Machine room energy-saving control method and device and storage medium

Abstract

The invention provides a machine room energy-saving control method and device and a storage medium. According to the method, training is carried out based on historical temperature data to obtain a temperature prediction model, the current machine room operation time is input into the temperature prediction model to obtain a target prediction temperature, and then air conditioner control information is obtained through genetic algorithm optimization based on the target prediction temperature and sensor temperature information. Each air conditioner in the machine room environment is controlled according to the air conditioner control information including the target set temperature and the target return air temperature of each air conditioner, and energy-saving adjustment is performed on the machine room environment in time, so that the training time is shortened, the real-time environment change is responded more quickly, the air conditioner setting is adjusted in real time to realize machine room energy saving, and an active optimization mechanism is adopted to improve the training efficiency. The limitation of large local optimization errors is avoided, it is ensured that the system can achieve the optimal efficiency in different situations, the energy utilization rate is improved, then the deployment cost is reduced, and the safety of the data center is improved through safety monitoring and maintenance strategies.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data center energy saving, and in particular to a machine room energy saving control method and device and a storage medium. BACKGROUND

[0002] As the core infrastructure of the digital era, the cooling control of the data machine room is very important for the stable and safe operation of the machine room. The energy consumption is mainly concentrated on the operation of the server and the cooling system.

[0003] In the prior art, the cooling control of the data machine room is generally controlled by an artificial intelligence automatic adjustment method, and is usually realized based on a reinforcement learning model.

[0004] However, the reinforcement learning model needs a large amount of training data and the training data types are various, the training time is long, and it cannot quickly respond to real-time environmental changes, thereby resulting in high deployment cost of the existing scheme. SUMMARY

[0005] The embodiments of the present application provide a machine room energy saving control method and device and a storage medium to solve the defect that the existing scheme has high deployment cost when cooling adjustment of the data machine room is performed due to long training time and lack of real effectiveness of training data.

[0006] In a first aspect, the embodiments of the present application provide a machine room energy saving control method, comprising:

[0007] obtaining historical temperature data of a machine room, wherein the historical temperature data is temperature data of a preset continuous time period in the machine room environment;

[0008] training a neural network based on a self-attention mechanism by taking the historical temperature data as training data to obtain a temperature prediction model;

[0009] obtaining sensor temperature information of each sensor in the machine room at the current machine room running time;

[0010] inputting the current machine room running time into the temperature prediction model to obtain a target predicted temperature;

[0011] obtaining air conditioner control information by a genetic algorithm based on the target predicted temperature and the sensor temperature information;

[0012] controlling each air conditioner in the machine room environment according to the air conditioner control information.

[0013] In a possible implementation, the obtaining of the air conditioner control information by the genetic algorithm based on the target predicted temperature and the sensor temperature information comprises:

[0014] obtaining energy consumption parameter information of the machine room;

[0015] based on the energy consumption parameter information, establishing a target function according to the target predicted temperature and the sensor temperature information;

[0016] controlling the convergence of the computer room environment temperature to the target predicted temperature through the target function;

[0017] obtaining the target set temperature and the target return air temperature of each air conditioner when the convergence is completed, and determining the target set temperature and the target return air temperature of each air conditioner as the air conditioner control information.

[0018] In a possible implementation, the air conditioner control information includes a target set temperature and a target return air temperature.

[0019] Correspondingly, the controlling the air conditioners in the computer room environment according to the air conditioner control information includes:

[0020] If the air conditioner is in an on state, obtaining the number of times that the target set temperature in the air conditioner control information exceeds a preset maximum safety temperature as a first over-target number;

[0021] obtaining the compressor working time length and the current return air temperature, and taking the first over-target number, the compressor working time length and the current return air temperature as on state information;

[0022] controlling the air conditioner to be turned off according to the on state information;

[0023] If the air conditioner is in an off state, obtaining the number of times that the target set temperature in the air conditioner control information is less than a preset minimum safety temperature as a second over-target number;

[0024] obtaining the current return air temperature, and taking the second over-target number and the current return air temperature as off state information;

[0025] controlling the air conditioner to be turned on according to the off state information.

[0026] In a possible implementation, the controlling the air conditioner to be turned off according to the on state information includes:

[0027] if the first over-target number is greater than a preset number, the compressor working time length is less than a preset working time length, and the current return air temperature is less than a preset maximum safety temperature, turning off the air conditioner;

[0028] otherwise, keeping the current state of the air conditioner.

[0029] In a possible implementation, the controlling the air conditioner to be turned on according to the off state information includes:

[0030] If the second number of times the target is exceeded is greater than the preset number, or if the current return air temperature is greater than the preset maximum safe temperature, the air conditioner is turned on;

[0031] Otherwise, maintain the current state of the air conditioner.

[0032] In one possible implementation, after controlling each air conditioner in the computer room environment according to the obtained air conditioning control information, the method further includes:

[0033] After control is achieved, the real-time return air temperature of each air conditioner and the real-time operating temperature of each sensor are obtained.

[0034] If the real-time return air temperature is detected to be greater than the preset abnormal return air temperature, or the real-time operating temperature is greater than the sensor's maximum temperature, then the corresponding air conditioner or sensor will be marked as an abnormal device.

[0035] Obtain the device information of the abnormal device, and generate alarm information based on the device information to remind the operation and maintenance personnel.

[0036] In one possible implementation, prior to acquiring the historical temperature data, the method further includes:

[0037] Multiple sets of test control parameters are generated using a random walk algorithm, and these test control parameters represent the target test states for each air conditioner.

[0038] Control each air conditioner in the computer room environment according to the test control parameters;

[0039] After a preset test time, the action time for achieving the target test state is obtained;

[0040] If the action time is less than the preset test time, the corresponding test control parameter is marked as a qualified test parameter, and the number of qualified test parameters is recorded to obtain the number of qualified tests.

[0041] If the number of qualified tests exceeds the qualified threshold, then the step of "acquiring historical temperature data" is executed.

[0042] If the number of qualified tests is less than or equal to the qualified threshold, maintenance information is generated to remind maintenance personnel.

[0043] Secondly, embodiments of this application provide a data center energy-saving control device, comprising:

[0044] The acquisition module is used to acquire historical temperature data of the computer room, wherein the historical temperature data is temperature data of a preset continuous time period in the computer room environment, and the historical temperature data is used as training data to train a self-attention neural network to obtain a temperature prediction model.

[0045] The processing module is used to obtain the sensor temperature information of each sensor in the current computer room during the current computer room operation time, and input the current computer room operation time into the temperature prediction model to obtain the target predicted temperature.

[0046] The control module is used to calculate air conditioning control information based on the target predicted temperature and the sensor temperature information using a genetic algorithm, and to control the set temperature and return air temperature of each air conditioner in the computer room environment according to the obtained air conditioning control information.

[0047] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0048] The memory stores computer-executed instructions;

[0049] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0050] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0051] The data center energy-saving control method, device, and storage medium provided in this application train a neural network based on a self-attention mechanism to obtain a temperature prediction model based on historical temperature data over a continuous period of time within the data center environment. Then, the current data center operating time is input into the temperature prediction model to obtain the target predicted temperature. Based on the target predicted temperature and sensor temperature information, a genetic algorithm is used to optimize and obtain air conditioning control information, including the target set temperature and target return air temperature of each air conditioner. The air conditioning control information is used to control each air conditioner within the data center environment, enabling timely energy-saving adjustments to the data center environment. Compared to existing solutions that obtain predicted temperatures based on reinforcement learning models, this application requires more direct and effective training data, shortens training time, responds more quickly to real-time environmental changes, and can instantly adjust air conditioning settings to achieve data center energy savings. This reduces deployment costs and improves data center security through security monitoring and maintenance strategies. Attached Figure Description

[0052] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0053] Figure 1 Flowchart of the data center energy-saving control method provided in this application Figure 1 ;

[0054] Figure 2 Flowchart of the data center energy-saving control method provided in this application Figure 2 ;

[0055] Figure 3 A schematic diagram of the energy-saving control device for the computer room provided in this application;

[0056] Figure 4 A schematic diagram of the structure of the electronic device provided in this application.

[0057] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0058] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0059] Traditional data centers, as core infrastructure in the digital age, support the information technology needs of various industries. Energy consumption in traditional data centers is primarily concentrated on server operation and cooling systems. However, due to their small size, aging equipment, and relatively low technological level, traditional manual adjustment methods are difficult to implement, have limited parameters, low frequency, and poor effectiveness—all key factors contributing to high energy consumption. Therefore, how to improve the energy efficiency of data centers at low cost has become a key concern for various enterprises and institutions.

[0060] Introducing artificial intelligence (AI) technology to improve energy efficiency is considered a key initiative. AI can achieve more intelligent and efficient energy use by monitoring data center operations in real time, predicting load demands, optimizing server configurations, and adjusting cooling systems. This approach not only helps reduce energy costs and improve data center operating efficiency but also reduces adverse environmental impacts, contributing to sustainable development.

[0061] Existing AI-based temperature control methods utilize historical data under random control parameters of air conditioners for model training. However, the impact of air conditioner temperature settings on the computer room environment is delayed, resulting in training data lacking real-world validity. Consequently, the trained model fails to achieve good generalization performance. In real-world conditions, air conditioners are affected by protection mechanisms, limiting the magnitude and frequency of adjustments. Furthermore, training through reinforcement learning models requires long training times, incurs high deployment costs, and suffers from significant computational latency. Moreover, the training data only utilizes historical best-matching results as the current optimal choice, preventing energy-saving effects from exceeding the energy-saving limits of traditional methods. This leads to a local optima, where the potential for further energy savings persists.

[0062] Therefore, this application provides a data center energy-saving control method, device, and storage medium. Based on historical temperature data, a self-attention mechanism-based neural network is trained to obtain a temperature prediction model. The current data center operating time is input into the temperature prediction model to obtain the target predicted temperature. Based on the target predicted temperature and sensor temperature information, a genetic algorithm is used to optimize and obtain air conditioning control information. By implementing the target set temperature and target return air temperature of each air conditioner indicated by the air conditioning control information, the air conditioning temperature of the data center is adjusted to achieve energy saving. Through the active optimization mechanism, the limitations of large local optimization errors are avoided, ensuring that the system can achieve optimal performance under different situations, shortening the training time, responding more quickly to real-time environmental changes, thereby reducing deployment costs, and improving the security of the data center through security monitoring and maintenance strategies.

[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0064] Figure 1 Flowchart of the data center energy-saving control method provided in this application Figure 1 ,like Figure 1 As shown, the method includes:

[0065] S101. Obtain historical temperature data and use the historical temperature data as training data to train a self-attention neural network to obtain a temperature prediction model, wherein the historical temperature data is temperature data of a continuous time period in the computer room.

[0066] Specifically, before acquiring a temperature prediction model based on historical temperature data, sensitivity testing is conducted to detect the sensitivity of various sensors and monitoring systems in the computer room environment to changes in air conditioning parameters. After all sensors and monitoring systems pass the test, historical temperature data is acquired, and sensitivity testing is conducted to provide reliable input data support for subsequent prediction and control modules. The preset model is the encoder part of the Transformer architecture. The encoder is a self-attention neural network that can effectively process time series data.

[0067] S102. Obtain the current operating time of the computer room and the current sensor temperature information of each sensor in the computer room. Input the current operating time of the computer room into the temperature prediction model to obtain the target predicted temperature.

[0068] Specifically, after obtaining the temperature prediction model, the current computer room operating time is input into the temperature prediction model. The time-temperature dependency relationship is captured by the self-attention mechanism in the temperature prediction model to obtain the predicted temperature data, i.e. the target predicted temperature. The temperature data recorded by each sensor in the current computer room environment is also obtained to perform inference and optimization based on the target predicted temperature and sensor temperature information.

[0069] S103. Based on the target predicted temperature and the sensor temperature information, obtain air conditioning control information through a genetic algorithm, and control the set temperature and return air temperature of each air conditioner in the computer room environment according to the air conditioning control information.

[0070] Specifically, after obtaining the target predicted temperature and sensor temperature information, the optimal air conditioning set temperature and return air temperature are obtained based on the target predicted temperature, the cooling standard required by the cabinet load, and the energy consumption index. This is also the target set temperature and target return air temperature of each air conditioner, which is to say, the air conditioning control information is obtained. The air conditioning control information can minimize air conditioning energy consumption while ensuring that the sensor temperature converges to the target predicted temperature.

[0071] The data center energy-saving control method provided in this application embodiment trains a self-attention mechanism-based neural network based on historical temperature data to obtain a temperature prediction model. The current data center operating time is input into the temperature prediction model to obtain the target predicted temperature. Based on the target predicted temperature and sensor temperature information, the air conditioning control information is obtained through optimization using a genetic algorithm. By implementing the target set temperature and target return air temperature of each air conditioner indicated by the air conditioning control information, the air conditioning temperature of the data center is adjusted to achieve energy saving. This shortens the training time, responds more quickly to real-time environmental changes, reduces deployment costs, and improves the security of the data center through security monitoring and maintenance strategies.

[0072] Figure 2 Flowchart of the data center energy-saving control method provided in this application Figure 2 ,like Figure 1 As shown, in this embodiment... Figure 3 Based on the examples, the energy-saving control method for computer rooms is described in detail. The method includes:

[0073] S201. Obtain the number of qualified tests using a random walk algorithm. If the number of qualified tests is greater than the qualified threshold, obtain the historical temperature data.

[0074] Specifically, the random walk algorithm is used to simulate the changes in air conditioning parameters and test the sensitivity of each air conditioner in the computer room environment at different speeds and amplitudes. Historical temperature data is obtained after the test is passed.

[0075] Among them, for the simulated test parameters, the range of parameters to be tested is determined, the maximum value and the minimum value of the test parameters are determined, and an initial test value r is randomly selected from the range of test parameters as the starting point of the test. In each iteration, the current parameter value is randomly increased, decreased or disturbed to obtain multiple sets of test control parameters. The test control parameters are used to indicate the target test state of each air conditioner, and each air conditioner in the computer room environment is controlled according to each test control parameter.

[0076] After a preset test time, the action time to achieve the target test state is obtained. If the action time is less than the preset test time, the corresponding test control parameter is marked as a qualified test parameter, and the number of qualified test parameters is recorded as the number of qualified test parameters. If the number of qualified test parameters is less than the qualified threshold, maintenance information is generated to remind the operation and maintenance personnel. Sensitivity testing can ensure efficient response to changes in the data center environment and provide a reliable foundation for energy efficiency optimization.

[0077] S202. Using the historical temperature data as training data, train the neural network based on the self-attention mechanism to obtain a temperature prediction model.

[0078] Specifically, after passing the sensitivity test, the original sensor data of each sensor in the computer room environment is obtained, and the original sensor data is cleaned, missing values ​​are handled and standardized to ensure the quality and consistency of the input sequence in order to obtain historical temperature data. The historical temperature data includes the temperature data of N sensors in the first T-1 minutes, that is, the historical temperature data includes temperature data of multiple time steps. These data form a time series set in chronological order.

[0079] Furthermore, each column in the time series dataset represents a time step, which includes a set S = {s} of temperature data from multiple time steps. (1) ,s(2) ,...,s (T) The time-step temperature data is used to indicate the temperature data sets at various times within a continuous preset time period. For each time-step temperature data set... The real-time temperature data set includes temperature data recorded by each sensor and air conditioning temperature information. The air conditioning temperature information is used to indicate the set temperature of each air conditioner in the computer room environment. Let be the data value collected by the i-th sensor at time t. This represents the set temperature value for the Lth air conditioner;

[0080] Furthermore, when training the preset Transformer model, T epochs are set for iterative training. At the end of each epoch, the model performance is evaluated using a validation set to determine whether the model is overfitting or underfitting. An early stopping strategy is adopted to prevent the model from overfitting, and the root mean square error (RMSE) is selected as the loss function, which can effectively measure the model performance in regression problems. This allows the trained temperature prediction model to establish correlations through different positions in time series data and use a self-attention mechanism to capture patterns and dependencies in the sequence, thereby enabling the prediction of the temperature value at time T.

[0081] Furthermore, the parameters of the preset Transformer model were fine-tuned. The number of layers in the TransformerEncoder was set, and the number of attention heads in each layer helped the model better capture key information from the input sequence. An appropriate number of encoder layers was selected to ensure the model had expressive power in feature extraction while avoiding overfitting. In the model's embedding part, the embedding dimension was set to ensure effective representation of the input data. For the fully connected layers, a reasonable number of hidden units was used to increase the model's non-linear fitting ability. For batch training, a moderate batch size was chosen to fully utilize hardware resources while maintaining model training stability. To prevent the model from rapidly converging to local minima, an appropriate initial learning rate was set, and a learning rate decay strategy was used to gradually reduce the learning rate for a more refined search of the parameter space.

[0082] S203. Obtain the current operating time of the computer room and the sensor temperature information of each sensor in the computer room, and input the current operating time of the computer room into the temperature prediction model to obtain the target predicted temperature;

[0083] Specifically, after obtaining the temperature prediction model, the current data center uptime is input into the encoder of the temperature prediction model. The model then passes through the first self-attention layer, where the self-attention mechanism calculates the association weights between each location and all other locations. After passing through the self-attention layer, the representation of each location undergoes non-linear transformation and feature extraction via a feedforward neural network. This layer processes and modifies the feature representation of each location. In the final layer of the encoder, global average pooling and other methods are used to merge the feature representations of all locations into a single vector. This vector is then passed through a fully connected layer for final prediction to obtain the target predicted temperature.

[0084] S204. Establish a target function based on the target predicted temperature and the sensor temperature information. Through the target function, control the convergence of the computer room ambient temperature to the target predicted temperature, and obtain the target set temperature and target return air temperature of each air conditioner when the convergence is completed.

[0085] Specifically, after obtaining the target predicted temperature and sensor temperature information, a genetic algorithm is used to converge the overall temperature of the computer room towards the target value, that is, all sensors converge towards the target temperature. The objective function is shown in the following formula:

[0086]

[0087] in, X represents the temperature of each sensor, that is, the current temperature of each sensor. target This indicates the target temperature for the entire computer room, which is also the target predicted temperature. This represents the set temperature of the l-th air conditioner at time t, which is also the target set temperature. This represents the return air temperature of the l-th air conditioner at time t, which is also the target return air temperature. Let represent the ambient temperature around the l-th air conditioner at time t. Here, α and β are weighting coefficients used to balance the importance of temperature convergence and energy minimization. E AC For energy consumption parameters, the impact of set temperature, return air temperature, and ambient temperature on air conditioning energy consumption was considered. δ1 and δ2 are the return air temperature correction coefficient and ambient temperature correction coefficient, respectively. A constraint was set on the set temperature range of the terminal air conditioner to ensure it remains within a reasonable range. The constraint conditions are as follows:

[0088]

[0089] in, The lower limit of the preset constraint temperature, To preset the upper limit of the constraint temperature, the set temperature and return air temperature of each air conditioner when convergence is completed are obtained as the target set temperature and target return air temperature, and the target set temperature and target return air temperature of each air conditioner are used as air conditioning control information.

[0090] S205. If the air conditioner is in the on state, obtain the on state information according to the air conditioner control information, and control the air conditioner to turn off according to the on state information.

[0091] Specifically, after obtaining the air conditioning control information including the target set temperature and the target return air temperature, the system obtains the number of times the target set temperature in the air conditioning control information exceeds the preset maximum safe temperature within the preset operating time after the air conditioning is turned on, which is taken as the first number of times the target exceeds the target. The system also obtains the compressor working time and the current return air temperature, and takes the first number of times the target exceeds the target, the compressor working time, and the current return air temperature as the on-state information.

[0092] Furthermore, using a preset running time of 30 minutes as a step, if the first number of times exceeding the target exceeds the preset number (e.g., 18 times), the compressor's operating time is less than the preset operating time, and the current return air temperature is less than the preset maximum safe temperature, the air conditioner is shut down; otherwise, the air conditioner remains in its current state. This shutdown logic is expressed by the following formula:

[0093]

[0094] in, Set the minimum safe temperature. Set the minimum safe temperature: For compressor operating time, This is the current return air temperature.

[0095] S206. If the air conditioner is in the off state, obtain the off state information according to the air conditioner control information, and control the air conditioner to turn on according to the off state information;

[0096] Specifically, after obtaining the air conditioning control information including the target set temperature and the target return air temperature, the number of times the target set temperature in the air conditioning control information is lower than the preset minimum safe temperature within the preset shutdown time after the air conditioning is turned off is obtained as the second target exceedance number. The current return air temperature is obtained and the second target exceedance number and the current return air temperature are used as shutdown status information.

[0097] Furthermore, taking the first 30 minutes of the preset running time as a step, if the number of times the second target is exceeded is greater than the preset number of 18 times, or if the return air temperature within the current three minutes is greater than the preset maximum safe temperature, the air conditioner is turned on; otherwise, the current state of the air conditioner is maintained. This is expressed by the following start-up logic formula.

[0098]

[0099] in, Set the minimum safe temperature. Set the minimum safe temperature: For compressor operating time, Based on the current return air temperature, by setting dynamic start-stop conditions for the air conditioners, each air conditioner can minimize energy consumption while maintaining the available cooling capacity for the load, effectively controlling the operating time of the air conditioning system.

[0100] S207. After control is implemented, obtain the real-time return air temperature of each air conditioner and the real-time operating temperature of each sensor. If it is detected that the real-time return air temperature is greater than the preset abnormal return air temperature, or the real-time operating temperature is greater than the maximum temperature of the sensor, generate an alarm message to remind the operation and maintenance personnel.

[0101] Specifically, air conditioners or sensors whose real-time return air temperature exceeds the preset abnormal return air temperature, or whose real-time operating temperature exceeds the sensor's maximum temperature, are marked as abnormal devices. The system obtains the identity information of the abnormal devices and generates alarm information based on the identity information to remind maintenance personnel. At the same time, it automatically switches to backup equipment or switches to a mode that reduces energy consumption to minimize the impact of the fault on the system operation. This enables return air temperature monitoring and high temperature monitoring of sensors, thereby improving the reliability of the air conditioning cooling system in the computer room environment and promptly detecting and resolving faults that may lead to energy waste or equipment damage.

[0102] Furthermore, it can also acquire equipment status information of air conditioners and sensors, and realize the detection of occasional air conditioner failures and monitoring of communication failures based on the equipment status information. At the same time, it can continuously record and analyze fault events, trace and analyze faults based on the recorded fault data, and provide references and improvement suggestions for optimizing the operation of air conditioning systems.

[0103] The data center energy-saving control method provided in this application embodiment trains a self-attention mechanism-based neural network based on historical temperature data to obtain a temperature prediction model. The current data center operating time is input into the temperature prediction model to obtain the target predicted temperature. Based on the target predicted temperature and sensor temperature information, a genetic algorithm is used to optimize and obtain air conditioning control information. By implementing the target set temperature and target return air temperature of each air conditioner as indicated by the air conditioning control information, the data center's air conditioning temperature is adjusted to achieve energy saving. The active optimization mechanism avoids the limitations of large local optimization errors, ensuring the system achieves optimal performance under different scenarios, shortening training time, and responding more quickly to real-time environmental changes, thereby reducing deployment costs. Furthermore, through security monitoring and maintenance strategies, the security of the data center is improved.

[0104] Figure 3 A schematic diagram of the structure of the energy-saving control device for the computer room provided in this application is shown below.Figure 4 As shown, the data center energy-saving control device 30 provided in this embodiment includes:

[0105] The acquisition module 301 is used to acquire historical temperature data of the computer room, wherein the historical temperature data is temperature data of a preset continuous time period in the computer room environment, and the historical temperature data is used as training data to train a neural network based on a self-attention mechanism to obtain a temperature prediction model.

[0106] Processing module 302 is used to obtain the sensor temperature information of each sensor in the current computer room during the current computer room operation time, and input the current computer room operation time into the temperature prediction model to obtain the target predicted temperature;

[0107] The control module 303 is used to calculate air conditioning control information based on the target predicted temperature and the sensor temperature information using a genetic algorithm, and to control each air conditioner in the computer room environment according to the air conditioning control information.

[0108] Furthermore, the acquisition module 301 is specifically used to acquire the energy consumption parameter information of the computer room;

[0109] Based on the energy consumption parameter information, an objective function is established according to the target predicted temperature and the sensor temperature information;

[0110] The objective function is used to control the convergence of the computer room ambient temperature towards the target predicted temperature.

[0111] The target set temperature and target return air temperature of each air conditioner are obtained when convergence is completed, and the target set temperature and target return air temperature of each air conditioner are determined as the air conditioning control information.

[0112] Furthermore, the control module 303, the air conditioning control information includes the target set temperature and the target return air temperature, specifically used to obtain the number of times the target set temperature in the air conditioning control information exceeds the preset maximum safe temperature when the air conditioner is in the on state, as the first number of times the target exceeds the target;

[0113] Obtain the compressor operating time and current return air temperature, and use the first number of times exceeding the target, the compressor operating time, and the current return air temperature as the on / off status information;

[0114] The air conditioner is controlled to turn off based on the on / off status information;

[0115] If the air conditioner is in the off state, the number of times the target set temperature in the air conditioner control information is less than the preset minimum safe temperature is obtained as the second number of times the target is exceeded;

[0116] Obtain the current return air temperature, and use the second number of times the target is exceeded and the current return air temperature as the shutdown status information;

[0117] The air conditioner is turned on based on the off status information.

[0118] Furthermore, the control module 303 is specifically used to shut down the air conditioner if the first number of times exceeding the target is greater than a preset number, the compressor working time is less than a preset working time, and the current return air temperature is less than a preset maximum safe temperature;

[0119] Otherwise, maintain the current state of the air conditioner.

[0120] Furthermore, the acquisition module 301 is specifically used to turn on the air conditioner if the number of times the second target is exceeded is greater than the preset number, or if the current return air temperature is greater than the preset maximum safe temperature;

[0121] Otherwise, maintain the current state of the air conditioner.

[0122] Furthermore, the control module 303 is also used to acquire the real-time return air temperature of each air conditioner after control is implemented, as well as the real-time operating temperature of each sensor.

[0123] If the real-time return air temperature is detected to be greater than the preset abnormal return air temperature, or the real-time operating temperature is detected to be greater than the sensor's maximum temperature, the corresponding air conditioner or sensor will be marked as an abnormal device, and the identity information of the abnormal device will be obtained. An alarm message will be generated based on the identity information to remind the operation and maintenance personnel.

[0124] Furthermore, the acquisition module 301 is also used to generate multiple sets of test control parameters through a random walk algorithm, wherein the test control parameters are the target test states of each air conditioner;

[0125] Control each air conditioner in the computer room environment according to the test control parameters;

[0126] After a preset test time, the action time for achieving the target test state is obtained;

[0127] If the action time is less than the preset test time, the corresponding test control parameter is marked as a qualified test parameter, and the number of qualified test parameters is recorded to obtain the number of qualified tests.

[0128] If the number of qualified tests exceeds the qualified threshold, then the step of "acquiring historical temperature data" is executed.

[0129] If the number of qualified tests is less than or equal to the qualified threshold, maintenance information is generated to remind maintenance personnel.

[0130] The data center energy-saving control device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0131] Figure 4 A schematic diagram of the structure of the electronic device provided in this application. ​ As shown, the electronic device 40 provided in this embodiment includes at least one processor 401 and a memory 402. Optionally, the device 40 further includes a communication component 403. The processor 401, memory 402, and communication component 403 are connected via a bus 404.

[0132] In a specific implementation, at least one processor 401 executes computer execution instructions stored in memory 402, causing at least one processor 401 to perform the above-described method.

[0133] The specific implementation process of processor 401 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0134] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0135] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0136] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0137] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0138] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0139] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0140] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0141] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0142] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0143] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0144] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0146] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for energy saving control of a machine room, characterized by, The method comprises the following steps: acquiring historical temperature data of a machine room, wherein the historical temperature data is temperature data of a preset continuous time period in the machine room environment; training a neural network based on a self-attention mechanism using the historical temperature data as training data to obtain a temperature prediction model; acquiring sensor temperature information of each sensor in the machine room at a current machine room running time; inputting the current machine room running time into the temperature prediction model to obtain a target predicted temperature; obtaining air conditioner control information through a genetic algorithm based on the target predicted temperature and the sensor temperature information; controlling each air conditioner in the machine room environment according to the air conditioner control information.

2. The method of claim 1, wherein, The method of obtaining air conditioner control information based on the target predicted temperature and the sensor temperature information through a genetic algorithm comprises the following steps: acquiring energy consumption parameter information of the machine room; establishing a target function based on the target predicted temperature and the sensor temperature information based on the energy consumption parameter information; controlling the temperature of the machine room environment to converge to the target predicted temperature through the target function; obtaining the target set temperature and the target return air temperature of each air conditioner when the convergence is completed, and determining the target set temperature and the target return air temperature of each air conditioner as the air conditioner control information.

3. The method of claim 1, wherein, The air conditioner control information comprises a target set temperature and a target return air temperature. Correspondingly, the method of controlling each air conditioner in the machine room environment according to the air conditioner control information comprises the following steps: if the air conditioner is in an on state, obtaining the number of times that the target set temperature in the air conditioner control information exceeds a preset maximum safe temperature as a first over-target number; obtaining the compressor working time and the current return air temperature, and taking the first over-target number, the compressor working time and the current return air temperature as on state information; controlling the air conditioner to be turned off according to the on state information; if the air conditioner is in an off state, obtaining the number of times that the target set temperature in the air conditioner control information is less than a preset minimum safe temperature as a second over-target number; obtaining the current return air temperature, and taking the second over-target number and the current return air temperature as off state information; controlling the air conditioner to be turned on according to the off state information.

4. The method of claim 3, wherein, The method of controlling the air conditioner to be turned off according to the on state information comprises the following steps: if the first over-target number is greater than a preset number, the compressor working time is less than a preset working time, and the current return air temperature is less than a preset maximum safe temperature, the air conditioner is turned off; otherwise, the current state of the air conditioner is maintained.

5. The method of claim 3, wherein, The method of controlling the air conditioner to be turned on according to the off state information comprises the following steps: if the second over-target number is greater than the preset number, or the current return air temperature is greater than a preset maximum safe temperature, the air conditioner is turned on; otherwise, the current state of the air conditioner is maintained.

6. The method of claim 1, wherein, After the air conditioners in the machine room environment are controlled according to the air conditioner control information, the method further comprises the following steps: acquiring real-time return air temperatures of each air conditioner and real-time working temperatures of each sensor after the control is implemented; if it is detected that there is a real-time return air temperature greater than a preset return air abnormal temperature, or a real-time working temperature greater than a sensor maximum temperature, the corresponding air conditioner or sensor is marked as an abnormal device. Obtain device information of the abnormal device, generate alarm information according to the device information to remind the operation and maintenance personnel.

7. The method according to any one of claims 1 to 7, characterized in that, Before the obtaining the historical temperature data, further comprising: Generate a plurality of test control parameters through a random walk algorithm, the test control parameters being target test states of each air conditioner; Control each air conditioner in the computer room environment according to the test control parameters; After a preset test time, obtain an action time for achieving the target test state; If the action time is less than the preset test time, mark the corresponding test control parameter as a qualified test parameter, and record the number of qualified test parameters to obtain a test qualified number; If the test qualified number is greater than a qualified threshold, execute the step of obtaining the historical temperature data; If the test qualified number is less than or equal to the qualified threshold, generate maintenance information to remind the operation and maintenance personnel.

8. A machine room energy saving control device, characterized by, Comprise: An acquisition module is used for acquiring historical temperature data of a computer room, wherein the historical temperature data is temperature data of a preset continuous time period in the computer room environment, the historical temperature data is used as training data to train a neural network based on a self-attention mechanism to obtain a temperature prediction model; A processing module is used for acquiring sensor temperature information of each sensor in the computer room at a current computer room running time, and inputting the current computer room running time into the temperature prediction model to obtain a target predicted temperature; A control module is used for calculating air conditioner control information based on the target predicted temperature and the sensor temperature information through a genetic algorithm, and controlling each air conditioner in the computer room environment according to the air conditioner control information.

9. An electronic device, comprising: Comprise: A memory and a processor; The memory stores computer execution instructions; The processor executes the computer execution instructions stored in the memory, so that the processor executes the method in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the method in any one of claims 1-7.

Images