An efficient computer room optimization design method based on hourly performance simulation

Abstract

The application relates to the technical field of data processing, and specifically provides an efficient machine room optimization design method based on hourly performance simulation, which comprises the following steps: S1, collecting relevant parameter data of machine room optimization design and performing pretreatment; S2, constructing a fitness function, and analyzing the change characteristics of the fitness function value of a hawk individual to construct trend fitness; S3, constructing adaptive fitness based on the trend fitness to analyze the difference degree of the fitness function value of an individual and the fitness function value of a group in the optimization process; S4, constructing an indication factor based on the adaptive fitness to analyze the position updating mode in the iteration process; and S5, improving the hawk optimization algorithm based on the indication factor to improve the optimization precision of the machine room environment; the application can jump out of a local optimal solution, find a potential global optimal solution in a solution space, and finally converge to the global optimal solution, so that the power of the obtained cooling equipment can guarantee long-term stable operation of machine room equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an efficient machine room optimization design method based on hourly performance simulation. BACKGROUND

[0002] In the design of an efficient machine room, the machine room contains multiple key devices such as servers, air conditioners, UPS power systems, monitoring devices, and communication devices, etc. These devices not only support the normal operation of the machine room, but also determine the energy consumption and environmental control efficiency of the machine room. With the increasing demand for computing and data processing tasks, the energy efficiency, reliability, and space utilization of the machine room have become key targets for optimization design. Therefore, optimizing the design of the machine room can not only improve the efficiency of the equipment and reduce energy consumption, but also effectively prolong the service life of the equipment and improve the overall operation efficiency of the machine room, meeting the growing demand for data processing. The optimization design process involves multiple factors, including device arrangement, energy management, cooling system design, etc. Through reasonable planning and scheduling, the comprehensive performance of the machine room can be greatly improved. In the current market, the Eagle Optimization Algorithm is used for machine room design optimization, which has strong global optimization capability, especially for high-latitude and large-scale machine room design problems. The global search stage of the Eagle Optimization Algorithm can quickly explore the solution space, and the local convergence stage can further improve the accuracy of the solution.

[0003] However, the traditional Eagle Optimization Algorithm has two position update methods in the global search stage and the local convergence stage during the optimization process. In each iteration, a random calculation method is selected to update the position. This random mode may not effectively find the global optimal solution when dealing with machine room design optimization problems, resulting in poor optimization effect of the algorithm. SUMMARY

[0004] In view of the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide an efficient machine room optimization design method based on hourly performance simulation, which is used to solve the above technical problems.

[0005] To achieve the above object and other related objects, the present application provides an efficient machine room optimization design method based on hourly performance simulation, which comprises:

[0006] Step S1: Collecting relevant parameter data for machine room optimization design and preprocessing.

[0007] Step S2: Constructing a fitness function and analyzing the change characteristics of the fitness function value of the Eagle individual to construct the trend fitness; the process of constructing the fitness function includes taking a time window with a time length of L, calculating the standard deviation of temperature data and humidity data in the time window respectively, and the fitness function is specifically expressed in the following form:

[0008]

[0009] wherein F represents the fitness function of the grey wolf optimization algorithm, V represents the gas flow rate, P represents the operating power of the cooling device, and respectively represent the standard deviation of the temperature data and the humidity data within the time window, 、 respectively represent the sum of the absolute values of the differences between the temperature data within the time window and the optimal operating temperature of the device and between the humidity data and the optimal operating humidity of the device, wherein the optimal operating temperature and the optimal operating humidity of the device can be obtained from the device manual;

[0010] The above fitness function is used as the fitness function of the grey wolf optimization algorithm, and by analyzing the change of the fitness function value of each grey wolf in the iteration process, the search efficiency and convergence speed of each grey wolf in the solution space are evaluated, that is, the improvement of the fitness function value indicates that the grey wolf is approaching the optimal solution, and the stagnation or decline of the fitness function value indicates that the grey wolf may be trapped in a local optimal solution, and the search strategy needs to be adjusted.

[0011] Step S3: based on the trend fitness, the difference degree of the individual fitness function value and the group fitness function value in the optimization process is analyzed to construct fitness.

[0012] Step S4: based on the fitness, the position updating mode in the iteration process is analyzed to construct an indicator factor.

[0013] Step S5: based on the indicator factor, the grey wolf optimization algorithm is improved to improve the optimization accuracy of the computer room environment.

[0014] In an embodiment of the present application, in step S1, the collection of the related parameter data of the computer room optimization design and the preprocessing thereof include:

[0015] Temperature sensors, humidity sensors, gas flow rate sensors and power sensors are arranged in the computer room; during the normal operation of the devices in the computer room, the temperature data and the humidity data in the computer room are collected by the temperature sensors and the humidity sensors, the gas flow rate data when the cooling device is cooling are collected by the gas flow rate sensors, and the power data when the cooling device is operating are collected by the power sensors;

[0016] The linear interpolation method is used to fill in the missing values of the temperature data, the humidity data, the gas flow rate data and the power data; at the same time, the maximum and minimum normalization method is used to normalize the filled data to obtain the related parameter data of the computer room optimization design after preprocessing.

[0017] In an embodiment of the present application, in step S2, the process of analyzing the change characteristics of the individual fitness function value of the eagle to construct the trend degree of goodness comprises: each eagle obtains the fitness function value at this iteration to reflect the cooling effect on the equipment room equipment at each iteration; taking the ith eagle at the jth iteration as an example, the sequence formed by the fitness function values from the 1st to the jth iteration is denoted as a fitness sequence, the fitness sequence is subjected to first-order difference processing to obtain a difference fitness sequence; the fitness sequence is taken as the input of the Mann-Kendall trend test algorithm, the hypothesis test condition is that the fitness sequence has an upward trend, and the output of the algorithm is the p-value of the fitness sequence having an upward trend, and the p-value size reflects the reliability of the hypothesis test condition being established.

[0018] In an embodiment of the present application, the calculation method of the trend degree of goodness is as follows:

[0019]

[0020]

[0021] wherein represents the upward factor of the fitness function value of the ith eagle at the iteration, , respectively represent the sum of positive items and the sum of absolute values of negative items in the difference fitness sequence, and 1 represents a hyperparameter for avoiding the influence of zero on the calculation of the numerator or the denominator; represents the trend degree of goodness of the ith eagle at the jth iteration, represents the p-value of the Mann-Kendall trend test corresponding to the fitness sequence corresponding to the ith eagle at the jth iteration, and exp() represents an exponential function with a natural constant as the base, which is used to avoid the denominator being zero to cause the calculation to be impossible, represents the maximum value of the upward factors corresponding to all eagles at the jth iteration;

[0022] In the process of using the eagle optimization algorithm to iteratively optimize the equipment room environment, if the upward trend of the fitness function value of the ith eagle relative to the group is more obvious and the upward degree is larger, it indicates that the eagle is flying in the direction of the optimal solution of the fitness function, that is, the change direction of the cooling equipment power is closer to the best working environment of the equipment room equipment at this time, and therefore the calculated trend degree of goodness is larger.

[0023] In an embodiment of the present application, in step S3, the difference between the individual fitness function value and the population fitness function value in the trend fitness analysis optimization process is used to construct the fitness degree, which includes: when the raven optimization algorithm is used to optimize the computer room environment, the change trend of the fitness function reflects the adjustment of the computer room environment optimization strategy, the more obvious the increase in the fitness function value, the more suitable the working environment and temperature control state of the computer room equipment, and the more stable the state of the equipment during work; therefore, during the optimization design process of the computer room environment by the raven optimization algorithm, when the number of iterations is small, the raven individual should have a high degree of freedom to explore in the solution space to find the optimal solution, and as the number of iterations increases, the optimization result should gradually converge to the optimal solution; meanwhile, the raven should adjust the optimization direction according to the population optimal solution during the optimization process, so as to find the global optimal solution.

[0024] In an embodiment of the present application, the fitness degree is calculated as follows:

[0025]

[0026] Wherein represents the fitness degree of the i-th raven at the j-th iteration, represents the fitness function value of the i-th raven at the j-th iteration, represents the maximum value of the fitness function value of the i-th raven in the first j iterations, represents the maximum value of the fitness function value of all ravens in the first j iterations, represents the trend fitness of the i-th raven at the j-th iteration, and M represents the maximum number of iterations; during the iteration optimization process of the raven individual, as the number of iterations gradually approaches the maximum number of iterations, the closer the optimal solution of the raven individual to the population optimal solution, the better the optimization direction of the raven individual, that is, the more suitable the power of the cooling equipment as the optimal cooling power of the computer room equipment during work, and therefore the more the iteration position should be updated based on the historical optimal iteration, and the greater the calculated fitness degree; if the optimal solution of the raven individual is greatly different from the population optimal solution, it indicates that the raven individual may fall into a local optimal solution during iteration, and the optimization effect of the raven individual is poor at this time, that is, the power of the cooling equipment is not suitable as the optimal cooling power of the computer room equipment during work, that is, the power of the cooling equipment is randomly disturbed to jump out of the local optimal solution, and then the power of the cooling equipment is updated for the next time, and therefore the smaller the calculated fitness degree.

[0027] In an embodiment of the present application, in step S4, the method for constructing the indication factor based on the position update in the iteration process of the fitness analysis comprises: in the process of optimizing the machine room environment using the algorithm, if the fitness function value of the hawk in the iteration process is greatly different from the optimal fitness function value of the group, it indicates that the hawk may fall into a local optimal solution in the current position iteration, at this time, random disturbance needs to be added to the current position to jump out of the local optimal solution, expand the optimization space, and facilitate finding the potential global optimal solution; if the fitness function value is close to the optimal fitness function value of the group, it indicates that the current position of the hawk is already optimal, and the next position should be updated based on the historical position information to avoid losing the global optimal solution.

[0028] In an embodiment of the present application, the calculation method of the indication factor based on the fitness is as follows:

[0029]

[0030] Wherein C represents the indication factor, the value of 1 indicates that the first position update method is selected, that is, the next position is updated based on the historical iteration data, the value of -1 indicates that the second position update method is selected, that is, the next position is updated by adding random disturbance, sgn() represents the sign function, the value is 1 when the data in the bracket is non-negative, and the value is -1 when the data in the bracket is negative, represents the fitness of the ith hawk in the jth iteration, represents the mean value of the fitness of the N hawks in the jth iteration; in the iteration process, if the difference between the ith hawk and the optimal solution of the group is too large, it indicates that the hawk should increase random disturbance in the subsequent iteration process, and expand the search range as much as possible, so as to facilitate finding the potential optimal solution and improve the cooling effect of the cooling equipment on the machine room equipment, therefore it is not suitable to continue iteration at the current position, and it is necessary to avoid falling into a local optimal solution, that is, the second position update method is selected to find the optimal solution, therefore the calculated indication factor is -1; if the ith hawk is close to the optimal solution of the group, it indicates that the hawk should converge around the optimal solution in the subsequent iteration process, so that the power of the cooling equipment after convergence can work in the best environment, at this time, the first position update method is selected, therefore the calculated indication factor is 1.

[0031] In an embodiment of the present application, in step S5, the method for improving the algorithm of the hawk optimization based on the indication factor and improving the optimization accuracy of the machine room environment comprises: in the process of optimizing the machine room environment by the cooling equipment, the fitness function is used as the fitness function of the algorithm of the hawk optimization, in the iteration process of the algorithm, the indication factor obtained by steps S1-S4 is used to direct the selection of the position update method of the hawk, the output of the algorithm is the optimal power of the cooling equipment at the current time, and the optimal power of the cooling equipment is used to optimize the machine room environment.

[0032] As described above, the high-efficiency machine room optimization design method based on hourly performance simulation provided by the application has the following beneficial effects:

[0033] When the application optimizes the machine room environment by using the eagle optimization algorithm, the fitness function is constructed, and then the change characteristics of the fitness function value of the eagle individual in the iteration process are analyzed to construct the trend fitness, which can evaluate the advantages and disadvantages of the fitness trend of the eagle individual in the iteration; the fitness is constructed by analyzing the difference between the fitness function value of the eagle individual and the group fitness function value based on the trend fitness, which can evaluate the advantages and disadvantages of the fitness of the eagle individual relative to the group fitness; in the iteration process, the difference between the fitness of the eagle individual and the average fitness of the group is compared based on the fitness, and the indicator is constructed, and the appropriate position updating mode is selected based on the indicator, so that in the iterative optimization process of the machine room environment, the eagle can jump out of the local optimal solution, find the potential global optimal solution in the solution space, and finally converge to the global optimal solution, so that the power of the cooling equipment can guarantee the long-term stable operation of the machine room equipment. DETAILED DESCRIPTION

[0034] The embodiments of the application are described below by specific examples, and those skilled in the art can easily understand other advantages and effects of the application from the content disclosed in the specification.

[0035] The application provides a high-efficiency machine room optimization design method based on hourly performance simulation, which comprises the following steps:

[0036] Step S1: collecting and preprocessing the related parameter data of machine room optimization design; specifically, temperature sensors, humidity sensors, gas flow rate sensors and power sensors are arranged in the machine room; during the normal operation of the equipment in the machine room, the temperature data and humidity data in the machine room are collected by the temperature sensors and humidity sensors, the gas flow rate when the cooling equipment is cooling is collected by the gas flow rate sensor, and the power data when the cooling equipment is running is collected by the power sensor; the data collection interval is 1 second in this application, and the implementer can select according to the situation. During the collection of the above data, missing values and other abnormal situations may occur due to environmental interference, sensor errors and other factors, in order to avoid the influence of missing values on the subsequent analysis process, the application uses linear interpolation method to fill the missing values; at the same time, in order to avoid the calculation results caused by different dimensions, the application uses the maximum and minimum normalization method to normalize the filled data; thus, the preprocessed related parameter data of machine room optimization design are obtained.

[0037] In this step, the linear interpolation method is a mathematical interpolation technique based on a two-point linear equation, suitable for cases where the function changes approximately linearly in a certain interval. The core idea is to construct a straight line equation by connecting two known data points (x0, y0) and (x1, y1), and then estimate the y value corresponding to any x in the interval. The maximum and minimum normalization is a linear transformation method that maps the data feature values to a fixed interval (such as [0, 1]) to standardize the data. This method is a common data preprocessing technique used to linearly convert data feature values to a specified range (usually [0, 1]); the normalized data retains the distribution pattern of the original data, but the numerical range is unified; the core purpose of using this method is to eliminate the influence of data dimension, make different features comparable, and improve the performance of machine learning algorithms. The linear interpolation method and the maximum and minimum normalization method are well-known techniques, and will not be described in detail here.

[0038] Step S2: Construct fitness function and analyze the change characteristics of the fitness function value of the eagle individual to construct trend fitness; specifically, when using the eagle optimization algorithm to optimize the computer room, the fitness function needs to be preset, and the process of constructing the fitness function in this application is as follows:

[0039] Take a time window of length L, and calculate the standard deviation of temperature data and humidity data in the time window, where L is 60 seconds in this application. The fitness function can be specifically represented as follows:

[0040]

[0041] Where F represents the fitness function of the eagle optimization algorithm, V represents the gas flow rate, P represents the operating power of the cooling equipment, and represent the standard deviation of temperature data and humidity data in the time window, respectively, , represent the sum of absolute values of the difference between temperature data and the best working temperature of the equipment, and the difference between humidity and the best working humidity of the equipment, respectively, where the best working temperature and the best working humidity of the equipment can be obtained from the equipment manual.

[0042] In the optimization design of the machine room, the environment temperature and humidity of the equipment in the working state should be kept stable as much as possible, and close to the optimal working state, which indicates that the optimization effect of the cooling equipment on the machine room environment is better, and thus the fitness function value calculated is higher. The above fitness function is used as the fitness function of the algorithm, and the preset parameters and values in the algorithm are as follows: the value of the number of eagles N is 50, and the value of the maximum iteration number M is 300. The above parameter values can be selected according to the actual situation. When the optimization algorithm is used to optimize the machine room environment, the search efficiency and convergence speed of each eagle in the solution space can be evaluated by analyzing the change of the fitness function value of each eagle in the iteration process, that is, the improvement of the fitness function value indicates that the eagle is approaching the optimal solution, and the stagnation or decline of the fitness function value indicates that the eagle may fall into a local optimal solution, and the search strategy needs to be adjusted.

[0043] Based on the above analysis, the trend fitness is constructed to reflect the change characteristics of the fitness function of the eagle in the iteration process, and the construction process of the trend fitness is as follows:

[0044] Each eagle will get the fitness function value at each iteration, which reflects the cooling effect on the machine room equipment. Taking the ith eagle at the jth iteration as an example, when the sequence composed of the fitness function values from the 1st iteration to the jth iteration is denoted as a fitness sequence, the fitness sequence is subjected to first-order difference processing to obtain a difference fitness sequence, the fitness sequence is used as the input of the Mann-Kendall trend test algorithm, the hypothesis test condition is that the fitness sequence has an upward trend, and the output of the algorithm is the p value of the existence of the upward trend of the fitness sequence, and the size of the p value reflects the reliability of the hypothesis test condition. The Mann-Kendall trend test algorithm is a commonly used non-parametric test method for testing whether the time series data has a trend. Its principle is to compare the size relationship between each point and its subsequent point in the data to determine whether there is a monotone increasing or decreasing trend in the data. The test process of the Mann-Kendall trend test algorithm is a known technology, which will not be described here.

[0045] Based on the above processing steps, the calculation method of the trend fitness in the application is as follows:

[0046]

[0047]

[0048] wherein indicates the upward factor of the fitness function value of the ith eagle at the iteration, ​respectively represent the sum of positive items and the sum of absolute values of negative items in the difference fitness sequence, 1 represents a hyperparameter, used to avoid the influence of zero numerator or denominator on calculation, and the implementer can select it according to the situation.

[0049] represents the trend fitness of the i-th eagle at the j-th iteration, represents the p value of the Mann-Kendall trend test corresponding to the fitness sequence corresponding to the i-th eagle at the j-th iteration, exp() represents an exponential function with a natural constant as the base, used to avoid the denominator being zero and causing calculation failure, represents the maximum value of the rising factor of all eagles at the j-th iteration. In particular, since there is no historical data at the 1st iteration, the preset values of the rising factor and the trend fitness are 1, facilitating subsequent calculation. In the process of using the eagle optimization algorithm to iteratively optimize the computer room environment, if the fitness function value of the i-th eagle has a more obvious rising trend and a greater rising degree relative to the group, it indicates that the eagle is flying towards the optimal solution of the fitness function, that is, the change direction of the cooling equipment power is closer to the best working environment of the computer room equipment, and therefore the calculated trend fitness is greater.

[0050] Step S3: constructing fitness degree based on the difference degree between the individual fitness function value and the group fitness function value in the optimization process according to the trend fitness; specifically, when the eagle optimization algorithm is used to optimize the computer room environment, the change trend of the fitness function reflects the adjustment of the computer room environment optimization strategy, and the more obvious the increase in the fitness function value, the more suitable the working environment and temperature control state of the computer room equipment, and the more stable the state of the equipment during work. Therefore, in the process of optimizing the computer room environment by using the eagle optimization algorithm, when the number of iterations is small, the eagle individual should have a high degree of freedom to explore the solution space as much as possible to find the optimal solution, and as the number of iterations increases, the optimization result should gradually converge to the optimal solution; meanwhile, the eagle should adjust the optimization direction according to the group optimal solution during the optimization process, so as to find the global optimal solution.

[0051] Based on the above analysis, the calculation method of the fitness degree in the present application is as follows:

[0052]

[0053] wherein represents the fitness degree of the i-th eagle at the j-th iteration, represents the fitness function value of the i-th eagle at the j-th iteration, represents the maximum value of the fitness function value of the i-th eagle in the first j iterations, represents the maximum value of the fitness function value of all eagles in the first j iterations, represents the trend fitness of the ith eagle at the jth iteration, M represents the maximum number of iterations, and in this application, the value is 300, and the implementer can select according to the actual situation.

[0054] In the iteration optimization process of the eagle individual, when the number of iterations gradually approaches the maximum number of iterations, the closer the optimal solution of the eagle individual to the group optimal solution, the better the optimization direction of the eagle individual, that is, the more suitable the power of the cooling device as the best cooling power of the computer room equipment in operation, and therefore the more the iteration position should be updated based on the historical optimal iteration condition, and the greater the calculated fitness; if the optimal solution of the eagle individual and the group optimal solution are quite different, it indicates that the eagle individual may fall into a local optimal solution during iteration, and the optimization effect of the eagle individual is poor, that is, the power of the cooling device is not suitable as the best cooling power of the computer room equipment in operation, that is, the power of the cooling device is increased with random disturbance to jump out of the local optimal solution, and then the power of the cooling device in the next iteration is updated, and therefore the smaller the calculated fitness.

[0055] Step S4: constructing an indication factor based on the position update mode in the iteration process based on the fitness; specifically, in the process of using the eagle optimization algorithm to optimize the computer room environment, if the fitness function value obtained by the eagle in the iteration process and the group optimal fitness function value are quite different, it indicates that the eagle may fall into a local optimal solution in the current position iteration, and at this time, random disturbance needs to be added to the current position to jump out of the local optimal solution and expand the optimization space, so as to find the potential global optimal solution; if the fitness function value and the group optimal fitness function value are relatively close, it indicates that the current position of the eagle is relatively optimal, and the next position should be updated based on the historical position information to avoid losing the global optimal solution.

[0056] Based on the above analysis, the indication factor is constructed based on the fitness in this application, and the calculation method is as follows:

[0057]

[0058] Wherein C represents the indication factor, the value of 1 represents selecting the first position update mode, that is, updating the next position based on the historical iteration data, the value of -1 represents selecting the second position update mode, that is, updating the next position by increasing random disturbance, sgn() represents the sign function, the value of 1 when the data in the bracket is non-negative, and the value of -1 when the data in the bracket is negative, represents the fitness of the ith eagle at the jth iteration, represents the mean value of the fitness of the N eagles at the jth iteration.

[0059] In the iteration process, if the i-th eagle is too different from the optimal solution of the group, it indicates that the eagle should increase the random disturbance in the subsequent iteration process to expand the search range as much as possible to find the potential optimal solution and improve the cooling effect of the cooling equipment on the equipment room. Therefore, it is not suitable to continue iteration at the current position to avoid falling into a local optimal solution. That is, the second position updating method is needed to find the optimal solution at this time, so the calculated indicator factor is -1. If the i-th eagle is close to the optimal solution of the group, it indicates that the eagle should converge around the optimal solution in the subsequent iteration process, so that the power of the cooling equipment obtained after convergence can work in the best environment. At this time, the first position updating method is needed, so the calculated indicator factor is 1.

[0060] Step S5: improving the eagle optimization algorithm based on the indicator factor to improve the optimization accuracy of the equipment room environment; specifically, when optimizing the equipment room environment by the cooling equipment, the fitness function is used as the fitness function of the eagle optimization algorithm. In the iteration process of the algorithm, the position updating method of the eagle is selected by the indicator factor obtained in the above steps. The output of the algorithm is the optimal cooling equipment power at the current time. The optimal power of the cooling equipment is used to optimize the equipment room environment, so that the obtained cooling equipment power can better meet the optimization needs of the current equipment room, improve the adaptability and optimization accuracy of the algorithm, avoid the optimization result deviating from the optimal solution caused by random selection of the position updating method, ensure that the temperature and humidity environmental parameters are in the ideal range with small fluctuations, the cooling system runs smoothly and has the best energy efficiency, and the equipment in the equipment room can work in the best environment to ensure the long-term stable operation of the equipment. The iteration process of the eagle optimization algorithm is based on the mathematical model of its hunting strategy. The optimization is realized by simulating the hunting behavior of eagles at different stages. The core iteration steps of the algorithm include: 1) initialization: randomly generate an initial population, the position of the eagle represents a potential solution in the solution space, and the fitness of each individual is calculated; 2) extended exploration stage: widely explore the search space by vertical diving high-altitude soaring, use Tent chaotic mapping to generate a diversified initial population to enhance the global search ability; 3) convergence optimization stage: adjust the moving speed by using the inertia weight strategy: accelerate the moving speed in the early stage to quickly approach the target area, and reduce the moving speed in the later stage to finely adjust the position to avoid falling into a local optimum; 4) position updating rule: according to the relative position of the eagle and the prey, i.e. the current solution and the optimal solution, the flight trajectory of the eagle is updated by a mathematical model to simulate behaviors such as diving and gliding; 5) termination condition: the algorithm is terminated when the preset iteration number is reached or the fitness meets a certain threshold. The algorithm balances global search and local optimization through group cooperation and adaptive mechanism, and is suitable for multi-objective optimization problems. The iteration process of the eagle optimization algorithm is a known technology, which will not be described in detail here.

[0061] To sum up, when the machine room environment is optimized by the application of the application, the fitness function is constructed, and then the change characteristics of the fitness function value of the hawk individual in the iteration process are analyzed to construct the trend fitness, so as to evaluate the advantages and disadvantages of the fitness trend of the hawk individual in the iteration; the difference between the fitness function value of the hawk individual and the group fitness function value is analyzed based on the trend fitness to construct the fitness fitness, so as to evaluate the advantages and disadvantages of the fitness of the hawk individual relative to the group fitness; in the iteration process, the difference between the fitness of the hawk individual and the group average fitness is compared based on the fitness fitness to construct the indicator factor, and the appropriate position updating mode is selected based on the indicator factor, so that in the iterative optimization process of the machine room environment, the hawk can jump out of the local optimal solution, find the potential global optimal solution in the solution space, and finally converge to the global optimal solution, so that the power of the obtained cooling equipment can guarantee the long-term stable operation of the machine room equipment. Therefore, the application effectively overcomes the various shortcomings in the prior art and has high industrial utilization value.

[0062] The above embodiments only exemplarily illustrate the principles and effects of the application, and are not used to limit the application. Any person skilled in the art can modify or change the above embodiments without departing from the spirit and scope of the application. Therefore, all equivalent modifications or changes completed by those skilled in the art without departing from the spirit and technical thought disclosed by the application should be covered by the claims of the application.

Claims

1. A high-efficiency data center optimization design method based on time-by-time performance simulation, characterized in that, The method includes: Step S1: Collect relevant parameter data for the optimized design of the computer room and perform preprocessing; Step S2: Construct a fitness function and analyze the changing characteristics of the fitness function values ​​of individual eagles to construct a trend goodness; Step S3: Construct fitness based on the difference between individual fitness function values ​​and population fitness function values ​​during the optimization process using trend goodness-of-fit analysis. The process of constructing the fitness function includes taking a time window of length L and calculating the standard deviations of temperature and humidity data within the time window. The fitness function is specifically expressed in the following form: Where F represents the fitness function of the Skyhawk optimization algorithm, V represents the gas flow rate, and P represents the operating power of the cooling equipment. and These represent the standard deviations of the temperature and humidity data within the time window, respectively. , These represent the sum of the absolute values ​​of the differences between the temperature data and the optimal operating temperature and humidity of the equipment within the time window, respectively. The optimal operating temperature and humidity of the equipment are obtained from the equipment instruction manual. Using the above fitness function as the fitness function of the Skyhawk optimization algorithm, by analyzing the changes in the fitness function value of each Skyhawk during the iteration process, we can evaluate the search efficiency and convergence speed of each Skyhawk in the solution space. That is, the increase in the fitness function value indicates that the Skyhawk is approaching the optimal solution, while the stagnation or decrease in the fitness function value indicates that the Skyhawk may be trapped in a local optimum and the search strategy needs to be adjusted. Step S4: Construct indicator factors based on the position update method during the fitness analysis iteration process; Step S5: Improve the Tianying optimization algorithm based on indicator factors to enhance the optimization accuracy of the data center environment.

2. The efficient data center optimization design method based on time-by-time performance simulation according to claim 1, characterized in that, In step S1, the collection and preprocessing of relevant parameter data for the optimized design of the data center includes: Temperature sensors, humidity sensors, gas flow rate sensors, and power sensors are installed in the computer room. During normal operation of the equipment in the computer room, temperature and humidity data are collected by the temperature and humidity sensors, gas flow rate data is collected by the gas flow rate sensor when the cooling equipment dissipates heat, and power data is collected by the power sensor when the cooling equipment is running. Linear interpolation was used to fill in missing values ​​in temperature, humidity, gas flow rate, and power data. At the same time, the max-min normalization method was used to normalize the filled data to obtain the relevant parameter data for the preprocessed computer room optimization design.

3. The efficient data center optimization design method based on time-by-time performance simulation according to claim 2, characterized in that, In step S2, the process of analyzing the changing characteristics of the fitness function values ​​of individual eagles to construct trend goodness includes: each eagle obtains its fitness function value at each iteration to reflect the cooling effect on the equipment in the computer room; taking the i-th eagle at the j-th iteration as an example, the sequence of fitness function values ​​from the 1st to the j-th iteration is denoted as the fitness sequence, and the fitness sequence is subjected to first-order differencing to obtain the differencing fitness sequence; the fitness sequence is used as the input of the Mann-Kendall trend test algorithm, the hypothesis test condition is that the fitness sequence has an upward trend, and the output of the algorithm is the p-value of the fitness sequence having an upward trend. The size of the p-value reflects the reliability of the hypothesis test condition.

4. The efficient data center optimization design method based on time-by-time performance simulation according to claim 3, characterized in that, The method for calculating the trend goodness of view is as follows: in This represents the increase factor of the fitness function value for the i-th eagle during iteration. , These represent the sum of positive terms and the sum of absolute values ​​of negative terms in the difference fitness sequence, respectively. 1 represents a hyperparameter used to avoid the numerator or denominator being zero and affecting the calculation. This represents the trend superiority of the i-th eagle in the j-th iteration. Let represent the p-value of the Mann-Kendall trend test corresponding to the fitness sequence of the i-th eagle at the j-th iteration, and exp() denotes an exponential function with the natural constant as the base, used to avoid the denominator being zero and thus impossible to calculate. This represents the maximum value of the increase factor corresponding to all eagles at the j-th iteration; In the process of iteratively optimizing the data center environment using the Skyhawk optimization algorithm, if the fitness function value of the i-th Skyhawk shows a more obvious upward trend and a greater degree of increase relative to the group, it indicates that the Skyhawk is flying towards the optimal solution of the fitness function. That is, the direction of change of the cooling equipment power is closer to the optimal working environment of the data center equipment, and therefore the calculated trend goodness is greater.

5. The efficient data center optimization design method based on time-by-time performance simulation according to claim 4, characterized in that, In step S3, the construction of fitness superiority based on the difference between individual fitness function values ​​and population fitness function values ​​during the optimization process based on trend superiority analysis includes: when using the Tianying optimization algorithm to optimize the data center environment, the changing trend of the fitness function reflects the adjustment of the data center environment optimization strategy. The more significant the increase in fitness function value, the more suitable the working environment and temperature control status of the data center equipment, and the more stable the equipment's working state. Therefore, in the process of optimizing the data center environment using the Tianying optimization algorithm, when the number of iterations is small, the Tianying individual should have a high degree of freedom to explore within the solution space as much as possible to find the optimal solution. As the number of iterations increases, the optimization result should gradually converge to the optimal solution. At the same time, during the optimization process, the Tianying should continuously adjust the optimization direction based on the population optimal solution in order to find the global optimal solution.

6. The efficient data center optimization design method based on time-by-time performance simulation according to claim 5, characterized in that, The method for calculating the fitness score is as follows: in This represents the fitness of the i-th eagle in the j-th iteration. This represents the fitness function value of the i-th eagle in the j-th iteration. This represents the maximum fitness function value of the i-th eagle during the first j iterations. This represents the maximum fitness function value of all Skyhawks in the first j iterations. Let M represent the trend goodness of the i-th eagle in the j-th iteration, and M represent the maximum number of iterations. During the individual eagle's iterative optimization process, as the number of iterations gradually approaches the maximum number of iterations, the closer the individual eagle's optimal solution is to the group's optimal solution, the better the eagle's optimization direction is. That is, the power of the cooling equipment is more suitable as the optimal cooling power for the computer room equipment. Therefore, the next iteration position should be updated based on the historical optimal iteration situation, and the calculated goodness of fitness will be greater. If the individual eagle's optimal solution differs greatly from the group's optimal solution, it indicates that the individual eagle may be trapped in a local optimum during iteration. In this case, the eagle's optimization effect is poor, meaning that the power of the cooling equipment is not suitable as the optimal cooling power for the computer room equipment. In this case, a random perturbation should be added to the power of the cooling equipment to make it jump out of the local optimum, and then the power of the cooling equipment should be updated for the next iteration. Therefore, the calculated goodness of fitness will be smaller.

7. The efficient data center optimization design method based on time-by-time performance simulation according to claim 6, characterized in that, In step S4, the method of constructing the indicator factor based on the position update during the fitness analysis iteration process includes: during the optimization of the data center environment using the Skyhawk optimization algorithm, if the fitness function value obtained by Skyhawk during the iteration process differs significantly from the optimal fitness function value of the population, it indicates that Skyhawk may be trapped in a local optimum when iterating at the current position. In this case, it is necessary to add random perturbation to the current position to escape the local optimum, expand the optimization space, and facilitate the search for potential global optima; if the fitness function value is close to the optimal fitness function value of the population, it indicates that the current position of Skyhawk is already relatively good, and the next position should be updated based on historical position information to avoid losing the global optimum.

8. The efficient data center optimization design method based on time-by-time performance simulation according to claim 7, characterized in that, The calculation method for constructing indicator factors based on fitness is as follows: Where C represents the indicator factor, a value of 1 indicates the selection of the first position update method, i.e., updating the next position based on historical iteration data, and a value of -1 indicates the selection of the second position update method, i.e., updating the next position by adding random perturbation. sgn() represents the sign function, taking a value of 1 when the data within the parentheses is non-negative, and a value of -1 when the data within the parentheses is negative. This represents the fitness of the i-th eagle in the j-th iteration. Let represent the mean fitness of N eagles in the j-th iteration. During the iteration process, if the i-th eagle differs too much from the group's optimal solution, it indicates that in subsequent iterations, the eagle should increase random perturbation to expand the search range as much as possible to find potential optimal solutions and improve the cooling effect of the cooling equipment on the computer room equipment. Therefore, it is not suitable to continue iterating at the current position to avoid getting trapped in a local optimum. In this case, the second position update method needs to be selected to find the optimal solution, and the calculated indicator factor is -1. If the i-th eagle is close to the group's optimal solution, it indicates that in subsequent iterations, the eagle should converge near the optimal solution so that the cooling equipment power obtained after convergence can work in the best environment. In this case, the first position update method needs to be selected, and the calculated indicator factor is 1.

9. The efficient data center optimization design method based on time-by-time performance simulation according to claim 8, characterized in that, In step S5, the improvement of the Tianying optimization algorithm based on the indicator factor to improve the optimization accuracy of the data center environment includes: when optimizing the data center environment through cooling equipment, the fitness function is used as the fitness function of the Tianying optimization algorithm. During the iteration process of the algorithm, the position update method of Tianying is selected directionally through the indicator factors obtained in steps S1-S4. The output of the algorithm is the optimal cooling equipment power at the current moment. The data center environment is optimized based on the obtained optimal cooling equipment power.

Images