A power distribution method for a computer room power supply system

Abstract

The present invention relates to the technical field of power distribution. More specifically, the present invention relates to a power distribution method for a power supply system in a computer room. The method includes: obtaining a reference data set of parameter items at each acquisition moment in the time series of operation data of bus-connected devices, where the parameter items include current and / or voltage; calculating the trend and prediction deviation of the parameter items at each acquisition moment, normalizing the product of the prediction deviation of the parameter items, a preset error factor, and the trend to obtain the error factor of the parameter items; using the error factor in the HTFE algorithm to obtain the predicted value of the parameter items at the next acquisition moment, and obtaining the power required by the device at the next acquisition moment through the product of the predicted values of each parameter item to achieve power distribution, effectively improving the accuracy of power distribution in the computer room power supply system.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution. More specifically, the present invention relates to a power distribution method for a computer room power supply system. Background Art

[0002] A computer room is the core place for data storage, processing and transmission, usually including relevant power supply equipment, environmental control equipment and security equipment, and its energy consumption accounts for a relatively large proportion of the total energy consumption of an enterprise. Therefore, improving energy utilization efficiency and realizing energy conservation and efficiency improvement in the computer room is one of the important goals of computer room management.

[0003] An uninterruptible power supply (UPS) power distribution cabinet is a special power distribution cabinet for a computer room, which distributes power to various devices connected to the power distribution cabinet bus according to a predetermined distribution ratio. There has been much research on UPS power distribution in the prior art. For example, the patent application document with the publication number CN107947337A discloses a power control device based on a UPS system, which includes an uninterruptible power supply module that can switch to a shutdown state and immediately supply backup power when the power supply of an external power supply device is interrupted; a power distribution module that supplies backup power to each power distribution area by pre-dividing the power distribution area of the power system; and a power control module that defines its emergency power supply level by detecting the charge of the battery and controls the supply of backup power to each power distribution area according to the priority order of the preset power distribution area and the emergency power supply level.

[0004] The above prior art distributes power to each power distribution area through a UPS. However, in the power distribution process, in order to ensure power supply during peak hours or sudden high loads, redundant power is usually reserved, resulting in power supply to equipment during non-peak hours exceeding the current actual power consumption demand, that is, there is a phenomenon of power surplus, increasing the energy consumption cost.

[0005] Based on this, how to accurately implement the power distribution of a computer room power supply system is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] To solve the above technical problem of how to accurately implement the power distribution of a computer room power supply system, the present invention proposes a power distribution method for a computer room power supply system, which includes the following steps:

[0007] Obtain a reference data set of parameter items at each acquisition moment in the operation data time series of the bus-connected devices, and the parameter items include current and / or voltage;

[0008] ;

[0009] , , are respectively the trend, the degree of fluctuation, and the value of the th parameter item at the i-th acquisition moment, is the number of data in the reference dataset, , are respectively the th and the th timestamps of the acquisition moments, is the value of the th parameter item at the th acquisition moment, is the exponential function with base e; the product of the prediction deviation of the parameter item, the preset error factor, and the trend is normalized to obtain the error factor of the parameter item; the error factor is used in the HTFE algorithm to obtain the predicted value of the parameter item at the next acquisition moment, and the required power of the device at the next acquisition moment is obtained by multiplying the predicted values of each parameter item to achieve power distribution.

[0010] The present invention takes into account that the standby power reserved to ensure power supply during peak hours or sudden high loads in the power distribution process will increase the energy consumption cost, and the power demands in different time periods are different, which will cause resource waste. Based on this, the present invention predicts the power demand of the device through the HTFE algorithm and adjusts the power of the device based on the predicted value, effectively improving the accuracy of power distribution and reducing resource waste. In this process, the present invention considers that the fixed error factor in the HTFE algorithm cannot adapt to the device operation data with various change trends, resulting in a low accuracy of the obtained prediction results; based on this, the present invention adjusts the error factor according to the trend change of the recent data of the device parameter item at each acquisition moment, so as to accurately obtain the predicted value of the power demand of the device and effectively improve the power distribution accuracy of the computer room power supply system.

[0011] According to a power distribution method for a computer room power supply system provided by the present invention, before obtaining the reference dataset of the parameter items at each acquisition moment in the operation data time series of the bus-connected devices, it further includes: collecting the parameter data generated during the operation of each device connected to the computer room UPS bus, and preprocessing the parameter data to obtain the parameter items at each acquisition moment in the operation data time series.

[0012] The present invention takes into account that there may be noise data or data missing in the originally collected parameter data, so the overall quality of the data is improved through preprocessing.

[0013] A power distribution method for a computer room power supply system provided by the present invention, the method for obtaining the reference data set of the parameter items includes: obtaining the historical data set of the parameter items; obtaining the number of data based on the size of the preset reference data set in the historical data set of the parameter items, so as to construct the reference data set of the parameter items.

[0014] The present invention analyzes by obtaining the historical data of the parameter item as its recent reference data set, reduces the data waiting time, and enables the accurate obtaining of the trend change of the parameter item based on this.

[0015] A power distribution method for a computer room power supply system provided by the present invention, the method for obtaining the fluctuation degree of the parameter items includes: normalizing the sum of the squares of the differences between the parameter item and each data in its reference data set to obtain the fluctuation degree of the parameter item.

[0016] The present invention can accurately obtain the fluctuation degree of the parameter item by obtaining the data difference between the parameter item and each parameter item in its reference data set. The larger the sum of the squares of the differences, the greater the fluctuation degree.

[0017] A power distribution method for a computer room power supply system provided by the present invention, the method for obtaining the prediction deviation of the parameter items includes: obtaining the initial prediction value of the parameter item through a prediction algorithm; recording the absolute value of the difference between the initial prediction value of the parameter item and the numerical value as the prediction deviation of the parameter item.

[0018] A power distribution method for a computer room power supply system provided by the present invention, the method of obtaining the power required by the device at the next acquisition moment by multiplying the prediction values of each parameter item to achieve power distribution includes: transmitting the power required by each device to the corresponding device through the corresponding bus.

[0019] A power distribution method for a computer room power supply system provided by the present invention, after realizing the power distribution, it further includes: performing abnormal monitoring and early warning on the parameter items of each device at each acquisition moment.

[0020] The present invention takes into account that there may be potential safety hazards when there are abnormalities in the parameter data generated during the operation of the device. Therefore, through abnormal monitoring and early warning, the staff is timely reminded to handle the abnormal situation.

[0021] A power distribution method for a computer room power supply system provided by the present invention, after performing abnormal monitoring and early warning on the parameter items of each device at each acquisition moment, it further includes: associatively storing the numerical values, required power, and abnormal monitoring results of the parameter items of each device at each acquisition moment.

[0022] The present invention has the following beneficial effects:

[0023] When the present invention realizes the power distribution of the computer room power supply system, it predicts the power demand of the equipment through the HTFE algorithm, and adjusts the power of the equipment based on the predicted value, effectively improving the accuracy of power distribution and reducing resource waste. In this process, the present invention considers that a fixed error factor cannot adapt to the operation data of various changing trends of the equipment, resulting in a low accuracy of the obtained prediction result; based on this, the present invention adjusts the error factor according to the trend change by obtaining the trend change of the recent data of the equipment parameter items at each acquisition moment, so as to accurately obtain the predicted value of the power demand of the equipment, and effectively improve the power distribution accuracy of the computer room power supply system. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By referring to the following detailed description with reference to the accompanying drawings, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0025] Figure 1 It is a schematic flow chart of a power distribution method for a computer room power supply system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] The following will describe in detail the specific implementation manners of the present invention with reference to the accompanying drawings.

[0028] The computer room is the core place for data storage, processing, and transmission, usually including relevant power supply equipment, environmental control equipment, and security equipment, and its energy consumption accounts for a relatively large proportion of the total energy consumption of the enterprise. Therefore, improving energy utilization efficiency and realizing energy conservation and efficiency improvement in the computer room are one of the important goals of computer room management.

[0029] The uninterruptible power supply (UPS) power distribution cabinet is a special power distribution cabinet for the computer room, which distributes power to various devices connected to the power distribution cabinet bus according to a predetermined distribution ratio. However, in the process of power distribution, in order to ensure the power supply during peak hours or sudden high loads, redundant power is usually reserved, resulting in the power supply of equipment during non-peak hours exceeding the current actual power consumption demand, that is, there is a phenomenon of power surplus, increasing the energy consumption cost.

[0030] Based on this, an embodiment of the present invention discloses a power distribution method for a computer room power supply system. By analyzing the historical voltage and current data of devices, the predicted demand power values of each device are obtained, and power resources are allocated to each device based on the predicted power demand values, which can effectively improve the accuracy of power distribution and avoid waste of power resources.

[0031] Please refer to Figure 1 as shown in Figure 1 FIG. 1, which is a schematic flowchart of a power distribution method for a computer room power supply system provided by an embodiment of the present invention. The method specifically includes the following steps.

[0032] S1: Obtain a reference data set of parameter items at each acquisition moment in the operation data time series of the bus-connected devices.

[0033] Among them, the parameter items include current and / or voltage.

[0034] It should be noted that in a computer room power supply system, the UPS power distribution cabinet is usually connected to the corresponding devices through its interfaces to maintain power supply stability. The UPS power distribution cabinet delivers power by connecting the corresponding devices through different buses of each interface. Therefore, the wiring information of each interface in the UPS power distribution cabinet can be obtained through the wiring records of the corresponding power distribution cabinet in the computer room management system, including the different bus numbers of each interface and the numbers of all devices connected to each bus. Then, by installing sensors on the devices, the acquisition of device parameter data is realized. The changes in the current and voltage of the devices directly determine the power demand of the devices.

[0035] Exemplarily, in an embodiment of the present invention, before obtaining the reference data set of parameter items at each acquisition moment in the operation data time series of the bus-connected devices, it further includes: collecting the parameter data generated during the operation of each device connected to the UPS bus in the computer room, and preprocessing the parameter data to obtain the parameter items at each acquisition moment in the operation data time series.

[0036] Among them, the preprocessing method can be data denoising, missing data interpolation, data format conversion, etc., which can be specifically set according to actual needs, and the embodiment of the present invention does not limit it too much here.

[0037] Specifically, voltage sensors are installed on the devices to collect voltage data, and current sensors are installed to collect current data; the acquisition frequency of the sensors can be set to once per second, which can be specifically set according to actual needs. At each acquisition moment, a voltage parameter value and a current parameter value are obtained, and the parameter values are preprocessed to obtain the parameter items in the operation data time series. The parameter items at each acquisition moment in the operation data time series correspond to each other one by one.

[0038] It should be further noted that in the 9th issue of the Journal of Computer Era in 2023, an article titled "Research on Time Series Prediction Method Based on Historical Trends and Prediction Errors" was published. The adaptive prediction algorithm for time series based on historical trends and prediction errors (Historical Trends Forecast Errors, abbreviated as HTFE) recorded in the article can analyze the trends in recent time series. By calculating the predicted value and the true value at the previous moment of the current moment, the prediction deviation at the current moment is obtained, and the prediction error is obtained by multiplying the prediction deviation by the error factor; it is continuously adjusted according to the prediction error and the current value at the current moment, and finally the predicted value at the next moment of the current moment is obtained.

[0039] Therefore, the embodiments of the present invention can predict the required power of the device through the HTFE algorithm.

[0040] However, the conventional HTFE algorithm uses a fixed error factor during the adjustment process. In practical applications, due to the significant differences in the operating modes of computer room equipment during peak and off-peak periods, the monitored data shows different trend changes. The fixed preset error factor is difficult to adapt to these changes. A larger error factor is highly sensitive to new data and can quickly respond to data changes, but it is prone to overfitting and ignores long-term trends; a smaller error factor is less sensitive to new data and can better capture long-term trends, but it responds slowly to short-term fluctuations. Therefore, the selection of the error factor affects the accuracy of device power prediction results and power distribution.

[0041] Based on this, the embodiments of the present invention analyze the recent data of the current parameter item through the HTFE algorithm, calculate its trend performance, and adjust the preset error factor based on the trend performance of the parameter item, so as to accurately obtain the predicted value of the parameter item at the next moment and realize the power prediction of the device.

[0042] Exemplarily, in the embodiments of the present invention, the method for obtaining the reference data set of the parameter item includes the following two possible implementation manners:

[0043] In one possible implementation manner, the historical data set of the parameter item can be obtained, and the number of data is obtained based on the size of the preset reference data set in the historical data set of the parameter item to construct the reference data set of the parameter item.

[0044] Among them, the size of the reference data set can be 10, and it can be specifically set according to actual needs. The size of the reference data set is the number of parameter items included in the reference data set, that is, the number of data. The finally obtained reference data set of the current parameter item does not include the current parameter item.

[0045] In this way, by obtaining historical data as the recent data of the current parameter item, the embodiment of the present invention can accurately obtain its change trend from the historical data, so as to accurately obtain its prediction result.

[0046] In another possible implementation manner, the number of data can be equally obtained based on the size of the preset reference data set on both sides of the parameter item to construct the reference data set of the parameter item.

[0047] In this way, by obtaining the parameter item data around the current parameter item as the recent data of the current parameter item, the embodiment of the present invention can accurately obtain the data change before and after the current parameter item, so as to accurately obtain its historical and future change trends and realize the prediction of the current parameter item.

[0048] Taking an example to illustrate the reference data set of the parameter item: the parameter item is voltage, and the size of the reference data set of the current voltage is 10. 10 voltage data can be obtained from the historical data set of the current voltage to construct the reference data set of the current voltage; or 5 voltage data can be obtained on both sides of the current voltage respectively, so as to construct the reference data set of the current voltage.

[0049] After obtaining the reference data set of the parameter item at each acquisition moment based on the above steps, continue to execute the following steps to analyze the trend performance in its reference data set.

[0050] S2: Obtain the fluctuation degree of the parameter item, and calculate the trend of each parameter item at each acquisition moment based on the fluctuation degree.

[0051] It should be noted that there are various devices in the computer room, and their operation models may have different peak periods, non-peak periods and stable periods. For example, for the server CPU in the computer room, the significant change in its recent data may indicate a significant increase or decrease in the user usage or access volume, resulting in the server device experiencing a rapid load fluctuation.

[0052] Based on this, the embodiment of the present invention can obtain its fluctuation degree by obtaining the recent data change situation of the device at the current moment. If the recent data of the parameter item fluctuates greatly at the current acquisition moment, it indicates that there is an obvious load trend change in the operation state of the device at the current moment.

[0053] Exemplarily, in the embodiment of the present invention, the method for obtaining the fluctuation degree of the parameter item includes: normalizing the sum of the squares of the differences between the parameter item and each data in its reference data set to obtain the fluctuation degree of the parameter item.

[0054] It can be understood that if the sum of the squares of the differences between the parameter item and each data in its reference data set at an acquisition moment is larger, it indicates that the difference between the parameter item and each parameter item in its reference data set is larger, the fluctuation degree is larger, and the corresponding fluctuation trend is more obvious.

[0055] It should be further noted that the greater the degree of fluctuation of a parameter item at the current acquisition moment, the more significant the data change in its recent data, but at the same time, it also means that the degree of dispersion in the recent data is greater. And the existence of multiple trend changes in the recent data will also affect the degree of dispersion and the degree of fluctuation. For example, a sharp increase or decrease in the user access volume will cause a significant upward or downward trend in the load of network devices.

[0056] Based on this, in order to further determine whether the data fluctuations in the recent data are fluctuations with consistent trends, the embodiments of the present invention obtain the estimated value of the current parameter item by weighting each parameter item in the reference dataset of the current parameter item, and obtain whether there is a significant trend change in the recent data of the current parameter item according to the deviation between the estimated value of the current parameter item and the true value. If there is a significant increasing or decreasing trend in the recent data, the estimated value of the current parameter item is relatively close to the true value.

[0057] Exemplarily, in the embodiments of the present invention, to determine the trend of each parameter item at each acquisition moment, the following relational expression can be specifically referred to:

[0058] ;

[0059] is the trend of the th parameter item at the i-th acquisition moment, is the degree of fluctuation of the th parameter item at the i-th acquisition moment, is the number of data in the reference dataset, is the timestamp of the th acquisition moment, is the timestamp of the th acquisition moment, is the value of the th parameter item at the th acquisition moment, is the value of the th parameter item at the i-th acquisition moment,

[0060] In the above formula, the difference between the timestamp of the acquisition moment of each parameter item in the reference dataset of the th parameter item and the timestamp of the acquisition moment of the initial parameter item in the reference dataset is its weight. The closer the parameter item in the reference dataset is to the i-th acquisition moment, the greater its weight.

[0061] represents the estimated value of the th parameter item at the i-th acquisition moment. If the If there is an obvious increasing or decreasing trend in the reference dataset of a parameter item, then the value is closer to the value of the th parameter item at the i-th acquisition moment. At this time, if the th parameter item has a large degree of fluctuation at the i-th acquisition moment, it means that the amplitude of this trend change is larger, that is, the th parameter item has stronger trendability at the i-th acquisition moment.

[0062] After obtaining the trendability of each parameter item at each acquisition moment based on the above steps, continue to execute the following steps.

[0063] S3: Normalize the product of the prediction deviation of the parameter item, the preset error factor and the trendability to obtain the error factor of the parameter item.

[0064] Among them, the preset error factor can be set to 0.5; the value of the preset error factor can be set according to actual needs in the range of 0 to 1, and the embodiments of the present invention do not limit this too much here.

[0065] It should be noted that the operation modes of the computer room equipment are significantly different during peak and non-peak periods, resulting in different trend changes in the data, and a fixed error factor is difficult to adapt to such changes. Based on the above steps, the embodiments of the present invention can analyze the recent data of the parameter item to obtain the trendability of the parameter item, and adjust the preset error factor through the trendability of the parameter item, so as to accurately obtain the error factor that conforms to the parameter item at the current moment.

[0066] Exemplarily, in the embodiments of the present invention, the method for obtaining the prediction deviation of a parameter item includes: obtaining the initial prediction value of the parameter item through a prediction algorithm; recording the absolute value of the difference between the initial prediction value of the parameter item and the value as the prediction deviation of the parameter item.

[0067] Exemplarily, when calculating the initial prediction value of the parameter item at the next acquisition moment of the current acquisition moment through the HTFE algorithm, the prediction deviation of the current acquisition moment can be obtained through the difference between the initial prediction value and the true value of the parameter item at the previous acquisition moment of the current acquisition moment; the prediction error is obtained by using the product of the prediction deviation and the preset error factor; and the initial prediction value of the parameter item at the next acquisition moment of the current acquisition moment is finally obtained by continuously adjusting according to the prediction error and the value of the parameter item at the current acquisition moment.

[0068] Among them, the specific steps of calculating the initial prediction value of the parameter item at each acquisition moment through the HTFE algorithm can be realized by the prior art, and the embodiments of the present invention do not elaborate here.

[0069] It should be noted that the HTFE algorithm mainly realizes the prediction of the next acquisition moment of the time series by adjusting the prediction deviation at the previous acquisition moment. Therefore, when calculating the initial prediction value of the first parameter item in the historical dataset of the current parameter item, there is no prediction error at the previous acquisition moment for adjustment.

[0070] Based on this, when obtaining the initial prediction value of the first parameter item through the prediction algorithm, prediction algorithms such as the moving average method and the exponential smoothing method can be used, and specific settings can be made according to actual needs. The embodiments of the present invention do not impose excessive restrictions here; the prediction algorithm for calculating the initial prediction values of other subsequent parameter items can use the HTFE algorithm.

[0071] Among them, the specific steps of obtaining the prediction value of the parameter item through the moving average method and the exponential smoothing method can be obtained through the prior art, and the embodiments of the present invention will not elaborate here.

[0072] After obtaining the prediction deviation of each parameter item based on the above steps, the error factor of the parameter item can be obtained through the following formula.

[0073] Exemplarily, the error factor of the parameter item is determined through the normalization processing of the product of the prediction deviation of the parameter item, the preset error factor, and the trend. Specifically, the following relational expression can be referred to:

[0074] ;

[0075] is the error factor of the th parameter item at the i-th acquisition moment, is the preset error factor, is the initial prediction value of the th parameter item at the i-th acquisition moment, is the th parameter item at the i-th acquisition moment, is the th parameter item at the i-th acquisition moment, is the normalization function.

[0076] In the above formula, represents the prediction deviation of the th parameter item at the i-th acquisition moment.

[0077] represents the correction weight for the preset error factor. The greater the deviation between the predicted value and the actual value of the current parameter item, the more it indicates that the preset error factor does not adapt to the data change of the parameter item at the current acquisition moment, resulting in lower prediction accuracy. At this time, if the trend of the parameter item is greater, the preset error factor needs to be increased to improve the sensitivity of the prediction algorithm to new data.

[0078] After obtaining the error factors of each parameter item at the acquisition moment based on the above steps, the predicted values of the parameter items at the next moment can be obtained in the HTFE algorithm based on the error factors of the parameter items at the acquisition moment.

[0079] S4: Use the error factors in the HTFE algorithm to obtain the predicted values of the parameter items at the next acquisition moment, and obtain the power required by the device at the next acquisition moment through the product of the predicted values of each parameter item, so as to achieve power distribution.

[0080] Among them, the specific steps of using the error factors in the HTFE algorithm to obtain the predicted values of the parameter items at the next acquisition moment are similar to the steps of calculating the initial predicted values of the parameter items through the HTFE algorithm above, and the specific implementation methods are recorded in the prior art. Therefore, the embodiments of the present invention will not be elaborated too much here.

[0081] It should be noted that there may be connections of multiple devices on the busbar of the power distribution cabinet. Therefore, it is necessary to analyze the prediction results of the electrical parameter data of all devices on each busbar, so as to predict in real time the power required by each device on the busbar at the next acquisition moment. Based on the prediction results of the total power required by each busbar, adjust the power transmission power of different busbars at the interface, so as to achieve the effect of optimizing power distribution and ensure that the device can reduce unnecessary energy consumption during peak and off-peak periods.

[0082] Exemplarily, in the embodiments of the present invention, the power required by the device at the next acquisition moment is obtained through the product of the predicted values of each parameter item to achieve power distribution, including: transmitting the power required by each device to the corresponding device through the corresponding busbar.

[0083] It can be understood that the parameter items are current and voltage. The predicted power of the device can be obtained through the product of the current predicted value and the voltage predicted value, and the distribution of power resources can be achieved based on the predicted power of the device.

[0084] It should be further noted that the power supply of the computer room is one of the core components for the normal operation of the enterprise. If there is an abnormality in the power supply of the computer room, there may be relatively large potential safety hazards.

[0085] Based on this, in the embodiments of the present invention, after achieving power distribution, it further includes: performing abnormal monitoring and early warning on the parameter items of each device at each acquisition moment.

[0086] Specifically, corresponding abnormal thresholds can be set for each parameter item, and the abnormal monitoring results of the device can be obtained through the comparison results of the numerical values of the parameter items with the abnormal thresholds.

[0087] Among them, the abnormal threshold can be specifically set according to the parameters of the device, and the embodiments of the present invention do not limit it too much here.

[0088] Exemplarily, in the embodiments of the present invention, for the parameter items of each device at each acquisition moment, abnormal monitoring and early warning are performed, and then it further includes: associatively storing the numerical values, required power, and abnormal monitoring results of the parameter items of each device at each acquisition moment.

[0089] After realizing the power distribution and abnormal monitoring of the device based on the above embodiments, relevant data can be associatively stored to facilitate data analysis by the staff.

[0090] It can be seen that in the embodiments of the present invention, when realizing the power distribution of the computer room power supply system, a reference data set of the parameter items at each acquisition moment in the time series of the operation data of the bus-connected devices can be obtained, and the parameter items include current and voltage;

[0091] ;

[0092] , , are respectively the trend, fluctuation degree, and numerical value of the th parameter item at the i-th acquisition moment, is the number of data items in the reference data set, , are respectively the th and the th timestamps of the acquisition moments, is the th numerical value of the th parameter item at the th acquisition moment; normalizing the product of the prediction deviation of the parameter item, the preset error factor, and the trend to obtain the error factor of the parameter item; using the error factor in the HTFE algorithm to obtain the predicted value of the parameter item at the next acquisition moment, and obtaining the required power of the device at the next acquisition moment through the product of the predicted values of each parameter item to achieve power distribution.

[0093] In this way, the embodiments of the present invention predict the power demand of the device through the HTFE algorithm and adjust the power of the device based on the predicted value, effectively improving the accuracy of power distribution and reducing resource waste. In this process, the embodiments of the present invention consider that a fixed error factor cannot adapt to the device operation data with various changing trends, resulting in a low accuracy of the obtained prediction results; based on this, the embodiments of the present invention adjust the error factor according to the trend change of the recent data of the device parameter items at each acquisition moment, so as to accurately obtain the predicted value of the power demand of the device, effectively improving the accuracy of power distribution of the computer room power supply system.

[0094] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A power distribution method for a computer room power supply system, characterized in that, Including: Obtain a reference data set of parameter items at each acquisition moment in the time series of operation data of the bus connection device, where the parameter items include current and / or voltage; ； , , are respectively the trend, the degree of fluctuation, and the value of the th parameter item at the i-th acquisition moment, is the number of data in the reference dataset, j is the independent variable of the acquisition moment in the reference dataset, , are respectively the th and the th timestamps of the acquisition moments, is the value of the th parameter item at the th acquisition moment, is the exponential function with base e; Normalize the product of the prediction deviation of the parameter item, the preset error factor, and the trendiness to obtain the error factor of the parameter item; Use the error factor in the HTFE algorithm to obtain the predicted value of the parameter item at the next acquisition moment, and obtain the power required by the device at the next acquisition moment through the product of the predicted values of each parameter item to achieve power distribution; Using the error factor in the HTFE algorithm to obtain the predicted value of the parameter item at the next acquisition moment includes: obtaining the prediction deviation at the current acquisition moment through the difference between the initial predicted value and the true value of the parameter item at the previous acquisition moment of the current acquisition moment; Obtain the prediction error by multiplying the prediction deviation by the preset error factor; continuously adjust according to the prediction error and the value of the parameter item at the current acquisition moment to obtain the predicted value of the parameter item at the next acquisition moment of the current acquisition moment.

2. The power distribution method of a computer room power supply system according to claim 1, characterized in that, Before obtaining the reference data set of parameter items at each acquisition moment in the time series of operation data of the bus connection device, it further includes: Collect the parameter data generated during the operation of each device connected to the UPS bus in the computer room, and preprocess the parameter data to obtain the parameter items at each acquisition moment in the time series of operation data.

3. The power distribution method of a computer room power supply system according to claim 1, characterized in that, The method for obtaining the reference data set of the parameter item includes: Obtain the historical data set of the parameter item; Based on the size of the preset reference data set, obtain the number of data in the historical data set of the parameter item to construct the reference data set of the parameter item.

4. A power distribution method for a computer room power supply system according to claim 1, characterized in that The method for obtaining the fluctuation degree of the parameter item includes: Normalize the sum of the squares of the differences between the parameter item and each data in its reference data set to obtain the fluctuation degree of the parameter item.

5. A power distribution method for a computer room power supply system according to claim 1, characterized in that The method for obtaining the prediction deviation of the parameter item includes: Obtain the initial predicted value of the parameter item through the prediction algorithm; record the absolute value of the difference between the initial predicted value of the parameter item and the value of the parameter item as the prediction deviation of the parameter item.

6. A power distribution method for a computer room power supply system according to claim 1, characterized in that The method of obtaining the power required by the device at the next acquisition moment through the product of the predicted values of each parameter item to achieve power distribution includes: Transmit the power required by each device to the corresponding device through the corresponding bus.

7. A power distribution method for a computer room power supply system according to claim 1, characterized in that, After achieving power distribution, it further includes: Conduct abnormal monitoring and early warning on the parameter items of each device at each acquisition moment.

8. A power distribution method for a computer room power supply system according to claim 7, characterized in that After conducting abnormal monitoring and early warning on the parameter items of each device at each acquisition moment, it further includes: Associatively store the values of the parameter items, the required power, and the abnormal monitoring results of each device at each acquisition moment.

Images