[0022] If the outdoor temperature is extremely low and the IT load is high, the operating mode is heat recovery heating mode;

[0023] If the outdoor temperature is extremely low and the IT load is low, the operating mode is direct heating mode;

[0024] If the outdoor temperature is low or cool, the operating mode is free cooling mode;

[0025] If the outdoor temperature is high or hot, the operating mode will be mechanical cooling mode.

[0026] Preferably, in step S4, the joint control includes forward-looking mode switching control based on prediction results, and PID fine-tuning control based on real-time feedback during mode operation.

[0027] The output of the PID controller is:

[0028] ;

[0029] in, , To set the temperature, This refers to the actual supply air temperature. , , These are PID parameters.

[0030] The present invention also provides a data center combined cooling and heating system, and a control method for implementing the above-mentioned data center combined cooling and heating system, comprising:

[0031] The data acquisition layer includes several indoor and outdoor sensors and equipment status sensors, which are used to collect environmental and equipment operation data in real time;

[0032] The edge processing and prediction layer includes a data preprocessing module, a historical database, a load forecasting model, and a pattern decision model, which are used for data processing, model building, and operational decision-making.

[0033] The intelligent execution layer, including the intelligent control core and the actuator driver module, is used to generate and issue control commands;

[0034] A combined cooling and heating air conditioning system includes hardware devices with multiple operating modes that execute corresponding operations in response to control commands.

[0035] Preferably, the combined cooling and heating air conditioning system further includes a heat recovery device for recovering waste heat generated by the data center and using it for heating or defrosting.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein when the processor executes the computer program, it implements the control method of the above-mentioned data center combined cooling and heating system.

[0037] The present invention achieves the following beneficial technical effects compared to the prior art:

[0038] This invention provides a combined cooling and heating system for data centers and its control method. By integrating historical data sensing, multi-step prediction, fuzzy decision-making, and real-time control, it achieves intelligent, efficient, and economical regulation of the data center's cooling and heating environment. The system possesses strong predictive and adaptive capabilities, enabling it to respond in advance to load and environmental changes, smoothly switch operating modes, and significantly improve system stability and energy efficiency. In particular, by introducing a heat recovery mechanism, it achieves the cascade utilization of waste heat, further reducing dependence on external energy sources and carbon emissions. Furthermore, the method combining the LSTM prediction model and the fuzzy logic decision-making model used in this invention effectively solves the control challenges of large time delays and nonlinearities in thermal systems, demonstrating good engineering applicability and promotional value. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart of the control method for a data center combined cooling and heating system provided by the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] The purpose of this invention is to provide a combined cooling and heating system for data centers and its control method, applicable to various types of data centers, especially with significant energy-saving effects in extremely cold regions. The system uses data-driven prediction and decision-making to achieve intelligent optimization control of the combined cooling and heating system, effectively improving energy efficiency and reducing operating costs and carbon emissions.

[0043] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Example 1:

[0045] This embodiment provides a data center combined cooling and heating system, including a data acquisition layer, an edge processing and prediction layer, an intelligent execution layer, and a combined cooling and heating air conditioning system.

[0046] Specifically, the data acquisition layer includes various sensors deployed indoors and outdoors, such as indoor return air temperature sensors, supply air temperature sensors, cabinet inlet and outlet air temperature sensors, outdoor dry and wet bulb temperature sensors, wind speed sensors, and solar radiation sensors. It also includes sensors that monitor equipment operating status, such as compressor power sensors, water pump frequency sensors, cooling tower outlet water temperature sensors, and valve opening sensors. These sensors are connected to the data acquisition and preprocessing module via fieldbus (such as Modbus or BACnet) or I / O modules.

[0047] Furthermore, the data acquisition and preprocessing module is responsible for cleaning, filtering, aligning, and normalizing the raw data.

[0048] In one specific embodiment, outlier removal employs the Raida criterion (3σ criterion), that is, for a given data point... If its deviation from the mean μ exceeds 3 times the standard deviation σ, it is judged as an outlier and replaced by a moving average; the filtering process uses a weighted moving average method, the formula of which is:

[0049] ;

[0050] in The weighting coefficients can be adjusted according to the actual data characteristics; normalization uses the minimum-maximum normalization method to map the data to the [0,1] interval, as shown in the formula:

[0051] ;

[0052] The processed data is stored in the historical database and then input into the joint power supply prediction model module.

[0053] Furthermore, the combined heat and power (CHP) forecasting model module includes an IT load forecasting model and a system mode decision model.

[0054] In one specific embodiment, the IT load forecasting model employs a Long Short-Term Memory (LSTM) network. Its input features include historical IT load, outdoor temperature, and time-periodic characteristics (e.g., converting timestamps into sine and cosine codes to reflect periodic patterns). The core computational process of the LSTM unit includes:

[0055] Forgotten Gate: ;

[0056] Input Gate: ; ;

[0057] Cell status update: ;

[0058] Output gate: ; ;

[0059] Where W and b are the model training parameters, and σ is the sigmoid activation function. The model is trained using historical data, with mean squared error (MSE) as the loss function, and the parameters are optimized using the backpropagation algorithm.

[0060] In one specific embodiment, the system mode decision model employs a fuzzy logic control system. Its inputs are the predicted IT load and outdoor temperature, and its output is the system operating mode. First, fuzzification processing is performed to convert precise input values ​​into fuzzy linguistic values. For example, IT load can be categorized as {low, medium, high}, and outdoor temperature as {extremely cold, cold, cool, moderate, warm, hot}. The fuzzy rule base contains multiple IF-THEN rules, for example:

[0061] If the outdoor temperature is extremely cold and the IT load is high, then the mode is heat recovery heating mode.

[0062] If the outdoor temperature is extremely cold and the IT load is low, then the mode is direct heating mode.

[0063] If the outdoor temperature is low or cool, THEN mode = free cooling mode;

[0064] If the outdoor temperature is high or hot, then the mode is mechanical cooling mode.

[0065] The inference engine employs the Mamdani fuzzy inference method, and defuzzification uses the centroid method to convert fuzzy outputs into precise mode instructions.

[0066] Furthermore, the intelligent control core receives instructions from the mode decision model and generates specific equipment control signals. The control strategy includes proactive mode switching and real-time fine-tuning. For example, before mode switching, the heat pump is started in advance for preheating or the cooling tower fan is turned off to achieve a smooth transition; during operation, a PID controller is used for closed-loop control of the supply air temperature, and its output is:

[0067] ;

[0068] in, , To set the temperature, This refers to the actual supply air temperature. , , These are PID parameters, which can be tuned using the Ziegler-Nichols method.

[0069] Furthermore, the actuator drive module converts digital control signals into analog or switching signals to drive various devices in the combined cooling and heating (CCHP) air conditioning system. The CCHP system includes multiple operating modes: mechanical refrigeration mode achieves cooling through a compressor, condenser, and evaporator; free cooling mode utilizes outdoor cold air for indirect cooling via a plate heat exchanger; direct heating mode heats the air supply through an electric heater or air source heat pump; and heat recovery heating mode uses a heat pump to upgrade waste heat from servers for heating or defrosting.

[0070] Furthermore, the system also has an online learning function, which regularly adds new data to the historical database and starts incremental training to update the LSTM model parameters to adapt to changes in load and environment.

[0071] Example 2:

[0072] This embodiment provides a control method for a data center combined cooling and heating system, such as... Figure 1 As shown, it includes:

[0073] S1. Collect historical operational and environmental data from the data center, and preprocess the collected data to obtain standardized time-series data. Historical operational data includes IT equipment power consumption, return air temperature, supply air temperature, compressor power, water pump frequency, cooling tower outlet water temperature, and valve opening. Environmental data includes outdoor dry-bulb temperature, wet-bulb temperature, humidity, wind speed, and solar radiation intensity. The preprocessing of the collected data includes outlier removal, filtering and smoothing, timestamp alignment, and data normalization. Outlier removal uses the Laida criterion, filtering and smoothing uses the weighted moving average method, and normalization uses the min-max normalization method.

[0074] S2. Based on standardized time series data, construct a load forecasting model for predicting IT equipment load and a pattern decision model for determining the operating mode of the decision system. The load forecasting model is a long short-term memory network model, whose input features include historical IT load, outdoor temperature, and time periodic encoding, and whose output is the future load. IT load forecast values ​​at each time point;

[0075] The cell state update formula for the LSTM model is:

[0076] ;

[0077] in, Output for the forget gate. For input gate output, This is the hidden state from the previous moment. For the current input, and These are the model parameters.

[0078] The mode decision model is a fuzzy logic control system. The inputs are the predicted IT load and outdoor temperature, and the output is the system operation mode. The fuzzy logic control system includes a fuzzification interface, a fuzzy rule base, an inference engine, and a defuzzification interface. The fuzzy rule base contains multiple IF-THEN rules, which are used to determine the operation mode based on the fuzzy linguistic values ​​of the input variables.

[0079] The fuzzy rule base includes:

[0080] If the outdoor temperature is extremely low and the IT load is high, the operating mode is heat recovery heating mode;

[0081] If the outdoor temperature is extremely low and the IT load is low, the operating mode is direct heating mode;

[0082] If the outdoor temperature is low or cool, the operating mode is free cooling mode;

[0083] If the outdoor temperature is high or hot, the operating mode will be mechanical cooling mode;

[0084] S3. Utilize the load forecasting model to predict the power consumption of IT equipment in the future period, and combine the prediction results with environmental data, input them into the mode decision model, and output the optimal operating mode instruction;

[0085] S4. Based on the optimal operating mode instruction, generate corresponding equipment control signals to jointly control each execution device in the combined cooling and heating system; wherein, the combined cooling and heating system includes mechanical refrigeration mode, free refrigeration mode, direct heating mode and heat recovery heating mode; the joint control includes forward-looking mode switching control based on prediction results, and PID fine-tuning control based on real-time feedback during mode operation;

[0086] The output of the PID controller is:

[0087]

[0088] in, , To set the temperature, This refers to the actual supply air temperature. , , These are PID parameters.

[0089] Taking a data center in a frigid region as an example, where outdoor temperatures can reach -20℃ in winter, the system predicts the IT load and outdoor temperature for the next 6 hours and decides to adopt a heat recovery heating mode. The control core starts the heat pump unit in advance, using waste heat from the servers to raise the temperature, and maintains the supply air temperature within the set value of 22℃±0.5℃ through a PID controller. Actual operation data shows that the energy saving rate exceeds 30% compared to traditional electric heating methods.

[0090] During the spring and autumn transition seasons, when outdoor temperatures range from 5-15℃, the system switches to a free cooling mode via fuzzy decision-making, utilizing cool outdoor air for cooling through plate heat exchangers, shutting down the compressor, and significantly reducing energy consumption. Simultaneously, PID control adjusts the water pump frequency and valve opening to maintain a stable computer room temperature.

[0091] During the high-temperature period in summer, the system switches to mechanical cooling mode and adjusts the outlet water temperature of the chiller unit in advance based on the forecast results to maximize the energy efficiency ratio while ensuring the cooling effect.

[0092] It should be noted that the above embodiments are merely illustrative examples, and the scope of protection of the present invention is not limited thereto. Those skilled in the art can adjust the sensor type, model structure, control parameters, etc., based on the concept of the present invention and in conjunction with specific application scenarios; all such modifications and substitutions fall within the scope of protection of the present invention.

[0093] This invention has illustrated its principles and implementation methods using specific examples. The descriptions of these embodiments are merely illustrative of the method and its core ideas; furthermore, those skilled in the art will recognize that modifications may be made to the specific implementation methods and application scope based on the principles of this invention. Therefore, the content of this specification should not be construed as limiting the invention.

Claims

1. A control method for a data center combined cooling and heating system, characterized in that, include: S1. Collect historical operational and environmental data from the data center, and preprocess the collected data to obtain standardized time series data; S2. Based on the standardized time series data, construct a load forecasting model for predicting IT equipment load and a mode decision model for determining the operating mode of the decision system; S3. The load forecasting model is used to predict the power consumption of IT equipment in the future period, and the prediction results are combined with environmental data and input into the mode decision model to output the optimal operating mode instruction. S4. Based on the optimal operating mode instruction, generate corresponding equipment control signals to jointly control each execution device in the combined cooling and heating system; wherein, the combined cooling and heating system includes mechanical refrigeration mode, free refrigeration mode, direct heating mode and heat recovery heating mode.

2. The control method for a data center combined cooling and heating system according to claim 1, characterized in that, In step S1, the historical operating data includes IT equipment power consumption, return air temperature, supply air temperature, compressor power, water pump frequency, cooling tower outlet water temperature, and valve opening; the environmental data includes outdoor dry bulb temperature, wet bulb temperature, humidity, wind speed, and solar radiation intensity.

3. The control method for a data center combined cooling and heating system according to claim 1, characterized in that, In step S1, the preprocessing of the collected data includes outlier removal, filtering and smoothing, timestamp alignment, and data normalization. Outlier removal uses the Laida criterion, filtering and smoothing uses the weighted moving average method, and normalization uses the min-max normalization method.

4. The control method for a data center combined cooling and heating system according to claim 1, characterized in that, In step S2, the load forecasting model is a long short-term memory network model. Its input features include historical IT load, outdoor temperature, and time periodic encoding, and its output is the future load forecast. The predicted IT load values ​​at each time point; the cell state update formula of the LSTM model is: ;in, Output for the forget gate. For input gate output, This is the hidden state from the previous moment. For the current input, and These are the model parameters.

5. The control method for a data center combined cooling and heating system according to claim 1, characterized in that, In step S2, the mode decision model is a fuzzy logic control system, with the input being the predicted IT load and outdoor temperature, and the output being the system operating mode; The fuzzy logic control system includes a fuzzification interface, a fuzzy rule base, an inference engine, and a defuzzification interface; The fuzzy rule base contains multiple IF-THEN rules, which are used to determine the operating mode based on the fuzzy linguistic values ​​of the input variables.

6. The control method for a data center combined cooling and heating system according to claim 5, characterized in that, The fuzzy rule base includes: If the outdoor temperature is extremely low and the IT load is high, the operating mode is heat recovery heating mode; If the outdoor temperature is extremely low and the IT load is low, the operating mode is direct heating mode; If the outdoor temperature is low or cool, the operating mode is free cooling mode; If the outdoor temperature is high or hot, the operating mode will be mechanical cooling mode.

7. The control method for a data center combined cooling and heating system according to claim 1, characterized in that, In step S4, the joint control includes forward-looking mode switching control based on prediction results, and PID fine-tuning control based on real-time feedback during mode operation. The output of the PID controller is: ;in, , To set the temperature, This refers to the actual supply air temperature. , , These are PID parameters.

8. A data center combined cooling and heating system, comprising a control method for implementing the data center combined cooling and heating system as described in any one of claims 1-7, characterized in that, include: The data acquisition layer includes several indoor and outdoor sensors and equipment status sensors, which are used to collect environmental and equipment operation data in real time; The edge processing and prediction layer includes a data preprocessing module, a historical database, a load forecasting model, and a pattern decision model, which are used for data processing, model building, and operational decision-making. The intelligent execution layer, including the intelligent control core and the actuator driver module, is used to generate and issue control commands; A combined cooling and heating air conditioning system includes hardware devices with multiple operating modes that execute corresponding operations in response to control commands.

9. The data center combined cooling and heating system according to claim 8, characterized in that, The combined cooling and heating air conditioning system also includes a heat recovery device for recovering waste heat generated by the data center and using it for heating or defrosting.

10. An electronic device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the control method for the data center combined cooling and heating system as described in any one of claims 1-7.

Images