Detailed Description
The embodiment of the application provides a Bluetooth circuit board self-adaptive power consumption management method and system. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For easy understanding, the following describes a specific flow of an embodiment of the present application, referring to fig. 1, and an embodiment of a bluetooth circuit board adaptive power consumption management method in an embodiment of the present application includes:
Step 101, setting power converters at different levels of a multi-layer circuit board, and performing real-time power consumption monitoring to obtain a first power data set of a first power converter and a second power data set of a second power converter;
It can be appreciated that the execution subject of the present application may be a bluetooth circuit board adaptive power consumption management system, and may also be a terminal or a server, which is not limited herein. The embodiment of the application is described by taking a server as an execution main body as an example.
Specifically, a top layer circuit board, a middle layer circuit board and a bottom layer circuit board in a multilayer circuit board are obtained, and the multilayer circuit board is used for Bluetooth communication. In order to effectively manage the power requirements of these levels, a first power converter is disposed between the top and middle circuit boards, while a second power converter is disposed between the middle and bottom circuit boards, so that the power can be more accurately distributed as needed, optimizing the energy efficiency of the overall system. Real-time current and voltage monitoring is performed. And for the first power converter, collecting initial data of current and voltage by the monitoring equipment to obtain first initial current data and first initial voltage data. And obtaining second initial current data and second initial voltage data through monitoring of the second power converter. To extract useful information from the initial data, the first and second initial current and voltage data are processed using principal component analysis techniques. Principal component analysis can effectively extract principal features from a large amount of data, reduce the complexity of the data while retaining the most critical information. By analysis, more representative data sets are obtained, namely first target current data and first target voltage data, and second target current data and second target voltage data. A power dataset is calculated based on the extracted target data. Using the first target current data and the first target voltage data, a first power data set of the first power converter is calculated, which data set directly reflects the power consumption situation between the top layer and the middle layer circuit board in actual operation. And likewise, calculating a second power data set by using the second target current data and the second target voltage data to obtain the power consumption condition between the middle-layer circuit board and the bottom-layer circuit board.
Feature extraction is performed on the initial current data and the voltage data monitored by the first and second power converters. Critical characteristic information such as peak value, valley value, average value and fluctuation rate of current, peak value, valley value, average value and fluctuation rate of voltage is extracted from the original monitoring data. By employing data processing techniques, such as signal processing algorithms, useful signals can be effectively separated from noise, ensuring that the extracted features are representative and practical. And the extracted current and voltage characteristics are subjected to linear characteristic transformation, and multidimensional data are converted into main components, so that subsequent processing and analysis are simplified. The linear feature transformation generally adopts a principal component analysis method, and components capable of representing the maximum variance of data are extracted by constructing a covariance matrix of the data, and the components are principal feature vectors of the data. After the linear feature transformation is completed, covariance matrices of the first current, the second current and the voltage are respectively constructed. The covariance matrix is constructed to analyze the correlation between different features, the correlation between which can provide important information on how to optimize and adjust the performance of the circuit board. By calculating the covariance of the feature vectors, it is known which features are correlated and which are independent. And (5) performing feature correlation calculation, and evaluating the correlation among the features by using a covariance matrix. By calculation it is determined which features have a greater influence on the current and voltage variations. On the basis of the feature correlation analysis, weights of each feature are calculated, and the weights reflect the importance of the features in circuit power consumption management. And carrying out feature fusion on the initially extracted current and voltage features by using the calculated feature weights. And integrating all the characteristics into one comprehensive current and voltage data in a weighted average mode to obtain target current data and target voltage data.
102, Respectively extracting power characteristics of a first power data set and a second power data set to obtain a first power characteristic set and a second power characteristic set;
Specifically, curve fitting is performed on the first and second power data sets respectively to obtain a first power curve and a second power curve describing the power consumption of the circuit board along with the time change. A mathematical model, such as a polynomial regression or a nonlinear model, is used to ensure that the curve reflects the actual power consumption data as accurately as possible. Based on the obtained power curve, feature extraction is performed, and key power features such as peak value, average power consumption, fluctuation amplitude and the like are identified from the simplified curve. And carrying out characteristic analysis on the first power curve and the second power curve to obtain a first initial power characteristic and a second initial power characteristic, wherein the characteristics reflect the specific conditions of each part of the circuit board on energy consumption management. And classifying the first initial power characteristics to obtain a plurality of first high-frequency power characteristics and a plurality of first low-frequency power characteristics, and classifying the second initial power characteristics to obtain a plurality of second high-frequency power characteristics and a plurality of second low-frequency power characteristics. The high frequency characteristics may be related to transient reactions and short term variations of the circuit board, while the low frequency characteristics may reflect longer term energy consumption trends and patterns. Classification helps the system to manage and optimize more specifically for different power consumption needs and performance expectations. By calculating the weights of the various features, their importance in overall power management is assessed, and the degree of influence of the features on the performance of the circuit board is assessed. And integrating the high-frequency power characteristics and the low-frequency power characteristics by using the weights to form a final first power characteristic set and a final second power characteristic set.
And carrying out Pearson correlation coefficient calculation on the first high-frequency power characteristic and the first low-frequency power characteristic, and determining the linear relation strength between different characteristics. The pearson correlation coefficient provides a measure of the degree of linear correlation of two variables, with a value between-1 and 1, where 1 represents a complete positive correlation, -1 represents a complete negative correlation, and 0 means no linear correlation. Similarly, pearson correlation coefficients are calculated for a second set of high frequency and low frequency power characteristics, resulting in a second pearson correlation coefficient. And setting the weight value of each characteristic according to the obtained pearson correlation coefficient. The high-correlation features obtain higher weight values, because the features are more important in describing the power consumption behavior of the circuit board, and the construction of the feature set is ensured to truly reflect the actual performance of the circuit board in different working states. The high frequency and low frequency power characteristics of the first and second sets are weighted using the set characteristic weights. Each feature will be adjusted according to its weight value to enhance the effect of those features that have a greater impact on the performance of the circuit board. And combining the weighted high-frequency power characteristics and the weighted low-frequency power characteristics to form a first power characteristic set and a second power characteristic set. The feature sets comprise all the screened and weighted key power features, so that the power consumption optimization strategy of the circuit board is accurate and strong in adaptability, and the power consumption optimization strategy can be dynamically adjusted according to actual requirements to achieve optimal energy efficiency.
Step 103, creating global power consumption optimization parameters of the multi-layer circuit board according to the first power feature set and the second power feature set through a dual Q network and a PSO algorithm.
Specifically, feature coding is performed on the first power feature set and the second power feature set respectively to obtain a first power feature vector and a second power feature vector, and complex power data is simplified into a mathematical expression form which is easier to calculate and analyze. And performing power optimization execution strategy analysis on the first power eigenvector and the second power eigenvector through the double Q network respectively. The dual Q network is a reinforcement learning algorithm that can avoid overestimation through two independent learning paths while maintaining stability during the learning process. In this process, the dual Q network outputs an initial power consumption optimization execution strategy of the first and second power converters by learning and simulating different power consumption management strategies. These strategies create an optimal power adjustment scheme based on real-time data and pre-set targets. To optimize these strategies and ensure their effectiveness in practical applications, bonus feedback parameter calculations are introduced. By evaluating the results of the operation of the first and second power converters, reward feedback parameters are calculated that reflect the effectiveness of each policy implementation and the success of power consumption savings. According to the feedback parameters, the dual Q network can update the initial power consumption optimization execution strategy, so that the dual Q network is more accurate and efficient. And calculating the power consumption parameters of the multi-layer circuit board according to the first target power consumption optimization execution strategy and the second target power consumption optimization execution strategy through a PSO algorithm. The PSO algorithm finds an optimal solution by simulating the searching behavior of the bird group, can effectively search in the global parameter space and quickly converge to an optimal power consumption management strategy. And combining the first target power consumption optimization execution strategy with the second target power consumption optimization execution strategy, and calculating global power consumption optimization parameters of the multi-layer circuit board by using a PSO algorithm. These parameters take into account the individual requirements of the circuit boards of each layer and ensure the maximization of the energy efficiency of the whole system by global optimization.
And generating a random initial value of the first power converter according to the first target power consumption optimization execution strategy through a PSO algorithm to obtain a first random initial value set, and generating a random initial value of the second power converter according to the second target power consumption optimization execution strategy to obtain a second random initial value set. The random initial value set provides necessary diversity foundation for particle swarm optimization, ensures that the algorithm can explore a wide area of the parameter space, and accordingly avoids missing a better global solution due to a local optimal solution. The first and second particle populations are constructed from a set of random initial values by a reverse particle propagation algorithm. The inverse particle propagation algorithm serves to optimize particle position and velocity, ensuring that particle groups can migrate effectively towards the region of potential optimal solution. Each particle represents one possible solution, and the entire population of particles constitutes the search space for the solution. The first and second particle populations are population partitioned to form a plurality of sub-particle populations. The searching process is refined, the searching efficiency is improved, and each sub-population can be deeply explored in a specific area of the parameter space. Population partitioning helps to finely manage resources so that each sub-population can adaptively adjust the search strategy in different directions, thereby more efficiently finding optimization parameters. And carrying out fitness calculation on each sub-particle population, and evaluating the effect of the solution represented by each particle on power consumption optimization. The fitness calculation is based on factors such as the degree of power consumption reduction, system stability and performance efficiency, ensuring that only the most suitable solution will be selected as the final optimization strategy, resulting in a first set of particle fitness and a second set of particle fitness. And generating power consumption optimization parameters of the first power converter and the second power converter according to the adaptability set. These parameters are based on the previously calculated optimal solutions aimed at minimizing the power consumption of the whole circuit board while maintaining the operating efficiency and stability of the system. And performing global optimization on the first power consumption optimization parameter and the second power consumption optimization parameter, and integrating all optimized parameters to form a global power consumption optimization parameter of the multi-layer circuit board.
In the embodiment of the application, the power converters are arranged at different levels of the multi-layer circuit board and are monitored in real time, so that the power consumption data of each layer of circuit board can be accurately obtained. This real-time monitoring ensures maximum efficiency of energy utilization, enabling the device to dynamically adjust power output according to the current operating conditions, thereby reducing inefficient energy consumption and extending the battery life of the device. And key features are effectively extracted from the original current and voltage data by applying a principal component analysis feature extraction technology. The deep data analysis method helps to more accurately understand the behaviors of the circuit board under different working conditions, and provides data support for subsequent power consumption optimization. By utilizing the dual Q network and the PSO algorithm, the method not only can create high-efficiency global power consumption optimization parameters, but also can dynamically adjust and optimize the power consumption configuration of the circuit board through the algorithm. The intelligent decision process remarkably improves the energy efficiency of the circuit board and reduces the energy waste caused by fixed or outdated power consumption configuration. By combining calculation of the reward feedback parameters and policy updating, the method realizes a self-learning power consumption management system. Through continuous learning and adjustment, the circuit board can constantly optimize its power consumption strategy in order to adapt to environmental change and equipment demand, promotes holistic performance and efficiency. By comprehensively considering the power consumption characteristics of each layer of circuit board and optimizing the execution strategy, the method can implement power consumption management on the whole circuit board level, thereby realizing more unified and coordinated energy management. The overall view management not only improves the energy efficiency, but also helps to keep the stability and reliability of the equipment, thereby realizing the self-adaptive power consumption management of the Bluetooth circuit board and improving the power consumption optimization effect of the multi-layer circuit board.
In a specific embodiment, the process of executing step 101 may specifically include the following steps:
(1) Acquiring a top layer circuit board, a middle layer circuit board and a bottom layer circuit board in a multi-layer circuit board, wherein the multi-layer circuit board is used for Bluetooth communication;
(2) A first power converter is arranged between the top layer circuit board and the middle layer circuit board, and a second power converter is arranged between the middle layer circuit board and the bottom layer circuit board;
(3) Current and voltage monitoring is conducted on the first power converter to obtain first initial current data and first initial voltage data, and current and voltage monitoring is conducted on the second power converter to obtain second initial current data and second initial voltage data;
(4) The method comprises the steps of respectively carrying out principal component characteristic analysis on first initial current data and first initial voltage data to obtain first target current data and first target voltage data, and respectively carrying out principal component characteristic analysis on second initial current data and second initial voltage data to obtain second target current data and second target voltage data;
(5) A first power data set of the first power converter is calculated based on the first target current data and the first target voltage data, and a second power data set of the second power converter is calculated based on the second target current data and the second target voltage data.
Specifically, a top layer circuit board, a middle layer circuit board and a bottom layer circuit board in a multilayer circuit board are obtained, and the multilayer circuit board is used for Bluetooth communication. The top circuit board is generally responsible for receiving and transmitting signals, while the middle and bottom circuit boards handle more data processing and power management functions. In order to effectively manage power distribution and consumption, first and second power converters are provided between the top and middle circuit boards and between the middle and bottom circuit boards, respectively. Each circuit board adjusts the power supply according to actual requirements, thereby optimizing the power consumption of the whole system. And monitoring the current and the voltage of the first power converter to obtain first initial current data and first initial voltage data, and monitoring the current and the voltage of the second power converter to obtain second initial current data and second initial voltage data. The first and second power converters are monitored to help the system discover any signs of anomalies or efficiency degradation in time. And carrying out principal component analysis on the initial data, extracting useful information and simplifying the data processing process. Principal component analysis reduces the complexity of the data by identifying major variables and patterns in the data, thereby more accurately identifying key factors that affect circuit board performance. For example, through principal component analysis, certain specific current and voltage patterns may be found to be directly related to system performance degradation, which can be used to adjust the mode of operation of the power converter to improve efficiency. After the principal component feature analysis is completed, first target current data and first target voltage data, and second target current data and second target voltage data are obtained based on the analysis result. Using these data, power data sets of the first and second power converters are calculated, which data sets reflect the power consumption characteristics of each power converter under certain operating conditions. For example, it is assumed that the power consumption requirement of the top circuit board increases due to the increase of applications when bluetooth communication is performed. The first power converter adjusts according to the first target current and voltage data after receiving the increased current and voltage demand signal to meet the power consumption demand of the top layer without overload. Similarly, if the middle circuit board processes more data, the second power converter will also adjust based on the second target current and voltage data, ensuring that the power supply is both satisfactory and maximally energy efficient.
In a specific embodiment, the executing step performs principal component feature analysis on the first initial current data and the first initial voltage data to obtain first target current data and first target voltage data, and performs principal component feature analysis on the second initial current data and the second initial voltage data to obtain second target current data and second target voltage data, which may specifically include the following steps:
(1) Respectively carrying out feature extraction on the first initial current data and the first initial voltage data to obtain a plurality of first initial current features and a plurality of first initial voltage features, and respectively carrying out feature extraction on the second initial current data and the second initial voltage data to obtain a plurality of second initial current features and a plurality of second initial voltage features;
(2) Respectively carrying out linear characteristic transformation on a plurality of first initial current characteristics and a plurality of first initial voltage characteristics to obtain a first current linear characteristic set and a first voltage linear characteristic set, and respectively carrying out linear characteristic transformation on a plurality of second initial current characteristics and a plurality of second initial voltage characteristics to obtain a second current linear characteristic set and a second voltage linear characteristic set;
(3) Constructing a first current covariance matrix and a first voltage covariance matrix which correspond to the first current linear feature set and the first voltage linear feature set, and constructing a second current covariance matrix and a second voltage covariance matrix which correspond to the second current covariance matrix and the second voltage covariance matrix which correspond to the second current linear feature set and the second voltage linear feature set;
(4) Performing characteristic correlation calculation on the first current covariance matrix and the first voltage covariance matrix to obtain a first current characteristic correlation and a first voltage characteristic correlation, and performing characteristic correlation calculation on the second current covariance matrix and the second voltage covariance matrix to obtain a second current characteristic correlation and a second voltage characteristic correlation;
(5) According to the first current characteristic correlation and the first voltage characteristic correlation, respectively calculating a first current principal component characteristic weight and a first voltage principal component characteristic weight corresponding to the first current linear characteristic set and the first voltage linear characteristic set, and according to the second current characteristic correlation and the second voltage characteristic correlation, respectively calculating a second current principal component characteristic weight and a second voltage principal component characteristic weight corresponding to the second current linear characteristic set and the second voltage linear characteristic set;
(6) And carrying out feature fusion on a plurality of first initial current features and a plurality of first initial voltage features according to the first current main component feature weights and the first voltage main component feature weights to obtain first target current data and first target voltage data, and carrying out feature fusion on a plurality of second initial current features and a plurality of second initial voltage features according to the second current main component feature weights and the second voltage main component feature weights to obtain second target current data and second target voltage data.
Specifically, feature extraction is performed on the initial current and voltage data collected by the first power converter and the second power converter. The current and voltage values of each converter are measured to capture the behavior of the converter under its operating conditions, such as peak, average, ripple, and minimum values of current and voltage, etc. The initial features are transformed linearly, usually by Principal Component Analysis (PCA) or the like. The primary features are extracted from the plurality of correlated original features to reduce the dimensionality of the data set while retaining the most important information. For example, if the raw data shows a high correlation between current and voltage characteristics, principal component analysis can help identify key factors that affect system performance, thereby simplifying the overall feature set. And constructing covariance matrixes of the current and the voltage. The covariance matrix helps to understand the linear relationship between the different features, i.e. which features have a strong correlation with other features. And calculating the eigenvalue and eigenvector of the covariance matrix to obtain the correlation of each feature. The high or low level of feature correlation represents the importance of each feature to the power consumption behavior of the circuit board. Features of high correlation will be given higher weights. And weighting and fusing all the features according to the weights calculated by the correlation of the features to form final target current data and target voltage data. Fusion ensures that the most important features are highlighted, while less critical features relatively reduce their impact. For example, if the principal component analysis shows that a particular current change pattern of the first power converter is highly correlated to a decrease in system performance, this feature will be weighted higher at the time of final fusion.
In a specific embodiment, the process of executing step 102 may specifically include the following steps:
(1) Performing curve fitting on the first power data set to obtain a first power curve, and performing curve fitting on the second power data set to obtain a second power curve;
(2) Extracting features of the first power curve to obtain a plurality of first initial power features, and extracting features of the second power curve to obtain a plurality of second initial power features;
(3) Performing feature classification on the plurality of first initial power features to obtain a plurality of first high-frequency power features and a plurality of first low-frequency power features, and performing feature classification on the plurality of second initial power features to obtain a plurality of second high-frequency power features and a plurality of second low-frequency power features;
(4) And performing feature weight calculation and feature set conversion on the plurality of first high-frequency power features and the plurality of first low-frequency power features to obtain a first power feature set, and performing feature weight calculation and feature set conversion on the plurality of second high-frequency power features and the plurality of second low-frequency power features to obtain a second power feature set.
In particular, curve fitting is performed on the first and second power data sets. The actual observed data points are approximated using a mathematical model, such as a polynomial regression or a nonlinear regression model. The purpose of curve fitting is to create a mathematical expression that describes the trend of power data over time or other variables. For example, if the first power data set shows a gradual increase in power consumption over time, the curve fit may reveal a linearly increasing or more complex exponential growth pattern. Similarly, the fitting of the second power data set may reveal different patterns depending on the specific power consumption behavior of different parts of the circuit board or different functional modules. And extracting features of the first power curve to obtain a plurality of first initial power features, and extracting features of the second power curve to obtain a plurality of second initial power features. Significant features such as peaks, valleys, averages, fluctuations, etc. are identified from the fitted curve. These features help to understand the power consumption dynamics of the circuit board more deeply. For example, a first power curve may show sudden increases in power consumption under certain conditions, which may be extracted as a key feature, while a second power curve may reveal periodic changes in power consumption. After feature extraction, these features are classified, typically into high frequency and low frequency power features. High frequency features are focused on transient behavior where power consumption changes rapidly, while low frequency features are focused on trends over longer time scales. And performing feature weight calculation and feature set conversion. The weight is assigned by analyzing the importance of each feature to the overall power consumption impact. Features of high influence will get higher weights reflecting their importance in power consumption optimization. For example, if a certain high frequency current characteristic is closely related to a decrease in system performance, the weight of this characteristic in the first set of power characteristics will be increased. And converting the initially extracted features into more comprehensive and more influential power feature sets through weights to obtain a first power feature set and a second power feature set, and guiding adjustment and optimization of the power converter, so that the overall optimization of the power consumption of the circuit board is realized.
In a specific embodiment, the performing step performs feature weight calculation and feature set conversion on the plurality of first high-frequency power features and the plurality of first low-frequency power features to obtain a first power feature set, and performs feature weight calculation and feature set conversion on the plurality of second high-frequency power features and the plurality of second low-frequency power features to obtain a second power feature set, and the process of obtaining the second power feature set may specifically include the following steps:
(1) The method comprises the steps of performing pearson correlation coefficient calculation on a plurality of first high-frequency power features and a plurality of first low-frequency power features to obtain a first pearson correlation coefficient, and performing pearson correlation coefficient calculation on a plurality of second high-frequency power features and a plurality of second low-frequency power features to obtain a second pearson correlation coefficient;
(2) Setting first characteristic weight values of a plurality of first high-frequency power characteristics and a plurality of first low-frequency power characteristics according to a first pearson correlation coefficient, and setting second characteristic weight values of a plurality of second high-frequency power characteristics and a plurality of second low-frequency power characteristics according to a second pearson correlation coefficient;
(3) Performing feature weighting on the first high-frequency power features and the first low-frequency power features according to the first feature weight values to obtain first high-frequency weighted features and first low-frequency weighted features, and performing feature weighting on the second high-frequency power features and the second low-frequency power features according to the second feature weight values to obtain second high-frequency weighted features and second low-frequency weighted features;
(4) A corresponding first set of power characteristics is generated from the plurality of first high frequency weighted characteristics and the plurality of first low frequency weighted characteristics, and a corresponding second set of power characteristics is generated from the plurality of second high frequency weighted characteristics and the plurality of second low frequency weighted characteristics.
Specifically, pearson correlation coefficient calculations are performed on the high frequency and low frequency power characteristics in the first and second power data sets. The pearson correlation coefficient is a statistical indicator that measures the strength of a linear relationship between two variables, ranging in value from-1 to 1, where 1 represents a complete positive correlation, -1 represents a complete negative correlation, and 0 represents no linear correlation. For example, for the first power data set, it is assumed that a high correlation of the rapid change of current (high frequency characteristic) with the stability of voltage (low frequency characteristic) is observed in the high power consumption mode. This correlation may indicate that the stable voltage of the circuit in the high power consumption state is critical to maintaining device performance. And setting corresponding weights for each high-frequency and low-frequency power characteristic through the calculated pearson correlation coefficient. These weights reflect the importance of the features in the overall power consumption behavior of the circuit. For example, if a certain high frequency current variation characteristic is highly correlated to a decrease in circuit performance, higher weights may be given because such a characteristic helps predict and avoid potential performance problems. And weighting the first and second power characteristics according to the set weight. By multiplying each feature by its corresponding weight, features that have a greater impact on circuit performance are emphasized while the impact of those less important features is reduced, resulting in weighted first and second high frequency weighted features and low frequency weighted features. The weighted features are combined into a comprehensive power feature set. All weighted features are integrated to form a comprehensive set of power features representative of the first and second power converters. For example, combining the high frequency weighting features and the low frequency weighting features in the first set of power features provides a comprehensive view of how the performance of the first power converter is affected by various factors under different operating conditions. Likewise, the second set of power characteristics is constructed in a similar manner to ensure that each power converter operates to achieve optimal energy efficiency and performance.
In a specific embodiment, the process of executing step 103 may specifically include the following steps:
(1) Performing feature coding on the first power feature set and the second power feature set respectively to obtain a first power feature vector and a second power feature vector;
(2) Performing power optimization execution strategy analysis on the first power feature vector and the second power feature vector through a dual Q network to obtain a first initial power consumption optimization execution strategy of the first power converter and a second initial power consumption optimization execution strategy of the second power converter;
(3) The method comprises the steps of performing bonus feedback parameter calculation on a first power converter and a second power converter to obtain a first bonus feedback parameter of the first power converter and a second bonus feedback parameter of the second power converter;
(4) Performing strategy updating on the first initial power consumption optimization execution strategy according to the first rewarding feedback parameter to obtain a first target power consumption optimization execution strategy, and performing strategy updating on the second initial power consumption optimization execution strategy according to the second rewarding feedback parameter to obtain a second target power consumption optimization execution strategy;
(5) And performing power consumption parameter calculation on the multi-layer circuit board according to the first target power consumption optimization execution strategy and the second target power consumption optimization execution strategy through a PSO algorithm to obtain global power consumption optimization parameters of the multi-layer circuit board.
Specifically, feature coding is performed on the first power feature set and the second power feature set respectively, so that a first power feature vector and a second power feature vector are obtained. For example, data of power consumption peaks, average power consumption, fluctuation, etc. are converted into a series of values, which can accurately describe the operation state of the power converter. If the first power converter exhibits a power consumption peak at high load, this information will be encoded into the first power feature vector. And analyzing the encoded power characteristic vector through a double Q network to form an initial power consumption optimization execution strategy. The dual Q network is a reinforcement learning algorithm that uses two value function estimators to reduce the bias from overestimation, thereby providing a more stable and accurate learning result. And analyzing each power characteristic vector to predict the influence of different adjustment strategies on power consumption and generating an initial power consumption optimization execution strategy of each converter. In order to further improve the effectiveness of the strategy, bonus feedback parameter calculation is performed. And evaluating the performance of each strategy according to the comparison of the actual operation data and the strategy prediction result. For example, if the first power converter has significantly reduced power consumption and increased stability after implementing a certain strategy, the strategy may obtain a higher bonus feedback parameter. These parameters directly affect the adjustment and optimization of the subsequent strategy. And carrying out strategy updating according to the rewarding feedback parameters. And the dual Q network adjusts a value function evaluator of the dual Q network according to the obtained feedback, and optimizes an original power consumption optimization execution strategy. The strategy of each converter is adjusted stepwise according to its actual performance to ensure that higher efficiency and lower power consumption can be achieved in future operation. And performing global power consumption parameter calculation on the whole multi-layer circuit board according to the updated first and second target power consumption optimization execution strategies through a PSO algorithm. PSO is a population intelligent algorithm, which simulates the searching behavior of a bird population to find the optimal solution. In this process, the algorithm will find the optimal combination of parameters in the parameter space that minimizes the overall power consumption.
In a specific embodiment, the process of executing step 106 may specifically include the following steps:
(1) Generating a random initial value of the first power converter according to a first target power consumption optimization execution strategy through a PSO algorithm to obtain a first random initial value set, and generating a random initial value of the second power converter according to a second target power consumption optimization execution strategy to obtain a second random initial value set;
(2) Carrying out particle population construction on the first random initial value set through a reverse particle propagation algorithm to obtain a first particle population, and carrying out particle population construction on the second random initial value set through the reverse particle propagation algorithm to obtain a second particle population;
(3) Performing population segmentation on the first particle population to obtain a plurality of first particle populations, and performing population segmentation on the second particle population to obtain a plurality of second particle populations;
(4) Respectively carrying out fitness calculation on a plurality of first particle populations to obtain a first particle fitness set corresponding to the first particle populations, and respectively carrying out fitness calculation on a plurality of second particle populations to obtain a second particle fitness set corresponding to the second particle populations;
(5) Generating a first power consumption optimization parameter of the first power converter according to the first particle fitness set, and generating a second power consumption optimization parameter of the second power converter according to the second particle fitness set;
(6) And performing global optimization on the first power consumption optimization parameter and the second power consumption optimization parameter to obtain the global power consumption optimization parameter of the multi-layer circuit board.
Specifically, a random initial value is generated for the first and second power converters by a PSO algorithm. The PSO algorithm starts its search process with a set of random solutions that are optimized by modeling the social behavior of the bird population during the iterative process of the algorithm. In this embodiment, each solution represents a set of possible power converter parameters, such as current limits, voltage settings, etc. For example, the first power converter may operate over a range of powers, and the set of random initial values generated by the PSO algorithm will cover various possible settings within this range. Particle population construction is performed on the randomly generated initial values using a reverse particle propagation algorithm. Each initial value is considered to be a particle and the whole collection forms a population of particles. Counter particle propagation is a process of adjusting particle position and velocity by iterative algorithm to find a better solution. For example, if a parameter setting represented by a certain particle results in a reduced power consumption while the performance remains unchanged, such a setting may be pushed in a more optimal direction in subsequent iterations. Particle population segmentation is performed to form a plurality of sub-particle populations, each of which is focused on a different portion of the search solution space. Population segmentation increases the likelihood of finding a globally optimal solution, preventing the algorithm from focusing prematurely on a locally optimal solution. Each sub-population performs independent search, but shares information of the globally optimal solution, thereby guiding the search process. And respectively carrying out fitness calculation on the plurality of first particle populations to obtain a first particle fitness set corresponding to the first particle populations, and respectively carrying out fitness calculation on the plurality of second particle populations to obtain a second particle fitness set corresponding to the second particle populations. Fitness calculations are typically based on the extent of power consumption reduction, improvement in system stability, and possibly performance gains. These fitness values provide a quantitative benefit assessment for each particle and its parameter setting, which is used to guide the final parameter selection. Based on the fitness sets of each particle population, optimal power consumption parameters for the first and second power converters are generated. These parameters reflect the best performing settings throughout the search. These parameters are applied to global optimization and all the optimal solutions are integrated to form a comprehensive power consumption optimization scheme for the multi-layer circuit board. Global optimization takes into account all relevant performance metrics and operational constraints to ensure that the circuit board can run with minimal power consumption under all operating conditions.
The method for managing adaptive power consumption of a bluetooth circuit board in the embodiment of the present application is described above, and the following describes a system for managing adaptive power consumption of a bluetooth circuit board in the embodiment of the present application, referring to fig. 2, an embodiment of the system for managing adaptive power consumption of a bluetooth circuit board in the embodiment of the present application includes:
the monitoring module 201 is configured to set power converters at different levels of the multi-layer circuit board and perform real-time power consumption monitoring to obtain a first power data set of the first power converter and a second power data set of the second power converter;
an extracting module 202, configured to extract power features of the first power data set and the second power data set, to obtain a first power feature set and a second power feature set;
a creating module 203, configured to create global power consumption optimization parameters of the multi-layer circuit board according to the first power feature set and the second power feature set through the dual Q network and the PSO algorithm.
Through the cooperation of the components, the power consumption data of each layer of circuit board can be accurately obtained through setting the power converters at different levels of the multi-layer circuit board and performing real-time monitoring. This real-time monitoring ensures maximum efficiency of energy utilization, enabling the device to dynamically adjust power output according to the current operating conditions, thereby reducing inefficient energy consumption and extending the battery life of the device. And key features are effectively extracted from the original current and voltage data by applying a principal component analysis feature extraction technology. The deep data analysis method helps to more accurately understand the behaviors of the circuit board under different working conditions, and provides data support for subsequent power consumption optimization. By utilizing the dual Q network and the PSO algorithm, the method not only can create high-efficiency global power consumption optimization parameters, but also can dynamically adjust and optimize the power consumption configuration of the circuit board through the algorithm. The intelligent decision process remarkably improves the energy efficiency of the circuit board and reduces the energy waste caused by fixed or outdated power consumption configuration. By combining calculation of the reward feedback parameters and policy updating, the method realizes a self-learning power consumption management system. Through continuous learning and adjustment, the circuit board can constantly optimize its power consumption strategy in order to adapt to environmental change and equipment demand, promotes holistic performance and efficiency. By comprehensively considering the power consumption characteristics of each layer of circuit board and optimizing the execution strategy, the method can implement power consumption management on the whole circuit board level, thereby realizing more unified and coordinated energy management. The overall view management not only improves the energy efficiency, but also helps to keep the stability and reliability of the equipment, thereby realizing the self-adaptive power consumption management of the Bluetooth circuit board and improving the power consumption optimization effect of the multi-layer circuit board.
The present application also provides a computer device, where the computer device includes a memory and a processor, where the memory stores computer readable instructions that, when executed by the processor, cause the processor to execute the steps of the bluetooth circuit board adaptive power consumption management method in the foregoing embodiments.
The present application also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, and may also be a volatile computer readable storage medium, where instructions are stored in the computer readable storage medium, where the instructions, when executed on a computer, cause the computer to perform the steps of the bluetooth circuit board adaptive power consumption management method.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, systems and units may refer to the corresponding processes in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. The storage medium includes a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk or an optical disk, etc. which can store the program code.
While the application has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that the foregoing embodiments may be modified or equivalents may be substituted for some of the features thereof, and that the modifications or substitutions do not depart from the spirit and scope of the embodiments of the application.