Intelligent power management and distribution method and system for computer integrated cabinet

Abstract

The present invention provides an intelligent power management and distribution method and system for an integrated computer cabinet, which relates to the field of power management technology, including obtaining the state parameters of each device in the integrated computer cabinet and establishing a three-dimensional equipment energy consumption model of the integrated computer cabinet. According to the three-dimensional equipment energy consumption model, a deep reinforcement learning algorithm is used to determine the total power demand and the optimal equipment layout plan of the integrated computer cabinet, and according to the total power demand, the power demand, priority and optimal equipment layout plan of each device, the power supply and heat dissipation of each device are jointly configured to generate a first power supply and heat dissipation plan. When an abnormality is detected in the integrated computer cabinet, a fault diagnosis algorithm and a fault prediction algorithm are used to determine the cause of the fault and the risk of the fault, adjust the first power supply and heat dissipation plan, and generate a second power supply and heat dissipation plan.

Description

Technical Field

[0001] The invention relates to power management technology, and in particular to an intelligent power management and distribution method and system for an electronic and computer integrated cabinet. Background Art

[0002] Computer integrated cabinet refers to a high-density IT infrastructure that integrates computing, storage, network and other equipment in one cabinet. As the number and performance of equipment in computer integrated cabinets continue to increase, the power consumption problem of cabinets has become increasingly prominent. Traditional cabinet power management and distribution methods have the following shortcomings:

[0003] Lack of real-time power consumption monitoring methods. The power consumption of equipment in the cabinet is dynamically changing. Traditional ammeters and electricity meters can only provide power consumption data at the level of the entire cabinet. They cannot accurately grasp the real-time power consumption status of a single device, making it difficult to carry out fine-grained power management.

[0004] The power distribution method is extensive. The traditional approach is usually to equip each device with an independent power module, or multiple devices share a large-capacity power supply. This distribution method lacks flexibility and optimization space, and is prone to waste and imbalance of power resources.

[0005] Lack of intelligent power scheduling strategy. The power consumption requirements of different devices in the cabinet vary, and there are significant time-varying characteristics. Traditional power management systems cannot dynamically adjust the power supply strategy according to the actual load and priority of the equipment, making it difficult to improve the energy efficiency of the power system while ensuring the power supply of the equipment.

[0006] Lack of effective protection measures for equipment abnormalities. When a short circuit, overload or other power failure occurs in individual equipment, traditional protection measures usually simply cut off the power supply to the entire cabinet, forcing other normal equipment to be powered off, affecting business continuity.

[0007] Unable to adapt to changes in the power supply environment. The power supply environment in the computer room may experience unexpected situations such as power outages and power grid fluctuations. Traditional power management systems lack the ability to respond to emergencies and cannot ensure continuous and reliable power supply to equipment in the cabinet. Summary of the invention

[0008] The embodiments of the present invention provide an intelligent power management and distribution method and system for an integrated computer cabinet, which can solve the problems in the prior art.

[0009] According to a first aspect of the embodiments of the present invention,

[0010] Provides intelligent power management and distribution methods for integrated computer cabinets, including:

[0011] Acquire the status parameters of each device in the electronic and computer integrated cabinet, the status parameters include power parameters, communication status parameters, power supply status parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the status parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices;

[0012] According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priorities, heat dissipation requirements and heat dissipation conditions of each device, a deep reinforcement learning algorithm is used to determine the total power requirements and optimal equipment layout plan of the integrated computer cabinet, and according to the total power requirements, the power requirements, priorities and optimal equipment layout plan of each device, the power supply and heat dissipation of each device are jointly configured to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, the power supply path and heat dissipation system of the integrated computer cabinet are controlled to power and dissipate heat for each device, and the power supply status, heat dissipation status and equipment operation status of the integrated computer cabinet are detected in real time;

[0013] If it is detected that the electronic integrated computer cabinet has power supply abnormality, heat dissipation abnormality or equipment operation abnormality, the fault diagnosis algorithm and the fault prediction algorithm are used to determine the fault cause and fault risk based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, and according to the fault cause and fault risk, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme. According to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device. At the same time, according to the fault prediction results, preventive measures are taken for high-risk equipment in advance until the electronic integrated computer cabinet resumes normal power supply and heat dissipation.

[0014] In an optional embodiment,

[0015] The state parameters of each device in the electronic and computer integrated cabinet are obtained, and the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters. The steps of establishing a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device include:

[0016] Using three-dimensional drawing software, according to the design drawings and actual dimensions of the electronic integrated cabinet, draw the outline and internal partitions of the electronic integrated cabinet, define the geometric parameters of the length, width and height of the electronic integrated cabinet, and generate a three-dimensional model of the equipment of the electronic integrated cabinet according to the installation position and arrangement of the equipment;

[0017] The power parameters of the equipment in the integrated computer cabinet are mapped to the color attributes of the three-dimensional model of the equipment. The color values ​​representing different power levels are generated according to the real-time power data of each equipment. The color values ​​are assigned to the color attributes of the corresponding equipment model to generate a three-dimensional heat map reflecting the power distribution of the equipment.

[0018] The communication links of the equipment in the integrated computer cabinet are mapped to the lines between the three-dimensional models of the equipment. According to the actual communication links between each device, three-dimensional lines representing the links are drawn between the corresponding equipment models. The color of the three-dimensional lines represents the type of the link, and the style of the three-dimensional lines represents the communication protocol, so as to generate a three-dimensional network diagram reflecting the communication topology of the equipment.

[0019] The power supply circuits of the equipment in the integrated computer cabinet are mapped to the attribute labels of the equipment 3D model. According to the power supply circuit to which each equipment is connected, attribute labels representing the power supply circuits are added to the corresponding equipment models. The colors or shapes of the labels represent different power supply circuits, and a 3D circuit diagram reflecting the power supply distribution of the equipment is generated.

[0020] The generated three-dimensional heat map, three-dimensional network diagram and three-dimensional circuit diagram are superimposed in the same three-dimensional space to construct a three-dimensional equipment energy consumption model that comprehensively reflects the energy consumption characteristics of the equipment inside the electronic and computer integration cabinet.

[0021] In an optional embodiment,

[0022] According to the total power demand of the integrated computer cabinet, the power demand of each device, the priority and the optimal device layout plan, the power supply and heat dissipation of each device are jointly configured, and the steps of generating the first power supply and heat dissipation plan include:

[0023] According to the total power demand of the integrated computer cabinet, the power demand and priority of each device, a multi-objective optimization algorithm is used to establish a mathematical model of equipment power supply. The power supply quality and energy efficiency of the equipment are taken as optimization goals, and the power demand and priority of the equipment are taken as constraints to construct a multi-objective optimization problem.

[0024] A non-dominated sorting genetic algorithm is used to solve the multi-objective optimization problem to obtain an optimal solution set of power supply voltage and current parameters, and a solution that meets the power supply quality and energy efficiency requirements of the equipment is selected from the optimal solution set according to the equipment priority as the optimal power supply voltage and current parameters. According to the optimal equipment layout plan and equipment heat dissipation requirements, a three-dimensional computational fluid dynamics model of the inside of the electronic and computer integrated cabinet is established using computational fluid dynamics simulation technology, and the three-dimensional computational fluid dynamics model includes an equipment model, an air duct model, and a cold source model;

[0025] Determine the boundary conditions and solution settings of the three-dimensional computational fluid dynamics model, obtain the temperature field, velocity field and pressure field distribution inside the electronic integrated cabinet, optimize the size, shape and layout of the air duct of the electronic integrated cabinet according to the computational fluid dynamics simulation results, adjust the position, number and cooling capacity of the cold source, maximize the use of natural cooling and forced cooling, and generate a first power supply and heat dissipation solution.

[0026] In an optional embodiment,

[0027] The method further comprises:

[0028] Continuously collect the temperature and refrigeration equipment energy consumption data of the electronic and computer integrated cabinet within a preset evaluation period, calculate the heat dissipation effect improvement rate within the evaluation period based on the collected temperature data, calculate the energy consumption saving rate within the evaluation period based on the collected energy consumption data, and judge the effectiveness of the heat dissipation optimization control according to the preset heat dissipation effect improvement rate threshold and energy consumption saving rate threshold; when the heat dissipation effect improvement rate is lower than the heat dissipation effect improvement rate threshold or the energy consumption saving rate is lower than the energy consumption saving rate threshold, trigger an alarm and adjust the heat dissipation optimization algorithm parameters, and re-execute the heat dissipation optimization control until the heat dissipation effect improvement rate and the energy consumption saving rate both reach the corresponding thresholds;

[0029] A global cost function is constructed as a weighted combination of heat dissipation cost, energy consumption cost and performance cost. The intelligent controller uses a heuristic optimization algorithm to generate a global collaborative control strategy with the global cost function as the optimization target. The output of the global collaborative control strategy includes reference control instructions and constraints of each control unit.

[0030] Each control unit uses a multi-agent reinforcement learning algorithm to learn the local optimal control strategy under the constraints of the global collaborative control strategy, and reports the learned local optimal control strategy to the intelligent controller. The intelligent controller evaluates the impact of the local optimal control strategy on the global cost function and uses the local optimal control strategy that is better than the current global collaborative control strategy to update the global collaborative control strategy.

[0031] Repeat the global collaborative strategy update of the intelligent controller until the global cost function value converges, the maximum number of iterations is reached, the computing resources are exhausted, or the time limit is reached. Output the current global collaborative control strategy as the optimal strategy for the intelligent control of heat dissipation of the integrated computer cabinet to generate an equipment heat dissipation optimization model.

[0032] In an optional embodiment,

[0033] The method further comprises:

[0034] A collaborative optimization method is adopted to comprehensively consider the mutual influence and constraint relationship between power supply and heat dissipation, couple the equipment power supply mathematical model and the heat dissipation optimization model, and establish a joint optimization model of power supply and heat dissipation. The total power consumption and heat dissipation performance of the electronic and computer integrated cabinet are taken as optimization goals, and the power demand, priority and heat dissipation demand of the equipment are taken as constraints. The joint optimization model is solved using an intelligent optimization algorithm, and the optimal power supply and heat dissipation parameter combination obtained by the solution is used as the control instruction for the power supply and heat dissipation of the equipment, so as to collaboratively adjust the power supply and heat dissipation status of the equipment.

[0035] In an optional embodiment,

[0036] According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priority, heat dissipation requirements of each device and the heat dissipation conditions of the electronic and computer integrated cabinet, the steps of applying the deep reinforcement learning algorithm include:

[0037] Define the reinforcement learning environment, represent the equipment layout state of the integrated computer cabinet as a three-dimensional matrix, each element of the matrix corresponds to a position in the integrated computer cabinet, the value of the element represents the type of equipment placed at the position, and all equipment layout combinations are used as the state space; define a set of equipment layout adjustment operations as the action space; design a reward function to evaluate the pros and cons of the equipment layout scheme, and comprehensively consider the impact of the equipment layout on the total power consumption and heat dissipation performance of the integrated computer cabinet;

[0038] Build a deep reinforcement learning model, select a deep reinforcement learning algorithm based on value function, build a deep neural network as the approximate function, the input of the deep neural network is the current device layout state, and the output is the estimated value of each action taken under this state. Use mean square error as the loss function, and use an optimization algorithm to update the parameters of the deep neural network by minimizing the loss function;

[0039] By using the deep reinforcement learning algorithm to interact with the environment, training data is generated and the transfer experience is stored in the experience replay cache. A batch of transfer experiences are randomly extracted from the experience replay cache, the predicted value and target value of the current deep neural network are calculated, and the error is calculated according to the loss function. The parameters of the deep neural network are updated by optimizing the deep reinforcement learning algorithm.

[0040] In an optional embodiment,

[0041] When it is detected that the electronic and computer integrated cabinet has power supply anomaly, heat dissipation anomaly or equipment operation anomaly, the steps of applying a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status include:

[0042] Monitor and collect real-time data on the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet; build a diagnostic rule base for common fault types, and when the power supply status, heat dissipation status or equipment operation status is detected to be abnormal, extract fault features from the real-time data, match the extracted fault features with the rules in the diagnostic rule base, find out the rules that meet the conditions, and output the corresponding fault causes in the rules as diagnostic results;

[0043] Use machine learning algorithms to train historical operation data and build a fault diagnosis model. The historical operation data includes power supply status, heat dissipation status, and equipment operation status data as well as corresponding fault type labels. Input the real-time data into the trained fault diagnosis model to automatically identify the fault type and cause.

[0044] The real-time data is sorted by timestamp to construct a multivariate time series data set, feature extraction is performed on the multivariate time series data set to extract prediction features reflecting precursors to faults, the prediction features are trained using a time series analysis algorithm, a fault prediction model is constructed, and the risk of faults occurring in the future is predicted based on the feature change trend of historical operation data;

[0045] The diagnosis results and prediction results based on the rules in the diagnosis rule base and the time series analysis algorithm are integrated to output the failure causes and failure risks of the electronic and computer integrated cabinet, and an alarm is triggered in advance according to the failure prediction results to prompt the taking of preventive measures.

[0046] According to a second aspect of the embodiments of the present invention,

[0047] The first unit is used to obtain the state parameters of each device in the electronic and computer integrated cabinet, the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices;

[0048] The second unit is used to determine the total power demand and optimal equipment layout plan of the electronic and computer integrated cabinet according to the three-dimensional equipment energy consumption model, combined with the power demand, equipment priority, heat dissipation demand and heat dissipation conditions of each device, and use a deep reinforcement learning algorithm, and jointly configure the power supply and heat dissipation of each device according to the total power demand, the power demand, priority and optimal equipment layout plan of each device to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, control the power supply path and heat dissipation system of the electronic and computer integrated cabinet, supply power and heat dissipation to each device, and detect the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet in real time;

[0049] The third unit is used to determine the cause of the fault and the risk of the fault by using a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, if it is detected that the electronic integrated computer cabinet has power supply anomalies, heat dissipation anomalies or equipment operation anomalies. According to the cause of the fault and the risk of the fault, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme. According to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device. At the same time, according to the fault prediction results, preventive measures are taken in advance for high-risk equipment until the electronic integrated computer cabinet resumes normal power supply and heat dissipation.

[0050] According to a third aspect of the embodiments of the present invention,

[0051] An electronic device is provided, comprising:

[0052] processor;

[0053] a memory for storing processor-executable instructions;

[0054] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0055] According to a fourth aspect of the embodiments of the present invention,

[0056] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0057] The intelligent power supply and heat dissipation optimization method for integrated computer cabinets proposed in this paper fully considers the equipment priority and optimal layout. On the premise of meeting the power and heat dissipation requirements of the equipment, it minimizes the energy consumption of the cabinet and improves the energy efficiency of the cabinet, providing an effective way for the green and intelligent development of integrated computer cabinets.

[0058] The intelligent heat dissipation control method for integrated computer cabinets proposed in this paper achieves global optimization of heat dissipation effect, energy consumption and performance, overcoming the limitations of traditional heat dissipation control methods. A heat dissipation effect evaluation mechanism is introduced to ensure the effectiveness and reliability of heat dissipation optimization. Heuristic optimization and reinforcement learning algorithms are used to adaptively adjust and optimize the heat dissipation control strategy, which has good robustness and intelligence. Through the interactive iteration of global coordination and local optimization, while meeting the heat dissipation requirements of each device, energy consumption is saved to the maximum extent and heat dissipation efficiency is improved. This method provides a novel and efficient solution for the heat dissipation control of integrated computer cabinets, which is of great significance for improving the overall performance and energy efficiency of the cabinet.

[0059] The collaborative optimization method for power supply and heat dissipation of the integrated computer cabinet proposed in this paper comprehensively considers the mutual influence and constraint relationship between power supply and heat dissipation, establishes a joint optimization model, uses intelligent optimization algorithm to solve, generates collaborative control instructions, and realizes the integrated optimization control of power supply and heat dissipation. Compared with the traditional method of optimizing power supply and heat dissipation separately, the collaborative optimization method can better balance the energy efficiency and performance of the cabinet, while meeting the power consumption and heat dissipation requirements of the equipment, minimizing the total power consumption and heat dissipation energy consumption of the cabinet, and improving the overall operation efficiency of the integrated computer cabinet. The collaborative optimization method provides a new idea and framework for the power supply and heat dissipation control of the integrated computer cabinet, which has broad application prospects and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of a flow chart of an intelligent power management and distribution method for an electronic and computer integrated cabinet according to an embodiment of the present invention;

[0061] Figure 2 The figure is a schematic diagram of the structure of the intelligent power management and distribution system for the integrated computer cabinet according to the embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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.

[0063] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0064] Figure 1 FIG. 1 is a flow chart of an intelligent power management and distribution method for an integrated computer cabinet according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0065] S1. Acquire the status parameters of each device in the electronic and computer integrated cabinet, the status parameters include power parameters, communication status parameters, power supply status parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the status parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices;

[0066] S2. According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priorities, heat dissipation requirements and heat dissipation conditions of each device, a deep reinforcement learning algorithm is used to determine the total power requirements and optimal equipment layout plan of the integrated computer cabinet, and according to the total power requirements, the power requirements, priorities and optimal equipment layout plan of each device, the power supply and heat dissipation of each device are jointly configured to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, the power supply path and heat dissipation system of the integrated computer cabinet are controlled to supply power and dissipate heat to each device, and the power supply status, heat dissipation status and equipment operation status of the integrated computer cabinet are detected in real time;

[0067] S3. If it is detected that the electronic integrated computer cabinet has power supply anomalies, heat dissipation anomalies or equipment operation anomalies, based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, a fault diagnosis algorithm and a fault prediction algorithm are used to determine the cause of the fault and the risk of the fault. According to the cause of the fault and the risk of the fault, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme. According to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device. At the same time, according to the fault prediction results, preventive measures are taken in advance for high-risk equipment until the electronic integrated computer cabinet resumes normal power supply and heat dissipation.

[0068] In an optional embodiment,

[0069] The state parameters of each device in the electronic and computer integrated cabinet are obtained, and the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters. The steps of establishing a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device include:

[0070] Using three-dimensional drawing software, according to the design drawings and actual dimensions of the electronic integrated cabinet, draw the outline and internal partitions of the electronic integrated cabinet, define the geometric parameters of the length, width and height of the electronic integrated cabinet, and generate a three-dimensional model of the equipment of the electronic integrated cabinet according to the installation position and arrangement of the equipment;

[0071] The power parameters of the equipment in the integrated computer cabinet are mapped to the color attributes of the three-dimensional model of the equipment. The color values ​​representing different power levels are generated according to the real-time power data of each equipment. The color values ​​are assigned to the color attributes of the corresponding equipment model to generate a three-dimensional heat map reflecting the power distribution of the equipment.

[0072] The communication links of the equipment in the integrated computer cabinet are mapped to the lines between the three-dimensional models of the equipment. According to the actual communication links between each device, three-dimensional lines representing the links are drawn between the corresponding equipment models. The color of the three-dimensional lines represents the type of the link, and the style of the three-dimensional lines represents the communication protocol, so as to generate a three-dimensional network diagram reflecting the communication topology of the equipment.

[0073] The power supply circuits of the equipment in the integrated computer cabinet are mapped to the attribute labels of the equipment 3D model. According to the power supply circuit to which each equipment is connected, attribute labels representing the power supply circuits are added to the corresponding equipment models. The colors or shapes of the labels represent different power supply circuits, and a 3D circuit diagram reflecting the power supply distribution of the equipment is generated.

[0074] The generated three-dimensional heat map, three-dimensional network diagram and three-dimensional circuit diagram are superimposed in the same three-dimensional space to construct a three-dimensional equipment energy consumption model that comprehensively reflects the energy consumption characteristics of the equipment inside the electronic and computer integration cabinet.

[0075] For example, this paper proposes a three-dimensional equipment energy consumption modeling method for an integrated computer cabinet. By obtaining the state parameters of each device in the cabinet and mapping these parameters to three-dimensional space, a three-dimensional equipment energy consumption model that fully reflects the energy consumption characteristics of the equipment is constructed. The specific technical implementation steps are as follows:

[0076] Obtain the status parameters of each device in the integrated computer cabinet. Through the sensors and monitoring system deployed in the cabinet, the power parameters, communication status parameters, power supply status parameters and device location parameters of each device are collected in real time. Among them, the power parameters can measure the real-time power of the device through the current sensor and voltage sensor; the communication status parameters include the communication link, communication protocol and communication flow between devices; the power supply status parameters include the power supply circuit and power supply voltage connected to the device; the device location parameters give the specific installation location of each device in the cabinet.

[0077] Construct a 3D geometric model of the integrated computer cabinet. Use 3D drawing software, such as AutoCAD and SolidWorks, to draw the cabinet's outline and internal partitions based on the design drawings and actual dimensions of the integrated computer cabinet. Define the cabinet's length, width, height and other geometric parameters. For example, a cabinet has a length of 1200 mm, a width of 800 mm, and a height of 2000 mm. Then, based on the installation location and layout of the equipment, generate a 3D model of each device in the cabinet's 3D geometric model, and arrange these device models according to the actual layout.

[0078] Generate a three-dimensional heat map of the device power distribution. Map the acquired device power parameters to the color attributes of the device's three-dimensional model. Based on the real-time power data of each device, generate color values ​​representing different power levels, for example, 0-500 watts corresponds to blue, 500-1000 watts corresponds to green, 1000-1500 watts corresponds to yellow, and above 1500 watts corresponds to red. Then, assign these color values ​​to the color attributes of the corresponding device model to obtain a three-dimensional heat map reflecting the device power distribution. Through the three-dimensional heat map, you can intuitively identify the high-power consumption areas and low-power consumption areas inside the cabinet.

[0079] Generate a 3D network diagram of the device communication topology. Map the acquired device communication links to lines between the device 3D models. For every two devices with communication links, draw a 3D line representing the link between their corresponding 3D models. Set the color and style of the 3D line according to the type of communication link and the communication protocol. For example, Ethernet links are represented by blue solid lines, fiber links are represented by red dashed lines, and UART links are represented by green dotted lines. The resulting 3D network diagram can clearly show the communication topology between devices.

[0080] Generate a three-dimensional circuit diagram of the equipment power supply distribution. Map the acquired equipment power supply circuit to the attribute label of the equipment three-dimensional model. For each device, add the attribute label representing the power supply circuit to the corresponding equipment three-dimensional model according to the power supply circuit to which it is connected. Different power supply circuits are distinguished by labels of different colors or shapes. For example, A power supply uses a yellow rectangular label, B power supply uses a blue circular label, and UPS power supply uses a red triangle label. The generated three-dimensional circuit diagram can vividly express the power supply distribution of the equipment.

[0081] Construct a three-dimensional equipment energy consumption model. In the same three-dimensional space, the generated three-dimensional heat map, the generated three-dimensional network map, and the generated three-dimensional circuit diagram are superimposed to obtain a three-dimensional equipment energy consumption model that fully reflects the energy consumption characteristics of the equipment inside the integrated computer cabinet. The model integrates multi-dimensional attribute information such as power, communication, power supply, and location of the equipment, providing a data basis for subsequent energy consumption analysis and optimization.

[0082] Energy consumption analysis and optimization are carried out based on the three-dimensional equipment energy consumption model. Cluster analysis is performed on the three-dimensional heat map to identify equipment groups with similar energy consumption levels, and to find out the high-energy consumption areas and low-energy consumption areas inside the cabinet, providing a basis for optimizing the equipment layout. The association rule mining algorithm is used to explore the correlation between equipment energy consumption parameters and location parameters. For example, the energy consumption of equipment in the upper part of the cabinet is generally higher, and the energy consumption of equipment close to the cold channel is lower. Using parametric simulation technology, different equipment layout schemes are simulated on the three-dimensional equipment energy consumption model. By adjusting parameters such as equipment position and cabinet structure, the impact of each scheme on energy consumption is evaluated, and the layout scheme with the best energy consumption is searched from many alternative schemes. For example, through simulation, it is found that concentrating high-power consumption equipment in the lower part of the cabinet can effectively reduce the heat accumulation on the top of the cabinet, thereby reducing the total energy consumption by 5%.

[0083] Through the above steps, the three-dimensional equipment energy consumption modeling method of the integrated computer cabinet proposed in this paper can accurately characterize the complex equipment energy consumption characteristics inside the cabinet, provide an intuitive and quantitative analysis tool for the energy-saving design and optimized operation of the cabinet, and has good engineering application value.

[0084] In an optional embodiment,

[0085] According to the total power demand of the integrated computer cabinet, the power demand of each device, the priority and the optimal device layout plan, the power supply and heat dissipation of each device are jointly configured, and the steps of generating the first power supply and heat dissipation plan include:

[0086] According to the total power demand of the integrated computer cabinet, the power demand and priority of each device, a multi-objective optimization algorithm is used to establish a mathematical model of equipment power supply. The power supply quality and energy efficiency of the equipment are taken as optimization goals, and the power demand and priority of the equipment are taken as constraints to construct a multi-objective optimization problem.

[0087] A non-dominated sorting genetic algorithm is used to solve the multi-objective optimization problem to obtain an optimal solution set of power supply voltage and current parameters, and a solution that meets the power supply quality and energy efficiency requirements of the equipment is selected from the optimal solution set according to the equipment priority as the optimal power supply voltage and current parameters. According to the optimal equipment layout plan and equipment heat dissipation requirements, a three-dimensional computational fluid dynamics model of the inside of the electronic and computer integrated cabinet is established using computational fluid dynamics simulation technology, and the three-dimensional computational fluid dynamics model includes an equipment model, an air duct model, and a cold source model;

[0088] Determine the boundary conditions and solution settings of the three-dimensional computational fluid dynamics model, obtain the temperature field, velocity field and pressure field distribution inside the electronic integrated cabinet, optimize the size, shape and layout of the air duct of the electronic integrated cabinet according to the computational fluid dynamics simulation results, adjust the position, number and cooling capacity of the cold source, maximize the use of natural cooling and forced cooling, and generate a first power supply and heat dissipation solution.

[0089] For example, this paper proposes an intelligent power supply and heat dissipation optimization method for an integrated computer cabinet. According to the total power demand of the cabinet, the power demand, priority and optimal equipment layout plan of each device, the power supply and heat dissipation of each device are jointly configured to generate the optimal power supply and heat dissipation plan. The specific technical implementation steps are as follows:

[0090] Establish a mathematical model for equipment power supply. First, determine the total power supply capacity of the cabinet based on the total power demand of the integrated computer cabinet. For example, a cabinet contains 20 devices with a total power demand of 10 kilowatts. Then, based on the power demand and priority of each device, a multi-objective optimization algorithm is used to establish a mathematical model for equipment power supply. This model takes the power supply quality and energy efficiency of the equipment as optimization goals, and the power demand and priority of the equipment as constraints to construct a multi-objective optimization problem. Among them, the power supply quality can be measured by indicators such as voltage stability and harmonic content, and the energy efficiency can be measured by indicators such as power conversion efficiency and power factor. The power demand of the equipment is determined according to its rated power, and the priority is assigned different weight coefficients according to the importance of the equipment. By reasonably setting the objective function and constraints, a complete mathematical model for equipment power supply is established.

[0091] Solve the mathematical model of equipment power supply. For the established multi-objective optimization problem, the non-dominated sorting genetic algorithm is used to solve it. This algorithm is a mature heuristic search algorithm that searches for the optimal solution in the solution space by simulating the biological evolution process. First, a set of power supply voltage and current parameters is randomly generated as the initial population. Then, the individuals in the population are non-dominated sorted and divided into different frontiers according to the quality of the individuals. Next, the individuals on different frontiers are selected, crossed and mutated to generate a new offspring population. After multiple iterations, the population is continuously optimized and finally converges to the optimal solution set. From the optimal solution set, according to the equipment priority, the solution that meets the equipment power supply quality and energy efficiency requirements is selected as the optimal power supply voltage and current parameters. For example, for core equipment with high priority, the power supply parameters with the highest voltage stability and the lowest harmonic content are selected; for non-core equipment with low priority, the power supply parameters with the best energy efficiency are selected.

[0092] Establish a three-dimensional computational fluid dynamics model inside the integrated computer cabinet. According to the optimal equipment layout plan and equipment heat dissipation requirements, the computational fluid dynamics simulation technology is used to establish a three-dimensional numerical model inside the integrated computer cabinet. The model consists of three main parts: equipment model, duct model and cold source model. The equipment model accurately depicts the shape and placement of the equipment in three-dimensional space according to the geometric dimensions and layout positions of each device. The duct model constructs the airflow channel network inside the cabinet based on the partitions, wiring troughs and other structures inside the cabinet. The cold source model determines the flow rate, temperature and other parameters of the cooling air according to the cooling equipment equipped in the cabinet, such as fans and air conditioners. Through reasonable simplification and meshing, a high-quality three-dimensional computational fluid dynamics model is generated to provide a basis for subsequent heat dissipation simulation analysis.

[0093] Conduct simulation analysis of heat dissipation inside the cabinet. Use the established three-dimensional computational fluid dynamics model to conduct simulation analysis of heat dissipation inside the cabinet. First, determine the boundary conditions and solution settings of the model, including equipment heat generation, inlet and outlet parameters, cold source conditions, etc. Then, use computational fluid dynamics numerical methods, such as finite element method and finite volume method, to solve the flow and heat exchange process inside the cabinet and obtain the temperature field, velocity field and pressure field distribution inside the cabinet. By analyzing the simulation results, evaluate the heat dissipation performance inside the cabinet and identify areas with weak heat dissipation. For example, through the temperature field distribution, it is found that the temperature near the top equipment of the cabinet is as high as 45 degrees Celsius, far exceeding the safe operating temperature of the equipment. According to the simulation results, optimize the air duct design and cold source configuration of the integrated computer cabinet, such as increasing the cross-sectional area of ​​the top air duct, increasing the number of fans near the top equipment, etc., to improve the heat dissipation conditions inside the cabinet.

[0094] Generate the optimal power supply and heat dissipation solution. The optimal power supply voltage and current parameters obtained in combination with the optimized cabinet heat dissipation configuration are used to generate the optimal power supply and heat dissipation solution. This solution minimizes power supply energy consumption and heat dissipation energy consumption while ensuring the power supply quality and reliability of the equipment, and balances the power consumption and heat dissipation requirements of the equipment. In terms of power supply, the optimal power supply voltage and current parameters are matched for equipment of different priorities to reduce power loss and conversion loss. In terms of heat dissipation, the air ducts and cold sources are arranged reasonably, the heat dissipation requirements of core equipment are prioritized, natural cooling is maximized, and active cooling energy consumption is reduced. For example, for a core device, the power supply voltage is 12V, the power supply current is 10A, the air duct cross-sectional area is 50 square centimeters, and 2 fans are equipped to ensure that the device temperature is below 30 degrees Celsius. For a non-core device, the power supply voltage is 5V, the power supply current is 1A, and natural convection heat dissipation is used to control the device temperature below 40 degrees Celsius. By optimizing and matching power supply and heat dissipation parameters, the overall energy-saving and efficient operation of the cabinet can be achieved.

[0095] Through the above steps, the intelligent power supply and heat dissipation optimization method for the integrated computer cabinet proposed in this paper fully considers the equipment priority and optimal layout. On the premise of meeting the power consumption and heat dissipation requirements of the equipment, it minimizes the energy consumption of the cabinet and improves the energy efficiency of the cabinet, providing an effective way for the green and intelligent development of the integrated computer cabinet.

[0096] In an optional embodiment,

[0097] The method further comprises:

[0098] Continuously collect the temperature and refrigeration equipment energy consumption data of the electronic and computer integrated cabinet within a preset evaluation period, calculate the heat dissipation effect improvement rate within the evaluation period based on the collected temperature data, calculate the energy consumption saving rate within the evaluation period based on the collected energy consumption data, and judge the effectiveness of the heat dissipation optimization control according to the preset heat dissipation effect improvement rate threshold and energy consumption saving rate threshold; when the heat dissipation effect improvement rate is lower than the heat dissipation effect improvement rate threshold or the energy consumption saving rate is lower than the energy consumption saving rate threshold, trigger an alarm and adjust the heat dissipation optimization algorithm parameters, and re-execute the heat dissipation optimization control until the heat dissipation effect improvement rate and the energy consumption saving rate both reach the corresponding thresholds;

[0099] A global cost function is constructed as a weighted combination of heat dissipation cost, energy consumption cost and performance cost. The intelligent controller uses a heuristic optimization algorithm to generate a global collaborative control strategy with the global cost function as the optimization target. The output of the global collaborative control strategy includes reference control instructions and constraints of each control unit.

[0100] Each control unit uses a multi-agent reinforcement learning algorithm to learn the local optimal control strategy under the constraints of the global collaborative control strategy, and reports the learned local optimal control strategy to the intelligent controller. The intelligent controller evaluates the impact of the local optimal control strategy on the global cost function and uses the local optimal control strategy that is better than the current global collaborative control strategy to update the global collaborative control strategy.

[0101] Repeat the global collaborative strategy update of the intelligent controller until the global cost function value converges, the maximum number of iterations is reached, the computing resources are exhausted, or the time limit is reached. Output the current global collaborative control strategy as the optimal strategy for the intelligent control of heat dissipation of the integrated computer cabinet to generate an equipment heat dissipation optimization model.

[0102] For example, based on the intelligent power supply and heat dissipation optimization method for the integrated computer cabinet proposed in this paper, in order to further improve the effectiveness and intelligence level of heat dissipation optimization, heat dissipation effect evaluation, global collaborative optimization and multi-agent reinforcement learning technologies are introduced to construct an intelligent heat dissipation control method for the integrated computer cabinet. The specific technical implementation steps are as follows:

[0103] Heat dissipation effect evaluation. During the execution of heat dissipation optimization control, the temperature of the integrated computer cabinet and the energy consumption data of the refrigeration equipment are continuously collected within a preset evaluation period. Based on the collected temperature data, the heat dissipation effect improvement rate within the evaluation period is calculated, that is, the percentage of decrease in the average temperature inside the cabinet before and after optimization. Based on the collected energy consumption data, the energy saving rate within the evaluation period is calculated, that is, the percentage of decrease in the energy consumption of the refrigeration equipment before and after optimization. The effectiveness of the heat dissipation optimization control is judged based on the preset heat dissipation effect improvement rate threshold and energy saving rate threshold. For example, the heat dissipation effect improvement rate threshold is set to 10%, and the energy saving rate threshold is set to 15%. If the average temperature inside the cabinet drops from 40 degrees Celsius to 35 degrees Celsius after optimization, the heat dissipation effect improvement rate is 12.5%, which is higher than the threshold; the energy consumption of the refrigeration equipment drops from 1000 watts to 800 watts, and the energy saving rate is 20%, which is also higher than the threshold, then the heat dissipation optimization control is considered to be effective. On the contrary, if the heat dissipation effect improvement rate or energy consumption saving rate is lower than the corresponding threshold, an alarm is triggered, indicating that the heat dissipation optimization control has not met expectations. It is necessary to adjust the heat dissipation optimization algorithm parameters and re-execute the heat dissipation optimization control until the heat dissipation effect improvement rate and energy consumption saving rate meet the requirements.

[0104] Construct a global cost function. In order to achieve the global optimal control of the heat dissipation of the integrated computer cabinet, a global cost function is constructed as a weighted combination of heat dissipation cost, energy consumption cost and performance cost. Among them, the heat dissipation cost represents the uniformity of temperature distribution and the number of hot spots inside the cabinet, the energy consumption cost represents the energy consumption level of the refrigeration equipment, and the performance cost represents the degree of influence of the heat dissipation status of the equipment on its performance. By reasonably setting the weight coefficients of the three costs, a global cost function is formed as a comprehensive indicator for evaluating the pros and cons of the heat dissipation control strategy. For example, the heat dissipation cost weight is 0.4, the energy consumption cost weight is 0.4, and the performance cost weight is 0.2. The heat dissipation cost is the difference between the highest temperature and the average temperature inside the cabinet, the energy consumption cost is the ratio of the energy consumption of the refrigeration equipment to the rated energy consumption, and the performance cost is the percentage of equipment performance degradation. The global cost function value is obtained by weighted summation. The smaller the value, the better the heat dissipation control strategy.

[0105] The intelligent controller generates a global collaborative control strategy. The integrated computer cabinet is equipped with an intelligent controller, which is responsible for generating a global collaborative control strategy and coordinating the work of each control unit. The intelligent controller uses a heuristic optimization algorithm to search for the optimal global collaborative control strategy with the global cost function as the optimization target. The heuristic optimization algorithm simulates the intelligent behavior of nature, such as ant colony foraging and particle swarm optimization, to find the optimal solution in a complex solution space. The output of the global collaborative control strategy includes reference control instructions and constraints of each control unit, such as the temperature setting value of the refrigeration equipment and the speed range of the fan. For example, the global collaborative control strategy output by the intelligent controller is: the temperature of the refrigeration equipment is set to 25 degrees Celsius, the speed range of the upper fan is 1000-2000 rpm, the speed range of the lower fan is 2000-3000 rpm, and the total power consumption of the cabinet does not exceed 2000 watts.

[0106] The control unit uses a reinforcement learning algorithm to optimize the local control strategy. Under the constraints of the global collaborative control strategy, the various control units of the integrated computer cabinet, such as the refrigeration equipment controller and the fan controller, use a multi-agent reinforcement learning algorithm to learn the local optimal control strategy. Reinforcement learning is an intelligent control method that does not require prior knowledge. It learns the optimal control behavior through continuous trial and error and feedback. Multi-agent reinforcement learning takes into account the interaction between different control units and converges to the global optimum through collaboration and competition. For example, the local optimal control strategy learned by the refrigeration equipment controller is: when the average temperature inside the cabinet is higher than 28 degrees Celsius, the temperature setting value of the refrigeration equipment is lowered by 2 degrees Celsius; when the average temperature is lower than 26 degrees Celsius, the temperature setting value is increased by 1 degree Celsius. The local optimal control strategy learned by the fan controller is: when the temperature of the upper device is higher than 35 degrees Celsius, the upper fan speed is increased by 500 rpm; when the temperature of the lower device is higher than 30 degrees Celsius, the lower fan speed is increased by 300 rpm.

[0107] The intelligent controller updates the global collaborative control strategy. Each control unit reports the learned local optimal control strategy to the intelligent controller. The intelligent controller evaluates the impact of the local optimal control strategy on the global cost function, that is, whether the value of the global cost function is better than the value under the current global collaborative control strategy under the local optimal control strategy. If the local optimal control strategy is better than the current global collaborative control strategy, it will be used to update the global collaborative control strategy, otherwise the current global collaborative control strategy will remain unchanged. For example, under the local optimal control strategy, the global cost function value is 0.8, which is better than 1.2 under the current global collaborative control strategy. The local optimal control strategy is adopted and the global collaborative control strategy is updated.

[0108] Repeat the iterative optimization and output the optimal heat dissipation control strategy. Repeat the intelligent controller to continuously update the global collaborative control strategy until one of the following termination conditions is met: the global cost function value converges, the maximum number of iterations is reached, the computing resources are exhausted, or the time limit is reached. After the optimization is terminated, the current global collaborative control strategy is output as the optimal strategy for the intelligent control of heat dissipation of the integrated computer cabinet, and the equipment heat dissipation optimization model is generated. For example, after 100 iterative optimizations, the global cost function value converges from the initial 2.5 to 0.6, reaching the maximum number of iterations, and the global collaborative control strategy at this time is output as the optimal strategy. The optimal strategy includes: the temperature of the refrigeration equipment is set to 24 degrees Celsius, the upper fan speed is 1800 rpm, the lower fan speed is 2500 rpm, and the total power consumption of the cabinet is 1600 watts. The optimal strategy is converted into an equipment heat dissipation optimization model to guide the heat dissipation control of the integrated computer cabinet.

[0109] Through the above steps, the intelligent heat dissipation control method for integrated computer cabinets proposed in this paper realizes the global optimization of heat dissipation effect, energy consumption and performance, and overcomes the limitations of traditional heat dissipation control methods. A heat dissipation effect evaluation mechanism is introduced to ensure the effectiveness and reliability of heat dissipation optimization. Heuristic optimization and reinforcement learning algorithms are used to adaptively adjust and optimize the heat dissipation control strategy, which has good robustness and intelligence. Through the interactive iteration of global coordination and local optimization, while meeting the heat dissipation requirements of each device, energy consumption is saved to the maximum extent and heat dissipation efficiency is improved. This method provides a novel and efficient solution for the heat dissipation control of integrated computer cabinets, which is of great significance for improving the overall performance and energy efficiency of the cabinet.

[0110] In an optional embodiment,

[0111] The method further comprises:

[0112] A collaborative optimization method is adopted to comprehensively consider the mutual influence and constraint relationship between power supply and heat dissipation, couple the equipment power supply mathematical model and the heat dissipation optimization model, and establish a joint optimization model of power supply and heat dissipation. The total power consumption and heat dissipation performance of the electronic and computer integrated cabinet are taken as optimization goals, and the power demand, priority and heat dissipation demand of the equipment are taken as constraints. The joint optimization model is solved using an intelligent optimization algorithm, and the optimal power supply and heat dissipation parameter combination obtained by the solution is used as the control instruction for the power supply and heat dissipation of the equipment, so as to collaboratively adjust the power supply and heat dissipation status of the equipment.

[0113] For example, based on the intelligent power supply and heat dissipation optimization method of the integrated computer cabinet proposed above, in order to further improve the energy efficiency and performance of the cabinet, this paper introduces the idea of ​​collaborative optimization, comprehensively considers the mutual influence and constraint relationship between power supply and heat dissipation, establishes a joint optimization model of power supply and heat dissipation, and realizes the coordinated regulation of the two. The specific technical implementation steps are as follows:

[0114] Couple the mathematical model of equipment power supply and the heat dissipation optimization model. The power supply and heat dissipation of equipment in the integrated computer cabinet have a close mutual influence and constraint relationship. On the one hand, the power supply status of the equipment determines its heat generation and heat dissipation requirements; on the other hand, the heat dissipation status of the equipment affects its operating temperature and power consumption level. In order to optimize power supply and heat dissipation in a coordinated manner, it is necessary to couple the mathematical models of the two. The mathematical model of equipment power supply describes the quantitative relationship between equipment power consumption and power supply voltage and current, and the heat dissipation optimization model describes the quantitative relationship between equipment temperature and heat dissipation parameters (such as fan speed, cooling power, etc.). By introducing coupling variables such as equipment heat generation and operating temperature into the two models, the relationship between power supply and heat dissipation is established. For example, the mathematical model of equipment power supply is: power consumption = voltage × current + heat generation coefficient × current^2, where the heat generation coefficient is related to the equipment operating temperature. The heat dissipation optimization model is: equipment temperature = ambient temperature + heat generation / heat dissipation coefficient, where the heat dissipation coefficient is related to heat dissipation parameters such as fan speed and cooling power. The two models are linked through the coupling variable of heat generation.

[0115] Establish a joint optimization model for power supply and heat dissipation. On the basis of the coupled equipment power supply mathematical model and heat dissipation optimization model, establish a joint optimization model for power supply and heat dissipation. The joint optimization model takes the total power consumption and heat dissipation performance of the integrated computer cabinet as the optimization target, while taking into account the power supply and heat dissipation requirements of the equipment. The optimization target can be expressed as a weighted combination, such as total power consumption × power consumption weight + heat dissipation cost × heat dissipation weight, where the power consumption weight and heat dissipation weight reflect the relative importance of the two targets. Constraints include equipment power requirements, priorities, and heat dissipation requirements. For example, the power requirement of equipment A is 500 watts, the priority is 1 (highest), and the heat dissipation requirement is that the temperature does not exceed 40 degrees Celsius; the power requirement of equipment B is 300 watts, the priority is 2, and the heat dissipation requirement is that the temperature does not exceed 45 degrees Celsius. The joint optimization model needs to find the optimal combination of power supply and heat dissipation parameters under the premise of meeting these constraints to minimize the total power consumption and heat dissipation cost.

[0116] Use intelligent optimization algorithms to solve the joint optimization model. Since there are many devices in the integrated computer cabinet, the combination space of power supply and heat dissipation parameters is very large, and traditional optimization methods are difficult to solve the joint optimization model. Therefore, it is necessary to use intelligent optimization algorithms, such as genetic algorithms and particle swarm optimization algorithms, to solve the joint optimization model. Intelligent optimization algorithms simulate the evolutionary mechanism of nature and use group intelligence and random search to find the optimal solution in a complex solution space. For example, a genetic algorithm is used to solve the joint optimization model, and the power supply and heat dissipation parameters are encoded into chromosomes. Through genetic operations such as selection, crossover, and mutation, the chromosomes are continuously optimized and finally converge to the optimal solution. During the solution process, it is necessary to design an appropriate fitness function to evaluate the quality of the chromosome, such as fitness = 1 / (total power consumption × power consumption weight + heat dissipation cost × heat dissipation weight). The greater the fitness, the better the chromosome. After multiple iterative evolutions, the genetic algorithm solves the optimal combination of power supply and heat dissipation parameters, that is, the optimal solution of the joint optimization model.

[0117] Generate control instructions for equipment power supply and heat dissipation. The optimal power supply and heat dissipation parameter combination obtained by solving the joint optimization model is converted into control instructions for equipment power supply and heat dissipation, which are used to guide the actual operation of the integrated computer cabinet. The control instructions include the power supply voltage and current of each device, as well as the speed of the cooling fan and the cooling power. For example, the optimal solution of the joint optimization model is: the power supply voltage of device A is 12V, the current is 40A, and the fan speed is 2000 rpm; the power supply voltage of device B is 5V, the current is 60A, and the fan speed is 3000 rpm; the total power consumption of the cabinet is 1800 watts. These parameters are converted into control instructions and sent to the intelligent power supply module and the heat dissipation control module to coordinately adjust the power supply and heat dissipation status of the equipment. In actual operation, the control instructions can be dynamically adjusted and optimized according to the working status and environmental conditions of the equipment to achieve real-time coordinated control of power supply and heat dissipation.

[0118] Evaluate the effect of collaborative optimization and continuously improve the optimization model. During the operation of the integrated computer cabinet, continuously monitor the power supply and heat dissipation status of the equipment, collect relevant data, and evaluate the effect of collaborative optimization. By comparing the total power consumption of the cabinet, the operating temperature of the equipment, the heat dissipation energy consumption and other indicators before and after optimization, quantitatively analyze the energy efficiency improvement and performance improvement brought by collaborative optimization. At the same time, based on the evaluation results, continuously improve and improve the joint optimization model of power supply and heat dissipation, adjust the optimization objectives, constraints and weight coefficients to make it closer to actual application needs. For example, if the total power consumption of the cabinet is reduced by 10% after optimization, but the operating temperature of some equipment exceeds the safety threshold, it is necessary to add temperature constraints in the joint optimization model and increase the heat dissipation weight to ensure the reliable operation of the equipment. Through continuous evaluation and improvement, the effect and robustness of collaborative optimization are continuously improved.

[0119] Through the above steps, the collaborative optimization method for power supply and heat dissipation of the integrated computer cabinet proposed in this paper comprehensively considers the mutual influence and constraint relationship between power supply and heat dissipation, establishes a joint optimization model, uses intelligent optimization algorithm to solve, generates collaborative control instructions, and realizes the integrated optimization control of power supply and heat dissipation. Compared with the traditional method of optimizing power supply and heat dissipation separately, the collaborative optimization method can better balance the energy efficiency and performance of the cabinet, while meeting the power consumption and heat dissipation requirements of the equipment, minimizing the total power consumption and heat dissipation energy consumption of the cabinet, and improving the overall operation efficiency of the integrated computer cabinet. The collaborative optimization method provides a new idea and framework for the power supply and heat dissipation control of the integrated computer cabinet, which has broad application prospects and promotion value.

[0120] In an optional embodiment,

[0121] According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priority, heat dissipation requirements of each device and the heat dissipation conditions of the electronic and computer integrated cabinet, the steps of applying the deep reinforcement learning algorithm include:

[0122] Define the reinforcement learning environment, represent the equipment layout state of the integrated computer cabinet as a three-dimensional matrix, each element of the matrix corresponds to a position in the integrated computer cabinet, the value of the element represents the type of equipment placed at the position, and all equipment layout combinations are used as the state space; define a set of equipment layout adjustment operations as the action space; design a reward function to evaluate the pros and cons of the equipment layout scheme, and comprehensively consider the impact of the equipment layout on the total power consumption and heat dissipation performance of the integrated computer cabinet;

[0123] Build a deep reinforcement learning model, select a deep reinforcement learning algorithm based on value function, build a deep neural network as the approximate function, the input of the deep neural network is the current device layout state, and the output is the estimated value of each action taken under this state. Use mean square error as the loss function, and use an optimization algorithm to update the parameters of the deep neural network by minimizing the loss function;

[0124] By using the deep reinforcement learning algorithm to interact with the environment, training data is generated and the transfer experience is stored in the experience replay cache. A batch of transfer experiences are randomly extracted from the experience replay cache, the predicted value and target value of the current deep neural network are calculated, and the error is calculated according to the loss function. The parameters of the deep neural network are updated by optimizing the deep reinforcement learning algorithm.

[0125] For example, this paper proposes a method for optimizing the layout of computer integrated cabinet equipment based on deep reinforcement learning, which comprehensively considers equipment energy consumption, priority, heat dissipation requirements and cabinet heat dissipation conditions, and automatically generates the optimal equipment layout plan through intelligent algorithms. The specific technical implementation steps are as follows:

[0126] Construct a three-dimensional equipment energy consumption model. In order to accurately evaluate the energy consumption and heat dissipation performance of different equipment layout schemes, it is necessary to construct a three-dimensional equipment energy consumption model. This model is based on the three-dimensional space of the integrated computer cabinet, considers the power requirements, working status, heat generation and other parameters of each device, and establishes the correlation between equipment energy consumption and layout position. For example, for a cabinet with a length × width × height of 60 × 80 × 200 cm, it can be divided into 12 × 16 × 10 three-dimensional grids, and the size of each grid is 5 × 5 × 20 cm. The power requirement of device A is 500 watts, and the heat generation during operation is 1000 joules / second. It is placed in the grid with coordinates (3, 5, 7), and the energy consumption value of the grid is 500 × 1000 = 500000 joules. By modeling and calculating the energy consumption of all devices, a three-dimensional equipment energy consumption matrix is ​​obtained, and each element of the matrix represents the energy consumption value of the corresponding grid.

[0127] Define the reinforcement learning environment. Model the equipment layout optimization problem of the integrated computer cabinet as a reinforcement learning environment. The state space is all possible equipment layout combinations, and each state is a three-dimensional matrix. The elements of the matrix represent the type of equipment placed at the corresponding position, such as 0 for an empty position, 1 for device A, 2 for device B, and so on. The action space is a set of equipment layout adjustment operations, such as moving a device from its current position to another empty position, or exchanging the positions of two devices. The reward function is used to evaluate the pros and cons of the equipment layout scheme, taking into account the impact of the layout on the total power consumption and heat dissipation performance of the cabinet. For example, the reward function can be designed as the weighted sum of the negative value of the total power consumption of the cabinet and the heat dissipation cost, and the weight coefficient reflects the relative importance of the two. When the total power consumption of the cabinet is reduced and the heat dissipation conditions are improved, the reward value is higher; otherwise, the reward value is lower. The goal of reinforcement learning is to find an equipment layout scheme that maximizes the cumulative reward.

[0128] Build a deep reinforcement learning model. Use a value-function-based deep reinforcement learning algorithm, such as the DQN (Deep Q-Network) algorithm, to build a deep neural network as an approximation of the Q function. The input of the network is the current device layout state matrix, and the output is the estimated Q value of each layout adjustment operation taken in this state. The network structure can use a convolutional neural network (CNN) or a three-dimensional CNN to effectively extract the spatial features of the state matrix. The loss function uses the mean square error (MSE), which is the square of the difference between the Q value predicted by the network and the target Q value. Update the network parameters by minimizing the loss function so that the predicted Q value is as close to the target Q value as possible. The optimization algorithm can be selected from stochastic gradient descent (SGD), Adam, etc. The learning rate is set to 0.001 and the batch size is set to 32.

[0129] Generate training data. Generate training data through the interaction between the agent and the reinforcement learning environment. Based on the current device layout state, the agent uses the ε-greedy strategy to select a layout adjustment operation to obtain the next state and immediate reward. Store this set of transfer experience (current state, operation, reward, next state) in the Experience Replay Buffer. The cache size is set to 10,000, which means that a maximum of 10,000 sets of transfer experience can be stored. When the cache is full, the new transfer experience will overwrite the earliest experience. By continuously interacting with the environment, the agent gradually explores the state space and accumulates valuable transfer experience.

[0130] Update the deep reinforcement learning model. Randomly extract a batch of transfer experiences (e.g., 128 groups) from the experience replay cache, and use this batch of experiences to train and update the deep neural network. For each group of experiences, calculate the predicted Q value of the action taken in the current state, and the estimated maximum Q value of the next state. According to the update formula of the Q learning algorithm, calculate the target Q value, which is the discounted sum of the immediate reward and the maximum Q value of the next state. Substitute the predicted Q value and the target Q value into the loss function, calculate the gradient of the network parameters, and update them using the optimization algorithm. Repeat this process many times (e.g., 1000 times) so that the predicted Q value of the network continues to approach the optimal Q value. During the training process, gradually reduce the value of ε from the initial 1.0 to 0.1 to balance exploration and exploitation.

[0131] The trained model is used to generate the optimal layout plan. After multiple rounds of training (e.g., 10,000 rounds), the deep reinforcement learning model eventually converges to a stable strategy. This strategy is applied to the actual layout optimization problem of the integrated computer cabinet equipment to automatically generate the optimal equipment layout plan. Specifically, starting from the initial layout state of the cabinet, the current state is input into the trained deep neural network to obtain the Q value estimate of each layout adjustment operation. The operation with the highest Q value is selected to obtain the next layout state. This process is repeated until a stable optimal layout state is reached. The layout plan is output to guide the actual equipment layout and installation.

[0132] Evaluate the optimization effect and continuously improve the model. Apply the equipment layout solution generated by deep reinforcement learning to the actual integrated computer cabinet to evaluate its energy consumption and heat dissipation performance. By comparing the total power consumption of the cabinet, equipment temperature distribution, heat dissipation efficiency and other indicators before and after optimization, quantitatively analyze the effect of layout optimization. At the same time, collect new data from actual applications, such as the actual energy consumption and temperature of the equipment, for continuous training and optimization of the deep reinforcement learning model. According to the evaluation results, adjust the model's hyperparameters, network structure, reward function design, etc., and continuously improve the model's generalization ability and practicality. Through continuous iterative optimization, the deep reinforcement learning model can adapt to different application scenarios of integrated computer cabinets and provide more accurate and efficient equipment layout optimization services.

[0133] Through the above steps, the computer integrated cabinet equipment layout optimization method based on deep reinforcement learning proposed in this paper can adaptively explore and optimize the equipment layout, comprehensively consider multiple factors such as equipment energy consumption, priority, and heat dissipation requirements, and automatically generate the optimal layout plan. Compared with traditional layout optimization methods based on rules or heuristic algorithms, deep reinforcement learning methods have stronger learning and adaptability and can handle more complex and dynamic application scenarios. At the same time, through the construction of a three-dimensional equipment energy consumption model and a reinforcement learning environment, this method can make full use of the spatial information of the computer integrated cabinet and improve the accuracy and efficiency of layout optimization. Deep reinforcement learning provides a new idea and method for the layout optimization of computer integrated cabinet equipment, which has broad application prospects.

[0134] In an optional embodiment,

[0135] When it is detected that the electronic and computer integrated cabinet has power supply anomaly, heat dissipation anomaly or equipment operation anomaly, the steps of applying a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status include:

[0136] Monitor and collect real-time data on the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet; build a diagnostic rule base for common fault types, and when the power supply status, heat dissipation status or equipment operation status is detected to be abnormal, extract fault features from the real-time data, match the extracted fault features with the rules in the diagnostic rule base, find out the rules that meet the conditions, and output the corresponding fault causes in the rules as diagnostic results;

[0137] Use machine learning algorithms to train historical operation data and build a fault diagnosis model. The historical operation data includes power supply status, heat dissipation status, and equipment operation status data as well as corresponding fault type labels. Input the real-time data into the trained fault diagnosis model to automatically identify the fault type and cause.

[0138] The real-time data is sorted by timestamp to construct a multivariate time series data set, feature extraction is performed on the multivariate time series data set to extract prediction features reflecting precursors to faults, the prediction features are trained using a time series analysis algorithm, a fault prediction model is constructed, and the risk of faults occurring in the future is predicted based on the feature change trend of historical operation data;

[0139] The diagnosis results and prediction results based on the rules in the diagnosis rule base and the time series analysis algorithm are integrated to output the failure causes and failure risks of the electronic and computer integrated cabinet, and an alarm is triggered in advance according to the failure prediction results to prompt the taking of preventive measures.

[0140] For example, this paper proposes an intelligent fault diagnosis and prediction method for integrated computer cabinets, which uses rule base, machine learning, time series analysis and other technologies to achieve rapid diagnosis and early warning of power supply, heat dissipation and equipment operation abnormalities. The specific technical implementation steps are as follows:

[0141] Real-time monitoring and data collection. Various sensors and monitoring devices are deployed in the integrated computer cabinet to collect data on power supply status, heat dissipation status and equipment operation status in real time. Power supply status data includes voltage, current, power factor, etc. Heat dissipation status data includes cabinet internal temperature, humidity, wind speed, etc. Equipment operation status data includes CPU utilization, memory occupancy, hard disk I / O, etc. Through industrial Ethernet or other communication protocols, the collected data is transmitted to the data processing platform in real time for storage and analysis. For example, the internal temperature of the cabinet is collected every 5 seconds, and a data point with a timestamp of "2023-05-20 14:30:00" and a temperature value of 35.6 degrees Celsius is obtained and stored in the time series database.

[0142] Fault diagnosis based on diagnostic rule base. Build a diagnostic rule base for common fault types. The rule base contains a set of diagnostic rules in the form of IF-THEN. Each rule consists of a fault characteristic condition and a corresponding fault cause. When the monitoring data meets the characteristic condition, the corresponding fault cause can be diagnosed. When the power supply status, heat dissipation status or equipment operation status is detected to be abnormal, the fault characteristics are extracted from the real-time data, the extracted characteristics are matched with the rules in the rule base, all the rules that meet the conditions are found, and the corresponding fault causes in the rules are output as the diagnosis results. For example, there is a rule in the rule base: "IF the internal temperature of the cabinet exceeds 40 degrees Celsius for 5 consecutive minutes THEN the cooling fan is faulty". When the internal temperature of the cabinet is monitored to exceed 40 degrees Celsius for 5 consecutive minutes, the diagnosis result is "cooling fan fault".

[0143] Fault diagnosis based on machine learning. Use machine learning algorithms such as decision trees, support vector machines, neural networks, etc. to train historical operation data and build a fault diagnosis model. Historical operation data includes power supply status, heat dissipation status, and equipment operation status data as well as corresponding fault type labels. Divide the historical data into training sets and test sets according to a certain ratio, use the training set data to train the model, and evaluate the diagnostic accuracy of the model on the test set. When the model reaches the expected performance indicators, it is deployed to the fault diagnosis system. When an abnormality is detected, the real-time monitoring data is input into the trained fault diagnosis model, and the model automatically identifies the fault type and cause and outputs the diagnosis result. For example, train a three-layer BP neural network model with 10 input nodes (corresponding to 10 monitoring indicators), 20 hidden layer nodes, and 5 output nodes (corresponding to 5 common fault types). When a set of abnormal monitoring data is input, the neural network outputs the fault type with the highest probability as "power module fault" with a probability of 0.85.

[0144] Fault prediction based on time series analysis. Sort the real-time monitoring data by timestamp to construct a multivariate time series data set. Each timestamp corresponds to a set of monitoring indicator values, forming a high-dimensional time series. Perform feature extraction on the multivariate time series data to extract predictive features that reflect the precursors to faults, such as mean, variance, trend, and periodicity. Use time series analysis algorithms, such as ARIMA and LSTM, to train the extracted predictive features and build a fault prediction model. Predict the risk probability of a fault in the future based on the characteristic change trend of historical operating data. For example, the time series of the temperature inside the cabinet is analyzed and it is found that it has a significant upward trend before the fault occurs. Train an LSTM model to predict the probability of a heat dissipation fault in the next hour based on the temperature changes in the past 24 hours. When the predicted probability exceeds the threshold (such as 0.7), an alarm is triggered to prompt preventive measures.

[0145] Combine the diagnosis and prediction results to output the cause of the fault and risk warning. Combine the diagnosis results based on the rule base and machine learning to output the most likely cause of the fault. At the same time, associate the fault prediction results obtained by time series analysis with the diagnosis results to determine the risk level of the fault. Provide corresponding processing suggestions and emergency plans based on the risk level. For example, when the diagnosis result is "power module failure" and the predicted probability of power failure in the next hour is 0.9, the system outputs: "Warning: Power module failure, high risk of failure, please stop the machine for maintenance immediately to avoid data loss and equipment damage!". Minimize the losses and impacts caused by failures through early warning and proactive processing.

[0146] Continuously optimize and update the fault diagnosis and prediction model. In actual applications, continuously collect the operating data and fault feedback information of the integrated computer cabinet to optimize and update the fault diagnosis and prediction model. Regularly evaluate the diagnostic accuracy and prediction accuracy of the model, and adapt the model to new fault modes and operating conditions through incremental learning or offline retraining. At the same time, continuously improve and expand the diagnostic rule base, incorporate more expert experience and case knowledge, and improve the coverage and diagnostic accuracy of the rule base. Through continuous optimization and updating, the performance and reliability of the fault diagnosis and prediction system are continuously improved, providing a strong guarantee for the safe and stable operation of the integrated computer cabinet.

[0147] Through the above steps, the intelligent fault diagnosis and prediction method for integrated computer cabinets proposed in this paper integrates multiple technical means such as rule base, machine learning and time series analysis. It can timely and accurately diagnose the power supply, heat dissipation and equipment operation faults of the cabinet, and warn of fault risks in advance, providing decision support for the rapid location and handling of faults. Compared with the traditional post-event passive processing method, this method can minimize the downtime and data loss of faults, and improve the overall availability and reliability of the cabinet. At the same time, through the data-driven modeling method, this method has good adaptability and scalability, and can continuously optimize and improve the performance of diagnosis and prediction. Intelligent fault diagnosis and prediction is a key technology to ensure the safe and reliable operation of integrated computer cabinets, and is of great significance to improving the intelligence level and operation and maintenance efficiency of cabinets.

[0148] Figure 2 FIG. 1 is a schematic diagram of the structure of the intelligent power management and distribution system for an integrated computer cabinet according to an embodiment of the present invention. Figure 2 As shown, the system comprises:

[0149] The first unit is used to obtain the state parameters of each device in the electronic and computer integrated cabinet, the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices;

[0150] The second unit is used to determine the total power demand and optimal equipment layout plan of the electronic and computer integrated cabinet according to the three-dimensional equipment energy consumption model, combined with the power demand, equipment priority, heat dissipation demand and heat dissipation conditions of each device, and use a deep reinforcement learning algorithm, and jointly configure the power supply and heat dissipation of each device according to the total power demand, the power demand, priority and optimal equipment layout plan of each device to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, control the power supply path and heat dissipation system of the electronic and computer integrated cabinet, supply power and heat dissipation to each device, and detect the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet in real time;

[0151] The third unit is used to determine the cause of the fault and the risk of the fault by using a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, if it is detected that the electronic integrated computer cabinet has power supply anomalies, heat dissipation anomalies or equipment operation anomalies. According to the cause of the fault and the risk of the fault, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme. According to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device. At the same time, according to the fault prediction results, preventive measures are taken in advance for high-risk equipment until the electronic integrated computer cabinet resumes normal power supply and heat dissipation.

[0152] According to a third aspect of the embodiments of the present invention,

[0153] An electronic device is provided, comprising:

[0154] processor;

[0155] a memory for storing processor-executable instructions;

[0156] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0157] According to a fourth aspect of the embodiments of the present invention,

[0158] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0159] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0160] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent power management and distribution method for an integrated computer cabinet, characterized in that: include: Acquire the status parameters of each device in the electronic and computer integrated cabinet, the status parameters include power parameters, communication status parameters, power supply status parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the status parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices; According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priorities, heat dissipation requirements and heat dissipation conditions of each device, a deep reinforcement learning algorithm is used to determine the total power requirements and optimal equipment layout plan of the integrated computer cabinet, and according to the total power requirements, the power requirements, priorities and optimal equipment layout plan of each device, the power supply and heat dissipation of each device are jointly configured to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, the power supply path and heat dissipation system of the integrated computer cabinet are controlled to power and dissipate heat for each device, and the power supply status, heat dissipation status and equipment operation status of the integrated computer cabinet are detected in real time; If it is detected that the electronic and computer integrated cabinet has power supply anomalies, heat dissipation anomalies or equipment operation anomalies, then based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, a fault diagnosis algorithm and a fault prediction algorithm are used to determine the cause of the fault and the risk of the fault, and according to the cause of the fault and the risk of the fault, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme, and according to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device, and at the same time, according to the fault prediction results, preventive measures are taken in advance for high-risk equipment until the electronic and computer integrated cabinet resumes normal power supply and heat dissipation; Continuously collect the temperature and refrigeration equipment energy consumption data of the electronic and computer integrated cabinet within a preset evaluation period, calculate the heat dissipation effect improvement rate within the evaluation period based on the collected temperature data, calculate the energy consumption saving rate within the evaluation period based on the collected energy consumption data, and judge the effectiveness of the heat dissipation optimization control according to the preset heat dissipation effect improvement rate threshold and energy consumption saving rate threshold; when the heat dissipation effect improvement rate is lower than the heat dissipation effect improvement rate threshold or the energy consumption saving rate is lower than the energy consumption saving rate threshold, trigger an alarm and adjust the heat dissipation optimization algorithm parameters, and re-execute the heat dissipation optimization control until the heat dissipation effect improvement rate and the energy consumption saving rate both reach the corresponding thresholds; A global cost function is constructed as a weighted combination of heat dissipation cost, energy consumption cost and performance cost. The intelligent controller uses a heuristic optimization algorithm to generate a global collaborative control strategy with the global cost function as the optimization target. The output of the global collaborative control strategy includes reference control instructions and constraints of each control unit. Each control unit uses a multi-agent reinforcement learning algorithm to learn the local optimal control strategy under the constraints of the global collaborative control strategy, and reports the learned local optimal control strategy to the intelligent controller. The intelligent controller evaluates the impact of the local optimal control strategy on the global cost function and uses the local optimal control strategy that is better than the current global collaborative control strategy to update the global collaborative control strategy. Repeat the global collaborative strategy update of the intelligent controller until the global cost function value converges, the maximum number of iterations is reached, the computing resources are exhausted, or the time limit is reached. Output the current global collaborative control strategy as the optimal strategy for the intelligent control of heat dissipation of the integrated computer cabinet to generate an equipment heat dissipation optimization model.

2. The method according to claim 1, characterized in that: The state parameters of each device in the electronic and computer integrated cabinet are obtained, and the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters. The steps of establishing a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device include: Using three-dimensional drawing software, according to the design drawings and actual dimensions of the electronic integrated cabinet, draw the outline and internal partitions of the electronic integrated cabinet, define the geometric parameters of the length, width and height of the electronic integrated cabinet, and generate a three-dimensional model of the equipment of the electronic integrated cabinet according to the installation position and arrangement of the equipment; The power parameters of the equipment in the integrated computer cabinet are mapped to the color attributes of the three-dimensional model of the equipment. The color values ​​representing different power levels are generated according to the real-time power data of each equipment. The color values ​​are assigned to the color attributes of the corresponding equipment model to generate a three-dimensional heat map reflecting the power distribution of the equipment. The communication links of the equipment in the integrated computer cabinet are mapped to the lines between the three-dimensional models of the equipment. According to the actual communication links between each device, three-dimensional lines representing the links are drawn between the corresponding equipment models. The color of the three-dimensional lines represents the type of the link, and the style of the three-dimensional lines represents the communication protocol, so as to generate a three-dimensional network diagram reflecting the communication topology of the equipment. The power supply circuits of the equipment in the integrated computer cabinet are mapped to the attribute labels of the equipment 3D model. According to the power supply circuit to which each equipment is connected, attribute labels representing the power supply circuits are added to the corresponding equipment models. The colors or shapes of the labels represent different power supply circuits, and a 3D circuit diagram reflecting the power supply distribution of the equipment is generated. The generated three-dimensional heat map, three-dimensional network diagram and three-dimensional circuit diagram are superimposed in the same three-dimensional space to construct a three-dimensional equipment energy consumption model that comprehensively reflects the energy consumption characteristics of the equipment inside the electronic and computer integration cabinet.

3. The method according to claim 2, characterized in that According to the total power demand of the integrated computer cabinet, the power demand of each device, the priority and the optimal device layout plan, the power supply and heat dissipation of each device are jointly configured, and the steps of generating the first power supply and heat dissipation plan include: According to the total power demand of the integrated computer cabinet, the power demand and priority of each device, a multi-objective optimization algorithm is used to establish a mathematical model of equipment power supply. The power supply quality and energy efficiency of the equipment are taken as optimization goals, and the power demand and priority of the equipment are taken as constraints to construct a multi-objective optimization problem. A non-dominated sorting genetic algorithm is used to solve the multi-objective optimization problem to obtain an optimal solution set of power supply voltage and current parameters, and a solution that meets the power supply quality and energy efficiency requirements of the equipment is selected from the optimal solution set according to the equipment priority as the optimal power supply voltage and current parameters. According to the optimal equipment layout plan and equipment heat dissipation requirements, a three-dimensional computational fluid dynamics model of the inside of the electronic and computer integrated cabinet is established using computational fluid dynamics simulation technology, and the three-dimensional computational fluid dynamics model includes an equipment model, an air duct model, and a cold source model; Determine the boundary conditions and solution settings of the three-dimensional computational fluid dynamics model, obtain the temperature field, velocity field and pressure field distribution inside the electronic integrated cabinet, optimize the size, shape and layout of the air duct of the electronic integrated cabinet according to the computational fluid dynamics simulation results, adjust the position, number and cooling capacity of the cold source, maximize the use of natural cooling and forced cooling, and generate a first power supply and heat dissipation solution.

4. The method according to claim 1, characterized in that The method further comprises: A collaborative optimization method is adopted to comprehensively consider the mutual influence and constraint relationship between power supply and heat dissipation, couple the equipment power supply mathematical model and the heat dissipation optimization model, and establish a joint optimization model of power supply and heat dissipation. The total power consumption and heat dissipation performance of the electronic and computer integrated cabinet are taken as optimization goals, and the power demand, priority and heat dissipation demand of the equipment are taken as constraints. The joint optimization model is solved using an intelligent optimization algorithm, and the optimal power supply and heat dissipation parameter combination obtained by the solution is used as the control instruction for the power supply and heat dissipation of the equipment, so as to collaboratively adjust the power supply and heat dissipation status of the equipment.

5. The method according to claim 1, characterized in that According to the three-dimensional equipment energy consumption model, combined with the power requirements, equipment priority, heat dissipation requirements of each device and the heat dissipation conditions of the electronic and computer integrated cabinet, the steps of applying the deep reinforcement learning algorithm include: Define the reinforcement learning environment, represent the equipment layout state of the integrated computer cabinet as a three-dimensional matrix, each element of the matrix corresponds to a position in the integrated computer cabinet, the value of the element represents the type of equipment placed at the position, and all equipment layout combinations are used as the state space; define a set of equipment layout adjustment operations as the action space; design a reward function to evaluate the pros and cons of the equipment layout scheme, and comprehensively consider the impact of the equipment layout on the total power consumption and heat dissipation performance of the integrated computer cabinet; Build a deep reinforcement learning model, select a deep reinforcement learning algorithm based on value function, build a deep neural network as the approximate function, the input of the deep neural network is the current device layout state, and the output is the estimated value of each action taken under this state. Use mean square error as the loss function, and use an optimization algorithm to update the parameters of the deep neural network by minimizing the loss function; By using the deep reinforcement learning algorithm to interact with the environment, training data is generated and the transfer experience is stored in the experience replay cache. A batch of transfer experiences are randomly extracted from the experience replay cache, the predicted value and target value of the current deep neural network are calculated, and the error is calculated according to the loss function. The parameters of the deep neural network are updated by optimizing the deep reinforcement learning algorithm.

6. The method according to claim 1, characterized in that When it is detected that the electronic and computer integrated cabinet has power supply anomaly, heat dissipation anomaly or equipment operation anomaly, the steps of applying a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status include: Monitor and collect real-time data on the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet; build a diagnostic rule base for common fault types, and when the power supply status, heat dissipation status or equipment operation status is detected to be abnormal, extract fault features from the real-time data, match the extracted fault features with the rules in the diagnostic rule base, find out the rules that meet the conditions, and output the corresponding fault causes in the rules as diagnostic results; Use machine learning algorithms to train historical operation data and build a fault diagnosis model. The historical operation data includes power supply status, heat dissipation status, and equipment operation status data as well as corresponding fault type labels. Input the real-time data into the trained fault diagnosis model to automatically identify the fault type and cause. The real-time data is sorted by timestamp to construct a multivariate time series data set, feature extraction is performed on the multivariate time series data set to extract prediction features reflecting precursors to faults, the prediction features are trained using a time series analysis algorithm, a fault prediction model is constructed, and the risk of faults occurring in the future is predicted based on the feature change trend of historical operation data; The diagnosis results and prediction results based on the rules in the diagnosis rule base and the time series analysis algorithm are integrated to output the failure causes and failure risks of the electronic and computer integrated cabinet, and an alarm is triggered in advance according to the failure prediction results to prompt the taking of preventive measures.

7. An intelligent power management and distribution system for an integrated computer cabinet, used to implement the method described in any one of claims 1 to 6, characterized in that: include: The first unit is used to obtain the state parameters of each device in the electronic and computer integrated cabinet, the state parameters include power parameters, communication state parameters, power supply state parameters and device location parameters, and establish a three-dimensional device energy consumption model of the electronic and computer integrated cabinet based on the state parameters of each device, the three-dimensional device energy consumption model includes the three-dimensional location distribution of each device, power demand, communication link, power supply path and thermal coupling relationship between devices; The second unit is used to determine the total power demand and optimal equipment layout plan of the electronic and computer integrated cabinet according to the three-dimensional equipment energy consumption model, combined with the power demand, equipment priority, heat dissipation demand and heat dissipation conditions of each device, and use a deep reinforcement learning algorithm, and jointly configure the power supply and heat dissipation of each device according to the total power demand, the power demand, priority and optimal equipment layout plan of each device to generate a first power supply and heat dissipation plan; according to the first power supply and heat dissipation plan, control the power supply path and heat dissipation system of the electronic and computer integrated cabinet, supply power and heat dissipation to each device, and detect the power supply status, heat dissipation status and equipment operation status of the electronic and computer integrated cabinet in real time; The third unit is used to determine the cause of the fault and the risk of the fault by using a fault diagnosis algorithm and a fault prediction algorithm based on the three-dimensional equipment energy consumption model and the real-time detected power supply status, heat dissipation status and equipment operation status, if it is detected that the electronic integrated computer cabinet has power supply anomalies, heat dissipation anomalies or equipment operation anomalies. According to the cause of the fault and the risk of the fault, the first power supply and heat dissipation scheme is adjusted to generate a second power supply and heat dissipation scheme. According to the second power supply and heat dissipation scheme, the power supply circuit and the heat dissipation system are controlled to re-supply and dissipate heat to each device. At the same time, according to the fault prediction results, preventive measures are taken in advance for high-risk equipment until the electronic integrated computer cabinet resumes normal power supply and heat dissipation.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images