Heat dissipation control method for edge intelligent computing servers based on data feature extraction

Abstract

The present disclosure relates to the field of printed circuit technology, and specifically to a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction. The method comprises: obtaining power consumption data and temperature data of power elements in an edge computing server; obtaining a potential temperature rise reliability coefficient of the power elements according to the power consumption data and temperature data of the power elements; obtaining a heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient; and adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control. The embodiment of the present disclosure can solve the technical problem that the heat dissipation of the edge intelligent server mainboard in the prior art uses an improper PID control algorithm to adjust the speed of the cooling fan, which will lead to a decrease in the heat dissipation performance of the entire device and the risk of causing equipment failure, thereby improving the heat dissipation performance of the device and reducing the risk of equipment failure.

Description

Technical Field

[0001] The present disclosure belongs to the field of printed circuit technology, and in particular relates to a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction. Background Art

[0002] Edge computing servers are advanced computing devices deployed near network physical devices or data sources. They integrate multiple core functions such as network connection, computing processing, data storage, and application services into one, forming an efficient and open working platform. This type of server is designed to meet a variety of application scenarios and needs to be able to adapt to various ambient temperature conditions from mild to harsh.

[0003] However, due to their high sensitivity to the environment, edge computing servers may face a higher risk of equipment failure at high temperatures. The heat of edge servers mainly comes from the CPU and other power components on the server motherboard. The server heat dissipation is mainly discharged from the device chassis by fans in a timely manner. If the heat dissipation effect of the environment where the device motherboard (printed circuit board) is located is not good, the continuous accumulation of heat will cause the temperature to continue to rise, which will affect the stability and work efficiency of the server operation. At present, the improper PID control algorithm used in the motherboard heat dissipation of edge intelligent servers to adjust the speed of the cooling fan may reduce the heat dissipation performance of the entire device and there is a risk of equipment failure. Summary of the invention

[0004] In order to solve the above problems, the embodiments of the present disclosure provide a heat dissipation control method for an edge intelligent computing server based on data feature extraction. The method aims to solve the technical problem that the heat dissipation of the mainboard of the edge intelligent server in the prior art uses an improper PID control algorithm to adjust the speed of the cooling fan, which will reduce the heat dissipation performance of the entire device and cause the risk of equipment failure.

[0005] The present disclosure provides a method for controlling heat dissipation of an edge-side intelligent computing server based on data feature extraction, the method comprising:

[0006] Obtain power consumption and temperature data of power components in edge computing servers;

[0007] Obtaining a potential temperature rise reliability coefficient of the power element according to the power consumption data and temperature data of the power element;

[0008] According to the potential temperature rise reliability coefficient, obtaining a heat dissipation enhancement significance coefficient of the server;

[0009] According to the heat dissipation enhancement significance coefficient, the exhaust volume of the fan of the server is adjusted to perform heat dissipation control.

[0010] Optionally, obtaining power consumption data and temperature data of power elements in the edge computing server includes:

[0011] In a first time window, the power consumption and temperature of any power element in the server are sampled at a first time interval, and the sampled values ​​are normalized to obtain the power consumption data and temperature data of the power element in the server.

[0012] Optionally, acquiring a potential temperature rise reliability coefficient of the power element according to the power consumption data and the temperature data of the power element includes:

[0013] Obtaining a heat transfer variability coefficient of the power element;

[0014] A potential temperature rise reliability coefficient of the power element is obtained according to the heat transfer variability coefficient.

[0015] Optionally, obtaining the heat transfer variability coefficient of the power element includes:

[0016] Acquire a power similarity coefficient of the power element according to the first power consumption data of the power element and the second power consumption data of the power element closest to the power element;

[0017] Acquire a temperature similarity coefficient of the power element according to the first temperature data of the power element and the second temperature data of the power element closest to the power element;

[0018] A heat transfer difference coefficient of the power element is obtained according to the power similarity coefficient and the temperature similarity coefficient.

[0019] Optionally, obtaining a potential temperature rise reliability coefficient of the power element according to the heat transfer variability coefficient includes:

[0020] Obtaining a power mutation abnormality coefficient of the power element;

[0021] A potential temperature rise reliability coefficient of the power element is obtained according to the power mutation abnormality coefficient and the heat transfer difference coefficient.

[0022] Optionally, obtaining the power mutation abnormality coefficient of the power element includes:

[0023] The power mutation abnormality coefficient of the power element is obtained by taking the mean of the Pettitt test statistics corresponding to the power consumption data of the power element in the first time window and the second time window, wherein the second time window is the previous time window of the first time window.

[0024] Optionally, acquiring the heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient includes:

[0025] Obtaining a power rise coefficient of the power element;

[0026] The heat dissipation enhancement significance coefficient of the power element is obtained according to the power rise coefficient and the potential temperature rise reliability coefficient.

[0027] Optionally, adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control includes:

[0028] Normalizing the heat dissipation enhancement significance coefficients of all power elements of the server within a first time window, and summing the results of the normalization processing to obtain a normalized heat dissipation enhancement significance coefficient of the server;

[0029] According to the normalized heat dissipation enhancement significance coefficient, the exhaust volume of the fan of the server is adjusted through a variable parameter PID control algorithm to perform heat dissipation control.

[0030] Optionally, the variable parameter PID control algorithm includes:

[0031] The proportional term parameter of the third time window is adjusted according to the normalized heat dissipation enhancement significance coefficient in the first time window, wherein the third time window is a next time window of the first time window.

[0032] Optionally, adjusting a proportional term parameter of a third time window according to the normalized heat dissipation enhancement significance coefficient in the first time window includes:

[0033] The sum of the normalized heat dissipation enhancement significance coefficient of the first time window and the proportional parameter of the first time window is used as the proportional parameter of the third time window.

[0034] In summary, the disclosed embodiment provides a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction, the method comprising: obtaining power consumption data and temperature data of power elements in the edge computing server; obtaining a potential temperature rise reliability coefficient of the power elements according to the power consumption data and temperature data of the power elements; obtaining a heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient; and adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control. The disclosed embodiment can solve the technical problem in the prior art that the mainboard heat dissipation of the edge intelligent server uses an improper PID control algorithm to adjust the speed of the heat dissipation fan, which will reduce the heat dissipation performance of the entire device and cause the risk of equipment failure, thereby improving the heat dissipation performance of the device and reducing the risk of equipment failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the implementation scheme of the present disclosure, the drawings required for use in the implementation scheme will be briefly introduced below. It should be understood that the drawings only show certain implementation schemes of the present disclosure and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on the drawings without paying creative work.

[0036] Figure 1 It is a flowchart of a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction according to an exemplary embodiment.

[0037] Figure 2 The present invention is a flowchart of a method for obtaining power consumption data and temperature data of power elements in an edge computing server according to an exemplary embodiment.

[0038] Figure 3 The present invention is a flow chart showing a method for obtaining a potential temperature rise reliability coefficient of a power element according to power consumption data and temperature data of the power element according to an exemplary embodiment.

[0039] Figure 4 The invention is a flow chart showing a method for obtaining a heat transfer variability coefficient of a power element according to an exemplary embodiment.

[0040] Figure 5 The present invention is a flow chart showing a method for obtaining a potential temperature rise reliability coefficient of a power element according to a heat transfer variability coefficient according to an exemplary embodiment.

[0041] Figure 6 The invention is a flow chart showing a method for obtaining a power mutation abnormality coefficient of a power element according to an exemplary embodiment.

[0042] Figure 7 The present invention is a flow chart showing a method for obtaining a heat dissipation enhancement significance coefficient of a server according to a potential temperature rise reliability coefficient according to an exemplary embodiment.

[0043] Figure 8 The present invention is a flow chart showing a method for adjusting the exhaust volume of a fan of a server according to a heat dissipation enhancement significance coefficient according to an exemplary embodiment.

[0044] Fig. 9 The figure is a flow chart of a variable parameter PID control algorithm according to an exemplary embodiment.

[0045] Fig.10 The present invention is a flow chart showing a method for adjusting a proportional term parameter of a third time window according to a normalized heat dissipation enhancement significance coefficient in a first time window according to an exemplary embodiment.

[0046] Fig.11 It is a schematic diagram showing a comparison of effects of a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction according to an exemplary embodiment. DETAILED DESCRIPTION

[0047] In order to clearly illustrate the technical features of the present solution, the present disclosure is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0048] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0049] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0050] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0051] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0052] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more". In the description of the present disclosure, unless otherwise specified, "multiple" means two or more than two, and other quantifiers are similar; "at least one item", "one or more items" or similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one item a can represent any number of a; for another example, one or more items of a, b and c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple; "and / or" is a kind of association relationship that describes the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural.

[0053] Although operations or steps are described in a specific order in the drawings in the embodiments of the present disclosure, it should not be understood that it is required to perform these operations or steps in the specific order shown or in a serial order, or to perform all the operations or steps shown to obtain the desired results. In the embodiments of the present disclosure, these operations or steps can be performed in series; these operations or steps can also be performed in parallel; or some of these operations or steps can be performed.

[0054] At the same time, it is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0055] First, the application scenarios of the present disclosure are described. Edge computing servers are advanced computing devices deployed near network physical devices or data sources. They integrate multiple core functions such as network connection, computing processing, data storage, and application services into one, forming an efficient and open working platform. This server is designed to meet a variety of application scenarios and needs to be able to adapt to various ambient temperature conditions from mild to harsh.

[0056] However, due to their high sensitivity to the environment, edge computing servers may face a higher risk of equipment failure at high temperatures. The heat of edge servers mainly comes from the CPU and other power components on the server motherboard. The server heat dissipation is mainly discharged from the device chassis by fans in a timely manner. If the heat dissipation effect of the environment where the device motherboard (printed circuit board) is located is not good, the continuous accumulation of heat will cause the temperature to continue to rise, which will affect the stability and work efficiency of the server operation. At present, the improper PID control algorithm used in the motherboard heat dissipation of edge intelligent servers to adjust the speed of the cooling fan may reduce the heat dissipation performance of the entire device and there is a risk of equipment failure.

[0057] In view of this, the embodiment of the present disclosure provides a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction. It aims to solve the technical problem that the heat dissipation of the mainboard of the edge intelligent server in the prior art uses an improper PID control algorithm to adjust the speed of the cooling fan, which will reduce the heat dissipation performance of the entire device and cause the risk of equipment failure. The present disclosure is explained below in conjunction with specific embodiments.

[0058] Figure 1 FIG. 1 is a flow chart showing a method for controlling heat dissipation of an edge-side intelligent computing server based on data feature extraction according to an exemplary embodiment. Figure 1 As shown, the embodiment of the present disclosure provides a method for controlling heat dissipation of an edge-side intelligent computing server based on data feature extraction, and the method may include the following steps:

[0059] In step S10, power consumption data and temperature data of power elements in the edge computing server are obtained.

[0060] In this step, the power consumption data and temperature data of the power elements in the edge computing server are obtained. Exemplarily, the power consumption and temperature of any power element in the server can be sampled at a first time interval (such as 0.1 seconds) in a first time window (such as a window of 2 seconds), and the sampled values ​​can be normalized to obtain the power consumption data and temperature data of the power elements in the server.

[0061] In step S20, a potential temperature rise reliability coefficient of the power element is obtained according to the power consumption data and temperature data of the power element.

[0062] In this step, the potential temperature rise reliability coefficient of the power element is obtained according to the power consumption data and temperature data of the power element. Exemplarily, the heat transfer difference coefficient of the power element can be obtained first, and then the potential temperature rise reliability coefficient of the power element can be obtained according to the heat transfer difference coefficient.

[0063] In step S30, the heat dissipation enhancement significance coefficient of the server is obtained according to the potential temperature rise reliability coefficient.

[0064] In this step, the heat dissipation enhancement significance coefficient of the server is obtained according to the potential temperature rise reliability coefficient. For example, the power rise coefficient of the power element can be obtained, and then the heat dissipation enhancement significance coefficient of the power element can be obtained according to the power rise coefficient and the potential temperature rise reliability coefficient, and then the heat dissipation enhancement significance coefficients of all power elements in the server are summed to obtain the heat dissipation enhancement significance coefficient of the server.

[0065] In step S40, the exhaust volume of the fan of the server is adjusted according to the heat dissipation enhancement significance coefficient to perform heat dissipation control.

[0066] In this step, the exhaust volume of the fan of the server is adjusted according to the heat dissipation enhancement significance coefficient to perform heat dissipation control. Exemplarily, the heat dissipation enhancement significance coefficient of the server can be normalized first to obtain the normalized heat dissipation enhancement significance coefficient of the server, and then the exhaust volume of the fan of the server is adjusted according to the normalized heat dissipation enhancement significance coefficient through a variable parameter PID control algorithm to perform heat dissipation control.

[0067] In summary, the disclosed embodiment provides a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction, the method comprising: obtaining power consumption data and temperature data of power elements in the edge computing server; obtaining a potential temperature rise reliability coefficient of the power elements according to the power consumption data and temperature data of the power elements; obtaining a heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient; and adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control. The disclosed embodiment can solve the technical problem in the prior art that the mainboard heat dissipation of the edge intelligent server uses an improper PID control algorithm to adjust the speed of the heat dissipation fan, which will reduce the heat dissipation performance of the entire device and cause the risk of equipment failure, thereby improving the heat dissipation performance of the device and reducing the risk of equipment failure.

[0068] Figure 2 FIG. 1 is a flow chart showing a method for obtaining power consumption data and temperature data of a power element in an edge computing server according to an exemplary embodiment. Figure 2 As shown, the acquisition of power consumption data and temperature data of power elements in the edge computing server may include the following steps:

[0069] In step S101, in a first time window, the power consumption and temperature of any power element in the server are sampled at a first time interval, and the sampled values ​​are normalized to obtain the power consumption data and temperature data of the power element in the server.

[0070] In this step, the power consumption and temperature of any power element in the server are sampled at a first time interval (e.g., 0.1 seconds) in a first time window (e.g., 2 seconds), and the sampled values ​​are normalized to obtain the power consumption data and temperature data of the power element in the server.

[0071] In a high-density edge computing server environment, the power consumption of a single server may be as high as hundreds of watts. If the temperature inside the server rises, this may shorten the life expectancy of the electronic components therein. In order to maintain a suitable operating temperature, the server is usually equipped with a fan to effectively discharge the heat generated during operation. In the server, high-power components such as the CPU, south bridge, north bridge, and graphics accelerator are the main heat sources. For example, the power consumption data of the power element can be obtained through the built-in files of the server system, and the server motherboard is equipped with a temperature sensor that can obtain the temperature data of the power element. Each power element can collect a set of power data and a set of temperature data.

[0072] Figure 3 FIG. 1 is a flow chart showing a method for obtaining a potential temperature rise reliability coefficient of a power element according to power consumption data and temperature data of the power element according to an exemplary embodiment. Figure 3 As shown, obtaining the potential temperature rise reliability coefficient of the power element according to the power consumption data and temperature data of the power element may include the following steps:

[0073] In step S201, the heat transfer variability coefficient of the power element is obtained.

[0074] In this step, the heat transfer difference coefficient of the power element is obtained. Exemplarily, the power similarity coefficient of the power element can be obtained based on the first power consumption data of the power element and the second power consumption data of the power element closest to the power element, and the temperature similarity coefficient of the power element can be obtained based on the first temperature data of the power element and the second temperature data of the power element closest to the power element, and then the heat transfer difference coefficient of the power element can be obtained based on the power similarity coefficient and the temperature similarity coefficient.

[0075] In step S202, a potential temperature rise reliability coefficient of the power element is obtained according to the heat transfer variability coefficient.

[0076] In this step, the potential temperature rise reliability coefficient of the power element is obtained according to the heat transfer difference coefficient. For example, the power mutation abnormality coefficient of the power element can be obtained first, and then the potential temperature rise reliability coefficient of the power element can be obtained according to the power mutation abnormality coefficient and the heat transfer difference coefficient.

[0077] Figure 4 FIG. 1 is a flow chart showing a method for obtaining a heat transfer variability coefficient of a power element according to an exemplary embodiment. Figure 4 As shown, the obtaining of the heat transfer variability coefficient of the power element may include the following steps:

[0078] In step S2011, a power similarity coefficient of the power element is acquired according to the first power consumption data of the power element and the second power consumption data of the power element closest to the power element.

[0079] In this step, according to the first power consumption data of the power element and the second power consumption data of the power element closest to the power element , get the power similarity coefficient of the power element . Exemplarily, the power similarity coefficient of the power element is obtained The following formulas may be included:

[0080] Formula 1

[0081] in, is the power similarity coefficient of the ith power element, and the power element closest to the ith power element is denoted as j. , Respectively represent the power consumption data of the i-th and j-th power components in the first time window, Indicates calculating the ED edit distance of two sets of power consumption data.

[0082] In step S2012, a temperature similarity coefficient of the power element is acquired according to the first temperature data of the power element and the second temperature data of the power element closest to the power element.

[0083] In this step, according to the first temperature data of the power element and the second temperature data of the power element closest to the power element , obtain the temperature similarity coefficient of the power component . Exemplarily, the temperature similarity coefficient of the power element is obtained The following formulas may be included:

[0084] Formula 2

[0085] in, is the temperature similarity coefficient of the ith power element, and the power element closest to the ith power element is denoted as j. , Respectively represent the temperature data of the i-th and j-th power components in the first time window, Indicates calculating the ED edit distance of two sets of temperature data.

[0086] In step S2013, the heat transfer difference coefficient of the power element is obtained according to the power similarity coefficient and the temperature similarity coefficient.

[0087] In this step, according to the power similarity coefficient and temperature similarity coefficient , obtain the heat transfer variability coefficient of the power element For example, the heat transfer variability coefficient of the power element is obtained: The following formulas may be included:

[0088] Formula 3

[0089] in, is the heat transfer variability coefficient of the i-th power element.

[0090] In edge servers, the design and working methods of high-power components mean that they are usually working at the same time to handle incoming tasks and jobs. When these power components are working, they consume electricity and generate heat, and the heavier the workload, the more heat is usually generated. Therefore, it can be concluded that if the temperature of a power component is higher, the power consumption of this power component should also be greater when it is running.

[0091] However, each power element can execute instructions independently, and different power elements may have different degrees of temperature rise, depending on their workload, heat dissipation efficiency, and their location on the motherboard. In some cases, although the operating power consumption of some power elements is not high, the detected temperature shows an upward trend. This phenomenon is usually caused by the presence of other high-power operating components nearby. The heat emitted by the high-power operating components will affect the surrounding environment. This phenomenon indicates that the current heat dissipation efficiency of the edge server is insufficient, and the accumulated heat cannot be discharged in time, resulting in poor overall heat dissipation.

[0092] When the power similarity coefficient The smaller the value, the closer the power consumption of two adjacent components in the time window is. The smaller it is, the closer the temperatures of the two adjacent components are. If the power similarity coefficient is calculated If it is too large or too small, the temperature similarity coefficient It is also too large or too small, that is, the power similarity coefficient Similarity coefficient with temperature positive correlation, then the heat transfer variability coefficient calculated at this time is The smaller it is, the better the local heat dissipation efficiency of the component in the server. The temperature similarity coefficient is small. If it is too large, the calculated heat transfer variability coefficient If the power similarity coefficient is larger, it means that the local heat dissipation efficiency of the power component is insufficient, and its temperature increase may be caused by heat transfer rather than power increase. The temperature similarity coefficient is too large. is too small, the calculated heat transfer variability coefficient If it gets bigger, it also means that the local heat dissipation efficiency of the power component is insufficient.

[0093] Figure 5 FIG. 1 is a flow chart showing a method for obtaining a potential temperature rise reliability coefficient of a power element according to a heat transfer variability coefficient according to an exemplary embodiment. Figure 5 As shown, obtaining the potential temperature rise reliability coefficient of the power element according to the heat transfer variability coefficient may include the following steps:

[0094] In step S2021, the power mutation abnormality coefficient of the power element is obtained.

[0095] In this step, the power mutation abnormality coefficient of the power element is obtained. For example, the power mutation abnormal coefficient of the power element can be obtained by the mean of the Pettitt test statistic corresponding to the power consumption data of the same power element in the first time window and the second time window. , where the second time window is the previous time window of the first time window, and the time length of the second time window can be the same as the time length of the first time window. The Pettitt test method detects mutation points by calculating a statistic. This statistic is based on the cumulative difference between the value of each data point in the sequence and the sequence average. The Pettitt test is a well-known technology, and the specific process will not be repeated here.

[0096] In step S2022, the potential temperature rise reliability coefficient of the power element is obtained according to the power sudden change abnormality coefficient and the heat transfer difference coefficient.

[0097] In this step, according to the power mutation abnormality coefficient and heat transfer variability coefficient , obtain the potential temperature rise reliability coefficient of the power component For example, the potential temperature rise reliability coefficient of the power element is obtained The following formulas may be included:

[0098] Formula 4

[0099] in, is the potential temperature rise reliability coefficient of the i-th power element, is the power mutation abnormality coefficient of the i-th power element, is the heat transfer variability coefficient of the i-th power element.

[0100] Power mutation abnormal coefficient The smaller the value, the less random the power data change is, and only short-term data changes exist. The smaller it is, the greater the degree to which the temperature change of the power component is affected by the power change. The calculated potential temperature rise reliability coefficient The higher the value, the more consistent the power change trend of the power element is with the temperature change trend in the window. This further indicates that the temperature rise of the power element is mainly caused by the increase in its power. The potential temperature rise reliability coefficient is calculated. It has the beneficial effect of improving the accuracy of warming trend prediction.

[0101] Figure 6 FIG. 1 is a flow chart showing a method for obtaining a power mutation abnormal coefficient of a power element according to an exemplary embodiment. Figure 6 As shown, the obtaining of the power mutation abnormal coefficient of the power element may include the following steps:

[0102] In step S20211, the power mutation anomaly coefficient of the power element is obtained by taking the mean of the Pettitt test statistics corresponding to the power consumption data of the power element in the first time window and the second time window, wherein the second time window is the previous time window of the first time window.

[0103] In this step, the power mutation abnormal coefficient of the power element can be obtained by the mean of the Pettitt test statistic corresponding to the power consumption data of the same power element in the first time window and the second time window. , wherein the second time window is the previous time window of the first time window, and the time length of the second time window may be the same as the time length of the first time window. Pre-heating is performed according to the rising trend of power in the edge server. If a rapid increase in power consumption is detected in the current time window, the exhaust volume of the fan is increased in advance to quickly dissipate heat. However, in the process of edge computing, the load changes of high-power consumption components are random, and the power data may have short-term rapid fluctuations. If most of the power data in the time window is in a stable state, and there are large fluctuations at the window boundary, it will cause great interference to the trend judgment. In order to detect whether there are short-term large fluctuations in power data, the Pettitt test algorithm can be used to calculate the possibility of mutations for each data point in the power consumption data sequence. The input of the Pettitt test is the power consumption data of the same power element in each time window, and the output is the statistic corresponding to each power consumption data point of the same power element. The Pettitt test is a well-known technology, and the specific process will not be repeated.

[0104] Figure 7 FIG. 1 is a flow chart showing a method for obtaining a heat dissipation enhancement significance coefficient of a server according to a potential temperature rise reliability coefficient according to an exemplary embodiment. Figure 7 As shown, obtaining the heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient may include the following steps:

[0105] In step S301, the power rise coefficient of the power element is obtained.

[0106] In this step, the power rise factor of the power element is obtained For example, the power rise coefficient of the i-th power element in each time window is The power rise coefficient can be determined by the existing Mann-Kendall trend test. The larger the value, the greater the possibility of rapid temperature rise of the server. The Mann-Kendall trend test method is used to obtain the standard normal distribution statistic Z corresponding to the power data in each time window. If Z>0, it means that the data has a significant upward trend. On the contrary, if Z<0, it means that the data has a significant downward trend. Therefore, in this embodiment, the standard normal distribution statistic obtained based on the algorithm is used as the power rise coefficient in the corresponding time window. .

[0107] In step S302, the heat dissipation enhancement significance coefficient of the power element is obtained according to the power rise coefficient and the potential temperature rise reliability coefficient.

[0108] In this step, according to the power rise factor and potential temperature rise reliability factor , obtain the heat dissipation enhancement significance coefficient of the power component . Exemplarily, the heat dissipation enhancement significance coefficient of the power element is obtained The following formulas may be included:

[0109] Formula 5

[0110] Heat dissipation enhancement significance coefficient of power components and power rise factor and potential temperature rise reliability factor Positive correlation, heat dissipation enhancement significance coefficient The larger it is, the more you need to increase the fan exhaust volume in time to improve the heat dissipation efficiency.

[0111] Figure 8 FIG. 1 is a flow chart showing a method for adjusting the exhaust volume of a fan of a server according to a heat dissipation enhancement significance coefficient according to an exemplary embodiment. Figure 8 As shown, adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control may include the following steps:

[0112] In step S401, the heat dissipation enhancement significance coefficients of all power elements of the server within a first time window are normalized, and the results of the normalization are summed to obtain a normalized heat dissipation enhancement significance coefficient of the server.

[0113] In this step, the heat dissipation enhancement significance coefficient of all power components of the server in the first time window is A normalization process is performed, and the results of the normalization process are summed to obtain a normalized heat dissipation enhancement significance coefficient K of the server.

[0114] In step S402, the exhaust volume of the fan of the server is adjusted according to the normalized heat dissipation enhancement significance coefficient through a variable parameter PID control algorithm to perform heat dissipation control.

[0115] In this step, according to the normalized heat dissipation enhancement significance coefficient K, the exhaust volume of the fan of the server is adjusted through a variable parameter PID control algorithm to perform heat dissipation control.

[0116] Fig. 9 FIG. 1 is a flow chart of a variable parameter PID control algorithm according to an exemplary embodiment. Fig. 9 As shown, the variable parameter PID control algorithm may include the following steps:

[0117] In step S4021, a proportional term parameter of a third time window is adjusted according to the normalized heat dissipation enhancement significance coefficient in the first time window, wherein the third time window is a next time window of the first time window.

[0118] In this step, the proportional term parameter P3 of the third time window is adjusted according to the normalized heat dissipation enhancement significance coefficient K in the first time window, wherein the third time window is the next time window of the first time window. The time length of the third time window can be the same as the time length of the first time window.

[0119] Fig.10 FIG. 1 is a flow chart showing a method for adjusting a proportional term parameter of a third time window according to a normalized heat dissipation enhancement significance coefficient in a first time window according to an exemplary embodiment. Fig.10 As shown, the adjusting the proportional term parameter of the third time window according to the normalized heat dissipation enhancement significance coefficient in the first time window may include the following steps:

[0120] In step S40211, the sum of the normalized heat dissipation enhancement significance coefficient of the first time window and the proportional parameter of the first time window is used as the proportional parameter of the third time window.

[0121] In this step, the sum of the normalized heat dissipation enhancement significance coefficient K of the first time window and the proportional parameter P1 of the first time window is used as the proportional parameter P3 of the third time window. That is, P3=K+P1. Among them, the input of the variable parameter PID control algorithm is the server power consumption data in each time window, and the output is the fan speed control amount. The fan adjusts the exhaust volume according to the fan speed control amount, thereby realizing the heat dissipation control of the edge side server.

[0122] The embodiment of the present disclosure analyzes the consistency relationship between the power consumption difference and the temperature difference between the power elements in the edge server to determine whether the temperature rise of the element is caused by the operation of the element itself or by heat transfer, and detects the local heat dissipation situation. Then, the rising trend of the power consumption of the element is pre-cooled, and the randomness of the power mutation of the power element in the edge server is analyzed by the mutation detection algorithm, which has the beneficial effect of improving the reliability of the prediction / detection of the rising power consumption / temperature rise trend of the element. The heat dissipation enhancement significance coefficient is obtained according to the degree of the rising power consumption trend. Due to the difference between the power and temperature of different elements and the short-term drastic change of power, the PID controller is not accurate enough to control the fan speed only according to the initial parameters. The present disclosure obtains the proportional term parameter of the PID control by an adaptive algorithm, and adjusts the exhaust volume of the fan by the output fan speed control amount to control the heat dissipation of the mainboard and the edge server. It makes up for the defect that the improper PID control parameters in the existing control method may lead to the reduction of the heat dissipation performance of the mainboard and the server. The embodiment of the present disclosure dynamically adjusts the proportional parameters of the PID control algorithm according to the normalized heat dissipation enhancement significance coefficient K, which can better realize heat dissipation control, improve the heat dissipation efficiency and timeliness of the mainboard, and make the operation of the server safer and more stable.

[0123] like Fig.11 As shown in FIG. 1 , the temperature stability of each power component on the same server is obtained by assigning the same task and using different PID temperature control algorithms. In the prior art, the stability of the result obtained by the PID control algorithm is evaluated by calculating the variance. Specifically, the fluctuation of the temperature data within a certain period of time after the control by different PID algorithms is characterized, that is, the temperature data of each power component within a set period of time is monitored after the control by different PID algorithms is calculated, the variance of all temperature data is calculated, and the variance is negatively correlated and normalized to obtain the temperature stability, which can be expressed as: ,in Indicates temperature stability, It represents the variance of the temperature data in the set time period after the corresponding PID algorithm is regulated. The length of the set time period can be 2 seconds, and Norm is the normalization function. It should be noted that the use of data variance to evaluate the stability of the control result of the PID algorithm is a known technology, and only a brief introduction is given here.

[0124] Fig.11 In the figure, A1 is the percentage of temperature stability of the power element obtained by adjusting the existing PID algorithm, A2 is the percentage of temperature stability of the power element obtained by the PID algorithm dynamically adjusted in this embodiment, the ordinate is the percentage of temperature stability, and the abscissa is the number of data sets tested. Fig.11 It can be seen that the temperature stability of the power element is higher after dynamic adjustment using this embodiment.

[0125] In summary, the disclosed embodiment provides a heat dissipation control method for an edge-side intelligent computing server based on data feature extraction, the method comprising: obtaining power consumption data and temperature data of power elements in the edge computing server; obtaining a potential temperature rise reliability coefficient of the power elements according to the power consumption data and temperature data of the power elements; obtaining a heat dissipation enhancement significance coefficient of the server according to the potential temperature rise reliability coefficient; and adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control. The disclosed embodiment can solve the technical problem in the prior art that the mainboard heat dissipation of the edge intelligent server uses an improper PID control algorithm to adjust the speed of the cooling fan, which will reduce the heat dissipation performance of the entire device and cause the risk of equipment failure, thereby improving the heat dissipation performance of the device and reducing the risk of equipment failure.

[0126] In some embodiments, the present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon. When the program instructions are executed by a processor, the steps of the edge-side intelligent computing server heat dissipation control method based on data feature extraction provided by the present disclosure are implemented.

[0127] In another exemplary embodiment, a computer program product is also provided, which includes a computer program that can be executed by a programmable electronic device, and the computer program has a code portion for executing the above-mentioned edge-side intelligent computing server heat dissipation control method based on data feature extraction when executed by the programmable electronic device.

[0128] The above-mentioned embodiments only express several implementation methods of the present disclosure, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present disclosure. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present disclosure, and these all belong to the protection scope of the present disclosure.

Claims

1. Heat dissipation control method for edge-side intelligent computing servers based on data feature extraction, It is characterized in that The method comprises: Obtain power consumption and temperature data of power components in edge computing servers; Obtaining a potential temperature rise reliability coefficient of the power element according to the power consumption data and temperature data of the power element; Obtaining a power rise coefficient of the power element, and obtaining a heat dissipation enhancement significance coefficient of the server according to the power rise coefficient and the potential temperature rise reliability coefficient; According to the heat dissipation enhancement significance coefficient, adjusting the exhaust volume of the fan of the server to perform heat dissipation control; The method for obtaining the potential temperature rise reliability coefficient includes: Obtaining a heat transfer difference coefficient of the power element; obtaining a power mutation abnormality coefficient of the power element; obtaining a potential temperature rise reliability coefficient of the power element according to the power mutation abnormality coefficient and the heat transfer difference coefficient; The obtaining of the heat transfer variability coefficient of the power element comprises: Acquire a power similarity coefficient of the power element according to the first power consumption data of the power element and the second power consumption data of the power element closest to the power element; Acquire a temperature similarity coefficient of the power element according to the first temperature data of the power element and the second temperature data of the power element closest to the power element; Acquiring a heat transfer difference coefficient of the power element according to the power similarity coefficient and the temperature similarity coefficient; The step of adjusting the exhaust volume of the fan of the server according to the heat dissipation enhancement significance coefficient to perform heat dissipation control includes: Normalizing the heat dissipation enhancement significance coefficients of all power elements of the server within a first time window, and summing the results of the normalization processing to obtain a normalized heat dissipation enhancement significance coefficient of the server; According to the normalized heat dissipation enhancement significance coefficient, the exhaust volume of the fan of the server is adjusted by a variable parameter PID control algorithm to perform heat dissipation control; The obtaining of the power similarity coefficient of the power element comprises: According to the first power consumption data of the power element and the second power consumption data of the power element closest to the power element , get the power similarity coefficient of the power element , including the following formula:. in, is the power similarity coefficient of the ith power element, and the power element closest to the ith power element is denoted as j. , Respectively represent the power consumption data of the i-th and j-th power components in the first time window, Indicates calculating the ED edit distance between two sets of power consumption data; The obtaining of the temperature similarity coefficient of the power element comprises: According to the first temperature data of the power element and the second temperature data of the power element closest to the power element , obtain the temperature similarity coefficient of the power component , including the following formula: in, is the temperature similarity coefficient of the ith power element, and the power element closest to the ith power element is denoted as j. , Respectively represent the temperature data of the i-th and j-th power components in the first time window, Indicates calculating the ED edit distance of two sets of temperature data; Obtaining the heat transfer difference coefficient of the power element according to the power similarity coefficient and the temperature similarity coefficient, including: in, is the heat transfer variability coefficient of the i-th power element; The obtaining of the power mutation abnormality coefficient of the power element comprises: Obtaining the power mutation abnormality coefficient of the power element by taking the mean of the Pettitt test statistics corresponding to the power consumption data of the same power element in the first time window and the second time window, wherein the second time window is the previous time window of the first time window; The obtaining of the potential temperature rise reliability coefficient of the power element comprises: in, is the potential temperature rise reliability coefficient of the i-th power element, is the power mutation abnormality coefficient of the i-th power element, is the heat transfer variability coefficient of the i-th power element; The step of obtaining the power rise coefficient of the power element and obtaining the heat dissipation enhancement significance coefficient of the power element according to the power rise coefficient and the potential temperature rise reliability coefficient includes: The power rise coefficient of the i-th power element in each time window is obtained through the standard normal distribution statistics in the Mann-Kendall trend test method. ; The power rise coefficient The potential temperature rise reliability factor The product between them is taken as the heat dissipation enhancement significance coefficient of the i-th power element; The variable parameter PID control algorithm comprises: Adjusting a proportional term parameter of a third time window according to the normalized heat dissipation enhancement significance coefficient in the first time window, wherein the third time window is a next time window of the first time window; The adjusting the proportional term parameter of the third time window according to the normalized heat dissipation enhancement significance coefficient in the first time window includes: taking the sum of the normalized heat dissipation enhancement significance coefficient of the first time window and the proportional parameter of the first time window as the proportional parameter of the third time window; The method for obtaining the power rise coefficient includes: The standard normal distribution statistic obtained based on the Mann-Kendall trend test algorithm is used as the power increase coefficient in the corresponding time window. The larger the power increase coefficient, the greater the possibility of rapid temperature rise of the server.

2. According to the method for controlling heat dissipation of edge-side intelligent computing servers based on data feature extraction according to claim 1, It is characterized in that The obtaining of power consumption data and temperature data of power elements in the edge computing server includes: In a first time window, the power consumption and temperature of any power element in the server are sampled at a first time interval, and the sampled values ​​are normalized to obtain the power consumption data and temperature data of the power element in the server.

Images