CN116147154A

CN116147154A - Method and device for adjusting temperature of air conditioner in machine room and electronic equipment

Info

Publication number: CN116147154A
Application number: CN202211353692.9A
Authority: CN
Inventors: 张晓峰; 张敏; 金志伟; 叶建安; 汤雯博
Original assignee: China Telecom Corp Ltd
Current assignee: China Telecom Corp Ltd
Priority date: 2022-11-01
Filing date: 2022-11-01
Publication date: 2023-05-23

Abstract

The application discloses a method and a device for adjusting the temperature of an air conditioner in a machine room, and belongs to the technical field of automatic control. The method comprises the following steps: and responding to temperature regulation of an air conditioner in the target machine room, and acquiring real-time data of target temperature of the target machine room and a preset influence factor, wherein the preset influence factor comprises: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor; solving the target control temperature of the air conditioner through a multi-dimensional temperature influence model of the machine room according to the real-time data and the target temperature; the air conditioner is temperature-regulated based on the target control temperature. The method realizes the automatic control of the temperature of the air conditioner of the machine room by combining the multidimensional factors influencing the temperature of the machine room, improves the instantaneity and the temperature regulation accuracy of the temperature regulation of the air conditioner of the machine room, and is beneficial to reducing the energy consumption of the air conditioner of the machine room.

Description

Method and device for adjusting temperature of air conditioner in machine room and electronic equipment

Technical Field

The present disclosure relates to the field of automatic control technologies, and in particular, to a method and an apparatus for adjusting a temperature of an air conditioner in a machine room, an electronic device, and a computer readable storage medium.

Background

The temperature of the air conditioner in the traditional machine room is manually adjusted. For example, the set value of the air conditioner temperature of the machine room is periodically adjusted manually according to experience, or the air conditioner temperature of the machine room is temporarily adjusted according to real-time temperature in the machine room collected by a sensor arranged in the machine room. When the temperature of the air conditioner in the machine room is regulated according to manual experience, the temperature of the air conditioner is regulated to a fixed value according to the periodic variation of seasons and outdoor air temperature. For example, the temperature of the air conditioner in the machine room in summer is set to 24 ℃, the temperature of the air conditioner in the machine room in winter is set to 26 ℃, the temperature set value is single, and the defect of overlarge energy consumption of the air conditioner in the machine room exists. In the prior art, according to the scheme of manually adjusting the temperature of the air conditioner in the machine room according to the indoor temperature acquired by the sensor, a large number of sensors are required to be arranged in the machine room, the hardware cost is high, in addition, timely adjustment cannot be achieved, and the defect of overlarge energy consumption exists.

It can be seen that the temperature adjusting method of the air conditioner in the machine room in the prior art needs to be improved.

Disclosure of Invention

The embodiment of the application provides a temperature regulation method and device for an air conditioner of a machine room, which can improve the instantaneity and the temperature regulation accuracy of the temperature regulation of the air conditioner of the machine room and is beneficial to reducing the energy consumption of the air conditioner of the machine room.

In a first aspect, an embodiment of the present application discloses a method for adjusting a temperature of an air conditioner in a machine room, including:

and responding to temperature regulation of an air conditioner in a target machine room, and acquiring real-time data of target temperature of the target machine room and a preset influence factor, wherein the preset influence factor comprises: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor;

taking the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room;

and adjusting the temperature of the air conditioner based on the target control temperature.

Optionally, after the real-time data and the target temperature are used as input of a pre-trained multi-dimensional temperature influence model of the machine room, the method further includes:

According to the real-time data and the target control temperature, estimating the temperature of the target machine room after the temperature adjustment is carried out by a pre-trained machine room temperature prediction model, and taking the estimated temperature as a predicted temperature;

after the air conditioner is temperature-regulated based on the target control temperature, the method further comprises:

obtaining the measured temperature of the target machine room;

and under the condition that the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold, tuning and optimizing the multi-dimensional temperature influence model of the machine room.

Optionally, the target machine room is divided into a plurality of three-dimensional subspaces, and the multi-dimensional temperature influence model of the machine room comprises: a sub-model corresponding to each of the preset influence factors, a feature generation sub-model, and a feature encoding network,

the step of solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room by taking the real-time data and the target temperature as inputs of the multi-dimensional temperature influence model of the machine room trained in advance comprises the following steps:

gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces through the submodels, so that single-factor space temperature influence values of the three-dimensional subspaces are obtained;

Fusing the single-factor space temperature influence values through the feature generation sub-model to obtain accumulated space temperature influence values of the preset influence factors on each three-dimensional subspace; the method comprises the steps of,

performing space dimension reduction on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix;

and performing coding mapping on the air conditioner temperature gain matrix through the characteristic coding network to obtain the real-time data and the target control temperature of the air conditioner corresponding to the target temperature.

Optionally, the submodel includes: expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of a machine room, wherein the air conditioner operation parameter factors comprise one or more of the following data: air conditioning temperature, fan revolution, wind direction angle, air conditioning operation time, air conditioning position, and thermal conductivity of air,

gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces, and the method comprises the following steps:

taking the value of the air conditioner operation parameter factor and the position of each three-dimensional subspace as first input data, and executing preset operation on the first input data through the first sub-model to obtain a single factor space temperature influence value of the air conditioner operation parameter factor on each three-dimensional subspace.

Optionally, the submodel includes: expressing a second sub-model of the effect of the IT device loading factor on the machine room temperature, the IT device loading factor comprising one or more of the following data: CPU utilization rate and memory utilization rate of the server, and IO consumption rate, position of the server, temperature of the server,

and taking the value of the IT equipment load factor in the target machine room, the position of each three-dimensional subspace and the heat conduction coefficient of air as second input data, and executing preset operation on the second input data through the second sub-model to obtain a single factor space temperature influence value of the IT equipment load factor on each three-dimensional subspace in the target machine room.

Optionally, the submodel includes: expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room, wherein the environmental factors comprise one or more of the following data: the outdoor temperature and the air humidity are controlled,

And taking the value of the environmental factor as third input data, and executing preset operation on the third input data through the third sub-model to obtain a single factor space temperature influence value of the environmental factor on each three-dimensional subspace in the target machine room.

Optionally, the performing the space dimension reduction processing on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix includes:

determining a projection plane of the three-dimensional subspace, wherein the projection plane is: a plane parallel to the side surface of the three-dimensional subspace in the horizontal direction, or a plane parallel to the side surface of the three-dimensional subspace in the vertical direction;

projecting the three-dimensional subspace onto the projection plane;

taking the average value of the accumulated space temperature influence values corresponding to the three-dimensional subspaces projected to the same area on the projection plane as the accumulated space temperature influence value corresponding to the corresponding area of the projection plane;

and obtaining an air conditioner temperature gain matrix according to the accumulated space temperature influence value corresponding to each region in the projection plane.

In a second aspect, an embodiment of the present application discloses a room air conditioner temperature adjusting device, including:

The system comprises a machine room target temperature and influence data acquisition module, a control module and a control module, wherein the machine room target temperature and influence data acquisition module is used for responding to temperature adjustment of an air conditioner in a target machine room and acquiring real-time data of the target temperature of the target machine room and a preset influence factor, and the preset influence factor comprises: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor;

the air conditioner temperature solving module is used for taking the real-time data and the target temperature as input of a pre-trained machine room multi-dimensional temperature influence model, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the machine room multi-dimensional temperature influence model;

and the air conditioner temperature adjusting module is used for adjusting the temperature of the air conditioner based on the target control temperature.

Optionally, the apparatus further includes:

the machine room temperature prediction module is used for predicting the machine room temperature of the target machine room after the temperature adjustment is implemented through a pre-trained machine room temperature prediction model according to the real-time data and the target control temperature, and taking the machine room temperature as a predicted temperature;

the machine room measured temperature acquisition module is used for acquiring the measured temperature of the target machine room after the air conditioner is subjected to temperature adjustment based on the target control temperature;

And the model tuning processing module is used for performing tuning training on the multi-dimensional temperature influence model of the machine room under the condition that the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold.

the air conditioner temperature solving module is further used for:

Optionally, the submodel includes: expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of a machine room, wherein the air conditioner operation parameter factors comprise one or more of the following data: the method comprises the steps of performing gain calculation on values of preset influence factors of three-dimensional subspaces through the submodel respectively to obtain single-factor space temperature influence values of the three-dimensional subspaces, wherein the single-factor space temperature influence values comprise:

Optionally, the submodel includes: expressing a second sub-model of the effect of the IT device loading factor on the machine room temperature, the IT device loading factor comprising one or more of the following data: the CPU utilization rate and the memory utilization rate of the server, the IO consumption rate, the position of the server and the temperature of the server, wherein gain calculation is respectively carried out on the values of preset influence factors of all three-dimensional subspaces through the submodel to obtain single-factor space temperature influence values of all three-dimensional subspaces, and the method comprises the following steps:

Optionally, the submodel includes: expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room, wherein the environmental factors comprise one or more of the following data: the gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces, and the method comprises the following steps:

projecting the three-dimensional subspace onto the projection plane;

In a third aspect, the embodiment of the application further discloses an electronic device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the computer program is executed by the processor to implement the method for adjusting the temperature of the air conditioner in the machine room.

In a fourth aspect, embodiments of the present application disclose a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the room air conditioner temperature adjustment method disclosed in embodiments of the present application.

According to the machine room air conditioner temperature adjusting method disclosed by the embodiment of the application, real-time data of the target temperature of the target machine room and the preset influence factor are obtained by responding to temperature adjustment of an air conditioner in the target machine room, wherein the preset influence factor comprises: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor; taking the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room; based on the target control temperature to the air conditioner carries out temperature regulation, realizes the temperature of the multi-dimensional factor automatic control computer lab air conditioner that combines the influence computer lab temperature, promotes real-time and the temperature regulation accuracy that can computer lab air conditioner temperature regulation, helps reducing the energy consumption of computer lab air conditioner, reaches the dynamic balance of computer lab temperature and air conditioner energy consumption.

The foregoing description is only an overview of the technical solutions of the present application, and may be implemented according to the content of the specification in order to make the technical means of the present application more clearly understood, and in order to make the above-mentioned and other objects, features and advantages of the present application more clearly understood, the following detailed description of the present application will be given.

Drawings

For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

Fig. 1 is one of flowcharts of a method for adjusting a temperature of an air conditioner in a machine room according to an embodiment of the present application;

FIG. 2 is a second flowchart of a method for adjusting the temperature of an air conditioner in a machine room according to an embodiment of the present disclosure;

FIG. 3 is a third flowchart of a method for adjusting the temperature of an air conditioner in a machine room according to an embodiment of the present disclosure;

fig. 4 is a schematic diagram illustrating three-dimensional space division of a machine room in the method for adjusting temperature of an air conditioner in the machine room according to the embodiment of the present application;

fig. 5 is a schematic diagram of a space dimension reduction effect in a temperature adjustment method of an air conditioner in a machine room according to an embodiment of the present application;

fig. 6 is a schematic diagram of a temperature interpolation principle in a temperature adjustment method of an air conditioner in a machine room according to an embodiment of the present application;

Fig. 7 is a schematic structural diagram of a temperature adjusting device of an air conditioner in a machine room according to an embodiment of the present application;

fig. 8 is a second schematic structural diagram of a temperature adjusting device of an air conditioner in a machine room according to an embodiment of the present disclosure;

FIG. 9 schematically illustrates a block diagram of an electronic device for performing a method according to the present application; and

fig. 10 schematically shows a memory unit for holding or carrying program code implementing the method according to the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

Next, a specific implementation manner of the method for adjusting the temperature of the air conditioner in the machine room disclosed in the embodiment of the present application is illustrated.

As shown in fig. 1, a method for adjusting temperature of an air conditioner in a machine room disclosed in an embodiment of the present application includes: step 120, step 130 and step 140.

Step 120, in response to temperature adjustment of an air conditioner in a target machine room, acquiring real-time data of a target temperature and a preset influence factor of the target machine room, wherein the preset influence factor comprises: an air conditioner operating parameter factor, an IT equipment load factor, and an environmental factor.

According to the machine room air conditioner temperature adjusting method, temperature adjustment is independently performed on each machine room. The target machine room is a machine room for air-conditioning temperature monitoring and air-conditioning temperature adjustment.

And a temperature acquisition device (such as a temperature sensor) is arranged at a part of the target machine room and is used for acquiring the real-time temperature of the target machine room. Typically, a plurality of temperature acquisition devices are disposed in or on the target machine room for acquiring real-time temperatures at different locations of the target machine room.

The arrangement position of air conditioners in a machine room, the number of the air conditioners, the running temperature of the air conditioners, the number of revolutions of a fan, the air outlet direction, the arrangement position, the number and the running state of IT equipment, the indoor and outdoor temperature of the machine room and other factors can influence the temperature of a certain position in the machine room. In the embodiment of the application, the multidimensional data influencing the temperature regulation effect of the air conditioner on the machine room are determined as influencing factors for determining the target temperature of the air conditioner, for example, the operation parameters of the air conditioner in the machine room, the environmental conditions of the machine room, the information of IT equipment running in the machine room and the like, and the preset influencing factors are defined according to the multidimensional data.

The air conditioner operation parameter factors included in the preset influence factors include: relevant parameters of an air conditioner in the machine room; the IT equipment load factors included in the preset influence factors include: relevant parameters of the server in the machine room; the environmental factors included in the preset influence factors include: outdoor temperature of the machine room, indoor and/or outdoor air humidity.

In practical applications, the temperature adjustment of the air conditioner in the target machine room may be performed periodically, for example, the machine room air conditioner temperature adjustment method is performed every 6 hours.

In other embodiments of the present application, the temperature adjustment of the air conditioner in the target machine room may also be triggered according to the temperature monitoring result of the target machine room. For example, when the temperature of the machine room collected by the temperature collection device in the machine room exceeds the preset temperature range, the air conditioner in the target machine room is subjected to temperature regulation.

For example, the air conditioner temperature adjustment condition may be set in advance. Wherein, the air conditioner temperature adjustment condition includes: the real-time temperature is smaller than a preset temperature threshold, or the absolute value of the difference between the real-time temperature and the preset temperature threshold is smaller than a preset temperature difference threshold. And the preset air conditioner temperature regulation conditions are determined according to the machine room maintenance requirements.

After the real-time temperature of the target machine room is obtained, the real-time temperature is matched with a preset air conditioner temperature regulation condition, and if the real-time condition meets the preset air conditioner temperature regulation condition, the running temperature of each air conditioner in the target machine room is regulated so as to achieve the aim of saving energy.

When the energy-saving strategy is implemented and the running temperature of the air conditioner is regulated, the temperature in the target machine room needs to be regulated to a certain degree, namely the target temperature of the target machine room is obtained. In the embodiment of the application, the target temperature of the target machine room can be determined according to the energy-saving requirement.

When the energy-saving strategy is implemented and the running temperature of the air conditioner is regulated, the real-time data of the preset influence factors are also required to be acquired. The value of the air conditioner operation parameter factor can be obtained through an air conditioner setting system; the IT device load factor can be obtained through an IT device operation management system of the target machine room; the environmental factors can be acquired by combining temperature and humidity acquisition equipment such as a sensor arranged in a target machine room; the measured temperature of the first machine room may be obtained by a temperature acquisition device deployed in the target machine room.

And 130, taking the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room.

When the energy-saving strategy is implemented to adjust the operation temperature of the air conditioner, the operation temperature of each air conditioner needs to be adjusted to a certain degree in advance, namely, the target control temperature is determined.

In the embodiment of the application, the air conditioner temperature meeting the multi-dimensional temperature influence model of the machine room can be obtained by executing the multi-dimensional temperature influence model of the machine room.

As shown in fig. 2, in some embodiments of the present application, before responding to temperature adjustment of an air conditioner in a target machine room, acquiring real-time data of a target temperature and a preset influence factor of the target machine room, the method further includes: step 110.

And step 110, training a multi-dimensional temperature influence model of the machine room according to the historical temperature of the target machine room and the value of a preset influence factor corresponding to the historical temperature.

In the embodiment of the application, firstly, according to the relation of the influence of the preset influence factor on the temperature of the machine room, a multi-dimensional temperature influence model of the machine room expressing the mapping relation between the value of the preset influence factor and the temperature of the machine room is established. The multi-dimensional temperature influence model of the machine room comprises: and each sub-model is used for calculating the gain of the value of the corresponding preset influence factor to the temperature of the machine room, and the gains of all the preset influence factors to the temperature of the machine room form an air conditioner temperature gain matrix. The air conditioner temperature gain matrix expresses the influence of the value of each preset influence factor on the temperature of the machine room. According to the analysis, training of the air conditioner temperature gain matrix can be achieved by training the multi-dimensional temperature influence model of the machine room based on the preset influence factors of the target machine room and the historical data of the machine room temperature.

Taking the machine room multidimensional temperature influence model as an example, the following expression is shown:

Δtp×t0×f (F (a), F (b), F (c))=troom; wherein a, b, c represent each preset influence factor, f (a), f (b), f (c) represent the sub-model corresponding to each preset influence factor, delta TP represents an air conditioner temperature gain matrix, the value of delta TP is calculated according to the output of each sub-model, troom represents the temperature of a machine room, T0 is a conversion matrix, and the value is constant. The above formula can show that the mapping relationship between a certain preset influence factor and the machine room temperature can be learned by bringing the values of a plurality of groups of preset influence factors and the machine room temperature into the above formula.

In the embodiment of the application, the matrix element value of the air conditioner temperature gain matrix Δtp is obtained by calculating the model parameter of the machine room multidimensional temperature influence model and the values of the preset influence factors according to a preset method, and based on the analysis, the process of training the machine room multidimensional temperature influence model is the process of learning the air conditioner temperature gain matrix Δtp.

Correspondingly, when the energy-saving strategy is implemented, the temperature Troom of the machine room is a known quantity, namely the target temperature, the value of the IT device load factor b is from the IT device load factor in the real-time data, the value of the environment factor c is from the environment factor in the real-time data, the value of the air conditioner operation parameter factor is from the air conditioner operation parameter factor in the real-time data, other values are maintained unchanged except the air conditioner temperature, namely the air conditioner temperature is needed to be solved in the air conditioner operation parameter factor a.

Based on the principle analysis, under the condition that the air conditioner temperature gain matrix is a determined value, all the real-time data except the air conditioner temperature are determined values, and the target temperature is determined values, the real-time data and the target temperature are used as input of a pre-trained computer room multi-dimensional temperature influence model, and the air conditioner temperature can be obtained by solving according to the operation relation expressed by the computer room multi-dimensional temperature influence model and is used as the target control temperature.

And 140, adjusting the temperature of the air conditioner based on the target control temperature.

After solving the target control temperature of the air conditioner, the temperature of the air conditioner can be adjusted according to the target control temperature so as to adjust the temperature of the target machine room, thereby achieving the effect of energy conservation.

In some embodiments of the present application, as shown in fig. 3, after the real-time data and the target temperature are used as input of a pre-trained multi-dimensional temperature influence model of the machine room, the method further includes: step 135.

And step 135, estimating the temperature of the target machine room after the temperature adjustment is implemented by a pre-trained machine room temperature prediction model according to the real-time data and the target control temperature, and taking the estimated temperature as a predicted temperature.

After the air conditioner is temperature-adjusted based on the target control temperature, as the air conditioner temperature is reset, the air conditioner operates at the reset temperature, and the room temperature may be changed. In theory, if the target control temperature of the air conditioner obtained by solving the steps is accurate, the temperature of the target machine room gradually approaches to the target temperature until the target temperature is reached.

In order to verify whether the target control temperature of the air conditioner obtained by solving is accurate, namely, whether the energy-saving strategy is effective, in the embodiment of the application, the temperature to be reached by the target machine room after the air conditioner temperature of the target machine room is predicted and regulated to the target control temperature is predicted and used as the machine room predicted temperature by training the machine room temperature prediction model for predicting the machine room temperature according to the air conditioner temperature in advance. And then, comparing the predicted temperature of the machine room with the temperature of the air conditioner to the target control temperature, and evaluating whether the energy-saving strategy is effective or not.

The machine room temperature prediction model is a neural network model trained based on historical data of a target machine room. The historical data for training the machine room temperature prediction model includes: and the actual temperature of the machine room, and the value of the preset influence factor when the target machine room is at the actual temperature of the machine room.

The process of training the machine room temperature prediction model is a process of establishing a mapping relationship between the value of the preset influence factor of the target machine room and the machine room temperature. Therefore, for the machine room temperature prediction model obtained through training, the corresponding machine room temperature can be predicted under the condition of giving the value of the preset influence factor.

The specific implementation manner of training the machine room temperature prediction model may refer to the prior art, and will not be described in detail in this embodiment of the present application.

Correspondingly, after the temperature adjustment of the air conditioner based on the target control temperature, the method further comprises: step 150 and step 160.

And step 150, obtaining the measured temperature of the target machine room.

After the energy-saving strategy is implemented on the target machine room, the air conditioner is waited to run for a period of time at the target control temperature, and then the real-time temperature of the target machine room can be acquired, so that the actually measured temperature is obtained. In some embodiments of the present application, the real-time temperature of the target machine room may be acquired by a temperature acquisition device deployed in the target machine room, and the measured temperature may be obtained according to the real-time temperature.

And 160, performing tuning training on the multi-dimensional temperature influence model of the machine room under the condition that the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold.

In the embodiment of the application, the energy-saving strategy can be secondarily optimized through the machine room temperature adaptation model. Alternatively, the machine room temperature adaptation model may be expressed as: beta(s) -P _Prediction α, wherein β(s) is the measured temperature of the target machine room obtained after implementation of the energy saving strategy, P _Prediction For the predicted temperature, α is a correction parameter. Alpha is a preset tuning threshold.

And the tuning threshold value alpha is used for modifying the influence factor, the optimal value of alpha is infinitely close to 0, and if the tuning threshold value is met, the condition that the air conditioner control temperature accuracy predicted by the machine room multidimensional temperature influence model based on the preset influence factor is highest is indicated.

In some embodiments of the present application, the correction parameters (i.e. tuning thresholds) may be preset according to the energy saving requirement of the machine room. For example, the tuning threshold may be set to 1. When the absolute value of the difference between the predicted temperature and the measured temperature of the target machine room is greater than or equal to a preset tuning threshold (for example, greater than or equal to 1), the solved target control temperature is considered to be inaccurate when the energy-saving strategy is executed, and optimization training is required to be performed on the air conditioner temperature gain matrix according to which the target control temperature is solved. And the training samples are updated to perform optimization training on the multi-dimensional temperature influence model of the machine room, so that the aim of optimizing an air conditioner temperature gain matrix is fulfilled. For example, the current machine room temperature is updated to the training sample for training the machine room multidimensional temperature influence model according to the real-time data of the currently acquired preset influence factors, and the machine room multidimensional temperature influence model is iteratively trained.

When the absolute value of the difference between the predicted temperature and the measured temperature of the target machine room is smaller than the preset tuning threshold, the executed energy saving strategy can be considered to be effective.

According to the machine room air conditioner temperature adjusting method disclosed by the embodiment of the application, real-time data of the target temperature of the target machine room and the preset influence factor are obtained by responding to temperature adjustment of an air conditioner in the target machine room, wherein the preset influence factor comprises: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor; taking the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room; based on the target control temperature to the air conditioner carries out temperature regulation, realizes the temperature of the multi-dimensional factor automatic control computer lab air conditioner that combines the influence computer lab temperature, promotes real-time and the temperature regulation degree of accuracy that can computer lab air conditioner temperature regulation, helps reducing the energy consumption of computer lab air conditioner, reaches the dynamic balance of computer lab temperature and air conditioner energy consumption.

Further, according to the machine room air conditioner temperature adjusting method disclosed by the embodiment of the application, real-time data of the target temperature of the target machine room and the preset influence factor are obtained by responding to temperature adjustment of an air conditioner in the target machine room, wherein the preset influence factor comprises: the method comprises the steps of pre-estimating the temperature of a machine room after the temperature adjustment of a target machine room through a pre-trained machine room temperature prediction model according to real-time data and the target control temperature after the air conditioner is subjected to temperature adjustment based on the target control temperature, wherein the pre-trained machine room temperature prediction model is used as a predicted temperature, acquiring the measured temperature of the target machine room after the temperature adjustment is performed after the air conditioner is subjected to temperature adjustment based on the target control temperature, and obtaining the predicted temperature and the measured temperature; and under the condition that the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold, tuning and optimizing the multi-dimensional temperature influence model of the machine room, thereby further improving the accuracy of the temperature adjustment of the air conditioner of the machine room and improving the energy-saving effect.

In order to make the temperature adjustment of the room air conditioner disclosed in the embodiments of the present application easier to understand, the following further illustrates a specific implementation manner of each step of the temperature adjustment method of the room air conditioner.

As mentioned above, the preset influence factor includes: an air conditioner operating parameter factor, an IT equipment load factor, and an environmental factor. Correspondingly, the multi-dimensional temperature influence model of the machine room comprises: and the submodels, the feature generation submodels and the feature coding network correspond to the preset influence factors. Wherein the sub-model corresponding to each preset influence factor comprises: a first sub-model corresponding to an air conditioner operating parameter factor, a second sub-model corresponding to an IT equipment load factor, and a third sub-model corresponding to an environmental factor.

In the embodiment of the application, corresponding sub-models are constructed according to the influence of various types of influence factors on the temperature of the machine room, so that corresponding operation is performed on the values of the corresponding types of influence factors through each sub-model, and the influence value of the values of each preset influence factor on the temperature of the machine room is obtained.

In some embodiments of the present application, the target machine room is divided into a plurality of three-dimensional subspaces.

In order to improve accuracy of temperature control, the whole three-dimensional space of the target machine room is divided into a plurality of local three-dimensional subspaces, as shown in fig. 4, a machine room temperature measured by a sensor in the target machine room, an air conditioner operation parameter (such as a supply and return air temperature, an air conditioner position, a fan rotating speed, an air outlet direction and the like), a machine room IT load (such as server power consumption, temperature, memory, deployment position and the like), an environment (such as outdoor temperature, air humidity and the like) and other multi-parameter indexes are obtained by using an existing data acquisition system (such as temperature acquisition equipment and an equipment management system) of the machine room through a grid method, and then, the influence value of the corresponding type influence factor on the temperature of each subspace in the machine room is calculated through each submodel.

Next, the execution principle of the multi-dimensional temperature influence model of the machine room, the execution principle of each sub-model and the training method of the air conditioner temperature gain matrix are explained by combining the training process of the multi-dimensional temperature influence model of the machine room.

In the foregoing step 110, training a multidimensional temperature influence model of the machine room according to the historical temperature of the target machine room and the value of the preset influence factor corresponding to the historical temperature includes: for each appointed historical moment, the following substep 1101, substep 1102 and substep 1103 are respectively executed, and an air conditioner temperature gain matrix corresponding to the appointed historical moment is obtained; thereafter, sub-step 1104 is performed.

Wherein the specified historical time may be a historical time of every 5 minutes in the past month.

When the multi-dimensional temperature influence model of the machine room is trained, the value of the preset influence factor of the target machine room at each historical moment and the temperature of the machine room are respectively taken as one set of historical data to obtain a plurality of sets of historical data, and then an air conditioner temperature gain matrix is respectively calculated for each set of historical data, so that an air conditioner temperature gain matrix can be obtained corresponding to each historical moment.

Specific embodiments of each sub-step are set forth below.

And a sub-step 1101 of performing gain calculation on the values of the preset influence factors of the three-dimensional subspaces at the appointed historical time through the sub-model to obtain single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical time.

In the process of training the multi-dimensional temperature influence model of the machine room, the influence value of each type of preset influence factor on each three-dimensional subspace in the target machine room is obtained after the corresponding value of the preset influence factor is calculated through each submodel. In the embodiment of the application, an influence value of a type of preset influence factor on the temperature of a certain three-dimensional subspace in a machine room is recorded as a single-factor space temperature influence value.

The following describes a specific implementation manner of each sub-model to obtain the influence value of each type of preset influence factor on each three-dimensional subspace in the target machine room after the calculation of the value of the corresponding preset influence factor.

1. First sub-model

The sub-model comprises: and expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of the machine room.

And calculating the historical data of the air conditioner operation parameter factors through the first sub-model, so that a single factor space temperature influence value of the air conditioner operation parameter factors on each three-dimensional subspace in the target machine room can be obtained.

In some embodiments of the present application, the air conditioner operating parameter factor includes one or more of the following data: the method comprises the steps of performing gain calculation on values of preset influence factors of each three-dimensional subspace at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of each three-dimensional subspace corresponding to the appointed historical moment, wherein the single-factor space temperature influence values comprise the following components: and taking the value of the air conditioner operation parameter factor at the appointed historical moment and the position of each three-dimensional subspace as first input data, and executing preset operation on the first input data through the first submodel to obtain a single factor space temperature influence value of the air conditioner operation parameter factor corresponding to the appointed historical moment on each three-dimensional subspace.

Based on the heat conduction model, an expression air conditioner temperature propagation consumption model can be constructed:

wherein p represents a three-dimensional subspace, i represents an air conditioner identifier in a target machine room, lambda is the heat conductivity coefficient of air, and T _i Air conditioning temperature v of the ith air conditioner _i Represents the number of revolutions, t, of a fan of an ith air conditioner _i Indicating the air conditioner operation time length of the ith air conditioner, O _i Represents the wind direction angle of the ith air conditioner, S _i,p Represents the distance between the ith air conditioner and the three-dimensional subspace P, O _ip The angle K of the three-dimensional subspace P relative to the ith air conditioner is represented ₁ As model parameters, Δt ₁ (p) represents a single-factor air space of the ith air conditioner to the three-dimensional subspace p in the target machine roomInter-temperature influence value.

Wherein the angle O of the three-dimensional subspace P relative to the ith air conditioner _ip The' can be calculated by the following formula:

distance S between ith air conditioner and three-dimensional subspace P _i,p The calculation can be made by the following formula:

in the above formula, (x) _i ，y _i ，z _i ) Representing the space position coordinate of the ith air conditioner in the target machine room (x) _p ，y _p ，z _p ) And representing the spatial position coordinates of the three-dimensional subspace P in the target machine room.

The first sub-model expressing the single-factor space temperature influence value of the air conditioner operation parameters of the n air conditioners on the three-dimensional subspace P can be expressed by the following formula:

Wherein n is the total number of air cavities in the target machine room, A ₁ And (p) representing the single-factor space temperature influence value of n air conditioners in the target machine room on the three-dimensional subspace p.

2. Second sub-model

The sub-model comprises: and expressing a second sub-model of the influence of the IT equipment load factor on the temperature of the machine room.

And operating the historical data of the IT equipment load factor through the second sub-model to obtain a single factor space temperature influence value of the IT equipment load factor on each three-dimensional subspace in the target machine room.

In some embodiments of the present application, the IT device load factor includes one or more of the following data: the calculating of gains is performed on the values of preset influence factors of the three-dimensional subspaces at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical moment, wherein the calculating comprises the following steps: and taking the value of the IT equipment load factor in the target machine room at the appointed historical moment, the position of each three-dimensional subspace and the heat conduction coefficient of air as second input data, and executing preset operation on the second input data through the second sub-model to obtain a single factor space temperature influence value of the IT equipment load factor corresponding to the appointed historical moment on each three-dimensional subspace in the target machine room.

The IT equipment load (such as a server) in the machine room can continuously release heat during the working process, thereby affecting the temperature of the machine room. Based on the thermal conduction model, an expression IT load heat source propagation consumption model can be constructed:

wherein p represents a three-dimensional subspace, j represents a server identifier in a target machine room, lambda is the heat conductivity coefficient of air, and T _j Represents the predicted temperature of the jth server based on the CPU usage and memory usage of the server and the IO consumption rate, S _j,p Representing the distance, K, between the jth server and the three-dimensional subspace P ₂ As model parameters, Δt ₂ And (p) representing the single-factor space temperature influence value of the jth server in the target machine room on the three-dimensional subspace p.

The second sub-model expressing the single-factor space temperature influence value of m IT load devices on the three-dimensional subspace P can be expressed by the following formula:

wherein m is the total number of servers in the target machine room, A ₂ (p) represents a single factor of m servers in the target machine room to the three-dimensional subspace pA space temperature influence value.

Wherein, the distance S between the jth server and the three-dimensional subspace P _j,p Is described throughout, and is not described in detail herein.

When the temperature of the server is predicted according to the CPU utilization rate and the memory utilization rate of the jth server and the IO consumption rate, the temperature of the server can be predicted by the following server temperature prediction model which is trained in advance:

T _j ＝W ₁ V _{cpu_j} +W ₂ V _{cache_j} +W ₃ V _{io_j} +T _{base_j} ；

Wherein V is _{cpu_j} CPU utilization of jth server is shown, V _{cache_j} Representing the memory utilization rate of the jth server, V _{io_j} Represents IO consumption rate, W of the jth server ₁ 、W ₂ And W is ₃ Representing the weight of the corresponding term, T being a constant obtained by training _{base_j} For the temperature of the jth server measured by the sensor, T _j The temperature of the server is predicted according to the CPU utilization rate, the memory utilization rate and the IO consumption rate of the jth server.

The server temperature is closely related to the service load, the key evaluation indexes of the server load are three indexes of CPU utilization rate, memory utilization rate and IO consumption, the actual temperature of the server is recorded through temperature sensing, and the temperature information of the server in each time period can be recorded in the running process of the server. The server temperature prediction model can be obtained by collecting CPU utilization rate, memory utilization rate and IO consumption rate of each server in the target machine room and the actual temperature of the server, and then training a regression model.

3. Third sub-model

The sub-model comprises: and expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room.

And calculating the corresponding historical data through a sub-model corresponding to each preset influence factor, so that a single factor space temperature influence value of the corresponding preset influence factor on each three-dimensional subspace in the target machine room can be obtained.

In some embodiments of the present application, the environmental factor includes one or more of the following data: the gain calculation is respectively performed on the values of the preset influence factors of the three-dimensional subspaces at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical moment, and the method comprises the following steps: and taking the value of the environmental factor at the appointed historical moment as third input data, and executing preset operation on the third input data through the third sub-model to obtain a single factor space temperature influence value of the environmental factor corresponding to the appointed historical moment on each three-dimensional subspace in the target machine room.

The outdoor temperature of the machine room and the air humidity have certain influence on the temperature in the machine room. Since the outdoor temperature value is in positive correlation with the temperature of the three-dimensional space in the machine room, and the humidity of the machine room is kept relatively stable, the humidity is a factor with small influence, and the third sub-model can be expressed by the following formula:

A ₃ (T _E ,H _E )＝K ₃ T _E +ε(H _E ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein T is _E Indicating the outdoor temperature of the target machine room, H _E Indicating the outdoor humidity, K of the target machine room ₃ As model parameters, A ₃ (T _E ,H _E ) And the single-factor space temperature influence value of the environmental factor of the target machine room on the three-dimensional subspace is represented.

In step 1102, for each of the specified historical time, the single factor space temperature influence values are fused to obtain an accumulated space temperature influence value of the preset influence factor on each three-dimensional subspace.

In the embodiment of the application, the temperature influence of various factors on the subspace of the machine room is fully considered, the single factor space temperature influence value calculated by the various submodels is fused, and the accumulated space temperature influence value of all preset influence factors on the three-dimensional subspaces is obtained. For example, the single-factor space temperature influence values calculated by the first sub-model, the second sub-model and the third sub-model may be summed, and then the sum is used as the cumulative space temperature influence value of all the preset influence factors on the three-dimensional subspace p. For example, the cumulative space temperature influence value Δtp of all preset influence factors on the three-dimensional subspace p can be calculated by the following formula:

ΔTP＝A ₁ (p)+A ₂ (p)+A ₃ (T _E ,H _E )。

according to the distribution position relation of the three-dimensional subspaces, the accumulated space temperature influence value corresponding to each subspace is used as a matrix element value, and a three-dimensional matrix formed by all preset influence factors on the accumulated space temperature influence values of all the three-dimensional subspaces in the target machine room can be obtained.

And step 1103, performing space dimension reduction processing on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix corresponding to the appointed historical moment.

And performing space dimension reduction processing on the accumulated space temperature influence value according to the space distribution of each three-dimensional subspace to obtain an influence value matrix of the preset influence factor on the two-dimensional space of the target machine room, and marking the influence value matrix as an air conditioner temperature gain matrix.

In some embodiments of the present application, performing a spatial dimension reduction process on the accumulated spatial temperature influence value to obtain an air conditioner temperature gain matrix corresponding to the specified historical moment, including: determining a projection plane of the three-dimensional subspace, wherein the projection plane is: a plane parallel to the side surface of the three-dimensional subspace in the horizontal direction, or a plane parallel to the side surface of the three-dimensional subspace in the vertical direction; projecting the three-dimensional subspace onto the projection plane; taking the average value of the accumulated space temperature influence values corresponding to the three-dimensional subspaces projected to the same area on the projection plane as the accumulated space temperature influence value corresponding to the corresponding area of the projection plane; and obtaining an air conditioner temperature gain matrix corresponding to the appointed historical moment according to the accumulated space temperature influence value corresponding to each region in the projection plane.

Taking the machine room subspace division schematic diagram shown in fig. 4 as an example, a plane in the vertical direction determined by the X axis and the Y axis may be selected as a projection plane, a plane in the horizontal direction determined by the X axis and the Z axis may be selected as a projection plane, a plane in the vertical direction determined by the Y axis and the Z axis may be selected as a projection plane, or a plane parallel to any one of the projection planes may be selected as a projection plane.

If each of the three-dimensional subspaces is projected toward the projection plane, it is seen that a plurality of subspaces located in a direction perpendicular to the projection plane are projected into the same grid of the projection plane. Therefore, according to the projection principle, each three-dimensional subspace projected to the same area on the projection plane can be determined, and then the average value of the accumulated space temperature influence values corresponding to each three-dimensional subspace of the same area is used as the accumulated space temperature influence value corresponding to the corresponding area of the projection plane.

Further by way of example, assume that the length, width and height of the target machine room are 3 meters and 3 meters respectively, and that the three-dimensional space division units are 1 meter, and according to the space coordinate system shown in fig. 4, the plane determined by the X axis and the Z axis in fig. 4 is taken as the projection plane. Then each 1 m x 1 m grid area on the projection plane is divided into 1 three-dimensional subspace every 1 m in the room space above the grid area, and Y subspaces are shared. As shown in fig. 5, for each grid region, the average value of the cumulative space temperature influence values of 3 subspaces in the machine room space above the grid region may be used as the cumulative space temperature influence value corresponding to the grid region. For example, an average value of the space temperature influence values Δtp1, Δtp2, and Δtp3 of the 3 subspaces projected to the grid region a is taken as the cumulative space temperature influence value corresponding to the grid region a.

By executing the above-mentioned space dimension-reduction processing, a two-dimensional matrix of the cumulative space temperature influence value corresponding to each grid region of the corresponding projection plane will be obtained. The two-dimensional matrix can be used as an air conditioner temperature gain matrix corresponding to the target machine room.

The dimension reduction method can obtain that the air conditioner temperature gain matrix expresses the temperature influence of preset influence factors on each subspace in the machine room.

The above-described sub-steps 1102 and 1103 are performed by a feature generation sub-model.

And step 1104, training a sub-model and the feature coding network corresponding to each preset influence factor according to the air conditioner temperature gain matrix corresponding to each appointed historical moment and the historical temperature of the target machine room.

According to the scheme, the values of the preset influence factors corresponding to different historical moments can be calculated to obtain an air-conditioning temperature gain matrix corresponding to the historical moments, the sequence of the air-conditioning temperature gain matrix and the historical temperature of the target machine room is input into a feature coding network, and feature coding and mapping are carried out by a convolution layer of a neural network model, so that a sub-model corresponding to each preset influence factor and the feature coding network can be trained.

And generating different air-conditioning temperature gain matrixes based on the same historical data by continuously optimizing the model parameters of each sub-model, so as to realize iterative training of the multi-dimensional temperature influence model of the machine room.

In the embodiment of the application, a characteristic coding network of a machine room multidimensional temperature influence model is constructed based on a SENet (Squeeze-and-Excitation Networks) network.

In the process of training a multi-dimensional temperature influence model of a machine room, taking 5 minutes as a period, selecting one month of machine room air conditioning parameters, server parameters, outdoor environment and actual historical temperature of the machine room, calculating a plurality of air conditioning temperature gain matrixes through the sub-model to obtain a plurality of air conditioning temperature gain matrix sequences, and constructing an original characteristic diagram; and then, carrying out feature compression through a Squeeze link of SENet.

And the SENet network performs space-time feature learning of each parameter of the environmental temperature of the machine room through three times of convolution training by spatial dimension reduction. Wherein, SEblock is expressed as:

wherein, represents the convolution operation,

convolution kernel representing three convolution operations, C representing a time series, U _c Representing the feature vector.

And then, realizing feature dimension reduction through a fully connected neural network. The process of dimension reduction can be expressed as:

Wherein H and W respectively represent the number of grids in two directions in the projection plane, namely H and W respectively correspond to the number of rows and columns of the air conditioner temperature gain matrix, F _sq Representing the convolution calculation process, Z _c The result is pooled for the average of the total area.

In order to limit the complexity of the model and promote the prediction of the air conditioning temperature of the conventional machine room, in the embodiment of the present application, two fully connected layers are used to parameterize the gating mechanism. And the dimension reduction is realized by using one connecting layer, so that the calculation is convenient, and the complexity of a model is reduced. And then, a dimension increasing layer is realized through a connecting layer, and the channel dimension of the output feature map is transferred.

In the specification link, in order to completely capture the channel dependency relationship, i.e. find the front-back relationship between different air conditioner temperature gain matrices, a simple gating mechanism with Sigmoid activation is used.

This function must meet two criteria: firstly, it must be flexible to operate and be able to learn the nonlinear relationship between channels; second, it must learn a non-exclusive relationship. In some embodiments of the present application, the following transformations are employed:

s＝F _ex (z,W)＝σ(g(z,W))＝σ(W ₂ δ(W ₁ z))；

in the above formula, delta represents a Relu activation function, sigma represents a Sigmoid activation function, wherein W ₁ And W is ₂ Is a model parameter.

After s is obtained, the final output of the se_block algorithm can be obtained by:

Q _c ＝F _scale (u _c ,s _c )＝s _c u _c ；

wherein Q is _c ＝[Q ₁ ,Q ₂ ,...]，u _c ∈R ^H×W ，F _scale Is the product on the channel, Q _c The target control temperature of the air conditioner corresponding to the target temperature of the machine room environment is obtained.

In the foregoing step 130, the real-time data and the target temperature are used as input of a pre-trained multi-dimensional temperature influence model of the machine room, and in the process of solving the real-time data and the target control temperature of the air conditioner corresponding to the target temperature through the multi-dimensional temperature influence model of the machine room, the obtained real-time data of the environment of the machine room and the target temperature of the machine room are input into the multi-dimensional temperature influence model of the machine room trained in advance, so that the obtained multi-dimensional temperature influence model of the machine room outputs a Q _c The target control temperature of the air conditioner is obtained.

As described above, the target machine room is divided into a plurality of three-dimensional subspaces, and the machine room multidimensional temperature influence model comprises: the method for solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room comprises the following steps of: gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces through the submodels, so that single-factor space temperature influence values of the three-dimensional subspaces are obtained; fusing the single-factor space temperature influence values through the feature generation sub-model to obtain accumulated space temperature influence values of the preset influence factors on each three-dimensional subspace; performing space dimension reduction on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix; and performing coding mapping on the air conditioner temperature gain matrix through the characteristic coding network to obtain the real-time data and the target control temperature of the air conditioner corresponding to the target temperature.

Gain calculation is performed on the values of the preset influence factors of the three-dimensional subspaces through the submodels respectively to obtain specific embodiments of single-factor space temperature influence values of the three-dimensional subspaces, and detailed description of model training stages is omitted here.

And fusing the single-factor space temperature influence values through the characteristic generation sub-model to obtain an accumulated space temperature influence value of the preset influence factors on each three-dimensional subspace, and performing space dimension reduction processing on the accumulated space temperature influence value to obtain a specific implementation mode of an air conditioner temperature gain matrix, wherein the specific implementation mode is described in a model training stage and is not repeated herein.

And inputting the obtained air conditioner temperature gain matrix into the characteristic coding network, wherein the machine room multidimensional temperature influence model is used for outputting a temperature sequence correspondingly, and the temperature sequence is the target control temperature of the air conditioner corresponding to the real-time data and the target temperature.

In some embodiments of the present application, the submodel comprises: expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of a machine room, wherein the air conditioner operation parameter factors comprise one or more of the following data: the method comprises the steps of performing gain calculation on values of preset influence factors of three-dimensional subspaces through the submodel respectively to obtain single-factor space temperature influence values of the three-dimensional subspaces, wherein the single-factor space temperature influence values comprise: taking the value of the air conditioner operation parameter factor and the position of each three-dimensional subspace as first input data, and executing preset operation on the first input data through the first sub-model to obtain a single factor space temperature influence value of the air conditioner operation parameter factor on each three-dimensional subspace.

In some embodiments of the present application, the submodel comprises: expressing a second sub-model of the effect of the IT device loading factor on the machine room temperature, the IT device loading factor comprising one or more of the following data: the CPU utilization rate and the memory utilization rate of the server, the IO consumption rate, the position of the server and the temperature of the server, wherein gain calculation is respectively carried out on the values of preset influence factors of all three-dimensional subspaces through the submodel to obtain single-factor space temperature influence values of all three-dimensional subspaces, and the method comprises the following steps: and taking the value of the IT equipment load factor in the target machine room, the position of each three-dimensional subspace and the heat conduction coefficient of air as second input data, and executing preset operation on the second input data through the second sub-model to obtain a single factor space temperature influence value of the IT equipment load factor on each three-dimensional subspace in the target machine room.

In some embodiments of the present application, the submodel comprises: expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room, wherein the environmental factors comprise one or more of the following data: the gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces, and the method comprises the following steps: and taking the value of the environmental factor as third input data, and executing preset operation on the third input data through the third sub-model to obtain a single factor space temperature influence value of the environmental factor on each three-dimensional subspace in the target machine room.

In some embodiments of the present application, the performing a spatial dimension reduction process on the accumulated spatial temperature influence value to obtain an air conditioner temperature gain matrix includes: determining a projection plane of the three-dimensional subspace, wherein the projection plane is: a plane parallel to the side surface of the three-dimensional subspace in the horizontal direction, or a plane parallel to the side surface of the three-dimensional subspace in the vertical direction; projecting the three-dimensional subspace onto the projection plane; taking the average value of the accumulated space temperature influence values corresponding to the three-dimensional subspaces projected to the same area on the projection plane as the accumulated space temperature influence value corresponding to the corresponding area of the projection plane; and obtaining an air conditioner temperature gain matrix according to the accumulated space temperature influence value corresponding to each region in the projection plane.

In the model application stage, the data processing process of each sub-model and the feature coding network in the model is the same as that of the model training stage, and the specific implementation of the foregoing steps refers to the related description of the model training stage, which is not repeated here.

In the embodiment of the application, in order to improve energy-saving effect, improve air conditioner temperature control's accuracy, carry out temperature regulation through the mode that divides into a plurality of three-dimensional subspaces with the computer lab space. Therefore, the size of the air conditioner temperature gain matrix is equivalent to the number of projection grids of the three-dimensional subspace of the target machine room on a projection plane. Taking an example that the target machine room comprises H×W×C three-dimensional subspaces, wherein H can represent the three-dimensional subspaces in the machine room length direction, W can represent the three-dimensional subspaces in the machine room width direction, and the air conditioner temperature gain matrix can be H×W. Correspondingly, the target temperature in the machine room is H×W temperature matrix.

After implementing the energy saving strategy (e.g. setting the target temperature of the room to 28 degrees, i.e. the value of the matrix element in the temperature matrix to 28), the real-time temperature of each of the aforementioned projection grids in the room needs to be measured for comparison with the predicted temperature.

As described above, to reduce hardware cost and complexity of deployment, the temperature acquisition device is typically deployed in a target machine room at a specified location, and the temperature acquisition device is not deployed in each three-dimensional subspace. Therefore, when the temperature of each three-dimensional subspace in the target machine room needs to be acquired, interpolation processing is required to be performed on the real-time temperature of the designated monitoring position in the target machine room acquired by the temperature acquisition equipment, so that the real-time temperature of each three-dimensional subspace is obtained and is used as the actual measurement temperature.

For example, obtaining the measured temperature of the target machine room includes: acquiring real-time temperature of a designated machine room position after the target machine room is subjected to temperature adjustment, wherein the real-time temperature is acquired by temperature acquisition equipment deployed in the target machine room; and carrying out temperature interpolation on the appointed space dimension according to the real-time temperature to obtain the actually measured temperature on the appointed space dimension in the target machine room. The appointed space dimension is consistent with the space dimension corresponding to the air conditioner temperature gain matrix.

For example, the real-time temperature acquired by the temperature acquisition device can be used,as the base temperature of the projection grid of the three-dimensional subspace corresponding to the temperature acquisition device in the projection plane, the temperatures of a plurality of projection grids in the projection plane can be obtained. Taking as an example a projection grid of a projection plane (as defined by the spatial coordinate axes X and Z) shown in FIG. 6, where T _{base_11} 、T _{base_12} 、T _{base_21} And T _{base_22} Projection plane coordinates (X) acquired by the temperature acquisition devices respectively ₁ ,Z ₁ )、(X ₁ ,Z ₂ )、(X ₂ ,Z ₁ ) And (X) ₂ ,Z ₂ ) Corresponding to the temperature in the projection grid. Then, for the projection grid without temperature record after the projection plane coordinates P (X, Z) corresponding to the three-dimensional subspace are projected to the projection plane, linear interpolation can be carried out according to the distance between the projection grid to be interpolated and the projection grid with known temperature in each dimension. Specifically, for example, the method for obtaining the base temperature of the projection grid where the projection plane coordinates P (X, Z) are located by interpolation is as follows:

first, linear interpolation is performed in the X direction by the following formula to obtain points (X, Z ₁ ) Temperature R of (2) ₁ Sum point (X, Z) ₂ ) Temperature R of (2) ₂ ：

Firstly, linear interpolation is carried out in the Z direction by the following formula to obtain the basic temperature T of a projection grid where the projection plane coordinates P (X, Z) are located _{base_P} ：

According to the method, temperature interpolation calculation is carried out on the projection grids of the subspace where the temperature acquisition equipment is not arranged for temperature acquisition, and after the base temperatures of all the projection grids are obtained, the base temperatures of all the projection grids form an actual measurement temperature matrix of the target machine room.

A large number of temperature sensors are required to be deployed in the traditional energy-saving method for temperature acquisition, investment cost is high, and the method can reduce the installation quantity of the temperature sensors and effectively reduce the deployment cost of the machine room by establishing a multi-dimensional gain temperature residual error model of the machine room and filling the local three-dimensional space temperature of the machine room by using a grid interpolation method.

In the case that the target machine room is divided into h×w×c three-dimensional subspaces, the machine room temperature prediction model used in the foregoing step 135 may be constructed based on SENet (Squeeze-and-Excitation Networks), and the machine room temperature prediction model is trained by taking the historical data of each preset influencing factor in the target machine room and the historical temperatures of each subspace of the target machine room as inputs, so as to learn the prediction capacities of the temperatures of each subspace of the machine room corresponding to different air-conditioning temperature setting values.

The training process and implementation principle of the machine room temperature prediction model refer to the training process and principle of the SENet network model in the prior art, and are not repeated in the embodiment of the present application.

When the temperature prediction is carried out, the real-time value of each preset influence factor of the target machine room and the target control temperature of the air conditioner obtained through the steps are input into a machine room temperature prediction model, and the machine room temperature prediction model outputs the predicted temperature of each subspace of the target machine room.

Further, according to the calculation requirement, the predicted temperature corresponding to each grid on the projection plane can be calculated according to the predicted temperature of each subspace.

In the case where the target machine room is divided into h×w×c three-dimensional subspaces, and the predicted target temperature of the machine room is set to the temperature matrix of h×w, the preset tuning threshold α may be set to the matrix of h×w, such as:

in this way, in the foregoing step 160, the predicted temperature and the measured temperature corresponding to each subspace in the target machine room are determined by using the matrix element values corresponding to the subspace, and it is determined whether to perform tuning training on the multi-dimensional temperature influence model of the machine room.

Those skilled in the art will appreciate that the above detailed description is merely one or more possible embodiments, and is not intended to limit the application to only possible embodiments.

Correspondingly, the embodiment of the application also discloses a temperature adjusting device of the air conditioner in the machine room, as shown in fig. 7, the device comprises:

the machine room target temperature and influence data obtaining module 720 is configured to obtain real-time data of a target temperature and a preset influence factor of the target machine room in response to temperature adjustment of an air conditioner in the target machine room, where the preset influence factor includes: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor;

The air conditioner temperature solving module 730 is configured to use the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solve, through the multi-dimensional temperature influence model of the machine room, a target control temperature of the air conditioner corresponding to the real-time data and the target temperature;

an air conditioner temperature adjustment module 740 for adjusting the temperature of the air conditioner based on the target control temperature.

In some embodiments of the present application, as shown in fig. 8, the apparatus further includes:

a machine room temperature prediction module 735, configured to predict, according to the real-time data and the target control temperature, a machine room temperature of the target machine room after the temperature adjustment is performed by using a pre-trained machine room temperature prediction model, as a predicted temperature;

the actually measured temperature obtaining module 750 is configured to obtain an actually measured temperature of the target machine room after performing temperature adjustment on the air conditioner based on the target control temperature;

and the model tuning processing module 760 is configured to perform tuning training on the multi-dimensional temperature influence model of the machine room when the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold.

Optionally, the target machine room is divided into a plurality of three-dimensional subspaces, and the multi-dimensional temperature influence model of the machine room comprises: the air conditioner temperature solving module 730 is further configured to:

projecting the three-dimensional subspace onto the projection plane;

In some embodiments of the present application, the apparatus further comprises: a machine room multidimensional temperature influence model training module (not shown in the figure);

the machine room multidimensional temperature influence model training module is used for training a machine room multidimensional temperature influence model according to the historical temperature of the target machine room and the value of a preset influence factor corresponding to the historical temperature.

Optionally, training a multidimensional temperature influence model of the machine room according to the historical temperature of the target machine room and the value of a preset influence factor corresponding to the historical temperature, including:

for each appointed historical moment, the following steps are respectively executed, and an air conditioner temperature gain matrix corresponding to the appointed historical moment is obtained:

gain calculation is respectively carried out on the values of the preset influence factors of the three-dimensional subspaces at the appointed historical moment through the submodel, so that single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical moment are obtained;

fusing the single factor space temperature influence values to obtain accumulated space temperature influence values of the preset influence factors on each three-dimensional subspace;

Performing space dimension reduction processing on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix corresponding to the appointed historical moment;

training a sub-model corresponding to each preset influence factor in the multi-dimensional temperature influence model of the machine room according to the air conditioner temperature gain matrix corresponding to each appointed historical moment and the historical temperature of the target machine room, and the feature coding network.

Optionally, the submodel includes: expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of a machine room, wherein the air conditioner operation parameter factors comprise one or more of the following data: the method comprises the steps of performing gain calculation on values of preset influence factors of each three-dimensional subspace at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of each three-dimensional subspace corresponding to the appointed historical moment, wherein the single-factor space temperature influence values comprise the following components:

and taking the value of the air conditioner operation parameter factor at the appointed historical moment and the position of each three-dimensional subspace as first input data, and executing preset operation on the first input data through the first submodel to obtain a single factor space temperature influence value of the air conditioner operation parameter factor corresponding to the appointed historical moment on each three-dimensional subspace.

Optionally, the submodel includes: expressing a second sub-model of the effect of the IT device loading factor on the machine room temperature, the IT device loading factor comprising one or more of the following data: the calculating of gains is performed on the values of preset influence factors of the three-dimensional subspaces at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical moment, wherein the calculating comprises the following steps:

and taking the value of the IT equipment load factor in the target machine room at the appointed historical moment, the position of each three-dimensional subspace and the heat conduction coefficient of air as second input data, and executing preset operation on the second input data through the second sub-model to obtain a single factor space temperature influence value of the IT equipment load factor corresponding to the appointed historical moment on each three-dimensional subspace in the target machine room.

Optionally, the submodel includes: expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room, wherein the environmental factors comprise one or more of the following data: the gain calculation is respectively performed on the values of the preset influence factors of the three-dimensional subspaces at the appointed historical moment through the submodel to obtain single-factor space temperature influence values of the three-dimensional subspaces corresponding to the appointed historical moment, and the method comprises the following steps:

And taking the value of the environmental factor at the appointed historical moment as third input data, and executing preset operation on the third input data through the third sub-model to obtain a single factor space temperature influence value of the environmental factor corresponding to the appointed historical moment on each three-dimensional subspace in the target machine room.

Optionally, the performing space dimension reduction processing on the accumulated space temperature influence value to obtain an air conditioner temperature gain matrix corresponding to the specified historical moment includes:

projecting the three-dimensional subspace onto the projection plane;

and obtaining an air conditioner temperature gain matrix corresponding to the appointed historical moment according to the accumulated space temperature influence value corresponding to each region in the projection plane.

The embodiment of the application discloses a temperature adjusting device for a machine room air conditioner, which is used for realizing the temperature adjusting method for the machine room air conditioner in the embodiment of the application, and specific implementation manners of each module of the device are not repeated, and reference can be made to specific implementation manners of corresponding steps in the embodiment of the method.

According to the machine room air conditioner temperature adjusting device disclosed by the embodiment of the application, real-time data of target temperature and preset influence factors of the target machine room are obtained by responding to temperature adjustment of an air conditioner in the target machine room, wherein the preset influence factors comprise: an air conditioner operation parameter factor, an IT equipment load factor and an environmental factor; taking the real-time data and the target temperature as input of a pre-trained multi-dimensional temperature influence model of the machine room, and solving the target control temperature of the air conditioner corresponding to the real-time data and the target temperature through the multi-dimensional temperature influence model of the machine room; based on the target control temperature to the air conditioner carries out temperature regulation, realizes the temperature of the multi-dimensional factor automatic control computer lab air conditioner that combines the influence computer lab temperature, promotes real-time and the temperature regulation degree of accuracy that can computer lab air conditioner temperature regulation, helps reducing the energy consumption of computer lab air conditioner, reaches the dynamic balance of computer lab temperature and air conditioner energy consumption.

Further, in the machine room air conditioner temperature adjusting device disclosed in the embodiment of the present application, after real-time data of a target temperature and the preset influencing factor of the target machine room are obtained in response to temperature adjustment of an air conditioner in the target machine room, and before the air conditioner is temperature-adjusted based on the target control temperature, the machine room temperature of the target machine room after the temperature adjustment is estimated by a pre-trained machine room temperature prediction model according to the real-time data and the target control temperature, and is used as a predicted temperature, and after the air conditioner is temperature-adjusted based on the target control temperature, an actual measurement temperature of the target machine room after the temperature adjustment is performed is obtained, and the predicted temperature and the actual measurement temperature are obtained; and under the condition that the absolute value of the difference between the predicted temperature and the measured temperature meets a preset tuning threshold, tuning and optimizing the multi-dimensional temperature influence model of the machine room, thereby further improving the accuracy of the temperature adjustment of the air conditioner of the machine room and improving the energy-saving effect.

In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described by differences from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other. For the device embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and reference is made to the description of the method embodiments for relevant points.

The above describes in detail a method and apparatus for adjusting temperature of an air conditioner in a machine room provided in the present application, and specific examples are applied to describe principles and embodiments of the present application, where the description of the above examples is only for helping to understand the method and a core idea of the present application; meanwhile, as those skilled in the art will have modifications in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.

The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

Various component embodiments of the present application may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functions of some or all of the components in an electronic device according to embodiments of the present application may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present application may also be embodied as an apparatus or device program (e.g., computer program and computer program product) for performing a portion or all of the methods described herein. Such a program embodying the present application may be stored on a computer readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.

For example, fig. 9 shows an electronic device in which a method according to the present application may be implemented. The electronic device may be a PC, a mobile terminal, a personal digital assistant, a tablet computer, etc. The electronic device conventionally comprises a processor 910 and a memory 920 and a program code 930 stored on said memory 920 and executable on the processor 910, said processor 910 implementing the method described in the above embodiments when said program code 930 is executed. The memory 920 may be a computer program product or a computer-readable medium. The memory 920 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read only memory), an EPROM, a hard disk, or a ROM. The memory 920 has a memory 9201 of program code 930 of a computer program for performing any of the method steps described above. For example, the memory space 9201 for the program code 930 may include individual computer programs for implementing the various steps in the above methods, respectively. The program code 930 is computer readable code. These computer programs may be read from or written to one or more computer program products. These computer program products comprise a program code carrier such as a hard disk, a Compact Disc (CD), a memory card or a floppy disk. The computer program comprises computer readable code which, when run on an electronic device, causes the electronic device to perform a method according to the above-described embodiments.

The embodiment of the application also discloses a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, realizes the steps of the computer room air conditioner temperature adjusting method according to the embodiment of the application.

Such a computer program product may be a computer readable storage medium, which may have memory segments, memory spaces, etc. arranged similarly to the memory 920 in the electronic device shown in fig. 9. The program code may be stored in the computer readable storage medium, for example, in a suitable form. The computer readable storage medium is typically a portable or fixed storage unit as described with reference to fig. 10. In general, the memory unit comprises computer readable code 930', which computer readable code 930' is code that is read by a processor, which code, when executed by the processor, implements the steps of the method described above.

Reference herein to "one embodiment," "an embodiment," or "one or more embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Furthermore, it is noted that the word examples "in one embodiment" herein do not necessarily all refer to the same embodiment.

In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the present application may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The application may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. The method for adjusting the temperature of the air conditioner in the machine room is characterized by comprising the following steps of:

2. The method according to claim 1, wherein after the real-time data and the target temperature are used as input of a pre-trained multi-dimensional temperature influence model of the machine room, and the real-time data and the target temperature are solved for the target control temperature of the air conditioner by the multi-dimensional temperature influence model of the machine room, the method further comprises:

obtaining the measured temperature of the target machine room;

3. The method according to claim 1 or 2, characterized in that the target machine room is divided into several three-dimensional subspaces, the machine room multidimensional temperature influence model comprising: a sub-model corresponding to each of the preset influence factors, a feature generation sub-model, and a feature encoding network,

4. A method according to claim 3, wherein the sub-model comprises: expressing a first sub-model of the influence of the air conditioner operation parameter factors on the temperature of a machine room, wherein the air conditioner operation parameter factors comprise one or more of the following data: air conditioning temperature, fan revolution, wind direction angle, air conditioning operation time, air conditioning position, and thermal conductivity of air,

5. A method according to claim 3, wherein the sub-model comprises: expressing a second sub-model of the effect of the IT device loading factor on the machine room temperature, the IT device loading factor comprising one or more of the following data: CPU utilization rate and memory utilization rate of the server, and IO consumption rate, position of the server, temperature of the server,

6. A method according to claim 3, wherein the sub-model comprises: expressing a third sub-model of the influence of the environmental factors on the temperature of the machine room, wherein the environmental factors comprise one or more of the following data: the outdoor temperature and the air humidity are controlled,

7. The method of claim 3, wherein performing the spatial dimension reduction on the cumulative spatial temperature impact value to obtain an air conditioner temperature gain matrix comprises:

projecting the three-dimensional subspace onto the projection plane;

8. The utility model provides a computer lab air conditioner temperature regulating device which characterized in that includes:

9. The apparatus as recited in claim 8, further comprising:

10. An electronic device comprising a memory, a processor and program code stored on the memory and executable on the processor, wherein the processor implements the room air conditioning temperature regulating method of any one of claims 1 to 7 when the program code is executed by the processor.

11. A computer readable storage medium having stored thereon program code, which when executed by a processor, implements the steps of the room air conditioning temperature regulating method of any of claims 1 to 7.