Intelligent terminal power consumption optimization method based on active and passive identification

Abstract

The invention relates to the technical field of intelligent terminals, in particular to an intelligent terminal power consumption optimization method based on active and passive identification, which comprises the following steps: passive identification acquires a first signal data set by means of a low-power-consumption wireless communication unit and based on a low-power-consumption wake-up mechanism, active identification period and passive cooperative on-demand wake-up are carried out, and cooperative data acquisition is carried out; carrying out standardized preprocessing on the data, and carrying out weighted fusion on passive electromagnetic and active characteristics according to scene priorities to generate a comprehensive scene vector; querying a predefined strategy mapping table based on the current scene category, obtaining a strategy identifier uniquely associated with the category, and loading a corresponding strategy execution parameter set from a strategy library according to the strategy identifier; analyzing the strategy parameter, generating a time sequence instruction, executing clock gating on the target hardware unit in sequence, and sending instructions by the power supply and the bus arbiter to complete power consumption state switching. The accuracy of scene judgment is improved through an active and passive fusion mechanism, and power consumption regulation and control deviation caused by scene misjudgment is effectively avoided.

Description

Technical Field

[0001] This invention relates to the field of smart terminal technology, and more specifically to a power consumption optimization method for smart terminals based on active and passive identification. Background Technology

[0002] As smart terminals become increasingly powerful, their power consumption also increases dramatically. Battery life has become one of the key technologies restricting user experience. Existing power management is mostly concentrated at the operating system level, using dynamic power management and dynamic voltage and frequency adjustment technologies based on timers or system load. For example, when the system is detected to be idle, the processor frequency is automatically reduced or it is put into a low-power sleep state. However, such solutions have significant lag. Their response begins after the user operation or computing task occurs, and they cannot predict the user's actual needs. Therefore, additional performance overhead and energy waste may occur during the state switching process.

[0003] Some power optimization collaborations have introduced context-aware technologies, such as using GPS positioning and base station information to infer the user's state and adjust power consumption strategies accordingly. These solutions have a relatively coarse granularity in determining the scenario and cannot accurately distinguish the user's usage state, resulting in the strategy execution process being too aggressive or conservative, which affects the user experience. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a method for optimizing the power consumption of smart terminals based on active and passive identification.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: a power consumption optimization method for intelligent terminals based on active and passive identification, comprising the following steps: S101. Passive identification uses a low-power wireless communication unit to set a dynamic scanning cycle based on the terminal screen display and power status. It collects the first signal dataset based on a low-power wake-up mechanism. The active identification cycle and passive identification cycle are coordinated to wake up on demand and perform coordinated data collection. S102. Standardize and preprocess the data, then weight and fuse passive electromagnetic and active features according to scene priority to generate a comprehensive scene vector. Calculate the cosine similarity with the preset historical scene fingerprint database template, and determine the usage scenario of the smart terminal according to the preset threshold. S103. Based on the current scenario category, query the predefined policy mapping table, obtain the policy identifier that is uniquely associated with the category, load the corresponding policy execution parameter set from the policy library according to the policy identifier, and send the policy execution parameter set to the unified power management controller of the terminal. S104. The power management controller parses the policy parameters, generates timing instructions, and sequentially executes clock gating, power supply and bus arbitrator instructions to the target hardware unit to complete the power consumption state switching.

[0006] In a preferred embodiment, in S101, the low-power wireless communication unit built into the terminal performs passive scanning in the background in a first cycle to acquire a first signal dataset. The first signal dataset constructs a complete electromagnetic environment feature vector by collecting basic parameters of the signal source, channel and protocol parameters, and signal stability parameters. The scanning process adopts an adaptive cycle setting strategy, which dynamically adjusts the first cycle of scanning according to the current working state and power level of the terminal. When the terminal is in an inactive state, a long cycle is used, and when it is in an active state, a short cycle is used. By balancing the real-time data acquisition and power consumption control requirements, the extra energy consumption caused by scanning too frequently is avoided. When each cycle arrives, the wireless communication unit is woken up to perform signal acquisition. The wireless communication unit is in a low-current sleep state during non-acquisition periods and is only briefly woken up at the beginning of each scanning cycle. After completing signal acquisition, it immediately returns to the sleep state. Simultaneously, the active sensing unit is activated, and collaborative data acquisition is performed in a second cycle matching the first cycle. The active sensing unit collects active scene datasets. The collected scene data is used to supplement and enhance the dimensions of environmental perception. The scene dataset mainly includes location information, motion state, environmental physical parameters, and user interaction behavior. Location information is obtained through the positioning system and converted into semantic location type. Motion state is obtained through acceleration data obtained by accelerometer and gyroscope. Motion state is divided according to acceleration variance: stationary state corresponds to small variance range, walking state corresponds to medium variance range, running state corresponds to large variance range, and riding state corresponds to uniform motion with small acceleration fluctuation. Environmental physical parameters are obtained through ambient light sensor to obtain light intensity and through temperature sensor to obtain ambient temperature. User interaction behavior is identified through the front-end application interface to identify the currently running application type and through touch event log to count screen touch frequency. Application type directly reflects user needs, touch frequency reflects user activity level, and user interaction behavior can ultimately confirm scene attributes. The initial value and adjustment of the second cycle are both realized through the terminal's sensor management interface. An intelligent adjustment mechanism is introduced. When the passive scanning recognition result reaches a high confidence level, the active acquisition cycle is automatically extended. In a preferred embodiment, in S102, the acquired first signal dataset and the active scene dataset are subjected to standardization preprocessing. The preprocessing process includes the following steps: S1. Standardize the continuous signal strength values ​​and linearly map them to a unified preset value range. The specific calculation process is as follows: in, This represents the standardized signal strength value. This represents the measured strength value of the wireless signal. This represents the standard minimum strength value of a wireless signal. This represents the typical maximum strength value of a wireless signal. This mapping process is calculated based on the theoretical upper and lower limits of the strength of different types of signals. S2. For discrete classification data, a unique hot coding method is used to convert it into a multi-dimensional binary vector for numerical processing and similarity calculation. S3. Outliers are identified and removed using criteria based on the normal distribution characteristics of the data. The collected raw data is cleaned to eliminate invalid data points that significantly deviate from the normal fluctuation range due to sudden external interference. (3) The criteria identify and remove outliers in continuous signal strength data. The specific calculation process is as follows: The formula for calculating the mean of the data is as follows: in, The mean of a data sample reflects the central tendency of the data. To calculate the standard deviation of continuous raw data, the specific formula is as follows: in, The standard deviation of the data sample reflects the dispersion of the data. The specific formula for calculating the outlier detection criteria is as follows: Values ​​identified as outliers are removed. This represents the outlier threshold based on the properties of a normal distribution; normal data will fall within [...]. Data exceeding the specified range is considered an outlier due to external interference. Based on the completed data preprocessing, a weighted fusion operation is performed to construct a comprehensive scene feature vector that can characterize the environment in which the terminal is located. Through differentiated weight allocation and linear fusion, a comprehensive vector that combines passive electromagnetic features and active scene features is constructed. The specific calculation formula is as follows: in, Represents the comprehensive scene feature vector. The weight coefficients represent the passive electromagnetic feature vectors, reflecting the contribution of passive electromagnetic features to scene differentiation. This represents the passive electromagnetic eigenvector, which includes the normalized signal strength and the encoded signal type. The weight coefficients represent the active location feature vectors, reflecting the contribution of the active location features. This represents an active location feature vector, containing the encoded location type. The weight coefficients represent the active motion feature vectors, reflecting the contribution of the active motion features. This represents an active motion feature vector, containing the encoded motion state and the standardized acceleration variance. The weight coefficients represent the active environment and behavioral feature vectors, reflecting the contribution of the active environment and behavioral features. The active environment and behavior feature vector includes the encoded application type and standardized illumination intensity. A historical scene fingerprint database is pre-built, storing corresponding feature templates according to scene category. Each scene's feature template is stored as a vector with the same dimensions as the comprehensive feature vector. The values ​​of each dimension of the vector correspond to the typical value range of the feature parameters in that scene, ensuring that the template vector and the comprehensive feature vector have the same format. Cosine similarity is calculated based on the real-time generated comprehensive scene feature vector and each standard template vector in the historical scene fingerprint database. The similarity is used to quantify the degree of matching between the real-time scene and the standard scene. The specific calculation formula is as follows: in, This represents the cosine similarity between the combined scene feature vector and the scene template vector. This represents a template vector for a certain type of scene in the historical scene fingerprint database, where m represents the dimension of the feature vector. This represents the k-th dimension value of the comprehensive scene feature vector. This represents the k-th dimension value of the scene template vector. The dot product of two vectors reflects the consistency of their directions. This represents the magnitude of the two vectors, used to normalize the dot product result; Based on the calculated highest similarity value, the final scene category is determined according to the preset multi-level judgment rules. If the highest similarity exceeds the first preset threshold, it is directly judged as the corresponding scene. If the highest similarity is within the range between the first preset high threshold and the second preset low threshold, a secondary judgment process is initiated. If the highest similarity is lower than the second preset low threshold, it is judged as an unknown scene, and the default conservative power management strategy is initiated.

[0007] In a preferred embodiment, in S103, after the terminal completes the current scene category determination, it uses the scene category as the search keyword and performs a matching search in the predefined policy mapping table through the terminal's mapping table query interface to obtain a policy identifier uniquely associated with the scene category. The terminal uses the policy identifier as an index to call the policy mapping table loading interface to locate and extract the corresponding policy execution parameter set in the policy library. The policy execution parameter set includes the target hardware module identifier, the target power consumption status identifier, and the triggering condition for state switching. The policy execution parameter set accesses the policy library according to the obtained policy identifier, reads and parses the corresponding data record. During the loading process, the parameter integrity check is automatically performed to check whether the policy execution parameter set contains a complete combination of the target hardware module identifier, the target power consumption status identifier, and the triggering condition for state switching. If one of the parameters is missing, it is determined that the policy execution parameter set is corrupted, and the backup policy execution parameter set corresponding to the scene is automatically loaded. If the parameters are complete, the policy execution parameter set is temporarily stored in the cache unit and transmitted to the terminal's unified power management controller. During transmission, a data frame encapsulation format with a check bit is used. The strategy execution parameter set is encapsulated into a fixed-length data frame according to a preset format, and a CRC check bit is added to the end of the data frame. After receiving the data frame, the power management controller first verifies whether there is a data transmission error through the CRC check bit. If the verification passes, the parameters in the data frame are parsed. If the verification fails, a transmission error is reported and a request is made to resend the parameter set. After receiving and verifying the parameter set, the power management controller immediately performs parameter parsing. According to the hardware unit to be controlled, an independent control task queue is established for each hardware unit. The state switching timing of each hardware unit is determined according to the trigger conditions. The control tasks of hardware units that switch immediately are added to the queue first. The control tasks of hardware units that switch after meeting specific conditions are temporarily stored and added to the queue after the trigger conditions are detected. The specific control parameters of each hardware unit are determined according to the target power consumption status identifier and converted into low-level instructions that the hardware unit can recognize.

[0008] In a preferred embodiment, in S104, the power management controller parses the strategy execution parameter set and generates a set of hardware control instructions with timing logic. The power management controller combines the parsed target state with the timing plan to generate standardized hardware control instructions. The instruction format includes the target hardware module identifier, opcode, parameter value, and execution timing number. All control instructions are sorted by timing number to form an ordered instruction queue, ensuring that they are triggered in a preset order during execution. A set of hardware control instructions is executed, and the power consumption state of the target hardware unit is switched from the current state to the state defined by the target power consumption state identifier according to the preset order. A first control word is written to the register of the target hardware unit. The internal clock gating unit is configured to manage dynamic power consumption. By writing a specific bit sequence, it controls the distribution of the clock tree inside the target hardware unit, selectively shutting down the clock signals of unnecessary functions, reducing its dynamic power consumption to zero. Secondly, it sends a second control word to the target power management unit to control the switching state of its power supply domain. By sending the second control word, it controls the power gating switch in the power supply network. It sends a third control command to the bus arbiter to actively notify the interconnect fabric to suspend all data transmission requests to the target hardware unit, placing it in an isolated state to prevent external devices from trying to access the slave device unit that has entered a low-power state and generating unnecessary wake-up events. After the first preset delay following the execution of the hardware control instruction, the status register of the target hardware unit is read to obtain its actual power consumption status value. The actual power consumption status value is compared with the expected status value of the target power consumption status identifier. If the comparison result is inconsistent, a policy execution exception signal is generated. Based on the policy execution exception signal, the backup policy execution process is triggered. The backup policy execution process is pre-stored in the backup partition of the policy library.

[0009] The beneficial effects of this invention are as follows: This invention generates hardware control instructions with timing logic through a unified power management controller, which directly operates the underlying clock gating, power supply and bus arbitrator of the hardware. Through the coordinated management of the dynamic power consumption, static power consumption and direct interaction of the hardware unit, the electrical safety of the state switching process is ensured. The accuracy of scene judgment is improved through the active-passive fusion mechanism, which effectively avoids the power consumption control deviation caused by scene misjudgment and fundamentally avoids unnecessary energy waste. Attached Figure Description

[0010] Figure 1 This is a flowchart of the present invention; Figure 2 This is a flowchart of the power consumption optimization control of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0013] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0014] like Figure 1 - Figure 2 This embodiment provides a power consumption optimization method for smart terminals based on active and passive identification, including the following steps: S101. Passive identification uses a low-power wireless communication unit to set a dynamic scanning cycle based on the terminal screen display and power status. It collects the first signal dataset based on a low-power wake-up mechanism. The active identification cycle and passive identification cycle are coordinated to wake up on demand and perform coordinated data collection. Furthermore, through the low-power wireless communication unit built into the terminal, passive scanning is performed in the background in the first cycle to acquire the first signal dataset. The first signal dataset constructs a complete electromagnetic environment feature vector by collecting basic parameters of the signal source, channel and protocol parameters, and signal stability parameters. The scanning process adopts an adaptive cycle setting strategy, which dynamically adjusts the first cycle of scanning according to the current working state and power level of the terminal. When the terminal is in an inactive state, a long cycle is used, and when it is in an active state, a short cycle is used. By balancing the real-time data acquisition and power consumption control requirements, the extra energy consumption caused by scanning too frequently is avoided. When each cycle arrives, the wireless communication unit is woken up to perform signal acquisition. The wireless communication unit is in a low-current sleep state during non-acquisition periods and is only briefly woken up at the beginning of each scanning cycle. After completing signal acquisition, it immediately returns to the sleep state. Simultaneously, the active sensing unit is activated, and coordinated data acquisition is performed in a second cycle matching the first cycle. The active sensing unit collects an active scene dataset. The collected scene data is used to supplement and enhance the dimensions of environmental perception. The scene dataset mainly includes location information, motion state, environmental physical parameters, and user interaction behavior. Location information is obtained through a positioning system and converted into a semantic location type. Motion state is obtained through acceleration data acquired by an accelerometer and gyroscope. Motion states are divided according to acceleration variance: stationary state corresponds to a small variance range, walking state corresponds to a medium variance range, running state corresponds to a large variance range, and riding state corresponds to a small variance range. The motion is uniform with small acceleration fluctuations. Environmental physical parameters are obtained by acquiring light intensity through an ambient light sensor and ambient temperature through a temperature sensor. User interaction behavior is identified by the front-end application interface to identify the currently running application type. The touch event log is used to count the screen touch frequency. The application type directly reflects user needs, and the touch frequency reflects user activity. User interaction behavior can ultimately confirm the scene attributes. The initial value and adjustment of the second cycle are both realized through the terminal's sensor management interface. An intelligent adjustment mechanism is introduced. When the passive scanning recognition result reaches a high confidence level, the active collection cycle is automatically extended to reduce redundant data collection. It should be noted that the low-power wireless communication unit uses Bluetooth, Wi-Fi, and cellular base station signal receivers as core scanning units. All three types of units have low power consumption characteristics and cover commonly used wireless signal types of terminals, which can comprehensively capture the characteristics of the surrounding electromagnetic environment and avoid scene misjudgment due to missing signal types.

[0015] It should be noted that the ambient light level is divided into low-light environment and strong-light environment based on the light intensity range. The characteristics of the environment are directly related to the usage requirements of the scene. For example, the low-light environment is a night scene, so the screen brightness needs to be increased. The strong-light environment may be an outdoor scene, so the anti-glare mode needs to be turned on. It can also help distinguish the scene and time of day.

[0016] Signal source basic parameter collection: covering signal type, signal strength value, and identification code. Among them, the signal type is used to distinguish the signal coverage area, and the signal strength value corresponds to the normal strength range according to the signal type. Since different types of signals have different propagation attenuation characteristics, the interval division can accurately reflect the distance between the signal source and the terminal (for example, when the signal strength is at a higher level at short distances, it means that the signal source is closer). The identification code is a unique identifier for the signal source and can be directly associated with the fixed signal characteristics of a specific scenario, providing a key basis for scenario matching.

[0017] Channel and protocol parameter acquisition: This includes Wi-Fi channel number, Bluetooth broadcast interval, base station frequency band, and signal modulation method. The channel and protocol parameters vary significantly in different scenarios. These parameters can help distinguish scenario attributes. The modulation method reflects the signal transmission rate requirements and indirectly reflects the communication load of the scenario.

[0018] Signal stability parameter acquisition: Record the signal intensity fluctuation value of multiple consecutive scans, and set the standard to judge the smaller fluctuation range as stable and the larger fluctuation range as unstable. Stable signals usually correspond to fixed scenes, while unstable signals correspond to moving scenes. This parameter can quickly and preliminarily determine whether the scene is in dynamic change, and provide a priority basis for subsequent fusion of active features.

[0019] S102. Standardize and preprocess the data, then weight and fuse passive electromagnetic and active features according to scene priority to generate a comprehensive scene vector. Calculate the cosine similarity with the preset historical scene fingerprint database template, and determine the usage scenario of the smart terminal according to the preset threshold. Furthermore, the acquired first signal dataset and active scene dataset are subjected to standardization preprocessing. The preprocessing process includes the following steps: S1. Standardize the continuous signal strength values ​​and linearly map them to a unified preset value range. The specific calculation process is as follows: in, This represents the standardized signal strength value. This represents the measured strength value of the wireless signal. This represents the standard minimum strength value of a wireless signal. This represents the typical maximum strength value of a wireless signal. This mapping process is calculated based on the theoretical upper and lower limits of the strength of different types of signals. S2. For discrete classification data, a unique hot coding method is used to convert it into a multi-dimensional binary vector for numerical processing and similarity calculation. S3. Outliers are identified and removed using criteria based on the normal distribution characteristics of the data. The collected raw data is cleaned to eliminate invalid data points that significantly deviate from the normal fluctuation range due to sudden external interference. (3) The criteria identify and remove outliers in continuous signal strength data. The specific calculation process is as follows: The formula for calculating the mean of the data is as follows: in, The mean of a data sample reflects the central tendency of the data. To calculate the standard deviation of continuous raw data, the specific formula is as follows: in, The standard deviation of the data sample reflects the dispersion of the data. The specific formula for calculating the outlier detection criteria is as follows: Values ​​identified as outliers are removed. This represents the outlier threshold based on the properties of a normal distribution; normal data will fall within [...]. Data exceeding the specified range is considered an outlier due to external interference. Based on the completed data preprocessing, a weighted fusion operation is performed to construct a comprehensive scene feature vector that can characterize the environment in which the terminal is located. Through differentiated weight allocation and linear fusion, a comprehensive vector that combines passive electromagnetic features and active scene features is constructed. The specific calculation formula is as follows: in, Represents the comprehensive scene feature vector. The weight coefficients represent the passive electromagnetic feature vectors, reflecting the contribution of passive electromagnetic features to scene differentiation. This represents the passive electromagnetic eigenvector, which includes the normalized signal strength and the encoded signal type. The weight coefficients represent the active location feature vectors, reflecting the contribution of the active location features. This represents an active location feature vector, containing the encoded location type. The weight coefficients represent the active motion feature vectors, reflecting the contribution of the active motion features. This represents an active motion feature vector, containing the encoded motion state and the standardized acceleration variance. The weight coefficients represent the active environment and behavioral feature vectors, reflecting the contribution of the active environment and behavioral features. The active environment and behavior feature vector includes the encoded application type and standardized illumination intensity. A historical scene fingerprint database is pre-built, storing corresponding feature templates according to scene category. Each scene's feature template is stored as a vector with the same dimensions as the comprehensive feature vector. The values ​​of each dimension of the vector correspond to the typical value range of the feature parameters in that scene, ensuring that the template vector and the comprehensive feature vector have the same format. Cosine similarity is calculated based on the real-time generated comprehensive scene feature vector and each standard template vector in the historical scene fingerprint database. The similarity is used to quantify the degree of matching between the real-time scene and the standard scene. The specific calculation formula is as follows: in, This represents the cosine similarity between the combined scene feature vector and the scene template vector. This represents a template vector for a certain type of scene in the historical scene fingerprint database, where m represents the dimension of the feature vector. This represents the k-th dimension value of the comprehensive scene feature vector. This represents the k-th dimension value of the scene template vector. The dot product of two vectors reflects the consistency of their directions. This represents the magnitude of the two vectors, used to normalize the dot product result; Based on the calculated highest similarity value, the final scene category is determined according to the preset multi-level judgment rules. If the highest similarity exceeds the first preset threshold, it is directly judged as the corresponding scene. If the highest similarity is within the range between the first preset high threshold and the second preset low threshold, a secondary judgment process is initiated. If the highest similarity is lower than the second preset low threshold, it is judged as an unknown scene, and the default conservative power management strategy is initiated.

[0020] It should be noted that the historical scenario fingerprint database includes five core scenarios: home, office, commuting, outdoor, and low-signal scenarios. The passive electromagnetic characteristics of the home scenario are: stable short-range signal and fixed Bluetooth device identifier. This is because the device location is fixed in the home environment, and the signal does not change frequently. The passive electromagnetic characteristics of the office scenario are: multiple short-range hotspots coexisting, stable wide-area signal, and constantly connected Bluetooth office devices. This is because there are multiple network devices and office peripherals in the office area. The passive electromagnetic characteristics of the commuting scenario are: frequent switching of wide-area signal and weak short-range signal. This is because the terminal moves with the user during commuting, moving away from fixed short-range devices and crossing multiple wide-area signal coverage areas. The passive electromagnetic characteristics of the outdoor scenario are: stable wide-area signal, no fixed short-range hotspots, and occasional temporary short-range devices. This is because there are no fixed network devices in the outdoor environment, only temporary device signals from other users. The passive electromagnetic characteristics of the low-signal scenario are: all signal strengths are at a low level, signal fluctuations are large, and there is no stable signal source. This is because signal coverage is poor in basements or remote areas.

[0021] S103. Based on the current scenario category, query the predefined policy mapping table, obtain the policy identifier that is uniquely associated with the category, load the corresponding policy execution parameter set from the policy library according to the policy identifier, and send the policy execution parameter set to the unified power management controller of the terminal. Furthermore, after the terminal completes the current scenario category determination, it uses the scenario category as the search keyword and performs a matching search in the predefined policy mapping table through the terminal's mapping table query interface to obtain the policy identifier uniquely associated with the scenario category. The terminal uses the policy identifier as an index to call the policy mapping table loading interface, locates and extracts the corresponding policy execution parameter set in the policy library. The policy execution parameter set includes the target hardware module identifier, the target power consumption status identifier, and the triggering condition for state switching. Based on the obtained policy identifier, the policy execution parameter set accesses the policy library, reads and parses the corresponding data records. During the loading process, parameter integrity verification is automatically performed to check whether the policy execution parameter set contains a complete combination of the target hardware module identifier, the target power consumption status identifier, and the triggering condition for state switching. If one of the parameters is missing, it is determined that the policy execution parameter set is corrupted, and the backup policy execution parameter set corresponding to the scenario is automatically loaded. If the parameters are complete, the policy execution parameter set is temporarily stored in the cache unit and transmitted to the terminal's unified power management controller. During transmission, a data frame encapsulation format with a check bit is used. The strategy execution parameter set is encapsulated into a fixed-length data frame according to a preset format, and a CRC check bit is added to the end of the data frame. After receiving the data frame, the power management controller first verifies whether there is a data transmission error through the CRC check bit. If the verification passes, the parameters in the data frame are parsed. If the verification fails, a transmission error is reported and a request is made to resend the parameter set. After receiving and verifying the parameter set, the power management controller immediately performs parameter parsing. According to the hardware unit to be controlled, an independent control task queue is established for each hardware unit. The state switching timing of each hardware unit is determined according to the trigger conditions. The control tasks of hardware units that switch immediately are added to the queue first. The control tasks of hardware units that switch after meeting specific conditions are temporarily stored and added to the queue after the trigger conditions are detected. The specific control parameters of each hardware unit are determined according to the target power consumption status identifier and converted into low-level instructions that the hardware unit can recognize.

[0022] It should be noted that the target power consumption state identifier refers to the specific power consumption state parameter that the hardware unit needs to switch to. This provides a clear control target for the unified power management controller, preventing the control results from deviating from the scenario requirements due to ambiguous parameters. The trigger condition refers to a predefined specific timing or judgment standard used to initiate the hardware unit's switch from the current power consumption state to the target power consumption state. Its core function is to ensure that hardware power consumption control is synchronized with scenario requirements, user behavior, or environmental changes, avoiding experience loss or power waste caused by control being too early or too late. The setting of this condition must be based on objective data that can be collected in real time by the terminal, and mainly falls into three categories: Instant trigger condition: This means that once the scenario category is determined, no additional judgment is needed, and the hardware state switch is initiated immediately. It is suitable for scenarios with clear scenario requirements and no tolerance for delay. For example, when a user moves from an office scenario to a commuting scenario, the Wi-Fi module needs to immediately switch from always on to sleep to avoid frequent Wi-Fi hotspot searches and power consumption during the commute. Environmental data trigger conditions: The controller determines whether to switch based on data collected by environmental sensors. For example, the screen module's trigger condition is when the detected ambient light intensity exceeds the preset strong light threshold. When the ambient light sensor detects that the light intensity exceeds the standard in real time, the controller immediately activates the screen's anti-glare mode and increases the brightness to ensure outdoor visibility. Application status triggering conditions: Determine whether to switch based on the running status of the foreground application.

[0023] S104. The power management controller parses the policy parameters, generates timing instructions, and sequentially sends instructions to the target hardware unit for clock gating, power supply, and bus arbitrator to complete the power consumption state switching. Furthermore, the power management controller parses the strategy execution parameter set and generates a set of hardware control instructions with timing logic. The power management controller combines the parsed target state with the timing plan to generate standardized hardware control instructions. The instruction format includes the target hardware module identifier, opcode, parameter value, and execution timing number. All control instructions are sorted by timing number to form an ordered instruction queue, ensuring that they are triggered in a preset order during execution. A set of hardware control instructions is executed, and according to the preset order, the power consumption state of the target hardware unit is switched from the current state to the state defined by the target power consumption state identifier. The first control word is written to the register of the target hardware unit to configure its internal... The clock gating unit manages dynamic power consumption. By writing a specific bit sequence, it controls the distribution of the clock tree inside the target hardware unit, selectively shutting down the clock signals of unnecessary functions to reduce its dynamic power consumption to zero. Secondly, it sends a second control word to the target power management unit to control the switching state of its power supply domain. By sending the second control word, it controls the power gating switch in the power supply network. Finally, it sends a third control command to the bus arbiter to actively notify the interconnect fabric to suspend all data transmission requests to the target hardware unit, placing it in an isolated state to prevent external devices from attempting to access the slave device unit that has entered a low-power state and generating unnecessary wake-up events. After the first preset delay following the execution of the hardware control instruction, the status register of the target hardware unit is read to obtain its actual power consumption status value. The actual power consumption status value is compared with the expected status value of the target power consumption status identifier. If the comparison result is inconsistent, a policy execution exception signal is generated. Based on the policy execution exception signal, the backup policy execution process is triggered. The backup policy execution process is pre-stored in the backup partition of the policy library.

[0024] It should be noted that the first preset delay is a reserved time window to ensure that the target hardware unit completes the state switch and reaches electrical stability. Its specific value depends on the physical characteristics of the target hardware unit.

[0025] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0026] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0027] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0028] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0029] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0030] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0031] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A power consumption optimization method for intelligent terminals based on active and passive identification, characterized in that, Includes the following steps: S101. Passive identification uses a low-power wireless communication unit to set a dynamic scanning cycle based on the terminal screen display and power status. It collects the first signal dataset based on a low-power wake-up mechanism. The active identification cycle and passive identification cycle are coordinated to wake up on demand and perform coordinated data collection. S102. Standardize and preprocess the data, then weight and fuse passive electromagnetic and active features according to scene priority to generate a comprehensive scene vector. Calculate the cosine similarity with the preset historical scene fingerprint database template, and determine the usage scenario of the smart terminal according to the preset threshold. S103. Based on the current scenario category, query the predefined policy mapping table, obtain the policy identifier that is uniquely associated with the category, load the corresponding policy execution parameter set from the policy library according to the policy identifier, and send the policy execution parameter set to the unified power management controller of the terminal. S104. The power management controller parses the policy parameters, generates timing instructions, and sequentially executes clock gating, electrical domain and bus arbitrator instructions to the target hardware unit to complete the power consumption state switching.

2. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 1, characterized in that, In step S101, the terminal's built-in low-power wireless communication unit performs a passive scan in the background in a first cycle to acquire a first signal dataset. The first signal dataset constructs a complete electromagnetic environment feature vector by collecting basic parameters of the signal source, channel and protocol parameters, and signal stability parameters. The scanning process adopts an adaptive cycle setting strategy, dynamically adjusting the first scan cycle according to the terminal's current working state and power level. When the terminal is in an inactive state, a long cycle is used, and when it is in an active state, a short cycle is used. By balancing the real-time data acquisition and power consumption control requirements, the wireless communication unit is woken up to perform signal acquisition when each cycle arrives. The wireless communication unit remains in a low-current sleep state during non-acquisition periods and is only briefly woken up at the beginning of each scan cycle. After completing signal acquisition, it immediately returns to the sleep state.

3. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 2, characterized in that, Simultaneously, the active sensing unit is activated, and collaborative data acquisition is performed in a second cycle matching the first cycle. The active sensing unit collects an active scene dataset. The collected scene data is used to supplement and enhance the dimensions of environmental perception. The scene dataset mainly includes location information, motion state, environmental physical parameters, and user interaction behavior. Location information is obtained through the positioning system and converted into semantic location type. Motion state is obtained through acceleration data acquired by accelerometer and gyroscope, and motion state is classified according to acceleration variance. Environmental physical parameters are obtained through ambient light sensor to acquire light intensity and through temperature sensor to acquire ambient temperature. User interaction behavior is identified through the foreground application interface to identify the currently running application type and through touch event log to count screen touch frequency. User interaction behavior can ultimately confirm scene attributes. The initial value and adjustment of the second cycle are both implemented through the terminal's sensor management interface, introducing an intelligent adjustment mechanism. When the recognition result of passive scanning reaches a high confidence level, the active acquisition cycle is automatically extended.

4. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 1, characterized in that, In step S102, the acquired first signal dataset and active scene dataset are subjected to standardization preprocessing. The preprocessing process includes the following steps: S1. Standardize the continuous signal strength values ​​and linearly map them to a unified preset value range. The specific calculation process is as follows: in, This represents the standardized signal strength value. This represents the measured strength value of the wireless signal. This represents the standard minimum strength value of a wireless signal. This represents the typical maximum strength value of a wireless signal. This mapping process is calculated based on the theoretical upper and lower limits of the strength of different types of signals. S2. For discrete classification data, a unique hot coding method is used to convert it into a multi-dimensional binary vector for numerical processing and similarity calculation. S3. Outliers are identified and removed using criteria based on the normal distribution characteristics of the data. The collected raw data is cleaned to eliminate invalid data points that significantly deviate from the normal fluctuation range due to sudden external interference. (3) The criteria identify and remove outliers in continuous signal strength data. The specific calculation process is as follows: The formula for calculating the mean of the data is as follows: in, The mean of a data sample reflects the central tendency of the data. To calculate the standard deviation of continuous raw data, the specific formula is as follows: in, The standard deviation of the data sample reflects the dispersion of the data. The specific formula for calculating the outlier detection criteria is as follows: Values ​​identified as outliers are removed. This represents the outlier threshold based on the properties of a normal distribution; normal data will fall within [...]. Data exceeding the specified range is considered an anomaly caused by external interference.

5. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 4, characterized in that, Based on the completed data preprocessing, a weighted fusion operation is performed to construct a comprehensive scene feature vector that can characterize the environment in which the terminal is located. Through differentiated weight allocation and linear fusion, a comprehensive vector that combines passive electromagnetic features and active scene features is constructed. The specific calculation formula is as follows: in, Represents the comprehensive scene feature vector. The weight coefficients represent the passive electromagnetic feature vectors, reflecting the contribution of passive electromagnetic features to scene differentiation. This represents the passive electromagnetic eigenvector, which includes the normalized signal strength and the encoded signal type. The weight coefficients represent the active location feature vectors, reflecting the contribution of the active location features. This represents an active location feature vector, containing the encoded location type. The weight coefficients represent the active motion feature vectors, reflecting the contribution of the active motion features. This represents an active motion feature vector, containing the encoded motion state and the standardized acceleration variance. The weight coefficients represent the active environment and behavioral feature vectors, reflecting the contribution of the active environment and behavioral features. The active environment and behavior feature vector is represented by the encoded application type and the standardized illumination intensity. A historical scene fingerprint database is pre-built, and the corresponding feature templates are stored in the historical scene fingerprint database according to the scene category. The feature template of each scene is stored as a vector with the same dimensions as the comprehensive feature vector. The values ​​of each dimension of the vector correspond to the typical value range of the feature parameters in the corresponding scene. Cosine similarity is calculated based on the comprehensive scene feature vector generated in real time and each standard template vector in the historical scene fingerprint database.

6. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 5, characterized in that, Similarity is used to quantify the degree of matching between a real-time scene and a standard scene. The specific calculation formula is as follows: in, This represents the cosine similarity between the combined scene feature vector and the scene template vector. This represents a template vector for a certain type of scene in the historical scene fingerprint database, where m represents the dimension of the feature vector. This represents the k-th dimension value of the comprehensive scene feature vector. This represents the k-th dimension value of the scene template vector. The dot product of two vectors reflects the consistency of their directions. This represents the magnitude of the two vectors, used to normalize the dot product result; Based on the calculated highest similarity value, the final scene category is determined according to the preset multi-level judgment rules. If the highest similarity exceeds the first preset threshold, it is directly judged as the corresponding scene. If the highest similarity is within the range between the first preset high threshold and the second preset low threshold, a secondary judgment process is initiated. If the highest similarity is lower than the second preset low threshold, it is judged as an unknown scene, and the default conservative power management strategy is initiated.

7. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 1, characterized in that, In S103, after the terminal completes the current scene category determination, it uses the scene category as the search keyword and performs a matching search in the predefined policy mapping table through the terminal's mapping table query interface to obtain the policy identifier that is uniquely associated with the scene category. The terminal uses the policy identifier as an index to call the policy mapping table loading interface to locate and extract the corresponding policy execution parameter set in the policy library. The policy execution parameter set includes the target hardware module identifier, the target power consumption status identifier, and the triggering conditions for state switching. Based on the obtained policy identifier, the policy execution parameter set accesses the policy library, reads and parses the corresponding data records. During the loading process, parameter integrity verification is automatically performed to check whether the policy execution parameter set contains a complete combination of the target hardware module identifier, the target power consumption status identifier, and the triggering conditions for state switching. If any one of the parameters is missing, the policy execution parameter set is determined to be corrupted, and the backup policy execution parameter set corresponding to the scenario is automatically loaded. If the parameters are complete, the policy execution parameter set is temporarily stored in the cache unit and then transmitted to the unified power management controller of the terminal.

8. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 7, characterized in that, During transmission, a data frame encapsulation and check bit format is used to encapsulate the strategy execution parameter set into a fixed-length data frame according to a preset format, and a CRC check bit is added to the end of the data frame. After receiving the data frame, the power management controller first verifies whether there is an error in the data transmission through the CRC check bit. If the verification passes, the parameters in the data frame are parsed. If the verification fails, a transmission error is reported and a request is made to resend the parameter set. After receiving and verifying the parameter set, the power management controller immediately performs parameter parsing. Based on the hardware unit to be controlled, it establishes an independent control task queue for each hardware unit. It determines the state switching timing of each hardware unit based on the trigger conditions, prioritizes adding control tasks of hardware units that switch immediately to the queue, and temporarily stores control tasks of hardware units that switch after meeting specific conditions. It adds them to the queue only after the trigger conditions are detected. Based on the target power consumption status identifier, it determines the specific control parameters of each hardware unit and converts them into low-level instructions that the hardware unit can recognize.

9. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 1, characterized in that, In step S104, the power management controller parses the strategy execution parameter set and generates a set of hardware control instructions with timing logic. The power management controller combines the parsed target state with the timing plan to generate standardized hardware control instructions. The instruction format includes the target hardware module identifier, opcode, parameter value, and execution timing number. All control instructions are sorted by timing number to form an ordered instruction queue, ensuring that they are triggered in a preset order during execution. A set of hardware control instructions is executed, and the power consumption state of the target hardware unit is switched from the current state to the state defined by the target power consumption state identifier according to the preset order. A first control word is written to the register of the target hardware unit to configure its internal clock gating unit for dynamic power consumption management. By writing a specific bit sequence, the distribution of the clock tree inside the target hardware unit is controlled, and the clock signals of unnecessary functions are selectively turned off, reducing its dynamic power consumption to zero. Next, a second control word is sent to the target power management unit to control the switching state of its power supply domain. By sending the second control word, the power gating switch in the power supply network is controlled. A third control instruction is sent to the bus arbiter to actively notify the interconnect fabric to suspend all data transmission requests to the target hardware unit and place it in an isolated state.

10. The power consumption optimization method for intelligent terminals based on active and passive identification according to claim 9, characterized in that, After the first preset delay following the execution of the hardware control instruction, the status register of the target hardware unit is read to obtain its actual power consumption status value. The actual power consumption status value is compared with the expected status value of the target power consumption status identifier. If the comparison result is inconsistent, a policy execution exception signal is generated. Based on the policy execution exception signal, the backup policy execution process is triggered. The backup policy execution process is pre-stored in the backup partition of the policy library.

Images