A method, device and computer equipment for predicting energy consumption of power-specific chips

Abstract

The present invention discloses a method for predicting the energy consumption of a power - specific chip, characterized in that: the method includes: determining the target energy - consumption data of the power - specific chip; constructing an energy - consumption prediction model adapted to the architecture of the power - specific chip based on the whale algorithm introducing a non - linear convergence factor α; the expression of the non - linear convergence factor α is: α = α int -(α int -α out )t 2 ; in the formula, α int is the preset initial value of α; α out is the preset final value of α; t is the predicted running time of the energy - consumption prediction model; processing the target energy - consumption data according to the energy - consumption prediction model to obtain the corresponding energy - consumption prediction result; solving the problem that the existing method for predicting the energy consumption of a power - specific chip cannot effectively predict the energy consumption of the chip, so that power enterprises cannot effectively cope with the sudden energy - consumption problems of the chip.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart grid, and in particular relates to a method, device and computer equipment for predicting energy consumption of a power-specific chip. Background Art

[0002] Power-specific chips are widely used in the core processors and microcontrollers of power companies. They are one of the core hardware devices of the power system and play a vital role in processing and controlling. The prediction of energy consumption of power-specific chips is mainly through collecting historical energy consumption data and combining relevant algorithms to scientifically predict energy consumption. In recent years, the use of new hardware equipment has led to abnormal operation of the power system. Traditional energy consumption prediction methods will not be able to effectively predict the energy consumption of chips, which also makes it impossible for power companies to effectively deal with the sudden energy consumption of chips and ensure the stable operation of the power system. Summary of the invention

[0003] The technical problem to be solved by the present invention is: to provide a method, device and computer equipment for predicting the energy consumption of a dedicated power chip, so as to solve the technical problems that the method for predicting the energy consumption of a dedicated power chip cannot effectively predict the energy consumption of the chip, making it impossible for power companies to effectively deal with the sudden energy consumption of the chip and unable to ensure the stable operation of the power system.

[0004] Technical solution of the present invention:

[0005] A method for predicting energy consumption of a power-specific chip, characterized in that the method comprises:

[0006] S1. Determine the target energy consumption data of the power-specific chip;

[0007] S2. Based on the whale algorithm with the introduction of nonlinear convergence factor α, an energy consumption prediction model adapted to the architecture of the power-specific chip is constructed; the expression of the nonlinear convergence factor α is:

[0008] α=α int -(α int -α out )t 2 ; (1)

[0009] In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model;

[0010] S3. Process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0011] The step of determining target energy consumption data of the power-specific chip includes:

[0012] S11, obtaining initial energy consumption data of the power-specific chip, and processing the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; the preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method;

[0013] S12. Determine target energy consumption data based on the obtained preprocessed data.

[0014] The power-specific chip includes a three-layer architecture.

[0015] The first layer architecture includes an AXI bus; at least one of a dynamic memory and a digital signal processor is provided on the AXI bus;

[0016] The second layer architecture includes an AHB bus connected to the AXI bus; the AHB bus is provided with at least one of an instruction interface and an interrupt controller;

[0017] The third layer architecture includes an APB bus connected to the AHB bus; the APB bus is provided with at least one of a timer, a serial port, a controller and an execution module.

[0018] The target energy consumption data is processed according to the energy consumption prediction model to obtain the corresponding energy consumption prediction result, including:

[0019] S31. In the current iteration process, when the target energy consumption data is processed according to the energy consumption prediction model, the energy consumption of the power-specific chip is predicted by the following formula based on the optimal solution of the whale group to obtain the corresponding first energy consumption prediction value:

[0020]

[0021] In formula (2), f is the energy consumption prediction value, N is the number of samples collected for the target energy consumption data, and x i is the actual energy consumption value determined based on the target energy consumption data; x i ' is the preset average predicted energy consumption value;

[0022] S32, when entering the next iteration, according to the maximum adaptive weight and the minimum adaptive weight of the model, the weight of the energy consumption prediction model is updated, and the location information of the whale group is reallocated, and the energy consumption prediction is re-performed based on the corresponding updated energy consumption prediction model to obtain the corresponding second energy consumption prediction value;

[0023] S33: When it is determined that the first energy consumption prediction value matches the second energy consumption prediction value, output a corresponding energy consumption prediction result.

[0024] Before step S31, the method further includes:

[0025] Initialize model parameters and randomly distribute whales; where:

[0026] The model parameters include at least one of the movement range of the whale group, the number of whales, the maximum number of data iterations of the model, the maximum adaptive weight, and the minimum adaptive weight;

[0027] The distribution location information of each whale corresponds to a standard kernel function, which is used to correct the calculation error of the model to ensure the accuracy of the predicted structure.

[0028] Step S32, according to the maximum adaptability weight and the minimum adaptability weight of the model, adjusts the weight of the energy consumption prediction model, including:

[0029] S321. According to the adaptive weight calculation method, the weight w of the energy consumption prediction model is adjusted by the following formula:

[0030] w=w min +(w max +w min )×k μ ; (3)

[0031] In the formula, w max is the preset maximum adaptive weight, w min is the preset minimum adaptive weight, k is the adjustment ratio of the adaptive weight, and μ is the relationship coefficient of the weight decrease.

[0032] A device for predicting energy consumption of a power-specific chip, the device comprising a data acquisition module, a model building module and an energy consumption prediction module, wherein:

[0033] A data acquisition module, used to determine the target energy consumption data of the power-specific chip;

[0034] The model building module is used to build an energy consumption prediction model adapted to the architecture of the power-specific chip based on the whale algorithm that introduces a nonlinear convergence factor α; wherein the expression of the nonlinear convergence factor α is:

[0035] α=α int -(α int -α out )t 2 ; (1)

[0036] In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model;

[0037] The energy consumption prediction module is used to process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0038] The data acquisition module is also used to acquire the initial energy consumption data of the power-specific chip, and process the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; the preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method; based on the obtained preprocessing data, the target energy consumption data is determined.

[0039] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods of claims 1 to 6 when executing the computer program.

[0040] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method according to any one of claims 1 to 6.

[0041] Beneficial effects of the present invention:

[0042] The present invention constructs an energy consumption prediction model for power-specific chips based on the whale algorithm that introduces a nonlinear convergence factor. When the value of the nonlinear convergence factor is larger and the convergence is slower, the searchable range of the energy consumption prediction model can be expanded, thereby improving the global searchability of the model. And as the number of iterations of the target energy consumption data increases, the nonlinear convergence factor shows a downward trend in value. This will also make the prediction efficiency of the energy consumption prediction model faster. Compared with the prior art, the present application has a higher prediction accuracy and stability, can effectively reduce the difficulty of chip energy consumption prediction, and provide theoretical data support for power allocation of the power system. It solves the technical problems that the prediction method of the energy consumption of power-specific chips in the prior art cannot effectively predict the energy consumption of the chip, making it impossible for power companies to effectively deal with the sudden energy consumption of the chip and unable to ensure the stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a method for predicting energy consumption of a power-specific chip in one embodiment of the present invention;

[0044] Figure 2 is a flow chart of a humpback whale's foraging process in one embodiment of the present invention;

[0045] Figure 3 A schematic diagram of the subject architecture of a power-specific chip in one embodiment of the present invention;

[0046] Figure 4 A detailed schematic diagram of a process for predicting energy consumption of a power-specific chip in one embodiment of the present invention;

[0047] Figure 5 A schematic diagram showing a comparison of prediction accuracy between the prediction method of the present application in one embodiment of the present invention, the traditional power-specific chip energy consumption prediction based on task scheduling, and the power-specific chip energy consumption prediction method based on least squares support vector machine;

[0048] Figure 6 A schematic diagram showing a comparison of prediction error rates of the prediction method of the present application in one embodiment of the present invention, the traditional power-specific chip energy consumption prediction based on task scheduling, and the power-specific chip energy consumption prediction method based on least squares support vector machine;

[0049] Figure 7 A schematic diagram showing a comparison of prediction stability between the prediction method of the present application in one embodiment of the present invention and the traditional power-specific chip energy consumption prediction method based on task scheduling and the power-specific chip energy consumption prediction method based on least squares support vector machine;

[0050] Figure 8 It is a system structure diagram of an energy consumption prediction device for a power-specific chip in one embodiment of the present invention. DETAILED DESCRIPTION

[0051] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0052] In one or more embodiments of the present invention, Figure 1 As shown, a method for predicting energy consumption of a power-specific chip is provided, and the method is applied to a computer device (the computer device can be a terminal or a server, and the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server can be an independent server or a server cluster composed of multiple servers) as an example for explanation, including the following steps:

[0053] S1. Determine the target energy consumption data of the power-specific chip.

[0054] Specifically, the computer device is connected to the energy consumption data collection device and obtains the initial energy consumption data of the power-specific chip. Due to the influence of the working environment of the power-specific chip and the energy consumption data collection device, the initial energy consumption data often has problems such as missing data and noise interference. This problem will not only increase the workload of energy consumption prediction, but also easily cause inaccurate prediction results. Therefore, before using the energy consumption prediction model constructed later to process the energy consumption data obtained by the computer device, it is necessary to pre-process the initial energy consumption data to avoid the occurrence of the above-mentioned problems. It should be noted that the pre-processing method adopted by the computer device can be flexibly set. For example, in one embodiment, when it is determined that there is noise interference, a moving average algorithm is used to perform noise reduction processing on the energy consumption data. The embodiment of the present application does not limit the pre-processing method.

[0055] In one embodiment, the target energy consumption data includes various data indicators that can reflect energy consumption, such as power consumption per unit time, etc., which is not limited in this embodiment of the present application.

[0056] S2. Based on the whale algorithm with the nonlinear convergence factor α, an energy consumption prediction model adapted to the architecture of the power-specific chip is constructed; wherein the expression of the nonlinear convergence factor α is:

[0057] α=α int -(α int -α out )t 2 ; (1)

[0058] In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model.

[0059] On one hand, it should be noted that the whale algorithm is a search algorithm based on the humpback whale's foraging behavior, in which the humpback whale's foraging behavior includes prey circling, hunting implementation, and random hunting. The algorithm process includes: the humpback whale group marks a target prey group and notifies other whale group members to jointly surround the prey. The whale group moves along a spiral path during the process of surrounding and hunting the prey, and seeks the best prey by continuously updating the whale group's position information. Figure 2 For your understanding, this embodiment of the present application will not be described in detail.

[0060] However, the traditional whale algorithm has certain limitations in the process of finding the optimal solution and cannot present the optimal solution finding process well. Therefore, based on the traditional whale algorithm, the embodiment of the present application introduces a nonlinear convergence factor α to improve the traditional whale algorithm in view of the structural characteristics of the power-specific chip, and constructs an energy consumption prediction model based on the improved whale algorithm, thereby realizing the optimal solution for the energy consumption prediction of the power-specific chip in turn.

[0061] On the other hand, it should be noted that when predicting the energy consumption of the initial power-specific chip, the larger the value of the nonlinear convergence factor α, the slower the model converges, the wider the search range of the prediction model, and the stronger the global search. And as the number of iterations of the energy consumption data increases, the convergence factor α tends to decrease in value, and the prediction efficiency of the model becomes faster.

[0062] S3. Process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0063] Specifically, during the corresponding iteration process, the computer device can directly output the energy consumption prediction result obtained during the current iteration process. Alternatively, in order to ensure the accuracy of the output result, the computer device can output the result based on the energy consumption prediction result calculated in the current and previous iterations while further ensuring the accuracy of the energy consumption prediction result obtained during the current iteration process. For specific implementation methods, please refer to Figure 3 , because the subsequent Figure 3 The following is a detailed description, and no further description is given of the embodiments of the present application.

[0064] The above-mentioned method for predicting energy consumption of power-specific chips constructs an energy consumption prediction model for power-specific chips based on the whale algorithm that introduces a nonlinear convergence factor. When the value of the nonlinear convergence factor is larger and the convergence is slower, the searchable range of the energy consumption prediction model can be expanded, thereby improving the global searchability of the model. And as the number of iterations of the target energy consumption data increases, the nonlinear convergence factor shows a downward trend in value. This will also make the prediction efficiency of the energy consumption prediction model faster. Compared with the prior art, the present application has higher prediction accuracy and stability, can effectively reduce the difficulty of chip energy consumption prediction, and provide theoretical data support for power allocation in power systems.

[0065] In one or more embodiments of the present invention, in step S1, determining target energy consumption data of the power-specific chip includes:

[0066] S11, obtaining initial energy consumption data of the power-specific chip, and processing the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; the preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method.

[0067] Specifically, the computer device eliminates redundant data in the initial energy consumption data by vertical and horizontal comparison, and after eliminating the redundant data, selectively performs data unification processing. In one embodiment, the data unification processing can be performed with reference to the following formula:

[0068]

[0069] Among them, h i ' represents the initial energy consumption data h after eliminating redundant data i , the corresponding value after unified processing; h max is the initial energy consumption data h i The maximum value that can be obtained; h min is the initial energy consumption data h i The minimum value that can be obtained.

[0070] Based on the above embodiment, after the unified processing is completed, considering the noise interference problem, in the current embodiment, the computer device uses a moving average algorithm to calculate the energy consumption data h obtained by the unified processing. i 'Perform noise reduction processing. The calculation principle is to take the current processing data as the center and calculate the arithmetic mean of the data before and after the data. The average value obtained is the arithmetic mean of the current processing data.

[0071] S12. Determine target energy consumption data based on the obtained preprocessed data.

[0072] Specifically, in the current embodiment, the computer device uses the preprocessed data obtained based on step S11 as the target energy consumption data, and inputs the target energy consumption data into the constructed energy consumption prediction model, which processes the target energy consumption data to realize the prediction of the energy consumption of the power-specific chip.

[0073] In the above embodiment, before performing energy consumption prediction, the acquired energy consumption data is preprocessed to avoid noise interference and data missing problems, improve the accuracy of the prediction results, and meet the application needs of power companies.

[0074] In one or more embodiments of the present invention, in step S2, the power-specific chip includes a three-layer architecture, wherein: the first layer architecture includes an AXI bus; the AXI bus is provided with a dynamic memory and at least one of a digital signal processor; the second layer architecture includes an AHB bus connected to the AXI bus; the AHB bus is provided with an instruction interface and at least one of an interrupt controller; the third layer architecture includes an APB bus connected to the AHB bus; the APB bus is provided with at least one of a timer, a serial port, a controller and an execution module.

[0075] For details, please refer to Figure 4 ,in:

[0076] (1) The architecture of the power-specific chip is an AXI bus, which includes two sets of dynamic memory eDRAN0 and core processing modules such as DSP (digital signal processor), where each core processing module has a separate instruction interface and data transmission interface. It should be noted that the above-mentioned instruction interface is used for internal transmission of the chip, and the data transmission interface can be connected to external devices (such as energy data acquisition equipment) for data sharing.

[0077] (2) The second layer of the architecture of the power-specific chip is the AHB bus, which is connected to the first layer main line and is constructed in a matrix structure and full interconnection mode. It contains four instruction interfaces. It should be noted that the instruction interface is responsible for processing and analyzing instructions from the first layer of architecture, and supports custom program compilation. In memory mode, the instructions of the first layer of architecture can be read and written. At the same time, it also contains an interrupt controller. When the chip has high energy consumption, unstable operation or failure, the interrupt controller can be activated to terminate the chip operation to ensure that other chip structures are not damaged.

[0078] (3) The third layer of the architecture of the power-specific chip is the APB bus, which includes multiple components such as a timer, a serial port, a controller, and an execution module. It is an important data processing module and is compatible with a data communication link. Users can access Ethernet to obtain data. At the same time, in order to ensure data security, in the current embodiment, mechanisms such as watchdogs and firewalls are used to defend against malicious attacks.

[0079] In one or more embodiments of the present invention, in step S3, the target energy consumption data is processed according to the energy consumption prediction model to obtain a corresponding energy consumption prediction result, including:

[0080] S31. In the current iteration process, when the target energy consumption data is processed according to the energy consumption prediction model, the energy consumption of the power-specific chip is predicted by the following formula based on the optimal solution of the whale group to obtain the corresponding first energy consumption prediction value:

[0081]

[0082] In formula (2), f is the energy consumption prediction value, N is the number of samples collected for the target energy consumption data, and x i is the actual energy consumption value determined based on the target energy consumption data; x i ' is the preset average predicted energy consumption value.

[0083] S32. When entering the next iteration, the weight of the energy consumption prediction model is updated according to the maximum adaptive weight and the minimum adaptive weight of the model, and the location information of the whale group is redistributed. The energy consumption prediction is re-performed based on the corresponding updated energy consumption prediction model to obtain the corresponding second energy consumption prediction value.

[0084] Specifically, to some extent, the selection of parameter α determines the prediction performance of the energy consumption prediction model. In the current embodiment, considering the influence of the whale group position vector in the improved whale algorithm on the prediction result, the prediction weight of the model is calculated by an adaptive weight calculation method, so as to adjust the model parameters in the data iteration, thereby improving the flexibility of the power-specific chip energy consumption prediction.

[0085] S33: When it is determined that the first energy consumption prediction value matches the second energy consumption prediction value, output a corresponding energy consumption prediction result.

[0086] For details, please refer to Figure 3 The computer device compares the first energy consumption prediction value and the second energy consumption prediction value, and when it is determined that the two are equal, the first energy consumption prediction value or the second energy consumption prediction value is output as the prediction result. Otherwise, it returns to step S31 and performs energy consumption prediction again.

[0087] In one embodiment, before outputting the prediction result, the computer device will also check whether the first energy consumption prediction value or the second energy consumption prediction value meets the preset output format. If so, the original data will be directly output; otherwise, the corresponding energy consumption prediction value will be converted into a numerical value, and the converted value will be output as the prediction result.

[0088] In one or more embodiments of the present invention, before step S31, the method further includes: initializing model parameters and randomly distributing whales; wherein: the model parameters include at least one of the moving range of the whale group, the number of whales, the maximum number of data iterations of the model, the maximum adaptive weight, and the minimum adaptive weight; the distribution position information of each whale corresponds to a standard kernel function, and the standard kernel function is used to correct the calculation error of the model to ensure the accuracy of the predicted structure.

[0089] In one or more embodiments of the present invention, step S32, adjusting the weight of the energy consumption prediction model according to the maximum adaptability weight and the minimum adaptability weight of the model, includes:

[0090] S321. According to the adaptive weight calculation method, the weight w of the energy consumption prediction model is adjusted by the following formula:

[0091] w=w min +(w max +w min )×k μ ; (3)

[0092] In the formula, w max is the preset maximum adaptive weight, w min is the preset minimum adaptive weight, k is the adjustment ratio of the adaptive weight, and μ is the relationship coefficient of the weight decrease.

[0093] It should be noted that the value of μ represents the downward trend of the adaptive weight, and the larger the value of μ, the greater the decrease in the weight w. When μ approaches 0, it means that the weight w has no obvious change.

[0094] In one embodiment, in order to verify the actual prediction performance of the present method, a comparative experiment was conducted, and in the experimental environment, the energy consumption data of a certain power-specific chip for one week was used as the experimental data sample set, and the prediction method of the present application was selected for experimental comparison with the traditional power-specific chip energy consumption prediction based on task scheduling and the power-specific chip energy consumption prediction method based on least squares support vector machine. The set experimental parameters are shown in Table 1 below.

[0095] Table 1 Experimental parameters

[0096] project parameter Data transmission interface IO Command interface SOI Power chip model AMD Athlon X4860k operating system Windows Protocol Stack Z-Stack Maximum data collection volume per time 10G Control Module CNA

[0097] According to the experimental parameters designed above, the prediction method of this application and the traditional prediction method are selected to predict the energy consumption of the power chip, and the prediction interval is 0.5h. In order to ensure the authenticity and reliability of the experimental results, the energy consumption prediction accuracy standard commonly used by the State Grid is selected: the prediction accuracy and prediction error rate are the basis for measuring the prediction accuracy of the three prediction methods, among which, the calculation method of the prediction accuracy and prediction error rate is as follows:

[0098]

[0099]

[0100] In the formula, E MAPE Indicates the prediction accuracy, E FA represents the prediction error rate, x i is the actual energy consumption value determined according to the target energy consumption data, xi ' is the preset average predicted energy consumption value, and N is the number of samples collected for the target energy consumption data. The comparison results of the prediction accuracy and prediction error rate of the three prediction methods are shown in Figure 5 and Figure 6 As shown. It can be seen from the above two figures that with the increase in the number of data iterations, the accuracy of the three prediction methods has decreased, and at the same time, the error rate began to rise. Among them, the accuracy of the prediction method based on task scheduling has a clear downward trend, with the lowest accuracy of 51%, the average accuracy of 75%, and the average error rate of 23%. The accuracy of the prediction method based on the least squares support vector machine has a relatively gentle downward trend, with the lowest accuracy of 69%, the average accuracy of 86%, and the average error rate of 15%. The accuracy of the prediction method based on the improved whale algorithm of the present application has no obvious downward trend, the general accuracy is higher than 93%, the error rate is lower than 10%, the average accuracy is 96%, and the average error rate is 6%. Its prediction accuracy is much higher than the other two traditional prediction methods.

[0101] In the current embodiment, in order to further verify the stability of the three power-specific chip energy consumption prediction methods, the time series description method is used to describe the data fluctuation state of the three prediction methods in the energy consumption prediction process. Among them, the faster the data fluctuation frequency and the larger the amplitude change range, the worse the stability of the prediction method. The prediction stability comparison results of the three prediction methods can be referred to. Figure 7 As can be seen from the above figure, within the 175 hours of observation, the data fluctuations of the prediction method based on task scheduling are more drastic, and the fluctuation changes are irregular, and the data fluctuation amplitude changes within the range of ±3. Figure 7 It can be seen that the prediction method based on the least squares support vector machine has a high frequency of data fluctuations in the early stage and is relatively gentle in the later stage, but the overall amplitude has no obvious change and is always maintained within the range of ±2. In contrast, the prediction method based on the improved whale algorithm in this application has regular data fluctuations and consistent frequency of change, with small overall fluctuations and a data fluctuation amplitude change range of ±1.

[0102] In summary, the energy consumption prediction method for power-specific chips based on the improved whale algorithm of the present application has high prediction accuracy and stability, and can accurately predict the energy consumption of power-specific chips. The traditional prediction method cannot properly process the collected energy consumption data, resulting in low prediction result accuracy, and the calculation method used is not flexible and the calculation process is complicated, which affects the overall stability. It can be concluded that the prediction performance of the research in this application is better and is more suitable for the energy consumption prediction work of power-specific chips.

[0103] Please refer to Figure 8The present application discloses a power-specific chip energy consumption prediction device 800, which includes a data acquisition module 801, a model building module 802, and an energy consumption prediction module 803, wherein:

[0104] The data acquisition module 801 is used to determine the target energy consumption data of the power-specific chip;

[0105] The model building module 802 is used to build an energy consumption prediction model that is compatible with the architecture of the power-specific chip based on the whale algorithm that introduces a nonlinear convergence factor α; wherein the expression of the nonlinear convergence factor α is:

[0106] α=α int -(α int -α out )t 2 ; (1)

[0107] In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model;

[0108] The energy consumption prediction module 803 is used to process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results.

[0109] In one of the embodiments, the data acquisition module is also used to acquire the initial energy consumption data of the power-specific chip, and process the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; the preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method; based on the obtained preprocessing data, the target energy consumption data is determined.

[0110] In one embodiment, the power-specific chip includes a three-layer architecture, wherein: the first layer architecture includes an AXI bus; the AXI bus is provided with a dynamic memory and at least one of a digital signal processor; the second layer architecture includes an AHB bus connected to the AXI bus; the AHB bus is provided with an instruction interface and at least one of an interrupt controller; the third layer architecture includes an APB bus connected to the AHB bus; the APB bus is provided with at least one of a timer, a serial port, a controller and an execution module.

[0111] In one embodiment, the energy consumption prediction module 803 is also used in the current iteration process. When processing the target energy consumption data according to the energy consumption prediction model, the energy consumption of the power-specific chip is predicted by the following formula based on the optimal solution of the whale group to obtain the corresponding first energy consumption prediction value:

[0112]

[0113] In formula (2), f is the energy consumption prediction value, N is the number of samples collected for the target energy consumption data, and x i is the actual energy consumption value determined based on the target energy consumption data; x i ' is the preset average predicted energy consumption value; when entering the next iteration, the weight of the energy consumption prediction model is updated and the location information of the whale group is redistributed according to the maximum adaptive weight and the minimum adaptive weight of the model, and the energy consumption prediction is re-performed based on the corresponding updated energy consumption prediction model to obtain the corresponding second energy consumption prediction value; when it is determined that the first energy consumption prediction value and the second energy consumption prediction value match, the corresponding energy consumption prediction result is output.

[0114] In one embodiment, the apparatus 800 further includes an initialization module, wherein:

[0115] The initialization module is used to initialize the model parameters and randomly distribute whales; wherein: the model parameters include the movement range of the whale group, the number of whales, the maximum number of data iterations of the model, the maximum adaptive weight, and at least one of the minimum adaptive weight; the distribution position information of each whale corresponds to a standard kernel function, which is used to correct the calculation error of the model to ensure the accuracy of the predicted structure.

[0116] In one embodiment, the energy consumption prediction module 803 is further configured to adjust the weight w of the energy consumption prediction model according to the following formula based on an adaptive weight calculation method:

[0117] w=w min +(w max +w min )×k μ ; (3)

[0118] In the formula, w max is the preset maximum adaptive weight, w min is the preset minimum adaptive weight, k is the adjustment ratio of the adaptive weight, and μ is the relationship coefficient of the weight decrease.

[0119] The above-mentioned power-specific chip energy consumption prediction device constructs an energy consumption prediction model for power-specific chips based on the whale algorithm that introduces a nonlinear convergence factor. When the value of the nonlinear convergence factor is larger and the convergence is slower, the searchable range of the energy consumption prediction model can be expanded, thereby improving the global searchability of the model. And as the number of iterations of the target energy consumption data increases, the nonlinear convergence factor shows a downward trend in value. This will also make the prediction efficiency of the energy consumption prediction model faster. Compared with the prior art, the present application has higher prediction accuracy and stability, can effectively reduce the difficulty of chip energy consumption prediction, and provide theoretical data support for power allocation in the power system.

[0120] In one or more embodiments of the present invention, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0121] The above-mentioned computer device constructs an energy consumption prediction model for power-specific chips based on the whale algorithm that introduces a nonlinear convergence factor. When the value of the nonlinear convergence factor is larger and the convergence is slower, the searchable range of the energy consumption prediction model can be expanded, thereby improving the global searchability of the model. And as the number of iterations of the target energy consumption data increases, the nonlinear convergence factor tends to decrease in value. This will also make the prediction efficiency of the energy consumption prediction model faster. Compared with the prior art, the present application has higher prediction accuracy and stability, can effectively reduce the difficulty of chip energy consumption prediction, and provide theoretical data support for power allocation in the power system.

[0122] In one or more embodiments of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0123] The above-mentioned storage medium constructs an energy consumption prediction model for power-specific chips based on the whale algorithm that introduces a nonlinear convergence factor. When the value of the nonlinear convergence factor is larger and the convergence is slower, the searchable range of the energy consumption prediction model can be expanded, thereby improving the global searchability of the model. And as the number of iterations of the target energy consumption data increases, the nonlinear convergence factor tends to decrease in value. This will also make the prediction efficiency of the energy consumption prediction model faster. Compared with the prior art, the present application has higher prediction accuracy and stability, can effectively reduce the difficulty of chip energy consumption prediction, and provide theoretical data support for power allocation in power systems.

Claims

1. A method for predicting energy consumption of a power-specific chip, characterized in that: The method comprises: S1. Determine the target energy consumption data of the power-specific chip; S2. Based on the whale algorithm with the introduction of nonlinear convergence factor α, an energy consumption prediction model adapted to the architecture of the power-specific chip is constructed; the expression of the nonlinear convergence factor α is: α＝α int -(α int -α out )t 2 ； (1) In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model; S3. Process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results; The target energy consumption data is processed according to the energy consumption prediction model to obtain the corresponding energy consumption prediction result, including: S31. In the current iteration process, when the target energy consumption data is processed according to the energy consumption prediction model, the energy consumption of the power-specific chip is predicted by the following formula based on the optimal solution of the whale group to obtain the corresponding first energy consumption prediction value: In formula (2), f is the energy consumption prediction value, N is the number of samples collected for the target energy consumption data, and x i is the actual energy consumption value determined based on the target energy consumption data; x i ' is the preset average predicted energy consumption value; S32, when entering the next iteration, according to the maximum adaptive weight and the minimum adaptive weight of the model, the weight of the energy consumption prediction model is updated, and the location information of the whale group is reallocated, and the energy consumption prediction is re-performed based on the corresponding updated energy consumption prediction model to obtain the corresponding second energy consumption prediction value; S33: When it is determined that the first energy consumption prediction value matches the second energy consumption prediction value, output a corresponding energy consumption prediction result.

2. The method for predicting energy consumption of a power-specific chip according to claim 1, characterized in that: The step of determining target energy consumption data of the power-specific chip includes: S11, obtaining initial energy consumption data of the power-specific chip, and processing the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; the preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method; S12. Determine target energy consumption data based on the obtained preprocessed data.

3. The method for predicting energy consumption of a power-specific chip according to claim 1, characterized in that: The power-specific chip includes a three-layer architecture. The first layer architecture includes an AXI bus; at least one of a dynamic memory and a digital signal processor is provided on the AXI bus; The second layer architecture includes an AHB bus connected to the AXI bus; the AHB bus is provided with at least one of an instruction interface and an interrupt controller; The third layer architecture includes an APB bus connected to the AHB bus; the APB bus is provided with at least one of a timer, a serial port, a controller and an execution module.

4. The method for predicting energy consumption of a power-specific chip according to claim 1, characterized in that: Before step S31, the method further includes: Initialize model parameters and randomly distribute whales; where: The model parameters include at least one of the movement range of the whale group, the number of whales, the maximum number of data iterations of the model, the maximum adaptive weight, and the minimum adaptive weight; The distribution location information of each whale corresponds to a standard kernel function, which is used to correct the calculation error of the model to ensure the accuracy of the predicted structure.

5. The method for predicting energy consumption of a power-specific chip according to claim 1, characterized in that: Step S32, according to the maximum adaptability weight and the minimum adaptability weight of the model, adjusts the weight of the energy consumption prediction model, including: S321. According to the adaptive weight calculation method, the weight w of the energy consumption prediction model is adjusted by the following formula: in=in min +(in max +in min )×k μ ; (3) In the formula, w max is the preset maximum adaptive weight, w min is the preset minimum adaptive weight, k is the adjustment ratio of the adaptive weight, and μ is the relationship coefficient of the weight decrease.

6. A power-specific chip energy consumption prediction device, characterized in that: The device comprises a data acquisition module, a model building module and an energy consumption prediction module, wherein: A data acquisition module, used to determine the target energy consumption data of the power-specific chip; The model building module is used to build an energy consumption prediction model adapted to the architecture of the power-specific chip based on the whale algorithm that introduces a nonlinear convergence factor α; wherein the expression of the nonlinear convergence factor α is: α＝α int -(α int -α out )t 2 ； (1) In formula (1), α int is the preset initial value of α; α out is the preset final value of α; t is the estimated running time of the energy consumption prediction model; An energy consumption prediction module, used to process the target energy consumption data according to the energy consumption prediction model to obtain corresponding energy consumption prediction results; The target energy consumption data is processed according to the energy consumption prediction model to obtain the corresponding energy consumption prediction result, including: S31. In the current iteration process, when the target energy consumption data is processed according to the energy consumption prediction model, the energy consumption of the power-specific chip is predicted by the following formula based on the optimal solution of the whale group to obtain the corresponding first energy consumption prediction value: In formula (2), f is the energy consumption prediction value, N is the number of samples collected for the target energy consumption data, and x i is the actual energy consumption value determined based on the target energy consumption data; x i ' is the preset average predicted energy consumption value; S32, when entering the next iteration, according to the maximum adaptive weight and the minimum adaptive weight of the model, the weight of the energy consumption prediction model is updated, and the location information of the whale group is reallocated, and the energy consumption prediction is re-performed based on the corresponding updated energy consumption prediction model to obtain the corresponding second energy consumption prediction value; S33: When it is determined that the first energy consumption prediction value matches the second energy consumption prediction value, output a corresponding energy consumption prediction result.

7. The device according to claim 6, characterized in that The data acquisition module is also used to acquire the initial energy consumption data of the power-specific chip, and process the initial energy consumption data according to a preset preprocessing method to obtain corresponding preprocessing data; The preprocessing method includes at least one of a vertical and horizontal comparison method, a unified processing method, and a noise reduction processing method; based on the obtained preprocessing data, the target energy consumption data is determined.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When a processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images