Optimization control method and device for refrigerating system of data center

Abstract

The invention relates to an optimal control method and device for a refrigerating system of a data center. The method comprises the following steps: S1, acquiring operation data of cold source side equipment and tail end side equipment; and S2, correspondingly training each equipment energy consumption model of the cold source side and the tail end side based on the operation data, and obtaining a system total energy consumption model according to each trained equipment energy consumption model. And training an unstable operation judgment model of the refrigeration system based on the operation data. And S3, based on the system total energy consumption model, calculating a total energy consumption value under different operation parameter combinations according to a preset optimization algorithm. And S4, judging whether the system is in an unstable operation state or not under different operation parameter combinations. And S5, taking the operation parameter combination which has the minimum total energy consumption value and is not in the unstable operation state as the optimal operation parameter combination. On the premise of guaranteeing the safety of the refrigerating system of the data center, the energy-saving effect of the refrigerating system can be maximized, the operation stability of the system can be guaranteed, and the operation life of equipment is effectively prolonged.

Description

Technical Field

[0001] The present invention relates to the field of data center control technology, and in particular to a data center refrigeration system optimization control method and device. Background Art

[0002] With the rapid development of new technologies such as artificial intelligence, 5G, cloud computing, and the Internet of Things, diversified computing demand scenarios such as high-tech, digital transformation, and terminal consumption continue to emerge. As the physical base for carrying such demand applications, the scale of data centers has been rapidly increased. With the large-scale construction of data centers, the energy consumption problem of data centers has gradually been exposed. Faced with the contradiction between high energy consumption and high growth of data centers, as well as the contradiction between increasingly stringent policies and growing demand, it is necessary to take a variety of measures to reduce the operating energy consumption of data centers and achieve energy-saving operation of data centers. As an important part of the data center, the refrigeration system undertakes the important task of ensuring the safe operation of the data center, and its energy consumption accounts for about 40%. Therefore, energy-saving control of the refrigeration system is of great significance to reducing the energy consumption of the entire data center.

[0003] However, the current energy-saving control of data center cooling systems often focuses on a single device, or a cold source system, or a terminal system. However, energy saving of a single device or a single system does not mean that the overall system efficiency is optimal. In fact, every change in a control parameter will affect the overall energy consumption of the system. Under each load condition, there is a set of control parameters that can minimize the overall energy consumption. Moreover, the relevant technology cannot guarantee that the system is in a stable operating state, which will bring the risk of shortening the operating life of the equipment. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a data center refrigeration system optimization control method and device.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for optimizing and controlling a refrigeration system of a data center, wherein the refrigeration system includes a cold source side device for providing a cold source and a terminal side device for heat dissipation, and the method includes the following steps:

[0006] S1. Obtaining the operating data of the cold source side equipment and the terminal side equipment;

[0007] S2, training energy consumption models of each device on the cold source side and the terminal side based on the operation data, and obtaining a total energy consumption model of the system according to the trained energy consumption models of each device; and training an unstable operation judgment model of the refrigeration system based on the operation data;

[0008] S3. Based on the system total energy consumption model, calculate the total energy consumption values ​​under different operating parameter combinations according to a preset optimization algorithm;

[0009] S4. Based on the unstable operation determination model, determine whether the system is in an unstable operation state under different operation parameter combinations;

[0010] S5. The operating parameter combination with the smallest total energy consumption value and not in an unstable operating state is taken as the optimal operating parameter combination.

[0011] Furthermore, in the data center refrigeration system optimization control method of the present invention, the method further includes:

[0012] S6. Send the optimal parameter combination to each device on the cold source side and the terminal side for execution.

[0013] Further, in the data center refrigeration system optimization control method of the present invention, in step S2, obtaining the system total energy consumption model according to the energy consumption models of the various devices includes:

[0014] The energy consumption models of the various devices are added together to obtain the total energy consumption model of the system.

[0015] Further, in the data center refrigeration system optimization control method of the present invention, the preset optimization algorithm includes preset constraints, and step S3 includes:

[0016] S31, using the system total energy consumption model as the objective function of the preset optimization algorithm;

[0017] S32, determining the parameter setting value of each device that meets the preset constraint condition, and forming a plurality of different operating parameter combinations; each operating parameter combination includes the parameter setting value of each device on the cold source side and the terminal side;

[0018] S33. Calculate the total energy consumption values ​​corresponding to different operating parameter combinations according to the objective function, and sort them from low to high according to the total energy consumption values.

[0019] Further, in the data center refrigeration system optimization control method of the present invention, the energy consumption model of each device on the cold source side and the terminal side obtained by training based on the operation data in step S2 includes:

[0020] S211, dividing the operating data into a training data set and a verification data set according to a preset ratio;

[0021] S212, using the training data set to train the energy consumption model of the corresponding equipment;

[0022] S213: Use the verification data set to verify the accuracy of the corresponding device energy consumption model. If the accuracy requirement is met, it is confirmed that the training of the device energy consumption model is completed.

[0023] Further, in the data center refrigeration system optimization control method of the present invention, step S2 also includes:

[0024] S214: monitor the accuracy of the energy consumption models of the various devices in real time, and retrain a device energy consumption model when the model does not meet the preset accuracy requirement.

[0025] Furthermore, in the data center refrigeration system optimization control method of the present invention, before step S2 and after step S1, it also includes:

[0026] The operation data is filtered, cleaned, aggregated and quality optimized.

[0027] Further, in the data center refrigeration system optimization control method of the present invention, in step S2, the working principle of the unstable operation determination model is as follows:

[0028] S221, according to the operating data, the indicator S DT Perform the calculation:

[0029] S DT =w1·DT Roc +w2 DT Dev ;

[0030] DT Roc =DT ratio / DT ratio_normal ;

[0031] DT Dev =(DT chiller_t -DT chiller_set ) / DT chiller_set ;

[0032] DT ratio =(DT chiller_t -DT chiller_t-TDT ) / T DT ;

[0033] w1+w2=1;

[0034] Among them, the indicator S DT DT is the criterion for determining unstable operation of the refrigeration system. Roc DT is the fluctuation of superheat of chiller compressor. Dev DT is the deviation of the superheat of the chiller compressor. ratio is the fluctuation rate of the superheat of the chiller compressor, DT ratio_normal is the standard value of the fluctuation rate of the superheat of the chiller compressor, w1 is the weight coefficient of the fluctuation rate of the superheat of the chiller compressor, w2 is the weight coefficient of the deviation of the superheat of the chiller compressor, T is the data collection period, TDT is a time period, DT chiller_t is the superheat of the chiller compressor at the current moment, DT chiller_t-TDT is T before the current moment DT The superheat of the chiller compressor at the moment, DT chiller_set It is the standard setting value of superheat of chiller compressor;

[0035] S222. Determine whether S DT Continuously greater than or equal to S DT_set ; If so, it is determined that the system is in an unstable operating state, and the system state STAT is recorded as 1. The corresponding operating parameter combination is the operating parameter combination of the unstable operating state; where S DT_set This is the set value for judging the unstable operation index.

[0036] Further, in the data center refrigeration system optimization control method of the present invention, in step S2, the unstable operation determination model of the refrigeration system is obtained based on the training of the operation data, including:

[0037] S223, combining the obtained system state STAT with the operation data to form a new data set, and dividing the new data set into a training data set and a verification data set according to a preset ratio;

[0038] S224, training the unstable operation determination model using a training data set;

[0039] S225 , using the verification data set to verify the accuracy of the unstable operation determination model, if the accuracy requirement is met, it is confirmed that the training of the unstable operation determination model is completed.

[0040] In addition, the present invention also provides a data center refrigeration system optimization control device, comprising:

[0041] A plurality of communication units, used for respectively receiving the operation data transmitted from the cold source side monitoring system and the terminal side monitoring system;

[0042] A data processing unit, connected to the communication unit, for optimizing the operation data;

[0043] A model training unit, connected to the data processing unit, for training the corresponding energy consumption models of the equipment on the cold source side and the terminal side and the unstable operation judgment model of the refrigeration system according to the optimized operation data;

[0044] An optimization control unit is connected to the model training unit, and is used to obtain a system total energy consumption model based on the trained energy consumption models of each device, and based on the system total energy consumption model, calculate the total energy consumption value under different operating parameter combinations according to a preset optimization algorithm, and based on the unstable operation judgment model, judge whether the system is in an unstable operating state under different operating parameter combinations, and take the operating parameter combination with the smallest total energy consumption value and not in an unstable operating state as the optimal operating parameter combination.

[0045] The data center refrigeration system optimization control method and device implementing the present invention have the following beneficial effects: the present invention can maximize the energy-saving effect of the refrigeration system while ensuring the safety of the data center refrigeration system by uniformly processing the operating data on the cold source side and the terminal side, that is, it can maximize the energy-saving potential of global optimization, and can also ensure the stability of system operation and effectively extend the operating life of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0047] Figure 1 It is a flow chart of a data center refrigeration system optimization control method provided by an embodiment of the present invention;

[0048] Figure 2 is a schematic diagram of a typical data center cooling system provided by an embodiment of the present invention;

[0049] Figure 3 It is a flow chart of a data center refrigeration system optimization control method provided by an embodiment of the present invention;

[0050] Figure 4 It is a flow chart of a data center refrigeration system optimization control method provided by an embodiment of the present invention;

[0051] Figure 5 It is a flow chart of a data center refrigeration system optimization control method provided by an embodiment of the present invention;

[0052] Figure 6 It is a flow chart of a data center refrigeration system optimization control method provided by an embodiment of the present invention;

[0053] Figure 7 It is a structural schematic diagram of a data center refrigeration system optimization control device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the specific implementation methods of the present invention are now described in detail with reference to the accompanying drawings. In the following description, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed", "set" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral body; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. When an element is referred to as being "on" or "under" another element, the element can be "directly" or "indirectly" located on the other element, or there may be one or more intermediate elements. The terms "first", "second", "third", etc. are only for the convenience of describing the present technical solution, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first", "second", "third", etc. can explicitly or implicitly include one or more of the features. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0055] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0056] The refrigeration system includes cold source side equipment for providing cold source and terminal side equipment for heat dissipation. That is, the data center refrigeration system consists of a cold source part that provides cold source and a terminal part that provides terminal heat dissipation. In current data centers, the data monitoring of the refrigeration system consists of two independent monitoring systems, of which the cold source side is monitored by the building automation control system, also known as the BA system, and the terminal side heat dissipation part is monitored by the power and environment monitoring system, also known as the dynamic environment system.

[0057] Related technologies use deep learning methods to establish energy consumption models for each device in the data center cooling system and use optimization algorithms to infer the operating parameter setting values ​​of each device under different load conditions when the total energy consumption of the system is the smallest, which can achieve the goal of overall energy saving of the system. However, the following two reasons limit the energy saving potential of this global optimization method and bring the risk of unstable system operation.

[0058] First, due to the information isolation and closed control between the two monitoring systems, it is impossible to achieve global energy-saving operation control of the refrigeration system, limiting the room for improvement in operating energy efficiency. Second, when the optimal operating parameter combination output by the global optimization method is used, although the energy consumption of the refrigeration system can be minimized, it cannot guarantee that the system is in a stable operating state, which will bring the risk of shortening the operating life of the equipment. Therefore, in order to ensure that the system is in a stable state, engineers usually strictly limit the upper and lower thresholds of the set parameters, but this also results in the global optimization method being unable to fully exert all the energy-saving potential.

[0059] In a preferred embodiment, reference Figure 1 The data center refrigeration system optimization control method of this embodiment includes the following steps:

[0060] S1. Obtain the operating data of the cold source side equipment and the terminal side equipment. Figure 2 The figure shows a schematic diagram of a typical data center refrigeration system. The cold source side equipment mainly includes cooling towers, cooling pumps, chillers, and refrigeration pumps. The terminal side equipment mainly includes precision air conditioners at each terminal. It can be understood that the acquired operating data can include at least two parts. The first part of the data can be obtained from the BA system of the data center, and can be collected by the BA system and uploaded to the communication unit 2 of this control device. The second part of the data can be obtained from the dynamic environment system of the data center, and can be collected by the dynamic environment system and uploaded to the communication unit 3 of this control device. Exemplarily, the first part of the data may include:

[0061] 1) Outdoor temperature t otdr , Outdoor relative humidity RH otdr .

[0062] 2) The chilled water inlet temperature of each chiller t chw_chiller_in , chiller outlet water temperature t chw_chiller_out , chilled water flow rate of chiller m chw_chiller , Cooling water inlet temperature of chiller t cw_chiller_in , Cooling water outlet temperature of chiller t cw_chiller_out , Cooling water flow rate of chiller m cw_chiller , the superheat DT of the compressor of the chiller chiller , Chiller operating current percentage CR chiller , Chiller operating power P chiller .

[0063] 3) Each chilled water pump inlet temperature t chw_pump_in , Chilled water pump outlet temperature t chw_pump_out , chilled water pump frequency f chw_pump , Chilled water pump operating power P chw_pump .

[0064] 4) Inlet water temperature of each cooling water pump tcw_pump_in , Cooling water pump outlet temperature t cw_pump_out , cooling water pump frequency f cw_pump , Cooling water pump operating power P cw_pump .

[0065] 5) Inlet water temperature of each cooling tower t w_ct_in , Cooling tower outlet water temperature t w_ct_out , cooling tower fan frequency f ct_fan , Cooling tower inlet wind speed v a_ct_fan , Cooling tower fan operating power P ct_fan .

[0066] The second part of the data may include: the water inlet temperature t of each terminal precision air conditioner w_pac_in , terminal precision air conditioning water outlet temperature t w_pac_out , Terminal precision air conditioning water flow m w_pac , Terminal precision air conditioning inlet temperature t a_pac_in , Terminal precision air conditioning outlet temperature t a_pac_out , terminal precision air conditioning fan frequency f pac_fan , Terminal precision air conditioning air supply speed v a_pac .

[0067] Alternatively, when the corresponding algorithm model requires one or more necessary data, the operating data will include the third part of data. That is, the third part of data comes from the operating data collected by the newly added sensors on site according to specific needs. For example, if the data necessary for the corresponding algorithm model is lacking, the corresponding sensors need to be added on site. For example, if the data on outdoor humidity is lacking, a new humidity sensor can be added; for example, if the data on the cooling tower inlet wind speed is lacking, a new wind speed sensor can be added. This part of the sensor data constitutes the third part of data. Data can be acquired by connecting the acquisition unit 4 of the control device to the newly added sensor.

[0068] In some embodiments, before step S2 and after step S1, the process further includes: filtering, cleaning, aggregating and quality optimization related processing of the operation data. That is, this step performs relevant optimization processing on the obtained operation data, exemplarily, with Figure 7For example, the first and second parts of the data can be processed by the communication data processing unit 5-2, including data filtering, cleaning, aggregation and quality optimization. The third part of the data is processed by the acquisition data processing unit 5-3, including data filtering, cleaning, aggregation and quality optimization. The optimized data is transmitted and synchronized to the data collection unit 5-1 by the communication processing unit 5-2 and the acquisition processing unit 5-3 respectively, and provided to the northbound device or system through the communication unit 1. It can be understood that the northbound device and system refer to some projects where there are other devices or other integrated systems that need to call the data generated by the method and device of the present invention. It has no affiliation with the refrigeration system described in the present invention and is independent of each other.

[0069] S2. Based on the operation data, the energy consumption models of the equipment on the cold source side and the terminal side are correspondingly trained, and the total energy consumption model of the system is obtained according to the trained energy consumption models of the equipment. In addition, the unstable operation judgment model of the refrigeration system is trained based on the operation data.

[0070] In some embodiments, the energy consumption models of the devices on the cold source side and the terminal side are obtained by training based on the operation data in step S2, including:

[0071] S211, dividing the running data into a training data set and a verification data set according to a preset ratio. That is, the data can be divided into a training data set and a verification data set in a random manner, wherein the default ratio of the training data set to the verification data set can be preferably 8:2. The ratio can be adjusted by the data set division ratio parameter.

[0072] S212: Use the training data set to train the corresponding equipment energy consumption model. That is, use the training data set to train the equipment model. The algorithm used in the training will not be described here in detail, and reference may be made to related technologies.

[0073] In a specific embodiment, the obtained device model is described in detail as follows:

[0074] Energy consumption model of chiller: P chiller_predict =f(t chw_chiller_in ,t chw_chiller_out ,m chw_chiller ,t cw_chiller_in ,t cw_chiller_out ,m cw_chiller, CR chiller ) (1)

[0075] Energy consumption model of chilled water pump:

[0076] P chw_pump_predict =f(t chw_pump_in ,t chw_pump_out ,m chw_chiller ,f chw_pump ) (2)

[0077] If missing data is missing, t chw_pump_in Can be obtained by chw_chiller_out Instead of; chw_pump_out Can be obtained by chw_chiller_in replace.

[0078] Energy consumption model of cooling water pump:

[0079] P cw_pump_predict =f(t cw_pump_in ,t cw_pump_out ,m cw_chiller ,f cw_pump ) (3)

[0080] If there is missing data or no corresponding sensor, t cw_pump_in Can be obtained by cw_chiller_out Instead of; cw_pump_out Can be obtained by cw_chiller_in replace.

[0081] Energy consumption model and heat dissipation model of cooling tower:

[0082] P ct_fan_predict =f(t w_ct_fan_in ,t w_ct_fan_out ,t otdr ,RH otdr ,v a_ct_fan ,f ct_fan ) (4)

[0083] Q ct_fan_predict =f(t w_ct_fan_in ,t w_ct_fan_out ,t otdr ,RH otdr ,v a_ct_fan ,f ct_fan ) (5)

[0084] If there is no missing data or no corresponding sensor, v a_ct_fan No need,t w_ct_fan_in Can be obtained by cw_chiller_out Instead, w_ct_fan_out Can be obtained by cw_chiller_in replace.

[0085] Energy consumption model and cooling capacity model of precision air conditioning:

[0086] P pac_predict =f(t w_pac_in ,t w_pac_out ,m w_pac ,t a_pac_in ,t a_pac_out ,f pac_fan ,v a_pac ) (6)

[0087] Q pac_predict =f(t w_pac_in ,t w_pac_out ,m w_pac ,t a_pac_in ,t a_pac_out ,f pac_fan ,v a_pac ) (7)

[0088] If there is no missing data or no corresponding sensor: a_pac It may not be necessary, or it can be calculated from the speed of the precision air conditioner fan; m w_pac It can be approximately calculated by the following formula:

[0089] m w_pac =Σ i Nchiller ·m chw_chiller_i / N pac_run (8)

[0090] Where i is the number of the chiller, N pac_run is the number of precision air conditioners in operation, N chiller is the number of chillers.

[0091] S213, using the verification data set to verify the accuracy of the corresponding device energy consumption model, if the accuracy requirement is met, then confirming that the equipment energy consumption model training is completed. It can be understood that after the accuracy requirement is met, the trained model is deployed to the equipment model unit 7-1.

[0092] Alternatively, refer to Figure 3 , step S2 further includes:

[0093] S214, monitor the accuracy of each device energy consumption model in real time, and retrain the energy consumption model of a device when it does not meet the preset accuracy requirements. In other words, monitor the accuracy of the device model in real time, and automatically retrain and deploy the model when the model accuracy continues to fail to meet the set requirements.

[0094] It should be noted that the working principle of the unstable operation determination model is as follows:

[0095] S221, according to the operating data of the indicator S DT Perform the calculation:

[0096] S DT =w1·DT Roc +w2 DT Dev

[0097] DT Roc =DT ratio / DT ratio_normal

[0098] DT Dev =(DT chiller_t -DT chiller_set ) / DT chiller_set

[0099] DT ratio =(DT chiller_t -DT chiller_t-TDT ) / T DT

[0100] w1+w2=1

[0101] Among them, the index S DT DT is the criterion for determining unstable operation of the refrigeration system. Roc DT is the fluctuation of the superheat of the chiller compressor, dimensionless unit. Dev DT is the deviation of the superheat of the chiller compressor, dimensionless unit. ratio DT is the fluctuation rate of the superheat of the chiller compressor, in ℃ / s. ratio_normal is the standard value of the fluctuation rate of the superheat of the chiller compressor, in ℃ / s, and the default value can be 0.02℃ / s. w1 is the weight coefficient of the fluctuation rate of the superheat of the chiller compressor, and the default value is 0.8. w2 is the weight coefficient of the deviation of the superheat of the chiller compressor, and the default value is 0.2. T is the data collection period, that is, the period of the timestamp, and the default value is 2 seconds. T DT It is a time period, which can be set. The recommended value is 5·T. chiller_t is the superheat of the chiller compressor at the current moment, DT chiller_t-TDT is T before the current moment DT The superheat of the chiller compressor at the moment, DT chiller_set The standard setting value for the superheat of the chiller compressor, usually 5°C.

[0102] S222. Determine whether S DT Continuously greater than or equal to S DT_set If yes, then the system is judged to be in an unstable operating state, and the system state STAT is recorded as 1. The corresponding operating parameter combination is the operating parameter combination of the unstable operating state. DT_set This is the set value for the unstable operation indicator. In other words, the following method is used to determine whether the system is in an unstable operation state: within a continuous 30·T time (multiples can be set, 30 is recommended) S DT Continuously greater than or equal to S DT_set , then the system is considered to be in an unstable state, the system state STAT is recorded as 1, and the rest are 0. At this time, the corresponding parameter operation combination is an unstable operation parameter combination, where SDT_set is a configurable unstable operation index judgment value, and the recommended value is 1.6. As an option, if: 1) the collection time S of a certain data DT <S DT_set If the system is in the process of adjusting operating parameters, the timing will be restarted.

[0103] This embodiment defines an indicator S DT As a criterion for determining unstable system operation, this indicator comprehensively reflects the fluctuations and deviations of the superheat of the chiller's compressor.

[0104] In some embodiments, reference Figure 4 In step S2, the unstable operation determination model of the refrigeration system is obtained by training based on the operation data, including:

[0105] S223, merging the obtained system state STAT with the operation data to form a new data set, and dividing the new data set into a training data set and a verification data set according to a preset ratio. In other words, the training data set and the verification data set can be divided into the training data set and the verification data set in a random manner, wherein the default ratio of the training data set to the verification data set is 8:2, which can be adjusted by the data set division ratio parameter.

[0106] S224: Train the unstable operation determination model using the training data set. It is understood that the algorithm used in the training is not specifically limited, and reference may be made to related technologies.

[0107] Exemplarily, the obtained unstable operation determination model is as follows:

[0108] STAT predict =f(N chiller ,N ct_fan ,t chw_chiller_out ,f chw_pump ,f cw_pump ,f ct_fan ,f pac_fan )

[0109] Where: N ct_fan The number of cooling towers that are turned on (if the system is Figure 7 The one-to-one form shown, then N ct_fan =N chiller ).

[0110] S225, using the verification data set to verify the accuracy of the unstable operation determination model, if the accuracy requirement is met, then confirming that the unstable operation determination model training is complete. Using the verification data set to verify the accuracy of the model, and after meeting the accuracy requirement, the trained model is deployed.

[0111] Alternatively, the method may further include: real-time monitoring of the accuracy of the unstable operation determination model, and automatically retraining and re-deploying when the model accuracy continues to fail to meet set requirements.

[0112] It is understandable that the corresponding accuracy requirements for different equipment energy consumption models may be different or the same, which is determined according to the needs of the user or the refrigeration system, and this solution does not make specific restrictions. Similarly, for the unstable operation judgment model of the refrigeration system, its accuracy requirements are also determined according to the needs of the user or the refrigeration system, and this solution does not make specific restrictions.

[0113] In some embodiments, in step S2, obtaining the total energy consumption model of the system according to the energy consumption models of each device includes: adding the energy consumption models of each device to obtain the total energy consumption model of the system. Based on the above embodiment, according to the energy consumption models of each device on the cold source side and the terminal side of the refrigeration system, the total energy consumption model of the refrigeration system is:

[0114]

[0115] S3. Based on the system total energy consumption model, the total energy consumption values ​​under different operating parameter combinations are calculated according to the preset optimization algorithm.

[0116] refer to Figure 5 The preset optimization algorithm includes preset constraints, and step S3 includes:

[0117] S31. Using the system total energy consumption model as the objective function of the preset optimization algorithm.

[0118] S32, determining the parameter setting value of each device that meets the preset constraint conditions, and forming multiple groups of different operating parameter combinations. Each group of operating parameter combinations includes the parameter setting values ​​of each device on the cold source side and the terminal side.

[0119] S33. Calculate the total energy consumption values ​​corresponding to different operating parameter combinations according to the objective function, and sort them from low to high according to the total energy consumption values.

[0120] In a specific implementation, the specific constraints of the optimization algorithm are as follows:

[0121] f chw_pump_min ≤f chw_pump ≤f chw_pump_max

[0122] f cw_pump_min ≤f cw_pump ≤f cw_pump_max ;

[0123] f ct_fan_min ≤f ct_fan ≤f ct_fan_max ;

[0124] f pac_fan_min ≤f pac_fan ≤f pac_fan_max ;

[0125] t chw_chiller_out_min ≤t chw_chiller_out ≤t chw_chiller_out_max ;

[0126] t a_pac_out_min ≤t a_pac_out ≤t a_pac_out_max ;

[0127] N chiller_min ≤N chiller ≤N chiller_max ;

[0128] N ct_fan_min ≤N ct_fan ≤N ct_fan_max ;

[0129] Q chw_chiller =m chw_chiller ·(t chw_chiller_in -t chw_chiller_out );

[0130] Q cw_chiller =m cw_chiller ·(t cw_chiller_out -t cw_chiller_in );

[0131]

[0132] The upper and lower limits of the specific constraints of the optimization algorithm can be set by the operation and maintenance personnel. In the method described in the present invention, each lower limit value can be significantly lower than the lower limit value of the traditional optimization algorithm, and each upper limit value can be significantly higher than the upper limit value of the traditional optimization algorithm, which can fully tap the energy-saving potential of the optimization algorithm.

[0133] The optimization algorithm is used to calculate the total energy consumption value under different combinations of optimized operating parameter settings and sort them from low to high. The optimized operating parameter settings are as follows:

[0134] 1) Number of refrigeration units turned on N Number of chillers turned on N chiller .

[0135] 2) Number of cooling towers open N ct_fan The setting value, (if the system Figure 2 The one-to-one form shown, then N ct_fan =N chiller ).

[0136] 3) Chiller outlet water temperature t chw_chiller_out The setting value is in oC.

[0137] 4) Chilled water pump frequency f chw_pump The setting value, unit is HZ.

[0138] 5) Cooling water pump frequency f cw_pump The setting value, unit is HZ.

[0139] 6) Cooling tower fan frequency f ct_fan , unit is HZ.

[0140] 7) Terminal precision air conditioning fan frequency f pac_fan , unit is HZ.

[0141] That is, the operating parameter combination setting value [N chiller ,N ct_fan， t chw_chiller_out, f chw_pump, f cw_pump, f ct_fan, f pac_fan 】. The specific optimization algorithm is not limited, and you can refer to related technologies.

[0142] S4. Based on the unstable operation determination model, determine whether the system is in an unstable operation state under different operation parameter combinations.

[0143] It can be understood that according to the unstable operation judgment model obtained, the obtained operating parameter combination setting values ​​[N chiller ,N ct_fan， t chw_chiller_out, f chw_pump, f cw_pump, f ct_fan, f pac_fan 】, to determine whether it is in an unstable operating state. As shown in Table 1, the first parameter combination that is not in the unstable operating area is selected as the optimal operating parameter combination, that is, the first numbered combination with STAT=0. The parameter combination numbered 2 in this table is the optimal operating parameter combination.

[0144] Table 1 Optimization algorithm table

[0145]

[0146] S5. The operating parameter combination with the smallest total energy consumption value and not in an unstable operating state is taken as the optimal operating parameter combination.

[0147] This embodiment can maximize the energy-saving effect of the refrigeration system while ensuring the safety of the data center refrigeration system by uniformly processing the operating data on the cold source side and the terminal side. That is, it can maximize the energy-saving potential of global optimization, ensure the stability of system operation, and effectively extend the operating life of the equipment.

[0148] In some embodiments, reference Figure 6 The method further includes: S6, sending the optimal parameter combination to each device on the cold source side and the terminal side for execution. Taking Table 1 as an example, the parameter combination numbered 2 in this table is the optimal operating parameter combination. The operating parameter combination numbered will be sent to the data set unit 5-1 for standard processing, and then sent to the data center BA system and the data center dynamic environment system through the communication unit 2 and the communication unit 3 respectively, and then set to the corresponding sub-device field controllers respectively.

[0149] It can be understood that this embodiment trains the energy consumption models of each device on the cold source side and the terminal side by obtaining the operating data of the equipment on the cold source side and the terminal side of the refrigeration system, so as to meet the cold load as a constraint, and sorts the operating parameter combinations from low to high according to the total energy consumption of the system through the optimization algorithm. A refrigeration system unstable operation judgment model is trained to infer in real time whether the system is in an unstable operating state under different operating parameter combinations. The first group of operating parameter combinations that are not in an unstable operating state, which are sorted from low to high in terms of the total energy consumption of the system, are sent down for execution to each device on the cold source side and the terminal side. Thereby, the energy-saving effect of the refrigeration system is maximized under the premise of ensuring the safety of the refrigeration system of the data center.

[0150] In some embodiments, the calculation result, i.e., the periodic storage of Table 1, can also be updated to the offline optimization unit 8 through the optimization calculation unit 7-3. At the same time, the failure detection unit 6 interacts with the optimization control unit 7 in real time. When the optimization control unit 7 fails, that is, the time of the real-time optimization calculation exceeds the set threshold (usually set to 2 minutes) or a preset emergency event occurs in the refrigeration system, it is immediately switched to the offline optimization unit 8 to calculate and infer the optimal operating parameter combination.

[0151] In another preferred embodiment, the data center refrigeration system optimization control device of this embodiment includes:

[0152] A plurality of communication units are used to receive the operation data transmitted from the cold source side monitoring system and the terminal side monitoring system respectively. A data processing unit is connected to the communication unit and is used to optimize the operation data. A model training unit is connected to the data processing unit and is used to train the energy consumption models of the corresponding equipment on the cold source side and the terminal side and the unstable operation judgment model of the refrigeration system according to the optimized operation data. An optimization control unit is connected to the model training unit and is used to obtain the system total energy consumption model according to the trained energy consumption models of each equipment, and based on the system total energy consumption model, calculate the total energy consumption value under different operation parameter combinations according to a preset optimization algorithm, and, based on the unstable operation judgment model, judge whether the system is in an unstable operation state under different operation parameter combinations, and take the operation parameter combination with the smallest total energy consumption value and not in an unstable operation state as the optimal operation parameter combination.

[0153] This embodiment can maximize the energy-saving effect of the refrigeration system while ensuring the safety of the data center refrigeration system by uniformly processing the operating data on the cold source side and the terminal side. That is, it can maximize the energy-saving potential of global optimization, ensure the stability of system operation, and effectively extend the operating life of the equipment.

[0154] In a specific embodiment, Figure 7 As shown, the device can obtain the operating data of the cold source side and terminal side equipment of the data center refrigeration system through the communication unit, and can perform tasks such as data processing, model training and algorithm operation, and provide a northbound communication unit for other systems to integrate, supporting commonly used communication protocols such as OPC and Modbus. The device consists of 9 units (specific functions are explained in the example), namely: communication unit 1, communication unit 2, communication unit 3, acquisition unit 4, data processing unit 5, failure detection unit 6, optimization control unit 7, offline optimization unit 8, model training unit 9. Among them:

[0155] 1) The data processing unit 5 is composed of three sub-units, namely: a data collection unit 5-1, a communication data processing unit 5-2, and a collection data processing unit 5-3.

[0156] 2) The optimization control unit 7 is composed of three sub-units, namely: an equipment model unit 7-1, an unstable operation determination model unit 7-2, and an optimization calculation unit 7-3.

[0157] 3) The model training unit 9 consists of two sub-units, namely: an equipment model training unit 9-1 and an unstable operation determination model training unit 9-2.

[0158] Regarding data collection and processing, communication unit 2, communication unit 3 and acquisition unit 4 obtain the operating data of the cold source side and terminal side equipment of the data center refrigeration system. The first part of the data is collected by the BA system and uploaded to the communication unit 2 of this device. The second part of the data is collected by the dynamic environment system and uploaded to the communication unit 3 of this device. The third part of the data is directly connected to the acquisition unit 4 of this device from the newly added sensor. The communication data processing unit 5-2 processes the first and second parts of the data, including data filtering, cleaning, aggregation and quality optimization. The acquisition data processing unit 5-3 processes the third part of the data, including data filtering, cleaning, aggregation and quality optimization. The optimized data is transmitted and synchronized by units 5-2 and 5-3 to the data collection unit 5-1 respectively, and can be provided to the northbound device or system through communication unit 1.

[0159] Regarding the training and deployment of the equipment energy consumption model, the data processed by the data processing unit 5 is uploaded from the data collection unit 5-1 to the equipment model training unit 9-1 for model training. Specifically, the data is divided into a training data set and a verification data set in a random manner, wherein the default ratio of the training data set and the verification data set is 8:2, which can be adjusted by the data set division ratio parameter. The parameter is synchronized to the data collection unit 5-1 and provided to the northbound device or system through the communication unit 1 for setting. The equipment model is trained using the training data set, and the algorithm used for training is not specifically limited. The hyperparameters involved in the algorithm used for training are synchronized to the data collection unit 5-1 and provided to the northbound device or system through the communication unit 1 for setting. The accuracy of the model is verified using the verification data set, and the trained model is deployed to the equipment model unit 7-1 after meeting the accuracy requirements. As an option, the failure detection unit 6 can monitor the accuracy of the equipment model in real time, and automatically retrain and deploy when the model accuracy continues to fail to meet the set requirements.

[0160] Regarding the training and deployment of the unstable operation judgment model, the data processed by the data processing unit 5 is uploaded from the data collection unit 5-1 to the unstable operation judgment model training unit 9-2 for model training. The working principle of the unstable operation judgment model is specifically referred to the above embodiment, which will not be repeated here. The obtained system state STAT is merged with the data uploaded by the previous data collection unit 5-1 to form a new data set, and is randomly divided into a training data set and a verification data set, wherein the default ratio of the training data set and the verification data set is 8:2, which can be adjusted by the data set segmentation ratio parameter. The parameter is synchronized to the data collection unit 5-1 and provided to the northbound device or system for setting. The unstable operation judgment model is trained using the training data set, and the algorithm used for training is not specifically limited. The hyperparameters involved in the algorithm used for training are synchronized to the data collection unit 5-1, and provided to the northbound device or system through the communication unit 1 for setting. The accuracy of the model is verified using the verification data set, and the trained model is deployed to the unstable operation judgment model unit 7-2 after meeting the accuracy requirements.

[0161] Regarding the optimization calculation unit 7-3, the hyper parameters involved in the algorithm are synchronized to the data collection unit 5-1, and provided to the northbound device or system for setting through the communication unit 1. The optimization calculation unit 7-3 determines the optimal operating parameter combination, and sends the optimal operating parameter combination to the data collection unit 5-1 for standard processing, and then sends it to the data center BA system and the data center dynamic environment system through the communication unit 2 and the communication unit 3 respectively, and then sets it to the corresponding sub-device field controllers.

[0162] Optionally, the optimization calculation unit 7-3 saves and updates the calculation result, i.e., Table 1, to the offline optimization unit 8 at regular intervals. At the same time, the failure detection unit 6 can also interact with the optimization control unit 7 in real time. When the optimization control unit 7 fails, i.e., the time of the real-time optimization calculation exceeds a set threshold (optionally set to 2 minutes) or a preset emergency occurs in the refrigeration system, it immediately switches to calculating and inferring the optimal operating parameter combination from the offline optimization unit 8.

[0163] In this embodiment, 1) unlike the previous global optimization algorithm that does not cover the terminal side equipment, the control device of the present invention can maximize the energy-saving potential of global optimization by uniformly processing the data on the cold source side and the terminal side. 2) unlike the previous global optimization algorithm that does not consider whether the system is in an unstable operating state, the new unstable area determination model and its combined algorithm can ensure the stability of the system operation and can greatly extend the operating life of the equipment. 3) unlike the previous global optimization algorithm that is not equipped with a failure detection unit, the method and device of the present invention can continue to maintain the continuity of the optimization algorithm under some abnormal circumstances.

[0164] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0165] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0166] It can be understood that the above embodiments only express the preferred implementation modes of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the patent scope of the present invention. It should be pointed out that, for ordinary technicians in this field, the above technical features can be freely combined without departing from the concept of the present invention, and several deformations and improvements can be made, which all belong to the protection scope of the present invention. Therefore, all equivalent changes and modifications made to the scope of the claims of the present invention should belong to the coverage of the claims of the present invention.

Claims

1. A method for optimizing and controlling a refrigeration system in a data center, wherein the refrigeration system comprises a cold source side device for providing a cold source and a terminal side device for heat dissipation, characterized in that: The method comprises the following steps: S1. Obtaining the operating data of the cold source side equipment and the terminal side equipment; S2, training energy consumption models of each device on the cold source side and the terminal side based on the operation data, and obtaining a total energy consumption model of the system according to the trained energy consumption models of each device; and training an unstable operation judgment model of the refrigeration system based on the operation data; S3. Based on the system total energy consumption model, calculate the total energy consumption values ​​under different operating parameter combinations according to a preset optimization algorithm; S4. Based on the unstable operation determination model, determine whether the system is in an unstable operation state under different combinations of operation parameters; S5. The operating parameter combination with the smallest total energy consumption value and not in an unstable operating state is taken as the optimal operating parameter combination.

2. The data center refrigeration system optimization control method according to claim 1, characterized in that: The method further includes: S6. Send the optimal parameter combination to each device on the cold source side and the terminal side for execution.

3. The data center refrigeration system optimization control method according to claim 1, characterized in that: In step S2, obtaining the total energy consumption model of the system according to the energy consumption models of the various devices includes: The energy consumption models of the various devices are added together to obtain the total energy consumption model of the system.

4. The data center refrigeration system optimization control method according to claim 3, characterized in that: The preset optimization algorithm includes preset constraints, and step S3 includes: S31, using the system total energy consumption model as the objective function of the preset optimization algorithm; S32, determining the parameter setting value of each device that meets the preset constraint condition, and forming a plurality of different operating parameter combinations; each operating parameter combination includes the parameter setting value of each device on the cold source side and the terminal side; S33. Calculate the total energy consumption values ​​corresponding to different operating parameter combinations according to the objective function, and sort them from low to high according to the total energy consumption values.

5. The data center refrigeration system optimization control method according to claim 1, characterized in that: In step S2, the energy consumption models of the equipment on the cold source side and the terminal side are obtained by training based on the operation data, including: S211, dividing the operating data into a training data set and a verification data set according to a preset ratio; S212, using the training data set to train the energy consumption model of the corresponding equipment; S213: Use the verification data set to verify the accuracy of the corresponding device energy consumption model. If the accuracy requirement is met, it is confirmed that the training of the device energy consumption model is completed.

6. The data center refrigeration system optimization control method according to claim 5, characterized in that: Step S2 also includes: S214: monitor the accuracy of the energy consumption models of the various devices in real time, and retrain a device energy consumption model when the model does not meet the preset accuracy requirement.

7. The data center refrigeration system optimization control method according to claim 1, characterized in that: Before step S2 and after step S1, the method further includes: The operation data is filtered, cleaned, aggregated and quality optimized.

8. The data center refrigeration system optimization control method according to claim 1, characterized in that: In step S2, the working principle of the unstable operation determination model is as follows: S221, according to the operating data, the indicator S DT Perform the calculation: S DT =w1·DT Roc +w2·DT Dev ; DT Roc =DT ratio / DT ratio_normal ; DT Dev =(DT chiller_t -DT chiller_set ) / DT chiller_set ; DT ratio =(DT chiller_t -DT chiller_t-TDT ) / T DT ; w1+w2=1; Among them, the indicator S DT DT is the criterion for determining unstable operation of the refrigeration system. Roc DT is the fluctuation of superheat of chiller compressor. Dev DT is the deviation of the superheat of the chiller compressor. ratio is the fluctuation rate of the superheat of the chiller compressor, DT ratio_normal is the standard value of the fluctuation rate of the superheat of the chiller compressor, w1 is the weight coefficient of the fluctuation rate of the superheat of the chiller compressor, w2 is the weight coefficient of the deviation of the superheat of the chiller compressor, T is the data collection period, T DT is a time period, DT chiller_t is the superheat of the chiller compressor at the current moment, DT chiller_t-TDT is T before the current moment DT The superheat of the chiller compressor at the moment, DT chiller_set It is the standard setting value of superheat of chiller compressor; S222. Determine whether S DT Continuously greater than or equal to S DT_set ; If so, it is determined that the system is in an unstable operating state, and the system state STAT is recorded as 1. The corresponding operating parameter combination is the operating parameter combination of the unstable operating state; where S DT_set This is the set value for judging the unstable operation index.

9. The data center refrigeration system optimization control method according to claim 8, characterized in that: In step S2, the unstable operation determination model of the refrigeration system is obtained by training based on the operation data, including: S223, combining the obtained system state STAT with the operation data to form a new data set, and dividing the new data set into a training data set and a verification data set according to a preset ratio; S224, training the unstable operation determination model using a training data set; S225 , using the verification data set to verify the accuracy of the unstable operation determination model, if the accuracy requirement is met, it is confirmed that the training of the unstable operation determination model is completed.

10. A data center refrigeration system optimization control device, characterized in that: include: A plurality of communication units, used for respectively receiving the operation data transmitted from the cold source side monitoring system and the terminal side monitoring system; A data processing unit, connected to the communication unit, for optimizing the operation data; A model training unit, connected to the data processing unit, for training the corresponding energy consumption models of the equipment on the cold source side and the terminal side and the unstable operation judgment model of the refrigeration system according to the optimized operation data; An optimization control unit is connected to the model training unit, and is used to obtain a system total energy consumption model based on the trained energy consumption models of each device, and based on the system total energy consumption model, calculate the total energy consumption value under different operating parameter combinations according to a preset optimization algorithm, and based on the unstable operation judgment model, judge whether the system is in an unstable operating state under different operating parameter combinations, and take the operating parameter combination with the smallest total energy consumption value and not in an unstable operating state as the optimal operating parameter combination.

Images