An adaptive power management method and system for bus-powered intelligent devices

By analyzing the current waveform characteristics of bus-powered intelligent devices and estimating the instruction computation overhead, a bus overcurrent status level identifier is generated. Combined with pulse width expansion and time-sharing sleep mode of sensor components, the problems of overcurrent risk and insufficient power supply in bus power supply management are solved, and dynamic power consumption management and efficient power utilization of devices are realized.

CN123113302APending Publication Date: 2026-11-17PLIMER INTELLIGENT TECH (SHANGHAI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611267043.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing bus power management methods cannot effectively distinguish between instantaneous current spikes and sustained overloads, leading to overcurrent risks or insufficient power supply. Furthermore, they fail to adaptively adjust to changes in the dynamic computational overhead of internal modules, resulting in low bus power utilization.

Method used

By acquiring the original current waveform records of the power bus and the list of instructions to be executed from the core control unit, waveform feature analysis and instruction computation overhead estimation are performed to generate a bus overcurrent status level identifier. Dynamic power consumption management is achieved by utilizing the pulse width expansion process and the time-sharing sleep-wake-up orchestration of sensor components.

Benefits of technology

It enables dynamic power consumption scheduling in a bus-powered environment, improving device functional continuity and power utilization efficiency, and avoiding overcurrent risks and power supply bottlenecks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113302A_ABST
    Figure CN123113302A_ABST
Patent Text Reader

Abstract

This invention provides an adaptive power management method and system for bus-powered intelligent devices, relating to the fields of bus power supply control and low-power embedded systems. The method involves analyzing the current waveform to obtain transient current spike amplitude characteristics and steady-state current duration characteristics, and estimating the computational overhead of the instruction task list to obtain a timing distribution description. The two sets of features are then fused to determine the overcurrent situation level, generating an overcurrent situation level identifier. Based on this identifier, the charging rate of the pulse width broadening process is adjusted to obtain a control instruction. The current waveform is reshaped using the spike amplitude characteristics and the control instruction, and a time-division sleep / wake-up orchestration of sensor components is performed based on the reshaped record to obtain an orchestration sequence. The orchestration sequence and the current waveform reshaped record are synchronously sent to the core control unit for dynamic power management scheduling. This invention achieves adaptive and refined power management in bus-powered scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of bus power supply control and low-power embedded system technology, and more specifically, to an adaptive power management method and system for bus-powered intelligent devices. Background Technology

[0002] Bus-powered intelligent devices are widely used in industrial automation, building control, and IoT sensing. They simultaneously transmit power and communicate signals via a single power bus, eliminating the need for independent power lines. However, in a bus-powered architecture, the available power of the device is entirely limited by the carrying capacity of the power bus. When multiple functional modules within the device operate simultaneously or alternately, the bus current demand may fluctuate drastically, leading to overcurrent risks or insufficient power supply. Existing bus power management methods typically employ fixed-threshold overcurrent protection or static power budget allocation strategies. For example, a fixed current upper limit threshold is preset, and power to some modules is cut off once the bus current exceeds this threshold, or the power consumption of each module is pre-allocated based on empirical data during device initialization. These methods have significant shortcomings: First, fixed-threshold strategies cannot distinguish between instantaneous current spikes and sustained overloads, easily misjudging normal transient current demands and unnecessarily interrupting device functionality. Second, static power budget allocation fails to consider the dynamic computational overhead changes of the core control unit's actual instruction execution tasks, and cannot adaptively adjust between instruction load peaks and troughs, resulting in low bus power utilization or power bottlenecks during peak periods. Furthermore, existing technologies lack a mechanism for jointly analyzing bus current waveform characteristics with internal instruction computation requirements, and also fail to coordinate current waveform reshaping with time-sharing sleep and wake-up of sensor components, making it impossible to achieve dynamic power consumption scheduling in a bus power supply environment while ensuring the integrity of device functions. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide an adaptive power management method for a bus-powered intelligent device, the method comprising: Obtain the original current waveform record conducted by the power bus, and obtain the list of instructions to be executed from the core control unit in the bus-powered intelligent device; The original current waveform record is subjected to waveform feature analysis processing to obtain the transient current peak amplitude characteristics and steady-state current duration characteristics of the original current waveform record. The instruction operation overhead of the task list to be executed is estimated to obtain the instruction operation overhead timing distribution description of the core control unit. The transient current spike amplitude characteristics and the steady-state current duration characteristics are combined with the instruction operation overhead timing distribution description to perform overcurrent situation fusion judgment processing, generate bus overcurrent situation level identifier, and perform charging rate regulation processing on the pulse width expansion process according to the bus overcurrent situation level identifier to obtain pulse width expansion charging rate regulation instruction. The transient current peak amplitude characteristics and the pulse width expansion charging rate control command are used to reshape the original current waveform record to generate a reshaped current waveform record. Based on the reshaped current waveform record, the sensor components in the bus-powered intelligent device are subjected to time-division sleep-wake-up orchestration to obtain the time-division sleep-wake-up orchestration sequence of the sensor components. The time-sharing sleep-wake-up sequence of the sensor components and the current waveform reshaping record are synchronously sent to the core control unit to perform dynamic power management scheduling.

[0004] Furthermore, embodiments of the present invention also provide an adaptive power management system for a bus-powered intelligent device, comprising: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the aforementioned adaptive power management method for a bus-powered intelligent device by executing the machine-executable instructions.

[0005] Based on the above, the process first synchronously acquires the original current waveform records conducted by the power bus and the list of instructions to be executed from the core control unit. Then, waveform feature analysis is performed on the original current waveform records to extract transient current spike amplitude characteristics and steady-state current duration characteristics. Simultaneously, instruction computation overhead is estimated from the list of instructions to be executed to obtain a time-series distribution description of computation overhead. Subsequently, the two sets of features are fused to determine the overcurrent situation, generating a bus overcurrent situation level identifier. Based on this identifier, the charging rate is adjusted according to the pulse width broadening process, thereby dynamically adjusting the bus charging strategy based on the real-time overcurrent situation rather than simply cutting off power. Furthermore, the original current waveform is reshaped using the transient current spike amplitude characteristics and the charging rate adjustment instructions. Based on the reshaped current waveform, the sensor components are scheduled for time-division sleep / wake-up, ensuring precise matching between the sensor component's operating timing and the bus current carrying capacity. Finally, the sensor time-sharing sleep-wake-up sequence and current waveform reshaping record are synchronously sent to the core control unit to perform dynamic power management scheduling. This deeply integrates the physical characteristics of the bus current waveform with the logical characteristics of the instruction operation overhead. The overcurrent status level is used as a bridge to drive the coordinated regulation of charging rate and sensor sleep. This changes the traditional coarse mode of fixed threshold protection and static power allocation. While ensuring the safety of bus power supply, it maximizes the continuity of device functions and power utilization efficiency. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the execution flow of the adaptive power management method for bus-powered intelligent devices provided in an embodiment of the present invention.

[0007] Figure 2 This is a schematic diagram of the waveform feature analysis, overcurrent situation fusion judgment and charging rate regulation scenario provided by the embodiments of the present invention.

[0008] Figure 3 This is a schematic diagram of the overall power management architecture and current waveform reshaping scenario of the bus-powered intelligent device provided in this embodiment of the invention.

[0009] Figure 4 This is a schematic diagram of a sensor time-sharing sleep-wake-up orchestration and dynamic power consumption management scheduling scenario provided in an embodiment of the present invention. Detailed Implementation

[0010] Figure 1 This is a flowchart illustrating an adaptive power management method for a bus-powered intelligent device according to an embodiment of the present invention. This embodiment provides an adaptive power management method for a bus-powered intelligent device, applied to an intelligent sensing node that simultaneously transmits electrical energy and data via a power bus. The intelligent sensing node is connected to a two-wire power bus and is powered by a bus power management unit, which draws power from the power bus to provide operating power to the core control unit, sensor components, communication module, and load switching component. The bus power management unit includes a current sampling resistor, a differential amplifier, an analog-to-digital converter, a pulse width expansion circuit, an energy storage capacitor bank, and a backup energy storage module. The core control unit is a microcontroller, integrating an instruction pipeline, an interrupt controller, and a clock management interface. The sensor components include a temperature sensor, a humidity sensor, a barometric pressure sensor, and an accelerometer. All data acquisition and processing involved in this embodiment are completed internally within the intelligent sensing node, without involving external transmission or storage of user privacy data.

[0011] Step S110: Obtain the original current waveform record of the power bus conduction and obtain the list of instructions to be executed from the core control unit in the bus-powered intelligent device.

[0012] In this embodiment, the bus power management unit converts the bus current into a voltage signal through a sampling resistor connected in series at the power bus input. This voltage signal is then amplified by a differential amplifier and input to the analog-to-digital converter (ADC). The ADC continuously samples the voltage signal at a sampling frequency Fs, where Fs covers the spectral range of transient spikes in the bus current. The digital sampling sequence output by the ADC is the original current waveform record, denoted as IW. IW is a one-dimensional time series with a length of N. The value at each sampling point is the current amplitude in mA, and the time interval between adjacent sampling points is Ts, where Ts = 1 / Fs in seconds. The core control unit sends its list of instructions to be executed to the bus power management unit via the internal bus. This list is denoted as TL, and TL is a structured data array. Each element of the array is an instruction record, containing an opcode field OP, an operand field OD, and a sequence number field SQ. The total number of instruction records in TL is denoted as M. The bus power management unit stores IW and TL in its internal data buffer.

[0013] Step S120: Perform waveform feature analysis on the original current waveform record to obtain the transient current peak amplitude characteristics and steady-state current duration characteristics of the original current waveform record, and perform instruction operation overhead estimation on the task list to be executed to obtain the instruction operation overhead timing distribution description of the core control unit.

[0014] Step S121: Perform current slope timing scan processing on the original current waveform record. When the current amplitude rise rate of adjacent sampling points in the original current waveform record exceeds the preset slope trigger threshold, generate a spike capture enable flag.

[0015] In this embodiment, the bus power management unit performs current slope timing scan processing on IW. From the 1st sampling point to the (N-1)th sampling point, the current amplitude rise rate of adjacent two sampling points is calculated one by one. For the nth sampling point and the (n+1)th sampling point, the rise rate SR is calculated as SR(n) = [IW(n+1) - IW(n)] / Ts. The unit of SR(n) is mA / s. The preset slope trigger threshold is denoted as ST. Each SR(n) is compared with ST. When SR(n) is greater than ST, a spike capture enable flag EN is generated and set to the active state.

[0016] Step S122: In response to the spike capture enable flag, perform peak lock tracking processing on the current amplitude jump section in the original current waveform record, and use the maximum current amplitude sampled in the current amplitude jump section as the transient current spike amplitude feature.

[0017] In this embodiment, when EN is set to the active state, the bus power management unit enters peak lock tracking state. Starting from the nth sampling point that triggered EN activation, the scanning continues in the direction of increasing sampling point number. The maximum current amplitude currently scanned is recorded as IP, with an initial value set to IW(n). For the mth subsequent sampling point, if IW(m) is greater than IP, IP is updated to IW(m). At the same time, SR(m) for the mth sampling point continues to be calculated. When SR(m) drops back to less than or equal to 0, the rise segment ends. The IP at this time is taken as the transient current peak amplitude feature corresponding to this peak capture, and recorded as PK. PK is stored in the feature parameter buffer.

[0018] Step S123: When the current amplitude rise rate of adjacent sampling points in the original current waveform record falls back to below the preset slope trigger threshold, the number of consecutive sampling points in the original current waveform record that remain within the rated current amplitude range is accumulated and the accumulated timing result is used as the steady-state current duration feature.

[0019] In this embodiment, after EN fails, the bus power supply management unit enters the steady-state current duration statistics state. Starting from the sampling point where SR(m) falls back to less than or equal to 0, the rated current amplitude range [IL, IH] is defined, where IL and IH are in mA. IW is scanned point by point. For the k-th sampling point, it is checked whether IW(k) satisfies IL≤IW(k)≤IH. If it does, the value of counter CS is incremented by 1. The scan continues until IW(k) exceeds the range or a new EN is triggered again. CS is multiplied by Ts to obtain the steady-state current duration feature TS, TS=CS×Ts, in seconds. TS is stored in the feature parameter buffer.

[0020] Step S124: Perform instruction category decomposition on the task list to be executed to obtain the computational complexity category identifier and instruction execution cycle number corresponding to each instruction.

[0021] In this embodiment, the bus power management unit performs instruction category decomposition processing on each instruction record in TL. The OPq of the q-th instruction record is retrieved and searched for in a preset instruction computation complexity mapping table. The computation complexity category identifiers are divided into three levels: CL, CM, and CH, corresponding to low, medium, and high complexity, respectively. For the q-th instruction, its computation complexity category identifier is denoted as CQ. Simultaneously, the instruction execution cycle number CY, in clock cycles, is obtained from the mapping table. CQ and CY are then associated and stored in the instruction computation overhead array.

[0022] Step S125: Construct a time-series distribution description of instruction computation overhead of the core control unit during the execution period of the instruction task list to be executed, based on the computational complexity category identifier and the instruction execution cycle number.

[0023] In this embodiment, the bus power management unit constructs an instruction overhead timing distribution description (DD) based on the CQ and CY of each instruction, combined with the current system clock frequency FC of the core control unit (in Hz). The execution time length Tq = CYq / FC of each instruction is calculated in seconds. The M instructions in TL are arranged on the time axis in SQ order, with the first instruction starting at the current time TN and ending at TN+T1, the second instruction starting at TN+T1 and ending at TN+T1+T2, and so on. The power consumption generated by each instruction within its execution time interval is obtained according to its CQ mapping: CL maps to PL, CM maps to PM, and CH maps to PH, all in mW. The execution time interval and power consumption of each instruction are combined into a record. All M instruction records are arranged in time axis order to form the DD.

[0024] Step S126: Record the transient current peak amplitude characteristics and steady-state current duration characteristics as independent feature parameters, and associate them with the instruction operation overhead timing distribution description and store them in the current feature history queue.

[0025] In this embodiment, the bus power supply management unit records PK and TS as two independent characteristic parameter variables. PK, TS, and DD are associated as a triplet record, and this triplet record is appended to the tail of the current characteristic history record queue QH. QH is a first-in-first-out circular queue with a maximum storage capacity of QM records. When QH is full, the newly written record overwrites the earliest written record at the head of the queue.

[0026] Step S130: The transient current spike amplitude characteristics and steady-state current duration characteristics are combined with the instruction operation overhead timing distribution description to perform overcurrent situation fusion judgment processing, generate bus overcurrent situation level identifier, and perform charging rate regulation processing on the pulse width expansion process according to the bus overcurrent situation level identifier to obtain the pulse width expansion charging rate regulation instruction.

[0027] Step S131: Standardize the transient current peak amplitude characteristics and steady-state current duration characteristics to obtain standardized transient current peak index and steady-state current duration index.

[0028] In this embodiment, the bus power supply management unit standardizes PK. The standardization method is to divide PK by a preset peak amplitude reference base PR, where PR is the upper limit of the allowable transient peak current amplitude of the smart sensor node under maximum rated load, in mA. The standardized transient current peak index SP is calculated as SP = PK / PR, and SP is dimensionless. TS is also standardized. The standardization method is to divide TS by a preset steady-state duration reference base TR, where TR is the expected steady-state current duration of the smart sensor node within a typical operating cycle, in seconds. The standardized steady-state current duration index SS is calculated as SS = TS / TR, and SS is dimensionless.

[0029] Step S132: Using the standardized transient current spike index and steady-state current duration index as inputs, perform current surge intensity assessment processing to generate the current surge intensity index.

[0030] In this embodiment, the bus power supply management unit uses SP and SS as input parameters to perform current surge intensity assessment. A weighted geometric mean model is used. The current surge intensity index IX is calculated as follows: A is the peak intensity weighting coefficient, B is the steady-state duration weighting coefficient, and A+B=1. IX is dimensionless.

[0031] Step S133: Extract the peak amplitude and duration of the computational overhead in the future time period from the instruction computational overhead timing distribution description, perform power supply current demand conversion on the peak amplitude of the computational overhead to obtain the predicted current surge peak amplitude, and perform time axis mapping on the duration of the computational overhead peak to obtain the predicted current surge duration.

[0032] In this embodiment, the bus power management unit extracts the computational overhead characteristics for the future time period from the DD (Distributed Controller). The future time period is defined as the time interval from the current time TN to TN+TW, where TW is the prediction time window length. The power consumption value sequence within this time interval is extracted from the DD, and the peak amplitude of the computational overhead PP is found, which is the maximum power consumption within this time interval, in mW. The power supply current demand is converted for PP, and the predicted current surge peak amplitude IP is calculated as follows: IP = PP / (VB × ET), where VB is the nominal voltage of the power bus in V, and ET is the power conversion efficiency of the core control unit, dimensionless. The unit of IP is mA. The duration corresponding to PP is obtained as the peak duration TP of the computational overhead. TP is time-axis mapped, and the predicted current surge duration TT is directly taken as TP, in seconds.

[0033] Step S134: Map the current surge intensity index, the predicted current surge peak amplitude, and the predicted current surge duration to a unified situation assessment space, perform multi-dimensional overcurrent situation aggregation processing, and generate a bus overcurrent situation level identifier.

[0034] In this embodiment, the bus power management unit maps the three indicators IX, IP, and TT to a unified situation assessment space. The mapping method is as follows: IX is multiplied by coefficient K1, IP is normalized and then multiplied by coefficient K2, and TT is normalized and then multiplied by coefficient K3. The IP normalization method is IP / IM, where IM is a preset normalization baseline value. The TT normalization method is TT / TM, where TM is a preset normalization baseline value. The comprehensive situation assessment value SV is calculated as follows: SV = K1 × IX + K2 × (IP / IM) + K3 × (TT / TM), where K1 + K2 + K3 = 1. The SV value range is mapped to the bus overcurrent situation level identifier LV. Multiple situation level classification thresholds are preset. When SV is less than the first situation threshold, LV is LO. When SV is between the first and second situation thresholds, LV is MD. When SV is greater than the second situation threshold, LV is HI.

[0035] Step S135: Based on the bus overcurrent status level identifier, control the number of selected charging current levels of the pulse width widening capacitor during the pulse width widening process, and generate a charging current level selection control command.

[0036] In this embodiment, the bus power management unit regulates the charging rate of the pulse width expansion process based on LV. The total number of charging current levels in the pulse width expansion circuit is denoted as NG, with each level corresponding to a fixed charging current IC. The number of charging current level selections NS is determined as follows: when LV is LO, NS is NL; when LV is MD, NS is NM; when LV is HI, NS is NH. NL is less than NM, less than NH, and less than or equal to NG. The bus power management unit generates a charging current level selection control instruction CG, where CG is a bitmask of length NG bits, with NS bits set to 1 and the rest set to 0.

[0037] Step S136: The charging rate of the stretching capacitor during the pulse width stretching process is adjusted by using the charging current level selection control command, and the start time of the discharge period of the stretching capacitor during the pulse width stretching process is delayed and matched according to the predicted current surge duration, so as to generate a pulse width stretching charging rate control command.

[0038] In this embodiment, the bus power supply management unit sends CG to the gating switch array of the pulse width expansion circuit, and the expansion capacitor is charged with the total charging current NS×IC. The start time of the discharge period of the expansion capacitor is delayed and matched according to TT. The discharge start time TD is calculated as: TD=TN+TO, where TO is the discharge delay offset. The discharge duration TU is TT×KP, where KP is the expansion coefficient, and KP is greater than 1. CG, TD, and TU are encapsulated into a pulse width expansion charging rate control instruction CP. CP contains three fields: the G bitmask field is assigned the value CG, the D start time field is assigned the value TD, and the U duration field is assigned the value TU.

[0039] Step S137: Associate and bind the bus overcurrent status level identifier with the pulse width expansion charging rate control command, generate an overcurrent status control associated record entry, and store the overcurrent status control associated record entry in the control history storage area.

[0040] In this embodiment, the bus power supply management unit associates and binds LV and CP, generating an overcurrent condition control association record entry RV=(LV, CP, TN). RV is then appended to the control history storage area.

[0041] In the steps of performing waveform feature parsing processing, such as Figure 2 As shown, the original current waveform record is first input into the waveform feature analysis processing stage. This stage extracts the transient current peak amplitude characteristics located at the peak position and the steady-state current duration characteristics located in the stable segment of the waveform. Simultaneously, the list of instructions to be executed is input into the instruction computation overhead estimation processing stage, which generates a time-series distribution description of the instruction computation overhead that increases in a stepwise manner over time. For example... Figure 2 As shown, the transient current spike amplitude characteristics, steady-state current duration characteristics, and instruction computation overhead timing distribution descriptions are all incorporated into the overcurrent status fusion and judgment processing stage via arrows. This stage outputs three evaluation parameters through comprehensive calculation: current surge intensity index, predicted current surge peak amplitude, and predicted current surge duration, and generates a bus overcurrent status level identifier based on these parameters.

[0042] like Figure 2 As shown, based on the bus overcurrent status level indicator, charging current level selection control is executed. The figure displays multiple charging rise curves with different slopes on a coordinate system showing the change of capacitor terminal voltage over time, representing charging current level 1, charging current level 2, and charging current level 3. This selection control, combined with discharge start-time delay matching, forms a pulse width broadening charging rate regulation command. This command internally includes two modules: charging rate adjustment and discharge start-time delay matching. Figure 2As shown, after the pulse width expansion charging rate control command is associated and bound with the bus overcurrent status level identifier, it is stored in the control history storage area as an overcurrent status control association record entry. Simultaneously, the figure also shows the backup energy storage access control signal, which acts on the backup energy storage module, triggering the backup energy storage module to access the power bus to supplement electrical energy.

[0043] Step S140: The original current waveform record is reshaped using the transient current spike amplitude characteristics and the pulse width expansion charging rate control command to generate a current waveform reshaped record. Based on the current waveform reshaped record, the sensor components in the bus-powered intelligent device are subjected to time-sharing sleep and wake-up orchestration to obtain the time-sharing sleep and wake-up orchestration sequence of the sensor components.

[0044] Step S141: According to the pulse width expansion charging rate control command, control the transient peak current in the original current waveform record to transfer charge to the expansion capacitor in the pulse width expansion process, and expand the transient peak current into a smooth supplementary current waveform in the time dimension to obtain the current waveform reconstructed record.

[0045] In this embodiment, the bus power management unit controls the pulse width expansion circuit according to the CP control. When a transient spike current occurs in IW, the expansion capacitor absorbs and stores part of the charge from the transient spike current, thus reducing the amplitude of the transient current spike on the power bus. The absorbed charge is released back to the power bus as a smooth discharge current during the TU period. The current component corresponding to the charge absorbed by the expansion capacitor is subtracted from the transient spike amplitude in IW, and then the discharge current component of the expansion capacitor is superimposed during the TU period to obtain the current waveform reshaping record IR. IR is a one-dimensional time series of the same length as IW, and the value at each sampling point is the reshaped current amplitude in mA.

[0046] Step S142: Perform trough interval identification processing on the current waveform reshaping record to identify the available sleep time window in the current waveform reshaping record where the current amplitude is lower than the sensor sleep current threshold.

[0047] In this embodiment, the bus power management unit performs trough interval identification processing on the IR. A preset sensor sleep current threshold IS is established. Each sampling point in the IR is traversed, and consecutive sampling point intervals with current amplitudes lower than IS are identified as trough intervals. For each trough interval, its start and end sampling point numbers are recorded and converted into start and end times. Each trough interval corresponds to a sleep available time window, denoted as WS, which includes the start time WS.start and the duration WS.dur.

[0048] Step S143: Based on the start time and duration of the available sleep time window, perform sleep time allocation processing on the sensor components in the bus-powered smart device to obtain the sleep time allocation plan for each sensor component.

[0049] In this embodiment, the intelligent sensing node includes U sensor components, denoted as SU1 to SUU. Each sensor component has independent sleep capability. All WS identified in step S142 are arranged in chronological order. A greedy allocation algorithm is used: for each WS, the sensor component with the most urgent sleep requirement is selected from those that have not been allocated a sleep period, and that WS is allocated to it as a sleep period. After allocation, the sleep start time and sleep duration of the sensor component are recorded. The sleep period allocation of all sensor components constitutes a sleep period allocation plan PL, where PL is an array of U elements, each containing the sleep start time and sleep duration.

[0050] Step S144: Sort the power-on transient current requirements of each sensor component in the bus-powered intelligent device to obtain the power-on transient current requirement sequence of the sensor components.

[0051] In this embodiment, each sensor component experiences transient current demand from power-on startup to reaching a stable operating state. The power-on transient current demand of the u-th sensor component SUu is denoted as IUu, in mA. The U sensor components are sorted in descending order of IUu, and the resulting sequence is the sensor component power-on transient current demand sequence SQ. Each element in SQ contains the sensor component identifier and the corresponding IU.

[0052] Step S145: Based on the transient current demand sequence of the sensor component upon power-on and the remaining power supply capacity at the peak moment in the current waveform reshaping record, the wake-up time staggered arrangement process is performed to obtain the time-sharing sleep wake-up arrangement sequence of the sensor component.

[0053] In this embodiment, the bus power management unit obtains the current amplitude at the peak moment in the IR and calculates the remaining power supply capacity RM of the power bus at the peak moment. RM is equal to the maximum power supply current of the bus minus the current amplitude at the corresponding moment in the IR, with the unit being mA. Starting with the sensor component with the largest IU in SQ, a wake-up time is scheduled for it. The wake-up time is preferentially selected when RM is greater than IUu and is staggered from other scheduled wake-up times. If RM is less than IUu at a certain time point, the wake-up time of that sensor component is postponed until RM is sufficiently recovered. The wake-up times and sleep times of all sensor components constitute the sensor component time-sharing sleep-wake-up arrangement sequence SS. SS is an array of U elements, each element containing the sensor component identifier, sleep start time, sleep end time (i.e., wake-up time).

[0054] Step S146: The time-sharing sleep and wake-up orchestration sequence of the sensor component is split into a sensor sleep trigger signal sequence and a sensor wake-up trigger signal sequence, and the sensor sleep trigger signal sequence and the sensor wake-up trigger signal sequence are written into the sensor sleep scheduling queue and the sensor wake-up scheduling queue, respectively.

[0055] In this embodiment, the bus power management unit extracts the sleep start time of each sensor component from the SS and arranges them in chronological order to form the sensor sleep trigger signal sequence SL. It also extracts the wake-up time of each sensor component from the SS and arranges them in chronological order to form the sensor wake-up trigger signal sequence SW. The SL is written to the sensor sleep scheduling queue QL, and the SW is written to the sensor wake-up scheduling queue QW. QL and QW are two timing scheduling queues within the bus power management unit, which automatically send sleep trigger signals or wake-up trigger signals to the corresponding sensor components when the corresponding time arrives.

[0056] Step S150: The time-sharing sleep-wake-up sequence of the sensor components and the current waveform reshaping record are synchronously sent to the core control unit to perform dynamic power management scheduling.

[0057] In this embodiment, the bus power supply management unit synchronously sends the SS and IR to the core control unit via the internal bus. After receiving the SS and IR, the core control unit loads them into the dynamic power management scheduler. Based on the sleep and wake-up times specified in the SS for each sensor component, the dynamic power management scheduler issues sleep or wake-up commands to the sensor components at the corresponding times. Simultaneously, based on the reshaped current waveform information in the IR, the dynamic power management scheduler adjusts the instruction execution cycle of the core control unit itself, matching the core control unit's operational power consumption curve with the reshaped current waveform, thus achieving overall dynamic power management scheduling for the bus-powered intelligent device.

[0058] In the steps of acquiring the original current waveform record and performing waveform reshaping, such as Figure 3 As shown, the power supply from the mains source provides power to the bus-powered intelligent device via the power bus. The power bus carries a record of the raw current waveform, which contains clear transient current spikes. The internal diagram of the bus-powered intelligent device shows three main components: the core control unit, the pulse width expansion process, and the sensor components. The core control unit contains a list of tasks to be executed, the pulse width expansion process contains the expansion capacitor, and the sensor components are represented by multiple black rectangular units.

[0059] like Figure 3As shown, transient current spikes in the original current waveform record are introduced into the pulse width broadening process through a charge transfer broadening mechanism (illustrated by a combination of dashed arrows and circular symbols in the figure), and after processing by the broadening capacitor, are reconstructed into a current waveform reconstructed record. The smooth waveform of this reconstructed current waveform record provides a basis for subsequent sleep execution. Figure 3 As shown, after waveform reshaping is completed, the bus power management scheduling unit receives dynamic power management scheduling signals from the core control unit and sensor components, and performs bus power management scheduling in conjunction with the current waveform reshaping record to maintain the stability and balance of bus power supply.

[0060] In the step of performing time-division sleep-wake-up orchestration on the sensor components based on the current waveform reshaping record, such as Figure 4 As shown, the sensor components mounted on the internal bus include sensor 1, sensor 2, sensor 3, and so on up to sensor N. The diagram clearly distinguishes the timeline states of each sensor component into wake-up state (solid rectangle) and sleep state (dashed hollow rectangle), and the wake-up times are staggered to allow each sensor component to alternately switch states on the timeline. Below the timeline is a continuous curve reflecting bus load fluctuations; the peaks of the curves indicate remaining power capacity, and the troughs indicate available sleep time windows. This orchestration operation ultimately generates a time-sharing sleep-wake sequence for the sensor components.

[0061] like Figure 4 As shown, the time-sharing sleep / wake-up orchestration sequence of the sensor component is transmitted to the core control unit. The core control unit is internally configured with a clock management interface and dynamic power management scheduling logic. In this stage, the time-sharing sleep / wake-up orchestration sequence of the sensor component is split into a sensor sleep trigger signal sequence and a sensor wake-up trigger signal sequence. These two sets of signal sequences are written into the sensor sleep scheduling queue and the sensor wake-up scheduling queue inside the core control unit, respectively. The core control unit executes various power scheduling operations according to the queue timing.

[0062] like Figure 4 As shown, during dynamic power management and scheduling, the diagram illustrates the RF front-end power consumption estimate for the communication module's thermal adjustment stage. Based on this estimate, a power consumption adjustment command for the communication module is generated. This command is used to adjust the communication module's transmit power level and data packet transmission interval in conjunction with the thermal adjustment stage. Simultaneously, the diagram shows the queuing status in the interrupt waiting queue for the response acceleration stage. Based on this, an interrupt response acceleration frequency configuration command is generated. This command operates on the clock management interface to improve the core control unit's interrupt handling response speed.

[0063] Step S210: During the execution of the task list of instructions to be executed by the core control unit, the actual current consumption waveform of the core control unit is collected and recorded synchronously.

[0064] In this embodiment, during the execution of instructions according to TL by the core control unit, the bus power management unit synchronously collects the actual current consumption of the core control unit's power supply pins. The actual current consumption waveform IA is collected and recorded as IW through a sampling resistor and an analog-to-digital converter connected in series in the core control unit's power supply circuit. IA and IW have the same sampling frequency Fs and time base.

[0065] Step S220: Convert the actual current consumption waveform record into an equivalent computational overhead timing distribution corresponding to the actual instruction execution amount.

[0066] In this embodiment, the bus power management unit converts the IA (Input / Output) into an equivalent operational overhead timing distribution (DA). The conversion method involves multiplying the current amplitude at each sampling point in the IA by the core control unit's power supply voltage VC to obtain the power consumption value at the corresponding moment, in mW. The power consumption values ​​at all sampling moments are arranged in time to form the DA.

[0067] Step S230: Perform deviation trend analysis on the equivalent computation overhead timing distribution and the instruction computation overhead timing distribution to generate a computation overhead deviation trend curve.

[0068] In this embodiment, the difference between DA and DD generated in step S125 is calculated point by point under the same time reference. For each sampling time, the deviation value DV is equal to the power consumption value in DA at that time minus the power consumption value in DD at that time, with the unit being mW. The DV values ​​of all sampling times are arranged by time to form the computational overhead deviation trend curve DC.

[0069] Step S240: Perform deviation trend identification processing on the computational cost deviation trend curve to generate a deviation trend category identifier.

[0070] In this embodiment, trend analysis is performed on DC. The first-order difference sequence of DC is calculated. If multiple consecutive values ​​in the first-order difference sequence are positive and the cumulative value of DC continues to increase, the deviation trend is determined to be a positive expanding trend, and the deviation trend category identifier TD is set to UP. If multiple consecutive values ​​are negative and the cumulative value of DC continues to decrease, the TD is set to DN. If the first-order difference fluctuates around zero, the TD is set to ST.

[0071] Step S250: Based on the deviation trend category identifier, the system clock frequency of the core control unit is adjusted by frequency ramping or ramping to obtain the system clock frequency adjustment command.

[0072] In this embodiment, the system clock frequency FC of the core control unit is adjusted according to TD. If TD is UP, it indicates that the actual power consumption is consistently higher than the estimate, and FC needs to be reduced to decrease computational overhead. If TD is DN, it indicates that the actual power consumption is consistently lower than the estimate, and FC can be increased to improve computational performance. If TD is ST, FC remains unchanged. The frequency-adjusted system clock frequency FN = FC + ΔF × DIR, where ΔF is the frequency adjustment step, DIR is -1 when TD is UP, +1 when TD is DN, and 0 when TD is ST. A system clock frequency adjustment instruction CF is generated, and CF contains a target frequency field assigned the value FN.

[0073] Step S260: Send the system clock frequency adjustment command to the clock management interface of the core control unit to adjust the operation and processing rhythm of the core control unit.

[0074] In this embodiment, the bus power supply management unit sends the CF (Clock Response) to the clock management interface of the core control unit via the internal bus. After receiving the CF, the clock management interface modifies the configuration parameters of the internal phase-locked loop or clock divider, adjusting the system clock frequency from FC (Functional Clock) to FN (Frequency Clock), thereby regulating the operation and processing rhythm of the core control unit.

[0075] Step S270: Obtain the control response feedback signal of the core control unit after executing the system clock frequency control command.

[0076] In this embodiment, after the core control unit completes the clock frequency switch, it returns a regulation response feedback signal FB to the bus power management unit. FB includes the actual system clock frequency FA after the switch and the timestamp of the switch completion.

[0077] Step S280: Correct the frequency modulation step amplitude of the system clock frequency control command according to the control response feedback signal, and generate the corrected system clock frequency control command.

[0078] In this embodiment, the bus power management unit corrects the frequency modulation step size ΔF based on the difference between the actual switching frequency FA and the target frequency FN in the FB. The corrected frequency modulation step size ΔFc = ΔF × (FN / FA). If frequency modulation is required again later, ΔFc is used instead of ΔF. A corrected system clock frequency control instruction CFc is generated, and the frequency modulation step size field in CFc is updated to ΔFc.

[0079] Step S290: Overwrite the currently executed frequency modulation parameter configuration in the clock management interface with the corrected system clock frequency modulation command.

[0080] In this embodiment, the bus power management unit sends the CFc to the clock management interface, overwriting the currently stored frequency modulation parameter configuration in the interface. Subsequent frequency modulation operations will be based on the corrected parameters.

[0081] Step S2100: Collect the actual instruction throughput of the core control unit after the execution of the corrected system clock frequency regulation instruction, calculate the throughput deviation between the actual instruction throughput and the expected instruction throughput in the instruction operation overhead timing distribution description, and perform feedback calibration on the prediction parameters of the instruction operation overhead prediction processing based on the throughput deviation calculation results.

[0082] In this embodiment, the actual number of instructions completed by the core control unit per unit time after CFc execution is collected as the actual instruction throughput RT. The expected instruction throughput ET is extracted from DD. The throughput deviation RD = RT / ET is calculated. If RD deviates from 1 by more than a preset tolerance, the power consumption mapping values ​​PL, PM, and PH used in step S125 are calibrated. The calibration method is: PL = PL × RD, PM = PM × RD, PH = PH × RD.

[0083] Step S310: Obtain the surge current spike signal generated by the load switching component in the bus-powered intelligent device at the moment of load switching.

[0084] In this embodiment, the intelligent sensing node controls the connection and disconnection of external high-power loads through a load switching component. The load switching component includes a power switch, which is a metal-oxide-semiconductor field-effect transistor or a relay. At the instant the load switching component performs a load switching operation, a surge current spike signal with a very short duration and an amplitude much higher than the rated operating current is generated on the power bus due to the charging and discharging effect of the parasitic capacitance at the load end or the back electromotive force of the inductive load. The bus power management unit captures this surge current spike signal at a sampling frequency Fs through a current sampling resistor and an analog-to-digital converter. The captured surge current spike signal is denoted as IS. IS is a one-dimensional time series with a sequence length denoted as NS. The value of each sampling point is the current amplitude in mA, and the sampling time interval is Ts.

[0085] Step S320: Perform pulse shaping processing on the surge current spike signal to generate a surge pulse trigger signal.

[0086] In this embodiment, the bus power management unit performs pulse shaping processing on the IS. The pulse shaping process uses a Schmitt trigger algorithm. The Schmitt trigger has two thresholds: a positive threshold VTH and a negative threshold VTL, where VTH is greater than VTL. The processing flow is as follows: Initialize the output state to low level. Traverse each sampling point in the IS. When the output state is low and the current amplitude at the current sampling point is greater than VTH, flip the output state to high level; when the output state is high and the current amplitude at the current sampling point is less than VTL, flip the output state to low level. The flipping times of the output state constitute a rectangular pulse sequence, with each rectangular pulse corresponding to a surge spike interval in the IS. This rectangular pulse sequence is the surge pulse trigger signal PT. PT is a Boolean sequence of the same length as the IS; a high level indicates that the current position is within the surge spike interval, and a low level indicates that the current position is outside the surge spike interval.

[0087] Step S330: Use the surge pulse trigger signal to trigger the fast current discharge process, and bypass the transient energy in the surge current spike signal to the energy storage capacitor bank through the fast current discharge process to obtain the residual current waveform after the surge energy is transferred.

[0088] In this embodiment, the bus power management unit monitors the edge transition of the PT (Power Transistor). When the PT transitions from low to high, a fast current discharge process is triggered. The fast current discharge process is executed as follows: the discharge path switch between the power bus and the energy storage capacitor bank is turned on. The discharge path switch is composed of a high-power metal-oxide-semiconductor field-effect transistor, and its on-resistance is denoted as RON (in Ω). After the discharge path is turned on, the surge current spike in IS flows into the energy storage capacitor bank through the discharge path. The equivalent series resistance of the energy storage capacitor bank is denoted as RESR (in Ω). The surge current spike generates a voltage drop across the discharge path impedance and the equivalent series resistance of the energy storage capacitor bank, and the surge energy is absorbed and stored by the energy storage capacitor bank. The current flowing through the power bus is the original surge current minus the discharge current flowing into the energy storage capacitor bank. The difference between the two constitutes the residual current waveform IR after the surge energy transfer. IR is a one-dimensional time series with the same length as IS. The value of each sampling point is the residual current amplitude, in mA. When the PT changes from high level to low level, the discharge path switch is turned off, and the rapid current discharge process ends.

[0089] Step S340: Perform secondary clamping processing on the residual current waveform to limit the voltage amplitude of the residual current waveform within the bus voltage safety threshold, thereby obtaining the safety clamping voltage signal.

[0090] In this embodiment, the bus power management unit performs secondary clamping processing on the IR. This secondary clamping processing is achieved through a transient voltage suppression diode or an active clamping circuit. The breakdown voltage of the transient voltage suppression diode is denoted as VBV, in volts (V), and VBV is less than the safe upper limit of the nominal voltage of the power bus. The current amplitude at each sampling point in the IR is multiplied by the equivalent impedance of the power bus to obtain the corresponding voltage amplitude. If the calculated voltage amplitude exceeds VBV, the voltage amplitude at that moment is clamped at VBV; otherwise, the original value is maintained. The voltage signal after clamping is the safety clamping voltage signal VC. VC is a one-dimensional time series with the same length as the IR, and the value at each sampling point is the voltage amplitude, in volts (V).

[0091] Step S350: Feedback the safety clamp voltage signal to the input of the power bus to maintain the stability of the bus voltage.

[0092] In this embodiment, the bus power supply management unit connects the voltage regulator (VC) to the voltage feedback loop at the power bus input. The voltage feedback loop includes an error amplifier and a pulse width modulation (PWM) controller. The error amplifier amplifies the difference between the VC and the nominal bus voltage and outputs an error signal. The PWM controller adjusts the duty cycle at the power bus input based on the error signal, ensuring the bus voltage quickly recovers to its nominal value after a surge. The feedback effect of the VC maintains the stability of the bus voltage during load switching transients.

[0093] Step S360: Collect the change in terminal voltage of the energy storage capacitor bank after absorbing transient energy.

[0094] In this embodiment, a voltage detection circuit is connected in parallel across the energy storage capacitor bank. This circuit acquires the terminal voltage of the energy storage capacitor bank in real time through a resistor divider network and an analog-to-digital converter. Before the fast current discharge process begins, the bus power supply management unit records the initial terminal voltage VC0 of the energy storage capacitor bank in volts (V). After the fast current discharge process ends, the final terminal voltage VC1 of the energy storage capacitor bank is recorded again in volts. The change in terminal voltage ΔV of the energy storage capacitor bank is calculated as ΔV = VC1 - VC0, in volts. ΔV reflects the magnitude of the transient energy absorbed by the energy storage capacitor bank during this surge impact.

[0095] Step S370: Adaptively adjust the duration of the fast current discharge process based on the change in the terminal voltage of the energy storage capacitor bank, and generate a discharge duration adjustment command.

[0096] In this embodiment, the discharge duration TD of the fast current discharge process is the time from the turn-on to the turn-off of the discharge path switch, in seconds. The bus power supply management unit adaptively adjusts TD based on ΔV. The adjustment logic is: the updated value of TD, TDnew = TDold × (1 + λ × (ΔV - ΔVt) / ΔVt), where TDold is the currently used discharge duration, λ is the adjustment step coefficient ranging from 0 to 1, and ΔVt is the target expected value of the voltage change at the energy storage capacitor bank, in volts (V). When ΔV is greater than ΔVt, it indicates that too much energy is absorbed, and the discharge duration is appropriately shortened to reduce energy absorption; when ΔV is less than ΔVt, it indicates that too little energy is absorbed, and the discharge duration is appropriately extended to increase energy absorption. A discharge duration adjustment instruction CD is generated, which contains a discharge duration field and is assigned the value TDnew.

[0097] Step S380: Write the discharge duration adjustment instruction into the discharge control parameter table of the fast current discharge process.

[0098] In this embodiment, the bus power management unit writes the CD to the discharge control parameter table of the fast current discharge process. The discharge control parameter table is a configuration data structure in the internal flash memory of the bus power management unit, which includes the discharge duration parameter, the discharge path switch on-resistance parameter, and the target voltage parameter of the energy storage capacitor bank. The writing of the CD overwrites the current value of the discharge duration parameter in the discharge control parameter table. Subsequent fast current discharge processes will use the updated TDnew as the on-resistance of the discharge path switch.

[0099] Step S410: Collect a set of historical current waveform records of the power bus in the bus-powered intelligent device over multiple consecutive monitoring cycles.

[0100] In this embodiment, the bus power supply management unit extracts a set of historical current waveform records HJ from the current characteristic history record queue QH, representing the past J consecutive monitoring cycles. Each monitoring cycle corresponds to a complete set of characteristic parameters, including the transient current peak amplitude feature PKj and the steady-state current duration feature TSj, where j ranges from 1 to J. HJ is a structure array, where each element is (PKj, TSj), for a total of J elements.

[0101] Step S420: Perform periodic energy distribution statistical processing on the historical current waveform record set to generate the energy consumption distribution pattern corresponding to each monitoring period.

[0102] In this embodiment, periodic energy distribution statistical processing is performed on the characteristic parameter pair (PKj, TSj) for each monitoring period in HJ. The numerical range of PK is divided into A intervals, with interval boundaries PA0, PA1, ..., PAA. The numerical range of TS is divided into B intervals, with interval boundaries TB0, TB1, ..., TBB. For the j-th monitoring period, the number of times (PKj, TSj) falls into each two-dimensional interval combination is counted, forming a two-dimensional frequency matrix MJ with A rows and B columns. MJ is the energy consumption distribution pattern corresponding to this monitoring period. A total of J energy consumption distribution patterns MJ1 to MJJ are generated.

[0103] Step S430: Perform mode shift analysis on the energy consumption distribution patterns of adjacent monitoring periods to obtain the energy consumption mode shift trend.

[0104] In this embodiment, mode shift analysis is performed on the energy consumption distribution patterns MJj and MJj+1 of adjacent monitoring periods. The difference in the Frobenius norms of the two matrices is calculated as the j-th mode shift Dj = ||MJj+1||F - ||MJj||F, where ||M||F represents the Frobenius norm of matrix M, which is equal to the square root of the sum of the squares of all elements of the matrix. The J-1 mode shifts Dj constitute the energy consumption mode shift trend sequence ET = [D1, D2, ..., DJ-1].

[0105] Step S440: Dynamically adjust the preset slope trigger threshold according to the energy consumption mode offset trend to generate the adjusted preset slope trigger threshold.

[0106] In this embodiment, the arithmetic mean DA of the most recent offsets of the ET sequence is calculated. The dynamic adjustment formula for the preset slope trigger threshold is: STa = ST × (1 + γ × DA), where ST is the currently used preset slope trigger threshold, STa is the adjusted preset slope trigger threshold, and γ is the adjustment gain coefficient, with a value ranging from -1 to 1. When DA is positive, it indicates that the energy consumption mode is shifting positively, and STa is increased to improve the peak capture trigger sensitivity; when DA is negative, it indicates that the energy consumption mode is shifting negatively, and STa is decreased to reduce the probability of false triggering.

[0107] Step S450: Override the currently used preset slope trigger threshold in the current slope timing scan processing with the adjusted preset slope trigger threshold.

[0108] In this embodiment, the bus power supply management unit writes STa into the slope trigger threshold register, overwriting the currently stored ST value in the register. In the subsequent step S121, when performing the current slope timing scan processing, the preset slope trigger threshold read from the slope trigger threshold register is the updated STa.

[0109] Step S460: Perform periodic peak density statistical processing on the historical current waveform record set to generate the peak occurrence density distribution corresponding to each monitoring period.

[0110] In this embodiment, periodic peak density statistical processing is performed on each monitoring period in HJ. For the j-th monitoring period, its PKj is compared with the preset peak density statistical reference value PR. If PKj is greater than PR, it indicates that the peak amplitude is significant. The number Cj of significant peaks with PKj greater than PR in the j-th monitoring period is counted. The monitoring duration of the j-th monitoring period is denoted as Lj, in seconds. The peak occurrence density DNj is calculated as: DNj = Cj / Lj, in seconds. J DNj constitute the peak occurrence density distribution sequence DS = [DN1, DN2, ..., DNJ].

[0111] Step S470: Dynamically adjust the minimum peak recognition width in the pulse width limiting process based on the peak occurrence density distribution to generate the adjusted minimum peak recognition width.

[0112] In this embodiment, the arithmetic mean DM of the DS sequence is calculated. The dynamic adjustment formula for the minimum spike recognition width is: WMa = WM × (DM / DR), where WM is the currently used minimum spike recognition width in seconds, WMa is the adjusted minimum spike recognition width, and DR is the reference value for spike occurrence density in times / second. When DM is greater than DR, WMa is increased accordingly to filter out more frequent narrow pulse interference; when DM is less than DR, WMa is decreased accordingly to ensure the ability to recognize sparse spikes.

[0113] Step S480: Write the adjusted minimum spike recognition width into the spike recognition width parameter table used in the pulse width limiting process.

[0114] In this embodiment, the bus power supply management unit writes WMa into the spike identification width parameter table, overwriting the currently stored WM value in the parameter table. In the subsequent peak lock tracking process in step S122, current amplitude jump segments with a duration less than WMa will be ignored as noise pulses and will not be identified as valid transient current spikes.

[0115] Step S510: Monitor the real-time voltage amplitude of the power bus in the bus-powered intelligent device. When the real-time voltage amplitude drops below the voltage drop warning threshold, trigger the backup energy storage access control signal.

[0116] In this embodiment, the bus power supply management unit continuously monitors the real-time voltage amplitude VB of the power bus through a voltage detection circuit, with the unit being volts (V). The voltage drop warning threshold is denoted as VL, which is set according to the minimum voltage requirement for normal operation of the intelligent sensing node. When VB is less than VL, the bus power supply management unit triggers the backup energy storage access control signal SC, which is an internal logic trigger signal.

[0117] Step S520: In response to the backup energy storage access control signal, the backup energy storage path is turned on, and the stored charge in the backup energy storage module is injected into the power bus through the backup energy storage path to obtain the power bus voltage after voltage compensation.

[0118] In this embodiment, when SC is triggered, the bus power management unit sends a turn-on command to the backup energy storage path switch. The backup energy storage path switch is composed of a P-channel metal-oxide-semiconductor field-effect transistor. Its gate is driven by the turn-on command, its source is connected to the positive terminal of the backup energy storage module, and its drain is connected to the power bus. After the path switch is turned on, the stored charge in the backup energy storage module is injected into the power bus through the path, providing supplementary current to the bus load. After the supplementary current is injected, the power bus voltage rises from VB to the voltage-compensated power bus voltage VCB, where VCB = VB + Qinj / Cbus, and Qinj is the injected charge in C, and Cbus is the equivalent capacitance of the power bus in F.

[0119] Step S530: When the voltage of the power bus after voltage compensation recovers to above the voltage drop warning threshold and maintains a preset recovery stabilization time, disconnect the backup energy storage path to cut off charge injection.

[0120] In this embodiment, when VCB rises to a level greater than or equal to VL, the bus power management unit starts a recovery stabilization timer, with the timer duration denoted as TH (seconds). If VCB remains above VL within the TH duration, a disconnect command is sent to the backup energy storage path switch upon timer expiration, shutting off the backup energy storage path and stopping charge injection. If VCB falls below VL again within the TH duration, the timer is reset and restarted.

[0121] Step S540: Monitor the remaining charge of the backup energy storage module during the charge injection process. When the remaining charge of the energy storage module is lower than the energy storage module's replenishment charging threshold, extract the available dormancy time window from the current waveform reshaping record.

[0122] In this embodiment, the bus power supply management unit monitors the remaining charge QR of the backup energy storage module using a coulomb counter or open-circuit voltage method, with the unit being C. The energy storage module's recharging threshold is denoted as QC. When QR is less than QC, the nearest unused sleep time window WS is selected from the set of sleep time windows generated in step S142, and its start time WS.start and duration WS.dur are extracted.

[0123] Step S550: Send a supplementary charging enable signal to the backup energy storage module within the available hibernation time window to recharge the backup energy storage module.

[0124] In this embodiment, at WS.start, the bus power management unit sends a supplementary charging enable signal EC to the supplementary charging path switch, thus activating the supplementary charging path. The supplementary charging path draws power from the power bus and charges the backup energy storage module through a constant current charging circuit. The constant current charging current is denoted as ICC, and its unit is mA. During the duration of WS.dur, the amount of charge replenished to the backup energy storage module is Qchg = ICC × WS.dur, and its unit is C.

[0125] Step S560: When the voltage at the end of the standby energy storage module reaches the full charge voltage threshold of the energy storage module, cancel the supplementary charging enable signal to terminate charge recharging.

[0126] In this embodiment, during the charge recharge process, the bus power management unit continuously monitors the terminal voltage VBK of the backup energy storage module. The full charge voltage threshold of the energy storage module is denoted as VF, in volts (V). When VBK is greater than or equal to VF, the supplementary charging enable signal EC is canceled, the supplementary charging path switch is turned off, and the charge recharge is terminated. If VBK does not reach VF within the duration WS.dur, EC is automatically canceled at the end of WS.dur.

[0127] Step S570: Write the start time and end time of charge recharging of the backup energy storage module into the backup energy storage maintenance log.

[0128] In this embodiment, the bus power supply management unit records the start time WS.start and the end time of this charge recharge as a log entry and writes it to the backup energy storage maintenance log. The backup energy storage maintenance log is stored in non-volatile memory and is used for backup energy storage module life assessment and maintenance plan formulation.

[0129] Step S610: Obtain the transmit power configuration information and receive sensitivity configuration information of the communication module in the bus-powered smart device during the data transmission and reception period.

[0130] In this embodiment, the intelligent sensing node includes a communication module, which is a wireless radio frequency transceiver or a bus transceiver. Before the start of the data transmission and reception period, the bus power supply management unit obtains the transmit power configuration information PW and the receive sensitivity configuration information SN through the configuration register read interface of the communication module. PW is the transmit power level or transmit power value, in dBm. SN is the receive sensitivity threshold value, in dBm.

[0131] Step S620: Estimate the power consumption of the radio frequency front-end of the communication module based on the transmit power configuration information and receive sensitivity configuration information to obtain the estimated power consumption of the radio frequency front-end during the communication period.

[0132] In this embodiment, the bus power management unit estimates the RF front-end power consumption of the communication module during data transmission and reception based on PW and SN. The estimation model is: RF = P0 + α × PW + β × |SN|, where P0 is the basic RF front-end power consumption in mW, α is the transmit power consumption coefficient (dimensionless), and β is the receive sensitivity consumption coefficient in mW / dBm. RF is the estimated RF front-end power consumption during the communication period, in mW.

[0133] Step S630: Match the estimated power consumption of the RF front-end during the communication period with the remaining power supply capacity of the corresponding period in the current waveform reshaping record to obtain the power supply matching result for the communication period.

[0134] In this embodiment, current waveform data within the corresponding time interval of the communication period is extracted from the current waveform reshaping record IR generated in step S141. The remaining power supply capacity RM of this time interval is equal to the maximum bus power supply current IM minus the arithmetic mean of the current amplitude in the IR for this period, with RM in mA. RF is converted to the equivalent current demand RFI = RF / VB, in mA. RFI is compared with RM. If RM is greater than or equal to RFI, the power supply matching result is sufficient; if RM is less than RFI, the power supply matching result is insufficient. The power supply matching result is denoted as MR.

[0135] Step S640: Based on the power supply matching results during the communication period, perform joint adjustment processing on the data packet transmission interval and transmission power level of the communication module to generate a power consumption adjustment command for the communication module.

[0136] In this embodiment, if MR is insufficient, the power consumption of the communication module needs to be jointly adjusted to match the power supply capacity. The joint adjustment strategy is as follows: increase the data packet transmission interval from the current value TG to TGa, where TGa = TG × (RM / RFI), and simultaneously decrease the transmit power level from the current value PW to PWa, where PWa = PW - ΔP, and ΔP is the step value for decreasing the transmit power level in dB. If MR is sufficient, TG and PW remain unchanged. A communication module power consumption adjustment instruction CM is generated, which includes an adjusted transmission interval field set to TGa and an adjusted transmit power level field set to PWa.

[0137] Step S650: Send the communication module power consumption adjustment command to the protocol stack control layer of the communication module to perform synchronous updates of transmission timing and power configuration.

[0138] In this embodiment, the bus power management unit sends the CM to the protocol stack control layer of the communication module via the internal bus. The protocol stack control layer parses the CM, extracts TGa and PWa, updates the transmit scheduling timer of the media access control layer and the transmit power control register of the physical layer, and completes the synchronous update of the transmission timing and power configuration.

[0139] Step S660: Collect communication link quality feedback parameters after the communication module executes the communication module power consumption adjustment command.

[0140] In this embodiment, after CM execution, the bus power management unit acquires the communication link quality feedback parameter LQ from the link quality monitoring register of the communication module. LQ includes Received Signal Strength Indicator (RSSI) in dBm and Signal-to-Noise Ratio (SNR) in dB.

[0141] Step S670: Perform reverse fine-tuning of the transmit power level based on the communication link quality feedback parameters to generate a transmit power level fine-tuning command.

[0142] In this embodiment, if the RSSI in LQ is lower than the preset lower limit of received signal strength RL or the SNR is lower than the preset lower limit of signal-to-noise ratio SL, it indicates that the excessive reduction in transmit power has led to a deterioration in the communication link quality, and the transmit power level needs to be fine-tuned in reverse. The transmit power level after reverse fine-tuning is PWb = PWa + ΔP. A transmit power level fine-tuning instruction FP is generated, where FP contains the fine-tuned transmit power level field assigned the value PWb.

[0143] Step S680: Write the transmit power level fine-tuning command into the radio frequency power control parameter table of the communication module.

[0144] In this embodiment, the bus power management unit writes FP into the radio frequency power control parameter table of the communication module, overwriting the current transmit power level value in the parameter table.

[0145] Step S690: Collect the communication bit error rate statistics after the communication module executes the transmit power level fine-tuning command, compare the communication bit error rate statistics with the upper limit of the communication bit error rate tolerance, and perform adaptive callback processing on the receiver sensitivity configuration information based on the comparison results.

[0146] In this embodiment, the communication bit error rate (BER) statistics of the communication module are collected after FP execution. The upper limit of the communication BER tolerance is denoted as BU. If the BER is greater than BU, an adaptive callback is performed on the receiver sensitivity configuration information SN. The callback method is to reduce the receiver sensitivity threshold, SNb = SN - ΔS, where ΔS is the receiver sensitivity callback step value in dB. Reducing the receiver sensitivity threshold can improve the receiver's demodulation capability for weak signals. SNb is written into the receiver sensitivity configuration register of the communication module.

[0147] For example, step S710: calculate the wake-up time interval of adjacent sensor components in the time-sharing sleep-wake-up sequence of sensor components, and obtain the wake-up time interval calculation result.

[0148] In this embodiment, the wake-up time of each sensor component is extracted from the time-sharing sleep-wake-up sequence SS generated in step S145. SS contains the wake-up times of U sensor components, arranged chronologically as WT1, WT2, ..., WTU. For the u-th and u+1-th sensor components, the wake-up time interval WUu = WTu+1 - WTu is calculated, in seconds. U-1 WUu constitute the wake-up time interval calculation result WUS.

[0149] Step S720: Analyze the transient load superposition effect at the corresponding moment in the current waveform reshaping record based on the wake-up time interval calculation results, and obtain the transient load superposition effect analysis results.

[0150] In this embodiment, the current waveform data corresponding to the wake-up time of the u-th and u+1-th sensor components are extracted from the IR generated in step S141. If WUu is less than the duration TTran of the power-on transient current of the sensor component, the power-on transient currents of the two sensor components overlap in time, resulting in a transient load superposition effect. The transient load superposition effect analysis result includes a Boolean flag indicating whether superposition exists and a predicted peak current value after superposition.

[0151] Step S730: Based on the analysis results of transient load superposition effect, rearrange and adjust the wake-up order of sensor components in the time-sharing sleep-wake-up arrangement sequence of sensor components to generate the rearranged time-sharing sleep-wake-up arrangement sequence of sensor components.

[0152] In this embodiment, if a transient load superposition effect exists, the wake-up order of the sensor components in SS is rearranged. The rearrangement rule is as follows: the wake-up time WTu+1 is postponed to WTu+TTran+TMargin, where TMargin is a preset safety protection interval in seconds. After the postponement, the wake-up time of all subsequent sensor components is correspondingly delayed. The rearranged sequence SSr is obtained.

[0153] Step S740: Resend the rearranged and adjusted sensor component time-sharing sleep-wake-up orchestration sequence to the core control unit to overwrite the original orchestration sequence.

[0154] In this embodiment, the bus power management unit sends SSr to the core control unit via the internal bus, and the core control unit uses SSr to overwrite the original orchestration sequence SS currently stored in the dynamic power management scheduler.

[0155] Step S750: Acquire the actual transient current superposition waveform generated by the sensor component when it performs a wake-up operation according to the rearranged sensor component time-sharing sleep-wake-up orchestration sequence.

[0156] In this embodiment, when the sensor component performs a wake-up operation according to SSr, the bus power supply management unit synchronously acquires the actual transient current superposition waveform IX generated on the power bus through the current sampling circuit. The acquisition parameters of IX are consistent with IW.

[0157] Step S760: Perform superposition conflict comparison processing on the actual transient current superimposed waveform and the current waveform reconstructed record, and perform secondary fine-tuning processing on the wake-up order based on the superposition conflict comparison results to generate the second fine-tuned wake-up order.

[0158] In this embodiment, the current waveforms at corresponding times in IX and IR are compared point by point. If the current amplitude at a certain sampling point in IX exceeds the preset allowable deviation ratio of the current amplitude at the corresponding sampling point in IR, it is determined that there is a superposition conflict at that time. The wake-up order is fine-tuned a second time based on the conflict time and conflict amplitude, and the fine-tuning amount is the equivalent time offset corresponding to the conflict amplitude. The wake-up order SX after the second fine-tuning is generated.

[0159] Step S770: Encode the second fine-tuned wake-up sequence into a wake-up sequence configuration frame and write it into the wake-up timing control parameter area of ​​the sensor component through the internal bus.

[0160] In this embodiment, SX is encoded into a configuration frame CFG according to the wake-up timing control interface protocol of the sensor components, and written into the wake-up timing control parameter area of ​​each sensor component through the internal bus.

[0161] Step S780: Update the predicted value of the transient current superimposed waveform corresponding to the wake-up sequence after the second fine-tuning to the load prediction data area at the corresponding moment in the current waveform reshaping record.

[0162] In this embodiment, based on the wake-up time and power-on transient current requirement IU of each sensor component in SX, the predicted transient current superposition value at each time moment is recalculated, and the calculation result is updated to the load prediction data area of ​​the corresponding time moment in IR. The updated IR will be used for waveform reshaping processing in subsequent monitoring cycles.

[0163] Step S810: During the pulse width expansion process, while expanding the transient peak current in the time dimension, the real-time terminal voltage of the expansion capacitor is simultaneously acquired.

[0164] In this embodiment, the pulse width expansion circuit includes a pulse width expansion capacitor. During the charging phase, the capacitor absorbs charge from transient peak currents, and during the discharging phase, it releases the charge with a smooth current. A voltage detection circuit is connected in parallel across the capacitor. This circuit synchronously samples the voltage across the capacitor at a sampling frequency Fv using a resistor divider network and an analog-to-digital converter. Fv is synchronized with Fs. The real-time voltage sequence of the sampled capacitor is denoted as VC, which is a one-dimensional time series. Each sampling point represents the voltage amplitude in volts (V). VC reflects the voltage change curve of the capacitor during charging and discharging.

[0165] Step S820: Perform real-time difference comparison processing between the real-time terminal voltage of the expansion capacitor and the target charging voltage of the expansion capacitor set in the pulse width expansion charging rate control command, and generate a charging difference signal for the expansion capacitor.

[0166] In this embodiment, the value of the target charging voltage field of the stretched capacitor is extracted from the pulse width stretching charging rate control command CP, denoted as VT, with units of V. During the charging phase, the difference between VC and VT is calculated at each sampling point. The formula for calculating the difference signal VD is VD = VC - VT, with units of V. A positive VD indicates that the current terminal voltage has exceeded the target voltage, while a negative VD indicates that the current terminal voltage has not yet reached the target voltage.

[0167] Step S830: Based on the difference signal of the extended capacitor charging, adjust the number of the currently selected charging current levels in the charging current level selection control command step by step to generate a dynamic fine-tuning control command for the charging current level.

[0168] In this embodiment, VD is compared with a preset difference tolerance range [VL, VH], where VL is negative and VH is positive. If VD is less than VL, it indicates that the charging progress is lagging, and the charging current needs to be increased, so the current number of selected gears NS is incremented by 1. If VD is greater than VH, it indicates that the charging progress is ahead, and the charging current needs to be decreased, so NS is decremented by 1. If VD is within the range [VL, VH], NS remains unchanged. The adjustment range of NS is limited by the total number of gears NG, and the adjusted NS value is between 1 and NG. The adjusted NS is encoded into a bitmask to generate a dynamic fine-tuning control instruction CD for the charging current gear, which contains the field of the fine-tuned number of selected gears.

[0169] Step S840: Use the dynamic fine-tuning control command of the charging current level to perform closed-loop regulation of the real-time charging current of the expansion capacitor, so that the voltage rise curve of the expansion capacitor approaches the target charging slope defined by the pulse width expansion charging rate control command.

[0170] In this embodiment, CD is sent to the charging current range selection switch array of the pulse width expansion circuit, and the number of conducting ranges is updated in real time. The real-time charging current ICH of the expansion capacitor is ICH = NS × IC, where IC is the single-range charging current. During charging, the rising slope of the voltage across the expansion capacitor is SRC = ICH / CC, where CC is the capacitance value of the expansion capacitor. By adjusting NS step by step, SRC tracks the target charging slope SRT, which is determined by the target charging voltage VT and the charging time TC, and SRT = VT / TC. Closed-loop regulation makes the rising curve of VC approximate the linear slope defined by SRT.

[0171] Step S850: When the real-time terminal voltage of the extended capacitor reaches the target charging voltage of the extended capacitor, the extended capacitor charging termination signal is triggered, and all selected charging current levels in the charging current level selection control command are canceled through the extended capacitor charging termination signal.

[0172] In this embodiment, when VC is greater than or equal to VT, the wide-band capacitor charging termination signal ST is triggered. In response to ST, NS is set to 0, a charging current level selection control command CG0 with all bitmasks zero is generated and sent to the selection switch array to turn off all charging current levels, thus ending the charging phase.

[0173] Step S860: After the charging termination signal of the extended capacitor is triggered, the discharge timing process of the extended capacitor is started, and the discharge start time of the extended capacitor is delayed and matched according to the predicted current surge duration, so that the discharge range of the extended capacitor and the predicted current surge duration are seamlessly connected on the time axis.

[0174] In this embodiment, the discharge timing process is initiated after ST is triggered. The start time TSH and end time TEH of the impact are obtained from the predicted current impact duration TT. The discharge start time TD is calculated as TD = TS - TC - TM, where TC is the charging duration and TM is the discharge advance margin. The discharge end time TE = TD + TU, where TU is the discharge duration. By adjusting TD, the discharge interval [TD, TE] and the impact interval [TSH, TEH] are seamlessly connected on the time axis, and the discharge current of the broadened capacitor begins to provide compensation precisely when the impact occurs.

[0175] Step S870: Within the discharge range of the extended capacitor, the discharge current of the extended capacitor is used to boost the current amplitude at the corresponding moment in the current waveform reconstructing record to generate the current waveform reconstructing record after the extended capacitor discharge compensation.

[0176] In this embodiment, within the discharge interval [TD, TE], the extended capacitor discharges with a constant discharge current IDC, where IDC = VC / RD, and RD is the total impedance of the discharge circuit. The current amplitude at each sampling point of IR within the discharge interval is added to IDC to obtain the current amplitude after boost compensation. The current amplitude at sampling points of IR outside the discharge interval remains unchanged. The waveform obtained after compensation is the reconstructed current waveform record IRC after the extended capacitor discharge compensation.

[0177] Step S880: The time periods that have undergone discharge compensation and the time periods that have not undergone discharge compensation in the current waveform reconstructed record after the expansion capacitor discharge compensation are spliced ​​together to obtain the continuous current waveform reconstructed record after compensation.

[0178] In this embodiment, the IRC is compensated for boosting within the discharge interval [TD, TE], but not outside the interval. At the interval boundary, linear interpolation is used to smooth the transition between adjacent sampling points, eliminating possible amplitude jumps at the interval boundary. The segments are spliced ​​together in chronological order to obtain the compensated continuous current waveform reconstructed record (IRF).

[0179] Step S890: Perform waveform residual extraction processing on the compensated continuous current waveform reshaping record and the original current waveform record to generate a waveform residual record, and feed the waveform residual record back to the overcurrent status fusion judgment processing to correct the generation logic of the bus overcurrent status level identifier.

[0180] In this embodiment, residuals are calculated point-by-point for IRF and IW under the same time base. The residual value is the current amplitude of IRF minus the current amplitude of IW. The residual values ​​of all sampling points constitute the waveform residual record RR. RR is fed back to the multi-dimensional overcurrent situation aggregation processing in step S134 as a correction factor in the calculation of the comprehensive situation assessment value SV. The corrected SV calculation formula is: SVc = SV + KR × |RR|avg, where |RR|avg is the arithmetic mean of the absolute values ​​of the residual records, and KR is the residual correction coefficient. SVc replaces SV to map the bus overcurrent situation level identifier LV, realizing feedback correction of the generation logic.

[0181] Step S910: Extract the instruction sequence to be executed by the core control unit from the list of instructions to be executed, and obtain the current pipeline fill status of the instruction pipeline in the core control unit.

[0182] In this embodiment, the first H instructions to be executed, sequenced SQ, are extracted sequentially from the task list TL, where H is the number of stages in the instruction pipeline. The current fill status PS of the instruction pipeline is read through the debug interface of the core control unit. PS is an H-bit bitmask, where each bit indicates whether the corresponding pipeline stage is occupied by a valid instruction.

[0183] Step S920: Perform instruction issue timing simulation and deduction processing on the instruction sequence to be executed and the current pipeline fill state to generate the power consumption switching time sequence of each stage of the pipeline and the corresponding equivalent computational overhead change during the execution of the instruction pipeline by the core control unit.

[0184] In this embodiment, starting with PS as the initial state, the instructions in SQ are sequentially deduced through the H-stage pipeline according to the pipelined issue rules. During the deduction process, the time point when each instruction enters each pipeline stage is recorded, as well as the power consumption increment corresponding to the activation of each pipeline stage. The power consumption increment of each pipeline stage is denoted as PW1 to PWH, in mW. When a pipeline stage pauses due to data or control dependencies, the number of pause cycles and the corresponding power consumption decrease are recorded. After the deduction is completed, the power consumption switching time point sequence TP for each stage of the pipeline and the corresponding equivalent computational overhead change sequence DP are output.

[0185] Step S930: Based on the power consumption switching time sequence of each stage of the pipeline and the corresponding equivalent computational overhead change, refine the predicted computational overhead value at the corresponding time point in the instruction computational overhead timing distribution description, and generate a pipeline-aware instruction computational overhead timing distribution description.

[0186] In this embodiment, the sampling positions corresponding to each time point of TP are located in DD, and the predicted computational overhead value at that position is replaced with the corresponding equivalent computational overhead change in DP. After replacement, the power consumption values ​​before and after that time point in DD are smoothly transitioned using linear interpolation. After all replacements are completed, a pipeline-aware instruction computational overhead timing distribution description DP is generated.

[0187] Step S940: Convert the pipeline-aware instruction computation overhead timing distribution description into the corresponding current demand timing distribution, and then perform peak cause correlation analysis with the transient current peak amplitude characteristics to identify pipeline stage switching-related current peaks caused by instruction pipeline stage switching.

[0188] In this embodiment, the power consumption values ​​at each time point in the DP are divided by VB and ET to convert them into current demand values, forming the current demand timing distribution IP. The times when the current demand rise rate in the IP exceeds ST are time-correlated and matched with the times when spikes occur in the IW. If the time deviation between the time of a spike occurrence and the time of a certain instruction pipeline stage switch in the IP is within a preset correlation window, then the spike is marked as a pipeline stage switch-related current spike caused by the instruction pipeline stage switch.

[0189] Step S950: Based on the identified pipeline stage switching associated current spike, pre-compensate the pulse width expansion charging rate control command, reduce the discharge current amplitude of the expansion capacitor in the pulse width expansion process before the pipeline stage switching associated current spike arrives, and generate a pipeline pre-compensated pulse width expansion charging rate control command.

[0190] In this embodiment, the associated current spike for each pipeline stage is switched, and its arrival time is denoted as TA. A preset lead time TD is placed before TA, and the amplitude of the broadened capacitor discharge current in the pulse width broadening charging rate control command CP is reduced from IDC to IDC×η, where η is the pre-compensation reduction coefficient, and η is less than 1. The reduced CP is the pipeline pre-compensation pulse width broadening charging rate control command CPp. CPp replaces CP for current waveform reshaping processing.

[0191] Step S960: Use the pipeline pre-compensation pulse width widening charging rate control command to replace the pulse width widening charging rate control command to perform current waveform reshaping processing on the original current waveform record.

[0192] Step S970: Collect the instruction pipeline pause cycle count during the execution of the instruction pre-compensation pulse width expansion charging rate control instruction by the core control unit, and perform offset correction on the simulation time window of the instruction launch timing simulation based on the instruction pipeline pause cycle count.

[0193] In this embodiment, the actual instruction pipeline pause cycle count SC during execution is collected. If SC is inconsistent with the pause cycle count obtained in step S920, the calculation time window is offset and corrected. The correction method is to offset the start time of the calculation time window by (SC actual - SC pushed) / FC to align the calculation with the actual execution.

[0194] Step S1010: Statistical analysis is performed on the time periods of missing sensor data corresponding to the sensor components in the time-sharing sleep-wake-up sequence of sensor components to obtain statistical information on the time periods of missing sensor data.

[0195] In this embodiment, the sleep periods of each sensor component are extracted from the SS, and each sleep period corresponds to a period of missing sensor data. The missing periods of each sensor component are summarized along the time axis to form the sensor data missing period statistics MS, which includes the start and end times of each missing period.

[0196] Step S1020: Based on the statistical information of missing sensor data periods, the upper limit of the sleep duration of each sensor component in the time-sharing sleep-wake-up sequence of sensor components is constrained and adjusted to generate the upper limit constraint condition of sleep duration.

[0197] In this embodiment, for each sensor component, the maximum allowable consecutive missing duration is calculated based on the total duration of the missing time period in its MS and the sampling frequency requirement. This maximum allowable consecutive missing duration is set as the upper limit of the sleep duration TU for that sensor component. The TUs of all sensor components constitute the upper limit constraint CU for the sleep duration.

[0198] Step S1030: Based on the upper limit constraint of sleep duration and the distribution of available sleep time windows in the current waveform reshaping record, the time-sharing sleep wake-up arrangement sequence of sensor components is rearranged to meet the sleep duration constraint, and a sleep duration constraint rearranged sequence is generated.

[0199] In this embodiment, under the premise of satisfying the CU (Cumulative Limits), the sleep time periods of each sensor component are reallocated to the available sleep time window in the IR (Infrared Registry). If the sleep time allocated to a sensor component exceeds the TU (Turnover Time), its sleep time period is split into multiple segments not exceeding the TU. The rearranged segments generate the sequence SSc.

[0200] Step S1040: In the sleep duration constraint rearrangement sequence, perform data type complementarity analysis on the sampling data types of sensor components with overlapping wake-up time windows to identify sensor component pairs with complementary sampling data types.

[0201] In this embodiment, for sensor components with overlapping wake-up time windows in the SSc, their sampling data types are analyzed. If the sampling data types of the two components are different but can complement each other after fusion, such as a temperature sensor and a humidity sensor, they are identified as a pair of sensor components with complementary sampling data types.

[0202] Step S1050: Perform wake-up time binding processing on sensor component pairs with complementary sampling data types, so that sensor component pairs with complementary sampling data types can synchronously complete sampling operations within the same wake-up time window, and generate a wake-up time bound sensor group sequence.

[0203] In this embodiment, the wake-up times of the identified complementary sensor component pairs are adjusted to be completely consistent, so that they wake up synchronously and complete sampling within the same time window. The adjusted sequence SSb is then generated.

[0204] Step S1060: Immediately after acquisition, the sampling data of each sensor component bound in the sensor group sequence at the wake-up time is fused and spliced ​​to generate a fused sensing data frame.

[0205] In this embodiment, after the bound sensor components are synchronously sampled, the bus power management unit splices the sampled data of the two into a fused sensing data frame FD, which includes a data type identifier and two data fields.

[0206] Step S1070: The independent sensor data frames of unbound sensor components in the fused sensor data frame and sleep duration constraint rearrangement sequence are time-aligned and assembled according to the timestamp to generate a missing sensor data stream.

[0207] In this embodiment, the independent data frames of the FD and unbound sensor components are assembled into a sensor data stream DF by sorting them according to their timestamps. The missing data in the DF due to sleep is partially compensated by the complementary information of the fused frames.

[0208] Step S1110: Perform sampling value change trend analysis on the continuous multiple sampling data of the sensor components in the bus-powered intelligent device to generate a sampling value change trend curve.

[0209] In this embodiment, the bus power supply management unit maintains a sampling data sliding window of length K for each sensor component, storing the most recent K sampled values. For the u-th sensor component, the data sequence in its sliding window is denoted as XU=[xu1, xu2, ..., xuK]. The sampling value change trend analysis uses the least squares linear fitting method, with the sampling sequence number as the independent variable and the sampled value as the dependent variable, fitting the linear equation x=au×t+bu. The slope au is calculated using the formula... , the intercept bu=(Σx−au×Σt) / K, where Σ represents the summation over K data points, and t is the sampling sequence number from 1 to K. au and bu constitute the sampling value variation trend curve TRu, the unit of au is the unit of sampling value per sampling period, and the unit of bu is the unit of sampling value.

[0210] Step S1120: Evaluate the predictability of the next sampling value of the sensor component according to the sampling value variation trend curve, and generate a sampling value predictability evaluation result.

[0211] In this embodiment, the fitted TRu is used to calculate the fitted value xfk of K sampling points in the sliding window, where xfk=au×k+bu, k=1, 2, ..., K. The fitted residual RMu=(1 / K)×Σ|xuk−xfk| is calculated, where RMu is the mean absolute residual, and the unit is the same as that of the sampling value. The preset residual threshold is denoted as RT. RMu is compared with RT. If RMu<RT, it indicates that the historical sampling values fluctuate slightly around the trend curve, the predictability of the next sampling value is high, and the predictability evaluation result PRu is marked as highly predictable. If RMu≥RT, it indicates that the historical sampling values fluctuate greatly, the predictability of the next sampling value is low, and PRu is marked as low predictable.

[0212] Step S1130: Perform flexible scaling processing on the sleep duration of the sensor component in the time-sharing sleep-wake scheduling sequence of the sensor component according to the sampling value predictability evaluation result, so as to obtain the time-sharing sleep-wake scheduling sequence of the sensor component after the flexible scaling of the sleep duration, wherein when the sampling value predictability evaluation result indicates that the fitted residual of the sampling value variation trend curve is less than the preset residual threshold, the sleep duration of the sensor component is extended, and when the sampling value predictability evaluation result indicates that the fitted residual of the sampling value variation trend curve is greater than the preset residual threshold, the sleep duration of the sensor component is shortened.

[0213] In this embodiment, the current sleep duration TSu of the u-th sensor component is extracted from the current sensor component time-sharing sleep-wake-up orchestration sequence SS, in seconds. Preset upper and lower limits for sleep duration scaling (TUmax and TUmin), as well as expansion and reduction coefficients (EU and DU, where EU > 1 and DU < 1). If PRu is highly predictable, the sleep duration of the sensor component is extended, with the new sleep duration TSunew = min(TSu × EU, TUmax). If PRu is low predictable, the sleep duration of the sensor component is shortened, with the new sleep duration TSunew = max(TSu × DU, TUmin). The sleep duration of the u-th sensor component in SS is updated to TSunew, while its wake-up time remains unchanged, and the sleep start time is shifted forward or backward accordingly. This elastic scaling process is performed on all U sensor components in SS to obtain the sensor component time-sharing sleep-wake-up orchestration sequence SSe with elastically scaled sleep durations.

[0214] Step S1140: Compare the actual sampled value obtained by the sensor component after the sleep duration elastic stretching process is awakened with the predicted sampled value calculated based on the sampled value change trend curve to generate the sampled value prediction deviation.

[0215] In this embodiment, for the u-th sensor component, after waking up according to the elastically stretched dormancy time TSunew, it performs its first sampling to obtain the actual sampled value xnew. The predicted sampled value at the wake-up time, xpred = au × (K + TSunew / Ts) + bu, is obtained by extrapolation using the TRu curve generated in step S1110, where Ts is the sampling period. The predicted deviation of the sampled value, PDu = |xnew - xpred|, is calculated in sampled value units.

[0216] Step S1150: When the sampled value prediction deviation exceeds the upper limit of the prediction deviation tolerance, the sensor component is marked as a sampling trend change sensor component, and the scaling step of the sleep duration elastic scaling process of the sampling trend change sensor component in the subsequent monitoring period is reduced and adjusted.

[0217] In this embodiment, a preset upper limit for prediction deviation tolerance, PU, ​​is set. If PDu > PU, the sensor component is marked as a sensor component with a sudden change in sampling trend, and the change marker MU is recorded as true. For the marked sensor component, when step S1130 is executed in a subsequent monitoring cycle, its expansion coefficient EU and reduction coefficient DU are adjusted towards 1, i.e., the scaling step amplitude is reduced. The adjustment method is EUnew = 1 + (EU - 1) × ρ, DUnew = 1 - (1 - DU) × ρ, where ρ is the amplitude reduction factor, 0 < ρ < 1. The adjusted EUnew and DUnew are stored in the sleep control parameter area of ​​the sensor component.

[0218] Step S1160: Perform a retrospective analysis on the current waveform segment in the current waveform reconstructing record corresponding to the moment of the sudden change in the sampling trend sensor component, and identify whether the sudden change in the sampling trend of the sampling trend sensor component is caused by the power bus current fluctuation.

[0219] In this embodiment, the time Tx when the sampling trend change sensor component changes is obtained, and the current waveform segment IWx within a preset analysis window before and after Tx is extracted from the current waveform reconstructed record IR. The peak-to-peak value (IPP) and standard deviation (ISTD) of the current amplitude within IWx are calculated. If the IPP exceeds the preset fluctuation amplitude threshold AP or the ISTD exceeds the preset fluctuation standard deviation threshold AS, it is determined that there is an abnormal fluctuation in the power bus current, and the sampling trend change may be caused by this current fluctuation. This analysis result is recorded as the associated cause of the sampling trend change.

[0220] Step S1210: During the process of dynamic power management scheduling by the core control unit, the queuing time of unprocessed interrupt request signals in the interrupt waiting queue of the core control unit is collected.

[0221] In this embodiment, the interrupt controller of the core control unit maintains an interrupt waiting queue, in which interrupt request signals that have not yet been processed by the core control unit are arranged according to priority. The bus power management unit reads the queuing time of each interrupt request signal in the interrupt waiting queue through the core control unit's debug interface or status register. For the r-th interrupt request signal in the interrupt waiting queue, its queuing time is denoted as WTr, which is equal to the current time minus the time when the interrupt request signal arrives at the interrupt controller, in seconds. The queuing times of all unprocessed interrupt request signals are summarized into a waiting time set WS.

[0222] Step S1220: Based on the queuing waiting time and the interrupt priority identifier of the interrupt request signal, map them to time urgency index and priority weight index respectively, and then perform a comprehensive quantitative evaluation to generate an evaluation value of the degree of emergency response requirement for interruption.

[0223] In this embodiment, for the r-th interrupt request signal, its interrupt priority identifier PRr is obtained. PRr is an integer value, and the smaller the value, the higher the priority. The queuing waiting time WTr is mapped to the time urgency index UTr, UTr = WTr / WTmax, where WTmax is the preset maximum allowable waiting time, and UTr ranges from 0 to 1. The interrupt priority identifier PRr is mapped to the priority weight index UPr, UPr = 1 / (PRr+1), and UPr ranges from 0 to 1. The comprehensive quantitative evaluation value Ur = α × UTr + β × UPr, where α and β are weighting coefficients, and α + β = 1. The maximum value of Ur among all unprocessed interrupt request signals is taken as the interrupt urgency response demand evaluation value UE.

[0224] Step S1230: Based on the emergency response demand assessment value and the remaining power supply capacity at the current moment in the current waveform reshaping record, the system clock frequency of the core control unit is instantaneously boosted. Within the preset instantaneous boost duration, the system clock frequency is boosted to the interrupt response acceleration frequency to obtain the interrupt response acceleration frequency configuration instruction.

[0225] In this embodiment, the current amplitude at the current moment is obtained from the current waveform reshaping record IR, and the current remaining power supply capacity RM is calculated. If the UE exceeds the preset emergency response threshold Uthr and RM is greater than the minimum power supply capacity required for instantaneous boost, then instantaneous boost processing of the system clock frequency is triggered. The duration of the instantaneous boost is denoted as TH, and the interrupt response acceleration frequency FH ​​= FC × (1 + UE × δ), where FC is the current system clock frequency and δ is the acceleration ratio coefficient. An interrupt response acceleration frequency configuration instruction CFi is generated, in which the target frequency field is assigned the value FH and the duration field is assigned the value TH.

[0226] Step S1240: Send the interrupt response acceleration frequency configuration command to the clock management interface of the core control unit, and restore the system clock frequency to the original frequency before the acceleration after the preset instantaneous boost duration expires.

[0227] In this embodiment, the bus power management unit sends the CFi to the clock management interface, which then switches the system clock frequency from FC to FH. Simultaneously, a duration timer is started, with a duration of TH. When the timer expires, the clock management interface automatically restores the system clock frequency from FH back to the original frequency FC before the boost.

[0228] Step S1250: During the period when the system clock frequency is at the interrupt response acceleration frequency, the current consumption increment of the power bus is tracked in real time to generate a current consumption increment curve during the interrupt response period with the same time base and current dimension as the current waveform reshaping record.

[0229] In this embodiment, during the frequency ramp-up period, the bus power management unit synchronously collects the actual current consumption IWf of the power bus at a sampling frequency Fs. The difference between IWf and the current amplitude of the corresponding time period in the current waveform reconstructed record IR is calculated point by point, with the increment value ΔIf = IWf - IR, in mA. The ΔIf of all sampling points constitutes the current consumption increment curve IC during the interrupt response period. IC and IR have the same time base and number of sampling points.

[0230] Step S1260: Overlay the current consumption increment curve during the interrupt response period with the corresponding time period in the current waveform reshaping record under the same time base, and determine whether the superimposed current amplitude triggers a new bus overcurrent status level indicator.

[0231] In this embodiment, the IC and IR are sampled point by point for the corresponding time period to obtain the superimposed current waveform IWs = IR + IC. The bus overcurrent status level indicator is recalculated for IWs according to the process of steps S131 to S134. If the newly calculated LV is higher than the original LV, a new bus overcurrent status level indicator is triggered.

[0232] Step S1270: When the superimposed current amplitude triggers a new bus overcurrent status level flag, perform emergency correction processing on the pulse width expansion charging rate control command, generate an emergency correction pulse width expansion charging rate control command, and immediately implement the emergency correction pulse width expansion charging rate control command.

[0233] In this embodiment, steps S135 and S136 are re-executed based on the newly triggered LV to generate a pulse width expansion charging rate control instruction CPe adapted to the new situation level. CPe is immediately written into the control register of the pulse width expansion circuit, overwriting the currently executing CP. The charging rate and discharge sequence of the expansion capacitor are updated according to CPe to cope with the additional current surge caused by the frequency increase.

[0234] Step S1310: Perform peak repetition period identification processing on the transient current peak amplitude characteristics and steady-state current duration characteristics. When the time interval between the continuously captured transient current peak amplitude characteristics shows a periodic repetition pattern, a periodic peak pattern identifier is generated.

[0235] In this embodiment, the transient current peak amplitude features PKz and the corresponding occurrence times Tz of the Z most recent peak events are extracted from the current feature history record queue QH, where z = 1, 2, ..., Z. The time interval ΔTz = Tz + 1 - Tz between adjacent peak events is calculated. The mean and standard deviation of ΔTz are calculated. If the ratio of the standard deviation to the mean is less than the preset periodicity judgment threshold RC, the time interval is judged to exhibit a periodic repetition pattern, and a periodic peak pattern identifier MP is generated as true. If the ratio is greater than or equal to RC, MP is false.

[0236] Step S1320: When the periodic spike pattern identifier is triggered, extract the spike repetition period value and spike amplitude fluctuation range corresponding to the periodic spike pattern, and construct a periodic spike template description based on the spike repetition period value and spike amplitude fluctuation range.

[0237] In this embodiment, when MP is true, the peak repetition period value PC is taken as the arithmetic mean of ΔTz, in seconds. The peak amplitude fluctuation range is taken as the interval between the minimum and maximum values ​​of PKz [PKmin, PKmax], in mA. A periodic peak template description MT = {period: PC, lower bound of amplitude: PKmin, upper bound of amplitude: PKmax} is constructed.

[0238] Step S1330: Use the periodic spike template description to perform component stripping on the periodic performance consumption component that matches the spike repetition period value in the instruction operation overhead timing distribution description, and separate the periodic performance consumption component from the instruction operation overhead timing distribution description to obtain the instruction operation overhead timing distribution description after stripping the periodic component.

[0239] In this embodiment, a spectral analysis is performed on the power consumption value sequence in the instruction overhead timing distribution description (DD) to detect the presence of a significant frequency component with a period of PC. If such a component exists, a band-stop filter is used to remove the frequency component corresponding to that period. The power consumption value sequence after filtering is the instruction overhead timing distribution description (DDs) after removing the periodic component.

[0240] Step S1340: The instruction operation overhead timing distribution description after stripping the periodic components is time-aligned with the periodic spike template description to generate a periodic spike cancellation timing scheduling scheme. The periodic spike cancellation timing scheduling scheme is used to instruct the pulse width broadening process to pre-start the charge broadening operation before the arrival time of the periodic spike.

[0241] In this embodiment, based on the predicted arrival time Tnext of the next periodic spike using DDs, and combined with the spike amplitude fluctuation range, the required pre-charge amount and pre-charge start time of the extended capacitor are calculated. A periodic spike cancellation timing schedule SCp is generated, which includes the pre-charge start time and the pre-charge target voltage.

[0242] Step S1350: Perform periodic synchronization correction processing on the pulse width expansion charging rate control command according to the periodic spike cancellation timing scheduling scheme, and periodically lock the charging start time of the expansion capacitor in the pulse width expansion charging rate control command with the predicted arrival time of the periodic spike to generate a periodic phase-locked pulse width expansion charging rate control command.

[0243] In this embodiment, the charging start time of the extended capacitor in CP is corrected to the pre-charging start time in SCp, and the charging target voltage is corrected to the pre-charging target voltage in SCp. The corrected command is the periodic phase-locked pulse width extension charging rate control command CPl. CPPl is phase-locked synchronized with the arrival time of the periodic spikes.

[0244] Step S1360: Use the periodic phase-locked loop pulse width expansion charging rate control command to replace the pulse width expansion charging rate control command to perform current waveform reshaping processing on the original current waveform record.

[0245] Step S1370: When the periodic spike mode indicator is not triggered again within a preset number of consecutive spike cycles, cancel the periodic spike mode indicator and resume the use of the pulse width expansion charging rate control command.

[0246] In this embodiment, a preset threshold of the number of consecutive triggers, NC, is used. If the ratio of the standard deviation to the mean of the time interval calculated in step S1310 exceeds RC within NC consecutive peak cycles, MP is canceled and set to false. The original CP is then used to replace CPl for current waveform reshaping.

[0247] Step S1410: Perform waveform frequency domain transformation on the current waveform reconstructed record to convert the current waveform reconstructed record from time domain representation to frequency domain representation, and obtain the current waveform frequency domain representation.

[0248] In this embodiment, a Fast Fourier Transform is performed on the current waveform reconstructed record IR to obtain the frequency domain representation FR of the current waveform. FR is a complex sequence, with each frequency point corresponding to amplitude and phase.

[0249] Step S1420: Perform frequency band energy concentration analysis on the frequency domain representation of the current waveform to identify the dominant frequency band intervals in the frequency domain representation of the current waveform where the energy concentration exceeds the frequency band energy concentration threshold.

[0250] In this embodiment, the spectrum range is divided into several frequency band intervals. For each frequency band interval, the proportion of the sum of squared amplitudes of the frequency points within the interval to the total sum of squared amplitudes is calculated as the energy concentration. Frequency band intervals with an energy concentration exceeding a preset threshold EC are marked as dominant frequency band intervals FB.

[0251] Step S1430: Based on the dominant frequency band range, perform frequency band matching adjustment processing on the sampling frequency of each sensor component in the time-division sleep-wake-up arrangement sequence of sensor components. Compress the sleep duration of sensor components whose sampling frequency is within the dominant frequency band range, and extend the sleep duration of sensor components whose sampling frequency is outside the dominant frequency band range to generate a frequency band matching sleep duration adjustment sequence.

[0252] In this embodiment, the sampling frequency of each sensor component is obtained, and it is checked whether it falls within the FB (Frequency Band Matching) range. If so, its sleep duration is shortened by multiplying it by the compression factor CF; otherwise, its sleep duration is lengthened by multiplying it by the expansion factor EF. A frequency band matching sleep duration adjustment sequence Sf is generated.

[0253] Step S1440: The wake-up times of sensor components whose sleep duration is compressed in the frequency band matching sleep duration adjustment sequence are rearranged so that the wake-up times avoid the current peak times corresponding to the dominant frequency band interval on the time axis, thereby generating a frequency band peak avoidance wake-up arrangement sequence.

[0254] Step S1450: Perform difference merging processing on the frequency band peak avoidance wake-up arrangement sequence and the sensor component time-division sleep wake-up arrangement sequence to generate a fused frequency band peak avoidance time-division sleep wake-up arrangement sequence.

[0255] Step S1510: Perform data change rate analysis on the sampled data stream of the sensor components in the bus-powered intelligent device to generate data change rate curves for each sensor component.

[0256] In this embodiment, the absolute value of the difference between adjacent sampled values ​​is calculated for the sampled data sequence of each sensor component and divided by the sampling period to obtain the data change rate sequence, and the data change rate curve VR is generated.

[0257] Step S1520: Based on the data change rate curve, classify the sampling urgency of the sensor components, and classify the sensor components whose data change rate exceeds the change rate threshold into the first sampling urgency group, and those that do not exceed the threshold into the second sampling urgency group.

[0258] Step S1530: Reduce the sleep time of the sensor components in the first sampling urgency group and expand the sleep time of the sensor components in the second sampling urgency group to generate a hierarchical sleep time allocation scheme based on urgency level.

[0259] Step S1540: Match the urgency-based hibernation duration allocation scheme with the available hibernation time window in the current waveform reshaping record. When the available hibernation time window cannot accommodate the reduced hibernation duration of the first sampling urgency group, postpone the wake-up time of the sensor components of the second sampling urgency group to release the remaining hibernation time window for use by the first sampling urgency group.

[0260] Step S1550: Update the time-sharing sleep-wake-up orchestration sequence of sensor components according to the sleep availability time window reallocation scheme, and generate the time-sharing sleep-wake-up orchestration sequence of sensor components for urgent sensing.

[0261] Step S1610: Perform a backtracking analysis of the historical generation records of the bus overcurrent status level identifier to extract the current characteristic evolution trajectory when the bus overcurrent status level identifier jumps from a low level to a high level.

[0262] In this embodiment, the PK and TS sequences before and after the LV level transition are extracted from the regulation history storage area, and the evolution trajectory EV of PK and TS over time is constructed.

[0263] Step S1620: Construct an overcurrent situation escalation early warning feature template based on the current characteristic evolution trajectory. The overcurrent situation escalation early warning feature template includes a standardized index of the slope of the transient current peak amplitude characteristic change and a standardized index of the compression rate of the steady-state current duration characteristic, which are precursors to the situation escalation.

[0264] In this embodiment, the rising slope SRp of PK and the compression rate CRt of TS before the transition are extracted from EV. After normalizing SRp and CRt, a warning feature template TMet is constructed.

[0265] Step S1630: The slope of the change of transient current peak amplitude and the compression rate of steady-state current duration characteristics captured in real time are compared with the standardized indicators in the template. When both fall into the warning trigger range, an overcurrent situation upgrade warning signal is generated.

[0266] Step S1640: In response to the overcurrent status upgrade warning signal, before the bus overcurrent status level indicator is actually upgraded, the pulse width expansion capacitor is pre-charged during the pulse width expansion process. The terminal voltage of the expansion capacitor is pre-increased to a preset pre-charge ratio of the target charging voltage to obtain the pre-charged expansion capacitor voltage.

[0267] Step S1650: Write the amount of precharge charge corresponding to the precharge widening capacitor voltage into the precharge status register of the pulse width widening process.

[0268] Step S1710: Classify the sampling data output interface types of sensor components with adjacent wake-up time windows in the time-sharing sleep-wake-up orchestration sequence of sensor components, and identify the set of sensor components that share the same data bus.

[0269] In this embodiment, for sensor components with adjacent wake-up time windows in SS, their sampling data output interface types are checked. Sensor components with the same interface type and connected to the same data bus are grouped into a sensor component set SB that shares the same data bus.

[0270] Step S1720: Perform bus occupancy time arrangement processing on the sampling data output time of each sensor component in the sensor component set that shares the same data bus, and arrange the sampling data output time of each sensor component serially on the time axis to generate a bus time-sharing occupancy arrangement sequence.

[0271] Step S1730: Associate and bind the sampling data output time of each sensor component in the time-division occupancy arrangement sequence with the wake-up time of each sensor component in the time-division sleep wake-up arrangement sequence, so that the wake-up time of the same sensor component and the sampling data output time maintain a preset output delay interval.

[0272] Step S1740: Fine-tune the sleep duration of each sensor component in the time-sharing sleep-wake-up arrangement sequence of sensor components according to the preset output delay interval, so that the sampling data output time of adjacent sensor components in the sensor component set sharing the same data bus does not cause bus conflict, and obtain a time-sharing sleep-wake-up arrangement sequence of sensor components without bus conflict.

[0273] Step S1750: Synchronously send the time-sharing sleep-wake-up sequence of the sensor components without bus conflicts and the time-sharing occupancy sequence of the bus to the core control unit to coordinate the sensor wake-up timing and the data bus occupancy timing.

[0274] In an exemplary embodiment, an adaptive power management system for a bus-powered intelligent device is provided. This system can be a terminal, server, etc., and its internal structure includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements an adaptive power management method for a bus-powered intelligent device. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the casing of the adaptive power management system of a bus-powered intelligent device, or an external keyboard, touchpad, or mouse, etc.

[0275] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. An adaptive power management method for a bus-powered intelligent device, the method comprising: The method includes: Obtain the original current waveform record conducted by the power bus, and obtain the list of instructions to be executed from the core control unit in the bus-powered intelligent device; The original current waveform record is subjected to waveform feature analysis processing to obtain the transient current peak amplitude characteristics and steady-state current duration characteristics of the original current waveform record. The instruction operation overhead of the task list to be executed is estimated to obtain the instruction operation overhead timing distribution description of the core control unit. The transient current spike amplitude characteristics and the steady-state current duration characteristics are combined with the instruction operation overhead timing distribution description to perform overcurrent situation fusion judgment processing, generate bus overcurrent situation level identifier, and perform charging rate regulation processing on the pulse width expansion process according to the bus overcurrent situation level identifier to obtain pulse width expansion charging rate regulation instruction. The transient current peak amplitude characteristics and the pulse width expansion charging rate control command are used to reshape the original current waveform record to generate a reshaped current waveform record. Based on the reshaped current waveform record, the sensor components in the bus-powered intelligent device are subjected to time-division sleep-wake-up orchestration to obtain the time-division sleep-wake-up orchestration sequence of the sensor components. The time-sharing sleep-wake-up sequence of the sensor components and the current waveform reshaping record are synchronously sent to the core control unit to perform dynamic power management scheduling.

2. The adaptive power management method of bus-powered intelligent device according to claim 1, wherein, The original current waveform record is subjected to waveform feature analysis processing to obtain the transient current peak amplitude characteristics and steady-state current duration characteristics. The list of tasks to be executed is then processed to estimate instruction computation overhead, resulting in a time-series distribution description of the instruction computation overhead of the core control unit, including: The original current waveform record is subjected to current slope timing scan processing. When the current amplitude jump rate of adjacent sampling points in the original current waveform record exceeds the preset slope trigger threshold, a spike capture enable flag is generated. In response to the spike capture enable flag, peak lock tracking processing is performed on the current amplitude jump section in the original current waveform record, and the maximum current amplitude sampled in the current amplitude jump section is used as the transient current spike amplitude feature. When the current amplitude rise rate of adjacent sampling points in the original current waveform record falls below the preset slope trigger threshold, the number of consecutive sampling points in the original current waveform record that remain within the rated current amplitude range is accumulated and the accumulated timing result is used as a feature of the steady-state current duration. The list of tasks to be executed is decomposed into instruction categories to obtain the computational complexity category identifier and the number of instruction execution cycles corresponding to each instruction. Based on the computational complexity category identifier and the instruction execution cycle number, construct a temporal distribution description of the instruction computational overhead of the core control unit during the execution period of the instruction task list to be executed; The transient current peak amplitude feature and the steady-state current duration feature are recorded as independent feature parameters, and the two are associated with the instruction operation overhead timing distribution description and stored in the current feature history queue. 3.The adaptive power management method of bus-powered intelligent devices according to claim 1, wherein, The process of fusing the transient current spike amplitude characteristics and the steady-state current duration characteristics with the instruction computation overhead timing distribution description to determine the overcurrent situation generates a bus overcurrent situation level identifier. Based on the bus overcurrent situation level identifier, the pulse width expansion process is then subjected to charging rate regulation to obtain a pulse width expansion charging rate regulation command, including: The transient current spike amplitude characteristics and the steady-state current duration characteristics are standardized to obtain standardized transient current spike indices and steady-state current duration indices. The standardized transient current spike index and steady-state current duration index are used as inputs to perform current surge intensity assessment and generate a current surge intensity index. Extract the peak amplitude and duration of the computational overhead in the future time period from the time sequence distribution description of the instruction computational overhead. Perform power supply current demand conversion on the peak amplitude of the computational overhead to obtain the predicted peak amplitude of the current surge. Perform time axis mapping on the duration of the peak computational overhead to obtain the predicted duration of the current surge. The current surge intensity index, the predicted current surge peak amplitude, and the predicted current surge duration are mapped to a unified situation assessment space, and multi-dimensional overcurrent situation aggregation processing is performed to generate a bus overcurrent situation level identifier. Based on the bus overcurrent status level identifier, the number of gating levels of the pulse width expansion capacitor charging current level in the pulse width expansion process is controlled, and a charging current level gating control command is generated. The charging rate of the stretching capacitor during the pulse width stretching process is adjusted by the charging current level selection control command, and the starting time of the discharge period of the stretching capacitor during the pulse width stretching process is delayed and matched according to the predicted current surge duration, so as to generate a pulse width stretching charging rate control command. The bus overcurrent status level identifier is associated with the pulse width expansion charging rate control command to generate an overcurrent status control associated record entry, and the overcurrent status control associated record entry is stored in the control history storage area.

4. The adaptive power management method of bus-powered intelligent devices according to claim 1, wherein, The process involves reshaping the original current waveform record using the transient current spike amplitude characteristics and the pulse width broadening charging rate control command to generate a reshaped current waveform record. Based on this reshaped record, a time-division sleep / wake-up orchestration process is then performed on the sensor components in the bus-powered intelligent device to obtain a time-division sleep / wake-up orchestration sequence for the sensor components, including: According to the pulse width expansion charging rate control command, the transient peak current in the original current waveform record is controlled to transfer charge to the expansion capacitor in the pulse width expansion process, and the transient peak current is expanded into a smooth supplementary current waveform in the time dimension to obtain the current waveform reconstructed record. The current waveform reshaping record is processed to identify the trough interval, and the available sleep time window in the current waveform reshaping record where the current amplitude is lower than the sensor sleep current threshold is identified. Based on the start time and duration of the available sleep time window, the sleep time allocation process is performed on the sensor components in the bus-powered smart device to obtain the sleep time allocation plan for each sensor component. The power-on transient current requirements of each sensor component in the bus-powered intelligent device are sorted to obtain a sequence of power-on transient current requirements of the sensor components. Based on the transient current demand sequence of the sensor component upon power-on and the remaining power supply capacity at the peak moment in the current waveform reshaping record, the wake-up time staggered arrangement process is performed to obtain the time-sharing sleep wake-up arrangement sequence of the sensor component. The time-sharing sleep-wake-up orchestration sequence of the sensor component is split into a sensor sleep trigger signal sequence and a sensor wake-up trigger signal sequence, and the sensor sleep trigger signal sequence and the sensor wake-up trigger signal sequence are written into the sensor sleep scheduling queue and the sensor wake-up scheduling queue, respectively.

5. The adaptive power management method of bus-powered intelligent device according to claim 1, wherein, The method further includes: During the execution of the task list of instructions to be executed by the core control unit, the actual current consumption waveform of the core control unit is collected and recorded simultaneously. The actual current consumption waveform record is converted into an equivalent computational overhead timing distribution corresponding to the actual instruction execution amount; The equivalent computation overhead timing distribution and the instruction computation overhead timing distribution are subjected to deviation trend analysis to generate a computation overhead deviation trend curve. The computational overhead deviation trend curve is subjected to deviation trend identification processing to generate a deviation trend category identifier; Based on the deviation trend category identifier, the system clock frequency of the core control unit is adjusted by frequency ramping or ramping to obtain a system clock frequency adjustment command; The system clock frequency adjustment command is sent to the clock management interface of the core control unit to adjust the operation and processing rhythm of the core control unit; Obtain the control response feedback signal of the core control unit after executing the system clock frequency control command; The frequency modulation step amplitude of the system clock frequency control command is corrected based on the control response feedback signal to generate a corrected system clock frequency control command. The corrected system clock frequency adjustment command overrides the currently executed frequency adjustment parameter configuration in the clock management interface; The actual instruction throughput of the core control unit after the execution of the corrected system clock frequency adjustment command is collected. The throughput deviation between the actual instruction throughput and the expected instruction throughput in the instruction operation overhead timing distribution description is calculated. The prediction parameters of the instruction operation overhead prediction processing are calibrated based on the throughput deviation calculation results.

6. The adaptive power management method of bus-powered intelligent devices according to claim 1, wherein, The method further includes: Acquire the surge current spike signal generated by the load switching component in the bus-powered intelligent device at the moment of load switching; The surge current spike signal is pulse shaped to generate a surge pulse trigger signal; The surge pulse trigger signal is used to trigger a fast current discharge process, and the transient energy in the surge current spike signal is bypassed to the energy storage capacitor bank through the fast current discharge process to obtain the residual current waveform after the surge energy is transferred. The residual current waveform is subjected to secondary clamping processing to limit the voltage amplitude of the residual current waveform within the bus voltage safety threshold, thereby obtaining a safety clamping voltage signal. The safety clamp voltage signal is fed back to the input of the power bus to maintain the stability of the bus voltage; The change in the terminal voltage of the energy storage capacitor bank after absorbing the transient energy is collected; The duration of the rapid current discharge process is adaptively adjusted based on the change in the terminal voltage of the energy storage capacitor bank, and a discharge duration adjustment command is generated. The discharge duration adjustment command is written into the discharge control parameter table of the fast current discharge process.

7. The adaptive power management method for a bus-powered intelligent device according to claim 2, characterized in that, The method further includes: Collect a set of historical current waveform records of the power bus in the bus-powered intelligent device during multiple consecutive monitoring cycles; The historical current waveform record set is subjected to periodic energy distribution statistical processing to generate energy consumption distribution patterns corresponding to each monitoring period; Model shift analysis was performed on the energy consumption distribution patterns of adjacent monitoring periods to obtain the energy consumption pattern shift trend. The preset slope trigger threshold is dynamically adjusted based on the energy consumption mode offset trend to generate an adjusted preset slope trigger threshold. The adjusted preset slope trigger threshold is overridden by the preset slope trigger threshold currently used in the current slope timing scan processing; The historical current waveform record set is subjected to periodic peak density statistical processing to generate the peak occurrence density distribution corresponding to each monitoring period; Based on the peak occurrence density distribution, the minimum peak recognition width in the pulse width limiting process is dynamically adjusted to generate the adjusted minimum peak recognition width. The adjusted minimum spike recognition width is written into the spike recognition width parameter table used in the pulse width limiting process.

8. The adaptive power management method of bus-powered intelligent devices according to claim 1, wherein, The method further includes: Monitor the real-time voltage amplitude of the power bus in the bus-powered intelligent device. When the real-time voltage amplitude drops below the voltage drop warning threshold, trigger the backup energy storage access control signal. In response to the backup energy storage access control signal, the backup energy storage path is turned on, and the stored charge in the backup energy storage module is injected into the power bus through the backup energy storage path to obtain the power bus voltage after voltage compensation. When the voltage of the power bus after voltage compensation recovers to above the voltage drop warning threshold and maintains a preset recovery stabilization time, the backup energy storage path is disconnected to cut off charge injection. Monitor the remaining charge of the backup energy storage module during the charge injection process. When the remaining charge of the energy storage module is lower than the energy storage module's recharge threshold, extract the available sleep time window from the current waveform reshaping record. During the dormancy availability window, a supplementary charging enable signal is sent to the backup energy storage module to perform charge recharging on the backup energy storage module. When the voltage at the terminal of the backup energy storage module reaches the full charge voltage threshold of the energy storage module, the supplementary charging enable signal is withdrawn to terminate the charge recharge. The start and end times of charge recharging of the backup energy storage module are written into the backup energy storage maintenance log.

9. The adaptive power management method of bus-powered intelligent device according to claim 1, wherein, The method further includes: Obtain the transmit power configuration information and receive sensitivity configuration information of the communication module in the bus-powered intelligent device during the data transmission and reception period; Based on the transmit power configuration information and the receive sensitivity configuration information, the power consumption of the radio frequency front-end of the communication module is estimated to obtain the estimated power consumption of the radio frequency front-end during the communication period. The estimated power consumption of the radio frequency front-end during the communication period is matched with the remaining power supply capacity of the corresponding period in the current waveform reshaping record to obtain the power supply matching result for the communication period. Based on the power supply matching result of the communication period, the data packet transmission interval and transmission power level of the communication module are jointly adjusted to generate a power consumption adjustment command for the communication module. The power consumption adjustment command of the communication module is sent to the protocol stack control layer of the communication module to perform synchronous updates of transmission timing and power configuration; Collect communication link quality feedback parameters after the communication module executes the power consumption adjustment command of the communication module; Based on the communication link quality feedback parameters, the transmit power level is fine-tuned in reverse to generate a transmit power level fine-tuning command. Write the transmit power level fine-tuning instruction into the radio frequency power control parameter table of the communication module; The communication error rate statistics of the communication module are collected after the fine-tuning command of the transmission power level is executed. The communication error rate statistics are compared with the upper limit of the communication error rate tolerance. Based on the comparison result, the receiving sensitivity configuration information is adaptively callback processed.

10. An adaptive power management system for a bus-powered intelligent device, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the adaptive power management method for a bus-powered smart device according to any one of claims 1 to 9 by executing the machine-executable instructions.