A data center based on an intelligent cabinet

Abstract

The present invention provides a data center based on intelligent cabinets, which includes a plurality of intelligent cabinets, a total refrigeration component, a plurality of sub-refrigeration components and a refrigeration scheduling component. The total refrigeration component is connected with a total refrigeration pipe, the sub-refrigeration components are connected with sub-refrigeration pipes, and the sub-refrigeration pipes are conductively connected to at least one intelligent cabinet. A plurality of groups of data acquisition devices are arranged in each intelligent cabinet, and each group of data acquisition devices includes a device data acquisition device and an environmental data acquisition device. A data processing device is arranged in each intelligent cabinet, and the data processing device is used to determine the refrigeration state in the intelligent cabinet based on the information collected by the plurality of groups of data acquisition devices. The refrigeration scheduling component is used to determine the optimal refrigeration parameters of the total refrigeration component based on the basic information of the load devices in each intelligent cabinet. The refrigeration scheduling component is also used to adjust the refrigeration parameters of the plurality of sub-refrigeration components in real time based on the refrigeration states in the plurality of intelligent cabinets, and has the advantage of reducing waste of refrigeration capacity.

Description

Technical Field

[0001] This specification relates to the field of cabinet heat dissipation, and particularly to a data center based on intelligent cabinets. Background Art

[0002] With the development of technology, the applications of IT devices such as servers, network switches, and mainframe computers are increasing continuously, the scale of data centers is expanding, and the heat dissipation density of IT devices has also increased significantly. Multiple load devices are often placed inside a cabinet, and multiple cabinets are placed side by side for convenient maintenance and management. To ensure that the cabinet can work normally, it is necessary to keep the cabinet within a normal temperature range at all times. The prior art uses excessive cooling capacity to cool the entire computer room. However, in this mode, the output power of the air conditioner is too large, and the cooling capacity is wasted seriously, especially when the load rate of the cabinet is low.

[0003] Therefore, it is necessary to provide a data center based on intelligent cabinets to achieve flexible scheduling of the cooling capacity of the data center and reduce the waste of cooling capacity. Summary of the Invention

[0004] One embodiment of this specification provides a data center based on intelligent cabinets, including a plurality of intelligent cabinets, a total cooling component, a plurality of sub-cooling components, and a cooling scheduling component. Among them, the total cooling component is connected with a total cooling pipe, and the total cooling pipe is conductively connected to each of the plurality of intelligent cabinets. The sub-cooling component is connected with a sub-cooling pipe, and the sub-cooling pipe is conductively connected to at least one of the intelligent cabinets. Each intelligent cabinet communicates with at least two of the sub-cooling components; the intelligent cabinet is used to place a plurality of load devices. Each intelligent cabinet is provided with multiple groups of data acquisition devices. Each group of data acquisition devices corresponds to a load device in the intelligent cabinet. Each group of data acquisition devices includes a device data acquisition device and an environmental data acquisition device. The device data acquisition device is used to obtain the device-related information of the corresponding load device, and the environmental data acquisition device is used to obtain the environmental-related information of the target area where the corresponding load device is located; a data processing device is provided in each intelligent cabinet. The multiple groups of data acquisition devices provided in the intelligent cabinet are electrically connected to the data processing device. The data processing device is used to determine the cooling state in the intelligent cabinet based on the information collected by the multiple groups of data acquisition devices; the cooling scheduling component is used to determine the optimal cooling parameters of the total cooling component based on the basic information of each intelligent cabinet, where the basic information of the intelligent cabinet includes the type of each load device placed in the intelligent cabinet, the placement position of each load device, and the internal structure of the intelligent cabinet; the cooling scheduling component is also used to adjust the cooling parameters of the multiple sub-cooling components in real time based on the cooling states in the multiple intelligent cabinets.

[0005] In some embodiments, the refrigeration scheduling component is configured to determine the optimal refrigeration parameters of the total refrigeration component based on the basic information of each intelligent cabinet, including: determining the optimal refrigeration parameters of the total refrigeration component through a first parameter determination model based on the external environmental temperature, external environmental humidity, the type of each load device placed in each intelligent cabinet, the placement location of each load device, and the internal structure of the intelligent cabinet, where the first parameter determination model is a machine learning model, and the first parameters include the optimal refrigeration parameters of the total refrigeration component.

[0006] In some embodiments, determining the conduction connection relationship between the sub-refrigeration components and at least one of the intelligent cabinets includes: generating a plurality of candidate connection schemes based on connection constraint conditions through a Monte Carlo model, where the candidate connection schemes include the conduction connection relationships between the plurality of sub-refrigeration components and the plurality of intelligent cabinets; generating a plurality of virtual operation states based on operation constraint conditions through the Monte Carlo model, where the plurality of virtual operation states include the refrigeration capacity requirements of the plurality of intelligent cabinets and the operation states of the plurality of sub-refrigeration components, and the operation state is a fault state or a normal operation state; for each candidate connection scheme, determining the total operation score of the candidate connection scheme in the plurality of virtual operation states; based on the total operation scores of each candidate connection scheme, determining a target connection scheme from the plurality of candidate connection schemes, and determining the conduction connection relationship between the sub-refrigeration components and at least one of the intelligent cabinets based on the target connection scheme.

[0007] In some embodiments, determining the total operation score of the candidate connection scheme in the plurality of virtual operation states includes: for each virtual operation state, determining the difference between the actual refrigeration supply of each intelligent cabinet and the refrigeration capacity requirement and the load of each sub-refrigeration component, and determining the operation score of the candidate connection scheme in the virtual operation state; based on the operation scores of the candidate connection scheme in each virtual operation state, determining the total operation score corresponding to the candidate connection scheme.

[0008] In some embodiments, the device data acquisition device at least includes a device temperature sensor; the environmental data acquisition device at least includes an environmental temperature sensor; an electric control valve is provided between the sub-refrigeration pipe and the intelligent cabinet.

[0009] In some embodiments, the data processing device determines the refrigeration state in the intelligent cabinet based on the information collected by the multiple sets of data acquisition devices, including: for each set of the data acquisition devices, obtaining a target device temperature sequence and a target ambient temperature sequence corresponding to the data acquisition device, wherein the target device temperature sequence includes the device temperatures obtained by the device temperature sensor at multiple time points in a target time period, and the target ambient temperature sequence includes the ambient temperatures obtained by the ambient temperature sensor at multiple time points in the target time period; performing denoising processing on the target device temperature sequence and the target ambient temperature sequence to generate a denoised target device temperature sequence and a denoised target ambient temperature sequence; determining the cooling capacity supplement requirement of the load device corresponding to the data acquisition device based on the denoised target device temperature sequence and the denoised target ambient temperature sequence; and determining the cooling capacity supplement requirement of the intelligent cabinet based on the cooling capacity supplement requirements of the multiple sets of data acquisition devices arranged in the intelligent cabinet.

[0010] In some embodiments, the performing denoising processing on the target device temperature sequence and the target ambient temperature sequence to generate a denoised target device temperature sequence and a denoised target ambient temperature sequence includes:

[0011] generating a target device temperature curve based on the target device temperature sequence and generating a target ambient temperature curve based on the target ambient temperature sequence; decomposing the target device temperature curve into at least one first intrinsic mode function component and a first residue; decomposing the target ambient temperature curve into at least one second intrinsic mode function component and a second residue; determining a first target intrinsic mode function component based on the at least one first intrinsic mode function component and a first residue through a first noise determination model, wherein the first target intrinsic mode function component is a first intrinsic mode function component containing noise; determining a second target intrinsic mode function component based on the at least one second intrinsic mode function component and a second residue through a second noise determination model, wherein the second target intrinsic mode function component is a second intrinsic mode function component containing noise; performing denoising processing on the first target intrinsic mode function component and the second target intrinsic mode function component based on the at least one first intrinsic mode function component and a first residue and the at least one second intrinsic mode function component and a second residue through a denoising model to obtain a denoised first target intrinsic mode function component and a denoised second target intrinsic mode function component; reconstructing the target device temperature curve based on the denoised first target intrinsic mode function component to generate the denoised target device temperature sequence; and reconstructing the target ambient temperature curve based on the denoised second target intrinsic mode function component to generate the denoised target ambient temperature sequence.

[0012] In some embodiments, the refrigeration scheduling component adjusts the refrigeration parameters of the plurality of sub-refrigeration components in real time based on the refrigeration states in the plurality of intelligent cabinets, including: for each of the intelligent cabinets, when the refrigeration capacity replenishment demand of the intelligent cabinet is greater than or equal to a preset refrigeration capacity replenishment demand threshold, based on the conduction connection relationship between the plurality of intelligent cabinets and the plurality of sub-refrigeration components, determining a target sub-refrigeration component from the plurality of sub-refrigeration components; determining the target refrigeration parameters of the target sub-refrigeration component based on the refrigeration capacity replenishment demand of the intelligent cabinet; opening the electric control valve of the sub-refrigeration pipe between the target sub-refrigeration component and the intelligent cabinet, and controlling the target sub-refrigeration component to refrigerate based on the target refrigeration parameters of the target sub-refrigeration component.

[0013] In some embodiments, the determining a target sub-refrigeration component from the plurality of sub-refrigeration components based on the conduction connection relationship between the plurality of intelligent cabinets and the plurality of sub-refrigeration components includes: determining a plurality of candidate sub-refrigeration components based on the conduction connection relationship between the plurality of intelligent cabinets and the plurality of sub-refrigeration components; determining a target sub-refrigeration component from the plurality of candidate sub-refrigeration components based on the relevant information of each candidate sub-refrigeration component.

[0014] In some embodiments, the determining a target sub-refrigeration component from the plurality of candidate sub-refrigeration components based on the relevant information of each candidate sub-refrigeration component includes: determining a matching score corresponding to each candidate sub-refrigeration component based on the connection distance, load rate, continuous refrigeration duration, and failure probability corresponding to each candidate sub-refrigeration component; determining a target sub-refrigeration component from the plurality of candidate sub-refrigeration components based on the matching scores corresponding to each candidate sub-refrigeration component. Description of the Drawings

[0015] This specification will further illustrate in the form of exemplary embodiments, and these exemplary embodiments will be described in detail through the drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, where:

[0016] Figure 1 is a schematic diagram of a module of a data center based on intelligent cabinets shown in some embodiments of this specification;

[0017] Figure 2 is a flowchart of determining the conduction connection relationship between a sub-refrigeration component and at least one intelligent cabinet shown in some embodiments of this specification;

[0018] Figure 3 is a flowchart of determining the refrigeration state in an intelligent cabinet shown in some embodiments of this specification;

[0019] Figure 4It is a flowchart for denoising the temperature sequence of a target device and the temperature sequence of a target environment as shown in some embodiments of this specification. Detailed implementation manners

[0020] To more clearly illustrate the technical solutions of the embodiments of this specification, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some examples or embodiments of this specification. For those of ordinary skill in the art, without creative efforts, this specification can also be applied to other similar scenarios based on these drawings. Unless obvious from the language context or otherwise stated, the same reference numerals in the figures represent the same structure or operation.

[0021] Figure 1 It is a schematic diagram of the modules of a multi-intelligent cabinet management system based on a data center as shown in some embodiments of this specification. As Figure 1 shown, the multi-intelligent cabinet management system based on the data center may include multiple intelligent cabinets, a total refrigeration component, multiple sub-refrigeration components, and a refrigeration scheduling component. Among them, the total refrigeration component is connected to a total refrigeration pipe, and the total refrigeration pipe is connected to each of the multiple intelligent cabinets in a conducting manner. The sub-refrigeration component is connected to a sub-refrigeration pipe, and the sub-refrigeration pipe is connected to at least one intelligent cabinet in a conducting manner. Each intelligent cabinet is connected to at least two sub-refrigeration components. An electric control valve is provided between the sub-refrigeration pipe and the intelligent cabinet.

[0022] Figure 2 It is a flowchart for determining the conducting connection relationship between the sub-refrigeration component and at least one intelligent cabinet as shown in some embodiments of this specification. As Figure 2 shown, in some embodiments, the conducting connection relationship between the sub-refrigeration component and at least one intelligent cabinet can be determined based on the following process:

[0023] Generate multiple candidate connection schemes based on the connection constraint conditions through a Monte Carlo model. Among them, the candidate connection schemes include the conducting connection relationships between multiple sub-refrigeration components and multiple intelligent cabinets;

[0024] Generate multiple virtual operating states based on the operating constraint conditions through a Monte Carlo model. Among the multiple virtual operating states, the refrigeration capacity requirements of multiple intelligent cabinets and the operating states of multiple sub-refrigeration components are included. The operating state is a fault state or a normal operating state;

[0025] For each candidate connection scheme, determine the total operating score of the candidate connection scheme under multiple virtual operating states;

[0026] Based on the total operating scores of each candidate connection scheme, determine the target connection scheme from multiple candidate connection schemes, and determine the conducting connection relationship between the sub-refrigeration component and at least one intelligent cabinet based on the target connection scheme.

[0027] Specifically, for each virtual operating state, determine the difference between the actual cooling supply and the cooling demand of each intelligent cabinet and the load of each sub-cooling component, and determine the running score of the candidate connection scheme in the virtual operating state; based on the running scores of the candidate connection scheme in each virtual operating state, determine the total running score corresponding to the candidate connection scheme.

[0028] Only as an example, the total running score corresponding to the candidate connection scheme can be calculated based on the following formula:

[0029]

[0030] where, S i is the total running score corresponding to the i-th candidate connection scheme, m is the total number of virtual operating states, n is the total number of intelligent cabinets, a1 and a2 are preset weights, b1 is a preset parameter and b1>0, is the difference between the actual cooling supply and the cooling demand of the l-th intelligent cabinet, is the preset load threshold of the x-th sub-cooling component, is the load of the x-th sub-cooling component.

[0031] It can be understood that the candidate connection scheme corresponding to the maximum value of the total running score can be used as the target connection scheme, and the sub-cooling components can be installed according to the target connection scheme. In the above way, the best conduction connection relationship between the sub-cooling components and at least one intelligent cabinet is determined, so that in various application scenarios, multiple sub-cooling components can still cooperate to complete the flexible cooling work of multiple intelligent cabinets, and the cooling effect on multiple intelligent cabinets can be ensured as much as possible.

[0032] The cooling scheduling component is used to determine the best cooling parameters of the total cooling components based on the basic information of each intelligent cabinet.

[0033] In some embodiments, the refrigeration scheduling component can determine the optimal refrigeration parameters of the total refrigeration component based on the first parameter determination model, the external environmental temperature, the external environmental humidity, the type of each load device placed in each intelligent cabinet, the placement location of each load device, and the internal structure of the intelligent cabinet. For example, the external environmental temperature, the external environmental humidity, the type of each load device placed in each intelligent cabinet, the placement location of each load device, and the internal structure of the intelligent cabinet can be input into the first parameter determination model, and the first parameter determination model can output the optimal refrigeration parameters of the total refrigeration component. Among them, the first parameter determination model can be a machine learning model such as an artificial neural network (ANN) model, a recurrent neural network (RNN) model, a long short-term memory network (LSTM) model, a bidirectional recurrent neural network (BRNN) model, etc. The basic information of the intelligent cabinet includes the type of each load device placed in the intelligent cabinet, the placement location of each load device, and the internal structure of the intelligent cabinet. The first parameter includes the optimal refrigeration parameters of the total refrigeration component.

[0034] Intelligent cabinets are used to place multiple load devices (such as servers, network switches, mainframe computers, etc.). Multiple groups of data acquisition devices are arranged in each intelligent cabinet. Each group of data acquisition devices corresponds to a load device in the intelligent cabinet. Each group of data acquisition devices includes a device data acquisition device and an environmental data acquisition device. The device data acquisition device is used to obtain device-related information of the corresponding load device, and the environmental data acquisition device is used to obtain environmental-related information of the target area where the corresponding load device is located. In some embodiments, the device data acquisition device at least includes a device temperature sensor, and the environmental data acquisition device at least includes an environmental temperature sensor.

[0035] A data processing device is arranged in each intelligent cabinet. The multiple groups of data acquisition devices arranged in the intelligent cabinet are all electrically connected to the data processing device. The data processing device is used to determine the refrigeration state in the intelligent cabinet based on the information collected by the multiple groups of data acquisition devices. Among them, the data processing device can include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc.

[0036] Figure 3 is a flowchart for determining the refrigeration state in an intelligent cabinet according to some embodiments of this specification, as Figure 3 shown. In some embodiments, the data processing device determines the refrigeration state in the intelligent cabinet based on the information collected by the multiple groups of data acquisition devices, which may include:

[0037] For each group of data acquisition devices,

[0038] Obtain the target device temperature sequence and the target ambient temperature sequence corresponding to the data acquisition device, where the target device temperature sequence includes the device temperatures obtained by the device temperature sensor at multiple time points within the target time period, and the target ambient temperature sequence includes the ambient temperatures obtained by the ambient temperature sensor at multiple time points within the target time period;

[0039] Perform denoising processing on the target device temperature sequence and the target ambient temperature sequence to generate a denoised target device temperature sequence and a denoised target ambient temperature sequence;

[0040] Based on the denoised target device temperature sequence and the denoised target ambient temperature sequence, determine the cooling capacity supplement requirement of the load device corresponding to the data acquisition device;

[0041] Based on the cooling capacity supplement requirements of multiple groups of data acquisition devices set within the intelligent cabinet, determine the cooling capacity supplement requirement of the intelligent cabinet.

[0042] Specifically, the data processing device can pre - establish a univariate regression model, where the independent variables of the univariate regression model are the device temperature sequence and the ambient temperature sequence, and the dependent variable of the univariate regression model is the cooling capacity supplement requirement. The denoised target device temperature sequence and the denoised target ambient temperature sequence can be input into the univariate regression model to determine the cooling capacity supplement requirement of the load device corresponding to the data acquisition device. Further, the sum of the cooling capacity supplement requirements of multiple groups of data acquisition devices set within the intelligent cabinet can be used as the cooling capacity supplement requirement of the intelligent cabinet.

[0043] It can be understood that by obtaining the target device temperature sequence and the target ambient temperature sequence, it is possible to relatively quickly and accurately determine whether the intelligent cabinet requires supplementary cooling by a sub - cooling component in addition to the total cooling component.

[0044] Figure 4 is a flowchart of performing denoising processing on the target device temperature sequence and the target ambient temperature sequence according to some embodiments of this specification. As Figure 4 shown, in some embodiments, performing denoising processing on the target device temperature sequence and the target ambient temperature sequence to generate a denoised target device temperature sequence and a denoised target ambient temperature sequence includes:

[0045] Generate a target device temperature curve based on the target device temperature sequence, and generate a target ambient temperature curve based on the target ambient temperature sequence;

[0046] Decompose the target device temperature curve into at least one first intrinsic mode function component and a first residual;

[0047] Decompose the target ambient temperature curve into at least one second intrinsic mode function component and a second residual;

[0048] Based on at least one first intrinsic mode component and a first residual, a first noise determination model determines a first target intrinsic mode component, where the first target intrinsic mode component is a first intrinsic mode component containing noise. The first noise determination model can be a machine learning model such as an artificial neural network (ANN) model, a recurrent neural network (RNN) model, a long short-term memory network (LSTM) model, or a bidirectional recurrent neural network (BRNN) model;

[0049] Based on at least one second intrinsic mode component and a second residual, a second noise determination model determines a second target intrinsic mode component, where the second target intrinsic mode component is a second intrinsic mode component containing noise. The second noise determination model can be a machine learning model such as an artificial neural network (ANN) model, a recurrent neural network (RNN) model, a long short-term memory network (LSTM) model, or a bidirectional recurrent neural network (BRNN) model;

[0050] Based on at least one first intrinsic mode component, a first residual, at least one second intrinsic mode component, and a second residual, a denoising model performs denoising processing on the first target intrinsic mode component and the second target intrinsic mode component to obtain the denoised first target intrinsic mode component and the denoised second target intrinsic mode component. The input of the denoising model can include the first target intrinsic mode component, other first intrinsic mode components, the first residual, the second target intrinsic mode component, other second intrinsic mode components, and the second residual. The output of the denoising model can include the denoised first target intrinsic mode component and the denoised second target intrinsic mode component. The denoising model can be a machine learning model such as an artificial neural network (ANN) model, a recurrent neural network (RNN) model, a long short-term memory network (LSTM) model, or a bidirectional recurrent neural network (BRNN) model;

[0051] Based on the denoised first target intrinsic mode component, reconstruct the target device temperature curve to generate a denoised target device temperature sequence;

[0052] Based on the denoised second target intrinsic mode component, reconstruct the target environmental temperature curve to generate a denoised target environmental temperature sequence.

[0053] It can be understood that through the above denoising method, the noise parts in the target device temperature sequence and the target ambient temperature sequence can be determined relatively accurately. At the same time, by denoising the first target intrinsic mode component and the second target intrinsic mode component simultaneously with the denoising model, the first target intrinsic mode component and the second target intrinsic mode component can be denoised more accurately based on the correlation relationship between at least one first intrinsic mode component and one first residual and at least one second intrinsic mode component and one second residual.

[0054] In some embodiments, the refrigeration scheduling component adjusts the refrigeration parameters of multiple sub-refrigeration components in real time based on the refrigeration states in multiple intelligent cabinets, including:

[0055] For each intelligent cabinet,

[0056] When the refrigeration capacity replenishment demand of the intelligent cabinet is greater than or equal to the preset refrigeration capacity replenishment demand threshold, the intelligent cabinet can be referred to as an intelligent cabinet that needs refrigeration replenishment. Based on the conduction connection relationship between multiple intelligent cabinets and multiple sub-refrigeration components, a target sub-refrigeration component is determined from the multiple sub-refrigeration components;

[0057] Based on the refrigeration capacity replenishment demand of the intelligent cabinet, the target refrigeration parameters of the target sub-refrigeration component are determined;

[0058] The electric control valve of the sub-refrigeration pipe between the target sub-refrigeration component and the intelligent cabinet is opened, and the target sub-refrigeration component is controlled to refrigerate based on the target refrigeration parameters of the target sub-refrigeration component.

[0059] In some embodiments, the refrigeration scheduling component can determine multiple candidate sub-refrigeration components based on the conduction connection relationship between multiple intelligent cabinets and multiple sub-refrigeration components; and determine the target sub-refrigeration component from the multiple candidate sub-refrigeration components based on the relevant information of each candidate sub-refrigeration component.

[0060] Specifically, the refrigeration scheduling component can determine the matching score corresponding to each candidate sub-refrigeration component based on the connection distance, load rate, continuous refrigeration duration, and failure probability corresponding to each candidate sub-refrigeration component; and determine the target sub-refrigeration component from the multiple candidate sub-refrigeration components based on the matching score corresponding to each candidate sub-refrigeration component. Among them, the connection distance corresponding to the candidate sub-refrigeration component is the length of the sub-refrigeration pipe between the candidate sub-refrigeration component and the intelligent cabinet that needs refrigeration replenishment, the load rate is the refrigeration capacity that the candidate sub-refrigeration component currently needs to provide for multiple intelligent cabinets, the continuous refrigeration duration is the cumulative working duration of the candidate sub-refrigeration component in the current period (for example, 1 month, 3 months, etc.), and the failure probability is the probability of the candidate sub-refrigeration component failing recently (for example, within one day, within one week, etc.).

[0061] As an example only, the matching score corresponding to the candidate sub-cooling component can be determined by the following formula:

[0062]

[0063] Wherein, M i is the matching score corresponding to the i-th candidate sub-cooling component, c1, c2, c3 and c4 are all preset weights, and D i is the normalized connection distance corresponding to the i-th candidate sub-cooling component, L i is the normalized load rate corresponding to the i-th candidate sub-cooling component, T i is the normalized continuous cooling duration corresponding to the i-th candidate sub-cooling component, and P i is the normalized failure probability corresponding to the i-th candidate sub-cooling component.

[0064] It can be understood that the refrigeration scheduling component can use the candidate sub-cooling component with the largest matching score as the target sub-cooling component.

[0065] In some embodiments, the refrigeration scheduling component can determine the target refrigeration parameters of the target sub-cooling component based on the refrigeration capacity supplement requirement of the intelligent cabinet through the second parameter determination model. Among them, the second parameter determination model can be a machine learning model such as an artificial neural network (ANN) model, a recurrent neural network (RNN) model, a long short-term memory network (LSTM) model, or a bidirectional recurrent neural network (BRNN) model.

[0066] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other deformations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be regarded as consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A data center based on an intelligent cabinet, characterized in that, It includes multiple intelligent cabinets, a total refrigeration component, multiple sub-refrigeration components, and a refrigeration scheduling component. Among them, the total refrigeration component is connected with a total refrigeration pipe, and the total refrigeration pipe is conductively connected to each of the multiple intelligent cabinets. The sub-refrigeration component is connected with a sub-refrigeration pipe, and the sub-refrigeration pipe is conductively connected to at least one of the intelligent cabinets. Each intelligent cabinet is communicated with at least two of the sub-refrigeration components; The intelligent cabinet is used to place multiple load devices. Multiple groups of data acquisition devices are arranged in each intelligent cabinet. Each group of data acquisition devices corresponds to one load device in the intelligent cabinet. Each group of data acquisition devices includes a device data acquisition device and an environment data acquisition device. The device data acquisition device is used to obtain the device-related information of the corresponding load device, and the environment data acquisition device is used to obtain the environment-related information of the target area where the corresponding load device is located; A data processing device is arranged in each intelligent cabinet. The multiple groups of data acquisition devices arranged in the intelligent cabinet are electrically connected to the data processing device. The data processing device is used to determine the refrigeration state in the intelligent cabinet based on the information collected by the multiple groups of data acquisition devices; The refrigeration scheduling component is used to determine the optimal refrigeration parameters of the total refrigeration component based on the basic information of each intelligent cabinet. The basic information of the intelligent cabinet includes the type of each load device placed in the intelligent cabinet, the placement position of each load device, and the internal structure of the intelligent cabinet; The refrigeration scheduling component is also used to adjust the refrigeration parameters of the multiple sub-refrigeration components in real time based on the refrigeration states in the multiple intelligent cabinets.

2. The data center based on an intelligent cabinet according to claim 1, wherein The refrigeration scheduling component is used to determine the optimal refrigeration parameters of the total refrigeration component based on the basic information of each intelligent cabinet, including: Through a first parameter determination model, based on the external environmental temperature, external environmental humidity, the type of each load device placed in each intelligent cabinet, the placement position of each load device, and the internal structure of the intelligent cabinet, determine the optimal refrigeration parameters of the total refrigeration component. Among them, the first parameter determination model is a machine learning model, and the first parameter includes the optimal refrigeration parameters of the total refrigeration component.

3. The data center based on an intelligent cabinet according to claim 1, characterized in that, Determining the conductive connection relationship between the sub-refrigeration component and at least one of the intelligent cabinets includes: Generating multiple candidate connection schemes through a Monte Carlo model based on connection constraint conditions. Among them, the candidate connection schemes include the conductive connection relationships between the multiple sub-refrigeration components and the multiple intelligent cabinets; Generating multiple virtual operating states through the Monte Carlo model based on operating constraint conditions. Among them, the multiple virtual operating states include the refrigeration capacity requirements of the multiple intelligent cabinets and the operating states of the multiple sub-refrigeration components. The operating state is a fault state or a normal operating state; For each of the candidate connection schemes, determine the total operating score of the candidate connection scheme under the multiple virtual operating states; Based on the total running scores of each of the candidate connection schemes, determine a target connection scheme from the multiple candidate connection schemes, and determine the conduction connection relationship between the sub-cooling component and at least one of the intelligent cabinets based on the target connection scheme.

4. A data center based on an intelligent cabinet according to claim 3, characterized in that, The determination of the total running scores of the candidate connection schemes in the multiple virtual operating states includes: For each of the virtual operating states, determine the difference between the actual cooling supply of each intelligent cabinet and the cooling demand, and the load of each sub-cooling component, and determine the running score of the candidate connection scheme in the virtual operating state; Based on the running scores of the candidate connection schemes in each of the virtual operating states, determine the total running score corresponding to the candidate connection scheme.

5. A data center based on an intelligent cabinet according to any one of claims 1-4, characterized in that, The device data acquisition device at least includes a device temperature sensor; The environmental data acquisition device at least includes an environmental temperature sensor; An electric control valve is provided between the sub-cooling pipe and the intelligent cabinet.

6. The data center based on an intelligent cabinet according to claim 5, characterized in that, The data processing device determines the cooling state in the intelligent cabinet based on the information collected by the multiple groups of data acquisition devices, including: For each group of the data acquisition devices, Obtain the target device temperature sequence and the target environmental temperature sequence corresponding to the data acquisition device, where the target device temperature sequence includes the device temperatures obtained by the device temperature sensor at multiple time points in a target time period, and the target environmental temperature sequence includes the environmental temperatures obtained by the environmental temperature sensor at multiple time points in the target time period; Perform denoising processing on the target device temperature sequence and the target environmental temperature sequence to generate a denoised target device temperature sequence and a denoised target environmental temperature sequence; Based on the denoised target device temperature sequence and the denoised target environmental temperature sequence, determine the cooling capacity replenishment demand of the load device corresponding to the data acquisition device; Based on the cooling capacity replenishment demands of the multiple groups of data acquisition devices provided in the intelligent cabinet, determine the cooling capacity replenishment demand of the intelligent cabinet.

7. A data center based on an intelligent cabinet according to claim 6, characterized in that, The performing denoising processing on the target device temperature sequence and the target environmental temperature sequence to generate a denoised target device temperature sequence and a denoised target environmental temperature sequence includes: Generate a target device temperature curve based on the target device temperature sequence, and generate a target environmental temperature curve based on the target environmental temperature sequence; Decompose the target device temperature curve into at least one first intrinsic mode function and a first residual; Decompose the target environmental temperature curve into at least one second intrinsic mode function and a second residual; Determine a first target intrinsic mode function through a first noise determination model based on the at least one first intrinsic mode function and a first residual, where the first target intrinsic mode function is a first intrinsic mode function containing noise; Determine a second target intrinsic mode function through a second noise determination model based on the at least one second intrinsic mode function and a second residual, where the second target intrinsic mode function is a second intrinsic mode function containing noise; The denoising model performs denoising processing on the first target intrinsic mode component and the second target intrinsic mode component based on the at least one first intrinsic mode component, one first residual, the at least one second intrinsic mode component, and one second residual, to obtain the denoised first target intrinsic mode component and the denoised second target intrinsic mode component; Reconstruct the target device temperature curve based on the denoised first target intrinsic mode component to generate the denoised target device temperature sequence; Reconstruct the target ambient temperature curve based on the denoised second target intrinsic mode component to generate the denoised target ambient temperature sequence.

8. The data center based on an intelligent cabinet according to claim 6, characterized in that, The refrigeration scheduling component adjusts the refrigeration parameters of the multiple sub-refrigeration components in real time based on the refrigeration states in the multiple intelligent cabinets, including: For each of the intelligent cabinets, When the refrigeration capacity replenishment demand of the intelligent cabinet is greater than or equal to the preset refrigeration capacity replenishment demand threshold, based on the conduction connection relationship between the multiple intelligent cabinets and the multiple sub-refrigeration components, determine the target sub-refrigeration component from the multiple sub-refrigeration components; Determine the target refrigeration parameters of the target sub-refrigeration component based on the refrigeration capacity replenishment demand of the intelligent cabinet; Open the electric control valve of the sub-refrigeration pipe between the target sub-refrigeration component and the intelligent cabinet, and control the target sub-refrigeration component to refrigerate based on the target refrigeration parameters of the target sub-refrigeration component.

9. The data center based on an intelligent cabinet according to claim 8, wherein The determining the target sub-refrigeration component from the multiple sub-refrigeration components based on the conduction connection relationship between the multiple intelligent cabinets and the multiple sub-refrigeration components includes: Determine multiple candidate sub-refrigeration components based on the conduction connection relationship between the multiple intelligent cabinets and the multiple sub-refrigeration components; Determine the target sub-refrigeration component from the multiple candidate sub-refrigeration components based on the relevant information of each candidate sub-refrigeration component.

10. A data center based on an intelligent cabinet according to claim 8, characterized in that, The determining the target sub-refrigeration component from the multiple candidate sub-refrigeration components based on the relevant information of each candidate sub-refrigeration component includes: Determine the matching score corresponding to the candidate sub-refrigeration component based on the communication distance, load rate, continuous refrigeration duration, and failure probability corresponding to each candidate sub-refrigeration component; Determine the target sub-refrigeration component from the multiple candidate sub-refrigeration components based on the matching score corresponding to each candidate sub-refrigeration component.

Images