Multi-objective optimization method, device and equipment of machine room cooling system, storage medium and computer program product

Abstract

This application relates to the field of air conditioning room control technology, and in particular to a multi-objective optimization method, apparatus, equipment, storage medium, and computer program product for a computer room cooling system. The method acquires multi-source information from the computer room cooling system and preprocesses it to obtain a multi-source input set. Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context. The key operational data in the decision context undergoes credibility assessment and correction processing to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. The candidate control suggestions are summarized and conflict resolved based on preset priority arbitration rules to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and these instructions are sent to the corresponding controllers when boundary checks pass.

Description

Technical Field

[0001] This application relates to the field of air conditioning room control technology, and in particular to a multi-objective optimization method, apparatus, equipment, storage medium and computer program product for a computer room cooling system. Background Technology

[0002] Data center server rooms are typically equipped with cooling equipment such as water-cooled terminal air conditioners to regulate and control the server room environment and equipment operating status. This control falls under the data center infrastructure management domain and focuses on optimizing the operation of the server room's water-cooled terminal air conditioning system. Existing server room water-cooled terminal air conditioning control strategies are mostly based on preset fixed operating parameters such as return air temperature and fan speed, or use traditional optimization algorithms to dynamically adjust finite objectives. During server room operation, in addition to real-time sensor and equipment operating parameters, there is also unstructured information such as server room CAD layout diagrams, maintenance logs, future load plans, and special settings in equipment manuals. However, the aforementioned existing methods often struggle to effectively understand and utilize this type of unstructured information, and lack global perception and reasoning support for complex and dynamic environments. Under multi-objective constraints, it is difficult to form consistent control decisions, resulting in poor overall energy efficiency of the server room cooling system. Therefore, improving the overall energy efficiency of server room cooling systems has become an urgent technical problem to be solved. Summary of the Invention

[0003] The main objective of this application is to provide a multi-objective optimization method, apparatus, equipment, storage medium, and computer program product for a computer room cooling system, aiming to solve the technical problem of how to improve the overall energy efficiency of a computer room cooling system.

[0004] To achieve the above objectives, this application provides a multi-objective optimization method for a computer room cooling system, the method comprising the following steps: Acquire multi-source information from the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set; Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context; The key operational data in the decision context are subjected to credibility assessment and correction processing to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. Based on preset priority arbitration rules, the set of candidate control suggestions is summarized and conflict resolved to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and the structured control instructions are sent to the corresponding controller when the boundary check passes.

[0005] In one embodiment, the step of acquiring multi-source information of the computer room cooling system and preprocessing the multi-source information to obtain a multi-source input set includes: The system acquires raw data from the data source of the data center cooling system and associates the raw data with the corresponding data source identifier and acquisition time sequence information to obtain the multi-source information. The data source includes at least structured operation data characterizing the data center environment and the operating status of the cooling equipment, as well as unstructured operation and maintenance information characterizing the data center operation knowledge. The structured runtime data is formatted and time-series aligned, and abnormal records in the structured runtime data are filtered out or corrected to obtain structured preprocessed data. The unstructured operation and maintenance information is parsed and standardized to obtain unstructured preprocessed data. The structured preprocessed data and the unstructured preprocessed data are then aggregated to obtain the multi-source input set.

[0006] In one embodiment, the step of constructing a decision context based on the multi-source input set according to a preset semantic organization template, and triggering an optimization task based on the decision context, includes: Based on the multi-source input set, structured operational data and unstructured operation and maintenance information are identified, and the structured operational data and unstructured operation and maintenance information are mapped to fields according to the information items required by the preset semantic organization template to obtain a set of context elements. Based on the set of context elements, the set of context elements is semantically integrated according to the preset semantic organization template to generate a unified decision context; The decision context is triggered based on preset trigger conditions, and the optimization task is triggered based on the decision context when the preset trigger conditions are met.

[0007] In one embodiment, the step of performing credibility assessment and correction processing on key operational data in the decision context to obtain a corrected decision context, and then performing task decomposition and task allocation on the optimization task based on the corrected decision context to obtain a set of candidate control suggestions, includes: Key operational data is extracted from the decision context, and the credibility of the key operational data is evaluated to generate an evaluation result. Based on the evaluation result, the key operational data is corrected, and the corrected key operational data is written back to the decision context to obtain a corrected decision context. Based on the correction decision context and the task objective corresponding to the optimization task, the optimization task is decomposed to obtain the task decomposition result, and task allocation information is generated according to the task decomposition result. Based on the task allocation information, the correction decision context is provided to the expertise processing unit corresponding to the task decomposition result, and the preliminary control suggestions output by the expertise processing unit based on the correction decision context are obtained. The preliminary control suggestions are then aggregated to obtain the candidate control suggestion set.

[0008] In one embodiment, the step of summarizing and resolving conflicts of the candidate control proposal set based on preset priority arbitration rules to obtain the target control strategy includes: Obtain each candidate control suggestion from the candidate control suggestion set, and perform normalization processing on each candidate control suggestion to obtain a candidate strategy element set; The candidate strategy element set is aggregated to generate an aggregated result, wherein the aggregated result is used to characterize the compatibility and conflict relationships among the candidate control suggestions. Based on the summarized results, conflict resolution is performed on the candidate strategy elements with conflicting relationships according to the preset priority arbitration rules, and the candidate strategy elements after conflict resolution are used to synthesize strategies to obtain the target control strategy.

[0009] In one embodiment, the step of parsing the target control strategy into structured control instructions and issuing the structured control instructions to the corresponding controller when the boundary check passes includes: Based on the target control strategy, strategy elements for characterizing the controlled object and control actions are extracted, and the strategy elements are filled and encoded according to a preset instruction data structure to obtain the structured control instruction. The structured control instructions are subjected to rule compliance checks to generate boundary verification results, wherein the rule compliance checks are used to determine whether the structured control instructions meet preset security rules and / or preset operating rules; When the boundary verification result indicates that the boundary verification is successful, the structured control command is encapsulated according to a preset industrial communication protocol and sent to the controller corresponding to the controlled object identifier in the structured control command.

[0010] Furthermore, to achieve the above objectives, this application also proposes a multi-objective optimization device for a computer room cooling system, the multi-objective optimization device for the computer room cooling system comprising: The preprocessing module is used to acquire multi-source information from the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set. The optimization triggering module is used to construct a decision context based on the multi-source input set according to a preset semantic organization template, and to trigger an optimization task based on the decision context; The candidate control module is used to perform credibility assessment and correction processing on the key operational data in the decision context to obtain a correction decision context, and to perform task decomposition and task allocation on the optimization task based on the correction decision context to obtain a set of candidate control suggestions. The control strategy module is used to summarize and resolve conflicts of the candidate control suggestion set based on preset priority arbitration rules to obtain the target control strategy; The target module is used to parse the target control strategy into structured control instructions and send the structured control instructions to the corresponding controller when the boundary check passes.

[0011] Furthermore, to achieve the above objectives, this application also proposes a multi-objective optimization device for a computer room cooling system. The device includes: a memory, a processor, and a multi-objective optimization program for the computer room cooling system stored in the memory and executable on the processor. The multi-objective optimization program for the computer room cooling system is configured to implement the steps of the multi-objective optimization method for the computer room cooling system as described in any of the above embodiments.

[0012] In addition, to achieve the above objectives, this application also proposes a storage medium storing a multi-objective optimization program for a computer room cooling system, wherein when the multi-objective optimization program for the computer room cooling system is executed by a processor, it implements the steps of the multi-objective optimization method for the computer room cooling system as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the multi-objective optimization method for the computer room cooling system described above.

[0014] This application obtains multi-source information from the computer room cooling system and preprocesses it to obtain a multi-source input set. Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context. The key operational data in the decision context is evaluated for credibility and corrected to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. The set of candidate control suggestions is summarized and conflict resolved based on preset priority arbitration rules to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and the structured control instructions are issued to the corresponding controller when the boundary verification passes. This application acquires and preprocesses multi-source information from the computer room cooling system to form a multi-source input set. Based on this, a unified decision context is constructed according to a preset semantic organization template, triggering optimization tasks and ensuring that optimization is based on consistent data semantics. Then, the key operational data in the decision context is assessed for reliability and corrected to obtain a corrected decision context, reducing deviation control caused by abnormal data. Based on the corrected decision context, the optimization task is decomposed and assigned to generate a set of candidate control suggestions. Subsequently, the set of candidate control suggestions is summarized and conflict resolved according to preset priority arbitration rules to obtain a consistent target control strategy. Finally, the target control strategy is parsed into structured control instructions and, after passing boundary verification, issued to the corresponding controllers. This reduces energy waste caused by ineffective adjustments, conflict oscillations, and execution mismatches, thereby improving the overall energy efficiency of the computer room cooling system. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the first embodiment of the multi-objective optimization method for the computer room cooling system of this application; Figure 2 This is a schematic diagram of a sub-process in the second embodiment of the multi-objective optimization method for the computer room cooling system of this application; Figure 3 This is a schematic diagram of a sub-process in the third embodiment of the multi-objective optimization method for the computer room cooling system of this application; Figure 4 This is a flowchart illustrating the overall process of multi-objective intelligent optimization in one embodiment of the multi-objective optimization method for the computer room cooling system of this application. Figure 5 This is a schematic diagram of the module structure of the multi-objective optimization device for the computer room cooling system according to an embodiment of this application; Figure 6 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the multi-objective optimization method of the computer room cooling system in the embodiments of this application.

[0016] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0018] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0019] It should be noted that data center server rooms are typically equipped with cooling equipment such as water-cooled terminal air conditioners to regulate and control the server room environment and equipment operating status. This control falls under the data center infrastructure management domain and focuses on optimizing the operation of the server room's water-cooled terminal air conditioning system. Existing server room water-cooled terminal air conditioning control strategies are mostly based on preset fixed operating parameters such as return air temperature and fan speed, or use traditional optimization algorithms to dynamically adjust finite objectives. During server room operation, in addition to real-time sensor and equipment operating parameters, there is also unstructured information such as server room CAD layout diagrams, maintenance logs, future load plans, and special settings in equipment manuals. However, the aforementioned existing methods often struggle to effectively understand and utilize this type of unstructured information, and lack global perception and reasoning support for complex and dynamic environments. Under multi-objective constraints, it is difficult to form consistent control decisions, resulting in poor overall energy efficiency of the server room cooling system. Therefore, how to improve the overall energy efficiency of server room cooling systems has become an urgent technical problem to be solved.

[0020] The main solution of this application is as follows: First, acquire multi-source information from the computer room cooling system and preprocess it to obtain a multi-source input set. Second, construct a decision context based on the multi-source input set according to a preset semantic organization template, and trigger optimization tasks based on the decision context. Third, perform credibility assessment and correction processing on key operational data in the decision context to obtain a corrected decision context, and decompose and assign optimization tasks based on the corrected decision context to obtain a set of candidate control suggestions. Fourth, summarize and resolve conflicts in the set of candidate control suggestions based on preset priority arbitration rules to obtain the target control strategy. Fifth, parse the target control strategy into structured control instructions and issue structured control instructions to the corresponding controller when the boundary verification passes.

[0021] This application acquires and preprocesses multi-source information from the computer room cooling system to form a multi-source input set. Based on this, a unified decision context is constructed according to a preset semantic organization template, triggering optimization tasks and ensuring that optimization is based on consistent data semantics. Then, the key operational data in the decision context is assessed for reliability and corrected to obtain a corrected decision context, reducing deviation control caused by abnormal data. Based on the corrected decision context, the optimization task is decomposed and assigned to generate a set of candidate control suggestions. Subsequently, the set of candidate control suggestions is summarized and conflict resolved according to preset priority arbitration rules to obtain a consistent target control strategy. Finally, the target control strategy is parsed into structured control instructions and, after passing boundary verification, issued to the corresponding controllers. This reduces energy waste caused by ineffective adjustments, conflict oscillations, and execution mismatches, thereby improving the overall energy efficiency of the computer room cooling system.

[0022] It should be noted that the executing entity of the method in this embodiment can be a computing service device with data processing, network communication, and program execution functions, or it can be a multi-objective optimization device of the aforementioned data center cooling system with the same or similar functions. This embodiment and the following embodiments will be described using a multi-objective optimization device of a data center cooling system as an example.

[0023] Based on this, a first embodiment of the multi-objective optimization method for the computer room cooling system of this application is proposed. Please refer to [the relevant documentation]. Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the multi-objective optimization method for the computer room cooling system of this application.

[0024] In this embodiment, the multi-objective optimization method for the computer room cooling system includes the following steps: S1: Obtain multi-source information from the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set; S2: Based on the multi-source input set, construct a decision context according to a preset semantic organization template, and trigger an optimization task based on the decision context; It should be noted that the data center cooling system is a collection of cooling equipment and its controlled objects used to regulate the temperature environment within the data center. Multi-source information is a collection of data center operation-related information from different data sources and of different types. The preset semantic organization template is a pre-defined template used to organize and express the multi-source input set according to a predetermined semantic structure. The multi-source input set is a data set formed by the convergence of preprocessed structured operational data and unstructured maintenance information, which can be directly used for subsequent semantic organization and task triggering. The decision context is a unified semantic information carrier obtained by integrating the multi-source input set according to the preset semantic organization template. Triggering the optimization task is based on the decision context meeting preset triggering methods or triggering conditions, initiating an optimization process oriented towards cooling control.

[0025] Specifically, during the operation of the data center cooling system, multi-source information related to its operation is collected. This multi-source information includes at least: structured operational data characterizing the data center environment and the operating status of the cooling equipment, and unstructured maintenance information characterizing data center operational knowledge. The structured operational data may include temperature sensor readings, air conditioning-related operating parameters, and load-related data from the equipment and service sides. The unstructured maintenance information may include data center layout diagrams, historical maintenance logs, and special setting scenarios in equipment manuals. After collecting the multi-source information, preprocessing is performed. This preprocessing includes at least standardizing the structured operational data to meet unified organization requirements and parsing and structuring the unstructured maintenance information to meet usability requirements. The preprocessed structured operational data and the preprocessed unstructured maintenance information are then aggregated to obtain a multi-source input set.

[0026] Furthermore, after obtaining the multi-source input set, the multi-source input set is semantically organized and integrated according to a preset semantic organization template. Operating status elements related to cooling control, maintenance knowledge elements, and information elements related to task triggering are embedded into the preset semantic organization template to form a unified, semantically rich decision context. Subsequently, a task trigger determination is performed based on the decision context, and an optimization task is initiated when a preset triggering method or triggering condition is met. This allows the optimization task to be triggered and executed based on the decision context as input, thereby achieving "constructing a decision context based on the multi-source input set according to the preset semantic organization template, and triggering an optimization task based on the decision context."

[0027] By collecting and preprocessing multi-source information to form a multi-source input set, subsequent optimization no longer relies on a single source or unstandardized data format, but is based on inputs that can be uniformly invoked. Then, according to a preset semantic organization template, the multi-source input set is integrated into a decision context with unified semantics, so that multi-source status elements and operation and maintenance knowledge elements are organized in a consistent structure and can be directly used for subsequent processing. Optimization tasks are triggered based on this decision context, so that optimization tasks are started when preset triggering methods or triggering conditions are met and the same decision context is used as input. This reduces ineffective optimization and control deviations caused by information dispersion, semantic inconsistency, or trigger mismatch, and provides a data and semantic foundation for the subsequent formation of control strategies that are more in line with the actual state of the data center.

[0028] S3: Perform credibility assessment and correction processing on the key operational data in the decision context to obtain a corrected decision context, and perform task decomposition and task allocation on the optimization task based on the corrected decision context to obtain a set of candidate control suggestions; It should be noted that key operational data refers to operational data that has a critical impact on control decisions within the decision-making context. Credibility assessment is the process of analyzing and judging the credibility of key operational data. Correction processing (error correction) is the process of correcting or refining key operational data based on the credibility assessment results. The optimization task is a system-triggered optimization processing task for computer room cooling control. Task decomposition involves breaking down the optimization task into separately processable subtasks based on the task objective and the current context. Task allocation assigns the subtasks obtained from task decomposition to matching specialized processing units / specialized agents for execution. The candidate control suggestion set is a collection of preliminary control suggestions output by each specialized processing unit / specialized agent after reasoning based on a shared context.

[0029] Specifically, after the optimization task enters the reasoning and decision-making stage, key operational data (such as key sensor data) is identified and extracted from the decision context. A credibility assessment is performed on the key operational data to determine whether there are abnormal fluctuations, missing data, or deviations that affect the reliability of the decision. Subsequently, based on the credibility assessment results, a correction process is performed on the key operational data, and the corrected key operational data is backfilled into the decision context to obtain a corrected decision context.

[0030] Furthermore, after obtaining the correction decision context, the optimization task is decomposed by combining the task objective of the optimization task with the current system state represented by the correction decision context, generating task allocation information corresponding to each sub-task; then, according to the task allocation information, the specialized processing unit / specialized agent matching the sub-task is called, so that it performs reasoning based on the shared correction decision context and outputs preliminary control suggestions respectively, and finally, the preliminary control suggestions are aggregated to form a candidate control suggestion set.

[0031] By performing credibility assessment and correction on key operational data in the decision-making context during the decision-making stage, control bias caused by abnormal or unreliable data entering the subsequent optimization inference chain can be reduced. This allows subsequent optimization tasks to use a more reliable corrected decision context as a unified input basis. Then, based on this corrected decision context, optimization tasks are decomposed and assigned, allowing different sub-tasks to infer and output control suggestions by matching specialized processing units under the same context. This converges multiple suggestions into a candidate control suggestion set, providing a more sufficient and consistent data and suggestion basis for subsequent aggregation arbitration and conflict resolution, thereby improving the stability and executability of the output control strategy.

[0032] S4: Based on preset priority arbitration rules, the candidate control suggestion set is summarized and conflict resolution is performed to obtain the target control strategy; S5: Parse the target control strategy into structured control instructions, and send the structured control instructions to the corresponding controller when the boundary check passes.

[0033] It should be noted that the candidate control suggestion set is the aggregated result of preliminary control suggestions output by multiple specialized processing units / agents based on a shared decision context. The preset priority arbitration rule is used to select and synthesize suggestions according to a predetermined priority order when multiple candidate suggestions exist and may be inconsistent. Conflict resolution refers to the process of selecting, coordinating, or replacing incompatible or contradictory suggestions in the candidate control suggestion set according to the preset priority arbitration rule, thereby eliminating conflicts. The target control strategy is the final consistent control strategy synthesized after aggregation and conflict resolution. The structured control instruction is structured instruction data that can be executed by the air conditioning system / cooling equipment, obtained by parsing and converting the target control strategy. Boundary verification refers to performing rule compliance checks or security interception checks on the structured control instructions before issuing them to determine whether the instructions meet preset safety or operational rules. The controller is the execution control unit corresponding to equipment such as water-cooled terminal air conditioners.

[0034] Specifically, after obtaining the set of candidate control suggestions, the suggestions in the set are first aggregated, and the information such as the controlled objects, control actions and constraints involved in each suggestion are uniformly organized to form a comparable and arbitrable aggregated result. Then, the aggregated result is arbitrated according to the preset priority arbitration rules, and conflict resolution is performed on the conflict-contradictory or incompatible candidate suggestions. The candidate suggestions retained after conflict resolution are then used for strategy synthesis to obtain the final consistent target control strategy.

[0035] Furthermore, after obtaining the target control strategy, the target control strategy is parsed and converted into structured control instructions through the instruction generation process, so that it meets the data organization form that can be executed on the equipment side; before the instructions are issued, boundary verification is performed on the structured control instructions to determine whether they comply with preset safety rules and / or preset operating rules; when the boundary verification passes, the structured control instructions are issued to the controller corresponding to the structured control instructions through the industrial communication protocol to drive the corresponding cooling equipment to perform operating setpoint adjustment.

[0036] By summarizing the set of candidate control recommendations and resolving conflicts according to preset priority arbitration rules, recommendations from multiple sources can be converged into a consistent target control strategy, reducing control cancellation or repeated adjustments caused by inconsistent recommendations. Furthermore, the target control strategy is parsed into structured control instructions and issued to the corresponding controllers after boundary verification, so that the control strategy can be stably implemented in the form of executable instructions, while preventing instructions that do not conform to preset safety or operating rules from entering the execution stage, thereby improving the consistency, executability and controllability of control output.

[0037] This embodiment acquires multi-source information from the computer room cooling system and preprocesses it to obtain a multi-source input set. Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context. The key operational data in the decision context is evaluated for reliability and corrected to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. The set of candidate control suggestions is summarized and conflict resolved based on preset priority arbitration rules to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and these instructions are sent to the corresponding controllers when boundary checks pass. This embodiment acquires and preprocesses multi-source information from the computer room cooling system to form a multi-source input set. Based on this, a unified decision context is constructed according to a preset semantic organization template, triggering optimization tasks and ensuring that optimization is based on consistent data semantics. Then, the key operational data in the decision context is assessed for reliability and corrected to obtain a corrected decision context, reducing deviation control caused by abnormal data. Based on the corrected decision context, the optimization task is decomposed and assigned to generate a set of candidate control suggestions. Subsequently, the candidate control suggestions are summarized and conflict resolved according to preset priority arbitration rules to obtain a consistent target control strategy. Finally, the target control strategy is parsed into structured control instructions and, after passing boundary verification, issued to the corresponding controller. This reduces energy waste caused by ineffective adjustments, conflict oscillations, and execution mismatches, thereby improving the overall energy efficiency of the computer room cooling system.

[0038] Based on the first embodiment described above, a second embodiment of the multi-objective optimization method for the computer room cooling system of this application is proposed. Please refer to... Figure 2 , Figure 2 This is a schematic diagram of a sub-process in the second embodiment of the multi-objective optimization method for the computer room cooling system of this application.

[0039] like Figure 2 As shown, in this embodiment, step S1 includes: S11: Obtain raw data from the data source of the computer room cooling system, and associate the raw data with the corresponding data source identifier and collection time sequence information to obtain the multi-source information. The data source includes at least structured operation data characterizing the computer room environment and the operating status of the cooling equipment, as well as unstructured operation and maintenance information characterizing the computer room operation knowledge. S12: Perform format unification and time sequence alignment processing on the structured running data, and filter or correct abnormal records in the structured running data to obtain structured preprocessed data. S13: The unstructured operation and maintenance information is parsed and standardized to obtain unstructured preprocessed data, and the structured preprocessed data and the unstructured preprocessed data are aggregated to obtain the multi-source input set.

[0040] It should be noted that a data source is an entity that can generate or provide information related to data center operations. Raw data is data directly collected from the data source and has not yet undergone cleaning, alignment, or standardization. Collection time-series information is information used to characterize the time-series attributes of the raw data, such as the collection time point, time sequence, or timestamp. Structured operational data is data related to the operational status of the data center environment and cooling equipment, expressed in a structured form such as fields and records. Unstructured operation and maintenance information is operation and maintenance-related information that is not presented in a fixed field structure but is used to characterize knowledge about data center operations. Time-series alignment is the alignment of structured operational data from different collection periods or different time bases along the time dimension. Structured preprocessed data is the result data obtained after structured operational data has undergone format standardization, time-series alignment, and anomaly handling. Parsing and standardization processing involves parsing the content of unstructured operation and maintenance information, extracting elements, and unifying its expression to give it an organized and aggregateable form.

[0041] Specifically, raw data is acquired from various data sources in the data center cooling system, and each raw data is associated with a data source identifier and acquisition time sequence information. This ensures that each data entry has a traceable source attribute and an aligned time attribute, thus forming multi-source information. This multi-source information includes at least structured operational data characterizing the data center environment and the operating status of the cooling equipment, as well as unstructured maintenance information characterizing data center operational knowledge. Subsequently, the structured operational data undergoes format standardization processing to ensure uniformity in field definitions, data types, and representation methods. Further time sequence alignment processing is performed to ensure consistent alignment of structured operational data from different sources along the time dimension. Based on this, abnormal records in the structured operational data are filtered out or corrected to obtain structured preprocessed data, meeting the input quality requirements for subsequent aggregation and retrieval.

[0042] Furthermore, after completing the structured preprocessing, the unstructured operation and maintenance information undergoes parsing and standardization. Key content elements within the unstructured information are analyzed, extracted, and standardized to obtain unstructured preprocessed data, enabling this information to participate in subsequent processing in an organized and associative form. Finally, the structured and unstructured preprocessed data are aggregated: during the aggregation process, relationships between data are established based on data source identifiers and acquisition time-series information, and the aggregation results are uniformly encapsulated and organized to obtain a multi-source input set.

[0043] By associating raw data with data source identifiers and collection time-series information to form multi-source information, subsequent processing is based on traceability and time alignment. Then, structured operational data undergoes format unification and time-series alignment, and abnormal records are filtered or corrected to obtain structured preprocessed data, ensuring that the structured data meets subsequent usage requirements in terms of representation consistency, time consistency, and input reliability. Simultaneously, unstructured operational information is parsed and standardized to obtain unstructured preprocessed data, giving unstructured information an organized and associative expression. Finally, the structured and unstructured preprocessed data are converged to obtain a multi-source input set, providing a consistent, usable, and associative input foundation for subsequent unified organization and decision-making based on the multi-source input set.

[0044] Based on the first embodiment described above, in this embodiment, step S2 includes: S21: Based on the multi-source input set, structured operation data and unstructured operation and maintenance information are identified, and the structured operation data and unstructured operation and maintenance information are mapped to fields according to the information items required by the preset semantic organization template to obtain a set of context elements; S22: Based on the set of context elements, semantically integrate the set of context elements according to the preset semantic organization template to generate a unified decision context; S23: Trigger the decision context based on preset trigger conditions, and trigger the optimization task based on the decision context when the preset trigger conditions are met.

[0045] It should be noted that information items are semantic fields / content elements defined in the preset semantic organization template, used to construct the decision context. Field mapping is the process of matching and transforming structured operational data fields and unstructured operational information elements from a multi-source input set according to the information items in the preset semantic organization template. The context element set is the set of elements obtained after field mapping that can be directly organized by the template. Semantic integration is the process of combining, associating, and organizing the context element set according to the preset semantic organization template to form a unified semantic structure. Trigger determination is the process of comparing the decision context with preset trigger conditions to determine whether the trigger requirements are met. Preset trigger conditions are a pre-defined set of conditions used to determine whether to start the optimization task.

[0046] Specifically, after obtaining the multi-source input set, the system first performs content recognition and classification to distinguish between structured operational data and unstructured operation and maintenance information. Then, it reads the set of information items defined in a preset semantic organization template and determines the required data source and expression form for each information item. Based on this, field-level mapping processing is performed on the structured operational data: fields corresponding to the information items are extracted, renamed, type-converted, or unit-consistentized to form structured elements that can be directly filled into the template. Simultaneously, element-level mapping processing is performed on the unstructured operation and maintenance information: operation and maintenance knowledge elements matching the information items are extracted from the unstructured operation and maintenance information and expressed in a standardized manner. Finally, the structured elements and operation and maintenance knowledge elements are summarized according to the information item dimension to obtain a context element set.

[0047] Furthermore, after obtaining the set of context elements, the set is semantically integrated according to a preset semantic organization template: the elements corresponding to each information item are combined and organized according to the hierarchical structure, relationships, and expression methods specified in the template to generate a unified decision context, ensuring that the decision context maintains semantic consistency and that information elements have referential relationships. Subsequently, the decision context is triggered based on preset triggering conditions: triggering elements for judgment are extracted from the decision context and matched with the preset triggering conditions; when the judgment result meets the preset triggering conditions, the optimization task is triggered using the decision context as the input basis for the optimization task.

[0048] By identifying structured operational data and unstructured maintenance information from a multi-source input set, and performing field mapping on the two types of information according to the information items of a preset semantic organization template to obtain a context element set, information from different sources and with different forms of expression can be extracted and organized in a unified element form required by the template. Then, based on the context element set, semantic integration is performed according to the preset semantic organization template to generate a unified decision context, so that subsequent processing faces an information carrier with consistent structure, relatable elements, and referentiality. Furthermore, the decision context is triggered based on preset trigger conditions, and the optimization task is triggered when the conditions are met. The optimization task is only started when the decision context has the information elements required for triggering and meets the trigger requirements, and the same decision context is used as the input basis, thereby reducing invalid processing and decision bias caused by misaligned information, inconsistent semantics, or mismatched triggers.

[0049] This embodiment acquires multi-source information from the computer room cooling system and preprocesses it to obtain a multi-source input set. Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context. The key operational data in the decision context is evaluated for reliability and corrected to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. The set of candidate control suggestions is summarized and conflict resolved based on preset priority arbitration rules to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and these instructions are sent to the corresponding controllers when boundary checks pass. This embodiment acquires and preprocesses multi-source information from the computer room cooling system to form a multi-source input set. Based on this, a unified decision context is constructed according to a preset semantic organization template, triggering optimization tasks and ensuring that optimization is based on consistent data semantics. Then, the key operational data in the decision context is assessed for reliability and corrected to obtain a corrected decision context, reducing deviation control caused by abnormal data. Based on the corrected decision context, the optimization task is decomposed and assigned to generate a set of candidate control suggestions. Subsequently, the candidate control suggestions are summarized and conflict resolved according to preset priority arbitration rules to obtain a consistent target control strategy. Finally, the target control strategy is parsed into structured control instructions and, after passing boundary verification, issued to the corresponding controller. This reduces energy waste caused by ineffective adjustments, conflict oscillations, and execution mismatches, thereby improving the overall energy efficiency of the computer room cooling system.

[0050] Based on the second embodiment described above, a third embodiment of the multi-objective optimization method for the computer room cooling system of this application is proposed. Please refer to... Figure 3 , Figure 3 This is a schematic diagram of a sub-process in the third embodiment of the multi-objective optimization method for the computer room cooling system of this application.

[0051] In this embodiment, step S3 includes: S31: Extract key operational data from the decision context, perform credibility assessment on the key operational data to generate assessment results, perform error correction processing on the key operational data based on the assessment results, and write back the error-corrected key operational data to the decision context to obtain a corrected decision context; S32: Based on the correction decision context and the task objective corresponding to the optimization task, the optimization task is decomposed to obtain the task decomposition result, and task allocation information is generated according to the task decomposition result; S33: Based on the task allocation information, provide the correction decision context to the expertise processing unit corresponding to the task decomposition result, obtain the preliminary control suggestions output by the expertise processing unit based on the correction decision context, and aggregate the preliminary control suggestions to obtain the candidate control suggestion set.

[0052] It should be noted that credibility assessment is the process of analyzing and judging the reliability, validity, or consistency of key operational data. The assessment result is the output information of the credibility assessment. The corrective decision context is a decision context containing the corrected key operational data. Task allocation information is the allocation result information used to indicate the correspondence between each subtask and its corresponding execution unit. A specialized processing unit is an execution unit that matches a specific subtask and possesses the corresponding processing capabilities. Preliminary control recommendations are the control recommendations output by the specialized processing unit after reasoning about its responsible subtask based on the corrective decision context.

[0053] Specifically, after obtaining the decision context, key operational data is extracted from it. This key operational data can be selected based on a preset set of key fields, data importance markers, or relevance to the task objective. Subsequently, a credibility assessment is performed on the key operational data to generate an assessment result. This credibility assessment may include consistency checks, rationality verification, missing data identification, or abnormal fluctuation identification to determine whether the key operational data meets the input requirements for subsequent optimization calculations. After generating the assessment result, error correction processing is performed on the key operational data based on the assessment result. This error correction processing may include correcting outliers, completing missing values, or unifying inconsistent fields. The corrected key operational data is then written back to the decision context to obtain a corrected decision context.

[0054] Furthermore, after obtaining the correction decision context, the optimization task is decomposed based on the task objective corresponding to the optimization task to obtain task decomposition results. These results characterize the processing scope, input dependencies, and output requirements of each sub-task. Task allocation information is then generated based on the task decomposition results to indicate the correspondence between each sub-task and its corresponding specialized processing unit. Subsequently, the correction decision context is provided to the specialized processing unit corresponding to the task decomposition results based on the task allocation information. This allows each specialized processing unit to process its corresponding sub-task with a consistent context input and output preliminary control suggestions. Finally, the preliminary control suggestions output by each specialized processing unit are aggregated to obtain a candidate control suggestion set for subsequent summarization and conflict resolution processing.

[0055] By first extracting key operational data from the decision context and conducting a credibility assessment, anomalies, missing data, or inconsistencies in the key data can be identified before task planning and suggestion generation. Based on the assessment results, the key operational data is corrected and written back into the decision context, thus forming a more reliable and consistent corrected decision context. Then, based on the corrected decision context and task objectives, the optimization task is decomposed and assigned, so that each sub-task generates preliminary control suggestions by matching specialized processing units on the same corrected context. The multiple suggestions are then aggregated to form a candidate control suggestion set, providing a more sufficient and consistent input basis for subsequent aggregation arbitration and conflict resolution, thereby reducing the risk of control deviations and decision inconsistencies caused by key data deviations or insufficient single-path suggestions.

[0056] Based on the second embodiment described above, in this embodiment, step S4 includes: S41: Obtain each candidate control suggestion in the candidate control suggestion set, and perform normalization processing on each candidate control suggestion to obtain a candidate strategy element set; S42: Perform a summary process based on the set of candidate strategy elements to generate a summary result, wherein the summary result is used to characterize the compatibility and conflict relationships among the candidate control suggestions; S43: Based on the summary results, conflict resolution is performed on the candidate strategy elements with conflicting relationships according to the preset priority arbitration rules, and the candidate strategy elements after conflict resolution are combined to obtain the target control strategy.

[0057] It should be noted that a candidate control suggestion is a single suggestion item in the set of candidate control suggestions. Normalization is the process of converting candidate control suggestions from different sources and with different expression formats into a unified data structure and a unified semantic definition. Compatibility refers to a relationship between candidate strategy elements that is not mutually exclusive in terms of controlled objects, control actions, or constraints, and can be simultaneously established or combined for execution. Conflict relationships refer to a relationship between candidate strategy elements that is mutually exclusive or contradictory in terms of controlled objects, control actions, or constraints; if used simultaneously, it will result in inconsistent or unexecutable relationships. Preset priority arbitration rules are a set of pre-defined rules used to select and decide on candidate strategy elements when conflict relationships exist. Strategy synthesis is the process of combining and assembling the candidate strategy elements retained after conflict resolution according to consistency and executability requirements to form the final control strategy.

[0058] Specifically, after obtaining the set of candidate control suggestions, each suggestion is retrieved individually, and normalization is performed on each suggestion to eliminate differences in expression, field definitions, and semantic phrasing. Normalization may include: converting the controlled object identifiers, control action types, control parameter items, and their constraint descriptions involved in the suggestion into unified fields; merging synonymous action descriptions into unified action semantics; and converting comparable parameter expressions into unified types and representations. After normalization, each candidate control suggestion is broken down or refined into arbitrable and composable strategy elements, which are then aggregated to form a set of candidate strategy elements, providing unified input for subsequent relationship determination and strategy synthesis.

[0059] Furthermore, a summary process is performed based on the candidate strategy element set to generate a summary result. During the summary process, strategy elements are merged and associated according to the controlled object dimension, control action dimension, or constraint dimension, and compatibility and conflict are determined between strategy elements, thus forming a structured representation of compatible and conflicting relationships in the summary result. Subsequently, based on the summary result, conflict resolution is performed on candidate strategy elements with conflicting relationships according to preset priority arbitration rules: conflict items are adjudicated and a set of strategy elements after conflict resolution is output; then, strategy synthesis is performed on the conflict-resolved strategy elements, combining compatible elements and unifying and integrating multiple elements on the same controlled object, ultimately obtaining the target control strategy.

[0060] By first standardizing each candidate control suggestion in the candidate control suggestion set and forming a candidate strategy element set, suggestions from multiple sources can be converged into arbitrable units with unified semantics and structure, reducing the incomparability or uncombinability caused by inconsistent expression. Then, based on the candidate strategy element set, a summary result representing compatible and conflicting relationships is generated, so that conflict determination has a clear basis and can be processed by subsequent rules. Furthermore, conflict relationships are resolved according to preset priority arbitration rules, and the resolved strategy elements are synthesized into strategies, which can converge multiple suggestions from a state of "parallel but potentially contradictory" to a "consistent and combinable" target control strategy, thereby reducing the risk of mutual cancellation or unexecutability between strategies, and providing a stable and unified strategy input basis for subsequent command issuance.

[0061] Based on the second embodiment described above, in this embodiment, step S5 includes: S51: Based on the target control strategy, extract strategy elements to characterize the controlled object and control action, and fill and encode the strategy elements according to the preset instruction data structure to obtain the structured control instruction; S52: Perform a rule compliance check on the structured control instructions to generate a boundary verification result, wherein the rule compliance check is used to determine whether the structured control instructions meet the preset security rules and / or preset operation rules; S53: When the boundary verification result indicates that the boundary verification is passed, the structured control command is encapsulated according to a preset industrial communication protocol and sent to the controller corresponding to the controlled object identifier in the structured control command.

[0062] It should be noted that field population is the process of writing the corresponding information from the strategy elements into the fields of the preset instruction data structure. Rule compliance checking is the process of verifying whether structured control instructions meet preset rules. Boundary verification results are the output information of the rule compliance check. Preset security rules are a set of rules pre-defined to constrain the security of instructions. Preset operation rules are a set of rules pre-defined to constrain the operability and compliance of instructions. Preset industrial communication protocols are pre-defined data communication protocol specifications used to encapsulate and transmit control instructions between the system and the controller.

[0063] Specifically, after obtaining the target control strategy, strategy elements representing the controlled object and control actions are first extracted from the target control strategy. These strategy elements include at least the controlled object identifier, the control action type, and control field information associated with the control action. Subsequently, the field set and field constraints defined in the preset instruction data structure are read, and the strategy elements are filled in according to the semantic correspondence of the fields, so that the strategy elements can be expressed in a unified field format specified by the instruction data structure. After the field filling is completed, the filled fields are encoded according to preset encoding rules to obtain structured control instructions, giving the structured control instructions a unified data representation that is transmissible and parsable.

[0064] Furthermore, upon receiving the structured control command, a rule compliance check is performed on the structured control command to generate a boundary verification result. This rule compliance check determines whether the structured control command meets preset safety rules and / or preset operating rules, and outputs a verification conclusion indicating pass or fail. If the boundary verification result indicates pass, the structured control command is encapsulated according to a preset industrial communication protocol to form a command message that meets the protocol specifications. This command message is then sent to the controller corresponding to the controlled object identifier in the structured control command, enabling the controller to perform control actions on the controlled object based on the structured control command.

[0065] By extracting the policy elements corresponding to the controlled object and control action from the target control strategy, and generating structured control instructions by filling in and encoding fields according to the preset instruction data structure, the control intent at the policy level can be converted into a unified instruction expression that the controller can parse and execute. Then, the structured control instructions are checked for rule compliance and boundary verification results are generated, so that instructions that do not meet the preset safety rules and / or preset operating rules are identified and intercepted before being issued, thereby reducing the risk of non-compliant instructions entering the execution stage. When the boundary verification is passed, the instructions are encapsulated according to the preset industrial communication protocol and sent to the corresponding controller, so that the instructions are stably transmitted and executed in a form that conforms to the communication specifications, thereby improving the executability, controllability and consistency between the control strategy and the equipment execution.

[0066] This embodiment acquires multi-source information from the computer room cooling system and preprocesses it to obtain a multi-source input set. Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context. The key operational data in the decision context is evaluated for reliability and corrected to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. The set of candidate control suggestions is summarized and conflict resolved based on preset priority arbitration rules to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and these instructions are sent to the corresponding controllers when boundary checks pass. This embodiment acquires and preprocesses multi-source information from the computer room cooling system to form a multi-source input set. Based on this, a unified decision context is constructed according to a preset semantic organization template, triggering optimization tasks and ensuring that optimization is based on consistent data semantics. Then, the key operational data in the decision context is assessed for reliability and corrected to obtain a corrected decision context, reducing deviation control caused by abnormal data. Based on the corrected decision context, the optimization task is decomposed and assigned to generate a set of candidate control suggestions. Subsequently, the candidate control suggestions are summarized and conflict resolved according to preset priority arbitration rules to obtain a consistent target control strategy. Finally, the target control strategy is parsed into structured control instructions and, after passing boundary verification, issued to the corresponding controller. This reduces energy waste caused by ineffective adjustments, conflict oscillations, and execution mismatches, thereby improving the overall energy efficiency of the computer room cooling system.

[0067] Please see Figure 4 , Figure 4 This is a flowchart illustrating the overall process of multi-objective intelligent optimization in one embodiment of the multi-objective optimization method for the computer room cooling system of this application. Figure 4 As shown, in one embodiment, the specific steps include: Step 1: Multi-source data acquisition and preprocessing. The system continuously collects multimodal data from the data center, including real-time sensor data such as temperature sensor readings for all cold and hot aisles, return air temperature, supply air temperature, fan speed, and water valve opening of air conditioners; equipment and business data such as rack utilization rate and real-time and predicted load of IT equipment; and unstructured knowledge such as data center CAD layout diagrams, historical operation and maintenance logs, and special settings scenarios in equipment manuals.

[0068] Step 2: Constructing the Decision Context and Task Triggering. The structured data and unstructured knowledge processed in Step 1 are integrated into a unified, semantically rich decision context according to a pre-defined prompt template. This context construction can be handled by a dedicated context management agent. The system triggers decision tasks periodically or based on specific events.

[0069] Step 3: Agent Collaborative Reasoning and Decision-Making. The decision-making process is activated. First, the data verification agent assesses the credibility and corrects errors in the key sensor data within the current context. Then, the scheduling agent decomposes the task based on the task objective and the current context, and invokes the corresponding specialized agents. Next, the cluster of specialized agents reason from their respective areas of expertise based on the shared context, outputting preliminary control suggestions and natural language explanations.

[0070] Step 4: Policy Arbitration and Instruction Generation. The arbitration mechanism aggregates the outputs of each agent and resolves conflicts according to preset priority rules, synthesizing a final, consistent control policy. Then, the instruction generation module parses the arbitrated natural language policy and converts it into structured control instructions executable by the air conditioning system. Instruction example: {"target_device": "ACU-01", "set_parameter": "return_air_temp", "set_value":24.5, "reason": "Balance energy efficiency and heat load"}.

[0071] Step 5: Secure Issuance and Execution of Control Commands. The control commands generated in Step 4 are sent to the corresponding water-cooled terminal air conditioning controllers via industrial communication protocols to adjust their operating setpoints. The system can be configured with a security interception layer to perform a final boundary check before the commands are issued.

[0072] Step 6: Effect Monitoring and Feedback Learning. The system continuously monitors the system status after the strategy is executed, including total power consumption, temperature field stability, and equipment operating status. This effect data is stored in association with the context of the decision-making process, forming a feedback knowledge base for subsequent prompt word iteration and model optimization.

[0073] This application also provides a multi-objective optimization device for a computer room cooling system. Please refer to... Figure 5 , Figure 5 This is a schematic diagram of the module structure of a multi-objective optimization device for a computer room cooling system according to an embodiment of this application. The multi-objective optimization device for the computer room cooling system includes: The preprocessing module 501 is used to acquire multi-source information of the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set; The optimization triggering module 502 is used to construct a decision context based on the multi-source input set according to a preset semantic organization template, and to trigger an optimization task based on the decision context; The candidate control module 503 is used to perform credibility assessment and correction processing on the key operational data in the decision context to obtain a correction decision context, and to perform task decomposition and task allocation on the optimization task based on the correction decision context to obtain a set of candidate control suggestions. Control strategy module 504 is used to summarize and resolve conflicts of the candidate control suggestion set based on preset priority arbitration rules to obtain the target control strategy; The target module 505 is used to parse the target control strategy into structured control instructions and send the structured control instructions to the corresponding controller when the boundary check passes.

[0074] The multi-objective optimization device for a data center cooling system provided in this application adopts the multi-objective optimization method for a data center cooling system in the above embodiments, and can solve the technical problem of how to improve the overall energy efficiency of a data center cooling system. Compared with the prior art, the beneficial effects of the multi-objective optimization device for a data center cooling system provided in this application are the same as the beneficial effects of the multi-objective optimization method for a data center cooling system provided in the above embodiments, and other technical features in the multi-objective optimization device for a data center cooling system are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0075] This application provides a multi-objective optimization device for a data center cooling system. The multi-objective optimization device for a data center cooling system 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 multi-objective optimization method for the data center cooling system in the above embodiments.

[0076] The following is for reference. Figure 6 , Figure 6 This is a schematic diagram of the hardware operating environment involved in the multi-objective optimization method of the computer room cooling system in the embodiments of this application. It shows a schematic diagram of the structure of the multi-objective optimization device suitable for implementing the multi-objective optimization method of the computer room cooling system in the embodiments of this application. Figure 6The multi-objective optimization device for the computer room cooling system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0077] like Figure 6 As shown, the multi-objective optimization device for the computer room cooling system may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the multi-objective optimization device for the computer room cooling system. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the multi-objective optimization equipment of the computer room cooling system to exchange data with other devices wirelessly or via wired communication. Although the figure shows a multi-objective optimization equipment for a computer room cooling system with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented or possessed alternatively.

[0078] In particular, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. When the computer program is executed by the processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0079] The multi-objective optimization device for a data center cooling system provided in this application employs the multi-objective optimization method for a data center cooling system described in the above embodiments, and can solve the technical problem of how to improve the overall energy efficiency of a data center cooling system. Compared with the prior art, the beneficial effects of the multi-objective optimization device for a data center cooling system provided in this application are the same as the beneficial effects of the multi-objective optimization method for a data center cooling system provided in the above embodiments, and other technical features in the multi-objective optimization device for a data center cooling system are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0080] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0082] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the multi-objective optimization method for the computer room cooling system in the above embodiments.

[0083] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the multi-objective optimization device of the computer room cooling system, the multi-objective optimization device of the computer room cooling system performs the following actions: acquires multi-source information from the computer room cooling system and preprocesses the multi-source information to obtain a multi-source input set; constructs a decision context based on the multi-source input set according to a preset semantic organization template, and triggers optimization tasks based on the decision context; performs credibility assessment and correction processing on key operational data in the decision context to obtain a corrected decision context, and decomposes and assigns optimization tasks based on the corrected decision context to obtain a candidate control suggestion set; summarizes and resolves conflicts in the candidate control suggestion set based on preset priority arbitration rules to obtain a target control strategy; parses the target control strategy into structured control instructions, and issues structured control instructions to the corresponding controller when the boundary check passes. Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via 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 (e.g., via the Internet using an Internet service provider).

[0084] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0085] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0086] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the multi-objective optimization method for the above-described computer room cooling system, which can solve the technical problem of how to improve the overall energy efficiency of the computer room cooling system. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the multi-objective optimization method for the computer room cooling system provided in the above embodiments, and will not be repeated here.

[0087] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-objective optimization method for a computer room cooling system as described above.

[0088] The computer program product provided in this application can solve the technical problem of how to improve the overall energy efficiency of a computer room cooling system. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the multi-objective optimization method for the computer room cooling system provided in the above embodiments, and will not be repeated here.

[0089] The above are merely preferred embodiments of this application and do not limit the scope of protection of this application. Any equivalent structural or procedural transformations made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of this application.

Claims

1. A multi-objective optimization method for a computer room cooling system, characterized in that, The method includes: Acquire multi-source information from the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set; Based on the multi-source input set, a decision context is constructed according to a preset semantic organization template, and an optimization task is triggered based on the decision context; The key operational data in the decision context are subjected to credibility assessment and correction processing to obtain a corrected decision context. Based on the corrected decision context, the optimization task is decomposed and assigned to obtain a set of candidate control suggestions. Based on preset priority arbitration rules, the set of candidate control suggestions is summarized and conflict resolved to obtain the target control strategy. The target control strategy is parsed into structured control instructions, and the structured control instructions are sent to the corresponding controller when the boundary check passes.

2. The method as described in claim 1, characterized in that, The step of acquiring multi-source information from the computer room cooling system and preprocessing the multi-source information to obtain a multi-source input set includes: The system acquires raw data from the data source of the data center cooling system and associates the raw data with the corresponding data source identifier and acquisition time sequence information to obtain the multi-source information. The data source includes at least structured operation data characterizing the data center environment and the operating status of the cooling equipment, as well as unstructured operation and maintenance information characterizing the data center operation knowledge. The structured runtime data is formatted and time-series aligned, and abnormal records in the structured runtime data are filtered out or corrected to obtain structured preprocessed data. The unstructured operation and maintenance information is parsed and standardized to obtain unstructured preprocessed data. The structured preprocessed data and the unstructured preprocessed data are then aggregated to obtain the multi-source input set.

3. The method as described in claim 1, characterized in that, The step of constructing a decision context based on the multi-source input set according to a preset semantic organization template, and triggering an optimization task based on the decision context, includes: Based on the multi-source input set, structured operational data and unstructured operation and maintenance information are identified, and the structured operational data and unstructured operation and maintenance information are mapped to fields according to the information items required by the preset semantic organization template to obtain a set of context elements. Based on the set of context elements, the set of context elements is semantically integrated according to the preset semantic organization template to generate a unified decision context; The decision context is triggered based on preset trigger conditions, and the optimization task is triggered based on the decision context when the preset trigger conditions are met.

4. The method as described in claim 1, characterized in that, The steps of performing credibility assessment and correction processing on key operational data in the decision context to obtain a corrected decision context, and then performing task decomposition and task allocation on the optimization task based on the corrected decision context to obtain a set of candidate control suggestions, include: Key operational data is extracted from the decision context, and the credibility of the key operational data is evaluated to generate an evaluation result. Based on the evaluation result, the key operational data is corrected, and the corrected key operational data is written back to the decision context to obtain a corrected decision context. Based on the correction decision context and the task objective corresponding to the optimization task, the optimization task is decomposed to obtain the task decomposition result, and task allocation information is generated according to the task decomposition result. Based on the task allocation information, the correction decision context is provided to the expertise processing unit corresponding to the task decomposition result, and the preliminary control suggestions output by the expertise processing unit based on the correction decision context are obtained. The preliminary control suggestions are then aggregated to obtain the candidate control suggestion set.

5. The method as described in claim 1, characterized in that, The step of summarizing and resolving conflicts in the candidate control suggestion set based on preset priority arbitration rules to obtain the target control strategy includes: Obtain each candidate control suggestion from the candidate control suggestion set, and perform normalization processing on each candidate control suggestion to obtain a candidate strategy element set; The candidate strategy element set is aggregated to generate an aggregated result, wherein the aggregated result is used to characterize the compatibility and conflict relationships among the candidate control suggestions. Based on the summarized results, conflict resolution is performed on the candidate strategy elements with conflicting relationships according to the preset priority arbitration rules, and the candidate strategy elements after conflict resolution are used to synthesize strategies to obtain the target control strategy.

6. The method as described in claim 1, characterized in that, The step of parsing the target control strategy into structured control instructions and issuing the structured control instructions to the corresponding controller when the boundary check passes includes: Based on the target control strategy, strategy elements for characterizing the controlled object and control actions are extracted, and the strategy elements are filled and encoded according to a preset instruction data structure to obtain the structured control instruction. The structured control instructions are subjected to rule compliance checks to generate boundary verification results, wherein the rule compliance checks are used to determine whether the structured control instructions meet preset security rules and / or preset operating rules; When the boundary verification result indicates that the boundary verification is successful, the structured control command is encapsulated according to a preset industrial communication protocol and sent to the controller corresponding to the controlled object identifier in the structured control command.

7. A multi-objective optimization device for a computer room cooling system, characterized in that, The device includes: The preprocessing module is used to acquire multi-source information from the computer room cooling system and preprocess the multi-source information to obtain a multi-source input set. The optimization triggering module is used to construct a decision context based on the multi-source input set according to a preset semantic organization template, and to trigger an optimization task based on the decision context; The candidate control module is used to perform credibility assessment and correction processing on the key operational data in the decision context to obtain a correction decision context, and to perform task decomposition and task allocation on the optimization task based on the correction decision context to obtain a set of candidate control suggestions. The control strategy module is used to summarize and resolve conflicts of the candidate control suggestion set based on preset priority arbitration rules to obtain the target control strategy; The target module is used to parse the target control strategy into structured control instructions and send the structured control instructions to the corresponding controller when the boundary check passes.

8. A multi-objective optimization device for a computer room cooling system, characterized in that, The device includes: a memory, a processor, and a multi-objective optimization program for a data center cooling system stored in the memory and executable on the processor, the multi-objective optimization program for the data center cooling system being configured to implement the steps of the multi-objective optimization method for a data center cooling system as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a multi-objective optimization program for a computer room cooling system, which, when executed by a processor, implements the steps of the multi-objective optimization method for a computer room cooling system as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the multi-objective optimization method for the computer room cooling system as described in any one of claims 1 to 6.

Images