Flexible power supply MCU dynamic frequency modulation control method and device

Abstract

The present invention relates to a flexible power supply MCU dynamic frequency modulation control method and device, comprising the following steps: determining a mode switching instruction of the load device based on a target working mode and a current working mode; extracting timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector; performing frequency band decomposition and prediction on the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain a load power consumption frequency band feature; performing a working frequency range mapping on the flexible power supply MCU based on the load power consumption frequency band feature to obtain a dynamic frequency adjustment sequence; inputting the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode, solving the technical problem that due to the huge differences in load characteristics under different application scenarios, the fixed-mode frequency adjustment mechanism cannot accurately match the actual requirements, resulting in low energy utilization efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of microcontrollers, and particularly to a dynamic frequency modulation control method and device for a flexible power supply MCU. Background Art

[0002] In the development of modern electronic devices, with the continuous enrichment and complication of functions, the performance requirements for the microcontroller unit (flexible power supply MCU) are getting higher and higher, and at the same time, the power consumption problem has become particularly prominent. Especially in battery-powered application scenarios such as mobile devices and Internet of Things sensors, how to effectively manage the working frequency of the flexible power supply MCU to achieve energy-saving purposes has become an important research direction. Traditional methods often adopt fixed frequency settings or simple dynamic voltage adjustment strategies to reduce power consumption, but these methods lack flexibility in dealing with load changes and cannot achieve the optimal energy efficiency.

[0003] The frequency modulation control schemes for flexible power supply MCUs in the prior art mostly rely on preset working modes and static power consumption models. This method is difficult to adapt to the complex load conditions and changes in the working environment in actual applications. In addition, due to the huge differences in load characteristics in different application scenarios, the fixed-mode frequency adjustment mechanism cannot accurately match the actual requirements, resulting in low energy utilization efficiency. Especially when dealing with sudden high-load tasks, the static adjustment strategy is prone to cause slow system response and affect the user experience. Therefore, it is particularly important to develop a technology that can flexibly adjust the working frequency of the flexible power supply MCU according to the real-time load situation.

[0004] In response to the above problems, the "dynamic frequency modulation control for flexible power supply MCU" method aims to achieve more refined power consumption management by intelligently analyzing the target working mode and current working mode of the load device, combined with advanced algorithm prediction and real-time adjustment technologies. This method not only considers immediate power consumption optimization but also focuses on long-term energy use efficiency improvement, attempting to overcome problems such as untimely response and poor adaptability in traditional frequency modulation control strategies. By extracting the timing characteristics of the mode switching instruction and based on this, accurately predicting and decomposing the real-time power consumption of the load device, finally realizing the dynamic adjustment of the working frequency of the flexible power supply MCU, thereby significantly improving the overall energy efficiency ratio of the system and meeting the requirements of diverse application scenarios. Summary of the Invention

[0005] The main object of the present invention is to provide a dynamic frequency modulation control method and device for a flexible power supply MCU, which solves the technical problem that due to the huge differences in load characteristics in different application scenarios, the fixed-mode frequency adjustment mechanism cannot accurately match the actual requirements, resulting in low energy utilization efficiency.

[0006] To achieve the above object, the present invention provides a flexible power supply MCU dynamic frequency modulation control method, including the following steps:

[0007] Read the target working mode and the current working mode of the preset load device, and determine the mode switching instruction of the load device based on the target working mode and the current working mode; extract the timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector; based on the mode instruction timing feature vector, perform frequency band decomposition and prediction on the real-time power consumption of the load device to obtain load power consumption frequency band characteristics; map the working frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence; input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode.

[0008] Further, determining the mode switching instruction of the load device based on the target working mode and the current working mode includes:

[0009] Perform state quantization encoding on the target working mode and the current working mode of the load device to obtain a target state code and a current state code, and perform a difference operation on the target state code and the current state code to obtain a state difference value;

[0010] Based on the state difference value, index the mode switching instruction set of the load device to obtain a mode switching instruction subset, and perform timing sorting on the mode switching instruction subset to obtain a sorted mode switching instruction sequence;

[0011] Perform instruction encoding compression on the sorted mode switching instruction sequence to obtain the mode switching instruction of the load device.

[0012] Further, extracting the timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector includes:

[0013] Perform time-domain segmented sampling on the mode switching instruction to obtain an instruction timing sampling sequence, and perform wavelet transform decomposition on the instruction timing sampling sequence to obtain multi-scale timing feature components;

[0014] Perform timing correlation analysis on the mode switching instruction based on the multi-scale timing feature components to obtain an instruction timing correlation matrix, and perform singular value decomposition on the instruction timing correlation matrix to obtain an instruction feature singular value sequence;

[0015] Perform feature dimensionality reduction mapping based on the singular value sequence of the instruction features to obtain a dimensionality-reduced feature subspace, and perform non-linear feature fusion on the dimensionality-reduced feature subspace to obtain the pattern instruction time series feature vector, where the pattern instruction time series feature vector includes an instruction energy consumption feature component, an instruction switching delay feature component, and an instruction execution cycle feature component.

[0016] Further, performing frequency band decomposition and prediction on the real-time power consumption of the load device based on the pattern instruction time series feature vector to obtain load power consumption frequency band features, including:

[0017] Perform multi-dimensional spectrum decomposition on the pattern instruction time series feature vector to obtain a power consumption frequency response feature set, and perform dynamic threshold segmentation on the power consumption frequency response feature set to obtain a power consumption frequency band boundary sequence;

[0018] Perform energy flow analysis on the power consumption frequency band boundary sequence through an adaptive filter to obtain a frequency band energy transfer matrix, and perform non-linear feature extraction on the frequency band energy transfer matrix to obtain a power consumption dynamic feature map;

[0019] Perform power consumption trend prediction on the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence, and perform time-frequency joint analysis on the power consumption prediction sequence to obtain a power consumption fluctuation feature set;

[0020] Perform multi-level decomposition on the power consumption fluctuation feature set to obtain a power consumption feature hierarchy tree, and perform frequency domain reconstruction on the power consumption feature hierarchy tree to obtain a frequency band power consumption mapping network;

[0021] Optimize and reorganize the frequency band power consumption mapping network through a dynamic planner to obtain a power consumption frequency band feature vector, and perform normalization processing on the power consumption frequency band feature vector to obtain load power consumption frequency band features; where the load power consumption frequency band features include frequency band power consumption feature values, frequency band stability data, and frequency band switching thresholds.

[0022] Further, performing power consumption trend prediction on the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence, including:

[0023] Perform feature hierarchical analysis on the power consumption dynamic feature map to obtain a multi-level power consumption feature sequence, and perform time domain expansion on the multi-level power consumption feature sequence to obtain a power consumption time series feature matrix, where the power consumption time series feature matrix includes a power consumption change trend, a power consumption mutation feature, and a power consumption steady state interval;

[0024] Perform key feature extraction on the power consumption time series feature matrix through a dynamic feature selector to obtain a power consumption key feature set, and perform time series correlation analysis on the power consumption key feature set to obtain a power consumption feature association network;

[0025] Based on the power consumption feature correlation network, perform power consumption evolution path analysis on the load device to obtain a power consumption evolution feature graph, and perform dynamic segmentation processing on the power consumption evolution feature graph to obtain a power consumption prediction feature group;

[0026] Perform a non-linear mapping transformation on the power consumption prediction feature group to obtain a power consumption prediction parameter matrix, and perform feature reconstruction on the power consumption prediction parameter matrix through a multi-dimensional feature fusion device to obtain a power consumption prediction vector;

[0027] Based on the power consumption prediction vector, perform time series expansion reconstruction to obtain a power consumption prediction time series table, and perform dynamic calibration on the power consumption prediction time series table to obtain a power consumption prediction sequence, where the power consumption prediction sequence includes a power consumption prediction trajectory and a power consumption prediction interval.

[0028] Further, the performing power consumption evolution path analysis on the load device based on the power consumption feature correlation network to obtain a power consumption evolution feature graph includes:

[0029] Perform topological structure analysis on the power consumption feature correlation network to obtain a power consumption feature transfer path set, and perform weight assignment on the power consumption feature transfer path set to obtain a power consumption path weight matrix;

[0030] Perform path optimization on the power consumption path weight matrix through a multi-dimensional path analyzer to obtain a key power consumption evolution path sequence, and perform time series expansion on the key power consumption evolution path sequence to obtain a power consumption evolution time series graph;

[0031] Perform feature propagation analysis based on the power consumption evolution time series graph to obtain a power consumption feature propagation network, and perform dynamic feature extraction on the power consumption feature propagation network to obtain a power consumption evolution feature set;

[0032] Perform multi-level reconstruction on the power consumption evolution feature set to obtain a power consumption evolution topological structure, and perform dynamic mapping on the power consumption evolution topological structure through a feature fusion processor to obtain a power consumption evolution feature graph, where the power consumption evolution feature graph includes a power consumption evolution trajectory, an evolution bifurcation point, and an evolution stable region.

[0033] Further, the mapping the operating frequency range of the flexible power supply MCU based on the load power consumption frequency band feature to obtain a dynamic frequency adjustment sequence includes:

[0034] Perform frequency domain deconstruction analysis on the load power consumption frequency band feature to obtain a frequency response feature matrix, and perform dynamic segmentation processing on the frequency response feature matrix to obtain a frequency allocation weight vector;

[0035] The frequency boundary constraint analysis is performed on the frequency allocation weight vector through a frequency mapping parser to obtain a frequency adjustment boundary set, and the frequency adjustment boundary set is dynamically optimized and reconstructed to obtain a frequency adjustment feature network;

[0036] Based on the frequency adjustment feature network, the working frequency space mapping of the flexible power supply MCU is performed to obtain a frequency mapping topology graph, and the timing characteristics of the frequency mapping topology graph are extracted to obtain a frequency adjustment feature sequence, where the frequency adjustment feature sequence includes a frequency adjustment step size, a frequency switching timing, and a frequency holding period;

[0037] The multi-dimensional dynamic reconstruction is performed on the frequency adjustment feature sequence to obtain a frequency adjustment control matrix, and the timing expansion of the frequency adjustment control matrix is performed through a frequency sequence optimizer to obtain a dynamic frequency adjustment sequence; where the dynamic frequency adjustment sequence includes a frequency adjustment trajectory, a frequency switching point, and a frequency stable interval.

[0038] The present invention also provides a dynamic frequency modulation control device for a flexible power supply MCU, including:

[0039] A reading module, configured to read a target working mode and a current working mode of a preset load device, and determine a mode switching instruction of the load device based on the target working mode and the current working mode; an extraction module, configured to extract timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector; a prediction module, configured to perform frequency band decomposition and prediction on the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain a load power consumption frequency band feature; a mapping module, configured to perform a working frequency range mapping on the flexible power supply MCU based on the load power consumption frequency band feature to obtain a dynamic frequency adjustment sequence; a control module, configured to input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode.

[0040] The present invention also provides a computer device, including a memory and a processor, where a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method described in any one of the above are implemented.

[0041] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0042] The flexible power supply MCU dynamic frequency modulation control method provided by the present invention includes the following steps: reading the target working mode and the current working mode of a preset load device, and determining a mode switching instruction for the load device based on the target working mode and the current working mode; extracting the timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector; decomposing and predicting the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain load power consumption frequency band characteristics; mapping the working frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence; inputting the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode, solving the technical problem that due to the huge differences in load characteristics under different application scenarios, the fixed-mode frequency adjustment mechanism cannot accurately match the actual requirements, resulting in low energy utilization efficiency, and realizing the method of inputting the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode, significantly improving the response speed of the system and the adaptability to external changes. It enables the system to quickly adapt to different working environments and task requirements while maintaining high efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic diagram of the steps of the flexible power supply MCU dynamic frequency modulation control method in an embodiment of the present invention;

[0044] Figure 2 is a structural block diagram of the flexible power supply MCU dynamic frequency modulation control device in an embodiment of the present invention;

[0045] Figure 3 is a structural schematic block diagram of a computer device in an embodiment of the present invention.

[0046] The implementation, functional features, and advantages of the objectives of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] As Figure 1 shown, Figure 1 is a schematic diagram of the steps of a flexible power supply MCU dynamic frequency modulation control method in an embodiment of the present invention;

[0049] An embodiment of the present invention provides a flexible power supply MCU dynamic frequency modulation control method, including the following steps:

[0050] Step S1: Read the target working mode and the current working mode of the preset load device, and determine the mode switching instruction of the load device based on the target working mode and the current working mode.

[0051] Specifically, in the above step, "reading the target working mode and the current working mode of the preset load device, and determining the mode switching instruction of the load device based on the target working mode and the current working mode" is the starting point of the entire flexible power supply MCU dynamic frequency modulation control method. Its core lies in generating accurate mode switching instructions by clarifying the working state of the load device. Specifically, this step first needs to obtain the target working mode and the current working mode of the load device from the system. The target working mode refers to the working state that the load device expects to reach, while the current working mode is the state in which the load device is actually operating currently. The difference between these two modes directly determines whether a mode switch is required and how to switch. For example, in the application scenario of an Internet of Things sensor, assume that the sensor is currently in the low-power standby mode (current working mode), but the system detects that a high-frequency data acquisition task needs to be executed (target working mode). At this time, a mode switching instruction needs to be generated to achieve the transition from the low-power standby mode to the high-frequency acquisition mode. To achieve this process, the system will read the relevant parameters of the target working mode and the current working mode through internal registers or external configuration files. These parameters may include information such as the working voltage, current demand, and task priority of the device. Subsequently, the system will analyze the specific switching requirements based on the difference between the two, such as whether it is necessary to increase the working frequency of the flexible power supply MCU, whether it is necessary to adjust the power supply power, etc. This analysis process not only depends on the comparison of the characteristics of the two modes but also needs to combine the actual operating environment and task requirements of the load device to ensure that the generated mode switching instruction meets the task requirements and can optimize the energy usage efficiency to the greatest extent. For example, in the battery management scenario of a mobile device, if the current working mode is the energy-saving mode and the target working mode is the high-performance mode, then the mode switching instruction may instruct the flexible power supply MCU to gradually increase the working frequency while coordinating the power management system to provide higher instantaneous power support for the device. In this way, not only can the smooth transition of device performance be ensured, but also the problems of energy waste or system instability caused by sudden power changes can be avoided. In short, through the accurate reading and comparative analysis of the target working mode and the current working mode, this step can generate scientific and reasonable mode switching instructions, laying a foundation for subsequent timing feature extraction and dynamic frequency adjustment, and also demonstrating strong flexibility and adaptability in practical applications.

[0052] Step S2: Extract the timing features of the mode switching instruction to obtain the mode instruction timing feature vector.

[0053] Specifically, in the above steps, "extracting the timing characteristics of the mode switching instruction to obtain the mode instruction timing feature vector" is a key link in the flexible power supply MCU dynamic frequency modulation control method. Its purpose is to deeply analyze the timing characteristics of the mode switching instruction and extract the feature information that can reflect the dynamic change law of the instruction. Specifically, this process first needs to regard the mode switching instruction as a time series signal, which contains a series of time-related operation information during the process of switching from the current working mode to the target working mode, such as the switching time point, the amplitude of frequency adjustment, and the trend of power change. By mathematically modeling and feature extraction of this information, a vector that can comprehensively describe the timing characteristics of the mode switching instruction can be generated, that is, the mode instruction timing feature vector. To achieve this goal, the system usually adopts relevant technologies in the field of signal processing, such as Fourier transform, wavelet transform, or autoregressive model, etc., to decompose and extract the timing characteristics in the mode switching instruction. For example, in the application scenario of Internet of Things sensors, assume that the load device needs to switch from the low-power standby mode to the high-frequency data acquisition mode. Then the mode switching instruction may contain multiple staged operation steps, such as gradually increasing the working frequency of the flexible power supply MCU, increasing the power output of the power supply, and starting the data acquisition module. These operations have a certain sequence and duration on the time axis. Through the timing feature extraction technology, the time distribution, frequency change rate, and power demand fluctuation of these operations can be quantified into specific numerical features and organized into a vector form. Taking the battery management of mobile devices as an example, when the device switches from the energy-saving mode to the high-performance mode, the mode switching instruction may involve frequency adjustment and power distribution operations at multiple time points. If the original instruction data is directly used, the efficiency of subsequent analysis and prediction may be low due to the complexity of the information. However, by extracting the timing characteristics of these instructions, the complex timing information can be compressed into a compact mode instruction timing feature vector, thus simplifying the subsequent processing flow. For example, the extracted feature vector may contain key information such as the average rate of frequency adjustment, the change slope of power demand, and the total duration of instruction execution. These information can not only reflect the overall trend of mode switching but also provide important reference for subsequent load power consumption prediction and dynamic frequency adjustment. In short, by extracting the timing characteristics of the mode switching instruction, the system can transform the complex timing information into a concise and representative mode instruction timing feature vector, which not only improves the efficiency of data processing but also lays a solid foundation for subsequent power consumption management and frequency optimization. This technology shows strong adaptability and practicality in actual applications, especially in scenarios that require frequent mode switching, and can significantly improve the response speed and energy utilization efficiency of the system.

[0054] Step S3, perform frequency band decomposition and prediction on the real-time power consumption of the load device based on the timing feature vector of the pattern instruction, to obtain the load power consumption frequency band feature.

[0055] Specifically, in the above steps, "decomposing and predicting the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain the load power consumption frequency band characteristics" is an important part of the flexible power supply MCU dynamic frequency modulation control method. The core is to analyze the mode instruction timing feature vector, combine the actual operation status of the load device, decompose the power consumption frequency band and predict the future trend, so as to extract the frequency band characteristics that can reflect the power consumption change law. Specifically, this process first needs to use the time distribution, frequency adjustment rate and power demand change information contained in the mode instruction timing feature vector to mathematically model and decompose the real-time power consumption signal of the load device. By decomposing the power consumption signal into components of different frequency bands, the main sources of power consumption fluctuations and their changing laws can be more clearly identified. In order to achieve this goal, the system usually uses frequency domain analysis technology, such as fast Fourier transform (FFT) or wavelet transform, to convert the real-time power consumption signal from the time domain to the frequency domain, and then separate the high-frequency and low-frequency components. For example, in the application scenario of IoT sensors, when the load device switches from low-power standby mode to high-frequency data acquisition mode, the power consumption signal may show obvious high-frequency fluctuations, which is caused by the increase in the operating frequency of the flexible power supply MCU and the startup of the data acquisition module. By performing frequency band decomposition on these power consumption signals, it can be identified that the high-frequency part mainly corresponds to the change in instantaneous power demand, while the low-frequency part reflects the long-term trend of the overall power consumption of the device. Subsequently, based on these decomposition results and the time characteristics in the mode instruction timing feature vector, the system predicts future power consumption changes and obtains the load power consumption frequency band characteristics. Taking the battery management of mobile devices as an example, when the device switches from energy-saving mode to high-performance mode, the power consumption signal may experience a process from stable to violent fluctuations. If the original power consumption data is directly used for analysis, it may be difficult to capture the subtle changes in it. However, by performing frequency band decomposition on the power consumption signal, the complex power consumption fluctuations can be decomposed into multiple frequency band components, and combined with information such as the frequency adjustment rate and power demand change slope in the mode instruction timing feature vector, the power consumption trend in the future period can be predicted. For example, the prediction results show that the high-frequency component may increase rapidly in the next few milliseconds, which indicates that the device is about to enter a short-term task processing stage with high power consumption; while the low-frequency component may remain relatively stable, reflecting that the long-term change trend of the overall power consumption of the device is relatively gentle. The extraction and prediction of this frequency band feature not only helps to grasp the power consumption dynamics of the device more accurately, but also provides an important reference for the subsequent mapping of the operating frequency range of the flexible power supply MCU. In short, through in-depth analysis of the mode instruction timing feature vector, combined with the frequency band decomposition and prediction of the real-time power consumption signal, the system can extract the load power consumption frequency band characteristics, so as to fully understand the power consumption change law of the device.This method demonstrates strong adaptability and practicality in actual applications. Especially in scenarios that require frequent pattern switching, it can significantly improve the accuracy of the system's energy efficiency management and response speed, laying a solid foundation for optimizing energy usage efficiency.

[0056] Step S4: Based on the load power consumption frequency band characteristics, map the operating frequency range of the flexible power supply MCU to obtain a dynamic frequency adjustment sequence.

[0057] Specifically, in the above steps, "mapping the working frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence" is a key link in the dynamic frequency modulation control method of the flexible power supply MCU. Its goal is to combine the load power consumption frequency band characteristics extracted in the previous step with the working frequency adjustment strategy of the flexible power supply MCU, so as to generate a dynamic frequency adjustment sequence that can adapt to real-time power consumption changes. Specifically, this process first needs to utilize the high-frequency and low-frequency component information contained in the load power consumption frequency band characteristics to analyze the impact of the power consumption requirements corresponding to different frequency bands on the working frequency of the flexible power supply MCU. For example, the high-frequency component usually reflects the change of instantaneous power demand, and may require a rapid increase in the working frequency of the flexible power supply MCU to meet the needs of burst tasks; while the low-frequency component reflects the long-term trend of the overall power consumption of the device, and the working frequency range of the flexible power supply MCU can be optimized accordingly to achieve more efficient energy management. To achieve this goal, the system will determine the frequency adjustment range suitable for the current working mode according to the specific parameters in the load power consumption frequency band characteristics, such as the high-frequency fluctuation amplitude and the low-frequency change slope, in combination with the hardware performance limitations and power consumption models of the flexible power supply MCU. For example, in the application scenario of Internet of Things sensors, when the device switches from the low-power standby mode to the high-frequency data acquisition mode, the load power consumption frequency band characteristics may show a significant increase in the high-frequency component, indicating that the device requires higher instantaneous computing power. At this time, the system will map the high-frequency component to a higher working frequency range of the flexible power supply MCU and generate a dynamic frequency adjustment sequence that gradually increases the frequency to ensure that the flexible power supply MCU can avoid energy waste while ensuring performance. At the same time, the analysis results of the low-frequency component are used to determine the working frequency fallback range of the flexible power supply MCU after completing short-term tasks, so as to achieve long-term energy optimization. Taking the battery management of mobile devices as an example, when the device switches from the energy-saving mode to the high-performance mode, the load power consumption frequency band characteristics may predict that the high-frequency component will increase rapidly in the next few milliseconds, while the low-frequency component remains relatively stable. Based on these characteristics, the system will map the high-frequency component to a higher working frequency range of the flexible power supply MCU and generate a dynamic frequency adjustment sequence that first rapidly increases the frequency and then gradually falls back. For example, when the user starts a graphics-intensive application, the working frequency of the flexible power supply MCU may increase from 500 MHz to 1.5 GHz in a short time, and then gradually decrease to 1.2 GHz according to the task requirements to maintain stable performance output. This dynamic frequency adjustment sequence can not only meet the real-time performance requirements of the device, but also minimize unnecessary energy consumption. In short, by deeply analyzing the load power consumption frequency band characteristics and mapping them to the working frequency range of the flexible power supply MCU, the system can generate a scientific and reasonable dynamic frequency adjustment sequence, thereby achieving precise control of the working state of the flexible power supply MCU.This method demonstrates strong flexibility and adaptability in practical applications. Especially in scenarios that require frequent pattern switching, it can significantly improve the system's response speed and energy efficiency management level, providing important technical support for optimizing energy usage efficiency.

[0058] Step S5: Input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode.

[0059] Specifically, in the above steps, "inputting the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode" is the final execution link of the flexible power supply MCU dynamic frequency modulation control method. The core lies in achieving precise switching of the working mode of the load device by applying the generated dynamic frequency adjustment sequence to the flexible power supply MCU. Specifically, this process first requires inputting the dynamic frequency adjustment sequence as an instruction into the frequency control unit of the flexible power supply MCU. This sequence contains information on the working frequency adjustment of the flexible power supply MCU at different time points, such as the amplitude of frequency increase, duration, and the speed of decline. These information directly guide the flexible power supply MCU on how to adjust its internal clock frequency, thereby driving the load device to complete a smooth transition from the current working mode to the target working mode. To achieve this goal, the system will utilize the hardware support mechanism of the flexible power supply MCU, such as dynamic voltage and frequency scaling (DVFS) technology or clock management unit (CMU), to convert the dynamic frequency adjustment sequence into specific frequency adjustment operations. For example, in the application scenario of Internet of Things sensors, when the device needs to switch from the low-power standby mode to the high-frequency data acquisition mode, the dynamic frequency adjustment sequence may instruct the flexible power supply MCU to quickly increase the working frequency from 200 MHz to 800 MHz to meet the startup requirements of the data acquisition module, and then gradually decline to 500 MHz to maintain stable acquisition performance. This frequency adjustment process can not only ensure the efficient operation of the device during the switching process but also avoid energy waste or system instability problems caused by sudden frequency changes. Taking the battery management of mobile devices as an example, when the user switches from the energy-saving mode to the high-performance mode, the dynamic frequency adjustment sequence may be input into the flexible power supply MCU to guide it to gradually increase the working frequency. For example, when the user starts a video playback application, the flexible power supply MCU may, according to the dynamic frequency adjustment sequence, quickly increase the working frequency from 500 MHz to 1.5 GHz to load the video content, and then gradually decrease to 1.2 GHz during the playback to balance performance and power consumption. This dynamic frequency adjustment strategy can not only meet the user's demand for high performance but also maximize the battery life of the device. In addition, since the dynamic frequency adjustment sequence is generated based on the load power consumption frequency band characteristics, its adjustment process has high pertinence and adaptability, and can achieve optimal energy management in a complex task environment. In short, by inputting the dynamic frequency adjustment sequence into the flexible power supply MCU, the system can effectively control the load device to switch from the current working mode to the target working mode, thereby achieving precise management of the device performance and power consumption. This method demonstrates strong practicality and flexibility in practical applications, especially in scenarios that require frequent mode switching, which can significantly improve the system response speed and energy utilization efficiency, providing important technical support for optimizing the overall system performance.

[0060] In a specific embodiment, determining the mode switching instruction of the load device based on the target working mode and the current working mode includes:

[0061] Performing state quantization encoding on the target working mode and the current working mode of the load device to obtain a target state code and a current state code, and performing a difference operation on the target state code and the current state code to obtain a state difference value;

[0062] Based on the state difference value, indexing the mode switching instruction set of the load device to obtain a subset of mode switching instructions, and performing timing sorting on the subset of mode switching instructions to obtain a sorted sequence of mode switching instructions;

[0063] Performing instruction encoding compression on the sorted sequence of mode switching instructions to obtain the mode switching instruction of the load device.

[0064] Specifically, in the above steps, "determining the mode switching instruction of the load device based on the target working mode and the current working mode" is a key link in the flexible power supply MCU dynamic frequency modulation control method. Its core lies in the process of state quantization encoding, differential operation, indexing the mode switching instruction set, and finally generating the mode switching instruction for the target working mode and the current working mode, so as to accurately describe the mode switching requirements of the load device. This process not only requires in-depth analysis of the working mode of the device, but also needs to combine specific algorithms and technical means to convert complex state information into an executable instruction sequence, thus laying a foundation for subsequent timing feature extraction and dynamic frequency adjustment. First, in order to perform state quantization encoding on the target working mode and the current working mode, the system maps the specific operating parameters of these two modes into a set of numerical state codes, namely the target state code and the current state code. These state codes usually contain key information related to the working mode, such as the working voltage of the device, current demand, task priority, and performance indicators. For example, in the application scenario of an Internet of Things sensor, assume that the current working mode is the low-power standby mode, and the target working mode is the high-frequency data acquisition mode. Then the current state code may contain lower working voltage and current demand parameters, while the target state code will have higher voltage and current demand values. Through this quantization encoding method, complex mode information can be converted into a concise and easy-to-process numerical form, thus providing a basis for subsequent differential operations. Next, the system performs a differential operation on the target state code and the current state code to obtain a state difference value. The core of this process is to calculate the difference between the two modes to quantify the demand intensity and direction of mode switching. For example, in the battery management scenario of a mobile device, if the current state code indicates that the device is in the energy-saving mode, and the target state code corresponds to the high-performance mode, then the state difference value may reflect specific requirements such as increasing the working frequency of the flexible power supply MCU, increasing the power output of the power supply, and starting more functional modules. This differential operation can not only intuitively reflect the difference between the two modes, but also provide a clear direction for subsequent mode switching instruction generation. Based on the state difference value, the system further indexes the mode switching instruction set of the load device to obtain a mode switching instruction subset. The mode switching instruction set is essentially a predefined database or rule library, which contains various possible mode switching requirements and their corresponding instruction sequences. For example, in the application of an Internet of Things sensor, the mode switching instruction set may contain operation instructions such as frequency adjustment, power distribution, and module startup required to switch from the low-power standby mode to the high-frequency data acquisition mode. Through the indexing of the state difference value, the system can screen out the instruction subset that best matches the current mode switching requirements from the huge instruction set.For example, if the state difference value indicates that it is necessary to quickly increase the operating frequency of the flexible power supply MCU and increase the power output, then the selected instruction subset may include a series of operation steps to gradually increase the frequency and power. Subsequently, the system will perform a timing sort on the mode switching instruction subset to obtain the sorted mode switching instruction sequence. The key point of this process is to reasonably arrange the time of each operation step in the instruction subset according to the actual requirements of mode switching to ensure the smoothness and efficiency of the mode switching process. For example, in the battery management scenario of a mobile device, when the device switches from the energy-saving mode to the high-performance mode, the mode switching instruction subset may include multiple operation steps, such as gradually increasing the operating frequency of the flexible power supply MCU, increasing the power output, and starting the data processing module. By performing a timing sort on these operation steps, it can be ensured that they are executed in a reasonable order and time interval. For example, first increase the operating frequency of the flexible power supply MCU to meet the computing requirements, then increase the power output to support high-frequency operation, and finally start the data processing module to complete task loading. This timing sort can not only avoid system instability problems caused by operation conflicts but also significantly improve the overall efficiency of mode switching. Finally, the system will perform instruction encoding compression on the sorted mode switching instruction sequence to obtain the mode switching instruction for the load device. The core of this process is to compress the complex instruction sequence into a compact and efficient instruction form through encoding technology, thereby reducing the occupation of storage space and improving the efficiency of instruction transmission and execution. For example, in the application of Internet of Things sensors, the sorted mode switching instruction sequence may include multiple phased operation steps, such as gradually increasing the frequency, adjusting the power, and starting the module. By encoding and compressing these instructions, they can be converted into a concise binary code or a data packet in a specific format for quick transmission and parsing in the system. For example, the compressed mode switching instruction may include key information such as the amplitude of frequency adjustment, the slope of power change, and the time interval of operation. These information can not only completely describe the requirements of mode switching but also provide important reference for subsequent timing feature extraction and dynamic frequency adjustment. For example, in the battery management scenario of a mobile device, assume that the device is currently in the energy-saving mode and the user suddenly starts a graphics-intensive application. At this time, the system needs to generate mode switching instructions to achieve the transition from the energy-saving mode to the high-performance mode. First, the system will perform state quantization encoding on the energy-saving mode and the high-performance mode to obtain the current state code and the target state code respectively. Then, by performing a difference operation on the two state codes, the system may obtain requirements such as increasing the operating frequency of the flexible power supply MCU to 1.5 GHz, increasing the power output to 3 W, and starting the GPU acceleration module. Based on these difference values, the system will index the mode switching instruction set and select an instruction subset that includes frequency adjustment, power distribution, and module start operations.Subsequently, the system will perform a timing sort on these instructions to ensure that the frequency is increased first, then the power is increased, and finally the module is started. Finally, the system will encode and compress the sorted instruction sequence to generate a compact mode switching instruction and input it into the flexible power supply MCU to guide the device to complete the mode switching. In summary, through state quantization encoding, differential operation, indexing the mode switching instruction set, timing sorting, and instruction encoding compression of the target working mode and the current working mode, the system can generate scientific and reasonable mode switching instructions, thereby achieving precise control of the working mode of the load device. This method demonstrates strong flexibility and adaptability in practical applications. Especially in scenarios where frequent mode switching is required, it can significantly improve the system's response speed and energy utilization efficiency, providing important technical support for optimizing the overall system performance.

[0065] In a specific embodiment, the mode switching instruction is sampled in the time domain in segments to obtain an instruction timing sampling sequence, and the instruction timing sampling sequence is decomposed by wavelet transform to obtain multi-scale timing feature components;

[0066] Based on the multi-scale timing feature components, timing correlation analysis is performed on the mode switching instruction to obtain an instruction timing correlation matrix, and the instruction timing correlation matrix is decomposed by singular value decomposition to obtain an instruction feature singular value sequence;

[0067] Based on the instruction feature singular value sequence, feature dimensionality reduction mapping is performed to obtain a dimensionality reduction feature subspace, and non-linear feature fusion is performed on the dimensionality reduction feature subspace to obtain the mode instruction timing feature vector, where the mode instruction timing feature vector includes an instruction energy consumption feature component, an instruction switching delay feature component, and an instruction execution cycle feature component.

[0068] Specifically, in the above steps, "extracting the timing feature of the mode switching instruction to obtain the mode instruction timing feature vector" is one of the core links in the dynamic frequency modulation control method of the flexible power supply MCU. Its purpose is to extract the feature information that can reflect the time series characteristics during the instruction execution process through in-depth analysis of the mode switching instruction. Specifically, this process includes multiple steps such as time-domain segmented sampling of the mode switching instruction, wavelet transform decomposition, timing correlation analysis, singular value decomposition, and feature dimensionality reduction mapping and non-linear feature fusion, so as to finally generate a vector that comprehensively describes the timing characteristics of the mode switching instruction. First, the system performs time-domain segmented sampling on the mode switching instruction to obtain the instruction timing sampling sequence. The key to this process is to discretize the continuous mode switching instruction according to the time interval, thus forming a series of instruction samples at a series of time points. For example, in the application scenario of Internet of Things sensors, when the device switches from the low-power standby mode to the high-frequency data acquisition mode, the mode switching instruction may include multiple operation steps, such as gradually increasing the working frequency of the flexible power supply MCU, adjusting the power output, and starting the data acquisition module. By performing time-domain segmented sampling on these instructions, they can be transformed into a series of discrete time series signals, and each time point corresponds to a specific operation state or parameter value. This way of time-domain segmented sampling not only simplifies the subsequent analysis process but also provides a basis for the extraction of multi-scale timing feature components. Then, the system performs wavelet transform decomposition on the instruction timing sampling sequence to obtain multi-scale timing feature components, including high-frequency instruction feature components, medium-frequency instruction feature components, and low-frequency instruction feature components. Wavelet transform is an effective signal processing technology that can decompose the original signal into a combination of different frequency components, thus revealing the multi-level structure inside the signal. For example, in the battery management scenario of mobile devices, assuming that the mode switching instruction includes a series of operation steps for switching from the energy-saving mode to the high-performance mode, through wavelet transform decomposition, it can be identified which parts of these operation steps correspond to instantaneous high-frequency changes (such as quickly increasing the working frequency of the flexible power supply MCU), which parts show relatively stable medium-frequency fluctuations (such as gradually increasing the power output), and which parts reflect the low-frequency components of the long-term trend (such as starting and stably operating the data processing module). The extraction of such multi-scale feature components not only helps to deeply understand the dynamic characteristics of the instruction but also provides a rich information source for the subsequent timing correlation analysis. Based on the multi-scale timing feature components, the system further performs timing correlation analysis on the mode switching instruction to obtain the instruction timing correlation matrix, and performs singular value decomposition on this matrix to obtain the instruction feature singular value sequence. Timing correlation analysis aims to explore the time dependence relationship, energy consumption correlation degree, and switching coupling degree between different operation steps during the instruction execution process.For example, in the scenario of Internet of Things sensors, the instruction timing correlation matrix may reflect the close connection between frequency adjustment and power distribution, as well as the energy demand fluctuations before and after the data acquisition module is started. By performing singular value decomposition on this matrix, complex correlation information can be compressed into a set of key singular value sequences. These singular values not only represent important features during the instruction execution process but also provide a basis for further feature dimensionality reduction mapping. Next, the system will perform feature dimensionality reduction mapping based on the instruction feature singular value sequence to obtain a reduced-dimensional feature subspace, and perform non-linear feature fusion on this subspace to generate a pattern instruction timing feature vector. The main goal of feature dimensionality reduction mapping is to reduce the data dimension while retaining the most important feature information, thereby improving the efficiency and accuracy of subsequent analysis. For example, in the battery management scenario of mobile devices, the instruction feature singular value sequence obtained after singular value decomposition may contain a large amount of redundant information. Through feature dimensionality reduction mapping, this information can be condensed into several key feature dimensions, such as the instruction energy consumption feature component, the instruction switching delay feature component, and the instruction execution cycle feature component. Subsequently, through non-linear feature fusion techniques such as principal component analysis (PCA) or independent component analysis (ICA), these feature dimensions can be integrated into a comprehensive pattern instruction timing feature vector, which can comprehensively describe the time series characteristics and internal correlations of pattern switching instructions. For example, in the battery management application scenario of mobile devices, assume that the user needs to switch from the energy-saving mode to the high-performance mode to run a graphics-intensive application. First, the system will perform time-domain segmented sampling on the pattern switching instruction to obtain a series of discrete instruction samples, and each sample represents the operation state at a certain moment. Then, these instruction samples are decomposed using wavelet transform to identify the high-frequency components (such as rapidly increasing the working frequency of the flexible power supply MCU), medium-frequency components (such as gradually increasing the power output), and low-frequency components (such as starting and stably operating the GPU acceleration module). Next, the system will perform time series correlation analysis on these multi-scale timing feature components, construct an instruction timing correlation matrix, and perform singular value decomposition on it to obtain the instruction feature singular value sequence. Finally, based on these singular value sequences, feature dimensionality reduction mapping is performed to generate a reduced-dimensional feature subspace, and it is converted into a pattern instruction timing feature vector through non-linear feature fusion technology. This vector not only contains the key energy consumption information, switching delay information, and execution cycle information during the instruction execution process but also provides an important reference basis for subsequent load power consumption prediction and dynamic frequency adjustment. In short, through a series of steps such as time-domain segmented sampling, wavelet transform decomposition, time series correlation analysis, singular value decomposition, feature dimensionality reduction mapping, and non-linear feature fusion of the pattern switching instruction, the system can generate a scientific and reasonable pattern instruction timing feature vector, thereby achieving an accurate description of the timing characteristics of the pattern switching instruction.This method demonstrates strong flexibility and adaptability in practical applications. Especially in scenarios that require frequent pattern switching, it can significantly improve the system's response speed and energy utilization efficiency, providing important technical support for optimizing the overall system performance.

[0069] In a specific embodiment, performing frequency band decomposition and prediction on the real-time power consumption of the load device based on the timing feature vector of the pattern instruction to obtain the load power consumption frequency band characteristics includes:

[0070] Performing multi-dimensional spectrum decomposition on the timing feature vector of the pattern instruction to obtain a power consumption frequency response feature set, and performing dynamic threshold segmentation on the power consumption frequency response feature set to obtain a power consumption frequency band boundary sequence;

[0071] Performing energy flow analysis on the power consumption frequency band boundary sequence through an adaptive filter to obtain a frequency band energy transfer matrix, and performing non-linear feature extraction on the frequency band energy transfer matrix to obtain a power consumption dynamic feature map;

[0072] Performing power consumption trend prediction on the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence, and performing time-frequency joint analysis on the power consumption prediction sequence to obtain a power consumption fluctuation feature set;

[0073] Performing multi-level decomposition on the power consumption fluctuation feature set to obtain a power consumption feature hierarchy tree, and performing frequency domain reconstruction on the power consumption feature hierarchy tree to obtain a frequency band power consumption mapping network;

[0074] Optimizing and reorganizing the frequency band power consumption mapping network through a dynamic planner to obtain a power consumption frequency band feature vector, and performing normalization processing on the power consumption frequency band feature vector to obtain the load power consumption frequency band characteristics; wherein, the load power consumption frequency band characteristics include frequency band power consumption characteristic values, frequency band stability data, and frequency band switching thresholds.

[0075] Specifically, in the above steps, "decomposing and predicting the real-time power consumption of the load device based on the timing feature vector of the pattern instruction to obtain the load power consumption frequency band feature" is one of the core steps in the flexible power supply MCU dynamic frequency modulation control method. Its purpose is to achieve precise decomposition and prediction of power consumption characteristics by deeply analyzing the timing feature vector of the pattern instruction and combining the real-time power consumption information of the load device. Specifically, this process includes multiple steps such as multi-dimensional spectrum decomposition, dynamic threshold segmentation, energy flow analysis, non-linear feature extraction, power consumption trend prediction, time-frequency joint analysis, multi-level decomposition, frequency domain reconstruction, and optimization and recombination, so as to finally generate a frequency band feature vector that comprehensively describes the load power consumption characteristics. First, the system performs multi-dimensional spectrum decomposition on the timing feature vector of the pattern instruction to obtain the power consumption frequency response feature set, and performs dynamic threshold segmentation on this feature set to obtain the power consumption frequency band boundary sequence. Multi-dimensional spectrum decomposition is a signal processing technology that can reveal the energy distribution of different frequency components by converting the original time series data into the frequency domain. For example, in the application scenario of Internet of Things sensors, assume that the load device switches from a low-power standby mode to a high-frequency data acquisition mode. The timing feature vector of the pattern instruction may contain multiple operation steps and their corresponding energy consumption information. Through multi-dimensional spectrum decomposition, it can be identified which parts of these operation steps correspond to high-frequency energy requirements (such as quickly increasing the working frequency of the flexible power supply MCU), which parts show medium-frequency fluctuations (such as adjusting the power supply output power), and which parts reflect low-frequency trends (such as starting and stabilizing the operation of the data acquisition module). Subsequently, using the dynamic threshold segmentation technology, the boundaries between different frequency bands can be determined according to the energy distribution to form the power consumption frequency band boundary sequence. This step not only helps to clarify the energy demand range of different frequency bands but also provides a basis for subsequent energy flow analysis. Then, the system performs energy flow analysis on the power consumption frequency band boundary sequence through an adaptive filter to obtain the frequency band energy transfer matrix, and performs non-linear feature extraction on this matrix to obtain the power consumption dynamic feature map. The adaptive filter can automatically adjust its parameters according to the characteristics of the input signal, thereby effectively filtering out noise and highlighting useful information. For example, in the battery management scenario of mobile devices, when the device switches from an energy-saving mode to a high-performance mode, the power consumption frequency band boundary sequence may show a trend of rapid increase in high-frequency band energy. Through the energy flow analysis of the adaptive filter, it can be quantified how these energies are transferred between different frequency bands and the frequency band energy transfer matrix can be constructed. Further, through non-linear feature extraction technology, key power consumption dynamic feature maps can be extracted from this matrix. These maps not only reflect the direction and intensity of energy transfer but also reveal the internal laws of power consumption changes. Based on the power consumption dynamic feature map, the system predicts the power consumption trend of the load device to obtain the power consumption prediction sequence, and performs time-frequency joint analysis on this sequence to obtain the power consumption fluctuation feature set.Power consumption trend prediction aims to infer the power consumption changes in the future period based on historical power consumption data, which is crucial for early energy management and scheduling. For example, in the application of Internet of Things sensors, assuming that the device is about to execute a series of short-term tasks, the power consumption dynamic feature map may show that the energy demand in the high-frequency band will increase significantly in the next few seconds and then gradually decline. By establishing an appropriate prediction model (such as ARIMA or neural network model), a power consumption prediction sequence can be generated, and combined with time-frequency joint analysis technology, further analyze the characteristics such as the periodicity, peaks, and valleys of power consumption fluctuations. These characteristics not only help to understand the overall trend of power consumption changes but also provide a basis for subsequent multi-level decomposition. Next, the system will perform multi-level decomposition on the power consumption fluctuation feature set to obtain a power consumption feature hierarchy tree, and perform frequency domain reconstruction on this tree to obtain a frequency band power consumption mapping network. Multi-level decomposition is an effective data analysis method. By decomposing a complex feature set into multiple hierarchical structures, the mutual relationships between different levels can be shown more clearly. For example, in the battery management scenario of mobile devices, the power consumption fluctuation feature set may contain fluctuation information on multiple time scales. Through multi-level decomposition, it can be organized into a hierarchical power consumption feature hierarchy tree, where each node represents the power consumption feature at a specific time scale. Subsequently, through frequency domain reconstruction technology, these features can be remapped back to the frequency domain to form a frequency band power consumption mapping network. This network not only contains the power consumption distribution of each frequency band but also reflects important information such as power consumption density gradient and frequency band switching cost. Finally, the system will optimize and reorganize the frequency band power consumption mapping network through a dynamic planner to obtain a power consumption frequency band feature vector, and perform normalization processing on this vector to obtain the load power consumption frequency band feature. The dynamic planner can effectively reorganize complex data structures while considering global optimality, thereby improving the overall performance of the system. For example, in the application scenario of Internet of Things sensors, the frequency band power consumption mapping network may show that the power consumption density of some frequency bands is relatively high, while that of other frequency bands is relatively low. Through the optimization and reorganization of the dynamic planner, an optimal frequency band allocation scheme that can meet the performance requirements and minimize energy consumption can be found, and then a power consumption frequency band feature vector is generated. Subsequently, through the normalization processing of this vector, the dimension difference between different frequency bands can be eliminated, ensuring that the final obtained load power consumption frequency band feature has a unified standard and comparability. For example, in the battery management application scenario of mobile devices, assume that the user needs to switch from the energy-saving mode to the high-performance mode to run a graphics-intensive application. First, the system will perform multi-dimensional spectrum decomposition on the mode instruction timing feature vector to identify the high-frequency band energy demand (such as quickly increasing the working frequency of the flexible power supply MCU), the mid-frequency band fluctuations (such as gradually increasing the power output), and the low-frequency band trend (such as starting and stably operating the GPU acceleration module). Then, the dynamic threshold segmentation technology is used to determine the boundaries between different frequency bands to form a power consumption frequency band boundary sequence.Next, energy flow analysis is performed through an adaptive filter to construct a frequency band energy transfer matrix, and a power consumption dynamic feature map is extracted therefrom. Based on these maps, the system can predict the power consumption changes over a period of time in the future, obtain a power consumption prediction sequence, and obtain a power consumption fluctuation feature set through time-frequency joint analysis. Subsequently, the power consumption fluctuation feature set is decomposed at multiple levels to form a power consumption feature hierarchy tree, and a frequency band power consumption mapping network is obtained through frequency domain reconstruction. Finally, the network is optimized and reorganized by a dynamic planner to generate a power consumption frequency band feature vector, and it is normalized to finally obtain the load power consumption frequency band feature. This feature not only includes the power consumption characteristics of each frequency band, but also reflects the stability and switching cost between frequency bands, providing an important reference basis for the subsequent flexible power supply MCU operating frequency adjustment. In summary, through a series of steps such as multi-dimensional spectrum decomposition, dynamic threshold segmentation, energy flow analysis, non-linear feature extraction, power consumption trend prediction, time-frequency joint analysis, multi-level decomposition, frequency domain reconstruction, and optimization and reorganization of the pattern instruction timing feature vector, the system can generate a scientific and reasonable load power consumption frequency band feature vector, thereby achieving an accurate description of the power consumption characteristics of the load device. This method demonstrates strong flexibility and adaptability in practical applications. Especially in scenarios that require frequent mode switching, it can significantly improve the system's response speed and energy utilization efficiency, providing important technical support for optimizing the overall system performance.

[0076] In a specific embodiment, the power consumption trend prediction of the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence includes:

[0077] Perform feature hierarchical analysis on the power consumption dynamic feature map to obtain a multi-level power consumption feature sequence, and perform time domain expansion on the multi-level power consumption feature sequence to obtain a power consumption time series feature matrix, where the power consumption time series feature matrix includes a power consumption change trend, a power consumption mutation feature, and a power consumption steady state interval;

[0078] Extract key features from the power consumption time series feature matrix through a dynamic feature selector to obtain a power consumption key feature set, and perform time series correlation analysis on the power consumption key feature set to obtain a power consumption feature association network;

[0079] Based on the power consumption feature association network, perform power consumption evolution path analysis on the load device to obtain a power consumption evolution feature map, and perform dynamic segmentation processing on the power consumption evolution feature map to obtain a power consumption prediction feature group;

[0080] Perform non-linear mapping transformation on the power consumption prediction feature group to obtain a power consumption prediction parameter matrix, and perform feature reconstruction on the power consumption prediction parameter matrix through a multi-dimensional feature fusion device to obtain a power consumption prediction vector;

[0081] Based on the power consumption prediction vector, perform time series expansion and reconstruction to obtain a power consumption prediction time series table, and perform dynamic calibration on the power consumption prediction time series table to obtain a power consumption prediction sequence, where the power consumption prediction sequence includes a power consumption prediction trajectory and a power consumption prediction interval.

[0082] Specifically, in the above steps, "performing power consumption trend prediction on the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence" is one of the core steps in the flexible power supply MCU dynamic frequency modulation control method. Its purpose is to achieve accurate prediction of future power consumption change trends through in-depth analysis of the power consumption dynamic feature map, combined with various algorithms and technical means. Specifically, this process includes multiple steps such as feature hierarchical analysis, time-domain expansion, key feature extraction, time-series correlation analysis, power consumption evolution path analysis, non-linear mapping transformation, and multi-dimensional feature fusion, so as to finally generate a prediction sequence that comprehensively describes future power consumption characteristics. First, the system performs feature hierarchical analysis on the power consumption dynamic feature map to obtain a multi-layer power consumption feature sequence, and performs time-domain expansion on this sequence to obtain a power consumption time-series feature matrix. Feature hierarchical analysis aims to decompose the complex power consumption dynamic feature map into multiple hierarchical structures, each layer representing power consumption characteristics within different time scales or frequency ranges. For example, in the application scenario of Internet of Things sensors, assuming that the load device switches from a low-power standby mode to a high-frequency data acquisition mode, the power consumption dynamic feature map may include high-frequency band energy requirements (such as rapidly increasing the working frequency of the flexible power supply MCU), medium-frequency band fluctuations (such as adjusting the power supply output power), and low-frequency band trends (such as starting and stabilizing the operation of the data acquisition module). Through feature hierarchical analysis, this information can be organized into a multi-layer structure, with each layer corresponding to a specific time scale or frequency range. Subsequently, by performing time-domain expansion on the multi-layer power consumption feature sequence, it can be converted into a two-dimensional power consumption time-series feature matrix, which contains important information such as power consumption change trends, power consumption mutation characteristics, and power consumption steady-state intervals. This time-domain expansion not only simplifies the subsequent data processing process but also provides a basis for key feature extraction. Next, the system performs key feature extraction on the power consumption time-series feature matrix through a dynamic feature selector to obtain a power consumption key feature set, and performs time-series correlation analysis on this set to obtain a power consumption feature correlation network. The dynamic feature selector can automatically identify the most representative and discriminative key features according to the characteristics of the input data, thereby improving the accuracy and efficiency of the prediction model. For example, in the battery management scenario of a mobile device, assuming that the power consumption time-series feature matrix shows that the high-frequency band energy requirements increase rapidly in the next few seconds, while the low-frequency band is relatively stable. Through the dynamic feature selector, features related to power consumption mutation, such as the frequency adjustment amplitude and the slope of the power demand change, can be extracted. Further, by performing time-series correlation analysis on these key features, a power consumption feature correlation network can be constructed. This network not only reveals the correlation strength between different features but also reflects internal relationships such as feature transmission delay and feature coupling degree. Based on the power consumption feature correlation network, the system performs power consumption evolution path analysis on the load device to obtain a power consumption evolution feature map, and performs dynamic segmentation processing on this map to obtain a power consumption prediction feature group.Power consumption evolution path analysis aims to infer the future power consumption change trajectory based on historical data, which is crucial for early energy management and scheduling. For example, in the application of Internet of Things sensors, assuming that the device is about to execute a series of short-term tasks, the power consumption feature correlation network may show that the energy demand in the high-frequency band will increase significantly in the next few seconds and then gradually decline. By establishing an appropriate prediction model (such as ARIMA or neural network model), a power consumption evolution feature map can be generated, and combined with dynamic segmentation technology, the entire evolution process can be divided into multiple stages, each stage corresponding to different power consumption evolution directions, change rates, and stability indicators. These features not only help to understand the overall trend of power consumption changes but also provide a basis for subsequent non-linear mapping transformation. Next, the system will perform a non-linear mapping transformation on the power consumption prediction feature group to obtain a power consumption prediction parameter matrix, and perform feature reconstruction on the prediction parameter matrix through a multi-dimensional feature fusion device to obtain a power consumption prediction vector. Non-linear mapping transformation is an effective data processing method. By mapping the original feature space to a new feature space, it can better capture the internal laws and complex relationships of the data. For example, in the battery management scenario of mobile devices, assuming that the power consumption prediction feature group shows that the energy demand in the high-frequency band will increase rapidly in the next period of time and then gradually stabilize. Through non-linear mapping transformation, these features can be transformed into a unified power consumption prediction parameter matrix, which not only contains key information such as predicted power consumption values, prediction time windows, and prediction confidence levels but also provides a basis for subsequent feature reconstruction. Further, through a multi-dimensional feature fusion device, these parameters can be integrated into a comprehensive power consumption prediction vector, which can comprehensively describe the future power consumption characteristics. Finally, the system will perform time series expansion reconstruction based on the power consumption prediction vector to obtain a power consumption prediction time series table, and perform dynamic calibration on this table to obtain a power consumption prediction sequence. Time series expansion reconstruction aims to reorganize the information in the prediction vector into a time series form for real-time monitoring and scheduling in practical applications. For example, in the application scenario of Internet of Things sensors, assuming that the power consumption prediction vector shows that the power consumption will experience a significant increase and then stabilize in the next few minutes. Through time series expansion reconstruction, this information can be transformed into a detailed power consumption prediction time series table, which lists the predicted power consumption values and their corresponding confidence levels at each time point. Subsequently, through dynamic calibration of this table, prediction errors can be eliminated and it is ensured that the finally obtained power consumption prediction sequence has high accuracy and reliability. This prediction sequence not only contains the power consumption prediction trajectory but also reflects the power consumption prediction interval, providing an important reference basis for subsequent dynamic frequency adjustment. For example, in the battery management application scenario of mobile devices, assume that the user needs to switch from the energy-saving mode to the high-performance mode to run a graphics-intensive application.First, the system performs feature hierarchical analysis on the power consumption dynamic feature map, identifying high-frequency energy demands (such as rapidly increasing the operating frequency of the flexible power supply MCU), mid-frequency fluctuations (such as gradually increasing the power output of the power supply), and low-frequency trends (such as starting and stably operating the GPU acceleration module). Then, by performing time-domain expansion on the multi-layer power consumption feature sequences, a power consumption time-series feature matrix is formed. Next, a dynamic feature selector is used to extract features related to power consumption mutations and construct a power consumption feature correlation network. Based on this network, the system can infer the future power consumption evolution path, obtain a power consumption evolution feature map, and through dynamic segmentation processing, form a power consumption prediction feature group. Subsequently, a non-linear mapping transformation is performed on the power consumption prediction feature group to obtain a power consumption prediction parameter matrix, which is integrated into a power consumption prediction vector through a multi-dimensional feature fusion device. Finally, based on this vector, time-series expansion reconstruction is carried out to obtain a power consumption prediction time-series table, and through dynamic calibration, a power consumption prediction sequence is ultimately formed. This prediction sequence not only includes the change trend and peak value of power consumption but also provides a reliable prediction confidence level, providing important support for the energy management and performance optimization of the system. In summary, through a series of steps such as feature hierarchical analysis, time-domain expansion, key feature extraction, time-series correlation analysis, power consumption evolution path analysis, non-linear mapping transformation, and multi-dimensional feature fusion on the power consumption dynamic feature map, the system can generate a scientific and reasonable power consumption prediction sequence, thereby achieving accurate prediction of future power consumption change trends. This method demonstrates strong flexibility and adaptability in practical applications, especially in scenarios that require frequent mode switching, significantly improving the system's response speed and energy utilization efficiency and providing important technical support for optimizing the overall system performance.

[0083] In a specific embodiment, the power consumption evolution path analysis of the load device based on the power consumption feature correlation network to obtain a power consumption evolution feature map includes:

[0084] Perform topological structure analysis on the power consumption feature correlation network to obtain a set of power consumption feature transfer paths, and perform weight assignment on the set of power consumption feature transfer paths to obtain a power consumption path weight matrix;

[0085] Through a multi-dimensional path analyzer, perform path optimization on the power consumption path weight matrix to obtain a key power consumption evolution path sequence, and perform time-domain expansion on the key power consumption evolution path sequence to obtain a power consumption evolution time-series diagram;

[0086] Based on the power consumption evolution time-series diagram, perform feature propagation analysis to obtain a power consumption feature propagation network, and perform dynamic feature extraction on the power consumption feature propagation network to obtain a power consumption evolution feature set;

[0087] Perform multi-level reconstruction on the power consumption evolution feature set to obtain a power consumption evolution topological structure, and perform dynamic mapping on the power consumption evolution topological structure through a feature fusion processor to obtain a power consumption evolution feature map, where the power consumption evolution feature map includes a power consumption evolution trajectory, an evolution bifurcation point, and an evolution stable region.

[0088] Specifically, in the above steps, "performing power consumption evolution path analysis on the load device based on the power consumption feature correlation network to obtain a power consumption evolution feature map" is one of the key steps in the flexible power supply MCU dynamic frequency modulation control method. Its purpose is to accurately depict the future power consumption change trajectory through in-depth analysis of the power consumption feature correlation network, combined with various algorithms and technical means. Specifically, this process includes multiple steps such as topological structure analysis, weight assignment, path optimization, timing unfolding, feature propagation analysis, multi-level reconstruction, and dynamic mapping, so as to finally generate a feature map that comprehensively describes the future power consumption evolution. First, the system performs topological structure analysis on the power consumption feature correlation network to obtain a set of power consumption feature transfer paths, and assigns weights to this set to obtain a power consumption path weight matrix. The topological structure analysis aims to reveal the connection relationships and transfer paths between power consumption features, which helps to understand the influence mechanism between different features. For example, in the application scenario of Internet of Things sensors, assume that the load device switches from a low-power standby mode to a high-frequency data acquisition mode. The power consumption feature correlation network may include multiple nodes (such as different functional modules) and edges (such as energy transfer relationships). Through topological structure analysis, the set of transfer paths between these nodes can be identified, and each path represents a specific energy transfer method. Subsequently, by assigning weights to the set of transfer paths, parameters such as the transfer efficiency, node connection strength, and branch diffusion coefficient of each path can be calculated to form a power consumption path weight matrix. This matrix not only quantifies the importance of different paths but also provides a basis for subsequent path optimization. Next, the system performs path optimization on the power consumption path weight matrix through a multi-dimensional path analyzer to obtain a sequence of key power consumption evolution paths, and unfolds the sequence in time to obtain a power consumption evolution timing diagram. Path optimization is an optimization technique that selects the most representative and influential path combinations by evaluating the weights and influencing factors of different paths. For example, in the battery management scenario of a mobile device, assume that the power consumption path weight matrix shows that certain paths have high transfer efficiency and strong node connection strength. Through the multi-dimensional path analyzer, a sequence of key power consumption evolution paths can be selected from them, and these paths represent the main energy transfer directions that are most likely to occur in the future for a period of time. Further, by unfolding the sequence of key power consumption evolution paths in time, it can be converted into a two-dimensional power consumption evolution timing diagram, which contains important information such as power consumption evolution nodes, evolution branch points, and evolution convergence points. This timing unfolding not only simplifies the subsequent data processing process but also provides a basis for feature propagation analysis. Based on the power consumption evolution timing diagram, the system performs feature propagation analysis to obtain a power consumption feature propagation network, and performs dynamic feature extraction on this network to obtain a set of power consumption evolution features. Feature propagation analysis aims to infer the future feature propagation trend based on historical data, which is crucial for early energy management and scheduling.For example, in the application of Internet of Things sensors, assume that the power consumption evolution time series diagram shows that the energy demand in the high-frequency band will rise rapidly in the next few seconds, while the low-frequency band remains relatively stable. By establishing an appropriate prediction model (such as ARIMA or neural network model), a power consumption feature propagation network can be generated, and combined with dynamic feature extraction technology, a key set of power consumption evolution features can be extracted. These features not only reflect the internal laws such as feature propagation rate, feature attenuation coefficient, and feature superposition intensity, but also provide a basis for subsequent multi-level reconstruction. Next, the system will perform multi-level reconstruction on the power consumption evolution feature set to obtain the power consumption evolution topological structure, and dynamically map the topological structure through a feature fusion processor to obtain the power consumption evolution feature map. Multi-level reconstruction is an effective data analysis method. By decomposing a complex feature set into multiple hierarchical structures, the mutual relationship between each level can be shown more clearly. For example, in the battery management scenario of mobile devices, the power consumption evolution feature set may contain feature information on multiple time scales. Through multi-level reconstruction, it can be organized into a hierarchical power consumption evolution topological structure, where each node represents the power consumption feature at a specific time scale. Subsequently, through the feature fusion processor, these features can be remapped to a new feature space to form the power consumption evolution feature map. This map not only contains the power consumption evolution trajectory, but also reflects important information such as evolution bifurcation points and evolution stable regions. For example, in the battery management application scenario of mobile devices, assume that the user needs to switch from the energy-saving mode to the high-performance mode to run a graphics-intensive application. First, the system will perform topological structure analysis on the power consumption feature correlation network to identify the main energy transfer paths (such as rapidly increasing the working frequency of the flexible power supply MCU, adjusting the power output, starting and stably operating the GPU acceleration module). Then, by assigning weights to these paths, a power consumption path weight matrix is formed, which quantifies the transfer efficiency, node connection strength, and branch diffusion coefficient of each path. Next, a multi-dimensional path analyzer is used to screen out the key power consumption evolution path sequence, and through time series unfolding, a power consumption evolution time series diagram is formed, which lists the power consumption evolution nodes, branch points, and convergence points at each time point. Based on the power consumption evolution time series diagram, the system can perform feature propagation analysis to obtain the power consumption feature propagation network, and extract the power consumption evolution feature set from it. These features not only reflect the rate and intensity of energy transfer, but also show the attenuation and superposition effects of the features. Subsequently, multi-level reconstruction is performed on the power consumption evolution feature set to form the power consumption evolution topological structure, and it is integrated into the power consumption evolution feature map through a feature fusion processor. This feature map not only contains the evolution trajectory of the power consumption, but also reveals the bifurcation points and stable regions in the evolution process, providing important support for the energy management and performance optimization of the system.In summary, through a series of steps such as topological structure analysis, weight assignment, path optimization, timing expansion, feature propagation analysis, multi-level reconstruction, and dynamic mapping of the power consumption feature correlation network, the system can generate a scientific and reasonable power consumption evolution feature map, thereby achieving an accurate depiction of the future power consumption change trajectory. This method demonstrates strong flexibility and adaptability in practical applications. Especially in scenarios that require frequent mode switching, it can significantly improve the system's response speed and energy utilization efficiency, providing important technical support for optimizing the overall system performance.

[0089] In a specific embodiment, the mapping of the operating frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence includes:

[0090] Performing frequency-domain deconstruction analysis on the load power consumption frequency band characteristics to obtain a frequency response feature matrix, and performing dynamic segmentation processing on the frequency response feature matrix to obtain a frequency allocation weight vector;

[0091] Performing frequency boundary constraint analysis on the frequency allocation weight vector through a frequency mapping resolver to obtain a frequency adjustment boundary set, and performing dynamic optimization and reconstruction on the frequency adjustment boundary set to obtain a frequency adjustment feature network;

[0092] Performing operating frequency space mapping on the flexible power supply MCU based on the frequency adjustment feature network to obtain a frequency mapping topology graph, and performing timing feature extraction on the frequency mapping topology graph to obtain a frequency adjustment feature sequence, where the frequency adjustment feature sequence includes a frequency adjustment step size, a frequency switching timing, and a frequency holding period;

[0093] Performing multi-dimensional dynamic reconstruction on the frequency adjustment feature sequence to obtain a frequency adjustment control matrix, and performing timing expansion on the frequency adjustment control matrix through a frequency sequence optimizer to obtain a dynamic frequency adjustment sequence; where the dynamic frequency adjustment sequence includes a frequency adjustment trajectory, a frequency switching point, and a frequency stable interval.

[0094] Specifically, in the above steps, "mapping the operating frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence" is one of the key links in the dynamic frequency modulation control method of the flexible power supply MCU. Its purpose is to achieve precise adjustment and optimization of the operating frequency of the flexible power supply MCU by deeply analyzing the load power consumption frequency band characteristics and combining various algorithms and technical means. Specifically, this process includes multiple steps such as frequency domain deconstruction analysis, dynamic segmentation processing, frequency boundary constraint analysis, dynamic optimization reconstruction, operating frequency space mapping, timing feature extraction, and multi-dimensional dynamic reconstruction, so as to finally generate a dynamic frequency adjustment sequence that comprehensively describes the operating frequency change trajectory of the flexible power supply MCU. First, the system will perform frequency domain deconstruction analysis on the load power consumption frequency band characteristics to obtain a frequency response characteristic matrix, and perform dynamic segmentation processing on this matrix to obtain a frequency allocation weight vector. The frequency domain deconstruction analysis aims to decompose the complex load power consumption frequency band characteristics into multiple frequency components and their corresponding response characteristics, which helps to understand the energy requirements of different frequency bands and their impact on the performance of the flexible power supply MCU. For example, in the application scenario of Internet of Things sensors, assuming that the load device switches from a low-power standby mode to a high-frequency data acquisition mode, the load power consumption frequency band characteristics may include high-frequency band energy requirements (such as quickly increasing the operating frequency of the flexible power supply MCU), medium-frequency band fluctuations (such as adjusting the power output), and low-frequency band trends (such as starting and stabilizing the operation of the data acquisition module). Through frequency domain deconstruction analysis, these frequency components and their response characteristics can be identified to form a power response characteristic matrix. Subsequently, by performing dynamic segmentation processing on this matrix, the weights of each frequency band can be allocated according to actual needs to form a frequency allocation weight vector. This vector not only quantifies the importance of different frequency bands but also provides a basis for subsequent frequency boundary constraint analysis. Then, the system will perform frequency boundary constraint analysis on the frequency allocation weight vector through a frequency mapping resolver to obtain a frequency adjustment boundary set, and perform dynamic optimization reconstruction on this set to obtain a frequency adjustment characteristic network. The frequency boundary constraint analysis aims to determine the operating range of each frequency band according to hardware limitations and application requirements to ensure the stability and reliability of the system. For example, in the battery management scenario of mobile devices, assuming that the frequency allocation weight vector shows that some frequency bands have high energy requirements. Through the frequency mapping resolver, the specific adjustment boundaries of these frequency bands can be determined to form a power adjustment boundary set. Further, by performing dynamic optimization reconstruction on the boundary set, a frequency adjustment characteristic network can be constructed. This network not only includes information such as frequency adjustment intervals, frequency switching thresholds, and frequency stability data but also provides a basis for subsequent operating frequency space mapping. Based on the frequency adjustment characteristic network, the system will perform operating frequency space mapping on the flexible power supply MCU to obtain a frequency mapping topology diagram, and perform timing feature extraction on this diagram to obtain a frequency adjustment characteristic sequence.The working frequency space mapping aims to convert the frequency regulation feature network into a specific frequency regulation scheme for real-time monitoring and scheduling in practical applications. For example, in the application of Internet of Things sensors, assume that the frequency regulation feature network shows that the energy demand in the high-frequency band will rise rapidly in the next few seconds, while the low-frequency band remains relatively stable. By establishing an appropriate frequency regulation model (such as linear regression or neural network model), a power mapping topology can be generated, which lists the frequency regulation intervals and switching thresholds at each time point. Subsequently, by extracting the temporal features of the frequency mapping topology, it can be converted into a detailed frequency regulation feature sequence, which contains important information such as frequency regulation step size, frequency switching timing, and frequency holding period. This temporal feature extraction not only simplifies the subsequent data processing flow but also provides a basis for multi-dimensional dynamic reconstruction. Next, the system will perform multi-dimensional dynamic reconstruction on the frequency regulation feature sequence to obtain a frequency regulation control matrix, and perform temporal unfolding on the control matrix through a frequency sequence optimizer to obtain a dynamic frequency regulation sequence. Multi-dimensional dynamic reconstruction is an effective data analysis method. By decomposing the complex frequency regulation feature sequence into multiple dimensional structures, the mutual relationships between dimensions can be shown more clearly. For example, in the battery management scenario of mobile devices, the frequency regulation feature sequence may contain frequency regulation information on multiple time scales. Through multi-dimensional dynamic reconstruction, it can be organized into a hierarchical frequency regulation control matrix, where each node represents the frequency regulation strategy at a specific time scale. Subsequently, through the frequency sequence optimizer, these strategies can be remapped to a new time series space to form a dynamic frequency regulation sequence. This sequence not only contains the frequency regulation trajectory but also reflects important information such as frequency switching points and frequency stable regions. For example, in the battery management application scenario of mobile devices, assume that the user needs to switch from the energy-saving mode to the high-performance mode to run a graphics-intensive application. First, the system will perform frequency domain deconstruction analysis on the load power consumption frequency band characteristics to identify the main energy demand frequency bands (such as rapidly increasing the working frequency of the flexible power supply MCU, adjusting the power output, starting and stably operating the GPU acceleration module). Then, by performing dynamic segmentation on these frequency bands, a power distribution weight vector is formed, which quantifies the importance of each frequency segment. Next, the frequency mapping resolver is used to determine the specific regulation boundaries of each frequency band to form a power regulation boundary set, and perform dynamic optimization reconstruction on it to form a frequency regulation feature network. Based on this network, the system can perform working frequency space mapping to obtain a frequency mapping topology, which lists the frequency regulation intervals and switching thresholds at each time point. Subsequently, by extracting the temporal features of the frequency mapping topology, a power regulation feature sequence can be formed, which contains information such as frequency regulation step size, frequency switching timing, and frequency holding period.Finally, a multi-dimensional dynamic reconstruction is performed on the frequency adjustment feature sequence to form a power adjustment control matrix, which is integrated into a dynamic frequency adjustment sequence by a frequency sequence optimizer. This sequence not only contains the frequency adjustment trajectory but also reveals the frequency switching points and frequency stable regions, providing important support for the energy management and performance optimization of the system. In summary, through a series of steps such as frequency domain deconstruction analysis, dynamic segmentation processing, frequency boundary constraint analysis, dynamic optimization reconstruction, working frequency space mapping, timing feature extraction, and multi-dimensional dynamic reconstruction of the load power consumption frequency band characteristics, the system can generate a scientific and reasonable dynamic frequency adjustment sequence, thereby achieving precise adjustment and optimization of the working frequency of the flexible power supply MCU. This method demonstrates strong flexibility and adaptability in practical applications, especially in scenarios that require frequent mode switching, significantly improving the system's response speed and energy utilization efficiency, and providing important technical support for optimizing the overall system performance.

[0095] The dynamic frequency modulation control method of the flexible power supply MCU in the embodiments of the present invention has been described above. Next, the dynamic frequency modulation control device of the flexible power supply MCU in the embodiments of the present invention will be described. Please refer to Figure 2 One embodiment of the dynamic frequency modulation control device of the flexible power supply MCU in the embodiments of the present invention includes:

[0096] A reading module 21, configured to read the target working mode and the current working mode of a preset load device, and determine a mode switching instruction of the load device based on the target working mode and the current working mode; an extraction module 22, configured to perform timing feature extraction on the mode switching instruction to obtain a mode instruction timing feature vector; a prediction module 23, configured to perform frequency band decomposition and prediction on the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain load power consumption frequency band characteristics; a mapping module 24, configured to perform a working frequency range mapping on the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence; and a control module 25, configured to input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode.

[0097] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the description in the above method embodiment, and details will not be repeated here.

[0098] Refer to Figure 3 In the embodiments of the present invention, a computer device is also provided. The internal structure of the computer device can be as Figure 3As shown in the figure. The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected via a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0099] Those skilled in the art can understand that Figure 3 the structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0100] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0101] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database, or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0102] It should be noted that in this text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, apparatus, article or method comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, apparatus, article or method. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, apparatus, article or method comprising such element.

[0103] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. A flexible power supply MCU dynamic frequency modulation control method, characterized in that It includes the following steps: Read the target working mode and the current working mode of the preset load device, and determine the mode switching instruction of the load device based on the target working mode and the current working mode; extract the timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector; decompose and predict the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain load power consumption frequency band characteristics; Map the working frequency range of the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence; Input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode; Perform state quantization encoding on the target working mode and the current working mode of the load device to obtain a target state code and a current state code, and perform differential operation on the target state code and the current state code to obtain a state difference value; Based on the state difference value, index the mode switching instruction set of the load device to obtain a mode switching instruction subset, and perform timing sorting on the mode switching instruction subset to obtain a sorted mode switching instruction sequence; Perform instruction coding compression on the sorted mode switching instruction sequence to obtain a mode switching instruction of the load device; The decomposing and predicting the real-time power consumption of the load device based on the mode instruction timing feature vector to obtain load power consumption frequency band characteristics includes: Perform multi-dimensional spectrum decomposition on the mode instruction timing feature vector to obtain a power consumption frequency response feature set, and perform dynamic threshold segmentation on the power consumption frequency response feature set to obtain a power consumption frequency band boundary sequence; Perform energy flow analysis on the power consumption frequency band boundary sequence through an adaptive filter to obtain a frequency band energy transfer matrix, and perform non-linear feature extraction on the frequency band energy transfer matrix to obtain a power consumption dynamic feature map; Perform power consumption trend prediction on the load device based on the power consumption dynamic feature map to obtain a power consumption prediction sequence, and perform time-frequency joint analysis on the power consumption prediction sequence to obtain a power consumption fluctuation feature set; Perform multi-level decomposition on the power consumption fluctuation feature set to obtain a power consumption feature hierarchy tree, and perform frequency domain reconstruction on the power consumption feature hierarchy tree to obtain a frequency band power consumption mapping network; Optimize and reorganize the frequency band power consumption mapping network through a dynamic planner to obtain a power consumption frequency band feature vector, and perform normalization processing on the power consumption frequency band feature vector to obtain load power consumption frequency band characteristics; wherein, the load power consumption frequency band characteristics include frequency band power consumption characteristic values, frequency band stability data, and frequency band switching thresholds.

2. The flexible power supply MCU dynamic frequency modulation control method according to claim 1, wherein The extracting the timing characteristics of the mode switching instruction to obtain a mode instruction timing feature vector includes: Perform time-domain segmented sampling on the mode switching instruction to obtain an instruction timing sampling sequence, and perform wavelet transform decomposition on the instruction timing sampling sequence to obtain multi-scale timing feature components; Perform temporal correlation analysis on the mode switching instruction based on the multi-scale temporal feature components to obtain an instruction temporal correlation matrix, and perform singular value decomposition on the instruction temporal correlation matrix to obtain an instruction feature singular value sequence; Perform feature dimensionality reduction mapping based on the instruction feature singular value sequence to obtain a dimensionality-reduced feature subspace, and perform non-linear feature fusion on the dimensionality-reduced feature subspace to obtain the mode instruction temporal feature vector, where the mode instruction temporal feature vector includes an instruction energy consumption feature component, an instruction switching delay feature component, and an instruction execution cycle feature component.

3. The flexible power supply MCU dynamic frequency modulation control method according to claim 1, characterized in that The power consumption trend prediction of the load device based on the power consumption dynamic feature map, obtaining a power consumption prediction sequence, includes: Perform feature hierarchical analysis on the power consumption dynamic feature map to obtain a multi-layer power consumption feature sequence, and perform time-domain expansion on the multi-layer power consumption feature sequence to obtain a power consumption temporal feature matrix, where the power consumption temporal feature matrix includes a power consumption change trend, a power consumption mutation feature, and a power consumption steady state interval; Extract key features from the power consumption temporal feature matrix through a dynamic feature selector to obtain a power consumption key feature set, and perform temporal correlation analysis on the power consumption key feature set to obtain a power consumption feature correlation network; Perform power consumption evolution path analysis on the load device based on the power consumption feature correlation network to obtain a power consumption evolution feature map, and perform dynamic segmentation processing on the power consumption evolution feature map to obtain a power consumption prediction feature group; Perform non-linear mapping transformation on the power consumption prediction feature group to obtain a power consumption prediction parameter matrix, and perform feature reconstruction on the power consumption prediction parameter matrix through a multi-dimensional feature fusion device to obtain a power consumption prediction vector; Perform time-series expansion and reconstruction based on the power consumption prediction vector to obtain a power consumption prediction time-series table, and perform dynamic calibration on the power consumption prediction time-series table to obtain a power consumption prediction sequence, where the power consumption prediction sequence includes a power consumption prediction trajectory and a power consumption prediction interval.

4. The flexible power supply MCU dynamic frequency modulation control method according to claim 3, wherein The power consumption evolution path analysis of the load device based on the power consumption feature correlation network, obtaining a power consumption evolution feature map, includes: Perform topological structure analysis on the power consumption feature correlation network to obtain a power consumption feature transfer path set, and perform weight assignment on the power consumption feature transfer path set to obtain a power consumption path weight matrix; Perform path optimization on the power consumption path weight matrix through a multi-dimensional path analyzer to obtain a key power consumption evolution path sequence, and perform time-series expansion on the key power consumption evolution path sequence to obtain a power consumption evolution time-series diagram; Perform feature propagation analysis based on the power consumption evolution time-series diagram to obtain a power consumption feature propagation network, and perform dynamic feature extraction on the power consumption feature propagation network to obtain a power consumption evolution feature set; Perform multi-level reconstruction on the power consumption evolution feature set to obtain a power consumption evolution topological structure, and perform dynamic mapping on the power consumption evolution topological structure through a feature fusion processor to obtain a power consumption evolution feature map, where the power consumption evolution feature map includes a power consumption evolution trajectory, an evolution bifurcation point, and an evolution stable region.

5. The flexible power supply MCU dynamic frequency modulation control method according to claim 1, wherein Performing a working frequency range mapping on the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence, including: Performing a frequency-domain deconstruction analysis on the load power consumption frequency band characteristics to obtain a frequency response characteristic matrix, and performing a dynamic segmentation process on the frequency response characteristic matrix to obtain a frequency allocation weight vector; Performing a frequency boundary constraint analysis on the frequency allocation weight vector through a frequency mapping parser to obtain a frequency adjustment boundary set, and performing a dynamic optimization and reconstruction on the frequency adjustment boundary set to obtain a frequency adjustment characteristic network; Performing a working frequency space mapping on the flexible power supply MCU based on the frequency adjustment characteristic network to obtain a frequency mapping topology diagram, and performing a timing characteristic extraction on the frequency mapping topology diagram to obtain a frequency adjustment characteristic sequence, wherein the frequency adjustment characteristic sequence includes a frequency adjustment step size, a frequency switching timing, and a frequency holding period; Performing a multi-dimensional dynamic reconstruction on the frequency adjustment characteristic sequence to obtain a frequency adjustment control matrix, and performing a timing expansion on the frequency adjustment control matrix through a frequency sequence optimizer to obtain a dynamic frequency adjustment sequence; wherein the dynamic frequency adjustment sequence includes a frequency adjustment trajectory, a frequency switching point, and a frequency stable interval.

6. A flexible power supply MCU dynamic frequency modulation control device, characterized in that, Including: A reading module, configured to read a target working mode and a current working mode of a preset load device, and determine a mode switching instruction of the load device based on the target working mode and the current working mode; An extraction module, configured to perform a timing characteristic extraction on the mode switching instruction to obtain a mode instruction timing characteristic vector; A prediction module, configured to perform a frequency band decomposition and prediction on the real-time power consumption of the load device based on the mode instruction timing characteristic vector to obtain load power consumption frequency band characteristics; A mapping module, configured to perform a working frequency range mapping on the flexible power supply MCU based on the load power consumption frequency band characteristics to obtain a dynamic frequency adjustment sequence; a control module, configured to input the dynamic frequency adjustment sequence into the flexible power supply MCU to control the load device to switch from the current working mode to the target working mode; Performing a state quantization encoding on the target working mode and the current working mode of the load device to obtain a target state code and a current state code, and performing a difference operation on the target state code and the current state code to obtain a state difference value; Indexing a mode switching instruction set of the load device based on the state difference value to obtain a mode switching instruction subset, and performing a timing sorting on the mode switching instruction subset to obtain a sorted mode switching instruction sequence; Performing an instruction encoding compression on the sorted mode switching instruction sequence to obtain a mode switching instruction of the load device; The performing a frequency band decomposition and prediction on the real-time power consumption of the load device based on the mode instruction timing characteristic vector to obtain load power consumption frequency band characteristics includes: Performing a multi-dimensional spectrum decomposition on the mode instruction timing characteristic vector to obtain a power consumption frequency response characteristic set, and performing a dynamic threshold segmentation on the power consumption frequency response characteristic set to obtain a power consumption frequency band boundary sequence; Perform energy flow analysis on the power consumption frequency band boundary sequence through an adaptive filter to obtain a frequency band energy transfer matrix, and perform non-linear feature extraction on the frequency band energy transfer matrix to obtain a power consumption dynamic feature map; Based on the power consumption dynamic feature map, perform power consumption trend prediction on the load device to obtain a power consumption prediction sequence, and perform time-frequency joint analysis on the power consumption prediction sequence to obtain a power consumption fluctuation feature set; Perform multi-level decomposition on the power consumption fluctuation feature set to obtain a power consumption feature hierarchy tree, and perform frequency domain reconstruction on the power consumption feature hierarchy tree to obtain a frequency band power consumption mapping network; Optimize and reorganize the frequency band power consumption mapping network through a dynamic planner to obtain a power consumption frequency band feature vector, and perform normalization processing on the power consumption frequency band feature vector to obtain a load power consumption frequency band feature; wherein, the load power consumption frequency band feature includes a frequency band power consumption feature value, frequency band stability data, and a frequency band switching threshold.

7. A computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

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

Images