A data center micro power supply capacity optimization method and system

Abstract

The application provides a data center micro power supply capacity optimization method and system, the method comprises the following steps: constructing a corresponding three-dimensional constraint parameter group in real time according to real-time working parameters; obtaining a first correlation coefficient between power balance parameters and equipment capacity parameters through linear correlation, a second correlation coefficient between equipment capacity parameters and voltage limit parameters through nonlinear correlation, and a third correlation coefficient between voltage limit parameters and power balance parameters through mapping correlation; obtaining a power output prediction value of a photovoltaic micro-grid, and combining the first correlation coefficient, the second correlation coefficient and the third correlation coefficient to construct a corresponding capacity optimization function in real time; optimizing a preset particle swarm algorithm to generate a corresponding target optimization algorithm, and solving the capacity optimization function through the target optimization algorithm to output a corresponding photovoltaic capacity optimization value in real time. The application can objectively and dynamically optimize the capacity of the photovoltaic micro power supply, and improves the optimization efficiency.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for optimizing the micro power supply capacity of a data center. Background Technology

[0002] As carbon reduction goals continue to advance, photovoltaics has become an important energy source for data center microgrids. Microgrids need to continuously achieve dynamic matching between energy storage and load through photovoltaic capacity optimization. This means that photovoltaic capacity optimization needs to be integrated into the overall operation of the microgrid, rather than matching a single power output.

[0003] As a coupled system, microgrids need to follow power balance constraints, equipment capacity constraints, and voltage limit constraints during operation, thus requiring comprehensive consideration for corresponding capacity optimization.

[0004] Furthermore, in practical applications, the above three types of constraints are not independent but will affect each other. Existing traditional static optimization methods either reduce photovoltaic utilization due to conservative design or cause equipment failure due to constraints, making it difficult to adapt to the above three types of constraint relationships, which correspondingly reduces the optimization efficiency of microgrid capacity. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a method and system for optimizing the capacity of microgrids in data centers, so as to solve the problem that the existing technology cannot effectively balance the three types of constraints of microgrids, which leads to a reduction in the efficiency of microgrid capacity optimization.

[0006] The first aspect of the present invention proposes:

[0007] A method for optimizing micro power supply capacity in a data center, wherein the method includes:

[0008] The real-time operating parameters of the photovoltaic microgrid are obtained, and a corresponding three-dimensional constraint parameter set is constructed in real time based on the real-time operating parameters. The three-dimensional constraint parameter set includes power balance parameters, equipment capacity parameters, and voltage limit parameters.

[0009] A first correlation coefficient between the power balance parameter and the equipment capacity parameter is obtained through linear correlation; a second correlation coefficient between the equipment capacity parameter and the voltage limiting parameter is obtained through nonlinear correlation; and a third correlation coefficient between the voltage limiting parameter and the power balance parameter is obtained through mapping correlation.

[0010] The predicted power output value of the photovoltaic microgrid is obtained, and the corresponding capacity optimization function is constructed in real time by combining the first correlation coefficient, the second correlation coefficient and the third correlation coefficient.

[0011] The preset particle swarm optimization algorithm is optimized using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate a corresponding target optimization algorithm. The capacity optimization function is then solved using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time.

[0012] The beneficial effects of this invention are as follows: By acquiring the real-time operating parameters of the photovoltaic microgrid, a three-dimensional constraint array for subsequent comprehensive analysis can be created. Based on this, the required first correlation coefficient, second correlation coefficient, and third correlation coefficient can be obtained sequentially, and the required capacity optimization function can be created. Based on this, the target optimization algorithm constructed in real time can be solved by the capacity optimization function, thereby objectively and accurately calculating the photovoltaic capacity optimization value adapted to the current photovoltaic microgrid. This allows for dynamic adjustment of the photovoltaic microgrid capacity, thus improving optimization efficiency.

[0013] Furthermore, the step of obtaining the predicted power output value of the photovoltaic microgrid and constructing the corresponding capacity optimization function in real time by combining the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient includes:

[0014] The power output data and corresponding meteorological observation data generated by the photovoltaic microgrid within a preset time period are obtained, and the meteorological observation data are decomposed into three feature dimensions: light intensity, ambient temperature and wind speed.

[0015] The power output data is trained by nonlinear mapping of the three feature dimensions to the power output data using the random forest algorithm to generate a corresponding power prediction model. The power prediction model is then used to output the predicted power output value that is compatible with the photovoltaic microgrid in real time.

[0016] The first correlation coefficient, the second correlation coefficient, and the third correlation coefficient are subjected to matrix orthogonalization to generate mutually independent parameter influence factors, and the capacity optimization function is constructed based on the parameter influence factors and the predicted power output value.

[0017] Furthermore, the step of constructing the capacity optimization function based on the parameter influence factor and the predicted power output value includes:

[0018] The parameter influence factors are constructed into an influence factor matrix according to the time series, and the power output prediction value is Fourier decomposed to obtain the fundamental component and harmonic component. The mapping relationship between the influence factor matrix and the fundamental component and harmonic component is constructed in real time.

[0019] Based on the mapping relationship, the parameter sensitivity coefficients for each time period are calculated in real time, and the short-circuit capacity parameters of the photovoltaic microgrid are obtained simultaneously, so as to construct the corresponding multi-dimensional constraint boundary in real time according to the parameter sensitivity coefficients and the short-circuit capacity parameters.

[0020] The influence factor matrix is ​​convolved with the multi-dimensional constraint boundary to generate a corresponding dynamic constraint vector. The capacity optimization function is then constructed in real time using the dynamic constraint vector as the boundary condition and maximizing the energy utilization rate of the photovoltaic microgrid as the objective.

[0021] Furthermore, the step of optimizing the preset particle swarm optimization algorithm using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate the corresponding target optimization algorithm includes:

[0022] The square of the first correlation coefficient is used as the reference value of the inertia weight in the preset particle swarm algorithm, and it is linearly decayed according to the number of iterations, with the decay rate being the ratio of the reference value to the total number of iterations.

[0023] Calculate the absolute value of the product of the second correlation coefficient and the voltage limit parameter, and use the absolute value of the product as the upper limit threshold of the particle velocity, and construct the penalty coefficient of particle fitness with the third correlation coefficient as the core.

[0024] The particle swarm optimization algorithm, after undergoing dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient, is used as the target optimization algorithm.

[0025] Furthermore, the step of using the particle swarm optimization algorithm after dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient as the target optimization algorithm includes:

[0026] The instantaneous changes in the power balance parameters are collected, and the instantaneous changes are convolved with the real-time attenuation value of the first correlation coefficient to generate the corresponding dynamic correction coefficient.

[0027] Based on the ratio of the real-time monitored value to the rated value of the equipment capacity parameter, the corresponding particle velocity adjustment threshold is determined, and the fluctuation frequency of the voltage limit parameter is extracted. The fluctuation frequency is then weighted and summed with the third correlation coefficient to obtain the particle fitness penalty coefficient.

[0028] The performance of the particle swarm optimization algorithm, after dynamic correction coefficient calibration, velocity threshold adjustment, and weighted penalty coefficient configuration, is verified to generate the target optimization algorithm accordingly.

[0029] Furthermore, the step of solving the capacity optimization function using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time includes:

[0030] The predicted power output value is divided into several sub-intervals according to the linear distribution characteristics, and the local optimization calculation of the target optimization algorithm is performed on each of the sub-intervals.

[0031] The nonlinear feature points of the second correlation coefficient are extracted as key verification nodes in the optimization process, and when the target optimization algorithm iterates to the key verification node, the candidate solutions output by the target optimization algorithm are collected.

[0032] The mapping correlation matrix of the third correlation coefficient is used as the screening template for the global optimal solution, and the candidate solutions output by the target optimization algorithm are substituted into the mapping correlation matrix for dimensionality adaptation verification, so as to select the photovoltaic capacity optimization value accordingly.

[0033] Furthermore, the step of substituting the candidate solutions output by the target optimization algorithm into the mapping correlation matrix for dimensionality adaptation verification, so as to select the optimized photovoltaic capacity value, includes:

[0034] The candidate solution is decomposed according to the three dimensions of the power balance parameter, the equipment capacity parameter and the voltage limit parameter to obtain a three-dimensional component group that fits the mapping correlation matrix.

[0035] Calculate the cosine similarity between the three-dimensional component group and the corresponding reference vector in the mapping association matrix, and select candidate solutions with a cosine similarity greater than a preset threshold as preliminary fitting solutions;

[0036] The preliminary fit solution is calculated using the capacity optimization function, and the candidate solution whose calculation result satisfies the global optimal condition is selected as the photovoltaic capacity optimization value.

[0037] The second aspect of the present invention proposes:

[0038] A data center micro power capacity optimization system, wherein the system comprises:

[0039] The acquisition module is used to acquire the real-time operating parameters of the photovoltaic microgrid and construct the corresponding three-dimensional constraint parameter set in real time based on the real-time operating parameters. The three-dimensional constraint parameter set includes power balance parameters, equipment capacity parameters and voltage limit parameters.

[0040] The processing module is used to obtain a first correlation coefficient between the power balance parameter and the equipment capacity parameter through linear correlation, a second correlation coefficient between the equipment capacity parameter and the voltage limit parameter through nonlinear correlation, and a third correlation coefficient between the voltage limit parameter and the power balance parameter through mapping correlation.

[0041] The module is used to obtain the predicted power output value of the photovoltaic microgrid and, in combination with the first correlation coefficient, the second correlation coefficient and the third correlation coefficient, construct the corresponding capacity optimization function in real time.

[0042] The optimization module is used to optimize the preset particle swarm optimization algorithm using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate a corresponding target optimization algorithm, and to solve the capacity optimization function using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time.

[0043] Furthermore, the building module is specifically used for:

[0044] The power output data and corresponding meteorological observation data generated by the photovoltaic microgrid within a preset time period are obtained, and the meteorological observation data are decomposed into three feature dimensions: light intensity, ambient temperature and wind speed.

[0045] The power output data is trained by nonlinear mapping of the three feature dimensions to the power output data using the random forest algorithm to generate a corresponding power prediction model. The power prediction model is then used to output the predicted power output value that is compatible with the photovoltaic microgrid in real time.

[0046] The first correlation coefficient, the second correlation coefficient, and the third correlation coefficient are subjected to matrix orthogonalization to generate mutually independent parameter influence factors, and the capacity optimization function is constructed based on the parameter influence factors and the predicted power output value.

[0047] Furthermore, the building module is specifically used for:

[0048] The parameter influence factors are constructed into an influence factor matrix according to the time series, and the power output prediction value is Fourier decomposed to obtain the fundamental component and harmonic component. The mapping relationship between the influence factor matrix and the fundamental component and harmonic component is constructed in real time.

[0049] Based on the mapping relationship, the parameter sensitivity coefficients for each time period are calculated in real time, and the short-circuit capacity parameters of the photovoltaic microgrid are obtained simultaneously, so as to construct the corresponding multi-dimensional constraint boundary in real time according to the parameter sensitivity coefficients and the short-circuit capacity parameters.

[0050] The influence factor matrix is ​​convolved with the multi-dimensional constraint boundary to generate a corresponding dynamic constraint vector. The capacity optimization function is then constructed in real time using the dynamic constraint vector as the boundary condition and maximizing the energy utilization rate of the photovoltaic microgrid as the objective.

[0051] Furthermore, the optimization module is specifically used for:

[0052] The square of the first correlation coefficient is used as the reference value of the inertia weight in the preset particle swarm algorithm, and it is linearly decayed according to the number of iterations, with the decay rate being the ratio of the reference value to the total number of iterations.

[0053] Calculate the absolute value of the product of the second correlation coefficient and the voltage limit parameter, and use the absolute value of the product as the upper limit threshold of the particle velocity, and construct the penalty coefficient of particle fitness with the third correlation coefficient as the core.

[0054] The particle swarm optimization algorithm, after undergoing dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient, is used as the target optimization algorithm.

[0055] Furthermore, the optimization module is specifically used for:

[0056] The instantaneous changes in the power balance parameters are collected, and the instantaneous changes are convolved with the real-time attenuation value of the first correlation coefficient to generate the corresponding dynamic correction coefficient.

[0057] Based on the ratio of the real-time monitored value to the rated value of the equipment capacity parameter, the corresponding particle velocity adjustment threshold is determined, and the fluctuation frequency of the voltage limit parameter is extracted. The fluctuation frequency is then weighted and summed with the third correlation coefficient to obtain the particle fitness penalty coefficient.

[0058] The performance of the particle swarm optimization algorithm, after dynamic correction coefficient calibration, velocity threshold adjustment, and weighted penalty coefficient configuration, is verified to generate the target optimization algorithm accordingly.

[0059] Furthermore, the optimization module is specifically used for:

[0060] The predicted power output value is divided into several sub-intervals according to the linear distribution characteristics, and the local optimization calculation of the target optimization algorithm is performed on each of the sub-intervals.

[0061] The nonlinear feature points of the second correlation coefficient are extracted as key verification nodes in the optimization process, and when the target optimization algorithm iterates to the key verification node, the candidate solutions output by the target optimization algorithm are collected.

[0062] The mapping correlation matrix of the third correlation coefficient is used as the screening template for the global optimal solution, and the candidate solutions output by the target optimization algorithm are substituted into the mapping correlation matrix for dimensionality adaptation verification, so as to select the photovoltaic capacity optimization value accordingly.

[0063] Furthermore, the optimization module is specifically used for:

[0064] The candidate solution is decomposed according to the three dimensions of the power balance parameter, the equipment capacity parameter and the voltage limit parameter to obtain a three-dimensional component group that fits the mapping correlation matrix.

[0065] Calculate the cosine similarity between the three-dimensional component group and the corresponding reference vector in the mapping association matrix, and select candidate solutions with a cosine similarity greater than a preset threshold as preliminary fitting solutions;

[0066] The preliminary fit solution is calculated using the capacity optimization function, and the candidate solution whose calculation result satisfies the global optimal condition is selected as the photovoltaic capacity optimization value.

[0067] The third aspect of the present invention proposes:

[0068] A computer includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the data center micro power capacity optimization method as described above.

[0069] The fourth aspect of the present invention proposes:

[0070] A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the data center micro power capacity optimization method as described above.

[0071] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0072] Figure 1 A flowchart of a data center micro power capacity optimization method provided in the first embodiment of the present invention;

[0073] Figure 2 The structural block diagram of the data center micro power capacity optimization system provided in the third embodiment of the present invention is shown.

[0074] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0075] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0076] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0077] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0078] Please see Figure 1 The image shows a data center micro-power supply capacity optimization method provided in the first embodiment of the present invention. The data center micro-power supply capacity optimization method provided in this embodiment can objectively and dynamically optimize the capacity of photovoltaic micro-power supplies, thereby improving the optimization efficiency.

[0079] Specifically, this embodiment provides:

[0080] A method for optimizing micro power supply capacity in a data center specifically includes the following steps:

[0081] Step S10: Obtain the real-time operating parameters of the photovoltaic microgrid, and construct the corresponding three-dimensional constraint parameter set in real time based on the real-time operating parameters. The three-dimensional constraint parameter set includes power balance parameters, equipment capacity parameters, and voltage limit parameters.

[0082] It is important to note that, firstly, real-time operational data of the photovoltaic microgrid (such as output power, equipment load, and node voltage) must be collected, and three types of core parameters are extracted to form a three-dimensional constraint parameter set. Specifically, the power balance parameters reflect the dynamic matching relationship between power generation and load demand in the microgrid (such as real-time power difference and energy storage charging / discharging capacity), and are the fundamental constraints for maintaining system stability. The equipment capacity parameters cover the rated capacity and maximum output limits of photovoltaic modules, energy storage devices, etc., determining the physical capability boundaries of the system. The voltage limit parameters include the voltage fluctuation range of each node and the allowable voltage deviation threshold, ensuring the safety of power transmission and the normal operation of equipment, thus facilitating subsequent processing.

[0083] Step S20: Obtain a first correlation coefficient between the power balance parameter and the equipment capacity parameter through linear correlation, obtain a second correlation coefficient between the equipment capacity parameter and the voltage limiting parameter through nonlinear correlation, and obtain a third correlation coefficient between the voltage limiting parameter and the power balance parameter through mapping correlation;

[0084] It should be noted that there is usually a linear relationship between power balance and equipment capacity (e.g., the larger the equipment capacity, the wider the adjustable power balance range), and the correlation strength is calculated using methods such as linear regression. This yields the first correlation coefficient mentioned above. The relationship between equipment capacity and voltage limitation is non-linear, influenced by factors such as line loss and load characteristics (e.g., capacity overload may cause a sudden voltage drop), and is solved using non-linear fitting (e.g., polynomial regression). This yields the second correlation coefficient mentioned above. Voltage limitation indirectly affects power balance by influencing power transmission efficiency, and a correlation model needs to be established using a mapping function (e.g., voltage-power loss curve). This yields the third correlation coefficient mentioned above, which is used for subsequent processing.

[0085] Step S30: Obtain the predicted power output value of the photovoltaic microgrid, and construct the corresponding capacity optimization function in real time by combining the first correlation coefficient, the second correlation coefficient and the third correlation coefficient;

[0086] It's important to note that predicting the future output of the photovoltaic system based on historical data and environmental factors (such as the difference in capacity demand between sunny and cloudy days) is a prerequisite for optimization. Based on this, the predicted power value is coupled with three-dimensional constraint parameters using correlation coefficients as weights to form an objective function that simultaneously satisfies power matching, equipment capacity, and voltage safety. This allows the generation of the necessary capacity optimization function for subsequent processing.

[0087] Step S40: The preset particle swarm optimization algorithm is optimized using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate a corresponding target optimization algorithm. The capacity optimization function is then solved using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time.

[0088] It should be noted that the inertia weight and particle velocity parameters of the PSO are dynamically adjusted using correlation coefficients to better adapt to the constraints of the microgrid. Based on this, the capacity optimization function is input into the target algorithm, and the photovoltaic capacity value that satisfies all constraints and optimizes system performance (such as energy efficiency) is found through iterative optimization. This method enables objective and dynamic optimization of the capacity of the Optics Valley microgrid, thereby improving optimization efficiency.

[0089] Second Embodiment

[0090] Furthermore, the step of obtaining the predicted power output value of the photovoltaic microgrid and constructing the corresponding capacity optimization function in real time by combining the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient includes:

[0091] The power output data and corresponding meteorological observation data generated by the photovoltaic microgrid within a preset time period are obtained, and the meteorological observation data are decomposed into three feature dimensions: light intensity, ambient temperature and wind speed.

[0092] The power output data is trained by nonlinear mapping of the three feature dimensions to the power output data using the random forest algorithm to generate a corresponding power prediction model. The power prediction model is then used to output the predicted power output value that is compatible with the photovoltaic microgrid in real time.

[0093] The first correlation coefficient, the second correlation coefficient, and the third correlation coefficient are subjected to matrix orthogonalization to generate mutually independent parameter influence factors, and the capacity optimization function is constructed based on the parameter influence factors and the predicted power output value.

[0094] It should be noted that the process involves collecting photovoltaic power output data and corresponding meteorological data for a preset time period (e.g., the past year). The meteorological data is broken down into three key features: light intensity directly determines the power generation efficiency of photovoltaic panels and is the core influencing factor; excessively high ambient temperature reduces the conversion efficiency of photovoltaic modules and requires separate modeling; and wind speed affects the heat dissipation efficiency of photovoltaic panels, indirectly affecting the output power. Based on this, a random forest algorithm is used to establish a nonlinear mapping between meteorological features and power output in real time. This is combined with existing neural networks to train a corresponding power prediction model in real time. Based on this, by inputting current meteorological data in real time, the power prediction model can output the predicted power output value for a future period (e.g., within one hour). To objectively and comprehensively construct the required capacity optimization function, the first, second, and third correlation coefficients are orthogonalized (e.g., using the Gram-Schmidt method) to eliminate multicollinearity among parameters and obtain mutually independent influencing factors. Based on this, the influencing factors are combined with the predicted power value to form a capacity optimization function with capacity as the variable and fused constraint weights, for subsequent processing.

[0095] Furthermore, the step of constructing the capacity optimization function based on the parameter influence factor and the predicted power output value includes:

[0096] The parameter influence factors are constructed into an influence factor matrix according to the time series, and the power output prediction value is Fourier decomposed to obtain the fundamental component and harmonic component. The mapping relationship between the influence factor matrix and the fundamental component and harmonic component is constructed in real time.

[0097] Based on the mapping relationship, the parameter sensitivity coefficients for each time period are calculated in real time, and the short-circuit capacity parameters of the photovoltaic microgrid are obtained simultaneously, so as to construct the corresponding multi-dimensional constraint boundary in real time according to the parameter sensitivity coefficients and the short-circuit capacity parameters.

[0098] The influence factor matrix is ​​convolved with the multi-dimensional constraint boundary to generate a corresponding dynamic constraint vector. The capacity optimization function is then constructed in real time using the dynamic constraint vector as the boundary condition and maximizing the energy utilization rate of the photovoltaic microgrid as the objective.

[0099] It should be noted that, firstly, the parameter influencing factors are arranged according to time series (e.g., 24 hours a day, 7 days a week) to construct an influencing factor matrix. This matrix reflects the changes in the impact of each constraint parameter on capacity optimization at different times (e.g., the influencing factor of power balance parameters is larger during strong daylight). Simultaneously, the predicted power output is Fourier decomposed, separating it into a fundamental component (corresponding to the stable power variation, such as the overall trend of daily output) and harmonic components (corresponding to the fluctuating power, such as short-term output fluctuations caused by cloud cover). Constructing the mapping relationship between the influencing factor matrix and the fundamental and harmonic components clarifies how the parameter influences at different times correspond to stable and fluctuating power variations (e.g., the influencing factor of the fundamental component should focus on long-term capacity matching, while the influencing factor of the harmonic component should focus on short-term capacity adjustment), providing multi-dimensional support for the subsequent construction of dynamic constraint boundaries. Based on this, and using the aforementioned mapping relationship, sensitivity analysis (e.g., changing the influence factor of a certain parameter and observing the magnitude of change in capacity optimization results) is used to calculate the parameter sensitivity coefficient for each time period. The larger the coefficient, the more sensitive the parameter is to capacity optimization during that time period (e.g., the sensitivity coefficient of the power balance parameter is larger during peak load periods). Simultaneously, the short-circuit capacity parameter of the photovoltaic microgrid is acquired (reflecting the system's ability to cope with short-circuit faults; the smaller the short-circuit capacity, the worse the system voltage stability). The parameter sensitivity coefficient is combined with the short-circuit capacity parameter to construct a multi-dimensional constraint boundary. For example, during periods with a large sensitivity coefficient and small short-circuit capacity, the constraint boundary of the voltage limiting parameter needs to be tightened to avoid voltage collapse due to improper capacity configuration. This constraint boundary can dynamically adapt to the system characteristics of different time periods, improving the practicality of the optimization function. Finally, the influence factor matrix is ​​convolved with the multi-dimensional constraint boundary. Essentially, this involves calculating the superposition effect of parameter influence and constraint boundary at different time periods using a sliding window to generate a dynamic constraint vector. Specifically, this vector reflects the comprehensive influence intensity of each constraint parameter within a given time period (e.g., in a time period with high power balance sensitivity and low short-circuit capacity, the corresponding dimension in the dynamic constraint vector will have a larger value, indicating stricter constraints). When constructing the capacity optimization function, the dynamic constraint vector is used as the boundary condition (ensuring that the optimization result does not violate the comprehensive constraints of each time period). The goal is to maximize the energy utilization rate of the photovoltaic microgrid (energy utilization rate = effective power generation / total installed capacity; maximizing this indicator can improve the economy and efficiency of capacity configuration). Through mathematical modeling (such as linear programming or nonlinear programming), the goal and constraints are combined to ultimately form a capacity optimization function that can adapt to dynamic operating scenarios.

[0100] Furthermore, the step of optimizing the preset particle swarm optimization algorithm using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate the corresponding target optimization algorithm includes:

[0101] The square of the first correlation coefficient is used as the reference value of the inertia weight in the preset particle swarm algorithm, and it is linearly decayed according to the number of iterations, with the decay rate being the ratio of the reference value to the total number of iterations.

[0102] Calculate the absolute value of the product of the second correlation coefficient and the voltage limit parameter, and use the absolute value of the product as the upper limit threshold of the particle velocity, and construct the penalty coefficient of particle fitness with the third correlation coefficient as the core.

[0103] The particle swarm optimization algorithm, after undergoing dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient, is used as the target optimization algorithm.

[0104] It's important to note that inertia weight is a key parameter in the particle swarm optimization algorithm that controls the initial velocity of particles. A larger weight indicates stronger global search capability (easier exploration of new solution spaces), while a smaller weight indicates stronger local optimization capability (easier convergence near the optimal solution). The square of the first correlation coefficient is used as the baseline value for the inertia weight because this coefficient reflects the linear relationship between power balance and equipment capacity, and its square amplifies the impact of this relationship on the algorithm (the stronger the relationship, the larger the baseline weight, and the wider the initial search range). Linear decay based on the number of iterations, with a decay rate of "baseline value / total number of iterations," ensures that the algorithm maintains a large inertia weight in the early stages of iteration (e.g., the first 50% of iterations) to fully explore the solution space; in the later stages (e.g., the last 50% of iterations), the weight gradually decreases, focusing on the vicinity of local optima, improving convergence speed and accuracy. For example, if the baseline value is 0.8 and the total number of iterations is 100, the weight decays by 0.008 with each iteration, gradually decreasing from 0.8 to 0.0, achieving a balance between search and convergence.

[0105] Based on this, particle velocity determines the particle's step size in the solution space. Excessive velocity can easily cause the particle to exceed the feasible solution range (e.g., capacity configuration exceeding equipment ratings). Therefore, an upper limit threshold needs to be set. The absolute value of the product of the second correlation coefficient and the voltage limit parameter is used as the upper limit threshold for velocity because the second correlation coefficient reflects the nonlinear relationship between equipment capacity and voltage. Combining it with the voltage limit parameter ensures that the velocity threshold adapts to voltage safety constraints (e.g., when voltage constraints are strict, a small absolute value of the product results in a low velocity threshold, a small particle step size, and avoids solutions violating voltage constraints). Simultaneously, a penalty coefficient for particle fitness is constructed using the third correlation coefficient as the core because the third correlation coefficient reflects the mapping relationship between voltage and power balance. When the solution corresponding to a particle violates voltage or power balance constraints, the penalty coefficient increases the fitness function value according to the magnitude of the third correlation coefficient (the larger the fitness value, the worse the feasibility of the solution), thereby guiding the particle to move towards the feasible solution region. For example, the larger the third correlation coefficient, the larger the penalty coefficient, and the more severe the penalty for solutions violating constraints, ensuring that the solution output by the algorithm conforms to the actual constraints. Finally, after optimization through three stages—"dynamic weight decay" (optimizing the balance between search and convergence), "adaptive velocity threshold constraint" (preventing solutions from exceeding constraints), and "directional configuration of penalty coefficients" (guiding solutions to move towards the feasible region)—the performance of the particle swarm optimization algorithm is significantly improved. It can extensively explore the solution space in the early stages of iteration and quickly converge to the optimal solution in the later stages, while also ensuring the feasibility of the solution through a penalty mechanism. Using this optimized algorithm as the target optimization algorithm can specifically address the characteristics of "numerous constraints and strong nonlinearity" in photovoltaic capacity optimization problems, providing efficient and accurate algorithmic support for subsequent solutions to the capacity optimization function, thus facilitating subsequent processing.

[0106] Furthermore, the step of using the particle swarm optimization algorithm after dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient as the target optimization algorithm includes:

[0107] The instantaneous changes in the power balance parameters are collected, and the instantaneous changes are convolved with the real-time attenuation value of the first correlation coefficient to generate the corresponding dynamic correction coefficient.

[0108] Based on the ratio of the real-time monitored value to the rated value of the equipment capacity parameter, the corresponding particle velocity adjustment threshold is determined, and the fluctuation frequency of the voltage limit parameter is extracted. The fluctuation frequency is then weighted and summed with the third correlation coefficient to obtain the particle fitness penalty coefficient.

[0109] The performance of the particle swarm optimization algorithm, after dynamic correction coefficient calibration, velocity threshold adjustment, and weighted penalty coefficient configuration, is verified to generate the target optimization algorithm accordingly.

[0110] It should be noted that power balance parameters may experience instantaneous changes due to factors such as sudden load changes and fluctuations in photovoltaic output (e.g., a sudden 10% increase in load at a certain moment causes a momentary rise in power balance parameters). If the inertial weights decay at a fixed rate, they may not be able to respond to such changes in a timely manner (e.g., when power fluctuations are large, a larger weight needs to be maintained to explore a better solution). Therefore, it is necessary to collect the instantaneous changes in power balance parameters and perform a convolution operation with the real-time decay value of the first correlation coefficient (the convolution operation can reflect the superimposed effect of the two), generating a dynamic correction coefficient: when the instantaneous change is large, the correction coefficient increases, which can slow down the decay rate of the inertial weights (e.g., the original decay was 0.008 each time, and after correction it is 0.005), allowing the algorithm to maintain a stronger global search capability; when the instantaneous change is small, the correction coefficient decreases, the weight decays normally, and the algorithm converges normally. Through the dynamic correction coefficient, the adjustment of the inertial weights not only depends on the number of iterations, but also adapts to the real-time changes in power balance, improving the dynamic adaptability of the algorithm.

[0111] Based on this, the particle velocity adjustment threshold needs to be determined according to the real-time status of the equipment capacity parameters: calculate the ratio of the real-time monitored value of the equipment capacity parameters to the rated value (e.g., real-time capacity of the energy storage battery / rated capacity = 0.8). If the ratio is greater than the preset threshold (e.g., 0.9), it indicates that the equipment is close to full capacity operation, and the velocity adjustment threshold needs to be lowered (to prevent particles from searching for solutions exceeding the equipment capacity); if the ratio is less than the preset threshold (e.g., 0.5), it indicates that the equipment has a large adjustment range, and the velocity threshold can be appropriately increased (to expand the search range). Simultaneously, the fluctuation frequency of the voltage limit parameters (e.g., the number of voltage fluctuations within 1 minute) reflects voltage stability. The higher the fluctuation frequency, the more unstable the voltage. A larger particle fitness penalty coefficient needs to be obtained through weighted summation (multiplying the fluctuation frequency by the third correlation coefficient and then summing the results) to more severely penalize solutions that violate voltage constraints. Through this adjustment, the algorithm's velocity control and penalty mechanism can adapt to the real-time operating status of the equipment and voltage, further improving the feasibility of the solution. Finally, the reliability of the algorithm was verified through multi-scenario testing. Specifically, the test scenarios included typical working conditions such as sunny days, cloudy days, and peak loads. The verification indicators included convergence speed (number of iterations), solution accuracy (deviation from the theoretical optimal solution), and constraint satisfaction rate. The algorithm that passed the verification was the final target optimization algorithm, which was then used for subsequent processing.

[0112] Furthermore, the step of solving the capacity optimization function using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time includes:

[0113] The predicted power output value is divided into several sub-intervals according to the linear distribution characteristics, and the local optimization calculation of the target optimization algorithm is performed on each of the sub-intervals.

[0114] The nonlinear feature points of the second correlation coefficient are extracted as key verification nodes in the optimization process, and when the target optimization algorithm iterates to the key verification node, the candidate solutions output by the target optimization algorithm are collected.

[0115] The mapping correlation matrix of the third correlation coefficient is used as the screening template for the global optimal solution, and the candidate solutions output by the target optimization algorithm are substituted into the mapping correlation matrix for dimensionality adaptation verification, so as to select the photovoltaic capacity optimization value accordingly.

[0116] It should be noted that the algorithm divides the predicted power output into sub-intervals based on their linear distribution characteristics (e.g., 0-50kW, 50-100kW). Local optimization is performed on each sub-interval to avoid ineffective searching in the large solution space. Based on this, nonlinear feature points of the second correlation coefficient (e.g., inflection points of the device capacity-voltage curve) are used as verification nodes. When the algorithm iterates to these nodes, a candidate set is collected to capture the optimal solution at the point of abrupt change in constraint relationship. Global verification is then performed through the mapping correlation matrix of the third correlation coefficient. Specifically, the matrix contains the ideal mapping relationship between voltage limit and power balance. The candidate solutions are substituted into the matrix for dimensionality adaptation verification to ensure that the solution satisfies the constraint correlation in the global scope. Finally, the photovoltaic capacity optimization value that satisfies both local optima and global constraints is selected for subsequent processing.

[0117] Furthermore, the step of substituting the candidate solutions output by the target optimization algorithm into the mapping correlation matrix for dimensionality adaptation verification, so as to select the optimized photovoltaic capacity value, includes:

[0118] The candidate solution is decomposed according to the three dimensions of the power balance parameter, the equipment capacity parameter and the voltage limit parameter to obtain a three-dimensional component group that fits the mapping correlation matrix.

[0119] Calculate the cosine similarity between the three-dimensional component group and the corresponding reference vector in the mapping association matrix, and select candidate solutions with a cosine similarity greater than a preset threshold as preliminary fitting solutions;

[0120] The preliminary fit solution is calculated using the capacity optimization function, and the candidate solution whose calculation result satisfies the global optimal condition is selected as the photovoltaic capacity optimization value.

[0121] It should be noted that the candidate solutions are structurally decomposed according to the three-dimensional constraint parameters, into power balance components, equipment capacity components, and voltage limit components, ensuring that each component corresponds one-to-one with the dimension of the mapping correlation matrix. Based on this, the cosine similarity between the three-dimensional component group and the matrix reference vector is calculated (the closer the value is to 1, the higher the matching degree). Candidate solutions with similarity higher than a preset threshold (0.85) are selected as preliminary adapted solutions. The preliminary adapted solutions are substituted into the optimization function to calculate the target value (such as energy utilization rate). The solution with the optimal target value and that satisfies all constraints is selected as the final photovoltaic capacity optimization value. This ensures that the result satisfies the three-dimensional constraint correlation while achieving the global optimality of system performance, thereby enabling objective and dynamic adjustment of the photovoltaic micro-power source capacity and improving optimization efficiency.

[0122] Please see Figure 2 The third embodiment of the present invention provides:

[0123] A data center micro power capacity optimization system, wherein the system comprises:

[0124] The acquisition module is used to acquire the real-time operating parameters of the photovoltaic microgrid and construct the corresponding three-dimensional constraint parameter set in real time based on the real-time operating parameters. The three-dimensional constraint parameter set includes power balance parameters, equipment capacity parameters and voltage limit parameters.

[0125] The processing module is used to obtain a first correlation coefficient between the power balance parameter and the equipment capacity parameter through linear correlation, a second correlation coefficient between the equipment capacity parameter and the voltage limit parameter through nonlinear correlation, and a third correlation coefficient between the voltage limit parameter and the power balance parameter through mapping correlation.

[0126] The module is used to obtain the predicted power output value of the photovoltaic microgrid and, in combination with the first correlation coefficient, the second correlation coefficient and the third correlation coefficient, construct the corresponding capacity optimization function in real time.

[0127] The optimization module is used to optimize the preset particle swarm optimization algorithm using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate a corresponding target optimization algorithm, and to solve the capacity optimization function using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time.

[0128] Furthermore, the building module is specifically used for:

[0129] The power output data and corresponding meteorological observation data generated by the photovoltaic microgrid within a preset time period are obtained, and the meteorological observation data are decomposed into three feature dimensions: light intensity, ambient temperature and wind speed.

[0130] The power output data is trained by nonlinear mapping of the three feature dimensions to the power output data using the random forest algorithm to generate a corresponding power prediction model. The power prediction model is then used to output the predicted power output value that is compatible with the photovoltaic microgrid in real time.

[0131] The first correlation coefficient, the second correlation coefficient, and the third correlation coefficient are subjected to matrix orthogonalization to generate mutually independent parameter influence factors, and the capacity optimization function is constructed based on the parameter influence factors and the predicted power output value.

[0132] Furthermore, the building module is specifically used for:

[0133] The parameter influence factors are constructed into an influence factor matrix according to the time series, and the power output prediction value is Fourier decomposed to obtain the fundamental component and harmonic component. The mapping relationship between the influence factor matrix and the fundamental component and harmonic component is constructed in real time.

[0134] Based on the mapping relationship, the parameter sensitivity coefficients for each time period are calculated in real time, and the short-circuit capacity parameters of the photovoltaic microgrid are obtained simultaneously, so as to construct the corresponding multi-dimensional constraint boundary in real time according to the parameter sensitivity coefficients and the short-circuit capacity parameters.

[0135] The influence factor matrix is ​​convolved with the multi-dimensional constraint boundary to generate a corresponding dynamic constraint vector. The capacity optimization function is then constructed in real time using the dynamic constraint vector as the boundary condition and maximizing the energy utilization rate of the photovoltaic microgrid as the objective.

[0136] Furthermore, the optimization module is specifically used for:

[0137] The square of the first correlation coefficient is used as the reference value of the inertia weight in the preset particle swarm algorithm, and it is linearly decayed according to the number of iterations, with the decay rate being the ratio of the reference value to the total number of iterations.

[0138] Calculate the absolute value of the product of the second correlation coefficient and the voltage limit parameter, and use the absolute value of the product as the upper limit threshold of the particle velocity, and construct the penalty coefficient of particle fitness with the third correlation coefficient as the core.

[0139] The particle swarm optimization algorithm, after undergoing dynamic weight decay, adaptive velocity threshold limitation, and directional configuration of penalty coefficient, is used as the target optimization algorithm.

[0140] Furthermore, the optimization module is specifically used for:

[0141] The instantaneous changes in the power balance parameters are collected, and the instantaneous changes are convolved with the real-time attenuation value of the first correlation coefficient to generate the corresponding dynamic correction coefficient.

[0142] Based on the ratio of the real-time monitored value to the rated value of the equipment capacity parameter, the corresponding particle velocity adjustment threshold is determined, and the fluctuation frequency of the voltage limit parameter is extracted. The fluctuation frequency is then weighted and summed with the third correlation coefficient to obtain the particle fitness penalty coefficient.

[0143] The performance of the particle swarm optimization algorithm, after dynamic correction coefficient calibration, velocity threshold adjustment, and weighted penalty coefficient configuration, is verified to generate the target optimization algorithm accordingly.

[0144] Furthermore, the optimization module is specifically used for:

[0145] The predicted power output value is divided into several sub-intervals according to the linear distribution characteristics, and the local optimization calculation of the target optimization algorithm is performed on each of the sub-intervals.

[0146] The nonlinear feature points of the second correlation coefficient are extracted as key verification nodes in the optimization process, and when the target optimization algorithm iterates to the key verification node, the candidate solutions output by the target optimization algorithm are collected.

[0147] The mapping correlation matrix of the third correlation coefficient is used as the screening template for the global optimal solution, and the candidate solutions output by the target optimization algorithm are substituted into the mapping correlation matrix for dimensionality adaptation verification, so as to select the photovoltaic capacity optimization value accordingly.

[0148] Furthermore, the optimization module is specifically used for:

[0149] The candidate solution is decomposed according to the three dimensions of the power balance parameter, the equipment capacity parameter and the voltage limit parameter to obtain a three-dimensional component group that fits the mapping correlation matrix.

[0150] Calculate the cosine similarity between the three-dimensional component group and the corresponding reference vector in the mapping association matrix, and select candidate solutions with a cosine similarity greater than a preset threshold as preliminary fitting solutions;

[0151] The preliminary fit solution is calculated using the capacity optimization function, and the candidate solution whose calculation result satisfies the global optimal condition is selected as the photovoltaic capacity optimization value.

[0152] The fourth embodiment of the present invention provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data center micro power capacity optimization method as described above.

[0153] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the data center micro power capacity optimization method as described above.

[0154] In summary, the data center micro-power capacity optimization method and system provided by the above embodiments of the present invention can objectively and dynamically optimize the capacity of photovoltaic microgrids, thereby improving the optimization efficiency.

[0155] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0156] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0157] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0158] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0159] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0160] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A method for data center micro power supply capacity optimization, characterized in that, The method comprises: acquiring real-time working parameters of a photovoltaic micro-grid, and constructing a corresponding three-dimensional constraint parameter group in real time according to the real-time working parameters, the three-dimensional constraint parameter group comprising a power balance parameter, a device capacity parameter and a voltage limit parameter; obtaining a first correlation coefficient between the power balance parameter and the device capacity parameter through linear correlation, a second correlation coefficient between the device capacity parameter and the voltage limit parameter through nonlinear correlation, and a third correlation coefficient between the voltage limit parameter and the power balance parameter through mapping correlation; acquiring a power output prediction value of the photovoltaic micro-grid, and constructing a corresponding capacity optimization function in real time in combination with the first correlation coefficient, the second correlation coefficient and the third correlation coefficient; optimizing a preset particle swarm algorithm through the first correlation coefficient, the second correlation coefficient and the third correlation coefficient to generate a corresponding target optimization algorithm, and solving the capacity optimization function through the target optimization algorithm to output a corresponding photovoltaic capacity optimization value in real time; the step of optimizing the preset particle swarm algorithm through the first correlation coefficient, the second correlation coefficient and the third correlation coefficient to generate the target optimization algorithm comprises: taking a square value of the first correlation coefficient as a benchmark value of an inertia weight in the preset particle swarm algorithm, and linearly decaying the benchmark value according to the number of iterations, the decay rate being a ratio of the benchmark value to the total number of iterations; calculating a product absolute value of the second correlation coefficient and the voltage limit parameter, and taking the product absolute value as an upper threshold value of particle speed, and constructing a penalty coefficient of particle fitness with the third correlation coefficient as the core; taking the particle swarm algorithm after dynamic weight decay, adaptive speed threshold limitation and directional configuration of the penalty coefficient as the target optimization algorithm; the step of taking the particle swarm algorithm after dynamic weight decay, adaptive speed threshold limitation and directional configuration of the penalty coefficient as the target optimization algorithm comprises: collecting an instantaneous change amount of the power balance parameter, and performing convolution operation on the instantaneous change amount and a real-time decay value of the first correlation coefficient to generate a corresponding dynamic correction coefficient; determining a particle speed adjustment threshold value according to a ratio of a real-time monitoring value of the device capacity parameter to a rated value, extracting a fluctuation frequency of the voltage limit parameter, and performing weighted summation on the fluctuation frequency and the third correlation coefficient to obtain a particle fitness penalty coefficient; verifying the performance of the particle swarm algorithm after dynamic correction coefficient calibration, speed threshold adjustment and weighted penalty coefficient configuration to correspondingly generate the target optimization algorithm.

2. The data center micro power supply capacity optimization method of claim 1, wherein, the step of acquiring the power output prediction value of the photovoltaic micro-grid and constructing the corresponding capacity optimization function in real time in combination with the first correlation coefficient, the second correlation coefficient and the third correlation coefficient comprises: acquiring power output data generated by the photovoltaic micro-grid within a preset time period and corresponding meteorological observation data, and correspondingly decomposing the meteorological observation data into three characteristic dimensions of light intensity, environmental temperature and wind speed; The three feature dimensions and the power output data are trained by a random forest algorithm for non-linear mapping to generate a corresponding power prediction model, and the power prediction model is used to output a power output prediction value adapted to the photovoltaic micro-grid in real time; The first correlation coefficient, the second correlation coefficient, and the third correlation coefficient are subjected to matrix orthogonalization processing to generate independent parameter influence factors, and the capacity optimization function is constructed according to the parameter influence factors and the power output prediction value.

3. The data center micro power supply capacity optimization method of claim 2, wherein, The step of constructing the capacity optimization function according to the parameter influence factors and the power output prediction value comprises: The parameter influence factors are constructed into an influence factor matrix in time sequence, and the power output prediction value is subjected to Fourier decomposition to obtain a fundamental component and a harmonic component, and a mapping relationship between the influence factor matrix and the fundamental component and the harmonic component is constructed in real time; Based on the mapping relationship, the parameter sensitivity coefficients of each time period are calculated in real time, and the short-circuit capacity parameters of the photovoltaic micro-grid are synchronously obtained, so as to construct a corresponding multi-dimensional constraint boundary according to the parameter sensitivity coefficients and the short-circuit capacity parameters in real time; The influence factor matrix and the multi-dimensional constraint boundary are subjected to convolution operation to generate a corresponding dynamic constraint vector, and the capacity optimization function is constructed in real time with the dynamic constraint vector as a boundary condition and the maximum energy utilization rate of the photovoltaic micro-grid as a target.

4. The data center micro power supply capacity optimization method of claim 1, wherein, The step of solving the capacity optimization function by the target optimization algorithm to output a corresponding photovoltaic capacity optimization value in real time comprises: The power output prediction value is divided into a plurality of sub-intervals according to linear distribution characteristics, and local optimization calculation of the target optimization algorithm is performed on each sub-interval; Non-linear feature points of the second correlation coefficient are extracted as key check nodes in the optimization process, and candidate solutions output by the target optimization algorithm are collected when the target optimization algorithm iterates to the key check nodes; A mapping correlation matrix of the third correlation coefficient is used as a screening template for a global optimal solution, and the candidate solutions output by the target optimization algorithm are substituted into the mapping correlation matrix for dimension adaptability verification to screen out the photovoltaic capacity optimization value.

5. The data center micro power supply capacity optimization method of claim 4, wherein, The step of substituting the candidate solutions output by the target optimization algorithm into the mapping correlation matrix for dimension adaptability verification to screen out the photovoltaic capacity optimization value comprises: The candidate solutions are disassembled according to three-dimensional dimensions of the power balance parameter, the equipment capacity parameter, and the voltage limit parameter to obtain a three-dimensional component group adapted to the mapping correlation matrix; Cosine similarity of the three-dimensional component group and a corresponding reference vector in the mapping correlation matrix is calculated, and candidate solutions with cosine similarity greater than a preset threshold are selected as preliminary adapted solutions; The preliminary adapted solutions are calculated by the capacity optimization function, and candidate solutions satisfying global optimal conditions in the calculation results are selected as the photovoltaic capacity optimization value.

6. A data center micro power supply capacity optimization system, characterized by, The system for implementing the data center micro power capacity optimization method as described in any one of claims 1 to 5 includes: The acquisition module is used to acquire the real-time operating parameters of the photovoltaic microgrid and construct the corresponding three-dimensional constraint parameter set in real time based on the real-time operating parameters. The three-dimensional constraint parameter set includes power balance parameters, equipment capacity parameters and voltage limit parameters. The processing module is used to obtain a first correlation coefficient between the power balance parameter and the equipment capacity parameter through linear correlation, a second correlation coefficient between the equipment capacity parameter and the voltage limit parameter through nonlinear correlation, and a third correlation coefficient between the voltage limit parameter and the power balance parameter through mapping correlation. The module is used to obtain the predicted power output value of the photovoltaic microgrid and, in combination with the first correlation coefficient, the second correlation coefficient and the third correlation coefficient, construct the corresponding capacity optimization function in real time. The optimization module is used to optimize the preset particle swarm optimization algorithm using the first correlation coefficient, the second correlation coefficient, and the third correlation coefficient to generate a corresponding target optimization algorithm, and to solve the capacity optimization function using the target optimization algorithm to output the corresponding photovoltaic capacity optimization value in real time.

7. A computer comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the data center micro power capacity optimization method as described in any one of claims 1 to 5.

8. A readable storage medium, having stored thereon a computer program, characterized in that, When executed by the processor, the program implements the data center micro power capacity optimization method as described in any one of claims 1 to 5.

Images