Energy-saving and efficient management and control system and method for computing power cabin based on artificial intelligence

Abstract

The invention relates to the technical field of computing power cabin management and control, in particular to an energy-saving and efficient computing power cabin management and control system and method based on artificial intelligence. Comprising a computing power cabin sensing unit, a load prediction operation unit, a resource scheduling generation unit, a heat dissipation scheduling generation unit, a scheme optimization unit and a scheme comparison control unit. The computing power cabin sensing unit is used for collecting the internal temperature and the external temperature of the computing power cabin, power consumption and the load state of a server; the load prediction operation unit is used for extracting historical operation data and real-time operation data of the server; the method comprises the following steps: acquiring a temperature range of normal operation of a server as a heat dissipation standard, acquiring a load state of the server according to a resource scheduling scheme, combining an internal temperature, an external temperature and the heat dissipation standard, performing heat dissipation power analysis by using AI, and formulating heat dissipation power which enables the internal temperature to be within the heat dissipation standard and consumes least energy.

Description

Technical Field

[0001] The present invention relates to the field of computing power cabin control technology, and specifically, to an energy-saving and efficient control system and method for a computing power cabin based on artificial intelligence. Background Art

[0002] In the era of digital economy, computing power has become an important driving force for economic growth and social progress. With the rapid development of technologies such as artificial intelligence, big data and cloud computing, traditional computing models can no longer meet the growing demand for computing power and complex application scenarios.

[0003] At present, the resource utilization rate of the traditional computing model is not high, which easily leads to overload of some computing nodes and idleness of other nodes, affecting the overall performance and response speed of the system. Secondly, in terms of energy consumption, there is a lack of effective energy-saving measures, resulting in high energy consumption of the computing cabin, which increases the operating costs of the enterprise. Therefore, an energy-saving and efficient management and control system and method of the computing cabin based on artificial intelligence is proposed. Summary of the invention

[0004] The purpose of the present invention is to provide an energy-saving and efficient management and control system and method for a computing cabin based on artificial intelligence to solve the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, one of the purposes of the present invention is to provide an energy-saving and efficient management and control system for a computing power cabin based on artificial intelligence, including a computing power cabin perception unit, a load prediction operation unit, a resource scheduling generation unit, a heat dissipation scheduling generation unit, a solution optimization unit, and a solution comparison control unit; The computing cabin sensing unit is used to collect the internal temperature and external temperature of the computing cabin, power consumption, and load status of the server; The load prediction operation unit is used to extract historical operation data and real-time operation data of the server, mine the rules and characteristics according to the historical operation data, and then perform load prediction analysis on the server to obtain predicted load data; The resource scheduling generation unit is used to obtain the performance indicators of the computing nodes, and then use artificial intelligence to perform load balancing analysis on the performance indicators in combination with the predicted load data obtained by the load prediction operation unit, so as to obtain a resource scheduling plan for the computing node resource allocation. The heat dissipation scheduling generation unit is used to perform heat dissipation power analysis using artificial intelligence according to the resource scheduling scheme obtained by the resource scheduling generation unit in combination with the internal temperature, the external temperature and the heat dissipation standard, obtain the heat dissipation scheduling scheme, and send it to the computing power cabin for heat dissipation power adjustment; The scheme optimization unit is used to adjust the power output of the server according to the resource scheduling scheme, and dynamically optimize the resource scheduling scheme according to the load status of the server; The scheme comparison control unit is used to synchronously predict power consumption when the resource scheduling scheme and the heat dissipation scheduling scheme are updated, and then compare the predicted power consumption with the actual power consumption of the unupdated scheme. When the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated scheme to perform normal operation analysis, and the scheme is rolled back and updated based on the analysis results.

[0006] As a further improvement of the present technical solution, the computing cabin sensing unit establishes a data transmission connection with the sensors inside the computing cabin, and comprehensively senses the computing cabin through the data fed back by the sensors, thereby obtaining the internal temperature of the computing cabin, power consumption, and load status of the server.

[0007] As a further improvement of the technical solution, the load prediction operation unit includes an operation data extraction module and a load prediction module; The operation data extraction module is used to extract the data transmission volume of the computing power cabin and divide it into historical operation data and real-time operation data; The load prediction module is used to mine patterns and characteristics based on historical operating data, find out the patterns and characteristics of the computing power cabin under different working conditions, and set a time interception threshold. According to the time interception threshold, data is intercepted in the historical operating data, and then the computing power load in the future is predicted and analyzed based on the intercepted data, and the predicted computing power load is used as the predicted load data.

[0008] As a further improvement of the present technical solution, the resource scheduling generation unit includes a performance acquisition module and a resource scheduling module; The performance acquisition module is used to acquire the performance index of each computing node, thereby acquiring the maximum load capacity of each computing node; The resource scheduling module is used to combine the predicted load data obtained by the load prediction module with the performance indicators of each computing node using AI to perform load balancing analysis. When the computing load is greater than the maximum load capacity, the resource configuration of the computing node is increased according to the computing load. Conversely, when the computing load is less than the maximum load capacity, the resource configuration of the computing node is reduced according to the computing load. The resource adjustments of each computing node are then collected to form a resource scheduling plan.

[0009] As a further improvement of the technical solution, the heat dissipation scheduling generation unit includes a heat dissipation standard acquisition module and a heat dissipation power analysis module; The heat dissipation standard acquisition module is used to obtain the temperature range of normal operation of the server and use it as the heat dissipation standard; The heat dissipation power analysis module is used to obtain the load status of the server according to the resource scheduling plan, and use AI to perform heat dissipation power analysis based on the load status of the server in combination with the internal temperature, external temperature and heat dissipation standard, so as to formulate a heat dissipation power that can make the internal temperature within the heat dissipation standard and consume the least energy, and use it as the heat dissipation scheduling plan, and then send it to the computing power cabin, so that the computing power cabin adjusts the heat dissipation power in the corresponding time period of the resource scheduling plan.

[0010] As a further improvement of the technical solution, the solution optimization unit obtains the operating load of the server according to the resource scheduling solution analysis, and dynamically adjusts according to the operating load. When the operating load is lower than the current server load, the power supply output voltage is automatically reduced when working according to the resource scheduling solution. When the load status of the server enters a dangerous state, the resource allocation of the resource scheduling plan is increased.

[0011] As a further improvement of the technical solution, the scheme comparison control unit includes a power prediction module and a scheme control module; The power prediction module is used to update and monitor the resource scheduling plan and the heat dissipation scheduling plan. When a plan is updated, the power consumption of the updated plan is predicted to obtain the predicted power consumption of the updated plan. The scheme control module is used to compare the predicted power consumption with the actual power consumption of the unupdated scheme. When the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated scheme to perform a normal operation analysis. When the analysis results show that it can operate normally, the scheme is rolled back and updated, and the updated scheme is canceled. Conversely, when the analysis results show that it cannot operate normally, the updated scheme is retained, and the artificial intelligence controls the computing power cabin according to the updated scheme. Secondly, when the predicted power consumption is less than the real-time power consumption, the artificial intelligence directly uses the updated one to control the computing power cabin.

[0012] As a further improvement of the present technical solution, the perception unit in the computing power cabin is the data collection layer of artificial intelligence, the load prediction operation unit is the data processing layer of artificial intelligence, the resource scheduling generation unit, the heat dissipation scheduling generation unit and the solution optimization unit are the intelligent generation solution layer of artificial intelligence, and the solution comparison control unit is the analysis and decision-making layer of artificial intelligence. By setting artificial intelligence inside the computing power cabin and giving the artificial intelligence the authority to control the operation of the computing power cabin, the working status of the computing power cabin is adjusted.

[0013] The second object of the present invention is to provide an energy-saving and efficient management method for a computing power cabin based on artificial intelligence, including the energy-saving and efficient management system for a computing power cabin based on artificial intelligence described in any one of the above, comprising the following steps: S1. Comprehensively perceive the data of the computing power cabin and then perform load prediction analysis; S2. Generate a resource scheduling plan based on the predicted load data, then generate a heat dissipation scheduling plan based on the resource scheduling plan, and adjust the power output of the server through the resource scheduling plan and dynamically optimize the resource scheduling plan according to the load status of the server; S3. Compare the power consumption of the updated solution with that of the non-updated solution, and roll back and update the solution according to the comparison result.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. An energy-saving and efficient management and control system and method for computing power cabins based on artificial intelligence. AI is used to predict the load conditions on different computing nodes, and a load balancing method is used to automatically disperse requests and traffic to avoid overloading of a single node. This not only improves system stability and response speed, but also makes the load of each computing node more balanced, further improving the utilization efficiency of computing resources.

[0015] 2. An energy-saving and efficient management and control system and method for a computing power cabin based on artificial intelligence. It dynamically adjusts the output voltage and current of the power supply according to the actual operating load of the server to achieve precise power supply. It automatically reduces the power supply output voltage when the server load is low, reduces useless energy consumption, and minimizes energy loss while ensuring the normal operation of the server.

[0016] 3. An energy-saving and efficient management and control system and method for a computing power cabin based on artificial intelligence. By obtaining the temperature range of normal operation of the server as the heat dissipation standard, the load status of the server is obtained according to the resource scheduling plan, and the heat dissipation power analysis is performed using AI in combination with the internal temperature, external temperature and heat dissipation standard. The heat dissipation power that can keep the internal temperature within the heat dissipation standard and consume the least energy is formulated. This ensures that the server operates within the appropriate temperature range, reduces damage to the hardware caused by excessively high or low temperatures, and improves the reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of the structure of the computing module of the present invention; Figure 2 It is the overall structural principle diagram of the present invention.

[0018] The meaning of each number in the figure is: 10. Computing cabin sensing unit; 20. Load prediction operation unit; 30. Resource scheduling generation unit; 40. Heat dissipation scheduling generation unit; 50. Solution optimization unit; 60. Solution comparison control unit. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] like Figure 1-Figure 2 As shown, one of the purposes of the present invention is to provide an energy-saving and efficient management and control system for a computing cabin based on artificial intelligence, including a computing cabin sensing unit 10, a load prediction operation unit 20, a resource scheduling generation unit 30, a heat dissipation scheduling generation unit 40, a solution optimization unit 50, and a solution comparison control unit 60; The computing cabin sensing unit 10 is used to collect the internal temperature and external temperature of the computing cabin, power consumption, and load status of the server; The computing cabin sensing unit 10 establishes a data transmission connection with the sensor inside the computing cabin, and comprehensively senses the computing cabin through the data fed back by the sensor, thereby obtaining the internal temperature, power consumption, and load status of the server in the computing cabin. The specific steps are as follows: Determine the required sensor types and specifications: Install temperature sensors, power consumption sensors, and server load monitoring sensors as required to ensure that relevant status data of the computing cabin can be accurately obtained; Temperature sensors should be reasonably distributed in different locations in the computing power cabin, such as near the heat-generating parts of the server, the air outlet of the cooling system, and other key locations, to fully reflect the temperature distribution in the cabin; Power consumption sensors are installed on the power input line to ensure that the total power consumption of the entire computing cabin can be accurately measured; Server load status monitoring can be achieved by installing software agents inside the server or using the server management interface to collect data.

[0021] Establish data connection: According to the type of sensor and application scenario, select a communication protocol such as Modbus, CAN bus, ZigBee, and set the corresponding communication parameters according to the selected communication protocol, so as to establish a physical connection between the sensor and the data receiving device according to the requirements of the communication protocol.

[0022] The load prediction operation unit 20 is used to extract the historical operation data and real-time operation data of the server, mine the rules and characteristics according to the historical operation data, and then perform load prediction analysis on the server to obtain predicted load data; The load prediction operation unit 20 includes an operation data extraction module and a load prediction module; The operation data extraction module is used to extract the data transmission volume of the computing power cabin and divide it into historical operation data and real-time operation data; The load prediction module is used to mine the rules and characteristics based on the historical operation data, find out the rules and characteristics of the computing power cabin under different working conditions, and set the time interception threshold. According to the time interception threshold, data is intercepted in the historical operation data, and then the computing power load in the future is predicted and analyzed based on the intercepted data. The predicted computing power load is then used as the predicted load data. The specific steps are as follows: Data extraction classification: The extracted data is divided into historical operation data and real-time operation data according to the time attribute. It can be classified according to the timestamp. For example, the data in the past period of time is marked as historical operation data, and the data being generated in the current and recent period of time is marked as real-time operation data.

[0023] Historical data mining stage: Normalize the historical operation data to make the data of different indicators comparable, and use data analysis algorithms, such as time series analysis and cluster analysis, to explore the laws and characteristics of the computing power cabin under different working conditions.

[0024] Data capture stage: Based on actual needs and experience, determine a reasonable time capture threshold, set it according to data fluctuations. The greater the fluctuation, the shorter the threshold time. Then, based on the set time capture threshold, capture data from historical operation data. Future computing load prediction: According to the data characteristics and requirements, select the prediction model, such as linear regression, support vector machine (SVM), neural network, etc., and then use the intercepted historical data to train the prediction model. Use the trained model to predict the computing load in the future, and use the predicted computing load as the predicted load data.

[0025] The resource scheduling generation unit 30 is used to obtain the performance indicators of the computing nodes, and then use artificial intelligence to perform load balancing analysis on the performance indicators in combination with the predicted load data obtained by the load prediction operation unit 20, so as to obtain a resource scheduling solution for resource allocation to the computing nodes; The resource scheduling generation unit 30 includes a performance acquisition module and a resource scheduling module; The performance acquisition module is used to obtain the performance indicators of each computing node, so as to obtain the maximum load capacity of each computing node; The resource scheduling module is used to combine the predicted load data obtained by the load prediction module with the performance indicators of each computing node using AI to perform load balancing analysis. When the computing load is greater than the maximum load capacity, the resource configuration of the computing node is increased according to the computing load. Conversely, when the computing load is less than the maximum load capacity, the resource configuration of the computing node is reduced according to the computing load. Then, the resource adjustments of each computing node are collected to form a resource scheduling plan. The specific steps are as follows: Obtain computing node performance indicators: Clearly identify the computing node performance indicators that need to be obtained, including CPU utilization, memory usage, and I / O rate. Through stress testing or historical data analysis, determine the maximum load capacity of each computing node under different performance indicators.

[0026] Load balancing analysis: Compare the predicted load data with the performance indicators of each computing node. For each computing node, determine whether its current performance indicators can meet the predicted load requirements.

[0027] Resource adjustment phase: When the computing load is greater than the maximum load capacity, determine the type and quantity of resources that need to be increased; when the computing load is less than the maximum load capacity, determine the type and quantity of resources that can be reduced.

[0028] Form a resource scheduling plan: Collect the resource adjustment information of each computing node, including the type and quantity of resources added or reduced, and the corresponding computing node identifier, and then generate a resource scheduling plan based on the collected resource adjustment information. The formula is as follows: ; Among them, Adjustment is the resource adjustment amount. When the result is a positive value, it means that resources need to be increased, and when it is a negative value, it means that resources need to be reduced. PredictedLoad is the predicted load, MaxLoad is the maximum load capacity, CurrentLoad is the current load, and ɑ is the adjustment coefficient, which is set according to the actual situation and represents different resource types.

[0029] The heat dissipation scheduling generation unit 40 is used to perform heat dissipation power analysis using artificial intelligence according to the resource scheduling scheme obtained by the resource scheduling generation unit 30 in combination with the internal temperature, the external temperature and the heat dissipation standard, obtain the heat dissipation scheduling scheme, and send it to the computing power cabin for heat dissipation power adjustment; The heat dissipation scheduling generation unit 40 includes a heat dissipation standard acquisition module and a heat dissipation power analysis module; The heat dissipation standard acquisition module is used to obtain the temperature range of normal operation of the server and use it as the heat dissipation standard; The heat dissipation power analysis module is used to obtain the load status of the server according to the resource scheduling plan, and use AI to perform heat dissipation power analysis based on the server load status combined with the internal temperature, external temperature and heat dissipation standard, so as to formulate a heat dissipation power that can keep the internal temperature within the heat dissipation standard and consume the least energy, and use it as a heat dissipation scheduling plan, and then send it to the computing power cabin, so that the computing power cabin can adjust the heat dissipation power in the corresponding period of the resource scheduling plan. The specific steps are as follows: Determine the normal operating temperature range of the server: During the operation of the server, use a temperature sensor to continuously monitor the internal temperature of the server, analyze the temperature data over a period of time, and determine the temperature range of the server in a stable operating state as the actual normal operating temperature range. Alternatively, the computing power cabin is set with the normal operating temperature range of the server.

[0030] Get server load status: Extract server load status information from the resource scheduling plan.

[0031] Heat dissipation power analysis: Obtain data such as the server's internal temperature, external temperature, and heat dissipation standards. The internal temperature can be obtained through the temperature sensor inside the server, and the external temperature can be obtained through the ambient temperature sensor. Then, use AI technology, such as machine learning algorithms, to establish a relationship model between heat dissipation power and the server's internal temperature, external temperature, and load status. Input the current server's internal temperature, external temperature, and load status into the trained model. The model predicts the heat dissipation power that can keep the internal temperature within the heat dissipation standard and consume the least energy based on the input data.

[0032] Formulate a heat dissipation scheduling plan: According to the results of the heat dissipation power analysis, formulate a specific heat dissipation power adjustment strategy. If the predicted heat dissipation power is higher than the current heat dissipation power, the power of the heat dissipation equipment needs to be increased. If the predicted heat dissipation power is lower than the current heat dissipation power, the power of the heat dissipation equipment can be reduced. Send cooling scheduling plan: After receiving the cooling scheduling plan, the control system of the computing power cabin adjusts the cooling power according to the plan in the corresponding period of the resource scheduling plan. The formula is as follows: ; Where P is the heat dissipation power, T internal is the internal temperature of the server, T standard_mid is the middle value of the standard temperature range for heat dissipation, T external is the external environment temperature, L is the server load status, k1, k2, k 3为 The adjustment coefficient can be obtained through historical data training and optimization, and is used to adjust the difference between the internal temperature and the standard middle value, the difference between the external and internal temperatures, and the influence of the load state on the heat dissipation power.

[0033] The scheme optimization unit 50 is used to adjust the power output of the server according to the resource scheduling scheme, and dynamically optimize the resource scheduling scheme according to the load status of the server; The solution optimization unit 50 obtains the running load of the server according to the resource scheduling solution analysis, and dynamically adjusts according to the running load. When the running load is lower than the current server load, the power supply output voltage is automatically reduced when working according to the resource scheduling solution. The specific steps are as follows: Analyze the server load: Use server monitoring tools or software to monitor the server load in real time. The indicators that can be monitored include CPU utilization, memory usage, network traffic, etc. Compare the server load conditions monitored in real time with the load conditions expected in the resource scheduling plan to determine whether the server's operating load is lower than the expected load in the resource scheduling plan.

[0034] Dynamically adjust the power supply output voltage: If the operating load is lower than the expected load in the resource scheduling plan, the power supply output voltage adjustment process will be started, and the power supply output voltage will be automatically reduced according to certain rules. The voltage can be gradually reduced while monitoring the stability and performance of the server.

[0035] When the server load status enters a dangerous state, the resource scheduling plan is configured to increase resources. The specific steps are as follows: Monitor server load status: Clearly define indicators used to determine whether the server load status has entered a dangerous state, such as CPU utilization, memory usage, disk I / O waiting time, etc. Set a dangerous threshold for each indicator. When the indicator exceeds the threshold, the server load status is considered to have entered a dangerous state.

[0036] Determine whether the server has entered a dangerous state: Compare the load status indicators monitored in real time with the set danger thresholds. If one or more indicators exceed the danger thresholds, the server load status is considered to have entered a dangerous state.

[0037] Dynamically optimize resource scheduling plans: When the server load status enters a dangerous state, analyze the resource requirements of the current server. According to the specific conditions of the load status indicators, determine which resources need to be added, such as the number of CPU cores, memory capacity, storage capacity, etc. According to the results of the resource demand analysis, adjust the resource scheduling plan and increase resource configuration to ensure that the server can cope with the high load state; If the CPU utilization is too high, you can increase the number of CPU cores assigned to the server; if the memory utilization is too high, you can increase the memory capacity of the server.

[0038] The scheme comparison control unit 60 is used to synchronously predict power consumption when the resource scheduling scheme and the heat dissipation scheduling scheme are updated, and then compare the predicted power consumption with the actual power consumption of the unupdated scheme. When the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated scheme to perform normal operation analysis, and the scheme is rolled back and updated according to the analysis results.

[0039] The scheme comparison control unit 60 includes a power prediction module and a scheme control module; The power prediction module is used to update and monitor the resource scheduling plan and the heat dissipation scheduling plan. When a plan is updated, the power consumption of the updated plan is predicted to obtain the predicted power consumption of the updated plan. The solution control module is used to compare the predicted power consumption with the actual power consumption of the unupdated solution. When the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated solution to perform normal operation analysis. When the analysis result shows that it can run normally, the solution is rolled back and updated, and the updated solution is canceled. On the contrary, when the analysis result shows that it cannot run normally, the updated solution is retained, and the artificial intelligence controls the computing power cabin according to the updated solution. Secondly, when the predicted power consumption is less than the real-time power consumption, the artificial intelligence directly uses the updated solution to control the computing power cabin. The specific steps are as follows: Establish a monitoring mechanism: Set up a monitoring mechanism for resource scheduling plans and heat dissipation scheduling plans, and check in real time or regularly whether the plans have been updated.

[0040] Power consumption prediction: When a solution update is detected, data related to power consumption prediction is collected, including server configuration, load conditions, cooling parameters, etc. The model trained with historical data is used to predict power consumption for the updated solution. The formula is as follows: ; Among them, Power predicted To predict power consumption, Load is the server load status which can be expressed as a numerical value, such as CPU utilization, CoolingPower is the power of the cooling device, and q, w, and e are coefficients which can be obtained through historical data training and used to adjust the relationship between power consumption and load status and cooling power.

[0041] Comparison and analysis phase: Compare the predicted power consumption with the actual power consumption when the plan is not updated; If the predicted power consumption is greater than the real-time power consumption, the real-time operation data can be analyzed for normal operation in combination with the unupdated solution. You can consider simulating the server operation under the unupdated solution to evaluate whether it can meet the current business needs; If the analysis results show that it can run normally, roll back the plan and cancel the updated plan; if the analysis results show that it cannot run normally, keep the updated plan.

[0042] Controlling the computing power cabin: If the predicted power consumption is less than the real-time power consumption, the AI ​​directly uses the updated plan to control the computing power cabin. If the updated plan is retained, the AI ​​controls the computing power cabin according to the updated plan.

[0043] In the computing power cabin, the sensing unit 10 is the data collection layer of artificial intelligence, the load prediction operation unit 20 is the data processing layer of artificial intelligence, the resource scheduling generation unit 30, the heat dissipation scheduling generation unit 40 and the solution optimization unit 50 are the intelligent generation solution layer of artificial intelligence, and the solution comparison control unit 60 is the analysis and decision-making layer of artificial intelligence. By setting up artificial intelligence inside the computing power cabin and giving artificial intelligence the authority to control the operation of the computing power cabin, the working status of the computing power cabin is adjusted.

[0044] The second object of the present invention is to provide an energy-saving and efficient management method for a computing power cabin based on artificial intelligence, including any one of the above-mentioned energy-saving and efficient management systems for a computing power cabin based on artificial intelligence, comprising the following steps: S1. Comprehensively perceive the data of the computing power cabin and then perform load prediction analysis; S2. Generate a resource scheduling plan based on the predicted load data, then generate a heat dissipation scheduling plan based on the resource scheduling plan, and adjust the power output of the server through the resource scheduling plan and dynamically optimize the resource scheduling plan according to the load status of the server; S3. Compare the power consumption of the updated solution with that of the non-updated solution, and roll back and update the solution according to the comparison result.

[0045] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An energy-saving and efficient management and control system for computing power cabin based on artificial intelligence, characterized by: It comprises a computing cabin sensing unit (10), a load prediction operation unit (20), a resource scheduling generation unit (30), a heat dissipation scheduling generation unit (40), a solution optimization unit (50) and a solution comparison control unit (60); The computing cabin sensing unit (10) is used to collect the internal temperature and external temperature of the computing cabin, power consumption, and load status of the server; The load prediction operation unit (20) is used to extract historical operation data and real-time operation data of the server, mine the rules and characteristics based on the historical operation data, and then perform load prediction analysis on the server to obtain predicted load data; The resource scheduling generation unit (30) is used to obtain performance indicators of computing nodes, and then use artificial intelligence to perform load balancing analysis on the performance indicators in combination with the predicted load data obtained by the load prediction operation unit (20), thereby obtaining a resource scheduling solution for resource allocation to computing nodes; The heat dissipation scheduling generation unit (40) is used to perform heat dissipation power analysis using artificial intelligence according to the resource scheduling scheme obtained by the resource scheduling generation unit (30) in combination with the internal temperature, the external temperature and the heat dissipation standard, obtain the heat dissipation scheduling scheme, and send it to the computing power cabin for heat dissipation power adjustment; The scheme optimization unit (50) is used to adjust the power output of the server according to the resource scheduling scheme, and dynamically optimize the resource scheduling scheme according to the load status of the server; The scheme comparison control unit (60) is used to synchronously predict power consumption when the resource scheduling scheme and the heat dissipation scheduling scheme are updated, and then compare the predicted power consumption with the actual power consumption of the unupdated scheme; when the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated scheme to perform normal operation analysis, and the scheme is rolled back and updated according to the analysis results.

2. The energy-saving and efficient management and control system for computing power cabin based on artificial intelligence according to claim 1 is characterized in that: The computing cabin sensing unit (10) establishes a data transmission connection with sensors inside the computing cabin, and comprehensively senses the computing cabin through data fed back by the sensors, thereby obtaining the internal temperature of the computing cabin, power consumption, and load status of the server.

3. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The load prediction operation unit (20) comprises an operation data extraction module and a load prediction module; The operation data extraction module is used to extract the data transmission volume of the computing power cabin and divide it into historical operation data and real-time operation data; The load prediction module is used to mine patterns and characteristics based on historical operating data, find out the patterns and characteristics of the computing power cabin under different working conditions, and set a time interception threshold. According to the time interception threshold, data is intercepted in the historical operating data, and then the computing power load in the future is predicted and analyzed based on the intercepted data, and the predicted computing power load is used as the predicted load data.

4. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The resource scheduling generation unit (30) comprises a performance acquisition module and a resource scheduling module; The performance acquisition module is used to acquire the performance index of each computing node, thereby acquiring the maximum load capacity of each computing node; The resource scheduling module is used to combine the predicted load data obtained by the load prediction module with the performance indicators of each computing node using AI to perform load balancing analysis. When the computing load is greater than the maximum load capacity, the resource configuration of the computing node is increased according to the computing load. Conversely, when the computing load is less than the maximum load capacity, the resource configuration of the computing node is reduced according to the computing load. The resource adjustments of each computing node are then collected to form a resource scheduling plan.

5. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The heat dissipation scheduling generation unit (40) comprises a heat dissipation standard acquisition module and a heat dissipation power analysis module; The heat dissipation standard acquisition module is used to obtain the temperature range of normal operation of the server and use it as the heat dissipation standard; The heat dissipation power analysis module is used to obtain the load status of the server according to the resource scheduling plan, and use AI to perform heat dissipation power analysis based on the load status of the server in combination with the internal temperature, external temperature and heat dissipation standard, so as to formulate a heat dissipation power that can make the internal temperature within the heat dissipation standard and consume the least energy, and use it as the heat dissipation scheduling plan, and then send it to the computing power cabin, so that the computing power cabin adjusts the heat dissipation power in the corresponding time period of the resource scheduling plan.

6. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The scheme optimization unit (50) obtains the operation load of the server according to the resource scheduling scheme analysis, and dynamically adjusts according to the operation load. When the operation load is lower than the current server load, the power supply output voltage is automatically reduced when working according to the resource scheduling scheme. When the load status of the server enters a dangerous state, the resource allocation of the resource scheduling plan is increased.

7. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The scheme comparison control unit (60) comprises a power prediction module and a scheme control module; The power prediction module is used to update and monitor the resource scheduling plan and the heat dissipation scheduling plan. When a plan is updated, the power consumption of the updated plan is predicted to obtain the predicted power consumption of the updated plan. The scheme control module is used to compare the predicted power consumption with the actual power consumption of the unupdated scheme. When the predicted power consumption is greater than the real-time power consumption, the real-time operation data is combined with the unupdated scheme to perform a normal operation analysis. When the analysis results show that it can operate normally, the scheme is rolled back and updated, and the updated scheme is canceled. Conversely, when the analysis results show that it cannot operate normally, the updated scheme is retained, and the artificial intelligence controls the computing power cabin according to the updated scheme. Secondly, when the predicted power consumption is less than the real-time power consumption, the artificial intelligence directly uses the updated one to control the computing power cabin.

8. The energy-saving and efficient management and control system for computing cabin based on artificial intelligence according to claim 1 is characterized in that: The computing cabin sensing unit (10) is the data collection layer of artificial intelligence, the load prediction operation unit (20) is the data processing layer of artificial intelligence, the resource scheduling generation unit (30), the heat dissipation scheduling generation unit (40) and the solution optimization unit (50) are the intelligent generation solution layer of artificial intelligence, and the solution comparison control unit (60) is the analysis and decision-making layer of artificial intelligence. By setting artificial intelligence inside the computing cabin and giving the artificial intelligence the operation control authority of the computing cabin, the working state of the computing cabin is adjusted.

9. A method for realizing energy-saving and efficient management and control of a computing power cabin based on artificial intelligence, comprising an energy-saving and efficient management and control system of a computing power cabin based on artificial intelligence as described in any one of claims 1 to 8, characterized in that: The steps include: S1. Comprehensively perceive the data of the computing power cabin and then perform load prediction analysis; S2. Generate a resource scheduling plan based on the predicted load data, then generate a heat dissipation scheduling plan based on the resource scheduling plan, and adjust the power output of the server through the resource scheduling plan and dynamically optimize the resource scheduling plan according to the load status of the server; S3. Compare the power consumption of the updated solution with that of the non-updated solution, and roll back and update the solution according to the comparison result.

Images