Energy consumption prediction and optimization method and related device of smart card chip

Abstract

The present application relates to a method and related device for predicting and optimizing energy consumption of a smart card chip. The method includes: obtaining the operation data based on the target working mode collected by the smart card chip within the target time window, performing time slice division and clustering processing on the operation data, and obtaining a multi-dimensional feature vector within the target time window; obtaining a trained energy consumption prediction model, inputting the multi-dimensional feature vector into the energy consumption prediction model for processing, and predicting a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip; determining each energy consumption unit of the smart card chip, and optimizing the energy consumption parameters based on the correlation between each energy consumption unit and the dynamic energy consumption distribution diagram to obtain the optimized energy consumption parameters. The present method can efficiently and effectively predict and optimize the energy consumption of smart card chips.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method for predicting and optimizing energy consumption of a smart card chip and related devices. Background Art

[0002] In the field of artificial intelligence technology, the performance and energy consumption of smart card chips in a wide range of application scenarios directly affect the reliability, stability and service life of the system. Based on this, how to accurately predict the energy consumption of smart card chips to achieve more efficient energy management and optimization has become a key problem that needs to be solved urgently.

[0003] However, related smart card chip energy consumption prediction methods usually rely on relatively single statistical models or traditional algorithms, and cannot fully utilize the complex dynamic data generated during the operation of the smart card chip. This makes it difficult to efficiently and effectively predict the energy consumption of the smart card chip, and it is difficult to effectively optimize the energy consumption of the smart card chip. Summary of the invention

[0004] Based on this, it is necessary to provide a smart card chip energy consumption prediction and optimization method, system, computer device and computer-readable storage medium to address the above technical problems, so as to efficiently and effectively predict and optimize the energy consumption of the smart card chip.

[0005] In a first aspect, the present application provides a method for predicting and optimizing energy consumption of a smart card chip, comprising:

[0006] Acquire operation data based on a target working mode collected by a smart card chip within a target time window, perform time slice division processing and clustering processing on the operation data, and obtain a multi-dimensional feature vector within the target time window;

[0007] Acquire a trained energy consumption prediction model, input the multi-dimensional feature vector into the energy consumption prediction model for processing, predict and obtain a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip, the dynamic energy consumption distribution diagram reflects the energy consumption timing characteristics of the smart card chip in the target working mode, and process the multi-dimensional feature vector including recursively updating the features of the multi-dimensional feature vector based on the correlation of the time series, obtaining the timing characteristics corresponding to each time point of the target time window, and mapping to obtain the timing characteristics corresponding to each time point of the future time period;

[0008] Determine each energy consumption unit of the smart card chip, and based on the association between each energy consumption unit and the dynamic energy consumption distribution diagram, optimize the energy consumption parameters to obtain optimized energy consumption parameters. The energy consumption unit reflects the energy consumption performance of each functional module in the smart card chip in the target working mode.

[0009] In a second aspect, the present application also provides an energy consumption prediction and optimization system for a smart card chip, comprising:

[0010] An acquisition module is used to acquire operation data based on a target working mode collected by a smart card chip within a target time window, and perform time slice division and clustering processing on the operation data to obtain a multi-dimensional feature vector within the target time window;

[0011] A prediction module, used to obtain a trained energy consumption prediction model, input the multi-dimensional feature vector into the energy consumption prediction model for processing, and predict a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip, wherein the dynamic energy consumption distribution diagram reflects the energy consumption timing characteristics of the smart card chip in the target working mode;

[0012] The optimization module is used to determine the various energy consumption units of the smart card chip, and based on the correlation between the various energy consumption units and the dynamic energy consumption distribution diagram, optimize the energy consumption parameters to obtain optimized energy consumption parameters. The energy consumption units reflect the energy consumption performance of each functional module in the smart card chip in the target working mode.

[0013] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above steps when executing the computer program.

[0014] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above steps when executed by a processor.

[0015] The energy consumption prediction and optimization method, system, computer device and computer-readable storage medium of the smart card chip mentioned above firstly divide the operation data of the smart card chip into time slices, so as to accurately capture the changing characteristics of the operation data in a short time within a smaller and finer-grained time range, and then cluster the operation data after the time slice division, so as to effectively convert the operation data into a multi-dimensional feature vector covering multi-dimensional information; secondly, the multi-dimensional feature vector is processed by the energy consumption prediction model to accurately predict the energy consumption parameters of the smart card chip, and the visual data expression is realized through the dynamic energy consumption distribution diagram; thirdly, based on the correlation between each energy consumption unit of the smart card chip and the dynamic energy consumption distribution diagram, the energy consumption parameters are optimized in a targeted manner in the dimensions of different software and hardware functional modules to obtain the optimized energy consumption parameters; based on this, the energy consumption parameters are efficiently and accurately predicted according to the operation data of the smart card chip, and the predicted energy consumption parameters are adaptively and effectively optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A schematic diagram of a process of predicting and optimizing the energy consumption of a smart card chip in one embodiment;

[0018] Figure 2 The figure is a structural block diagram of an energy consumption prediction and optimization system for a smart card chip in one embodiment. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] In one embodiment, Figure 1 As shown, a method for predicting and optimizing energy consumption of a smart card chip is provided. This embodiment uses the method applied to a server as an example. It can be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps S101 to S103.

[0021] Step S101, obtaining the operation data based on the target working mode collected by the smart card chip within the target time window, performing time slice division processing and clustering processing on the operation data, and obtaining a multi-dimensional feature vector within the target time window.

[0022] Among them, the smart card chip refers to a microprocessor chip integrated in a smart card, which is used to provide a secure and reliable computing environment and efficient data processing capabilities, such as the security chip in a financial payment card.

[0023] Among them, the target time window represents the specific time range for collecting the operation data of the smart card chip, such as collecting the operation data corresponding to the current period or a historical period; the operation data represents the performance or operation-related data generated by the smart card chip during operation, such as processor load, memory usage, communication rate and other data; the operation data based on the target working mode represents the operation data generated by the smart card chip in a specific task scenario, which is used to reflect the dynamic performance of the smart card chip in a specific task scenario, such as the dynamic performance in task scenarios such as data encryption and decryption, identity authentication, storage reading and writing, and data transmission.

[0024] Among them, the multi-dimensional feature vector represents a high-dimensional data structure formed by time slice division and clustering of the operation data of the smart card chip, which is used to comprehensively represent the multi-dimensional operation characteristics of the smart card chip within the target time window, such as information containing multiple dimensions such as CPU utilization, input / output rate, and current fluctuation.

[0025] Exemplarily, the operation data of the smart card chip within the target time window can be collected through the internal sensors of the smart card chip or the external monitoring equipment. The operation data includes multi-dimensional indicators of the smart card chip in a specific working mode, such as processor load, memory usage, input and output operation frequency, current and voltage fluctuations, etc.; the operation data is divided into time slices, that is, the operation data of the entire target time window is decomposed into data in several time slices of equal length, so as to ensure that the data in each time slice can reflect the operation status of the smart card chip within a smaller and fine-grained time range; the data in each time slice is clustered, for example, the core information is extracted from the data in each time slice through statistical features such as mean, extreme value, standard value, etc., so as to form a feature vector covering multi-dimensional information in the target time window, that is, a multi-dimensional feature vector.

[0026] Step S102, obtain the trained energy consumption prediction model, input the multi-dimensional feature vector into the energy consumption prediction model for processing, predict and obtain the dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip, the dynamic energy consumption distribution diagram reflects the energy consumption timing characteristics of the smart card chip in the target working mode, and the processing of the multi-dimensional feature vector includes recursive updating of the features of the multi-dimensional feature vector based on the correlation of the time series, obtaining the timing characteristics corresponding to each time point of the target time window, and mapping the timing characteristics corresponding to each time point of the future time period.

[0027] Among them, the energy consumption prediction model represents a mathematical model built based on a machine learning or deep learning algorithm, which is used to learn the nonlinear relationship between the operating data and energy consumption of the smart card chip.

[0028] The energy consumption parameters represent quantitative indicators related to energy consumption during the operation of the smart card chip, such as energy consumption per unit time, cumulative energy consumption, energy utilization rate, etc.

[0029] The dynamic energy consumption distribution diagram represents a graphical data representation generated according to the predicted energy consumption parameters of the smart card chip, and is used to show the changing trend of the energy consumption of the smart card chip over time in the future period.

[0030] Exemplarily, the multi-dimensional feature vector is input into the energy consumption prediction model, which parses each dimension of the multi-dimensional feature vector layer by layer through an internal computing network, and finally outputs the prediction result corresponding to the energy consumption parameter in the future time period through the mapping relationship between the input features and the output energy consumption parameters learned in the training phase. The prediction result is represented by a graphical data through a dynamic energy consumption distribution diagram. For example, in the dynamic energy consumption distribution diagram, time is used as the horizontal axis and the energy consumption parameter is used as the vertical axis, which reflects the energy consumption change trend of the smart card chip in the future time period.

[0031] Exemplarily, in the process of processing the multidimensional feature vector, the features of the multidimensional feature vector at each time point in the target time window are recursively updated in combination with the correlation of the time series, so as to fully capture the dynamic change trend of the features in the time dimension. For example, the feature state of the multidimensional feature vector at each time point in the target time window is used as input, and combined with time series models such as recurrent neural network (RNN) or long short-term memory network (LSTM), the features are gradually adjusted to generate the time series features of each time point in the target time window; furthermore, through the time series mapping method, the time series features of the target time window are projected to the future time period, so as to generate the time series features of each time point in the future time period, so that the feature data can be extended to the prediction stage, so as to achieve the effective deduction and capture of the feature trend of the future time period.

[0032] Step S103, determine each energy consumption unit of the smart card chip, and optimize the energy consumption parameters based on the correlation between each energy consumption unit and the dynamic energy consumption distribution diagram to obtain optimized energy consumption parameters. The energy consumption unit reflects the energy consumption performance of each functional module in the smart card chip in the target working mode.

[0033] Among them, the energy consumption unit represents the quantitative expression of the energy consumption characteristics of each functional module in the smart card chip under the target working mode. It is the basic analysis unit of the energy consumption of the smart card chip, such as the energy consumption characteristics of storage and retrieval of data corresponding to the storage module (such as energy consumption when writing data, energy consumption when reading data, the impact of data access rate on energy consumption, etc.), the energy consumption characteristics of communication connection and data transmission corresponding to the communication module (such as energy consumption of sending and receiving signals, the impact of data transmission rate and transmission distance on energy consumption, etc.), and the energy consumption characteristics of data encryption corresponding to the encryption module (such as the impact of data volume and algorithm complexity on energy consumption, etc.).

[0034] Exemplarily, according to the static energy consumption characteristics of each functional module of the smart card chip under the target working mode, the energy consumption units corresponding to each functional module are determined, and based on the changing trends of different energy consumption parameters of the smart card chip in the future time period shown in the dynamic energy consumption distribution diagram, the contribution of each energy consumption unit to the matching energy consumption parameters is identified, thereby determining the correlation between each energy consumption unit and the energy consumption parameters in the dynamic energy consumption distribution diagram; further, based on the correlation between each energy consumption unit and the energy consumption parameters in the dynamic energy consumption distribution diagram, the energy consumption parameters corresponding to each energy consumption unit are optimized to obtain the optimized energy consumption parameters.

[0035] In the above-mentioned smart card chip energy consumption prediction and optimization method, first, the operation data of the smart card chip is divided into time slices, so as to accurately capture the change characteristics of the operation data in a short time within a smaller and finer-grained time range, and then the operation data after the time slice division is clustered, so as to effectively convert the operation data into a multi-dimensional feature vector covering multi-dimensional information; secondly, the multi-dimensional feature vector is processed by the energy consumption prediction model to accurately predict the energy consumption parameters of the smart card chip, and the visual data expression is realized through the dynamic energy consumption distribution diagram; thirdly, based on the correlation between each energy consumption unit of the smart card chip and the dynamic energy consumption distribution diagram, the energy consumption parameters are optimized in a targeted manner in the dimensions of different software and hardware functional modules to obtain the optimized energy consumption parameters; based on this, the energy consumption parameters are efficiently and accurately predicted according to the operation data of the smart card chip, and the predicted energy consumption parameters are adaptively and effectively optimized.

[0036] In an exemplary embodiment, time slice division and clustering are performed on the operation data to obtain a multi-dimensional feature vector within a target time window, including steps S201 to S202.

[0037] Step S201, divide the operating data within the target time window into time slices according to a preset time slice length to obtain multiple time slice data. The time slice length is determined according to the working mode and data sampling frequency of the smart card chip. The time slice data includes the operation instruction data, power consumption measurement data and working status data of the smart card chip corresponding to the time slice length.

[0038] Among them, the operation instruction data represents a set of task instructions executed by the smart card chip within the time slice, such as encryption operation instructions, data transmission instructions or storage access instructions executed by the smart card chip within the time slice; the power consumption measurement data represents the energy information consumed by the smart card chip within the time slice, such as power, current and voltage; the working status data represents the operating status of the smart card chip within the time slice, such as high-load operating status, low-load operating status, etc.

[0039] Exemplarily, all operating data within the target time window are divided according to a preset time slice length to form multiple continuous and equal-length time slice data. Each time slice data represents a local performance of the operating status of the smart card chip within the time slice length, which contains multi-dimensional information describing the operating status, such as operation instruction data, power consumption measurement data, and working status data.

[0040] Optionally, under high-load operating conditions, the operating data of the smart card chip changes rapidly, or the operating data collected at a high sampling frequency has a lower time accuracy, so a smaller time slice length can be set to capture rapid dynamic changes in the operating data; while under low-load operating conditions, the operating data of the smart card chip changes more smoothly, or the operating data collected at a low sampling frequency has a higher time accuracy, so a larger time slice length can be set to reduce unnecessary computing consumption.

[0041] Step S202, perform feature extraction processing on multiple time slice data respectively to obtain feature information corresponding to the multiple time slice data respectively, perform clustering processing on feature information with similar features based on a clustering analysis algorithm to obtain a multi-dimensional feature vector within the target time window.

[0042] The characteristic information corresponding to the time slice data is used to quantitatively describe the data characteristics of different dimensions such as the operation instruction data, power consumption measurement data, and working status data of the smart card chip within the time slice.

[0043] The cluster analysis algorithm refers to an algorithm for clustering and analyzing data to efficiently identify data with similar features, such as the K-means clustering algorithm.

[0044] Exemplarily, the characteristic information corresponding to the operation instruction data is used to quantitatively describe the complexity, frequency and efficiency of the smart card chip executing tasks within the time slice, for example, statistical characteristics such as the sum, average value, and maximum value of the number of instructions, and dynamic characteristics such as the frequency change of instruction calls or the dynamic ratio of instruction type distribution; the characteristic information corresponding to the power consumption measurement data is used to quantitatively describe the operating energy consumption of the smart card chip within the time slice, for example, statistical characteristics such as average power consumption, maximum power consumption, and standard deviation of power consumption, and dynamic characteristics such as frequency components of power consumption signals and power consumption fluctuations within a cycle; the characteristic information corresponding to the working state data is used to quantitatively describe the state changes and switching frequency of the smart card chip within the time slice, for example, statistical characteristics such as the duration of a certain state and the total number of state switches, and dynamic characteristics such as the frequency distribution of state switches.

[0045] Exemplarily, feature extraction is performed on multiple time slice data respectively to obtain feature information corresponding to each time slice data in different dimensions, and the feature information is clustered based on a clustering analysis algorithm. For example, in a K-means clustering algorithm, feature information of different dimensions corresponding to the same time slice data is clustered according to the Euclidean distance to integrate feature information of similar features in the time slice data. Based on this, an integrated feature vector corresponding to the time slice data is obtained. Based on the integrated feature vectors corresponding to each time slice data, a multi-dimensional feature vector within the target time window is obtained.

[0046] Optionally, based on the strong correlation between high-frequency instructions and power consumption, relevant features in the operation instruction data dimension (such as the frequency and number of operation instructions, etc.) and relevant features in the energy consumption measurement data dimension (such as the power consumption fluctuation range, etc.) can be used as feature information of similar features; based on the strong correlation between power consumption and working state switching, relevant features in the energy consumption measurement data dimension (such as the maximum and minimum power consumption values, etc.) and relevant features in the working state data dimension (such as the working state switching time, switching frequency, etc.) can be used as feature information of similar features.

[0047] In this embodiment, firstly, based on the time slice length dynamically determined by the working mode of the smart card chip and the data sampling frequency, the operating data within the target time window is adaptively divided into time slices, so as to effectively obtain multiple time slice data; secondly, effective feature information is obtained by performing feature extraction processing on the time slice data, and then the feature information of similar features is clustered, so as to obtain a multi-dimensional feature vector within the target time window based on multi-dimensional data integration to efficiently extract.

[0048] In an exemplary embodiment, the multi-dimensional feature vector is input into the energy consumption prediction model for processing, and a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip is predicted, including steps S301 to S302.

[0049] Step S301, input the multi-dimensional feature vector into the energy consumption prediction model, recursively calculate the multi-dimensional feature vector according to the time series, and obtain the time series feature mapping results of each expected time point within the expected time window. The time series feature mapping results include the feature vector change information and feature vector time association information corresponding to each expected time point, and the starting time point of the expected time window matches the ending time point of the target time window.

[0050] Among them, the time series feature mapping result represents the mapping result generated by combining the multi-dimensional feature vector with the time series characteristics through the energy consumption prediction model, which is used to describe the dynamic change and time dependence of the feature vector at each expected time point within the expected time window; the expected time window represents the time range for predicting the energy consumption of the smart card chip in the future period, that is, extending from the end time point of the target time window to the preset time period in the future; the expected time point represents each time point within the expected time window.

[0051] Among them, the feature vector change information represents the dynamic changes of the feature vectors at each expected time point in the time series, such as the increase, decrease or fluctuation characteristics of the feature value. For example, from the expected time point t1 to the expected time point t2, the CPU load increases from 30% to 70%, and the current fluctuation increases from 0.5A to 0.8A. These changes constitute the feature vector change information; the feature vector time correlation information represents the correlation between the changes of the feature vectors at each expected time point and the time series, such as the feature changes at a specific time position. For example, a feature suddenly increases at time point t1 and stabilizes at time point t2. This time correlation reflects the temporal correlation effect of a specific task or event on the energy consumption of the smart card chip.

[0052] Exemplarily, based on an algorithm model with time series processing capabilities, such as a recurrent neural network (RNN) or a long short-term memory network (LSTM), the multi-dimensional feature vector is recursively calculated according to the time series, that is, by processing the input features at each time point, the correlation of the input features in the time dimension is captured, thereby generating the time series feature mapping results of each expected time point in the expected time window; furthermore, these time series feature mapping results include feature vector change information and feature vector time association information, the feature vector change information reflects the dynamic change trend of the feature in the time dimension, and the feature vector time association information reflects the mapping relationship between the feature change and the time point; furthermore, the starting time point of the expected time window matches the ending time point of the target time window, thereby ensuring the continuity and logical consistency of the prediction time range.

[0053] Step S302, based on the time series feature mapping result, the energy consumption value corresponding to each expected time point is calculated, and the energy consumption values ​​corresponding to each expected time point are combined in chronological order to generate a dynamic energy consumption distribution diagram corresponding to the expected time window.

[0054] Among them, the energy consumption value at the expected time point represents the energy consumption of the smart card chip at the corresponding expected time point, that is, it is used to quantitatively describe the energy consumption performance of the smart card chip at each moment during the operation of the expected time window, such as the real-time power consumption at a certain expected time point (i.e., instantaneous power consumption), the total energy consumption accumulated from the expected time window (i.e., cumulative power consumption), and the ratio of energy consumption at a certain expected time point to the performance of the smart card chip (i.e., energy efficiency).

[0055] Exemplarily, the time series feature mapping results corresponding to each expected time point are based on the nonlinear relationship between the features learned by the energy consumption prediction model and the energy consumption, and the feature vector change information and the feature vector time correlation information corresponding to each expected time point are respectively mapped to the energy consumption parameters of the corresponding expected time point to generate a sequence of energy consumption values ​​for all expected time points within the expected time window; furthermore, the energy consumption values ​​corresponding to each expected time point are combined in chronological order to form a complete time series data, which describes the dynamic changes in energy consumption of the smart card chip within the entire expected time window, and then the complete time series data is visualized to obtain a dynamic energy consumption distribution graph with time represented by the horizontal axis and energy consumption value represented by the vertical axis. In the dynamic energy consumption distribution graph, the peak value, valley value, rising or falling rate of each energy consumption parameter and the overall distribution trend of energy consumption can be seen.

[0056] In this embodiment, the energy consumption prediction model is used to recursively calculate the multi-dimensional feature vector according to the time series, so as to accurately capture the input features at each time point in the target time window, thereby effectively generating the timing feature mapping results of each expected time point in the expected time window; furthermore, the energy consumption value corresponding to each expected time point is calculated based on the timing feature mapping results, and then the dynamic energy consumption distribution diagram corresponding to the expected time window is generated with data reliability.

[0057] In an exemplary embodiment, the multi-dimensional feature vector is recursively calculated according to the time series to obtain the time series feature mapping results of each expected time point in the expected time window, including steps S401 to S404.

[0058] Step S401, decompose the multidimensional feature vector according to the time series to obtain the feature state of the multidimensional feature vector at each time point in the target time window. The feature state at each time point reflects the operating behavior and task execution status of the smart card chip at the corresponding time point.

[0059] Among them, the characteristic state of a time point represents a set of operating data features at each time point decomposed in a multi-dimensional feature vector, such as a characteristic state used to describe an operating behavior such as a task or working mode performed by a smart card chip at a certain time point, and a characteristic state used to describe a task completion degree, task load status, and other task execution conditions of a smart card chip at a certain time point.

[0060] Exemplarily, the complex multi-dimensional feature vector is decomposed according to the time series, that is, the data in the entire target time window is divided into several independent time point feature states. The feature state of each time point reflects the fine-grained information corresponding to the operating behavior and task execution status of the smart card chip at the corresponding time point.

[0061] Step S402 , combining the feature state of the multi-dimensional feature vector at each time point with the feature state at the previous time point to obtain the time series feature input corresponding to each time point.

[0062] Exemplarily, the feature state of the multidimensional feature vector at each time point is combined with the feature state at the previous time point to capture the temporal correlation of the feature state at each time point, thereby constructing the temporal feature inputs at each time point. These temporal feature inputs not only retain the independent features at a single time point, but also integrate the dynamic change characteristics of the time series.

[0063] Step S403 , based on the time series feature input corresponding to each time point, recursively update the feature vector of the multi-dimensional feature vector at each time point to obtain the time series feature output corresponding to each time point.

[0064] Exemplarily, the time series feature input corresponding to each time point is recursively updated to the corresponding time series feature output, that is, in the time series feature input that combines the feature state of the current time point and the feature state of the previous time point, the feature vector of the current time point is dynamically adjusted based on the feature state of the previous time point, and the adjusted feature vector of the current time point is used as the time series feature output corresponding to the current time point; based on this, the time series feature output corresponding to each time point is obtained to capture the long-term dependency of the multi-dimensional feature vector in the time dimension.

[0065] Step S404, based on the correlation between the time series of the target time window and the time series of the expected time window, the time series feature outputs corresponding to each time point in the target time window are mapped to the time series features corresponding to each expected time point in the expected time window, and the time series features of consecutive expected time points are smoothed to obtain the time series feature mapping results of each expected time point in the expected time window.

[0066] Among them, the correlation between the time series of the target time window and the time series of the expected time window represents the similarity between the logical association of the two time windows on the time axis and their feature change trends.

[0067] Exemplarily, first, determine the correlation between the time series of the target time window and the time series of the expected time window, that is, based on the characteristic trend of the target time window in the entire time window or at the end of the window, such as the characteristic trend of periodic fluctuations, linear growth or sudden changes, infer the continuation of the characteristic trend in the expected time window; secondly, based on the correlation between the time series of the target time window and the time series of the expected time window, from the starting time point of the expected time window, the time series feature output corresponding to each time point in the target time window is mapped to the time series feature corresponding to each expected time point in the expected time window, so as to continue the characteristic trend of the target time window into the expected time window; thirdly, in order to eliminate possible fluctuations or outliers introduced by recursive calculations, the time series features of continuous expected time points are smoothed, such as sliding average, low-pass filtering, regularization adjustment and other processing methods, to obtain the time series feature mapping results of each expected time point in the expected time window.

[0068] In this embodiment, first, the multi-dimensional feature vector is decomposed according to the time series to obtain the feature state at each time point in the target time window, so as to efficiently convert the complex multi-dimensional feature vector into the feature state of the time point granularity; secondly, the feature state of the multi-dimensional feature vector at each time point is combined with the feature state of the previous time point to obtain the time series feature input corresponding to each time point, so as to efficiently integrate the dynamic change characteristics of the time series; thirdly, the feature vector of the multi-dimensional feature vector at each time point is recursively updated according to the time series feature input corresponding to each time point, so as to obtain the time series feature corresponding to each time point. Output is performed to effectively capture the long-term dependency of the multidimensional feature vector in the time dimension; secondly, according to the correlation between the time series of the target time window and the time series of the expected time window, the time series feature output corresponding to each time point in the target time window is mapped to the time series feature corresponding to each expected time point in the expected time window, thereby adaptively extending the feature trend of the target time window to the expected time window; thirdly, the time series feature output is smoothed to obtain the time series feature mapping results at each expected time point in the expected time window, thereby improving the stability and consistency of the time series features at each expected time point in the expected time window.

[0069] In an exemplary embodiment, the energy consumption value corresponding to each expected time point is calculated based on the time series feature mapping result, including steps S501 to S502.

[0070] Step S501, obtaining the energy consumption contribution corresponding to each feature dimension, and performing weight allocation processing on the feature value of each expected time point in the time series feature mapping result based on the energy consumption contribution corresponding to each feature dimension, to obtain the weighted feature value corresponding to each expected time point.

[0071] Among them, each feature dimension represents an energy consumption indicator of the smart card chip, and multiple feature dimensions are used to combine and describe the various energy consumption performances of the smart card chip at various time points, such as different feature dimensions such as processor load, storage module access frequency, and task switching frequency; the energy consumption contribution degree represents a quantitative indicator of the degree of influence of the feature dimension on the energy consumption of the smart card chip, which can represent the correlation coefficient between each feature dimension and the energy consumption parameter obtained by analyzing historical data, or the weight coefficient generated by the energy consumption prediction model during the training process. A high energy consumption contribution degree can be set for a feature with higher energy consumption to highlight the energy consumption performance of the feature in the energy consumption calculation process, and a low energy consumption contribution degree can be set for a feature with lower energy consumption to suppress the energy consumption performance of the feature in the energy consumption calculation process; further, the energy consumption contribution degree can be adaptively preset for a specific task scenario. For example, in an encryption task, a higher energy consumption contribution degree is set for the operating indicators involved in the data encryption process of the encryption module.

[0072] Exemplarily, the energy consumption contribution corresponding to each feature dimension is obtained, and the feature value of each expected time point is weighted based on the energy consumption contribution corresponding to each feature dimension. That is, in the feature values ​​of the time series feature mapping results at each expected time point, the energy consumption contribution corresponding to a certain feature dimension to a certain feature value at the current expected time point is determined, and the feature value is weighted based on the determined energy consumption contribution to obtain the weighted feature value corresponding to the feature value of the current expected time point on the feature dimension; based on this, the weighted feature values ​​corresponding to each expected time point in different feature dimensions are obtained.

[0073] Step S502 , performing multiple nonlinear transformation processes on the weighted characteristic values ​​corresponding to each expected time point, to obtain the energy consumption value corresponding to each expected time point.

[0074] Exemplarily, multiple nonlinear transformations can be performed on the weighted eigenvalues ​​corresponding to each expected time point through activation functions (such as ReLU, Sigmoid or Tanh functions) or other nonlinear operations to obtain energy consumption values ​​corresponding to each expected time point in different feature dimensions, that is, energy consumption values ​​corresponding to different energy consumption parameters; in the transformation process, each nonlinear transformation operation further extracts and maps the weighted eigenvalues, and multiple nonlinear transformation operations can superimpose multiple layers of nonlinear mapping, gradually refining from simple feature relationships to complex energy consumption characteristics.

[0075] In this embodiment, firstly, based on the energy consumption contribution corresponding to each feature dimension, the feature value of each expected time point in the time series feature mapping result is weighted, and the weighted feature value corresponding to each expected time point in different feature dimensions is obtained, thereby highlighting the influence of the features with high energy consumption contribution on the energy consumption calculation, while suppressing the interference of the features with low energy consumption contribution, thereby ensuring the accuracy and credibility of the subsequent energy consumption calculation results; secondly, the weighted feature value corresponding to each expected time point is subjected to multiple nonlinear transformation processing, and the energy consumption value corresponding to each expected time point in different energy consumption parameters is obtained, thereby fully capturing the nonlinear correlation between the features, and further ensuring the accuracy and credibility of the energy consumption calculation results.

[0076] In an exemplary embodiment, based on the association between each energy consumption unit and the dynamic energy consumption distribution diagram, the energy consumption parameters are optimized to obtain optimized energy consumption parameters, including steps S601 to S603.

[0077] Step S601 , dividing the data in the dynamic energy consumption distribution diagram according to each energy consumption unit to obtain energy consumption data sets corresponding to each energy consumption unit.

[0078] The energy consumption data set refers to a set of energy consumption data associated with a specific energy consumption unit extracted and divided from a dynamic energy consumption distribution diagram, which can be used to describe the energy consumption performance and change trend of the specific energy consumption unit within an expected time window.

[0079] Exemplarily, each energy consumption unit represents the energy consumption characteristics of a specific functional module in the smart card chip. Based on this, the comprehensive energy consumption data in the dynamic energy consumption distribution diagram is split into independent energy consumption data sets according to different energy consumption units, so that each energy consumption unit corresponds to a set of independent energy consumption data. Each energy consumption data set not only retains the changing characteristics in the time series, but also refines the energy consumption sources of different modules.

[0080] Step S602 performs feature decomposition processing on the energy consumption data sets corresponding to each energy consumption unit to obtain the behavior characteristics of each energy consumption unit within the expected time window. The behavior characteristics represent a feature set that describes the correlation between the operating state and the energy consumption state of the energy consumption unit within the expected time window.

[0081] Among them, the behavioral characteristics represent the quantitative description of the correlation between the operating state and energy consumption state of the energy consumption unit within the expected time window; for example, the behavioral characteristics of the energy consumption unit belonging to the processor module may include the quantitative correlation between the operating state and energy consumption such as task load, operating frequency, and instruction type; for another example, the behavioral characteristics of the energy consumption unit belonging to the storage module may include the quantitative correlation between the operating state and energy consumption such as data access frequency, operation type, and access mode.

[0082] Exemplarily, the principal component analysis method (PCA) can be used to reduce the dimension and extract features of the energy consumption data set, so as to obtain the behavioral characteristics of each energy consumption unit in the expected time window based on the analysis of the energy consumption data set of each energy consumption unit, and then obtain the correlation between the operating status and energy consumption status of each functional module corresponding to the smart card chip within the expected time window.

[0083] Step S603: Optimizing the energy consumption parameters of each energy consumption unit according to the behavior characteristics of each energy consumption unit within the expected time window to obtain optimized energy consumption parameters corresponding to the smart card chip.

[0084] Exemplarily, based on established optimization goals, such as minimizing overall energy consumption, improving energy utilization efficiency, and achieving a balance between energy consumption and performance, and in combination with the behavioral characteristics of each energy consumption unit within an expected time window, the energy consumption parameters corresponding to each energy consumption unit can be optimized to obtain the optimized energy consumption parameters corresponding to the smart card chip; for example, for a processor module, energy consumption can be reduced by lowering the operating frequency, adjusting task scheduling, and other means to obtain optimized energy consumption parameters; for another example, for a storage module, energy consumption can be reduced by optimizing the data access sequence or reducing unnecessary read and write operations, and other means to obtain optimized energy consumption parameters.

[0085] In this embodiment, first, the data in the dynamic energy consumption distribution diagram is divided and processed according to each energy consumption unit, and the energy consumption data set corresponding to each energy consumption unit is adaptively obtained, so as to achieve targeted optimization of energy consumption parameters according to the dimension of the energy consumption unit; secondly, the energy consumption data set corresponding to each energy consumption unit is feature decomposed, and the behavioral characteristics of each energy consumption unit are obtained at the feature level, and the quantitative correlation between the operating state and the energy consumption state of each energy consumption unit is effectively obtained; thirdly, according to the behavioral characteristics of each energy consumption unit, the energy consumption parameters of each energy consumption unit are optimized at the feature level in a targeted manner, so as to efficiently obtain the optimized energy consumption parameters corresponding to the smart card chip.

[0086] In an exemplary embodiment, according to the behavior characteristics of each energy consumption unit within the expected time window, the energy consumption parameters of each energy consumption unit are optimized to obtain the optimized energy consumption parameters corresponding to the smart card chip, including steps S701 to S702.

[0087] Step S701, obtaining an optimization objective function pre-constructed based on the behavior characteristics and energy consumption parameters of each energy consumption unit, wherein the optimization objective function includes evaluation indicators corresponding to the operation status and operation constraints of each energy consumption unit, the energy consumption status, and the energy consumption constraints;

[0088] Among them, the optimization objective function represents a mathematical model for quantifying the operating status, energy consumption status and relationship between related constraints of each energy consumption unit, and at the same time defines the optimization target corresponding to the energy consumption parameters; in the optimization process, the optimization objective function is used to guide the direction of adjusting the energy consumption parameters to gradually approach the optimal solution, so that the operating performance and energy consumption performance of the smart card chip reach the set goals.

[0089] Among them, the operating status represents a set of indicators that describe the operating characteristics of the energy consumption unit during operation, such as the load rate of the processor module, the access frequency of the storage module, etc.; the operating constraints represent the restrictions that the energy consumption unit needs to meet during operation, which are used to ensure that the smart card chip is within the normal working allowable range in actual use, such as the operating frequency of the processor module cannot exceed the maximum frequency limit of the hardware, and the access rate of the storage module needs to be within the range supported by the hardware.

[0090] Among them, the energy consumption state represents a set of energy consumption indicators that describe the energy consumption characteristics of the energy consumption unit under a specific operating state, for example, the instantaneous power consumption of the processor module when running at high load is 2.5W, and the cumulative energy consumption of the storage module in continuous writing operation is 10J; the energy consumption constraint represents the restriction conditions that the energy consumption unit needs to meet under a specific operating state, which is used to ensure that the smart card chip is within the allowable energy consumption range in actual use, for example, the total power consumption of the processor module cannot exceed 5W, and the peak energy consumption of the storage module needs to be lower than 3W.

[0091] Step S702, in the optimization objective function, the energy consumption parameters of each energy consumption unit are iteratively updated according to the behavioral characteristics of each energy consumption unit until a set of energy consumption parameters that meets the global optimization conditions corresponding to the optimization objective function is obtained, and the energy consumption parameter set is used as the optimized energy consumption parameters corresponding to the smart card chip.

[0092] Among them, the global optimization condition represents the logical condition that the optimization objective function satisfies when reaching the overall optimum in the process of adjusting the energy consumption parameters of all energy consumption units; for example, when the optimization goal is to minimize the total energy consumption, the global optimization condition may mean that the value of the optimization objective function reaches the global minimum and does not violate the operating constraints and energy consumption constraints of any energy consumption unit.

[0093] Exemplarily, during the optimization process, the energy consumption parameters are iteratively updated in combination with the optimization objective function, wherein in each iteration, the value of the optimization objective function is calculated according to the behavioral characteristics of the current energy consumption parameters, and then the energy consumption parameters are adjusted in combination with the update gradient of the optimization objective function, so that the value of the optimization objective function changes in a more optimal direction; for example, in the process of adjusting the frequency of the processor module, the frequency may be reduced to reduce power consumption, and its impact on the task completion time is evaluated at the same time. Furthermore, if the performance degradation caused by reducing the frequency exceeds an acceptable range, the frequency adjustment will be limited to a specific range; after each iteration, it is detected whether the energy consumption parameter set currently composed of the energy consumption parameters of all energy consumption units meets the global optimization conditions defined by the optimization objective function, such as detecting whether the reduction rate of the total power consumption obtained based on the comprehensive combination of all energy consumption parameters reaches a set threshold, or whether the energy consumption utilization efficiency obtained based on the comprehensive combination of all energy consumption parameters reaches a set threshold, etc.; when the energy consumption parameter set of all energy consumption units simultaneously meets the global optimization conditions of the optimization objective function, it means that all energy consumption parameters have reached the overall optimum, so that the value of the optimization objective function reaches the set optimization goal, the iterative process ends, and the current energy consumption parameter set is used as the optimized energy consumption parameter.

[0094] Optionally, the optimization adjustment of the energy consumption parameters of each energy consumption unit may affect the energy consumption parameters of other energy consumption units. Based on this, the global optimization condition requires that the energy consumption parameters of all energy consumption units reach the overall optimum in coordination with each other. For example, when the processor module reduces the frequency, it must be ensured that the access delay of the storage module will not lead to additional energy consumption growth.

[0095] In this embodiment, based on a pre-constructed optimization objective function, the energy consumption parameters of each energy consumption unit are iteratively updated according to the behavioral characteristics of each energy consumption unit until a set of energy consumption parameters that meets the global optimization conditions corresponding to the optimization objective function is obtained, thereby rationally allocating resources to each energy consumption unit based on a global level to accurately obtain the optimized energy consumption parameters corresponding to the smart card chip.

[0096] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0097] Based on the same inventive concept, the embodiment of the present application also provides a smart card chip energy consumption prediction and optimization system for implementing the above-mentioned smart card chip energy consumption prediction and optimization method. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the energy consumption prediction and optimization system embodiments of one or more smart card chips provided below can refer to the limitations of the energy consumption prediction and optimization method of the smart card chip above, and will not be repeated here.

[0098] In an exemplary embodiment, Figure 2 As shown, a smart card chip energy consumption prediction and optimization system is provided, including: an acquisition module 201, a prediction module 202 and an optimization module 203, wherein:

[0099] The acquisition module 201 is used to acquire the operation data based on the target working mode collected by the smart card chip within the target time window, perform time slice division and clustering processing on the operation data, and obtain a multi-dimensional feature vector within the target time window;

[0100] The prediction module 202 is used to obtain a trained energy consumption prediction model, input the multi-dimensional feature vector into the energy consumption prediction model for processing, and predict a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip. The dynamic energy consumption distribution diagram reflects the energy consumption time series characteristics of the smart card chip in the target working mode. The processing of the multi-dimensional feature vector includes recursive updating of the multi-dimensional feature vector based on the correlation of the time series, obtaining the time series characteristics corresponding to each time point of the target time window, and mapping to obtain the time series characteristics corresponding to each time point of the future time period;

[0101] The optimization module 203 is used to determine the various energy consumption units of the smart card chip, and based on the correlation between the various energy consumption units and the dynamic energy consumption distribution diagram, optimize the energy consumption parameters to obtain optimized energy consumption parameters. The energy consumption units reflect the energy consumption performance of each functional module in the smart card chip in the target working mode.

[0102] In an exemplary embodiment, the acquisition module 201 is also used to: divide the operating data within the target time window into time slices according to a preset time slice length to obtain multiple time slice data, where the time slice length is determined based on the working mode of the smart card chip and the data sampling frequency, and the time slice data includes the operation instruction data, power consumption measurement data and working status data of the smart card chip corresponding to the time slice length; perform feature extraction on the multiple time slice data to obtain feature information corresponding to the multiple time slice data, cluster the feature information with similar features based on a clustering analysis algorithm to obtain a multi-dimensional feature vector within the target time window.

[0103] In an exemplary embodiment, the prediction module 202 is also used to: input the multi-dimensional feature vector into the energy consumption prediction model, recursively calculate the multi-dimensional feature vector according to the time series, and obtain the time series feature mapping results of each expected time point within the expected time window, the time series feature mapping results include the feature vector change information and the feature vector time association information corresponding to each expected time point, and the starting time point of the expected time window matches the end time point of the target time window; based on the time series feature mapping results, the energy consumption value corresponding to each expected time point is calculated, and the energy consumption values ​​corresponding to each expected time point are combined in chronological order to generate a dynamic energy consumption distribution diagram corresponding to the expected time window.

[0104] In an exemplary embodiment, the prediction module 202 is also used to: decompose the multidimensional feature vector according to the time series to obtain the feature state of the multidimensional feature vector at each time point in the target time window, and the feature state at each time point reflects the operating behavior and task execution status of the smart card chip at the corresponding time point; combine the feature state of the multidimensional feature vector at each time point with the feature state of the previous time point to obtain the time series feature input corresponding to each time point; recursively update the feature vector of the multidimensional feature vector at each time point according to the time series feature input corresponding to each time point, and obtain the time series feature output corresponding to each time point; according to the correlation between the time series of the target time window and the time series of the expected time window, map the time series feature output corresponding to each time point in the target time window to the time series features corresponding to each expected time point in the expected time window, and smooth the time series features of consecutive expected time points to obtain the time series feature mapping results of each expected time point in the expected time window.

[0105] In an exemplary embodiment, the prediction module 202 is also used to: obtain the energy consumption contribution corresponding to each feature dimension, perform weight allocation processing on the characteristic values ​​of each expected time point in the time series feature mapping result based on the energy consumption contribution corresponding to each feature dimension, and obtain the weighted characteristic values ​​corresponding to each expected time point; perform multiple nonlinear transformation processing on the weighted characteristic values ​​corresponding to each expected time point, and obtain the energy consumption value corresponding to each expected time point.

[0106] In an exemplary embodiment, the optimization module 203 is also used to: divide the data in the dynamic energy consumption distribution diagram according to each energy consumption unit to obtain the energy consumption data set corresponding to each energy consumption unit; perform feature decomposition processing on the energy consumption data set corresponding to each energy consumption unit to obtain the behavior characteristics of each energy consumption unit within the expected time window, and the behavior characteristics represent the feature set that describes the correlation between the operating status and the energy consumption status of the energy consumption unit within the expected time window; optimize the energy consumption parameters of each energy consumption unit according to the behavior characteristics of each energy consumption unit within the expected time window to obtain the optimized energy consumption parameters corresponding to the smart card chip.

[0107] In an exemplary embodiment, the optimization module 203 is also used to: obtain an optimization objective function pre-constructed based on the behavioral characteristics and energy consumption parameters of each energy consumption unit, the optimization objective function including the operating status of each energy consumption unit and the evaluation indicators corresponding to the operating constraints, energy consumption status and energy consumption constraints; in the optimization objective function, the energy consumption parameters of each energy consumption unit are iteratively updated according to the behavioral characteristics of each energy consumption unit until a set of energy consumption parameters that meets the global optimization conditions corresponding to the optimization objective function is obtained, and the energy consumption parameter set is used as the optimized energy consumption parameters corresponding to the smart card chip.

[0108] Each module in the energy consumption prediction and optimization system of the smart card chip can be implemented in whole or in part by software, hardware and their combination. Each module can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0109] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in any of the above embodiments when executing the computer program.

[0110] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in any of the above embodiments are implemented.

[0111] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and 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-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0112] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0113] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for predicting and optimizing energy consumption of a smart card chip, characterized in that: The method comprises: Acquire operation data based on a target working mode collected by a smart card chip within a target time window, perform time slice division processing and clustering processing on the operation data, and obtain a multi-dimensional feature vector within the target time window; Obtaining a trained energy consumption prediction model, inputting the multi-dimensional feature vector into the energy consumption prediction model for processing, and predicting a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip, wherein the dynamic energy consumption distribution diagram reflects the energy consumption timing characteristics of the smart card chip in the target working mode; Determine each energy consumption unit of the smart card chip, and optimize the energy consumption parameters to obtain optimized energy consumption parameters based on the association between each energy consumption unit and the dynamic energy consumption distribution diagram, wherein the energy consumption unit reflects the energy consumption performance of each functional module in the smart card chip in the target working mode; The step of inputting the multi-dimensional feature vector into the energy consumption prediction model for processing and predicting a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip includes: Input the multi-dimensional feature vector into the energy consumption prediction model, recursively calculate the multi-dimensional feature vector according to the time series, and obtain the time series feature mapping results of each expected time point within the expected time window, wherein the time series feature mapping results include the feature vector change information and the feature vector time association information corresponding to each expected time point, and the starting time point of the expected time window matches the ending time point of the target time window; The energy consumption value corresponding to each expected time point is calculated based on the time series feature mapping result, and the energy consumption values ​​corresponding to each expected time point are combined in chronological order to generate a dynamic energy consumption distribution diagram corresponding to the expected time window.

2. The method according to claim 1, characterized in that The performing time slice division processing and clustering processing on the operation data to obtain a multi-dimensional feature vector within the target time window includes: The operation data in the target time window is divided into time slices according to a preset time slice length to obtain a plurality of time slice data, wherein the time slice length is determined according to the working mode and data sampling frequency of the smart card chip, and the time slice data includes the operation instruction data, power consumption measurement data and working status data of the smart card chip corresponding to the time slice length; Feature extraction processing is performed on multiple time slice data to obtain feature information corresponding to the multiple time slice data respectively, and feature information with similar features is clustered based on a clustering analysis algorithm to obtain a multi-dimensional feature vector within the target time window.

3. The method according to claim 1, characterized in that The recursive calculation of the multi-dimensional feature vector according to the time series to obtain the time series feature mapping results of each expected time point within the expected time window includes: Decomposing the multidimensional feature vector according to the time series to obtain the feature state of each time point of the multidimensional feature vector within the target time window, wherein the feature state of each time point reflects the operation behavior and task execution status of the smart card chip at the corresponding time point; Combining the feature state of the multi-dimensional feature vector at each time point with the feature state at the previous time point to obtain the time series feature input corresponding to each time point; According to the time series feature input corresponding to each time point, the feature vector of the multi-dimensional feature vector at each time point is recursively updated to obtain the time series feature output corresponding to each time point; According to the correlation between the time series of the target time window and the time series of the expected time window, the time series feature outputs corresponding to each time point in the target time window are mapped to the time series features corresponding to each expected time point in the expected time window, and the time series features of consecutive expected time points are smoothed to obtain the time series feature mapping results of each expected time point in the expected time window.

4. The method according to claim 1, characterized in that: The energy consumption value corresponding to each expected time point is calculated based on the time series feature mapping result, including: Obtaining the energy consumption contribution corresponding to each feature dimension, and performing weight distribution processing on the feature values ​​of each expected time point in the time series feature mapping result based on the energy consumption contribution corresponding to each feature dimension, to obtain the weighted feature values ​​corresponding to each expected time point; The weighted eigenvalues ​​corresponding to each expected time point are subjected to multiple nonlinear transformation processes to obtain the energy consumption value corresponding to each expected time point.

5. The method according to claim 1, characterized in that The optimizing process of the energy consumption parameters based on the association between each energy consumption unit and the dynamic energy consumption distribution diagram to obtain the optimized energy consumption parameters includes: Divide the data in the dynamic energy consumption distribution diagram according to each energy consumption unit to obtain energy consumption data sets corresponding to each energy consumption unit; Performing feature decomposition processing on the energy consumption data sets corresponding to each energy consumption unit, to obtain the behavior characteristics of each energy consumption unit within the expected time window, wherein the behavior characteristics represent a feature set describing the correlation between the operating state and the energy consumption state of the energy consumption unit within the expected time window; According to the behavior characteristics of each energy consumption unit within the expected time window, the energy consumption parameters of each energy consumption unit are optimized to obtain the optimized energy consumption parameters corresponding to the smart card chip.

6. The method according to claim 5, characterized in that The step of optimizing the energy consumption parameters of each energy consumption unit according to the behavior characteristics of each energy consumption unit within the expected time window to obtain the optimized energy consumption parameters corresponding to the smart card chip includes: Obtaining an optimization objective function pre-constructed based on the behavior characteristics and energy consumption parameters of each energy consumption unit, wherein the optimization objective function includes evaluation indicators corresponding to the operation status and operation constraints of each energy consumption unit, the energy consumption status, and the energy consumption constraints; In the optimization objective function, the energy consumption parameters of each energy consumption unit are iteratively updated according to the behavioral characteristics of each energy consumption unit until a set of energy consumption parameters that meets the global optimization conditions corresponding to the optimization objective function is obtained, and the energy consumption parameter set is used as the optimized energy consumption parameters corresponding to the smart card chip.

7. A smart card chip energy consumption prediction and optimization system, characterized in that: The system comprises: An acquisition module is used to acquire operation data based on a target working mode collected by a smart card chip within a target time window, and perform time slice division and clustering processing on the operation data to obtain a multi-dimensional feature vector within the target time window; A prediction module, used to obtain a trained energy consumption prediction model, input the multi-dimensional feature vector into the energy consumption prediction model for processing, and predict a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip, wherein the dynamic energy consumption distribution diagram reflects the energy consumption timing characteristics of the smart card chip in the target working mode; an optimization module, used to determine each energy consumption unit of the smart card chip, and optimize the energy consumption parameters to obtain optimized energy consumption parameters based on the association between each energy consumption unit and the dynamic energy consumption distribution diagram, wherein the energy consumption unit reflects the energy consumption performance of each functional module in the smart card chip in the target working mode; Among them, the prediction module is also used to: input the multi-dimensional feature vector into the energy consumption prediction model, recursively calculate the multi-dimensional feature vector according to the time series, and obtain the time series feature mapping results of each expected time point within the expected time window, the time series feature mapping results include the feature vector change information and feature vector time association information corresponding to each expected time point, and the starting time point of the expected time window matches the end time point of the target time window; based on the time series feature mapping results, the energy consumption value corresponding to each expected time point is calculated, and the energy consumption values ​​corresponding to each expected time point are combined in chronological order to generate a dynamic energy consumption distribution diagram corresponding to the expected time window.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images