A water-cooling energy-saving control method and system based on real-time data

Abstract

The present invention provides a water-cooled energy-saving control method and system based on real-time data, which relates to the field of energy-saving refrigeration and is applied to the water-cooled system of a data center. The water-cooled system includes multiple ventilation ducts and at least one water-cooled unit. At least one air outlet is provided on the ventilation duct, and a heat dissipation valve is arranged on the air outlet. The system includes: a pipeline setting module for determining the installation positions of the multiple ventilation ducts in the data center; a monitoring and determination module for determining multiple data collection points in the data center; a data collection module including multiple data collection components for collecting the environmental information of the data collection points where they are located; a refrigeration scheduling module for determining optimal refrigeration scheduling parameters based on the environmental information of the multiple data collection points collected by the multiple data collection components, wherein the optimal refrigeration scheduling parameters at least include the operating parameters of each water-cooled unit and the opening parameters of each heat dissipation valve, having the advantage of realizing intelligent energy-saving refrigeration of the data center.

Description

Technical Field

[0001] The present invention relates to the field of energy-saving refrigeration, and particularly relates to a water-cooled energy-saving control method and system based on real-time data. Background Art

[0002] With the rapid development and popularization of information technology, the number of data centers and their energy consumption levels have increased exponentially. The air-conditioning cooling demand of data centers has also increased significantly. Data centers continuously generate heat and have a large amount of heat. Refrigeration is required throughout the year. The energy consumption of the refrigeration system in data centers accounts for nearly half of the total energy consumption. Therefore, the energy consumption problem of the refrigeration system in data centers has attracted more and more attention.

[0003] The energy-saving control strategies of existing water-cooled machines in data centers are often configured by energy engineers based on engineering experience. However, the heat generation situation of the equipment in data centers is complex and changeable. A single strategy often cannot adapt to environmental changes, which has a greater impact on the effectiveness of the energy-saving control strategy and even endangers the safe operation of data centers.

[0004] Therefore, there is a need to provide a water-cooled energy-saving control method based on real-time data to achieve intelligent energy-saving refrigeration in data centers. Summary of the Invention

[0005] The present invention provides a water-cooled energy-saving control system based on real-time data, which is applied to the water-cooled system of a data center. The water-cooled system includes multiple ventilation ducts and at least one water-cooled unit. At least one air outlet is provided on the ventilation duct, and a heat dissipation valve is arranged on the air outlet. The system includes: a pipeline setting module, configured to determine the installation positions of the multiple ventilation ducts in the data center, where the installation positions of the multiple ventilation ducts in the data center are used to indicate the installation of the multiple ventilation ducts; a monitoring and determination module, configured to determine multiple data collection points in the data center; a data collection module, including multiple data collection components, where the positions of the multiple data collection components are determined based on the multiple data collection points in the data center, and the data collection components are used to collect the environmental information of the data collection points where they are located; a refrigeration scheduling module, configured to determine optimal refrigeration scheduling parameters based on the environmental information of the multiple data collection points collected by the multiple data collection components, where the optimal refrigeration scheduling parameters at least include the operating parameters of each water-cooled unit and the opening parameters of each heat dissipation valve.

[0006] Further, the duct setting module determines the installation positions of the multiple ventilation ducts in the data center, including: obtaining the image information of the data center; determining the position information of multiple devices in the data center according to the image information of the data center; determining the heat generation information of multiple areas in the data center based on the position information of multiple devices in the data center and the average heat generation information of each device; obtaining the position information of multiple devices and the heat generation information of multiple areas of multiple sample data centers; for each of the sample data centers, calculating the distribution similarity between the data center and the sample data center based on the position information of multiple devices and the heat generation information of multiple areas in the data center and the position information of multiple devices and the heat generation information of multiple areas in the sample data center; determining similar sample data centers from the multiple sample data centers based on the distribution similarity between the data center and each of the sample data centers; obtaining the ventilation duct installation position information of the similar sample data centers; and determining the installation positions of the multiple ventilation ducts in the data center based on the ventilation duct installation position information of the similar sample data centers and the heat generation information of multiple areas in the data center.

[0007] Further, the monitoring and determining module determines multiple data collection points of the data center, including: obtaining the data collection point distribution information of the similar sample data centers; determining multiple candidate data collection points of the data center based on the data collection point distribution information of the similar sample data centers; collecting test environment information at the multiple candidate data collection points; and screening the multiple candidate data collection points of the data center based on the test environment information collected at the multiple candidate data collection points to determine the multiple data collection points of the data center.

[0008] Further, the monitoring and determining module screens the multiple candidate data collection points of the data center based on the test environment information collected at the multiple candidate data collection points to determine the multiple data collection points of the data center, including: determining the temperature fluctuation parameter and humidity fluctuation parameter of each candidate data collection point based on the test environment information collected at the multiple candidate data collection points; and screening the multiple candidate data collection points of the data center based on the temperature fluctuation parameter and humidity fluctuation parameter of each candidate data collection point to determine the multiple data collection points of the data center. Specifically, the temperature fluctuation parameter of the candidate data collection point is calculated according to the following formula: where, S (i，temperature) is the temperature fluctuation parameter of the i-th candidate data collection point, T (i，t) is the test environment temperature at the t-th test time point collected at the i-th candidate data collection point, T (i，presets)$T_{i}$ is the preset standard temperature corresponding to the $i$-th candidate data collection point, and $T$ is the total number of test time points corresponding to the test environment information; the humidity fluctuation parameter of the candidate data collection point is calculated based on the following formula: Where, $S$ (i，h umidity) is the humidity fluctuation parameter of the $i$-th candidate data collection point, and $H$ (i，t) is the test environment humidity at the $t$-th test time point collected at the $i$-th candidate data collection point, and $H$ (i，presets) is the preset standard humidity corresponding to the $i$-th candidate data collection point;

[0009] The candidate data collection points with temperature fluctuation parameters greater than the preset temperature fluctuation parameter threshold and / or humidity fluctuation parameters greater than the preset humidity fluctuation parameter threshold are used as the data collection points of the data center.

[0010] Furthermore, the refrigeration scheduling module is further configured to: establish a temperature correlation relationship and a humidity correlation relationship between any two of the data collection points based on the test environment information collected at the multiple candidate data collection points.

[0011] Furthermore, the environment information includes environment humidity information and environment temperature information; the refrigeration scheduling module determines the optimal refrigeration scheduling parameters based on the environment information of multiple data collection points collected by the multiple data collection components, including: for each of the data collection points, based on the temperature correlation relationship and the humidity correlation relationship between any two of the data collection points, determine the temperature-correlated data collection point and the humidity-correlated data collection point of the data collection point, and based on the environment temperature information collected at the temperature-correlated data collection point and the environment humidity information collected at the humidity-correlated data collection point, perform data preprocessing on the environment humidity information and the environment temperature information collected at the data collection point to generate the preprocessed environment information corresponding to the data collection point; determine the optimal refrigeration scheduling parameters based on the preprocessed environment information corresponding to each of the data collection points.

[0012] Further, based on the ambient temperature information collected at the temperature - related data acquisition point and the ambient humidity information collected at the humidity - related data acquisition point, the refrigeration scheduling module performs data pre - processing on the ambient humidity information and ambient temperature information collected at the data acquisition point to generate the pre - processed ambient information corresponding to the data acquisition point, including: generating the current temperature sequence and the current humidity sequence of the data acquisition point based on the ambient humidity information and ambient temperature information of the data acquisition point collected at multiple time points in the current acquisition cycle; determining the current temperature fluctuation sequence of the data acquisition point based on the current temperature sequence of the data acquisition point, where an element of the current temperature fluctuation sequence is the temperature fluctuation parameter of the data acquisition point in a time period of the current acquisition cycle; determining the current humidity fluctuation sequence of the data acquisition point based on the current humidity sequence of the data acquisition point, where an element of the current humidity fluctuation sequence is the humidity fluctuation parameter of the data acquisition point in a time period of the current acquisition cycle; generating the current temperature sequence of the temperature - related data acquisition point based on the ambient temperature information of the temperature - related data acquisition point collected at multiple time points in the current acquisition cycle, and determining the current temperature fluctuation sequence of the temperature - related data acquisition point based on the current temperature sequence of the temperature - related data acquisition point; generating the current humidity sequence of the humidity - related data acquisition point based on the ambient humidity information of the humidity - related data acquisition point collected at multiple time points in the current acquisition cycle, and determining the current humidity fluctuation sequence of the humidity - related data acquisition point based on the current humidity sequence of the humidity - related data acquisition point; performing empirical mode decomposition on the current temperature fluctuation sequence of the data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current temperature fluctuation sequence of the data acquisition point; performing empirical mode decomposition on the current humidity fluctuation sequence of the data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current humidity fluctuation sequence of the data acquisition point; performing empirical mode decomposition on the current temperature sequence of the temperature - related data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current temperature sequence of the temperature - related data acquisition point; performing empirical mode decomposition on the current humidity sequence of the humidity - related data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current humidity sequence of the humidity - related data acquisition point;Based on at least one intrinsic mode component and residual corresponding to the current temperature fluctuation sequence of the data acquisition point, at least one intrinsic mode component and residual corresponding to the current humidity fluctuation sequence of the data acquisition point, at least one intrinsic mode component and residual corresponding to the current temperature sequence of the temperature-associated data acquisition point, and at least one intrinsic mode component and residual corresponding to the current humidity sequence of the humidity-associated data acquisition point, the data correction model performs data denoising and data completion on the environmental humidity information and environmental temperature information collected at the data acquisition point, and generates the preprocessed environmental information corresponding to the data acquisition point.

[0013] Further, the refrigeration scheduling module determines optimal refrigeration scheduling parameters based on the environmental information of multiple data acquisition points collected by the multiple data acquisition components, including: determining the optimal refrigeration scheduling parameters through a parameter determination model based on the preprocessed environmental information corresponding to each data acquisition point.

[0014] Further, the system further includes a fault diagnosis module, which is configured to perform fault diagnosis based on the environmental information of multiple data acquisition points collected by the multiple data acquisition components and the wind speed information of each air outlet during the operation of the water-cooled system based on the optimal refrigeration scheduling parameters.

[0015] The present invention provides a water-cooled energy-saving control method based on real-time data, which is applied to a water-cooled system in a data center. The water-cooled system includes multiple ventilation ducts and at least one water-cooled unit. At least one air outlet is provided on the ventilation duct, and a heat dissipation valve is arranged on the air outlet. The method includes: determining the installation positions of the multiple ventilation ducts in the data center, where the installation positions of the multiple ventilation ducts in the data center are used to indicate the installation of the multiple ventilation ducts; determining multiple data acquisition points in the data center; arranging multiple data acquisition components based on the multiple data acquisition points in the data center, where the data acquisition components are used to collect the environmental information of the data acquisition points where they are located; determining optimal refrigeration scheduling parameters based on the environmental information of multiple data acquisition points collected by the multiple data acquisition components, where the optimal refrigeration scheduling parameters at least include the operating parameters of each water-cooled unit and the opening parameters of each heat dissipation valve.

[0016] Compared with the prior art, the water-cooled energy-saving control method based on real-time data provided by the present invention has at least the following beneficial effects:

[0017] 1. By determining the installation positions of multiple ventilation ducts in the data center, the ventilation structure is optimized. On this basis, multiple data collection points are set, and based on the environmental information of multiple data collection points collected by multiple data collection components, the refrigeration scheduling parameters of the water-cooling system in the data center are adjusted flexibly and in real time, realizing intelligent energy-saving refrigeration in the data center.

[0018] 2. First, based on the data collection point distribution information of similar sample data centers, multiple candidate data collection points in the data center are quickly determined. On this basis, based on the temperature fluctuation parameters and humidity fluctuation parameters of each candidate data collection point, multiple candidate data collection points in the data center are screened to determine multiple data collection points in the data center that are prone to temperature fluctuations and humidity fluctuations, reducing the acquisition of invalid data and improving the efficiency of parameter determination.

[0019] 3. Through the data correction model, based on at least one intrinsic mode component and residual corresponding to the current temperature fluctuation sequence of the data collection point, at least one intrinsic mode component and residual corresponding to the current humidity fluctuation sequence of the data collection point, at least one intrinsic mode component and residual corresponding to the current temperature sequence of the temperature-correlated data collection point, and at least one intrinsic mode component and residual corresponding to the current humidity sequence of the humidity-correlated data collection point, data denoising and data completion are performed on the environmental humidity information and environmental temperature information collected at the data collection point, generating relatively accurate preprocessed environmental information corresponding to the data collection point, avoiding the situation where inaccurate data caused by a single sensor failure leads to a low matching degree of the subsequently determined optimal refrigeration scheduling parameters. Description of the Drawings

[0020] This specification will be further described by way 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:

[0021] Figure 1 is a schematic diagram of the modules of a water-cooling energy-saving control system based on real-time data shown in some embodiments of this specification;

[0022] Figure 2 is a schematic flow diagram of determining the installation positions of multiple ventilation ducts in the data center shown in some embodiments of this specification;

[0023] Figure 3 is a schematic flow diagram of generating preprocessed environmental information corresponding to a data collection point shown in some embodiments of this specification;

[0024] Figure 4 is a schematic flow diagram of a water-cooling energy-saving control method based on real-time data shown in some embodiments of this specification. Detailed implementation manners

[0025] 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.

[0026] Figure 1 is a schematic diagram of the modules of a water-cooled energy-saving control system based on real-time data shown in some embodiments of this specification. A water-cooled energy-saving control system based on real-time data can be applied to the water-cooled system of a data center. Among them, the water-cooled system includes multiple ventilation ducts and at least one water-cooled unit. At least one air outlet is provided on the ventilation duct, and a heat dissipation valve is provided on the air outlet. As Figure 1 shown, a water-cooled energy-saving control system based on real-time data may include a pipeline setting module, a monitoring and determination module, a data acquisition module, a refrigeration scheduling module, and a fault diagnosis module.

[0027] The pipeline setting module can be used to determine the installation positions of multiple ventilation ducts in the data center.

[0028] Among them, the installation positions of multiple ventilation ducts in the data center are used to indicate the installation of multiple ventilation ducts.

[0029] Figure 2 is a schematic flowchart of determining the installation positions of multiple ventilation ducts in the data center shown in some embodiments of this specification. As Figure 2 shown, in some embodiments, the pipeline setting module determines the installation positions of multiple ventilation ducts in the data center, including:

[0030] Obtain the image information of the data center. Specifically, an image acquisition device can be used to acquire the image information of different positions of the data center;

[0031] According to the image information of the data center, determine the position information of multiple devices in the data center. Specifically, the position information of multiple devices in the data center can be determined through an image recognition model based on the image information of the data center. Among them, the image recognition model can be a MobileNetv1 model, an EfficientNet model, etc.;

[0032] Based on the location information of multiple devices in the data center and the average heat generation information of each device, determine the heat generation information of multiple regions in the data center. Specifically, for each device, the heat generation per unit time of sample devices of the same model as this device in different operating states can be obtained, and the average value of the heat generation per unit time of the sample devices in different operating states is calculated, so as to calculate the average heat generation information of this device. Further, based on the location information of each device, cluster the multiple devices in the data center to determine multiple regions. For each region, the sum of the average heat generation information of the devices included in this region is used as the heat generation of this region;

[0033] Obtain the location information of multiple devices and the heat generation information of multiple regions in multiple sample data centers. Among them, the sample data center can be a data center with the installation position of the ventilation duct and data collection points planned in advance. The sample data center is a virtual data center, a real data center, or a physical model of a data center, etc.;

[0034] For each sample data center, based on the location information of multiple devices in the data center, the heat generation information of multiple regions, and the location information of multiple devices in the sample data center and the heat generation information of multiple regions, calculate the distribution similarity between the data center and the sample data center. Specifically, based on the location information of multiple devices in the data center, determine the relative position relationship between any two devices in the data center. Based on the location information of multiple regions in the data center, determine the relative position relationship between multiple regions in the data center. Based on the location information of multiple devices in the sample data center, determine the relative position relationship between any two devices in the sample data center. Based on the location information of multiple regions in the sample data center, determine the relative position relationship between multiple regions in the sample data center. Through the similarity determination model, based on the relative position relationship between any two devices in the data center, the relative position relationship between multiple regions, and the heat generation information of multiple regions, and the relative position relationship between any two devices in the sample data center, the relative position relationship between multiple regions, and the heat generation information of multiple regions, calculate the distribution similarity between the data center and the sample data center. Among them, the similarity determination model can be a BP neural network model improved based on the genetic algorithm;

[0035] Based on the distribution similarity between the data center and each sample data center, determine the similar sample data center from multiple sample data centers. Specifically, the sample data center with the largest distribution similarity can be used as the similar sample data center;

[0036] Obtain the ventilation duct installation position information of the similar sample data center;

[0037] Based on the installation location information of ventilation ducts in a similar sample data center and the heat generation information of multiple areas in the data center, determine the installation locations of multiple ventilation ducts in the data center. Specifically, an initial installation plan for multiple ventilation ducts in the data center can be generated based on the installation location information of ventilation ducts in a similar sample data center, and then the initial installation plan can be fine-tuned based on the heat generation information of multiple areas in the data center. For example, the ventilation ducts can be adjusted to be closer to the areas with higher heat generation.

[0038] The basic problems that need to be solved by the improved BP neural network model using genetic algorithms include chromosome encoding, construction of fitness functions, design of selection operators, crossover operators, and mutation operators, and the combination of the optimal individual and the algorithm.

[0039] The learning process of a neural network is a process of optimizing and learning the two continuous parameters of the network weights and thresholds. If the initial parameters are not selected properly, the algorithm is prone to falling into a local optimum. The genetic algorithm is used to determine the initial parameters of the network to avoid the defect of the algorithm falling into a local optimum. In the process of chromosome encoding, if binary encoding is used for the parameters, the encoding string will be too long, and it needs to be restored to a real number during decoding, which will affect the learning accuracy of the network and the running time of the algorithm. Therefore, real number encoding is adopted in this chapter.

[0040] Genetic algorithms basically do not utilize external information in evolutionary search. Only based on the fitness function, the fitness values of each individual in the population are used for search, and the fitness values are used to judge the excellence of individuals. Therefore, the selection of the fitness function is crucial, directly affecting the convergence speed of the genetic algorithm and whether the optimal solution can be found. Generally speaking, the fitness function is transformed from the objective function. Define the network error function:

[0041]

[0042] Among them, E A is the target fitness function, E is the individual fitness evaluation function, P is the extreme value, is the position of the Kth particle in the d and y dimensions, m is the total number of particles. Based on the fact that the smaller the objective function value, the larger the fitness value, and the larger the objective function value, the smaller the fitness value, the fitness function should take the reciprocal of the objective function, that is, the fitness function is:

[0043] F(E A ) = 1 / E A

[0044] The selection strategy uses the proportional selection operator adopted in genetic algorithms. Let the population size be M, and the fitness of individual i be F i , then the probability that individual i is selected is P i is:

[0045]

[0046] Among them, P i is the selection probability determined by affinity.

[0047] Since real - number coding is adopted, the crossover operator adopts the arithmetic crossover strategy. Suppose there are two individuals Arithmetic crossover is performed between them, and the two new individuals generated after the crossover operation are:

[0048]

[0049] The mutation operator adopts the uniform mutation strategy. Suppose there is an individual, if X = x1x2…x k …x l is the mutation point, and its value range is [U k min , U k max . After performing the uniform mutation operation on the individual at this point, a new individual X = x1x2…x′ k …x l can be obtained, where the new gene value of the mutation point is:

[0050]

[0051] Among them, x′ k is the mutation point, is the minimum mutation point, is the maximum mutation point, and r is a random number that conforms to the uniform probability distribution within [0, 1].

[0052] After the genetic algorithm training is completed, find the individual with the largest fitness value, decode each component of this individual into the corresponding parameter value, and then train the BP neural network model until the BP neural network model meets the end condition.

[0053] The monitoring and determination module can be used to determine multiple data acquisition points in the data center. Among them, the data acquisition point can be a certain location in the data center.

[0054] In some embodiments, the monitoring and determination module determines multiple data acquisition points in the data center, including:

[0055] Obtain the distribution information of data acquisition points in the similar sample data center. Specifically, the distribution information of data acquisition points can include the location information of data acquisition points in the similar sample data center;

[0056] Based on the data collection point distribution information of the similar sample data center, determine multiple candidate data collection points for the data center. Specifically, the positions of multiple candidate data collection points can be determined according to the position information of the data collection points of the similar sample data center.

[0057] Collect test environment information at multiple candidate data collection points. Specifically, the test environment information may include test environment temperature and test environment humidity.

[0058] Based on the test environment information collected at multiple candidate data collection points, screen the multiple candidate data collection points for the data center to determine multiple data collection points for the data center.

[0059] In some embodiments, the monitoring and determination module screens the multiple candidate data collection points for the data center based on the test environment information collected at multiple candidate data collection points to determine multiple data collection points for the data center, including:

[0060] Based on the test environment information collected at multiple candidate data collection points, determine the temperature fluctuation parameter and humidity fluctuation parameter of each candidate data collection point.

[0061] Based on the temperature fluctuation parameter and humidity fluctuation parameter of each candidate data collection point, screen the multiple candidate data collection points for the data center to determine multiple data collection points for the data center.

[0062] Specifically, the temperature fluctuation parameter of the candidate data collection point can be calculated based on the following formula:

[0063]

[0064] where S (i，temperature) is the temperature fluctuation parameter of the i-th candidate data collection point, T (i，t) is the test environment temperature at the t-th test time point collected at the i-th candidate data collection point, T (i，presets) is the preset standard temperature corresponding to the i-th candidate data collection point, and T is the total number of test time points corresponding to the test environment information.

[0065] The humidity fluctuation parameter of the candidate data collection point can be calculated based on the following formula:

[0066]

[0067] where S (i，h umidity ) is the humidity fluctuation parameter of the i-th candidate data collection point, H (i，t) is the test environment humidity at the t-th test time point collected at the i-th candidate data collection point, H (i，presets) is the preset standard humidity corresponding to the i-th candidate data collection point.

[0068] Candidate data collection points with a temperature fluctuation parameter greater than a preset temperature fluctuation parameter threshold and / or a humidity fluctuation parameter greater than a preset humidity fluctuation parameter threshold can be used as data collection points for the data center.

[0069] The data collection module may include multiple data collection components.

[0070] Among them, the positions of the multiple data collection components are determined based on multiple data collection points of the data center, and the data collection components are used to collect the environmental information of the data collection points where they are located.

[0071] In some embodiments, the environmental information includes environmental humidity information and environmental temperature information. The data collection component may include multiple sensors. For example, the data collection component may include a temperature sensor and a humidity sensor.

[0072] The refrigeration scheduling module can be used to determine optimal refrigeration scheduling parameters based on the environmental information of multiple data collection points collected by multiple data collection components.

[0073] Among them, the optimal refrigeration scheduling parameters at least include the operating parameters of each water-cooled chiller and the opening parameters of each heat dissipation valve. The operating parameters of the water-cooled chiller may include the load rate, the frequency of each water pump, and the rotational speed of each fan, etc.

[0074] In some embodiments, the refrigeration scheduling module is further configured to: establish a temperature correlation relationship and a humidity correlation relationship between any two data collection points based on the test environmental information collected at multiple candidate data collection points.

[0075] Specifically, based on the test environmental information collected at multiple candidate data collection points, the temperature fluctuation parameter and the humidity fluctuation parameter of each candidate data collection point in each test time period can be determined, and a test temperature fluctuation sequence and a test humidity fluctuation sequence are generated. Among them, the test temperature fluctuation sequence can be formed by arranging the temperature fluctuation parameters of the candidate data collection points in each test time period in chronological order, and the test humidity fluctuation sequence can be formed by arranging the humidity fluctuation parameters of the candidate data collection points in each test time period in chronological order. For any two data collection points, the test temperature fluctuation sequence similarity of the test temperature fluctuation sequences of the two data collection points can be calculated. When the test temperature fluctuation sequence similarity is greater than a preset test temperature fluctuation sequence similarity threshold, there is a temperature correlation relationship between the two data collection points. The test humidity fluctuation sequence similarity of the test humidity fluctuation sequences of the two data collection points can be calculated. When the test humidity fluctuation sequence similarity is greater than a preset test humidity fluctuation sequence similarity threshold, there is a humidity correlation relationship between the two data collection points.

[0076] For example, the similarity of the test temperature fluctuation sequences of two data acquisition points can be calculated based on the following formula:

[0077]

[0078] where S (i，j) is the similarity of the test temperature fluctuation sequences of the i-th data acquisition point and the j-th data acquisition point, M is a preset parameter, and V (i，n) is the value of the n-th element of the test temperature fluctuation sequence of the i-th data acquisition point, and V (j，n) is the value of the n-th element of the test temperature fluctuation sequence of the j-th data acquisition point, and N is the total number of elements of the test temperature fluctuation sequence.

[0079] In some embodiments, the refrigeration scheduling module determines the optimal refrigeration scheduling parameters based on the environmental information of multiple data acquisition points collected by multiple data acquisition components, including:

[0080] For each data acquisition point, based on the temperature correlation relationship and humidity correlation relationship between any two data acquisition points, determine the temperature-correlated data acquisition point and humidity-correlated data acquisition point of the data acquisition point. Based on the ambient temperature information collected at the temperature-correlated data acquisition point and the ambient humidity information collected at the humidity-correlated data acquisition point, perform data preprocessing on the ambient humidity information and ambient temperature information collected at the data acquisition point to generate the preprocessed environmental information corresponding to the data acquisition point;

[0081] Determine the optimal refrigeration scheduling parameters based on the preprocessed environmental information corresponding to each data acquisition point.

[0082] Figure 3 is a schematic flowchart of generating the preprocessed environmental information corresponding to the data acquisition point according to some embodiments of this specification. As Figure 3 shown, in some embodiments, the refrigeration scheduling module performs data preprocessing on the ambient humidity information and ambient temperature information collected at the data acquisition point based on the ambient temperature information collected at the temperature-correlated data acquisition point and the ambient humidity information collected at the humidity-correlated data acquisition point to generate the preprocessed environmental information corresponding to the data acquisition point, including:

[0083] Based on the ambient humidity information and ambient temperature information of the data acquisition point collected at multiple time points in the current acquisition cycle, generate the current temperature sequence and current humidity sequence of the data acquisition point;

[0084] Based on the current temperature sequence of the data acquisition point, determine the current temperature fluctuation sequence of the data acquisition point, where an element of the current temperature fluctuation sequence is the temperature fluctuation parameter of the data acquisition point in a time period of the current acquisition cycle;

[0085] Based on the current humidity sequence of the data acquisition point, determine the current humidity fluctuation sequence of the data acquisition point, where an element of the current humidity fluctuation sequence is the humidity fluctuation parameter of the data acquisition point in a time period of the current acquisition cycle;

[0086] Based on the ambient temperature information of the temperature-related data acquisition point collected at multiple time points in the current acquisition cycle, generate the current temperature sequence of the temperature-related data acquisition point, and based on the current temperature sequence of the temperature-related data acquisition point, determine the current temperature fluctuation sequence of the temperature-related data acquisition point;

[0087] Based on the ambient humidity information of the humidity-related data acquisition point collected at multiple time points in the current acquisition cycle, generate the current humidity sequence of the humidity-related data acquisition point, and based on the current humidity sequence of the humidity-related data acquisition point, determine the current humidity fluctuation sequence of the humidity-related data acquisition point;

[0088] Perform empirical mode decomposition on the current temperature fluctuation sequence of the data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current temperature fluctuation sequence of the data acquisition point;

[0089] Perform empirical mode decomposition on the current humidity fluctuation sequence of the data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current humidity fluctuation sequence of the data acquisition point;

[0090] Perform empirical mode decomposition on the current temperature sequence of the temperature-related data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current temperature sequence of the temperature-related data acquisition point;

[0091] Perform empirical mode decomposition on the current humidity sequence of the humidity-related data acquisition point to generate at least one intrinsic mode function component and a residual corresponding to the current humidity sequence of the humidity-related data acquisition point;

[0092] Based on at least one intrinsic mode component and residual corresponding to the current temperature fluctuation sequence of the data acquisition point, at least one intrinsic mode component and residual corresponding to the current humidity fluctuation sequence of the data acquisition point, at least one intrinsic mode component and residual corresponding to the current temperature sequence of the temperature-related data acquisition point, and at least one intrinsic mode component and residual corresponding to the current humidity sequence of the humidity-related data acquisition point, the data correction model performs data denoising and data completion on the ambient humidity information and ambient temperature information collected at the data acquisition point, and generates preprocessed ambient information corresponding to the data acquisition point. Among them, the data correction model can be a machine learning model such as an Artificial Neural Network (ANN) model, a Recurrent Neural Networks (RNN) model, a Long Short-Term Memory (LSTM) model, or a Bidirectional Recurrent Neural Network (BRNN) model.

[0093] In some embodiments, the refrigeration scheduling module determines the optimal refrigeration scheduling parameters based on the ambient information of multiple data acquisition points collected by multiple data acquisition components, including:

[0094] The parameter determination model determines the optimal refrigeration scheduling parameters based on the preprocessed ambient information corresponding to each data acquisition point. The parameter determination model can be a GRU model. GRU (Gate Recurrent Unit) is a type of Recurrent Neural Network (RNN). Like LSTM (Long-Short Term Memory), it is also proposed to solve problems such as long-term memory and gradients in backpropagation. In many cases, the actual performance of GRU and LSTM is similar. Compared with LSTM, using GRU can achieve comparable results, and it is easier to train, which can greatly improve the training efficiency. Therefore, GRU is often preferred in many cases. The input-output structure of GRU is similar to that of an ordinary RNN, and its internal idea is similar to that of LSTM. Compared with LSTM, GRU has one less "gate control" inside, with fewer parameters, but it can still achieve functions comparable to LSTM.

[0095] GRU processes the hidden state h at the previous moment t-1 and the external input information x at the current moment t through two multiplicative gates, namely the reset gate and the update gate, so as to update the hidden state h at the current moment t to implement model training, and the calculation rules are as follows:

[0096] z t =σ(W (z)x t +U (z) h t-1 )

[0097] r t =σ(W (r) x t +U (r) h t-1 )

[0098] h′ t =tanh(Wx t +r t ⊙Uh t-1 )

[0099]

[0100] wherein, W (z) 、W (r) 、U (z) 、U are response input weight matrices; h t is the current memory content; ⊙ is the Hadamard operation, tanh() is the hyperbolic tangent function; σ is the Sigmoid activation function,

[0101] The fault diagnosis module can be used to perform fault diagnosis based on the environmental information of multiple data collection points collected by multiple data collection components and the wind speed information of each air outlet during the operation of the water cooling system based on the optimal refrigeration scheduling parameters.

[0102] Specifically, according to the optimal refrigeration scheduling parameters, during the operation of the water cooling system based on the optimal refrigeration scheduling parameters, the environmental temperature and environmental humidity of multiple data collection points can be predicted, the differences between the predicted environmental temperature and environmental humidity of multiple data collection points and the environmental information of multiple data collection points collected by multiple data collection components can be calculated to determine the refrigeration effect, and based on the wind speed information of each air outlet, it can be determined whether the opening degree of each cooling valve is abnormal.

[0103] Figure 4 is a schematic flowchart of a water cooling energy-saving control method based on real-time data according to some embodiments of the present specification. A water cooling energy-saving control method based on real-time data can be applied to a water cooling energy-saving control system based on real-time data, as Figure 4 shown, a water cooling energy-saving control method based on real-time data may include the following steps:

[0104] Step 410, determining the installation positions of multiple ventilation ducts in the data center, wherein the installation positions of the multiple ventilation ducts in the data center are used to indicate the installation of the multiple ventilation ducts;

[0105] Step 420: Determine multiple data collection points in the data center;

[0106] Step 430: Set multiple data collection components based on the multiple data collection points in the data center. Among them, the data collection components are used to collect the environmental information of the data collection points where they are located;

[0107] Step 440: Determine the optimal refrigeration scheduling parameters based on the environmental information of the multiple data collection points collected by the multiple data collection components. Among them, the optimal refrigeration scheduling parameters at least include the operating parameters of each water-cooled chiller and the opening parameters of each cooling valve.

[0108] For more descriptions of a water-cooled energy-saving control method based on real-time data, reference can be made to the relevant descriptions of a water-cooled energy-saving control system based on real-time data, which will not be elaborated here.

[0109] 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 water cooling energy-saving control system based on real-time data, applied to the water cooling system of a data center, wherein: The water cooling system comprises a plurality of ventilation ducts and at least one water cooling unit, the ventilation duct is provided with at least one air outlet, and the air outlet is provided with a heat dissipation valve, and is characterized in that it comprises: a duct setting module, used to determine installation positions of the plurality of ventilation ducts in the data center, wherein the installation positions of the plurality of ventilation ducts in the data center are used to indicate installation of the plurality of ventilation ducts; A monitoring and determining module, used to determine a plurality of data collection points of the data center; A data collection module, comprising a plurality of data collection components, wherein the positions of the plurality of data collection components are determined based on a plurality of data collection points of the data center, and the data collection components are used to collect environmental information of the data collection points; A refrigeration scheduling module, used to determine optimal refrigeration scheduling parameters based on the environmental information of multiple data collection points collected by the multiple data collection components, wherein the optimal refrigeration scheduling parameters at least include the operating parameters of each of the water cooling units and the opening parameters of each of the heat dissipation valves; The pipeline setting module determines the installation positions of the plurality of ventilation pipelines in the data center, including: Acquiring image information of the data center; Determining location information of a plurality of devices in the data center according to the image information of the data center; Determine the heat generation information of multiple areas of the data center based on the location information of multiple devices in the data center and the average heat generation information of each of the devices; Obtain location information of multiple devices in multiple sample data centers and heat generation information of multiple areas; For each of the sample data centers, based on the location information of multiple devices and the calorific value information of multiple areas of the data center and the location information of multiple devices and the calorific value information of multiple areas of the sample data center, calculate the distribution similarity between the data center and the sample data center; Determining similar sample data centers from the multiple sample data centers based on the distribution similarity between the data center and each of the sample data centers; Obtaining ventilation duct installation location information of the similar sample data center; Based on the ventilation duct installation position information of the similar sample data center and the heating value information of multiple areas of the data center, the installation positions of the multiple ventilation ducts in the data center are determined.

2. A water cooling energy-saving control system based on real-time data according to claim 1, characterized in that: The monitoring and determining module determines a plurality of data collection points of the data center, including: Obtaining data collection point distribution information of the similar sample data center; Determine multiple candidate data collection points of the data center based on the data collection point distribution information of the similar sample data center; Collecting test environment information at the plurality of candidate data collection points; Based on the test environment information collected at the multiple candidate data collection points, multiple candidate data collection points of the data center are screened to determine the multiple data collection points of the data center.

3. A water cooling energy-saving control system based on real-time data according to claim 2, characterized in that: The monitoring and determining module screens multiple candidate data collection points of the data center based on the test environment information collected at the multiple candidate data collection points to determine the multiple data collection points of the data center, including: Based on the test environment information collected at the plurality of candidate data collection points, determining a temperature fluctuation parameter and a humidity fluctuation parameter of each of the candidate data collection points; Based on the temperature fluctuation parameter and the humidity fluctuation parameter of each of the candidate data collection points, multiple candidate data collection points of the data center are screened to determine multiple data collection points of the data center; Specifically, the temperature fluctuation parameters of the candidate data collection points are calculated based on the following formula: in, is the temperature fluctuation parameter of the i-th candidate data collection point, is the test environment temperature at the tth test time point collected at the i-th candidate data collection point, is the preset standard temperature corresponding to the i-th candidate data collection point, The total number of test time points corresponding to the test environment information; The humidity fluctuation parameters of the candidate data collection points are calculated based on the following formula: in, is the humidity fluctuation parameter of the i-th candidate data collection point, is the test environment humidity at the t-th test time point collected at the i-th candidate data collection point, is the preset standard humidity corresponding to the i-th candidate data collection point; The candidate data collection points whose temperature fluctuation parameters are greater than a preset temperature fluctuation parameter threshold and / or whose humidity fluctuation parameters are greater than a preset humidity fluctuation parameter threshold are used as data collection points of the data center.

4. A water cooling energy-saving control system based on real-time data according to claim 2 or 3, characterized in that: The refrigeration scheduling module is also used for: Based on the test environment information collected at the plurality of candidate data collection points, a temperature association relationship and a humidity association relationship between any two of the data collection points are established.

5. The water cooling energy-saving control system based on real-time data according to claim 4 is characterized in that: The environmental information includes environmental humidity information and environmental temperature information; The refrigeration scheduling module determines the optimal refrigeration scheduling parameters based on the environmental information of the multiple data collection points collected by the multiple data collection components, including: For each of the data collection points, based on the temperature correlation relationship and the humidity correlation relationship between any two of the data collection points, determine the temperature-related data collection point and the humidity-related data collection point of the data collection point, and based on the ambient temperature information collected at the temperature-related data collection point and the ambient humidity information collected at the humidity-related data collection point, perform data preprocessing on the ambient humidity information and ambient temperature information collected at the data collection point to generate preprocessed environmental information corresponding to the data collection point; Based on the pre-processed environmental information corresponding to each of the data collection points, optimal refrigeration scheduling parameters are determined.

6. The water cooling energy-saving control system based on real-time data according to claim 5 is characterized in that: The refrigeration scheduling module performs data preprocessing on the ambient humidity information and ambient temperature information collected at the data collection point based on the ambient temperature information collected at the temperature-related data collection point and the ambient humidity information collected at the humidity-related data collection point, and generates preprocessed environmental information corresponding to the data collection point, including: Based on the ambient humidity information and ambient temperature information of the data collection point collected at multiple time points in the current collection cycle, generating a current temperature sequence and a current humidity sequence of the data collection point; Based on the current temperature sequence of the data acquisition point, determine the current temperature fluctuation sequence of the data acquisition point, wherein an element of the current temperature fluctuation sequence is a temperature fluctuation parameter of the data acquisition point in a time period of the current acquisition cycle; Based on the current humidity sequence of the data collection point, determining the current humidity fluctuation sequence of the data collection point, wherein an element of the current humidity fluctuation sequence is a humidity fluctuation parameter of the data collection point in a time period of the current collection cycle; Based on the ambient temperature information of the temperature-related data collection point collected at multiple time points in the current collection period, a current temperature sequence of the temperature-related data collection point is generated, and based on the current temperature sequence of the temperature-related data collection point, a current temperature fluctuation sequence of the temperature-related data collection point is determined; Based on the environmental humidity information of the humidity-related data collection point collected at multiple time points in the current collection period, a current humidity sequence of the humidity-related data collection point is generated, and based on the current humidity sequence of the humidity-related data collection point, a current humidity fluctuation sequence of the humidity-related data collection point is determined; Performing empirical mode decomposition on the current temperature fluctuation sequence of the data collection point to generate at least one intrinsic mode component and residual corresponding to the current temperature fluctuation sequence of the data collection point; Performing empirical mode decomposition on the current humidity fluctuation sequence of the data collection point to generate at least one intrinsic mode component and residual corresponding to the current humidity fluctuation sequence of the data collection point; Performing empirical mode decomposition on the current temperature sequence of the temperature-related data collection point to generate at least one intrinsic mode component and residual corresponding to the current temperature sequence of the temperature-related data collection point; Performing empirical mode decomposition on the current humidity sequence of the humidity-related data collection point to generate at least one intrinsic mode component and residual corresponding to the current humidity sequence of the humidity-related data collection point; Through the data correction model, based on at least one intrinsic modal component and residual corresponding to the current temperature fluctuation sequence of the data collection point, at least one intrinsic modal component and residual corresponding to the current humidity fluctuation sequence of the data collection point, at least one intrinsic modal component and residual corresponding to the current temperature sequence of the temperature-related data collection point, and at least one intrinsic modal component and residual corresponding to the current humidity sequence of the humidity-related data collection point, data denoising and data completion are performed on the ambient humidity information and ambient temperature information collected at the data collection point to generate preprocessed environmental information corresponding to the data collection point.

7. The water cooling energy-saving control system based on real-time data according to claim 6 is characterized in that: The refrigeration scheduling module determines the optimal refrigeration scheduling parameters based on the environmental information of the multiple data collection points collected by the multiple data collection components, including: The optimal refrigeration scheduling parameters are determined by a parameter determination model based on the pre-processed environmental information corresponding to each of the data collection points.

8. A water cooling energy-saving control system based on real-time data according to any one of claims 1 to 3, characterized in that: It also includes a fault diagnosis module for performing fault diagnosis based on the environmental information of multiple data collection points collected by the multiple data collection components and the wind speed information of each air outlet during the operation of the water cooling system based on the optimal refrigeration scheduling parameters.

9. A water cooling energy-saving control method based on real-time data, applied to a water cooling system of a data center, and operated in a water cooling energy-saving control system based on real-time data as described in any one of claims 1 to 8, wherein: The water cooling system comprises a plurality of ventilation ducts and at least one water cooling unit, the ventilation duct is provided with at least one air outlet, and the air outlet is provided with a heat dissipation valve, and is characterized in that it comprises: Determining installation positions of the plurality of ventilation ducts in the data center, wherein the installation positions of the plurality of ventilation ducts in the data center are used to indicate installation of the plurality of ventilation ducts; determining a plurality of data collection points of the data center; A plurality of data collection components are set based on the plurality of data collection points of the data center, wherein the data collection components are used to collect environmental information of the data collection points; Based on the environmental information of multiple data collection points collected by the multiple data collection components, optimal refrigeration scheduling parameters are determined, wherein the optimal refrigeration scheduling parameters at least include the operating parameters of each of the water cooling units and the opening parameters of each of the heat dissipation valves.

Images