Energy-saving control method and system for air-cooled air-conditioning type computer room

Abstract

The present invention relates to an energy-saving control method and system for an air-cooled air-conditioning type computer room, and belongs to the technical field of computer room energy-saving control. The method comprises: collecting basic environmental data of an air-cooled air-conditioning computer room, fusing the air-conditioning operation status data and air-conditioning power data therein into structural data; constructing a training set and an input sequence according to the structural data, establishing a resource allocation model and an energy consumption loss objective function, iteratively training the resource allocation model, and updating the resource allocation model according to the training results to obtain a control quantity output sequence; setting a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, and fuzzy controlling the refrigeration system according to the temperature environment parameters and the steady-state temperature target value to obtain a fuzzy control quantity; converting the control quantity output sequence and the fuzzy control quantity into a control signal to perform real-time control on the air-cooled air conditioner. Dynamic matching of the refrigeration requirements of the equipment in the computer room is achieved, extending the service life of the unit, and reducing energy consumption.

Description

Technical Field

[0001] The invention belongs to the technical field of energy-saving control of computer rooms, and in particular relates to an energy-saving control method and system for an air-cooled air-conditioning type computer room. Background Art

[0002] In the existing air-cooled air-conditioning room, due to the low efficiency of equipment operation and backward management methods, energy waste is serious, which is mainly reflected in the air-conditioning energy consumption: the air conditioners in the same room operate independently, and are generally all set to the on-state, which consumes a lot of power; the air conditioner is set to too low a temperature, lasts too long, and is not adjusted or shut down in time after reaching the temperature, etc., which will increase the load of the air conditioner and lead to increased power consumption; for the problem of high energy consumption and waste of air conditioners, the existing technology generally adopts the method of adding energy-saving control cabinets to achieve energy-saving transformation, but this technology will have certain pain points and difficulties, which are mainly reflected in the following aspects: each air conditioner needs It is necessary to add an energy-saving control cabinet separately, and each energy-saving control cabinet requires a space to be placed. However, the design layout of an air-cooled air-conditioning room is generally compact, and the energy-saving control cabinet may not be placed next to the air conditioner; the energy-saving control cabinet requires relevant control modifications to the air conditioner, which may involve modifications and adjustments to the air-conditioning equipment, involving additional costs such as cooperation from the air-conditioning manufacturer; different brands of air-conditioning manufacturers may have different support for equipment modifications, and risks are prone to occur after the modification; since the air conditioner needs to be modified or new pipelines and other related engineering means are needed, and this type of room is in the production link, all aspects need to coordinate the construction time, and the overall modification cycle is relatively long. Summary of the invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides an energy-saving control method and system for an air-cooled air-conditioning type computer room;

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] S1: collecting temperature environment parameters of the air-cooled air-conditioning room, air inlet and outlet temperatures of the load cabinet, air-conditioning operation status data, and air-conditioning power data, and integrating the air-conditioning operation status data and the air-conditioning power data into structure data;

[0006] S2: construct a training set and an input sequence according to the structural data; establish a resource configuration model and an energy consumption loss objective function, and calculate the upper and lower bounds of the number of air conditioners participating in the switch control according to the training set and the energy consumption loss objective function; set a training integrated network according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations according to the target loss value; take the lower bound of the number of air conditioners participating in the air conditioner control switch as the initial quantity, and calculate the corresponding resource configuration overhead by accumulation; iterate the resource configuration model according to the number of training iterations to obtain the optimal resource configuration plan; update the representation parameters of the resource configuration model according to the training results, and obtain the control quantity output sequence through the trained resource configuration model and the input sequence;

[0007] S3: setting a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, selecting an input quantity and a transformation coefficient of a fuzzy controller according to the temperature environment parameter, setting a fuzzy domain according to the input quantity, converting the input quantity to the fuzzy domain through the transformation coefficient, tracking the input quantity and calculating the fuzzy error and error change rate corresponding to the input quantity and the output control quantity, and calculating a membership function according to the fuzzy error and the error change rate; establishing a fuzzy rule base according to expert experience, constructing a fuzzy controller through the fuzzy rule base and the membership function, obtaining a fuzzy quantity through the fuzzy controller, and performing a clarification process on the fuzzy quantity to obtain a fuzzy control quantity;

[0008] S4: Initialize the controller of the environmental control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the machine room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, adjust the temperature of the terminal air conditioner through the 485 communication interface according to the fuzzy control quantity.

[0009] Specifically, the temperature environment parameters are collected by sensors and transmitted to the controller of the environmental control box. The sensor adopts a single bus data format, and the bidirectional transmission of input and output is completed by a single data pin port. The single complete data transmission is 24 bytes; the air-conditioning operation status data is transmitted to the controller of the environmental control box by the serial communication module of the terminal air-conditioning equipment.

[0010] As a preferred technical solution of the present invention, the resource allocation model is:

[0011] A resource allocation model is constructed based on the structural data and a target loss value is preset. The calculation formula is:

[0012]

[0013] Where C is the resource configuration overhead, is the working power of the air conditioner of type t in the running state, is the working power of the air conditioner of type t when not in operation, n k is the number of air conditioners in operation, n s is the number of air conditioners not in operation, t i is the processing time of iterative training, and count is the number of iterations;

[0014] A resource allocation constraint is established according to the resource allocation model and the target loss value, wherein: T is the total time of iterative training, f loss (count,n k )=l,f loss is the minimum loss function, l is the target loss value, r is the maximum resource allocation ratio under non-interference conditions;

[0015] The upper and lower bounds of the number of air conditioners participating in the air conditioner control switch are calculated according to the resource configuration constraint and the resource configuration model, and the calculation formula is:

[0016]

[0017] in, is the upper bound of the number of air conditioners involved in switch control, is the lower bound of the number of air conditioners involved in switch control, w i is the number of switch controls in a single iteration, b 0 , b 1 is the model coefficient, c k is the resource allocation cost of the air conditioner involved in switch control, T is the total time of iterative training, l is the target loss value, u is the resource allocation ratio generated in the previous iteration, and n s is the number of air conditioners not in operation, b s is the total amount of air conditioning resources, and g is the model training parameter.

[0018] Specifically, the model training method is:

[0019] Setting a training integrated network according to the resource configuration model, wherein the input of the training integrated network is a target loss value, a minimum loss function, and a total time of iterative training;

[0020] Initialize the optimal resource configuration overhead, the number and type of air conditioners controlled by switches; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations using the target loss value;

[0021] In a single iteration process, the instance parameters of the air conditioner are obtained, and the maximum resource configuration ratio is calculated to obtain the upper and lower bounds of the number of air conditioners participating in the air conditioner control switch; the lower bound of the number of air conditioners participating in the air conditioner control switch is used as the initial amount, and the corresponding resource configuration overhead is calculated by accumulation. If the resource configuration overhead is less than or equal to the optimal resource configuration overhead, the resource configuration overhead is used as the optimal resource configuration overhead and the corresponding resource configuration scheme is recorded until the accumulated amount exceeds the upper bound of the number of air conditioners participating in the air conditioner control switch, and the single iteration process is terminated.

[0022] An air-cooled air-conditioning type computer room energy-saving control system, including: a data acquisition module, an AI algorithm module, a fuzzy control module, and a control output module:

[0023] The data acquisition module is used to collect the temperature environment parameters of the air-cooled air-conditioning room, the air inlet and outlet temperatures of the load cabinet, the air-conditioning operation status data, and the air-conditioning power data, and merge the air-conditioning operation status data and the air-conditioning power data into structure data;

[0024] The AI ​​algorithm module is used to construct a training set and an input sequence according to the structural data; establish a resource configuration model and an energy consumption loss objective function, and calculate the upper and lower bounds of the number of air conditioners participating in the switch control according to the training set and the energy consumption loss objective function; set a training integrated network according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations according to the target loss value; take the lower bound of the number of air conditioners participating in the air conditioner control switch as the initial quantity, and calculate the corresponding resource configuration overhead by accumulation; iterate the resource configuration model according to the number of training iterations to obtain the optimal resource configuration plan; update the representation parameters of the resource configuration model according to the training results, and obtain the control quantity output sequence through the trained resource configuration model and the input sequence;

[0025] The fuzzy control module is used to set a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, select the input quantity and transformation coefficient of the fuzzy controller through the temperature environment parameters, set a fuzzy domain according to the input quantity, convert the input quantity to the fuzzy domain through the transformation coefficient, track the input quantity and calculate the fuzzy error and error change rate corresponding to the input quantity and the output control quantity, and calculate the membership function according to the fuzzy error and error change rate; establish a fuzzy rule base according to expert experience, construct a fuzzy controller through the fuzzy rule base and the membership function, obtain a fuzzy quantity through the fuzzy controller, and perform clarification processing on the fuzzy quantity to obtain a fuzzy control quantity;

[0026] The control output module is used to initialize the controller of the environmental control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the computer room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, the temperature of the terminal air conditioner is adjusted through the 485 communication interface according to the fuzzy control quantity.

[0027] The beneficial effects of the present invention are:

[0028] (1) By setting up AI algorithm modules, energy-saving transformation can be completed without engineering measures or with minimal impact on all aspects; group control of air conditioners can be realized to achieve functions such as master-slave, cascading, and patrol, avoid competitive operation, and reduce energy consumption; dynamically adjust the unit operation time according to actual needs, reduce unnecessary operation time of redundant units, and put redundant units in the group into a dormant backup state to reduce energy consumption; intelligently control the operation status of each air-conditioning unit in the group in real time according to the temperature and humidity conditions of the computer room and cabinet, IT load, etc., reduce the risks caused by human intervention, and reduce energy consumption;

[0029] (2) By connecting multiple cabinet temperature sensors, the air inlet temperature of the main equipment area is actively detected, and the air conditioning in the corresponding area is linked to the room to perform real-time optimization, dynamically matching the cooling needs of the equipment in the room to achieve energy saving; it can effectively balance the operating time of each air-conditioning unit, avoid long-term operation of a single air-conditioning unit, and avoid frequent start and stop of the compressor, thereby extending the service life of the unit and reducing energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0031] Figure 1 It is a flow chart of an energy-saving control method for an air-cooled air-conditioning type computer room in the present invention;

[0032] Figure 2 The present invention is a structural block diagram of an air-cooled air-conditioning type computer room energy-saving control system. DETAILED DESCRIPTION

[0033] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0034] See also Figure 1-2 , an energy-saving control method and system for an air-cooled air-conditioning type computer room;

[0035] S1: collecting temperature environment parameters of the air-cooled air-conditioning room, air inlet and outlet temperatures of the load cabinet, air-conditioning operation status data, and air-conditioning power data, and integrating the air-conditioning operation status data and the air-conditioning power data into structure data;

[0036] S2: construct a training set and an input sequence according to the structural data; establish a resource configuration model and an energy consumption loss objective function, and calculate the upper and lower bounds of the number of air conditioners participating in the switch control according to the training set and the energy consumption loss objective function; set a training integrated network according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations according to the target loss value; take the lower bound of the number of air conditioners participating in the air conditioner control switch as the initial quantity, and calculate the corresponding resource configuration overhead by accumulation; iterate the resource configuration model according to the number of training iterations to obtain the optimal resource configuration plan; update the representation parameters of the resource configuration model according to the training results, and obtain the control quantity output sequence through the trained resource configuration model and the input sequence;

[0037] S3: Set a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, select the input and transformation coefficient of the fuzzy controller through the temperature environment parameters, set a fuzzy domain according to the input, convert the input to the fuzzy domain through the transformation coefficient, track the input and calculate the fuzzy error and error change rate corresponding to the input and output control quantity, and calculate the membership function according to the fuzzy error and error change rate; establish a fuzzy rule base according to expert experience, construct a fuzzy controller through the fuzzy rule base and the membership function, obtain a fuzzy quantity through the fuzzy controller, and clarify the fuzzy quantity to obtain a fuzzy control quantity.

[0038] S4: Initialize the controller of the environment control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the machine room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, adjust the temperature of the terminal air conditioner through the 485 communication interface according to the fuzzy control quantity;

[0039] Specifically, the temperature environment parameters are collected by sensors and transmitted to the controller of the environmental control box. The sensor adopts a single bus data format, and the bidirectional transmission of input and output is completed by a single data pin port. The single complete data transmission is 24 bytes; the air-conditioning operation status data is transmitted to the controller of the environmental control box by the serial communication module of the terminal air-conditioning equipment.

[0040] In this embodiment, the environmental parameters, load conditions and air conditioning operation status data of the air-cooled air-conditioning room and cabinet are collected through the data acquisition module, and then the relevant data are input into the AI ​​algorithm module. After training, the AI ​​algorithm finds the air-cooled air-conditioning setting scheme with the best energy consumption, automatically adjusts the operation parameters and start and stop of the air-cooled air-conditioning, and finally outputs the control signal to the air-cooled air-conditioning through the control output module, which increases the real-time monitoring of the environmental parameters and equipment operation status data of the air-cooled air-conditioning room. When an abnormal situation occurs, the AI ​​algorithm module can timely discover and take corresponding measures to deal with it; during operation, when the running air-conditioning equipment fails, it can automatically compensate for the standby air-conditioning equipment, and give priority to the standby air-conditioning equipment with a short running time. Different models of energy-saving algorithm boxes are selected according to the room management requirements, different room layouts, different budgets, etc. In the single-device start-stop mode, each air conditioner can be started and stopped separately. The target return air temperature can be manually set for each air conditioner.

[0041] As a preferred technical solution of the present invention, the resource allocation model is:

[0042] A resource allocation model is constructed based on the structural data and a target loss value is preset. The calculation formula is:

[0043]

[0044] Where C is the resource configuration overhead, is the working power of the air conditioner of type t in the running state, is the working power of the air conditioner of type t when not in operation, n k is the number of air conditioners in operation, n s is the number of air conditioners not in operation, t iis the processing time of iterative training, and count is the number of iterations;

[0045] A resource allocation constraint is established according to the resource allocation model and the target loss value, wherein: T is the total time of iterative training, f loss (count,n k )=l,f loss is the minimum loss function, l is the target loss value, r is the maximum resource allocation ratio under non-interference conditions;

[0046] The upper and lower bounds of the number of air conditioners participating in the air conditioner control switch are calculated according to the resource configuration constraint and the resource configuration model, and the calculation formula is:

[0047]

[0048] in, is the upper bound of the number of air conditioners involved in switch control, is the lower bound of the number of air conditioners involved in switch control, w i is the number of switch controls in a single iteration, b 0 , b 1 is the model coefficient, c k is the resource allocation cost of the air conditioner involved in switch control, T is the total time of iterative training, l is the target loss value, u is the resource allocation ratio generated in the previous iteration, and n s is the number of air conditioners not in operation, b s is the total amount of air conditioning resources, and g is the model training parameter.

[0049] In this embodiment, the air conditioners with the shortest running time and normal status are started first. The air conditioners in the faulty state are deleted from the control group. After the system is in a steady state, the statistical time is counted and the standby air conditioners are started first. The air conditioner with the shortest cumulative running time is started first, and then the air conditioner with the longest cumulative running time is stopped. The target return air temperature of the air conditioner is intelligently adjusted through the AI ​​algorithm. When the average temperature of the computer room is low, the target return air temperature of the air conditioner can be increased; when the temperature of the computer room is high, the target return air temperature of the air conditioner can be reduced; the upper and lower limits of the target return air temperature are set.

[0050] Specifically, the model training method is:

[0051] Setting a training integrated network according to the resource configuration model, wherein the input of the training integrated network is a target loss value, a minimum loss function, and a total time of iterative training;

[0052] Initialize the optimal resource configuration overhead, the number and type of air conditioners controlled by switches; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations using the target loss value;

[0053] In a single iteration process, the instance parameters of the air conditioner are obtained, and the maximum resource configuration ratio is calculated to obtain the upper and lower bounds of the number of air conditioners participating in the air conditioner control switch; the lower bound of the number of air conditioners participating in the air conditioner control switch is used as the initial amount, and the corresponding resource configuration overhead is calculated by accumulation. If the resource configuration overhead is less than or equal to the optimal resource configuration overhead, the resource configuration overhead is used as the optimal resource configuration overhead and the corresponding resource configuration scheme is recorded until the accumulated amount exceeds the upper bound of the number of air conditioners participating in the air conditioner control switch, and the single iteration process is terminated.

[0054] An air-cooled air-conditioning type computer room energy-saving control system, including: a data acquisition module, an AI algorithm module, a fuzzy control module, and a control output module:

[0055] The data acquisition module is used to collect the temperature environment parameters of the air-cooled air-conditioning room, the air inlet and outlet temperatures of the load cabinet, the air-conditioning operation status data, and the air-conditioning power data, and merge the air-conditioning operation status data and the air-conditioning power data into structure data;

[0056] The AI ​​algorithm module is used to construct a training set and an input sequence according to the structural data; establish a resource configuration model and an energy consumption loss objective function, and calculate the upper and lower bounds of the number of air conditioners participating in the switch control according to the training set and the energy consumption loss objective function; set a training integrated network according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations according to the target loss value; take the lower bound of the number of air conditioners participating in the air conditioner control switch as the initial quantity, and calculate the corresponding resource configuration overhead by accumulation; iterate the resource configuration model according to the number of training iterations to obtain the optimal resource configuration plan; update the representation parameters of the resource configuration model according to the training results, and obtain the control quantity output sequence through the trained resource configuration model and the input sequence;

[0057] The fuzzy control module is used to set a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, select the input quantity and transformation coefficient of the fuzzy controller through the temperature environment parameters, set a fuzzy domain according to the input quantity, convert the input quantity to the fuzzy domain through the transformation coefficient, track the input quantity and calculate the fuzzy error and error change rate corresponding to the input quantity and the output control quantity, and calculate the membership function according to the fuzzy error and error change rate; establish a fuzzy rule base according to expert experience, construct a fuzzy controller through the fuzzy rule base and the membership function, obtain a fuzzy quantity through the fuzzy controller, and perform clarification processing on the fuzzy quantity to obtain a fuzzy control quantity;

[0058] The control output module is used to initialize the controller of the environmental control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the computer room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, the temperature of the terminal air conditioner is adjusted through the 485 communication interface according to the fuzzy control quantity.

[0059] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A method for energy saving control of an air-cooled air-conditioning type computer room, characterized in that: include: S1: collecting temperature environment parameters of the air-cooled air-conditioning room, air inlet and outlet temperatures of the load cabinet, air-conditioning operation status data, and air-conditioning power data, and integrating the air-conditioning operation status data and the air-conditioning power data into structure data; S2: constructing a training set and an input sequence according to the structural data; establishing a resource allocation model and an energy consumption loss objective function, wherein the resource allocation model construction method is: A resource allocation model is constructed based on the structural data and a target loss value is preset. The calculation formula is: Where C is the resource configuration overhead, is the working power of the air conditioner of type t in the running state, is the working power of the air conditioner of type t when not in operation, n k is the number of air conditioners in operation, n s is the number of air conditioners not in operation, t i is the processing time of iterative training, and count is the number of iterations; A resource allocation constraint is established according to the resource allocation model and the target loss value, wherein: T is the total time of iterative training, f loss (count,n k )=l,f loss is the minimum loss function, l is the target loss value, r is the maximum resource allocation ratio under non-interference conditions; The upper and lower bounds of the number of air conditioners participating in the air conditioner control switch are calculated according to the resource configuration constraint and the resource configuration model, and the calculation formula is: in, is the upper bound of the number of air conditioners involved in switch control, is the lower bound of the number of air conditioners involved in switch control, w i is the number of switch controls in a single iteration, b0 and b1 are model coefficients, and c k is the resource allocation cost of the air conditioner involved in switch control, T is the total time of iterative training, l is the target loss value, u is the resource allocation ratio generated in the previous iteration, and n s is the number of air conditioners not in operation, b s is the total amount of air conditioning resources, g is the model training parameter; The upper and lower bounds of the number of air conditioners participating in the switch control are calculated according to the training set and the energy consumption loss objective function; a training integrated network is set according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; the resource configuration model is pre-run in the training integrated network to obtain model training parameters, and the number of training iterations is calculated according to the target loss value; the lower bound of the number of air conditioners participating in the air conditioner control switch is used as the initial quantity, and the corresponding resource configuration overhead is calculated by accumulation; the resource configuration model is iterated repeatedly according to the number of training iterations to obtain the optimal resource configuration plan; the representation parameters of the resource configuration model are updated according to the training results, and the control quantity output sequence is obtained through the trained resource configuration model and the input sequence; S3: setting a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, selecting an input quantity and a transformation coefficient of a fuzzy controller according to the temperature environment parameter, setting a fuzzy domain according to the input quantity, converting the input quantity to the fuzzy domain through the transformation coefficient, tracking the input quantity and calculating the fuzzy error and error change rate corresponding to the input quantity and the output control quantity, and calculating a membership function according to the fuzzy error and the error change rate; establishing a fuzzy rule base according to expert experience, constructing a fuzzy controller through the fuzzy rule base and the membership function, obtaining a fuzzy quantity through the fuzzy controller, and performing a clarification process on the fuzzy quantity to obtain a fuzzy control quantity; S4: Initialize the controller of the environmental control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the machine room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, adjust the temperature of the terminal air conditioner through the 485 communication interface according to the fuzzy control quantity.

2. The energy-saving control method for an air-cooled air-conditioning type computer room according to claim 1, characterized in that: The temperature environment parameters are collected by sensors and transmitted to the controller of the environmental control box. The sensor adopts a single bus data format, and the bidirectional transmission of input and output is completed by a single data pin port. The single complete data transmission is 24 bytes; the air-conditioning operation status data is transmitted to the controller of the environmental control box by the serial port communication module of the terminal air-conditioning equipment.

3. The energy-saving control method for an air-cooled air-conditioning type computer room according to claim 1, characterized in that: The model training method is: Setting a training integrated network according to the resource configuration model, wherein the input of the training integrated network is a target loss value, a minimum loss function, and a total time of iterative training; Initialize the optimal resource configuration overhead, the number and type of air conditioners controlled by switches; pre-run the resource configuration model in the training integrated network to obtain model training parameters, and calculate the number of training iterations using the target loss value; In a single iteration, the instance parameters of the air conditioner are obtained, and the maximum resource allocation ratio is calculated to obtain the upper and lower bounds of the number of air conditioners participating in the air conditioner control switch; Taking the lower bound of the number of air conditioners participating in the air-conditioning control switch as the initial quantity, the corresponding resource configuration overhead is calculated by accumulation. If the resource configuration overhead is less than or equal to the optimal resource configuration overhead, the resource configuration overhead is used as the optimal resource configuration overhead and the corresponding resource configuration scheme is recorded until the accumulated amount exceeds the upper bound of the number of air conditioners participating in the air-conditioning control switch, and the single iteration process is terminated; the resource configuration model is iterated cyclically according to the number of training iterations to obtain the optimal resource configuration scheme, including air conditioner type, quantity and location parameters.

4. An air-cooled air-conditioning type computer room energy-saving control system, used to execute the air-cooled air-conditioning type computer room energy-saving control method as described in any one of claims 1 to 3, characterized in that: include: Data acquisition module, AI algorithm module, fuzzy control module, control output module: The data acquisition module is used to collect the temperature environment parameters of the air-cooled air-conditioning room, the air inlet and outlet temperatures of the load cabinet, the air-conditioning operation status data, and the air-conditioning power data, and merge the air-conditioning operation status data and the air-conditioning power data into structure data; The AI ​​algorithm module is used to construct a training set and an input sequence according to the structural data; establish a resource allocation model and an energy consumption loss objective function, and the resource allocation model construction method is: A resource allocation model is constructed based on the structural data and a target loss value is preset. The calculation formula is: Where C is the resource configuration overhead, is the working power of the air conditioner of type t in the running state, is the working power of the air conditioner of type t when not in operation, n k is the number of air conditioners in operation, n s is the number of air conditioners not in operation, t i is the processing time of iterative training, and count is the number of iterations; A resource allocation constraint is established according to the resource allocation model and the target loss value, wherein: T is the total time of iterative training, f loss (count,n k )=l,f loss is the minimum loss function, l is the target loss value, r is the maximum resource allocation ratio under non-interference conditions; The upper and lower bounds of the number of air conditioners participating in the air conditioner control switch are calculated according to the resource configuration constraint and the resource configuration model, and the calculation formula is: in, is the upper bound of the number of air conditioners involved in switch control, is the lower bound of the number of air conditioners involved in switch control, w i is the number of switch controls in a single iteration, b0 and b1 are model coefficients, and c k is the resource allocation cost of the air conditioner involved in switch control, T is the total time of iterative training, l is the target loss value, u is the resource allocation ratio generated in the previous iteration, and n s is the number of air conditioners not in operation, b s is the total amount of air conditioning resources, g is the model training parameter; The upper and lower bounds of the number of air conditioners participating in the switch control are calculated according to the training set and the energy consumption loss objective function; a training integrated network is set according to the resource configuration model, and the input of the training integrated network is the target loss value, the minimum loss function, and the total time of iterative training; the resource configuration model is pre-run in the training integrated network to obtain model training parameters, and the number of training iterations is calculated according to the target loss value; the lower bound of the number of air conditioners participating in the air conditioner control switch is used as the initial quantity, and the corresponding resource configuration overhead is calculated by accumulation; the resource configuration model is iterated repeatedly according to the number of training iterations to obtain the optimal resource configuration plan; the representation parameters of the resource configuration model are updated according to the training results, and the control quantity output sequence is obtained through the trained resource configuration model and the input sequence; The fuzzy control module is used to set a steady-state temperature target value according to the inlet and outlet temperatures of the load cabinet, select the input quantity and transformation coefficient of the fuzzy controller through the temperature environment parameters, set a fuzzy domain according to the input quantity, convert the input quantity to the fuzzy domain through the transformation coefficient, track the input quantity and calculate the fuzzy error and error change rate corresponding to the input quantity and the output control quantity, and calculate the membership function according to the fuzzy error and error change rate; establish a fuzzy rule base according to expert experience, construct a fuzzy controller through the fuzzy rule base and the membership function, obtain a fuzzy quantity through the fuzzy controller, and perform clarification processing on the fuzzy quantity to obtain a fuzzy control quantity; The control output module is used to initialize the controller of the environmental control box according to the fuzzy control quantity and the control quantity output sequence, obtain the configuration scheme and the number of start and stop units of the terminal air conditioner in the computer room through the control quantity output sequence, set the patrol time, and start and stop the corresponding terminal air conditioner according to the control quantity output sequence. If the absolute value of the difference between the inlet and outlet temperature of the load cabinet and the steady-state temperature target value is less than the preset temperature change threshold, the temperature of the terminal air conditioner is adjusted through the 485 communication interface according to the fuzzy control quantity.

Images