Machine room air conditioner energy saving method, device and equipment based on large language model and storage medium

Abstract

The invention discloses a machine room air conditioner energy saving method, device and equipment based on a large language model and a storage medium, and relates to the technical field of air conditioner energy saving. Constructing a prompt text according to the real-time environment state information of the machine room; the prompt text is input into a large language model, and an air conditioner control strategy set is obtained; the air conditioner control strategy set is verified through a prediction model, and a target strategy is obtained; and performing energy-saving control on the air conditioner in the machine room through the target strategy. According to the method, the input prompt text of the large language model is constructed, the air conditioner control strategy set is generated, the optimal strategy is screened out, the control strategy is iteratively optimized according to the execution result of the optimal strategy, and the universality, robustness and intelligent degree of machine room air conditioner control can be improved.

Description

Technical Field

[0001] This application relates to the technical field of air conditioner energy saving, and particularly to an energy saving method, device, equipment and storage medium for computer room air conditioners based on large language models. Background Art

[0002] With the rapid development of modern information technology, computer rooms, as key areas for data processing, transmission and storage, environmental control is crucial. It relies on multiple precision air conditioners to maintain specific temperature, humidity and other parameters to ensure the stable operation and service life of internal equipment. In terms of computer room air conditioner control, there are mainly two traditional control methods. One is a pure expert experience-based mode, relying on professionals to determine air conditioner operation parameters and control logic based on past experiences. However, due to significant differences in the scale, equipment layout and heat dissipation requirements of computer rooms, the versatility of this method is poor, and the energy saving effect is not ideal. The other is that although the optimization algorithm strategy can calculate relatively optimal air conditioner control parameters based on the real-time situation of the computer room, once there are equipment updates, layout changes or overall replacements in the computer room, the original algorithm model is difficult to quickly adapt to the new environment, has low robustness, and cannot always maintain good control performance under changing working conditions, resulting in the difficulty of achieving the energy saving goal of air conditioners, and restricting its role in the long-term operation and maintenance of computer rooms. Summary of the Invention

[0003] The main purpose of this application is to provide an energy saving method, device, equipment and storage medium for computer room air conditioners based on large language models, aiming to solve the technical problem of how to improve the versatility, robustness and intelligence level of computer room air conditioner control.

[0004] To achieve the above purpose, this application proposes an energy saving method for computer room air conditioners based on large language models, and the method includes:

[0005] Obtain the real-time environmental status information of the computer room;

[0006] Construct a prompt text according to the real-time environmental status information of the computer room;

[0007] Input the prompt text into the large language model to obtain a set of air conditioner control strategies;

[0008] Verify the set of air conditioner control strategies through a prediction model to obtain a target strategy;

[0009] Perform energy saving control of the computer room air conditioner through the target strategy.

[0010] In one embodiment, the step of constructing a prompt text according to the real-time environmental status information of the computer room includes:

[0011] According to the real-time environmental status information of the computer room, obtain the current environmental information and equipment operation information;

[0012] Obtain the current status description text according to the current environmental information and the device operation information;

[0013] Obtain the control instruction text, the inference logic text, the core target text, and the reward function;

[0014] Obtain the prompt text according to the current status description text, the control instruction text, the inference logic text, the core target text, and the reward function.

[0015] In one embodiment, the step of obtaining the control instruction text, the inference logic text, the core target text, and the reward function includes:

[0016] Obtain the device operation requirements, energy-saving goals, temperature data, energy consumption data of the computer room, and the thought chain control template of the large language model;

[0017] Obtain the control instruction text according to the device operation requirements and the energy-saving goals;

[0018] Obtain the inference logic text according to the thought chain control template;

[0019] Set the core target text for the air conditioner control in the computer room;

[0020] Design the reward function according to the temperature data and the energy consumption data.

[0021] In one embodiment, the step of designing the reward function according to the temperature data and the energy consumption data includes:

[0022] Design the temperature deviation penalty factor according to the temperature data;

[0023] Design the energy consumption reward factor according to the energy consumption data;

[0024] Obtain the reward function according to the temperature deviation penalty factor and the energy consumption reward factor.

[0025] In one embodiment, the step of verifying the air conditioner control strategy set through the prediction model to obtain the target strategy includes:

[0026] Obtain the air conditioner parameter adjustment information in the air conditioner control strategy set;

[0027] Input the real-time environmental status information of the computer room and the air conditioner parameter adjustment information into the prediction model to obtain the temperature change trend and the energy consumption estimation value;

[0028] Analyze the execution effect of the air conditioner control strategy set according to the temperature change trend and the energy consumption estimation value, and obtain the target strategy according to the execution effect.

[0029] In one embodiment, the steps of controlling the energy saving of the computer room air conditioner through the target policy include:

[0030] Adjust the operating parameters of the computer room air conditioner through the target policy;

[0031] Control the operation of the computer room air conditioner according to the operating parameters;

[0032] When the computer room air conditioner is operating, monitor and record the operating data of the computer room and the computer room air conditioner in real time to achieve energy saving control of the computer room air conditioner.

[0033] In one embodiment, after the steps of controlling the energy saving of the computer room air conditioner through the target policy, it includes:

[0034] Obtain the actual execution result data and control process information of the target policy;

[0035] Convert the actual execution result data and the control process information into natural language feedback information;

[0036] Obtain target data according to the natural language feedback information;

[0037] Update the current status description text and the parameters of the reward function in the prompt text according to the target data to obtain an updated text;

[0038] Perform feedback iterative learning on the large language model through the updated text to optimize the inference logic and policy generation mechanism of the large language model.

[0039] In addition, to achieve the above object, the present application also proposes a computer room air conditioner energy saving device based on a large language model, and the device includes:

[0040] A data acquisition module for acquiring real-time environmental status information of the computer room;

[0041] A text construction module for constructing a prompt text according to the real-time environmental status information of the computer room;

[0042] A policy generation module for inputting the prompt text into a large language model to obtain a set of air conditioner control policies;

[0043] A verification and prediction module for verifying the set of air conditioner control policies through a prediction model to obtain a target policy;

[0044] A policy execution module for controlling the energy saving of the computer room air conditioner through the target policy.

[0045] In addition, to achieve the above object, the present application also provides an energy-saving device for computer room air conditioners based on a large language model, where the device includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the energy-saving method for computer room air conditioners based on a large language model as described above.

[0046] In addition, to achieve the above object, the present application also provides a storage medium, where the storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium, and when the computer program is executed by a processor, it implements the steps of the energy-saving method for computer room air conditioners based on a large language model as described above.

[0047] In addition, to achieve the above object, the present application also provides a computer program product, where the computer program product includes a computer program, and when the computer program is executed by a processor, it implements the steps of the energy-saving method for computer room air conditioners based on a large language model as described above.

[0048] One or more technical solutions proposed by the present application have at least the following technical effects:

[0049] By constructing the input prompt text of the large language model, generating a set of air conditioner control strategies and screening out the optimal strategy, and iteratively optimizing the control strategy according to the execution result of the optimal strategy, the technical problems of poor generality, insufficient robustness, and difficulty in balancing environmental conditions and energy conservation of traditional computer room air conditioner control methods are overcome. Compared with the prior art, it can improve the generality, robustness, and intelligence level of computer room air conditioner control. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0051] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is a schematic flowchart provided for Embodiment 1 of the energy-saving method for computer room air conditioners based on a large language model of the present application;

[0053] Figure 2 It is a schematic flowchart provided for Embodiment 2 of the energy-saving method for computer room air conditioners based on a large language model of the present application;

[0054] Figure 3Schematic flowchart of Prompt construction for the computer room air conditioning energy saving method based on large language model provided in the second embodiment of the present application;

[0055] Figure 4 Schematic flowchart provided for the third embodiment of the computer room air conditioning energy saving method based on large language model of the present application;

[0056] Figure 5 Schematic flowchart provided for the fourth embodiment of the computer room air conditioning energy saving method based on large language model of the present application;

[0057] Figure 6 Schematic flowchart provided for the fifth embodiment of the computer room air conditioning energy saving method based on large language model of the present application;

[0058] Figure 7 Schematic flowchart of the feedback mechanism construction for the computer room air conditioning energy saving method based on large language model provided in the fifth embodiment of the present application;

[0059] Figure 8 Schematic flowchart of the computer room air conditioning energy saving method based on large language model provided in the embodiment of the present application;

[0060] Figure 9 Schematic diagram of the module structure of the computer room air conditioning energy saving device based on large language model in the embodiment of the present application;

[0061] Figure 10 Schematic diagram of the device structure of the hardware operating environment involved in the computer room air conditioning energy saving method based on large language model in the embodiment of the present application.

[0062] The realization of the purpose, functional characteristics and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0063] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0064] In order to better understand the technical solutions of the present application, the following will be described in detail in combination with the accompanying drawings of the specification and the specific implementation manners.

[0065] The main solution of the embodiment of the present application is: obtaining the real-time environmental status information of the computer room; constructing a prompt text according to the real-time environmental status information of the computer room; inputting the prompt text into a large language model to obtain a set of air conditioning control strategies; verifying the set of air conditioning control strategies through a prediction model to obtain a target strategy; and performing computer room air conditioning energy saving control through the target strategy.

[0066] In this embodiment, for the convenience of description, the internal actuator of the computer room air conditioning energy saving system based on the large language model is used as the execution subject for elaboration below.

[0067] Due to the poor generality and insufficient robustness of the existing computer room air conditioning control methods, it is difficult to balance environmental conditions and energy saving.

[0068] This application provides a solution. By constructing the input prompt text of the large language model, generating an air conditioning control strategy set and screening out the optimal strategy, and iteratively optimizing the control strategy according to the execution result of the optimal strategy, it can improve the generality, robustness and intelligence level of the computer room air conditioning control.

[0069] As can be seen from the above embodiments, in this application, by constructing the input prompt text of the large language model, generating an air conditioning control strategy set and screening out the optimal strategy, and iteratively optimizing the control strategy according to the execution result of the optimal strategy, it overcomes the technical problems of poor generality and insufficient robustness of the traditional computer room air conditioning control methods, and it is difficult to balance environmental conditions and energy saving, and can improve the generality, robustness and intelligence level of the computer room air conditioning control.

[0070] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions. The following takes the internal actuator of the computer room air conditioning energy saving system based on the large language model as an example to illustrate this embodiment and the following embodiments.

[0071] Based on this, the embodiment of this application provides a computer room air conditioning energy saving method based on the large language model, referring to Figure 1 , Figure 1 is the flow chart of the first embodiment of the computer room air conditioning energy saving method based on the large language model of this application.

[0072] In this embodiment, the computer room air conditioning energy saving method based on the large language model includes steps S10 to S50:

[0073] Step S10, obtaining the real-time environmental status information of the computer room.

[0074] It should be noted that the real-time environmental status information of the computer room includes real-time environmental status data such as temperature distribution, humidity data, and equipment operating power in different areas of the computer room, and also includes air conditioning operation parameter data such as cooling capacity, air volume, air speed, and energy consumption of the air conditioner. The real-time environmental status information of the computer room can be collected by installing various sensors in the inner cylinder area of the computer room.

[0075] Step S20, constructing a prompt text according to the real-time environmental status information of the computer room.

[0076] It should be noted that the prompt text, that is, Prompt, is a kind of input information given to a model or system, which is used to guide the model to generate specific outputs or perform specific tasks. Its main functions include guiding the model output. For example, when the prompt "Write a poem about spring" is input into a language model, the model will generate a poem related to spring; restricting the scope of the generated content. For example, using the prompt "Describe a cat in a humorous style", the content generated by the model will revolve around the cat and adopt a humorous style; stimulating specific capabilities of the model. For example, when the prompt of a math calculation problem is input into the model, the model will use its math calculation ability to give an answer.

[0077] Before constructing the Prompt, determine parameters such as the initial target temperature range and energy consumption limit for the computer room air conditioner control. You can first set the initial Prompt of the model, including the description of the initial state of the computer room, basic control instructions, a preliminary thinking chain control template, the initial target, the default reward function, and the feedback information part of the air conditioner. Then, improve the initial Prompt according to the real-time environmental state information of the computer room and other relevant data. Finally, obtain the Prompt that can be used for model input.

[0078] In step S30, input the said prompt text into a large language model to obtain a set of air conditioner control strategies.

[0079] It should be noted that the large language model (Large Language Model, LLM) is an artificial intelligence technology based on deep learning and is also one of the core research contents of natural language processing. It can understand the natural language text input by people, analyze semantic, syntactic and other information in it, and then generate high-quality and logical text content based on a large amount of training data and the language patterns it has learned, such as answering various questions, etc.

[0080] Additionally, it should be noted that the air conditioner control strategy set is a set of strategies composed of a series of air conditioner control strategies. The air conditioner control strategy is to regulate the operation of the air conditioner system by using various methods and technologies to achieve the goals of comfort, energy conservation, and high efficiency. In terms of temperature control, the indoor temperature can be monitored by a temperature sensor. When it is higher or lower than the set value, it will automatically cool or heat to maintain a constant temperature. There is also adaptive temperature control, which will dynamically adjust the set value in combination with the indoor and outdoor environments, the human thermal comfort requirements, different time periods, and personnel activities. Humidity control includes independent humidity control. Using a humidity sensor, when the humidity is abnormal, the dehumidification or humidification function will be started. There is also coordinated control with temperature, and the environment temperature and humidity conditions are jointly adjusted to create a comfortable thermal and humid environment. The wind speed control will adjust the wind speed according to the air conditioner load. When the load is large, the wind speed will be increased, and when the load is small, the wind speed will be decreased. It can also be intelligently adjusted according to the indoor personnel position and activity conditions to avoid direct blowing of strong wind on the human body. The operation mode control has an economic operation mode, which focuses on energy conservation and optimizes the operation parameters and time to reduce consumption. There is also a comfortable operation mode, which comprehensively considers multiple factors to create a comfortable indoor environment.

[0081] Step S40, verify the air conditioner control strategy set through the prediction model to obtain the target strategy.

[0082] It should be noted that the prediction model is a temperature and energy consumption prediction model pre-trained through a large amount of historical data. This model is trained through deep learning based on the environmental parameters, air conditioner control parameters, and corresponding temperature changes and energy consumption data in the historical operation data of the computer room. It can simulate and predict the temperature change trend and energy consumption estimation value of the computer room after executing the strategy according to the input air conditioner control strategy, and evaluate the effectiveness and feasibility of the strategy.

[0083] Additionally, it should be noted that the target strategy is the air conditioner control strategy with the best execution effect in the air conditioner control strategy set obtained after verifying the air conditioner control strategy set.

[0084] Step S50, perform energy-saving control on the computer room air conditioner through the target strategy.

[0085] Apply the target strategy to the actual control of the computer room environment, directly adjust the operation parameters of the precision air conditioner in the computer room. During the execution of the strategy, the temperature change situation in each area of the computer room, the energy consumption data of the air conditioner, and other relevant operation parameters can also be monitored in real time, and the data change curve and key data points are recorded. After each execution of the target strategy, the result data such as the temperature change curve, energy consumption change curve, and operation parameter change of the air conditioner equipment in the computer room are converted into feedback information described in natural language. Such feedback information can facilitate the understanding of the LLM and its learning from it. Whether it is successful control experience or unsatisfactory control results, they can provide valuable reference basis for its subsequent strategy generation, and promote the continuous optimization of the control logic and strategy generation ability of the LLM.

[0086] This embodiment provides an energy-saving method for computer room air conditioners based on large language models. Since this application generates an air conditioner control strategy set and filters out the optimal strategy by constructing the input prompt text of the large language model, and iteratively optimizes the control strategy according to the execution results of the optimal strategy, it overcomes the technical problems of poor generality, insufficient robustness, and difficulty in balancing environmental conditions and energy conservation in traditional computer room air conditioner control methods, and can improve the generality, robustness, and intelligence level of computer room air conditioner control.

[0087] Based on the first embodiment of this application, in the second embodiment of this application, the same or similar content as the above-mentioned embodiment 1 can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 2 , step S20 includes steps S21 to S24:

[0088] Step S21, obtain the current environmental information and equipment operation information according to the real-time environmental status information of the computer room.

[0089] It should be noted that the current environmental information includes the temperature distribution and humidity data in the real-time environmental status information of the computer room, and the equipment operation information includes the real-time operation power data of various types of equipment (such as servers, storage devices, etc.) in the computer room in the real-time environmental status information of the computer room.

[0090] Step S22, obtain the current status description text according to the current environmental information and the equipment operation information.

[0091] Organize the current environmental information and equipment operation information into a standard text description format, such as "The temperature in area A of the computer room is 25°C, the humidity is 45%, the total operation power of the servers is 50kW; the temperature in area B is 26°C, the humidity is 43%, and the operation power of the storage device is 30kW...", to obtain the current status description text, so that the LLM can comprehensively and accurately understand the current environment and equipment operation status of the computer room.

[0092] Step S23, obtain the control instruction text, reasoning logic text, core target text, and reward function.

[0093] It should be noted that the control instruction text is the text for formulating clear air conditioner control instructions based on the operation requirements and energy-saving goals of the computer room equipment. The reasoning logic text is the text of the reasoning logic of the LLM planned according to the preset thinking chain control template. The core target text is the text describing the core target of the computer room air conditioner control.

[0094] Additionally, it should be noted that the reward function is a function that maps the state, action, and next state of the agent to a scalar reward value. Its role is to provide the learning goal and direction for the agent, enabling it to maximize the long-term cumulative reward by choosing actions, and learning beneficial or adverse behaviors based on the reward feedback to adjust the strategy. When designing, it should be consistent with the task goal, the reward value needs to be measurable, have appropriate sparsity, and consider environmental constraints. A scientific and reasonable reward function is constructed based on the Gaussian distribution, and through the reward and punishment mechanism, the LLM is guided to generate a better control strategy, enabling it to actively pursue the balance between energy consumption reduction and temperature stability during the control process.

[0095] In a feasible implementation manner, step S23 includes steps S231 to S235:

[0096] Step S231, obtain the equipment operation requirements, energy-saving goals, temperature data, energy consumption data of the computer room, and the thought chain control template of the large language model.

[0097] It should be noted that the equipment operation in the computer room can include multiple key aspects. For example, in terms of the environment, the temperature needs to be maintained at around 22°C - 24°C. If it is too high, it will be difficult for the equipment to dissipate heat and accelerate the aging of components. If it is too low, it may cause water vapor condensation and trigger a short circuit. The humidity should be controlled at 40% - 60%. If the humidity is too high, it is easy for the equipment to get damp and moldy, affecting insulation. If it is too low, it may generate static electricity and damage electronic components. The energy-saving goal refers to the specific goal set within a certain period to reduce energy consumption, improve energy utilization efficiency, reduce energy waste, and reduce the environmental impact related to energy. It usually has clear quantitative indicators and time nodes. The temperature data includes the target temperature range of the computer room and the measured value of the actual temperature. The energy consumption data includes the energy consumption values of the computer room air conditioner at different times.

[0098] Additionally, it should be noted that the thought chain control template of the large language model, that is, the CoT (Chain of Thought) control template, is a prompting technique designed to guide the model to solve complex problems through step-by-step reasoning rather than directly generating answers. It is usually constructed in the form of natural language, decomposing the problem-solving process into a series of intermediate reasoning steps, allowing the model to think and reason step by step according to these steps, and finally obtain the answer.

[0099] Step S232, obtain the control instruction text according to the equipment operation requirements and the energy-saving goal.

[0100] Based on the operating requirements and energy-saving goals of the computer room equipment, clear air-conditioning control instructions are formulated. For example, "On the premise of ensuring that the overall temperature of the computer room does not exceed 28 °C, gradually reduce the air-conditioning cooling power, and give priority to adjusting the air-conditioning units with higher energy efficiency ratios...", so that the LLM clearly knows what type of control operations need to be carried out and how to set the reasoning steps as "First, analyze the reasons for the uneven temperature distribution in the computer room and determine the areas that need to be key-adjusted; then calculate the adjustment range of the air-conditioning cooling capacity according to the temperature difference between regions; then evaluate the impact on the overall energy consumption after the adjustment; finally, comprehensively consider to obtain a specific combination of air-conditioning control strategies...".

[0101] Step S233, obtain the reasoning logic text according to the thinking chain control template.

[0102] According to the preset CoT control template, plan the LLM reasoning logic, such as "First, analyze the reasons for the uneven temperature distribution in the computer room and determine the areas that need to be key-adjusted; then calculate the adjustment range of the air-conditioning cooling capacity according to the temperature difference between regions; then evaluate the impact on the overall energy consumption after the adjustment; finally, comprehensively consider to obtain a specific combination of air-conditioning control strategies...".

[0103] Step S234, set the core target text for the computer room air-conditioning control.

[0104] Set the core goal of the computer room air-conditioning control, such as "Stabilize the average temperature of the computer room between 26 - 28 °C, and at the same time reduce the overall energy consumption of the air-conditioning system by 5 - 10% compared with the previous cycle", so that the LLM always optimizes and weighs around this goal when generating strategies.

[0105] Step S235, design a reward function according to the temperature data and the energy consumption data.

[0106] Scientific and reasonable design of the reward function can accurately quantify the advantages and disadvantages of the control strategy and guide the LLM to learn and optimize, as well as convert the execution result into natural language feedback information and use it for updating the Prompt and the LLM iterative learning mechanism to achieve the self-optimizing cycle of the control process.

[0107] In a feasible implementation manner, step S235 includes steps A10 - A30:

[0108] Step A10, design a temperature deviation penalty factor according to the temperature data.

[0109] Based on the temperature data, assume that the target temperature range of the computer room is (T min , T max ), at time t, the actual temperature measurement values of each area in the computer room are T i (t) (i = 1, 2,..., n, n is the number of areas), calculate the sum of the absolute values of the temperature deviation of each area from the target range, and the formula is as follows:

[0110]

[0111] Among them, D T (t) is the sum of the absolute values of the temperature deviation of each area from the target range, and T i (t) is the measured value of the actual temperature of each area in the computer room, and T target is the median of the target temperature range, that is, T target =(T min +T max ) / 2.

[0112] Define the temperature deviation penalty factor, and the formula is as follows:

[0113]

[0114] Among them, P T (t) is the temperature deviation penalty factor, and C T is a constant set according to the scale of the computer room and the temperature sensitivity, and is used to normalize the penalty factor so that P T (t) takes values between 0 and 1.

[0115] The larger the value of P T (t), the more serious the temperature deviation and the greater the penalty.

[0116] Step A20, design an energy consumption reward factor according to the energy consumption data.

[0117] Based on the energy consumption data, calculate the energy consumption reduction ratio, and the formula is as follows:

[0118]

[0119] Among them, r E (t) is the energy consumption reduction ratio, the energy consumption of the air conditioning system at time t is E(t), and the energy consumption at the previous moment is E(t - 1).

[0120] Define the energy consumption reward factor P E (t)=min(r E (t), 1), that is, if the energy consumption reduction ratio r E (t) exceeds 1 (to prevent abnormal data), then the energy consumption reward factor takes 1, and the larger this value, the better the energy consumption reduction effect and the greater the reward.

[0121] Step A30, obtain a reward function according to the temperature deviation penalty factor and the energy consumption reward factor.

[0122] Define a reward function according to the temperature deviation penalty factor and the energy consumption reward factor, and the formula is as follows:

[0123] R(t) = αP E (t) - (1 - α)P T (t)

[0124] Where α is the weight coefficient, with a value range between 0 and 1, used to balance the relative importance of the energy consumption reduction reward and the temperature deviation penalty.

[0125] When α is larger, more emphasis is placed on the reward for energy consumption reduction; when α is smaller, more attention is paid to the penalty for temperature control.

[0126] By designing the reward and penalty mechanism, it is possible to guide the LLM to generate better control strategies, enabling it to actively pursue the balance between energy consumption reduction and temperature stability during the control process.

[0127] By constructing the relevant content in the Prompt, a basis is provided for generating strategies for the LLM.

[0128] Step S24, obtain the prompt text according to the current state description text, the control instruction text, the inference logic text, the core target text, and the reward function.

[0129] Write the current state description text, the control instruction text, the inference logic text, the core target text, and the reward function into the Prompt, so as to input the constructed Prompt into the LLM. The LLM generates a set of air-conditioning control strategies. At the same time, after the strategy is executed, feedback information can also be obtained according to the execution effect of the strategy to update the Prompt for iterative learning of the LLM.

[0130] As Figure 3 shown in the schematic diagram of the brief process of Prompt construction, it includes: describing the current state (weather, computer room environment, etc.); providing control instructions (air-conditioning switch, set temperature, etc.); CoT template (planning the inference logic of the LLM); clarifying the goal (maintaining a reasonable temperature and low energy consumption); reward function (design of the reward function based on the Gaussian distribution); feedback information (evaluating the control strategy effect and information update).

[0131] This embodiment provides an energy-saving method for computer room air conditioners based on a large language model. According to the real-time environment state information of the computer room, the current environment information and equipment operation information are obtained; according to the current environment information and the equipment operation information, the current state description text is obtained; the control instruction text, the inference logic text, the core target text, and the reward function are obtained; the prompt text is obtained according to the current state description text, the control instruction text, the inference logic text, the core target text, and the reward function, which can improve the generality of computer room air conditioner control.

[0132] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar content as that in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 4 , step S40 includes steps S41 to S43:

[0133] Step S41, obtain the air-conditioning parameter adjustment information in the air-conditioning control strategy set.

[0134] It should be noted that the air-conditioning parameter adjustment information is the information used to adjust the air-conditioning parameters, which may include the air-conditioning temperature parameter range, the air-conditioning wind speed parameter range, etc.

[0135] Step S42, input the real-time environmental status information of the computer room and the air-conditioning parameter adjustment information into the prediction model to obtain the temperature change trend and the energy consumption estimation value.

[0136] It should be noted that the temperature change trend refers to the dynamic trend of the temperature in the computer room rising or falling over time after implementing the corresponding air-conditioning parameter adjustment strategy. The energy consumption estimation value is based on the air-conditioning parameter adjustment strategy, considering the power consumption corresponding to each air-conditioning parameter, such as the energy consumption differences of various components at different wind speeds, modes, and temperature settings, combined with factors such as the space size and heat load of the computer room, and is obtained by simulation calculation of the energy that the computer room air-conditioning system is expected to consume within a certain period of time.

[0137] Input the real-time environmental status information of the computer room and the air-conditioning parameter adjustment information into the prediction model. After receiving these inputs, the prediction model will perform simulation calculations based on its built-in algorithms and relevant physical principles, etc. Regarding the temperature change trend, it can comprehensively consider the impact of the air-conditioning cooling or heating capacity on the computer room environmental temperature under different parameter settings. For example, when reducing the air-conditioning temperature setting value, increasing the wind speed, or switching to a stronger cooling mode, it can simulate how the temperature in the computer room will decrease at what rate and gradually approach the newly set target temperature, or how the temperature will rise to a stable state in the heating scenario. At the same time, it can also consider the role of factors such as the overall heat dissipation situation and space layout of the computer room on the temperature change, presenting the temperature change curve over time and clearly showing the trend of temperature increase or decrease. For the energy consumption estimation value, the model will combine the power consumption corresponding to each air-conditioning parameter, such as higher wind speed consuming more energy than lower wind speed, the energy consumption differences of key components such as compressors at different temperature settings in the cooling mode, and the current space size and heat load of the computer room, and estimate the energy consumption of the entire computer room air-conditioning system within a certain period of time after implementing this strategy through calculation.

[0138] Step S43, analyze the execution effect of the air-conditioning control strategy set according to the temperature change trend and the energy consumption estimation value, and obtain the target strategy according to the execution effect.

[0139] Compare the effects after executing each policy in the air-conditioning control policy set with the preset temperature target range and energy consumption limit. For the comparison of the temperature target range, check whether the actual temperature change trend in the computer room after executing the policy is stably within the preset temperature range. If the actual temperature always fluctuates within the target range, it indicates that the policy performs well in temperature control and can meet the requirements of computer room equipment for a suitable working temperature. For example, if the computer room equipment requires the working temperature to be between 20°C and 25°C, and after executing a certain policy, the computer room temperature can be stabilized within this range and will not significantly exceed or fall below this range, then the policy is effective in temperature control. Regarding the consideration of energy consumption limit, compare the estimated energy consumption value after executing the policy with the preset energy consumption upper limit. If the estimated energy consumption value is lower than or equal to the energy consumption limit, it indicates that the policy meets the requirements in terms of energy conservation. For example, if the preset computer room energy consumption shall not exceed [X] kWh per month, and after executing a certain policy, the energy consumption value estimated by the model is within this limit, it means that the policy is feasible in terms of energy conservation. By conducting such comparative analysis on multiple policies, the best-performing policy can be selected from numerous policies as the target policy. This target policy should perform well in both temperature control and energy conservation. It can not only ensure that the computer room temperature is always within a suitable range to guarantee the normal operation and lifespan of the equipment, but also minimize energy consumption, reduce operating costs, and achieve efficient management and sustainable development of the computer room.

[0140] This embodiment provides an energy-saving method for computer room air conditioners based on a large language model, which obtains the air-conditioning parameter adjustment information in the air-conditioning control policy set; inputs the real-time environmental state information of the computer room and the air-conditioning parameter adjustment information into a prediction model to obtain the temperature change trend and the estimated energy consumption value; analyzes the execution effect of the air-conditioning control policy set according to the temperature change trend and the estimated energy consumption value, and obtains the target policy according to the execution effect, which can improve the generality of computer room air-conditioning control.

[0141] Based on the first embodiment of the present application, in the fourth embodiment of the present application, the same or similar content as that in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 5 , step S50 includes steps S51 to S53:

[0142] Step S51, adjust the operating parameters of the computer room air conditioner through the target policy.

[0143] It should be noted that the operating parameters of the computer room air conditioner mainly include temperature parameters, humidity parameters, pressure parameters, flow parameters, etc.

[0144] After determining the target strategy, it can be applied to the actual control link of the computer room environment to directly adjust the operating parameters of the precision air conditioner in the computer room to achieve the expected environmental control effect. In actual operation, adjust the temperature setting value of the air conditioner according to the temperature parameter set in the target strategy to ensure that it meets the requirements of the target strategy. For example, accurately set the temperature to the value most suitable for the stable operation of the computer room equipment, so that the temperature in the computer room can be maintained within the ideal range, ensuring that various hardware such as servers and storage devices do not experience performance degradation, failures, etc. due to too high or too low temperature. At the same time, according to the humidity target range specified in the target strategy, adjust the humidity setting, dehumidification function, etc. of the precision air conditioner accordingly, so that the air humidity in the computer room remains at an appropriate level, avoiding short circuits caused by electronic components getting damp due to too high humidity, or damage to equipment caused by static electricity due to too low humidity. In terms of pressure parameters, according to the relevant requirements of the suction pressure, discharge pressure, evaporation pressure, and condensation pressure involved in the target strategy, by adjusting means such as the operating frequency of the compressor and the opening of the valve, maintain the pressure of each link of the refrigeration system in a reasonable state, ensure the normal and efficient operation of the air conditioner refrigeration cycle, and ensure that it can continuously and stably provide suitable temperature adjustment effects for the computer room. For flow parameters, according to the standards of the refrigerant flow and air flow planned in the target strategy, adjust the throttling device, fan speed, etc. in the air conditioning system to ensure that the refrigerant can circulate at an appropriate flow rate in the system, and the air can also form a good circulation flow in the computer room, so as to achieve an ideal state in all aspects such as temperature and humidity and air circulation in the entire computer room environment, and create a high-quality and stable operating environment for many precision devices in the computer room.

[0145] Step S52, control the operation of the computer room air conditioner according to the operating parameters.

[0146] After adjusting the operating parameters, the operation of the computer room air conditioner can be controlled according to these parameters. During the operation process, the real-time status of the computer room and the computer room air conditioner can be combined, and the operating parameters of the air conditioner can be adaptively adjusted according to the target strategy to make the air conditioner achieve a better operating effect.

[0147] Step S53, when the computer room air conditioner is operating, monitor and record the operating data of the computer room and the computer room air conditioner in real time to achieve energy-saving control of the computer room air conditioner.

[0148] It should be noted that the operating data are data such as the environmental changes in the computer room and the energy consumption of the computer room air conditioner during the execution of the target strategy.

[0149] During the execution of the strategy, monitor the temperature change conditions in each area of the computer room, the energy consumption data of the air conditioner, and other relevant operating parameters in real time, and record the data change curve and key data points.

[0150] This embodiment provides an energy-saving method for computer room air conditioners based on large language models, which obtains the air conditioner parameter adjustment information in the air conditioner control strategy set; inputs the real-time environmental status information of the computer room and the air conditioner parameter adjustment information into a prediction model to obtain the temperature change trend and the estimated energy consumption value; analyzes the execution effect of the air conditioner control strategy set according to the temperature change trend and the estimated energy consumption value, and obtains the target strategy according to the execution effect, which can improve the generality of computer room air conditioner control.

[0151] Based on the first embodiment of the present application, in the fifth embodiment of the present application, the same or similar content as that in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, please refer to Figure 6 , after step S50, it includes steps S01 to S05:

[0152] Step S01, obtain the actual execution result data and control process information of the target strategy.

[0153] It should be noted that the actual execution result data includes data such as the actual change of the temperature in the computer room, the actual consumption value of the energy consumption, and the final state of the operation parameters of the air conditioner equipment. The control process information is the relevant information of the air conditioner operation during the execution of the strategy.

[0154] To obtain the actual execution result data of the air conditioner control strategy such as the actual change of the temperature in the computer room and the actual consumption value of the energy consumption, the temperature data at the location can be collected in real time through the sensors installed in the computer room, and the temperature data can be presented in the form of charts, etc., to intuitively show how the temperature in the computer room changes over time. Regarding the actual consumption value of the energy consumption, it can be obtained by means of an electricity monitor installed on the air conditioner power supply line, which can accurately record the electricity consumption of the air conditioner at different operation periods. By summarizing these electricity data, such as by time periods of days, weeks, months, etc., the actual consumption of the air conditioner energy consumption can be known. At the same time, the operation parameters of the air conditioner equipment can be obtained through the display and recording function of the air conditioner itself, such as viewing the set temperature, actual room temperature, operation mode, wind speed gear, etc. displayed in real time on the operation panel.

[0155] Step S02, convert the actual execution result data and the control process information into natural language feedback information.

[0156] It should be noted that the natural language feedback information is the text information obtained by natural language description. The natural language description refers to the description that can clearly, accurately and effectively convey specific information, ideas, emotions or content, and conforms to language norms and logic, and has certain value and understanding significance for readers or listeners.

[0157] Convert the actual execution result data and control process information into meaningful natural language descriptions. For example, "After the execution of this control strategy, the temperature in Area A of the computer room increased from 24°C to 27.5°C within 1 hour, the energy consumption decreased by 8%, the set temperature of Air Conditioner No. 1 was adjusted from 23°C to 26°C, and the set temperature of Air Conditioner No. 2 was adjusted from 25°C to 26.5°C...", and then generate a feedback information text based on the natural language description, that is, the natural language feedback information.

[0158] Step S03, obtain target data according to the natural language feedback information.

[0159] It should be noted that the target data is the actual execution result data and control process information about the target strategy extracted from the natural language feedback information, including data such as temperature change values, energy consumption change values, and air conditioner parameter changes.

[0160] Step S04, update the current status description text and the parameters of the reward function in the prompt text according to the target data to obtain an updated text.

[0161] Update the Prompt according to the objective function. The updated content includes the current status description text and the parameters of the reward function of the Prompt, as well as the control instruction text, the inference logic text, and the core objective text.

[0162] Step S05, perform feedback iterative learning on the large language model through the updated text to optimize the inference logic and strategy generation mechanism of the large language model.

[0163] Input the updated Prompt into the LLM for feedback iterative learning. The LLM analyzes the deficiencies and successful experiences of the control strategy based on the rewards and punishments in the feedback information, and adjusts its own inference logic and strategy generation mechanism. Both successful control experiences and unsatisfactory control results can provide valuable reference bases for its subsequent strategy generation, promoting its continuous optimization of the control logic and strategy generation ability, and preparing for generating better strategies in the next control cycle. At the same time, the steps of constructing the Prompt to the LLM iterative learning can be repeated according to the set time period or trigger conditions to continuously optimize the computer room air conditioner control strategy, realize the dynamic adjustment and adaptive optimization of the computer room air conditioner energy-saving control, and ensure that the computer room environment is always in a stable and energy-saving state.

[0164] As Figure 7 shown is a schematic diagram of the brief process of constructing the feedback mechanism, including: collecting the control strategy execution result data; converting the data into natural language descriptions; generating feedback information based on the natural language descriptions; extracting key data from the feedback information; updating relevant content in the Prompt.

[0165] This embodiment provides a method for energy saving of computer room air conditioners based on large language models, which includes obtaining the actual execution result data and control process information of the target strategy; converting the actual execution result data and the control process information into natural language feedback information; obtaining target data according to the natural language feedback information; updating the current state description text and the parameters of the reward function in the prompt text according to the target data to obtain an updated text; and performing feedback iterative learning on the large language model through the updated text to optimize the inference logic and strategy generation mechanism of the large language model, so as to improve the generality, robustness, and intelligence level of computer room air conditioner control.

[0166] Exemplarily, to facilitate understanding of the implementation process of the method for energy saving of computer room air conditioners based on large language models obtained by combining Embodiments 1, 2, 3, 4, and 5, please refer to Figure 8 , Figure 8 A brief flow schematic diagram of the method for energy saving of computer room air conditioners based on large language models is provided, specifically:

[0167] State acquisition; Prompt construction; Policy set generation; Prediction verification; Action execution; Feedback reception.

[0168] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the method for energy saving of computer room air conditioners based on large language models of this application. Any simple transformations in more forms based on this technical concept are within the protection scope of this application.

[0169] This application also provides a device for energy saving of computer room air conditioners based on large language models. Please refer to Figure 9 , and the device includes:

[0170] A data acquisition module 10 for obtaining real-time environmental state information of the computer room;

[0171] A text construction module 20 for constructing a prompt text according to the real-time environmental state information of the computer room;

[0172] A policy generation module 30 for inputting the prompt text into a large language model to obtain an air conditioner control policy set;

[0173] A verification and prediction module 40 for verifying the air conditioner control policy set through a prediction model to obtain a target policy;

[0174] A policy execution module 50 for performing energy saving control of the computer room air conditioner through the target policy.

[0175] In one embodiment, the text construction module 20 is further configured to obtain the current environmental information and equipment operation information according to the real-time environmental status information of the computer room; obtain the current status description text according to the current environmental information and the equipment operation information; acquire the control instruction text, the inference logic text, the core target text, and the reward function; and obtain the prompt text according to the current status description text, the control instruction text, the inference logic text, the core target text, and the reward function.

[0176] In one embodiment, the text construction module 20 is further configured to obtain the equipment operation requirements, energy-saving targets, temperature data, energy consumption data of the computer room, and the thought chain control template of the large language model; obtain the control instruction text according to the equipment operation requirements and the energy-saving targets; obtain the inference logic text according to the thought chain control template; set the core target text for the computer room air conditioner control; and design the reward function according to the temperature data and the energy consumption data.

[0177] In one embodiment, the text construction module 20 is further configured to design a temperature deviation penalty factor according to the temperature data; design an energy consumption reward factor according to the energy consumption data; and obtain the reward function according to the temperature deviation penalty factor and the energy consumption reward factor.

[0178] In one embodiment, the verification and prediction module 40 is further configured to obtain the air conditioner parameter adjustment information in the air conditioner control strategy set; input the real-time environmental status information of the computer room and the air conditioner parameter adjustment information into the prediction model to obtain the temperature change trend and the energy consumption estimation value; analyze the execution effect of the air conditioner control strategy set according to the temperature change trend and the energy consumption estimation value, and obtain the target strategy according to the execution effect.

[0179] In one embodiment, the strategy execution module 50 is further configured to adjust the operation parameters of the computer room air conditioner through the target strategy; control the operation of the computer room air conditioner according to the operation parameters; and during the operation of the computer room air conditioner, monitor and record the operation data of the computer room and the computer room air conditioner in real time to achieve energy-saving control of the computer room air conditioner.

[0180] In one embodiment, the strategy execution module 50 is further configured to obtain the actual execution result data and control process information of the target strategy; convert the actual execution result data and the control process information into natural language feedback information; obtain the target data according to the natural language feedback information; update the parameters of the current status description text and the reward function in the prompt text according to the target data to obtain the updated text; and perform feedback iterative learning on the large language model through the updated text to optimize the inference logic and strategy generation mechanism of the large language model.

[0181] The energy-saving device for computer room air conditioners based on large language models provided by this application adopts the energy-saving method for computer room air conditioners based on large language models in the above embodiments, and can solve the technical problem of how to improve the generality, robustness and intelligence level of computer room air conditioner control. Compared with the prior art, the beneficial effects of the energy-saving device for computer room air conditioners based on large language models provided by this application are the same as those of the energy-saving method for computer room air conditioners based on large language models provided by the above embodiments, and other technical features in the energy-saving device for computer room air conditioners based on large language models are the same as the features disclosed in the above embodiment methods, which will not be elaborated here.

[0182] This application provides an energy-saving device for computer room air conditioners based on large language models. The energy-saving device for computer room air conditioners based on large language models includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the energy-saving method for computer room air conditioners based on large language models in the first embodiment above.

[0183] The following refers to Figure 10 , which shows a schematic structural diagram of an energy-saving device for computer room air conditioners based on large language models suitable for implementing the embodiments of this application. The energy-saving device for computer room air conditioners based on large language models in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions: tablet computers), PMPs (Portable Media Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 10 The shown energy-saving device for computer room air conditioners based on large language models is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of this application.

[0184] As Figure 10As shown, the energy-saving device for computer room air conditioners based on large language models may include a processing device 1001 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 into the random access memory (RAM: Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the energy-saving device for computer room air conditioners based on large language models are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the energy-saving device for computer room air conditioners based on large language models to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an energy-saving device for computer room air conditioners based on large language models with various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems can be alternatively implemented or had.

[0185] Particularly, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments disclosed in the present application are executed.

[0186] The energy-saving device for computer room air conditioners based on large language models provided by this application adopts the energy-saving method for computer room air conditioners based on large language models in the above embodiments, and can solve the technical problem of how to improve the generality, robustness, and intelligence level of computer room air conditioner control. Compared with the prior art, the beneficial effects of the energy-saving device for computer room air conditioners based on large language models provided by this application are the same as those of the energy-saving method for computer room air conditioners based on large language models provided by the above embodiments, and other technical features in the energy-saving device for computer room air conditioners based on large language models are the same as the features disclosed in the method of the previous embodiment, which will not be elaborated here.

[0187] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0188] As described above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claimed rights.

[0189] This application provides a computer-readable storage medium with computer-readable program instructions (i.e., computer programs) stored thereon, and the computer-readable program instructions are used to execute the energy-saving method for computer room air conditioners based on large language models in the above embodiments.

[0190] The computer-readable storage medium provided by the present application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, device, or component. The program code contained on the computer-readable storage medium can be transmitted by any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0191] The above computer-readable storage medium can be included in the energy-saving device for computer room air conditioning based on a large language model; or it can exist independently without being assembled into the energy-saving device for computer room air conditioning based on a large language model.

[0192] The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed by the energy-saving device for computer room air conditioning based on a large language model, the energy-saving device for computer room air conditioning based on a large language model is enabled to: obtain real-time environmental status information of the computer room; construct a prompt text according to the real-time environmental status information of the computer room; input the prompt text into the large language model to obtain a set of air conditioning control strategies; verify the set of air conditioning control strategies through a prediction model to obtain a target strategy; and perform energy-saving control of the computer room air conditioning through the target strategy.

[0193] Computer program code for performing the operations of this application can be written in one or more programming languages or combinations thereof. The above-mentioned programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, by connecting through the Internet using an Internet service provider).

[0194] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of the code, and this module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0195] The modules described in the embodiments of this application can be implemented in software or in hardware. Among them, the name of the module does not constitute a limitation to the unit itself in some cases.

[0196] The readable storage medium provided in this application is a computer-readable storage medium. The computer-readable storage medium stores computer-readable program instructions (i.e., computer programs) for performing the above-mentioned computer room air-conditioning energy-saving method based on a large language model, and can solve the technical problem of how to improve the versatility, robustness, and intelligence level of computer room air-conditioning control. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the computer room air-conditioning energy-saving method based on a large language model provided in the above embodiments, and will not be elaborated here.

[0197] The present application also provides a computer program product, including a computer program, which when executed by a processor implements the steps of the above-mentioned computer room air-conditioning energy-saving method based on a large language model.

[0198] The computer program product provided by the present application can solve the technical problem of how to improve the generality, robustness and intelligence of computer room air-conditioning control. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as those of the computer room air-conditioning energy-saving method based on a large language model provided in the above embodiments, and will not be elaborated here.

[0199] The above are only partial embodiments of the present application, and thus do not limit the patent scope of the present application. Any equivalent structural transformation made under the technical concept of the present application by using the content of the specification and drawings of the present application, or any direct / indirect application in other related technical fields is included in the patent protection scope of the present application.

Claims

1. A computer room air conditioning energy saving method based on a large language model, characterized in that: The method comprises: Obtain real-time environmental status information of the computer room; Constructing a prompt text according to the real-time environmental status information of the computer room; Inputting the prompt text into a large language model to obtain an air conditioning control strategy set; Verifying the air conditioning control strategy set through a prediction model to obtain a target strategy; The target strategy is used to control the energy saving of the computer room air conditioner.

2. The method according to claim 1, characterized in that The step of constructing a prompt text according to the real-time environment status information of the computer room comprises: According to the real-time environmental status information of the computer room, current environmental information and equipment operation information are obtained; Obtaining a current status description text according to the current environment information and the device operation information; Obtain control instruction text, reasoning logic text, core target text and reward function; A prompt text is obtained according to the current state description text, the control instruction text, the reasoning logic text, the core target text and the reward function.

3. The method according to claim 2, characterized in that The step of obtaining the control instruction text, the reasoning logic text, the core target text and the reward function comprises: Obtain the equipment operation requirements, energy-saving targets, temperature data, energy consumption data and the thinking chain control template of the large language model in the computer room; Obtaining a control instruction text according to the equipment operation requirements and the energy-saving target; Obtaining a reasoning logic text according to the thought chain control template; Set the core target text for computer room air conditioning control; A reward function is designed according to the temperature data and the energy consumption data.

4. The method according to claim 3, characterized in that The step of designing a reward function according to the temperature data and the energy consumption data comprises: designing a temperature deviation penalty factor according to the temperature data; designing an energy consumption bonus factor according to the energy consumption data; A reward function is obtained according to the temperature deviation penalty factor and the energy consumption reward factor.

5. The method according to claim 1, characterized in that The step of verifying the air conditioning control strategy set by using the prediction model to obtain the target strategy includes: Acquiring air conditioning parameter adjustment information in the air conditioning control strategy set; Input the real-time environmental status information of the computer room and the air conditioning parameter adjustment information into a prediction model to obtain a temperature change trend and an estimated energy consumption value; The execution effect of the air conditioning control strategy set is analyzed according to the temperature change trend and the energy consumption estimation value, and the target strategy is obtained according to the execution effect.

6. The method according to claim 1, characterized in that The step of performing energy-saving control of the computer room air conditioner by using the target strategy includes: Adjusting the operating parameters of the computer room air conditioner through the target strategy; Controlling the operation of the computer room air conditioner according to the operating parameters; When the computer room air conditioner is running, the operating data of the computer room and the computer room air conditioner are monitored and recorded in real time to achieve energy-saving control of the computer room air conditioner.

7. The method according to any one of claims 1 to 6, characterized in that After the step of performing energy-saving control of the computer room air conditioner by using the target strategy, the method further comprises: Obtaining actual execution result data and control process information of the target strategy; Converting the actual execution result data and the control process information into natural language feedback information; Obtaining target data according to the natural language feedback information; updating the current state description text and the parameters of the reward function in the prompt text according to the target data to obtain an updated text; Feedback iterative learning is performed on the large language model through the updated text to optimize the reasoning logic and strategy generation mechanism of the large language model.

8. A computer room air conditioning energy saving device based on a large language model, characterized in that: The device comprises: Data acquisition module, used to obtain real-time environmental status information of the computer room; A text construction module, used to construct a prompt text according to the real-time environmental status information of the computer room; A strategy generation module, used for inputting the prompt text into a large language model to obtain an air conditioning control strategy set; A verification prediction module, used to verify the air conditioning control strategy set through a prediction model to obtain a target strategy; The strategy execution module is used to perform energy-saving control of the computer room air conditioner through the target strategy.

9. A computer room air conditioning energy-saving device based on a large language model, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the computer room air conditioning energy saving method based on a large language model as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the computer room air conditioning energy saving method based on a large language model as described in any one of claims 1 to 7 are implemented.

Images