CN116466672B

CN116466672B - Data center machine room parameter regulation and control method based on artificial intelligence and related device

Info

Publication number: CN116466672B
Application number: CN202310691920.1A
Authority: CN
Inventors: 李逸伦; 张超; 李原洲
Original assignee: Shenzhen Baoteng Internet Technology Co ltd
Current assignee: Shenzhen Baoteng Internet Technology Co ltd
Priority date: 2023-06-13
Filing date: 2023-06-13
Publication date: 2023-11-10
Anticipated expiration: 2043-06-13
Also published as: CN116466672A

Abstract

The invention relates to the field of artificial intelligence, and discloses a data center machine room parameter regulation and control method and a related device based on artificial intelligence, which are used for realizing intelligent parameter regulation and control of a data center machine room and improving the accuracy of parameter regulation and control. The method comprises the following steps: inputting wind system operation data into a plurality of first training models to perform model training to obtain a first model set, and inputting water system operation data into a plurality of preset second training models to perform model training to obtain a second model set; taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and the second layer decision element model to obtain a target parameter regulation model; vector conversion is carried out on the second operation data and the machine room environment data to obtain a target operation vector; and inputting the target operation vector into a target parameter regulation model to perform machine room parameter regulation analysis, so as to obtain a tower air quantity regulation parameter and a cooling water quantity regulation parameter.

Description

Data center machine room parameter regulation and control method based on artificial intelligence and related device

Technical Field

The invention relates to the field of artificial intelligence, in particular to a data center machine room parameter regulation and control method and a related device based on artificial intelligence.

Background

Data centers are an important infrastructure of the current digital society, and the energy resources and operation costs required for the data centers are continuously increased, so that the improvement of the energy utilization efficiency of the data centers has become an urgent task. Under the background, the parameter regulation and control method of the data center machine room can help a data center administrator to realize accurate machine room parameter regulation and control through an intelligent technology so as to reduce energy consumption and operation cost of the machine room.

However, the current data center machine room parameter regulation and control technology adopts a periodical regulation and control strategy, and cannot accurately reflect real-time machine room environment change and energy consumption conditions, so that energy waste and energy consumption imbalance are caused. In addition, due to the complexity and uncertainty of the data center room, many existing data acquisition and analysis methods may cause inaccurate data or larger errors, thereby affecting the accuracy and effect of parameter regulation.

Disclosure of Invention

The invention provides a data center machine room parameter regulation and control method and a related device based on artificial intelligence, which are used for realizing intelligent parameter regulation and control of a data center machine room and improving the accuracy of parameter regulation and control.

The first aspect of the invention provides a data center room parameter regulation method based on artificial intelligence, which comprises the following steps:

Acquiring first operation data of a machine room temperature control system, and classifying attributes of the first operation data to obtain wind system operation data and water system operation data;

inputting the wind system operation data into a plurality of preset first training models to perform model training to obtain a first model set, and inputting the water system operation data into a plurality of preset second training models to perform model training to obtain a second model set;

taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model;

acquiring second operation data to be processed and machine room environment data, and performing vector conversion on the second operation data and the machine room environment data to obtain a target operation vector;

inputting the target operation vector into the target parameter regulation model for carrying out machine room parameter regulation analysis to obtain a tower air quantity regulation parameter and a cooling water quantity regulation parameter;

and generating a parameter regulation scheme of the machine room temperature control system according to the tower air volume regulation parameter and the cooling water volume regulation parameter, and monitoring the optimal temperature and the lowest power of the machine room temperature control system according to the parameter regulation scheme to obtain the optimal temperature point and the lowest total power of the machine room temperature control system.

With reference to the first aspect, in a first implementation manner of the first aspect of the present invention, the obtaining first operation data of the machine room temperature control system and classifying attributes of the first operation data to obtain wind system operation data and water system operation data includes:

acquiring first operation data of a machine room temperature control system, and acquiring a first attribute identifier of a stroke system and a second attribute identifier of a water system of the machine room temperature control system;

classifying and extracting the first operation data according to the first attribute identifier to obtain wind system operation data, wherein the wind system operation data comprises: cooling tower fan frequency, air volume data and air system IT data;

classifying and extracting the first operation data according to the second attribute identifier to obtain water system operation data, wherein the water system operation data comprises: host cooling water pump frequency, water flow data, and water system IT data.

With reference to the first aspect, in a second implementation manner of the first aspect of the present invention, the inputting the wind system operation data into a preset plurality of first training models to perform model training to obtain a first model set, and inputting the water system operation data into a preset plurality of second training models to perform model training to obtain a second model set, includes:

Respectively carrying out model training on a plurality of preset first training models according to cooling tower fan frequency, air volume data and air system IT data in the air system operation data to obtain a plurality of first prediction results;

setting the model weight of each first training model according to the plurality of first prediction results to obtain the first model weight of each first training model;

performing model integration on the plurality of first training models according to the first model weights to obtain a first model set;

respectively carrying out model training on a plurality of preset second training models according to the frequency of a host cooling water pump, water flow data and water system IT data in the water system operation data to obtain a plurality of second prediction results;

setting the model weight of each second training model according to the plurality of second prediction results to obtain the second model weight of each second training model;

and carrying out model integration on the plurality of second training models according to the second model weights to obtain a second model set.

With reference to the first aspect, in a third implementation manner of the first aspect of the present invention, the taking the first model set and the second model set as a first layer decision base model, and performing model fusion on the first layer decision base model and a preset second layer decision meta model to obtain a target parameter regulation model includes:

Model merging is carried out on the first model set and the second model set, and the first model set and the second model set are used as a first layer decision base model;

and carrying out model connection on the first layer of decision base model and the preset second layer of decision element model through a preset middle layer to obtain a target parameter regulation model.

With reference to the first aspect, in a fourth implementation manner of the first aspect of the present invention, the obtaining second operation data to be processed and machine room environment data, and performing vector conversion on the second operation data and the machine room environment data to obtain a target operation vector includes:

acquiring second operation data to be processed and acquiring machine room environment data of the machine room temperature control system, wherein the machine room environment data comprises wet bulb temperature, wet bulb humidity and outdoor temperature;

vector encoding is carried out on the second operation data to obtain an initial operation vector, and vector encoding is carried out on the machine room environment data to obtain an environment evaluation vector;

and vector splicing is carried out on the initial operation vector and the environment evaluation vector to obtain a target operation vector.

With reference to the first aspect, in a fifth implementation manner of the first aspect of the present invention, inputting the target operation vector into the target parameter regulation model to perform machine room parameter regulation analysis, to obtain a tower air volume regulation parameter and a cooling water volume regulation parameter, includes:

Inputting the target operation vector into the target parameter regulation model, wherein the target parameter regulation model comprises: a first layer of decision base model, a middle layer of decision base model and a second layer of decision base model;

performing feature extraction and predictive analysis on the target operation vector through the first layer decision base model to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value;

and inputting the tower air volume parameter predicted value and the cooling water volume parameter predicted value into the second layer decision element model through the middle layer for parameter verification to obtain a tower air volume regulation parameter and a cooling water volume regulation parameter.

With reference to the first aspect, in a sixth implementation manner of the first aspect of the present invention, the generating a parameter regulation scheme of the machine room temperature control system according to the tower air volume regulation parameter and the cooling water volume regulation parameter, and performing optimal temperature and minimum power monitoring on the machine room temperature control system according to the parameter regulation scheme to obtain an optimal temperature point and a minimum total power of the machine room temperature control system, includes:

determining the current tower air quantity and the current cooling water quantity of the machine room temperature control system according to the second operation data;

Generating a first regulation scheme according to the tower air volume regulation parameter and the current tower air volume, and generating a second regulation scheme according to the cooling water volume regulation parameter and the current cooling water volume;

generating a parameter regulation scheme of the machine room temperature control system according to the first regulation scheme and the second regulation scheme;

performing optimal temperature monitoring on the machine room temperature control system according to the parameter regulation scheme to obtain an optimal temperature point;

and acquiring the minimum total power corresponding to the machine room temperature control system according to the optimal temperature point.

The second aspect of the invention provides an artificial intelligence-based data center room parameter regulation and control device, which comprises:

the acquisition module is used for acquiring first operation data of the machine room temperature control system, classifying the first operation data in attribute, and acquiring wind system operation data and water system operation data;

the training module is used for inputting the operation data of the wind system into a plurality of preset first training models to perform model training to obtain a first model set, and inputting the operation data of the water system into a plurality of preset second training models to perform model training to obtain a second model set;

The fusion module is used for taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model;

the conversion module is used for acquiring second operation data to be processed and machine room environment data, and carrying out vector conversion on the second operation data and the machine room environment data to obtain a target operation vector;

the analysis module is used for inputting the target operation vector into the target parameter regulation model to carry out machine room parameter regulation analysis, so as to obtain tower air quantity regulation parameters and cooling water quantity regulation parameters;

and the regulation and control module is used for generating a parameter regulation and control scheme of the machine room temperature control system according to the tower air volume regulation and control parameters and the cooling water volume regulation and control parameters, and monitoring the optimal temperature and the lowest power of the machine room temperature control system according to the parameter regulation and control scheme to obtain the optimal temperature point and the lowest total power of the machine room temperature control system.

The third aspect of the invention provides an artificial intelligence-based data center machine room parameter regulation and control device, which comprises: a memory and at least one processor, the memory having instructions stored therein; and the at least one processor invokes the instructions in the memory to enable the artificial intelligence-based data center room parameter regulating equipment to execute the artificial intelligence-based data center room parameter regulating method.

A fourth aspect of the present invention provides a computer readable storage medium having instructions stored therein that, when executed on a computer, cause the computer to perform the above-described artificial intelligence based data center room parameter tuning method.

In the technical scheme provided by the invention, the wind system operation data is input into a plurality of first training models to perform model training to obtain a first model set, and the water system operation data is input into a plurality of preset second training models to perform model training to obtain a second model set; taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and the second layer decision element model to obtain a target parameter regulation model; vector conversion is carried out on the second operation data and the machine room environment data to obtain a target operation vector; the invention realizes real-time monitoring and accurate regulation of parameters such as temperature, humidity, energy consumption and the like of a machine room, thereby improving the energy utilization efficiency and energy saving effect of a data center, reducing the operation cost of the data center, adopting an AI intelligent decision technology, automatically adjusting the operation parameters of a machine room temperature control system, reducing the energy consumption of the temperature control system, reducing the PUE value of the data center, improving the operation efficiency of the machine room, utilizing a multi-model fusion technology, more accurately predicting the environment and energy consumption condition of the machine room, improving the accuracy of data analysis and regulation, further adopting an optimization control strategy, finding the optimal temperature point and the minimum total power of the temperature control system of the machine room, improving the regulation effect of the machine room, realizing the intelligent parameter regulation of the machine room of the data center, and improving the accuracy of parameter regulation.

Drawings

FIG. 1 is a schematic diagram of an embodiment of a method for regulating parameters of a data center room based on artificial intelligence in an embodiment of the invention;

FIG. 2 is a flow chart of model training in an embodiment of the invention;

FIG. 3 is a flow chart of model fusion in an embodiment of the invention;

FIG. 4 is a flow chart of vector conversion according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of an embodiment of an artificial intelligence-based data center room parameter control device according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of an embodiment of an artificial intelligence-based data center room parameter regulation apparatus according to an embodiment of the present invention.

Detailed Description

The embodiment of the invention provides a data center machine room parameter regulation and control method based on artificial intelligence and a related device, which are used for realizing intelligent parameter regulation and control of a data center machine room and improving the accuracy of parameter regulation and control. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For ease of understanding, a specific flow of an embodiment of the present invention is described below, referring to fig. 1, and an embodiment of a method for adjusting parameters of a data center room based on artificial intelligence in the embodiment of the present invention includes:

s101, acquiring first operation data of a machine room temperature control system, and classifying attributes of the first operation data to obtain wind system operation data and water system operation data;

it can be understood that the execution subject of the present invention may be an artificial intelligence based data center room parameter adjusting device, and may also be a terminal or a server, which is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example.

Specifically, the server installs a temperature control system, adds a sensor and a control device, and connects them to the network. And acquiring data generated when the temperature control system is operated for the first time, wherein the data comprise a series of parameters such as temperature, humidity, wind speed, water pressure and the like. And classifying the attribute of the acquired first operation data. The classification algorithm may be used to classify the attributes, classifying the data by various attributes, such as temperature, humidity, and wind and water pressure. And obtaining wind system operation data and water system operation data according to the obtained attribute classification result. For example, the wind speed and wind direction data are used as wind system operation data, and the water pressure and water flow data are used as water system operation data. The data is monitored and analyzed and compared with historical data to predict future operational trends of the temperature control system. Therefore, the problems can be found and solved in advance before the problems occur in the machine room, and the normal operation of the machine room is ensured. The data is visualized for analysis and understanding by the user. Various charts and reports may be used to present data, such as bar charts, line charts, real-time monitoring, and the like. For example, a room temperature control system is an electronic control system that monitors real-time temperature, humidity, etc. parameters and adjusts temperature via a wind system and a water system. And for the machine room temperature control system, acquiring first operation data of the temperature control system, and classifying the data to obtain wind system operation data and water system operation data. Firstly, the temperature control system is installed in a machine room, so that first operation data are obtained, the data are classified, for example, temperature and humidity data are used as one group, wind speed and wind direction data are used as another group, and water flow and water pressure data are used as the last group. Next, the obtained data is visualized. The real-time monitoring system can be used for displaying the change conditions of temperature and humidity, the line graph is used for displaying temperature and humidity data and the like, so that the operation condition of the machine room temperature control system can be more intuitively known, the problem can be rapidly positioned and solved when the problem occurs, and the high-efficiency operation of the machine room temperature control system is ensured.

S102, inputting wind system operation data into a plurality of preset first training models to perform model training to obtain a first model set, and inputting water system operation data into a plurality of preset second training models to perform model training to obtain a second model set;

specifically, the server needs to perform some preprocessing, such as data cleaning, data normalization, and the like, before inputting the wind system or water system operation data into the model, so as to ensure the accuracy of model training. A number of different classification and regression models, such as linear regression, logistic regression, decision trees, support vector machines, etc., may typically be selected to suit the machine learning model of the current dataset. According to different data distribution and prediction requirements, a more suitable model is selected. Training is performed on a plurality of first (or second) training models, respectively, using the selected models, to form a first (or second) set of models. In the training process, the data set needs to be divided, and the data set is generally divided according to a training set, a verification set, a test set and the like, so that generalization of the model is ensured. During training, a validation set is used to evaluate the performance of each model, such as accuracy, AUC, F1-Score, and the like. Based on these metrics, an optimal model is selected for subsequent prediction and practical application. In the model training process, model tuning is very necessary. Different hyper-parameters, different loss functions or optimizers may be tried to optimize the performance of the model. For example, consider the classification and prediction of wind and water system operational data for a large plant. The method comprises the steps that a plurality of preset first training models are input into wind system operation data to perform model training, and a first model set is obtained. To meet the predicted demand and data distribution, different types of machine learning models such as logistic regression, decision trees, SVMs, etc. can be selected, such as wind, temperature, humidity, etc. Data preprocessing, such as missing value filling, normalization, etc., is required before model training. In the training process, the data set is divided into a training set, a verification set and a test set so as to evaluate and select the model. And finally, selecting an optimal model according to the index, and performing corresponding model tuning operation. Similarly, the model training is performed by inputting the water system operation data into a preset plurality of second training models, and the optimal model is selected from the available model set, which is a similar process. The links of data preprocessing, model selection, model training, model evaluation, selection and the like are all needed to be completed in sequence.

S103, taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model;

it should be noted that, the first model set and the second model set are used as the first layer decision base model, and the data set is used to tune the first layer decision base model, so as to realize a better prediction result. The first layer of model integration may use various integration methods, such as voting, stacking, learning, etc., to achieve performance optimization of the integrated model. And selecting and extracting the result obtained by integrating the first layer of model and other characteristics, and inputting the result and other characteristics into a preset second layer of decision meta-model for training to obtain a target parameter regulation model. After the model fusion is completed, parameter tuning is performed on the model. Common methods are grid search, genetic algorithm, etc. These methods can select the best parameters for the model in the hyper-parametric domain and maximize the performance of the model. For example, wind speed and yaw angle data are input into a preset first model set and second model set, and model integration is performed. After the first layer is integrated, a more accurate prediction result is obtained, the prediction result and other features are combined together, and the prediction result and other features are input into a preset second layer decision meta-model for model training. After the model fusion is completed, parameter tuning is performed on the model. Genetic algorithms may be used to select the best superparameters for the model to maximize the performance of the model. The finally obtained target parameter regulation model can be used for predicting and adjusting wind speed and yaw angle data of the wind power station so as to ensure normal operation of the wind power station. The first model set and the second model set are used as a first layer decision base model, and model fusion is carried out by using a preset second layer decision element model to obtain a target parameter regulation model, so that the accuracy and the prediction performance of the model can be greatly improved.

S104, acquiring second operation data to be processed and machine room environment data, and performing vector conversion on the second operation data and the machine room environment data to obtain a target operation vector;

specifically, second operation data to be processed and machine room environment data are obtained. The second operational data and the machine room environmental data include various parameters such as temperature, humidity, pressure, current, voltage, etc. These data may be measured by sensors or monitoring devices. Some pre-processing is performed on the data before vector conversion. For example, operations such as data cleansing, missing value filling, data normalization, etc. are required to ensure data quality and consistency. And vector conversion is carried out on the second operation data and the machine room environment data. Typically, feature extraction is performed on the data prior to vector conversion to reduce vector dimensions and improve the accuracy of the vector representation. Various conversion methods such as Principal Component Analysis (PCA), linear Discriminant Analysis (LDA), and the like may be used. And obtaining a target operation vector by carrying out vector conversion on the second operation data and the machine room environment data. For example, consider CPU load and memory usage data for several servers in a machine room, as well as temperature and humidity data for the machine room. To obtain the target operation vector, vector conversion is required for these data. And carrying out data fusion on the CPU load and memory utilization rate data and the temperature and humidity data to obtain a complete target data set. And then preprocessing such as filling in data missing values, data normalization and the like. Feature extraction and transformation is then performed, for example using PCA. Through these steps, a target running vector, which is a vector containing all variables, can be obtained. The vector is input into a subsequent system, and target results such as fault detection, system control and the like can be obtained.

S105, inputting the target operation vector into a target parameter regulation model for carrying out machine room parameter regulation analysis to obtain a tower air quantity regulation parameter and a cooling water quantity regulation parameter;

specifically, the server target parameter regulation model: a target parameter tuning model is constructed, which may be a machine learning model or other type of mathematical model constructed from historical data and prior knowledge. Target operation vector input: and inputting the target operation vector to be processed into the target parameter regulation model so as to carry out machine room parameter regulation analysis. Analysis results: and obtaining the tower air quantity regulation parameters and the cooling water quantity regulation parameters according to the analysis result of the target parameter regulation model. These parameters can be used to adjust the status of the associated equipment or control system, thereby ensuring stable operation of the machine room and saving energy. For example, assume that the machine room includes a high power computer and a laser cutter. The target operating vector of the machine room may include various parameters such as temperature, humidity, machine load, battery power, etc. And inputting the target operation vector into a preset target parameter regulation model to perform machine room parameter regulation analysis. And obtaining the tower air quantity regulation parameters and the cooling water quantity regulation parameters through model analysis.

S106, generating a parameter regulation scheme of the machine room temperature control system according to the tower air volume regulation parameter and the cooling water volume regulation parameter, and monitoring the optimal temperature and the lowest power of the machine room temperature control system according to the parameter regulation scheme to obtain the optimal temperature point and the lowest total power of the machine room temperature control system.

Specifically, according to the tower air volume regulation parameters and the cooling water volume regulation parameters, a parameter regulation scheme of the machine room temperature control system is generated through machine learning, mathematical optimization and other technologies. The schedule may include a number of parameters such as fan speed, water pump flow, cooling water temperature, etc. And according to the parameter regulation scheme, the optimal temperature and the lowest power of the machine room temperature control system are monitored. The process may be implemented by a sensor or monitoring system, such as a temperature sensor, a power meter, or the like. And analyzing the optimal temperature point and the minimum total power of the temperature control system of the machine room according to the monitoring data and the regulation scheme. These data can be used to improve the performance of the temperature control system, optimize the decisions of the regulation scheme, etc. For example, assuming that one machine room includes a plurality of servers and other devices, a parameter regulation scheme of the machine room temperature control system is generated according to the tower air volume regulation parameter and the cooling water volume regulation parameter. For example, the control of the machine room temperature may be achieved by increasing the fan speed and adjusting the cooling water flow. And then, monitoring the optimal temperature and the lowest power of the machine room temperature control system through a temperature sensor and a power meter to obtain the optimal temperature point and the lowest total power of the machine room temperature control system.

In the embodiment of the invention, the operation data of the wind system is input into a plurality of first training models to perform model training to obtain a first model set, and the operation data of the water system is input into a plurality of preset second training models to perform model training to obtain a second model set; taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and the second layer decision element model to obtain a target parameter regulation model; vector conversion is carried out on the second operation data and the machine room environment data to obtain a target operation vector; the invention realizes real-time monitoring and accurate regulation of parameters such as temperature, humidity, energy consumption and the like of a machine room, thereby improving the energy utilization efficiency and energy saving effect of a data center, reducing the operation cost of the data center, adopting an AI intelligent decision technology, automatically adjusting the operation parameters of a machine room temperature control system, reducing the energy consumption of the temperature control system, reducing the PUE value of the data center, improving the operation efficiency of the machine room, utilizing a multi-model fusion technology, more accurately predicting the environment and energy consumption condition of the machine room, improving the accuracy of data analysis and regulation, further adopting an optimization control strategy, finding the optimal temperature point and the minimum total power of the temperature control system of the machine room, improving the regulation effect of the machine room, realizing the intelligent parameter regulation of the machine room of the data center, and improving the accuracy of parameter regulation.

In a specific embodiment, the process of executing step S101 may specifically include the following steps:

(1) Acquiring first operation data of a machine room temperature control system, and acquiring a first attribute identifier of a stroke system and a second attribute identifier of a water system of the machine room temperature control system;

(2) Classifying and extracting the first operation data according to the first attribute identifier to obtain wind system operation data, wherein the wind system operation data comprises: cooling tower fan frequency, air volume data and air system IT data;

(3) The first operation data is classified and extracted according to the second attribute identification to obtain water system operation data, wherein the water system operation data comprises: host cooling water pump frequency, water flow data, and water system IT data.

Specifically, the server obtains first operation data of the machine room temperature control system. The first operational data includes a variety of parameters such as temperature, humidity, power, current, etc. Can be measured by a sensor or a monitoring device. And then acquiring a first attribute identifier of the air flow system and a second attribute identifier of the water system of the machine room temperature control system. These attribute identifications may be determined based on system architecture and device specifications. Data classification and extraction: and classifying and extracting the first operation data according to the first attribute identification of the wind system to obtain the operation data of the wind system. The wind system operation data comprises cooling tower fan frequency, air volume data, wind system IT data and the like. In particular, assume that the room temperature control system includes a wind system and a water system. The first attribute identification of the wind system may be cooling tower fan frequency and the second attribute identification of the water system may be water pump speed. According to the attribute identifiers, the first operation data can be classified and extracted to obtain wind system operation data. For example, the wind system operation data may include cooling tower fan frequency, air volume data, and wind system IT data. Such data may be used to monitor and control the wind system, such as to detect fan operating conditions, adjust air volume, etc.

In a specific embodiment, as shown in fig. 2, the process of executing step S102 may specifically include the following steps:

s201, respectively carrying out model training on a plurality of preset first training models according to cooling tower fan frequency, air volume data and air system IT data in the air system operation data to obtain a plurality of first prediction results;

s202, setting the model weight of each first training model according to a plurality of first prediction results to obtain the first model weight of each first training model;

s203, carrying out model integration on a plurality of first training models according to the first model weight to obtain a first model set;

s204, respectively carrying out model training on a plurality of preset second training models according to the frequency of the host cooling water pump, water flow data and water system IT data in the water system operation data to obtain a plurality of second prediction results;

s205, setting the model weight of each second training model according to a plurality of second prediction results to obtain the second model weight of each second training model;

s206, performing model integration on the plurality of second training models according to the second model weights to obtain a second model set.

Specifically, the server obtains the fan frequency and the air volume data of the cooling tower and the IT data of the air system in the operation data of the air system, and can obtain the fan frequency and the air volume data and the IT data of the air system through a sensor or monitoring equipment. Then, a plurality of first training models, which are typically built by algorithms such as neural networks, support vector machines, etc., need to be preset. Each model corresponds to a model weight for controlling the degree of preference of the model for training data. And respectively carrying out model training on the plurality of first training models according to the cooling tower fan frequency, the air volume data and the air system IT data in the air system operation data so as to obtain a plurality of first prediction results. Specifically, a conventional supervised learning method such as a back propagation algorithm or a support vector regression algorithm may be used. Each model obtained after training can be used for the next prediction. The model weights for each first training model are then set to obtain first model weights for each model. And calculating according to the prediction result so that the weight better reflects the better fitting degree of each model in the wind system. For example, the weight may be set according to an index such as an average absolute error (MAE) of the prediction result and a variance of the prediction result. For example, assume that two first training models are preset, where model A uses a support vector machine algorithm and model B uses a neural network algorithm. And respectively inputting the wind system operation data into the two models for training to obtain the prediction results of the two models. The MAE index may then be used to calculate model weights for the two models so that the weights better reflect a better fit of each model in the wind system. For example, if model A's MAE value is smaller, then model weights may be increased in model B to make model B more predictive of the wind system. And carrying out model integration on the plurality of first training models according to the first model weight to obtain a first model set. Model integration may use a variety of methods, such as simple averaging, weighted averaging, bagging, and the like. Wherein the weighted average method can perform weighted average on the model according to the weight so as to obtain more accurate results. Then, host cooling water pump frequency, water flow data and water system IT data in the water system operation data are acquired, and these data can be acquired by sensors or monitoring devices. Further, a plurality of second training models are preset, and the models are usually built by algorithms such as a neural network, a support vector machine and the like. Each model corresponds to a model weight for controlling the degree of preference of the model for training data. And respectively carrying out model training on the plurality of second training models according to the frequency of the host cooling water pump, the water flow data and the IT data of the water system in the water system operation data so as to obtain a plurality of second prediction results. Specifically, a conventional supervised learning method such as a back propagation algorithm or a support vector regression algorithm may be used. Each model obtained after training can be used for the next prediction. For example, two second training models are preset, where model C uses a neural network algorithm and model D uses a support vector machine algorithm. And respectively inputting the operation data of the water system into the two models for training to obtain the prediction results of the two models. Then, the prediction results can be combined by using a model integration method such as a simple average method and the like to obtain more accurate prediction results. And setting the model weight of each second training model according to the plurality of second prediction results so as to obtain the second model weight of each model. This may be calculated based on the predictions so that the weights better reflect the better fit of each model to the water system. For example, the weight may be set according to an index such as an average absolute error (MAE) of the prediction result and a variance of the prediction result. And then, carrying out model integration on the plurality of second training models according to the second model weights to obtain a second model set. This may use various methods such as a simple averaging method, a weighted averaging method, bagging, and the like. Wherein the weighted average method can perform weighted average on the model according to the weight so as to obtain more accurate results. For example, assume that two second training models are preset, where model E uses a support vector machine algorithm and model F uses a neural network algorithm. And respectively inputting the water system operation data into the two models for training to obtain the prediction results of the two models. The MAE index may then be used to calculate model weights for the two models so that the weights better reflect the better fit of each model in the water system. For example, if the MAE value of model E is smaller, then the model weight may be increased in model F to make model F more predictive of the water system. Further, the prediction results may be combined by a weighted average method or the like to obtain a more accurate prediction result.

In a specific embodiment, as shown in fig. 3, the process of executing step S103 may specifically include the following steps:

s301, carrying out model combination on the first model set and the second model set, and taking the first model set and the second model set as a first layer decision base model;

s302, performing model connection on the first layer of decision base model and the preset second layer of decision element model through a preset middle layer to obtain a target parameter regulation model.

Specifically, the server performs model merging on the first model set and the second model set to obtain a first layer decision base model. Model merging may be based on a variety of model integration methods, such as simple averaging, bagging, stacking, etc., and a method appropriate for the problem may be selected to achieve model integration. And then, presetting a middle layer, and carrying out model connection on the first layer decision base model and a preset second layer decision element model. In this intermediate layer, various methods such as neural networks, bayesian optimization, and the like can be used. There is a need for an objective and method of explicit model joining. And further, training a second layer of decision meta model to connect the first layer of decision base model and the second layer of model with the middle layer to obtain the target parameter regulation model. For training of the second-layer model, some more complex methods, such as reinforcement learning, genetic algorithm, deep learning, etc., need to be used. For example, assume that two model sets are combined to obtain a first layer decision base model. The first layer decision base model and the second layer decision base model may then be connected using a neural network as an intermediate layer. Here, the second layer model may be a neural network that requires training data and optimization methods to adjust its parameters and weights. According to the training result, a final target parameter regulation model can be obtained, and can be used for predicting parameters of the water system and the wind system and other related data.

In a specific embodiment, as shown in fig. 4, the process of executing step S104 may specifically include the following steps:

s401, acquiring second operation data to be processed and acquiring machine room environment data of a machine room temperature control system, wherein the machine room environment data comprise wet bulb temperature, wet bulb humidity and outdoor temperature;

s402, vector encoding is carried out on the second operation data to obtain an initial operation vector, and vector encoding is carried out on the machine room environment data to obtain an environment evaluation vector;

s403, vector stitching is carried out on the initial operation vector and the environment evaluation vector, and a target operation vector is obtained.

Specifically, the server first obtains second operation data to be processed and machine room environment data of the machine room temperature control system. Such data may typically be acquired by sensors or monitoring devices. Further, the second operation data and the room environment data are vector-encoded and are represented in a vector format in a computer. The vector encoding can be performed by a variety of methods, such as One-Hot encoding, bag of words model, LDA, etc. The data here may be normalized or normalized to ensure vector uniformity. The vector code of the second operational data is converted into an initial operational vector. This vector may include a number of operating variables that need to be monitored, such as the frequency of the water pump, the output of the pressure sensor, the output of the flow sensor, etc. And (3) vector coding of the machine room environment data is converted into an environment evaluation vector. This vector typically includes a number of operating variables of the machine room environment, such as wet bulb temperature, wet bulb humidity, outdoor temperature, etc. For example, assume that second operation data to be processed and room environment data of a room temperature control system are acquired. The second operation data comprise the frequency of the water pump, the output of the pressure sensor and the output of the flow sensor, and the machine room environment data comprise wet bulb temperature, wet bulb humidity and outdoor temperature. Converting these data into vector format requires normalization or normalization to ensure vector uniformity. Wherein the initial operating vector may include a vector formed by the frequency of the water pump, the output of the pressure sensor, and the output of the flow sensor. The environmental evaluation vector may include a vector formed by the wet bulb temperature, the wet bulb humidity, and the outdoor temperature. Generally, the initial operation vector refers to the operation characteristics of the model in the initial state, the environment evaluation vector represents the evaluation characteristics of the external environment, and the target operation vector represents the desired final operation characteristics. And combining the initial operation vector and the environment evaluation vector by a vector splicing method. Specifically, when vector stitching is performed, the dimensions of the initial running vector and the environment evaluation vector are unified, and then the two vectors are stitched together. In this way a new vector is obtained which contains all the information of the initial run vector and the environment evaluation vector. Finally, the new vector is used to derive the target run vector. In general, this new vector may be trained by a neural network or the like model to obtain the target operational vector. Specifically, the new vector may be input into a neural network and then trained in a back-propagation manner until a stable and efficient target operational vector is obtained.

In a specific embodiment, the process of executing step S105 may specifically include the following steps:

(1) Inputting the target operation vector into a target parameter regulation model, wherein the target parameter regulation model comprises: a first layer of decision base model, a middle layer of decision base model and a second layer of decision base model;

(2) Performing feature extraction and predictive analysis on the target operation vector through the first layer decision base model to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value;

(3) And inputting the tower air quantity parameter predicted value and the cooling water quantity parameter predicted value into a second layer of decision element model through the middle layer for parameter verification to obtain the tower air quantity regulation parameter and the cooling water quantity regulation parameter.

Specifically, the target parameter regulation model generally comprises three levels, namely a first-level decision base model, a middle-level decision element model and a second-level decision element model. The first layer decision base model is responsible for carrying out feature extraction and predictive analysis on a target operation vector to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value; the intermediate layer integrates the predicted values to obtain a group of comprehensive predicted values; and the second layer of decision meta model converts the comprehensive predicted value into an actual regulation parameter and outputs the actual regulation parameter to the control system. Next, the target operation vector is input into the first layer decision base model for feature extraction and predictive analysis. In particular, the target operating vector may be considered as a multi-dimensional vector, where each dimension represents a characteristic, such as temperature, humidity, flow, etc. Then, the vector can be trained through a neural network and other models, so that predicted values corresponding to the tower air quantity parameter and the cooling water quantity parameter are obtained. When the first layer decision base model is used for feature extraction and predictive analysis, a neural network and other models are used for extracting and screening each feature in the target operation vector to obtain a new vector which contains features related to tower air quantity parameters and cooling water quantity parameters. The extracted characteristics are converted into corresponding predicted values, for example, the temperature characteristics are converted into corresponding tower air quantity parameter predicted values, and the flow characteristics are converted into corresponding cooling water quantity parameter predicted values. And analyzing and predicting the vector after the feature transformation through a neural network and other models to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value. Obtaining a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value; finally, the tower air quantity parameter predicted value and the cooling water quantity parameter predicted value can be obtained through feature extraction and predictive analysis of the first layer of decision base model. The predicted values are used as the input of the middle layer, a group of comprehensive predicted values are obtained after integration, and then the predicted values are converted into actual regulation and control parameters through a second layer of decision element model and are output to a control system. The middle layer is generally a neural network or other models, after the first layer decision base model obtains the tower air volume parameter predicted value and the cooling water volume parameter predicted value, the predicted values are integrated to obtain a group of comprehensive predicted values, and then the predicted values are input into the second layer decision element model for parameter verification. And then, transmitting the tower air quantity parameter predicted value and the cooling water quantity parameter predicted value obtained by the first layer decision base model to the middle layer for integration. Specifically, these predictions are considered as inputs to the middle layer, which are then trained and integrated using a neural network or other model to obtain a set of comprehensive predictions. And finally, inputting the comprehensive predicted value obtained by the middle layer into a second layer of decision element model for parameter verification, thereby obtaining the tower air quantity regulation and control parameters and the cooling water quantity regulation and control parameters. Specifically, the main function of the second layer decision element model is to analyze and check the input of the intermediate layer to determine the final tower air volume regulation parameters and cooling water volume regulation parameters. When verification is carried out, parameters such as a weighting coefficient and the like are set according to actual conditions, so that the weights of predicted values of different dimensions are determined, and the final regulation and control parameters are determined. For example, assume that a water pump control system is designed to automatically control the flow and pressure of a water pump. And taking the tower air quantity parameter predicted value and the cooling water quantity parameter predicted value as the input of the middle layer, and integrating the tower air quantity parameter predicted value and the cooling water quantity parameter predicted value through a neural network to obtain a group of comprehensive predicted values. And then, inputting the comprehensive predicted value into a second layer of decision meta-model for verification, and determining final water pump flow and pressure regulation parameters according to weights of the predicted values of different dimensions.

In a specific embodiment, the process of executing step S106 may specifically include the following steps:

(1) Determining the current tower air quantity and the current cooling water quantity of the machine room temperature control system according to the second operation data;

(2) Generating a first regulation scheme according to the tower air volume regulation parameter and the current tower air volume, and generating a second regulation scheme according to the cooling water volume regulation parameter and the current cooling water volume;

(3) Generating a parameter regulation scheme of the machine room temperature control system according to the first regulation scheme and the second regulation scheme;

(4) Performing optimal temperature monitoring on the machine room temperature control system according to the parameter regulation and control scheme to obtain an optimal temperature point;

(5) And obtaining the minimum total power corresponding to the temperature control system of the machine room according to the optimal temperature point.

Specifically, the current tower air volume and cooling water volume are determined according to the second operation data. Specifically, parameters such as temperature, humidity, pressure and the like in the machine room are monitored by devices such as sensors and the like, and the parameters are converted into predicted values of tower air quantity and cooling water quantity. Then, the current tower air quantity and cooling water quantity are determined according to the predicted values so as to facilitate subsequent parameter regulation and temperature monitoring. Next, a set of regulation schemes is generated based on the tower air volume regulation parameters and the cooling water volume regulation parameters. Specifically, a mathematical model or other method is utilized to generate a first regulation scheme and a second regulation scheme according to the current tower air volume and cooling water volume, and the tower air volume regulation parameter and cooling water volume regulation parameter. These regulation schemes can be used to control the temperature in the machine room to ensure that the temperature in the machine room is always within a suitable range. Next, a set of regulation schemes is generated based on the tower air volume regulation parameters and the cooling water volume regulation parameters. Specifically, a mathematical model or other method is utilized to generate a first regulation scheme and a second regulation scheme according to the current tower air volume and cooling water volume, and the tower air volume regulation parameter and cooling water volume regulation parameter. These regulation schemes can be used to control the temperature in the machine room to ensure that the temperature in the machine room is always within a suitable range. And then, carrying out optimal temperature monitoring on the machine room temperature control system. Specifically, the temperature in the machine room is monitored by using a sensor or other equipment, and the temperature data are input into a machine room temperature control system for analysis and processing. And analyzing the data to obtain the temperature change trend in the machine room, the relation between the temperature change trend and the regulation parameters, and the regulation parameters corresponding to the optimal temperature point. And finally, acquiring the minimum total power corresponding to the temperature control system of the machine room according to the optimal temperature point. Specifically, the optimal modulation scheme is determined by calculating the energy consumption required for the different modulation schemes and comparing the energy consumption magnitudes of these schemes. Then, according to the optimal regulation scheme, the total power required by the machine room temperature control system is calculated, and the minimum total power corresponding to the machine room temperature control system at the moment is obtained.

The method for regulating parameters of an artificial intelligence-based data center room in the embodiment of the present invention is described above, and the device for regulating parameters of an artificial intelligence-based data center room in the embodiment of the present invention is described below, referring to fig. 5, and one embodiment of the device for regulating parameters of an artificial intelligence-based data center room in the embodiment of the present invention includes:

the acquiring module 501 is configured to acquire first operation data of a machine room temperature control system, and classify attributes of the first operation data to obtain wind system operation data and water system operation data;

the training module 502 is configured to input the wind system operation data into a preset plurality of first training models to perform model training to obtain a first model set, and input the water system operation data into a preset plurality of second training models to perform model training to obtain a second model set;

a fusion module 503, configured to take the first model set and the second model set as a first layer decision base model, and perform model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model;

the conversion module 504 is configured to obtain second operation data to be processed and machine room environment data, and perform vector conversion on the second operation data and the machine room environment data to obtain a target operation vector;

The analysis module 505 is configured to input the target operation vector into the target parameter regulation model to perform machine room parameter regulation analysis, so as to obtain a tower air volume regulation parameter and a cooling water volume regulation parameter;

and the regulation and control module 506 is configured to generate a parameter regulation and control scheme of the machine room temperature control system according to the tower air volume regulation and control parameter and the cooling water volume regulation and control parameter, and monitor the optimal temperature and the lowest power of the machine room temperature control system according to the parameter regulation and control scheme, so as to obtain an optimal temperature point and the lowest total power of the machine room temperature control system.

Through the cooperative cooperation of the components, the wind system operation data is input into a plurality of first training models to perform model training to obtain a first model set, and the water system operation data is input into a plurality of preset second training models to perform model training to obtain a second model set; taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and the second layer decision element model to obtain a target parameter regulation model; vector conversion is carried out on the second operation data and the machine room environment data to obtain a target operation vector; the invention realizes real-time monitoring and accurate regulation of parameters such as temperature, humidity, energy consumption and the like of a machine room, thereby improving the energy utilization efficiency and energy saving effect of a data center, reducing the operation cost of the data center, adopting an AI intelligent decision technology, automatically adjusting the operation parameters of a machine room temperature control system, reducing the energy consumption of the temperature control system, reducing the PUE value of the data center, improving the operation efficiency of the machine room, utilizing a multi-model fusion technology, more accurately predicting the environment and energy consumption condition of the machine room, improving the accuracy of data analysis and regulation, further adopting an optimization control strategy, finding the optimal temperature point and the minimum total power of the temperature control system of the machine room, improving the regulation effect of the machine room, realizing the intelligent parameter regulation of the machine room of the data center, and improving the accuracy of parameter regulation.

Fig. 5 above describes the parameter control device of the data center room based on artificial intelligence in the embodiment of the present invention in detail from the perspective of a modularized functional entity, and the parameter control device of the data center room based on artificial intelligence in the embodiment of the present invention is described in detail from the perspective of hardware processing.

Fig. 6 is a schematic structural diagram of an artificial intelligence-based data center room parameter regulating apparatus 600 according to an embodiment of the present invention, where the artificial intelligence-based data center room parameter regulating apparatus 600 may have relatively large differences according to different configurations or performances, and may include one or more processors (central processing units, CPU) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) storing applications 633 or data 632. Wherein the memory 620 and the storage medium 630 may be transitory or persistent storage. The program stored on the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations on the artificial intelligence based data center room parameter tuning device 600. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the artificial intelligence based data center room parameter regulating apparatus 600.

The artificial intelligence based data center room parameter tuning device 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input/output interfaces 660, and/or one or more operating systems 631, such as Windows Serve, mac OS X, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the artificial intelligence based data center room parameter regulating apparatus structure illustrated in fig. 6 is not limiting and may include more or fewer components than illustrated, or may combine certain components, or a different arrangement of components.

The invention also provides an artificial intelligence-based data center room parameter regulating device, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the artificial intelligence-based data center room parameter regulating method in the above embodiments.

The invention also provides a computer readable storage medium, which can be a nonvolatile computer readable storage medium, and can also be a volatile computer readable storage medium, wherein the computer readable storage medium stores instructions, and when the instructions run on a computer, the instructions cause the computer to execute the steps of the data center room parameter regulation method based on artificial intelligence.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random acceS memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will 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 technical solutions of the embodiments of the present invention.

Claims

1. The data center machine room parameter regulation and control method based on the artificial intelligence is characterized by comprising the following steps of:

taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model, wherein the method specifically comprises the following steps of: model merging is carried out on the first model set and the second model set, and the first model set and the second model set are used as a first layer decision base model; through a preset middle layer, carrying out model connection on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model; the second layer decision meta-model is a neural network, and parameters and weights of the second layer decision meta-model are adjusted by using training data and an optimization method;

inputting the target operation vector into the target parameter regulation model for carrying out machine room parameter regulation analysis to obtain a tower air quantity regulation parameter and a cooling water quantity regulation parameter, wherein the method specifically comprises the following steps of: inputting the target operation vector into the target parameter regulation model, wherein the target parameter regulation model comprises: a first layer of decision base model, a middle layer of decision base model and a second layer of decision base model; performing feature extraction and predictive analysis on the target operation vector through the first layer decision base model to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value; inputting the tower air volume parameter predicted value and the cooling water volume parameter predicted value into the second layer decision element model through the middle layer for parameter verification to obtain a tower air volume regulation parameter and a cooling water volume regulation parameter;

2. The method for regulating parameters of a data center room based on artificial intelligence according to claim 1, wherein the steps of obtaining first operation data of a room temperature control system, classifying attributes of the first operation data, and obtaining wind system operation data and water system operation data comprise:

3. The method for controlling parameters of a data center room based on artificial intelligence according to claim 2, wherein the inputting the wind system operation data into a preset plurality of first training models to perform model training to obtain a first model set, and inputting the water system operation data into a preset plurality of second training models to perform model training to obtain a second model set, includes:

4. The method for adjusting parameters of a data center machine room based on artificial intelligence according to claim 1, wherein the steps of obtaining second operation data to be processed and machine room environment data, and performing vector conversion on the second operation data and the machine room environment data to obtain a target operation vector include:

5. The method for regulating parameters of a data center room based on artificial intelligence according to claim 1, wherein the generating a parameter regulating scheme of the room temperature control system according to the tower air volume regulating parameter and the cooling water volume regulating parameter, and monitoring the best temperature and the lowest power of the room temperature control system according to the parameter regulating scheme, to obtain the best temperature point and the lowest total power of the room temperature control system, comprises:

6. Data center computer lab parameter regulation and control device based on artificial intelligence, its characterized in that, data center computer lab parameter regulation and control device based on artificial intelligence includes:

the fusion module is used for taking the first model set and the second model set as a first layer decision base model, and carrying out model fusion on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model, and specifically comprises the following steps: model merging is carried out on the first model set and the second model set, and the first model set and the second model set are used as a first layer decision base model; through a preset middle layer, carrying out model connection on the first layer decision base model and a preset second layer decision element model to obtain a target parameter regulation model; the second layer decision meta-model is a neural network, and parameters and weights of the second layer decision meta-model are adjusted by using training data and an optimization method;

the analysis module is used for inputting the target operation vector into the target parameter regulation model to carry out machine room parameter regulation analysis, so as to obtain a tower air quantity regulation parameter and a cooling water quantity regulation parameter, and specifically comprises the following steps: inputting the target operation vector into the target parameter regulation model, wherein the target parameter regulation model comprises: a first layer of decision base model, a middle layer of decision base model and a second layer of decision base model; performing feature extraction and predictive analysis on the target operation vector through the first layer decision base model to obtain a tower air quantity parameter predicted value and a cooling water quantity parameter predicted value; inputting the tower air volume parameter predicted value and the cooling water volume parameter predicted value into the second layer decision element model through the middle layer for parameter verification to obtain a tower air volume regulation parameter and a cooling water volume regulation parameter;

7. Data center computer lab parameter regulation and control equipment based on artificial intelligence, its characterized in that, data center computer lab parameter regulation and control equipment based on artificial intelligence includes: a memory and at least one processor, the memory having instructions stored therein;

the at least one processor invoking the instructions in the memory to cause the artificial intelligence based data center room parameter tuning apparatus to perform the artificial intelligence based data center room parameter tuning method of any one of claims 1-5.

8. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the artificial intelligence based data center room parameter tuning method of any one of claims 1-5.