Inertial navigation low-power consumption power supply optimization method and system

Abstract

The application relates to the technical field of power supply of inertial navigation, and discloses an inertial navigation low-power-supply optimization method and system. The method comprises the following steps: collecting original sampling data of a gyroscope and an accelerometer of an inertial navigation system, extracting multi-scale motion features, combining task priorities to judge a carrier motion state, and constructing a power grading configuration table; power is dispatched on a time-sharing and on-demand basis, a power state ring buffer queue is constructed, energy budget balancing is performed in combination with a battery state of charge, a forward-looking power scheduling plan is generated, a voltage regulation control and energy recovery topology graph is constructed, power supply is executed cycle by cycle, and closed-loop correction is realized through navigation accuracy verification. The application reduces power jump loss through multi-dimensional motion state judgment, realizes energy recovery among devices, and reduces the overall power consumption of the system while ensuring navigation accuracy.

Description

Technical Field

[0001] This invention relates to the field of inertial navigation power supply technology, and in particular to an optimized method and system for low-power power supply in inertial navigation. Background Technology

[0002] In integrated navigation systems, inertial navigation devices (INS) serve as the core navigation units, working in conjunction with auxiliary navigation signals such as satellites and odometry to achieve high-precision positioning and attitude calculation. Existing INS power supply optimization schemes often use the macroscopic motion state of the carrier as the basis for power adjustment, dividing different power operating modes to regulate the power supply to the INS. The core objective is to reduce the overall system energy consumption while meeting navigation accuracy requirements. However, the adjustment logic of these schemes relies solely on motion state thresholds and does not incorporate parameters such as the status of auxiliary navigation signals and the real-time error status of the INS.

[0003] During urban commuting operations, engineers discovered that when a vehicle's in-vehicle navigation system briefly stopped at an intersection with satellite signal, the system failed to adjust its power based on the signal status. Instead, when the system briefly lost satellite signal upon entering an underground parking garage, significant navigation drift occurred, and recalibration took a considerable amount of time after signal recovery. In one instance, the navigation positioning error exceeded the system's tolerance range after only tens of seconds in an underground parking garage without auxiliary signals, causing navigational deviations and forcing the driver to stop and recalibrate. On-site debugging and analysis revealed that the root cause of the problem was that the existing power supply scheme only differentiated power modes based on the vehicle's macroscopic motion, failing to adjust power according to the real-time availability of auxiliary navigation signals and the real-time error status of the inertial navigation system. When auxiliary signals were available, the system failed to reduce power, resulting in unnecessary energy redundancy. Furthermore, when auxiliary signals were temporarily interrupted, the zero-bias instability of the inertial devices was not properly suppressed, leading to rapid error accumulation and a decrease in positioning accuracy. These types of problems shorten the equipment's endurance in the field, and navigation deviations during periods without auxiliary signals cannot be corrected in time, requiring a lot of time for post-calibration, which seriously affects the operational efficiency and reliability of the navigation system. Summary of the Invention

[0004] This invention proposes a low-power power supply optimization method for inertial navigation, comprising:

[0005] S1: Collect raw angular velocity and raw acceleration sampling data from the gyroscope and accelerometer in the inertial navigation system, extract multi-scale motion features and determine the state of the raw angular velocity and raw acceleration sampling data, obtain the carrier motion state classification label, and construct a power classification configuration table based on the carrier motion state classification label;

[0006] Further, step S1 includes:

[0007] S11: Using the gyroscope and accelerometer in the inertial navigation system, collect the three-axis raw angular velocity sampling data and three-axis raw acceleration sampling data of the carrier in real time at the system's rated sampling frequency; at the same time, obtain the task priority parameters in the current navigation task command;

[0008] S12: Perform multi-scale motion feature extraction on the original angular velocity sampling data and the original acceleration sampling data to obtain a set of carrier motion feature parameters;

[0009] Further, step S12 includes:

[0010] S121: Calculate the root mean square value of angular velocity and the peak value of angular velocity change rate within a preset sliding time window for the raw angular velocity sampling data of the three axes; calculate the root mean square value of acceleration and the peak value of acceleration change rate within the same sliding time window for the raw acceleration sampling data of the three axes.

[0011] S122: Calculate the short-time energy spectral density for the original angular velocity sampling data and the original acceleration sampling data within the sliding time window, and extract the angular velocity energy concentration; the angular velocity energy concentration is the ratio of the energy of the angular velocity signal in the preset frequency band to the total frequency band energy;

[0012] S123: Combine the root mean square value of angular velocity, peak value of angular velocity change rate, root mean square value of acceleration, peak value of acceleration change rate, and angular velocity energy concentration into a set of carrier motion characteristic parameters;

[0013] S13: Based on the set of carrier motion feature parameters and combined with the task priority parameter, motion state is determined and a carrier motion state hierarchical label is generated;

[0014] Further, step S13 includes:

[0015] S131: The core working mode determination condition is set as follows: the root mean square value of angular velocity is greater than or equal to the first angular velocity threshold, or the root mean square value of acceleration is greater than or equal to the first acceleration threshold, or the task priority parameter is marked as a precise positioning task; the standby mode determination condition is set as follows: the root mean square value of angular velocity is less than the second angular velocity threshold, and the root mean square value of acceleration is less than the second acceleration threshold, and the peak value of the rate of change of angular velocity is less than the first rate of change of angular velocity threshold, and the peak value of the rate of change of acceleration is less than the first rate of change of acceleration threshold; wherein the first angular velocity threshold is greater than the second angular velocity threshold, and the first acceleration threshold is greater than the second acceleration threshold;

[0016] S132: The state that does not meet the core working mode determination condition and does not meet the standby mode determination condition is determined as the normal working mode; the state that meets the core working mode determination condition is marked as the core working mode label, the state that meets the standby mode determination condition is marked as the standby mode label, and the normal working mode is marked as the normal working mode label; the core working mode label, the normal working mode label and the standby mode label are collectively referred to as the carrier motion state classification label.

[0017] S133: Introduce a state switching hysteresis protection mechanism. Specifically, when a parameter in the set of carrier motion characteristic parameters crosses the boundary from one mode determination interval to an adjacent mode determination interval, the carrier motion state classification label will only be updated after the parameter has been continuously kept in the boundary interval for a duration exceeding the preset hysteresis confirmation duration, so as to avoid frequent switching at the state boundary.

[0018] S14: Construct a power classification configuration table based on the carrier motion state classification label;

[0019] Further, step S14 includes:

[0020] S141: Establish a component list for each core component in the inertial navigation system, including a gyroscope, an accelerometer, and a navigation computer; for each core component in the component list, set its rated power supply value under the core working mode label, its downgraded power supply value under the normal working mode label, and its minimum standby power supply value under the standby mode label.

[0021] S142: For each core device in the device list, set its rated sampling frequency under the core working mode label, the reduced sampling frequency under the normal working mode label, and the intermittent sampling frequency under the standby mode label; the intermittent sampling frequency corresponds to the intermittent power supply cycle, which includes a power supply activation period and a power supply sleep period.

[0022] S143: Arrange the power supply power value and sampling frequency of each core device under each carrier motion state classification label by row as core device and column as carrier motion state classification label to construct a power classification configuration table; each item in the power classification configuration table includes device identifier, corresponding carrier motion state classification label, power supply power value, sampling frequency and intermittent power supply cycle parameter.

[0023] S2: Based on the power level configuration table, perform time-sharing on-demand power supply scheduling for each core component of the inertial navigation system, generate a real-time power command sequence for each core component, and construct a circular buffer queue of device power status based on the real-time power command sequence;

[0024] Further, step S2 includes:

[0025] S21: Read the power grading configuration table, and according to the current carrier motion state grading label, query the power supply power value and sampling frequency of each core device that matches the current carrier motion state grading label in the power grading configuration table, and generate the power allocation instruction group for the current period.

[0026] S22: Execute time-sharing on-demand control logic for power supply scheduling of each core device in the power allocation instruction group to generate a real-time power instruction sequence;

[0027] Further, step S22 includes:

[0028] S221: For the gyroscope, when the carrier motion state classification label is the standby mode label, according to the intermittent power supply cycle parameters in the power classification configuration table, the power supply command corresponding to the reduced power supply value is output to the gyroscope during the power supply activation period, and the zero power command is output to the gyroscope during the power supply sleep period; when the carrier motion state classification label is the normal working mode label, if the gyroscope type is a MEMS gyroscope, the corresponding reduced power supply value command is output according to the reduced sampling frequency; if the gyroscope type is a fiber optic gyroscope or a laser gyroscope, the power supply of the redundant detection channels is turned off, only the power supply of the core measurement channel is retained, and the power supply value command corresponding to the core measurement channel is output.

[0029] S222: For the accelerometer, when the carrier motion status classification label is the standby mode label, the same intermittent power supply control logic as the gyroscope is executed according to the intermittent power supply cycle parameters; when the carrier motion status classification label is the normal working mode label, the power supply power value instruction is output according to the reduced sampling frequency.

[0030] S223: For the navigation computer, dynamic frequency modulation power supply control is adopted. Specifically, the current computing load rate of the navigation computer's processor is obtained in real time. When the carrier motion status classification label is the standby mode label and the computing load rate is lower than the preset low load threshold, the processor clock frequency is reduced to the preset minimum operating frequency, and the minimum holding power value command corresponding to the minimum operating frequency is output. When the carrier motion status classification label is the normal operating mode label, the target operating frequency is calculated by linear interpolation between the minimum operating frequency and the rated operating frequency based on the computing load rate, and the power supply value command corresponding to the target operating frequency is output.

[0031] S224: Arrange the power supply value instructions generated by each core device in each control cycle according to the timestamp to form a real-time power instruction sequence; each instruction in the real-time power instruction sequence includes a timestamp, device identifier, target power supply value, target sampling frequency and channel control flag;

[0032] S23: Construct a circular buffer queue for device power status based on real-time power command sequence;

[0033] Further, step S23 includes:

[0034] S231: In the memory of the power supply optimization control system, a circular buffer space of fixed length L is allocated, and the write pointer and read pointer are initialized to point to the first position of the circular buffer space; the fixed length L is the preset length of the historical power status recording window;

[0035] S232: After generating a real-time power command sequence in each control cycle, the current target power supply value of each core device in the real-time power command sequence, the current carrier motion state classification label and the current timestamp are encapsulated into a power state record node and written into the position pointed to by the write pointer in the circular buffer space. The write pointer moves one position to the right. When the write pointer reaches the end of the circular buffer space, it wraps back to the first position to cover the earliest power state record node.

[0036] S233: The device power status circular cache queue is the circular cache space in which power status record nodes are written; the device power status circular cache queue stores the power status record nodes of the most recent L control cycles, and each power status record node contains the target power supply value of all core devices, the carrier motion status classification label and the timestamp; the device power status circular cache queue supports traversal reading by timestamp order and query by device identifier index;

[0037] S3: Read the device power status circular buffer queue, extract the historical power change trend of each core device, combine it with the state of charge data of the power supply battery, generate a forward-looking power scheduling plan through the energy budget balancing algorithm, and construct a voltage regulation control and energy recovery instruction topology based on the forward-looking power scheduling plan;

[0038] Further, step S3 includes:

[0039] S31: Read all power status record nodes of the most recent L control cycles from the device power status circular buffer queue in timestamp order, and extract the historical power change trend of each core device.

[0040] Further, step S31 includes:

[0041] S311: For each power status record node stored in the device power status circular buffer queue, group them according to device identifier to obtain the time series of the target power supply value of each core device in the most recent L control cycles;

[0042] S312: Perform linear regression fitting on the time series of the target power supply value of each core device to calculate the power change slope and the standard deviation of power fluctuation; use the power change slope and the standard deviation of power fluctuation as parameters describing the historical power change trend of the core device.

[0043] S313: Calculate the overall power consumption rate of the system based on the historical power change trend description parameters of each core component. The overall power consumption rate of the system is the sum of the power change slopes of all core components.

[0044] S32: Collect the state of charge data of the power supply battery, combine it with the overall power consumption rate of the system, and generate a forward-looking power scheduling plan through the energy budget balancing algorithm;

[0045] Further, step S32 includes:

[0046] S321: Collects the current state of charge percentage, current output voltage, and current discharge current of the power supply battery, calculates the remaining available energy of the power supply battery, and estimates the expected remaining driving time under the current power consumption mode based on the overall power consumption rate of the system and the remaining available energy.

[0047] S322: Compare the estimated remaining battery life with the preset remaining task duration; if the estimated remaining battery life is greater than or equal to the preset safety multiple of the remaining task duration, maintain the power supply value in the current power grading configuration table unchanged; if the estimated remaining battery life is less than the preset safety multiple of the remaining task duration, start the energy budget balancing algorithm.

[0048] S323: The energy budget balancing algorithm includes the following steps: calculating the energy gap value, where the energy gap value is the product of the remaining task duration and the current total system power minus the remaining available energy; based on the energy gap value and the power fluctuation standard deviation in the historical power change trend description parameters of each core device, prioritizing power reduction allocation to core devices with larger power fluctuation standard deviations in descending order; for each core device to be reduced, calculating its power reduction amount, where the power reduction amount does not exceed the difference between the current target power supply value and the minimum maintained power supply value of the core device; and accumulating the power reduction amounts of each core device until the total power reduction can compensate for the energy gap value.

[0049] S324: The target power supply value of each core device after adjustment by the energy budget equalization algorithm, the adjustment effective time interval, and the power transition slope limit are encapsulated into a forward-looking power scheduling plan; the forward-looking power scheduling plan includes the planned power supply value sequence of each core device in several future control cycles and the corresponding time interval.

[0050] S33: Based on a forward-looking power scheduling plan, construct a topology diagram for voltage regulation control and energy recovery commands;

[0051] Further, step S33 includes:

[0052] S331: Based on the planned power supply value sequence of each core device in the forward-looking power scheduling plan, calculate the target output voltage value of each core device in each control cycle; the target output voltage value is determined by looking up the power-voltage mapping table of the device; for each core device, the difference between the target output voltage values ​​of adjacent control cycles is recorded as the voltage jump variable;

[0053] S332: Construct a set of nodes for a voltage regulation control and energy recovery command topology graph; the voltage regulation control and energy recovery command topology graph is a directed acyclic graph; each node in the node set represents a power supply operation for a core device within a control cycle; each node contains the following attributes: device identifier, control cycle number, target output voltage value, voltage jump variable, and device type identifier;

[0054] S333: Construct a set of directed edges for the voltage regulation control and energy recovery command topology graph; the directed edges represent the execution dependencies and energy transfer relationships between nodes; specifically, within the same control cycle, if the voltage jump variable of one core device is negative (i.e., a buck operation is executed), and the voltage jump variable of another core device is positive (i.e., a boost operation is executed), then an energy recovery directed edge is added from the buck operation node to the boost operation node. The attribute of this energy recovery directed edge includes an estimated value of recoverable energy; the estimated value of recoverable energy is calculated based on the inductor energy storage and capacitor energy storage of the corresponding device of the buck operation node; between adjacent control cycles, a timing-dependent directed edge is added from the preceding control cycle node to the following control cycle node of the same core device, indicating the timing execution order of the power supply operation of that device;

[0055] S334: Perform topology sorting on the voltage regulation control and energy recovery command topology graph to determine the execution priority sequence of all nodes; within each control cycle, prioritize the execution of the buck operation node with a negative voltage jump variable, so that the recoverable energy released by it can be transferred to the node that needs to perform the boost operation through the energy recovery directed edge, thereby reducing the energy directly drawn from the power supply battery;

[0056] S4: Read the voltage regulation control and energy recovery instruction topology diagram, drive the wide voltage adaptive power supply chip and energy recovery circuit to perform cycle-by-cycle voltage regulation power supply operation and energy recovery operation according to the execution priority sequence of the topology sorting, and at the same time feed the execution results back to the inertial navigation main controller for navigation accuracy verification and power supply strategy closed-loop correction.

[0057] Further, step S4 includes:

[0058] S41: According to the execution priority sequence determined by the topology sorting in the voltage regulation control and energy recovery instruction topology diagram, in each control cycle, read all nodes and their attributes corresponding to the current control cycle in sequence;

[0059] S42: For nodes with negative voltage jump variables in the current control cycle, drive the wide-voltage adaptive power supply chip to perform a step-down operation, adjusting the power supply voltage of the corresponding core device from the current voltage to the target output voltage value; during the transient process of the step-down operation, activate the energy recovery circuit, collect the transient current released by the inductor and capacitor in the device power supply circuit, convert the transient current into DC feedback current through the energy recovery circuit, and charge the power supply battery or energy storage capacitor; record the actual recovered energy value;

[0060] S43: For nodes with positive voltage jump variables in the current control cycle, check whether there is a directed edge for energy recovery pointing to the node in the voltage regulation control and energy recovery command topology graph; if there is a directed edge for energy recovery, prioritize using the actual recovered energy value corresponding to the directed edge for energy recovery to provide part of the energy for the boost operation of the core device, and supplement the insufficient part by the power supply battery; drive the wide voltage adaptive power supply chip to adjust the power supply voltage of the corresponding core device from the current voltage to the target output voltage value;

[0061] S44: For nodes where the voltage jump variable is zero in the current control cycle, maintain the output voltage of the wide voltage adaptive power supply chip unchanged, and only check whether the deviation between the power supply voltage and the target output voltage value is within the preset voltage stability tolerance range. If it exceeds the voltage stability tolerance range, perform fine-tuning compensation.

[0062] S45: Summarize the actual power supply voltage, actual power supply, and actual energy recovery value of all core devices after the current control cycle is completed into the power supply execution result of the current cycle and feed it back to the inertial navigation main controller;

[0063] S46: After receiving the power supply execution result of the current cycle, the inertial navigation main controller performs navigation accuracy verification and power supply strategy closed-loop correction.

[0064] Further, step S46 includes:

[0065] S461: The inertial navigation main controller evaluates the positioning accuracy and attitude accuracy indicators of the current navigation solution output based on the actual power supply and actual sampling frequency of each core device in the current cycle; it compares the positioning accuracy indicator with the preset lower limit of positioning accuracy and the attitude accuracy indicator with the preset lower limit of attitude accuracy.

[0066] S462: If the positioning accuracy index is lower than the lower limit of positioning accuracy or the attitude accuracy index is lower than the lower limit of attitude accuracy, it is determined that the current power supply strategy causes insufficient navigation accuracy. The inertial navigation main controller sends an accuracy protection command to the state discrimination link in step S1, forcibly upgrading the current carrier motion state classification label by one level, that is, upgrading the standby mode label to the normal working mode label, and the normal working mode label to the core working mode label, triggering the power supply parameters of the corresponding higher power level in the power classification configuration table to ensure that the navigation accuracy is restored to above the lower limit of accuracy.

[0067] S463: If the positioning accuracy index is higher than or equal to the lower limit of positioning accuracy and the attitude accuracy index is higher than or equal to the lower limit of attitude accuracy, then the current power supply strategy is determined to meet the navigation accuracy requirements, and the current power supply strategy is maintained unchanged; at the same time, the actual recovered energy value in the current cycle power supply execution result is added to the system energy recovery cumulative value, the state of charge data of the power supply battery is updated, and the updated remaining available energy is provided for the energy budget balancing algorithm in the next cycle step S3.

[0068] This invention provides an inertial navigation low-power power supply optimization system, comprising:

[0069] Data acquisition and state discrimination module: used to acquire raw angular velocity sampling data and raw acceleration sampling data of gyroscope and accelerometer in inertial navigation system, perform multi-scale motion feature extraction and state discrimination on the raw angular velocity sampling data and raw acceleration sampling data, obtain carrier motion state classification labels, and construct power classification configuration table based on the carrier motion state classification labels;

[0070] Time-sharing power supply scheduling module: Based on the power classification configuration table, it performs time-sharing on-demand power supply scheduling for each core component of the inertial navigation system, generates a real-time power command sequence for each core component, and constructs a circular buffer queue of device power status based on the real-time power command sequence;

[0071] Energy Budget and Scheduling Planning Module: This module is used to read the power status circular buffer queue of the devices, extract the historical power change trends of each core device, combine the state of charge data of the power supply battery, generate a forward-looking power scheduling plan through the energy budget balancing algorithm, and construct a voltage regulation control and energy recovery instruction topology based on the forward-looking power scheduling plan.

[0072] Voltage regulation execution and closed-loop correction module: It is used to read the voltage regulation control and energy recovery command topology diagram, drive the wide voltage adaptive power supply chip and energy recovery circuit to perform cycle-by-cycle voltage regulation power supply operation and energy recovery operation according to the execution priority sequence of the topology sorting, and at the same time feed back the execution results to the inertial navigation main controller for navigation accuracy verification and power supply strategy closed-loop correction.

[0073] Compared with existing technologies, the beneficial effects of this invention are as follows: Addressing the problem of power redundancy and low energy utilization caused by constant full-power power supply in mobile inertial navigation systems, this invention collects raw data from gyroscopes and accelerometers, extracts multi-scale motion features, and generates a carrier motion state classification label through a hysteresis protection mechanism. It then constructs a power classification configuration table, enabling dynamic adjustment of the power supply power and sampling frequency of each core component as needed. This effectively reduces the ineffective power consumption of sensors and navigation computers, avoids the risk of misjudgment based on single features, and reduces switching losses in the power supply circuit through state switching hysteresis protection.

[0074] This invention employs differentiated time-sharing and on-demand control for different types of core components, fully exploring the energy-saving potential of various components. Simultaneously, it constructs a circular buffer queue for component power status and, combined with battery state of charge, generates a forward-looking scheduling plan through an energy budget balancing algorithm. This pre-balances energy consumption with task endurance requirements, preventing energy depletion mid-task. Furthermore, by constructing a voltage regulation control and energy recovery command topology, this invention achieves energy redistribution among different power supply channels, recovering and utilizing the stored energy released during voltage reduction transients, significantly improving the overall system energy utilization rate.

[0075] This invention also incorporates a navigation accuracy verification and closed-loop correction mechanism. When the power supply strategy affects navigation accuracy, it automatically increases the power level, ensuring a dynamic balance between navigation accuracy and power consumption optimization. This invention represents an optimization solution in the field of power supply and distribution, effectively improving the energy efficiency and operational reliability of inertial navigation systems. Attached Figure Description

[0076] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0077] Figure 1 This is a flowchart of a low-power power supply optimization method for inertial navigation according to the present invention;

[0078] Figure 2 This is a schematic diagram of the state switching hysteresis protection in the embodiment;

[0079] Figure 3 This is an example diagram of time-sharing on-demand power supply scheduling in the embodiments;

[0080] Figure 4 This is a schematic diagram of the circular buffer queue for device power status in the embodiment;

[0081] Figure 5 This is a topology diagram of voltage regulation control and energy recovery commands in the embodiment;

[0082] Figure 6 This is a functional block diagram of an inertial navigation low-power power supply optimization system according to the present invention. Detailed Implementation

[0083] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0084] Example

[0085] Please see Figure 1 As shown, this embodiment provides a low-power power supply optimization method for inertial navigation, including:

[0086] S1: Collect raw angular velocity and raw acceleration sampling data from the gyroscope and accelerometer in the inertial navigation system, extract multi-scale motion features and determine the state of the raw angular velocity and raw acceleration sampling data, obtain the carrier motion state classification label, and construct a power classification configuration table based on the carrier motion state classification label;

[0087] Further, step S1 includes:

[0088] S11: Using the gyroscope and accelerometer in the inertial navigation system, collect the three-axis raw angular velocity sampling data and three-axis raw acceleration sampling data of the carrier in real time at the system's rated sampling frequency; at the same time, obtain the task priority parameters in the current navigation task command;

[0089] Further, step S11 includes:

[0090] S111: In the inertial measurement unit of the inertial navigation system, gyroscopes and accelerometers are deployed along the three orthogonal axes of the carrier coordinate system. The gyroscopes measure the angular velocity of the carrier around each axis, and the accelerometers measure the specific force acceleration of the carrier along each axis. The raw angular velocity and acceleration data of the three axes are synchronously acquired at the system's rated sampling frequency. The rated sampling frequency is the highest sampling frequency specified in the inertial navigation system design specifications, ensuring that the sensors meet the Nyquist sampling theorem's bandwidth requirements for the carrier motion signal in all operating modes. Synchronous acquisition means that the gyroscopes and accelerometers use the same sampling clock signal as a reference, complete data sampling at the same time, and record the timestamp, ensuring that the raw angular velocity and acceleration data of the three axes are strictly aligned on the time axis. It should be noted that when the system is in normal operating mode or standby mode, the sampling frequency of the sensor's navigation data output is executed according to the down-frequency sampling frequency or intermittent sampling frequency in step S142. However, the monitoring channel used for motion state discrimination always operates in an independent low-power wake-up sampling mode. This wake-up sampling mode continuously collects data at a fixed monitoring sampling frequency not higher than the system's rated sampling frequency. The monitoring sampling frequency is determined based on twice the typical switching frequency of the carrier's motion state. For example, the monitoring sampling frequency can be set to 10 Hz to 50 Hz, and its power consumption is much lower than that at the rated sampling frequency. For example, for a MEMS inertial measurement unit, the system's rated sampling frequency can be set to 200 Hz; for a fiber optic gyroscope inertial measurement unit, the system's rated sampling frequency can be set to 1000 Hz.

[0091] S112: Read the current navigation mission instruction from the mission scheduling module of the inertial navigation main controller and extract the mission priority parameter. The mission priority parameter is a preset mission level identifier in the navigation mission instruction, including two values: a precise positioning mission identifier and a regular navigation mission identifier. When the vehicle performs a mission requiring high-precision positioning, the mission priority parameter is marked as a precise positioning mission identifier; when the vehicle performs a regular cruise or transfer mission, the mission priority parameter is marked as a regular navigation mission identifier. The mission priority parameter is set based on pre-configuration by the upper-level mission planning system according to operational or operational requirements, and is not determined autonomously by the power supply optimization system.

[0092] Specifically, the synchronous acquisition operation in S11 is the data source entry point for the entire power supply optimization method. In inertial navigation systems, gyroscopes and accelerometers are the core sensors for sensing the motion state of the carrier, and their output data directly reflects the angular and linear motion characteristics of the carrier. For mobile inertial navigation carriers that rely on battery power, the sensor sampling process itself is one of the main sources of power consumption. Therefore, the dynamic adjustment of the sampling frequency in subsequent steps must be based on the raw data acquired here. Synchronous acquisition ensures a strict correspondence between angular velocity data and acceleration data in the time dimension. When the carrier performs a maneuver, the changes in angular velocity and acceleration have a time coupling characteristic. If there is a time deviation between the two types of data, the motion feature extraction in the subsequent step S12 will not be able to accurately reflect the true motion state of the carrier, leading to misjudgment of the motion state. The monitoring channel used for motion state judgment operates independently of the navigation data output channel in a low-power wake-up sampling mode, ensuring that the changes in the carrier's motion state can still be continuously sensed during sensor frequency reduction or intermittent power supply, avoiding failure of motion state judgment due to data loss. The introduction of task priority parameters enables the power supply strategy to be linked with the requirements of upper-level tasks, avoiding the problem of insufficient navigation accuracy due to reduced power when performing high-precision tasks.

[0093] S12: Perform multi-scale motion feature extraction on the original angular velocity sampling data and the original acceleration sampling data to obtain a set of carrier motion feature parameters;

[0094] Further, step S12 includes:

[0095] S121: Calculate the root mean square (RMS) value of angular velocity and the peak value of angular velocity change rate within a preset sliding time window for the triaxial raw angular velocity sampling data. Calculate the RMS value of acceleration and the peak value of acceleration change rate within the same sliding time window for the triaxial raw acceleration sampling data. When the system is in normal operating mode or standby mode, step S121 uses the data collected by the monitoring channel for feature calculation. The window length of the sliding time window is determined based on the typical switching cycle of the carrier's motion state. The window length should be greater than the typical transition time of the carrier from stationary to maneuvering to avoid capturing incomplete motion transition processes. For example, the window length of the sliding time window can be set to 500 milliseconds to 2000 milliseconds. The RMS value of angular velocity is calculated as follows: take the average of the sum of the squares of the angular velocities of each axis of the triaxial raw angular velocity sampling data within the sliding time window, and then take the arithmetic square root. The RMS value of angular velocity characterizes the overall angular motion intensity of the carrier within this time window. The calculation method for the peak value of the angular velocity change rate is as follows: First-order difference operations are performed on the angular velocity sequences of each axis of the triaxial raw angular velocity sampling data within the sliding time window. The maximum value among the absolute values ​​of the first-order differences for each axis is taken as the peak value of the angular velocity change rate for that axis. Then, the maximum value of the three-axis angular velocity change rate peak values ​​is taken as the peak angular velocity change rate. The peak angular velocity change rate reflects the intensity of the carrier's angular motion; when the carrier makes sharp turns or rapid attitude adjustments, the peak angular velocity change rate increases significantly. The calculation methods for the root mean square value of acceleration and the peak value of acceleration change rate are the same as those for the root mean square value of angular velocity and the peak angular velocity change rate, except that the input data is replaced with triaxial raw acceleration sampling data.

[0096] S122: Calculate the short-time energy spectral density for the raw angular velocity sampling data and raw acceleration sampling data within the sliding time window, and extract the angular velocity energy concentration. The calculation process of the short-time energy spectral density is as follows: apply a Hanning window function to the time-domain signal within the sliding time window for windowing processing, perform a discrete Fourier transform on the windowed signal to obtain a frequency domain representation, and calculate the square of the amplitude of each frequency component in the frequency domain representation as a power spectral density estimate. The angular velocity energy concentration is defined as the ratio of the power spectral density integral value of the triaxial raw angular velocity sampling data within a preset motion-related frequency band to the power spectral density integral value across the entire frequency band. The preset motion-related frequency band is determined according to the carrier type. Its upper limit frequency should cover the highest angular velocity change frequency of typical carrier maneuvers, and its lower limit frequency should exclude the influence of sensor DC bias and low-frequency drift. For example, for UAV carriers, the motion-related frequency band can be set to 0.5 Hz to 20 Hz; for vehicle-mounted carriers, the motion-related frequency band can be set to 0.1 Hz to 5 Hz. The physical meaning of angular velocity energy concentration is as follows: when the carrier is in a meaningful motion state, the energy of its angular velocity signal is mainly concentrated in the motion-related frequency band, and the angular velocity energy concentration is close to 1; when the carrier is stationary, the gyroscope output is mainly a noise signal, and the noise energy is distributed in a wide frequency band, and the angular velocity energy concentration is much less than 1.

[0097] S123: The root mean square value of angular velocity, peak value of angular velocity change rate, root mean square value of acceleration, peak value of acceleration change rate, and angular velocity energy concentration are combined to form a set of carrier motion characteristic parameters. The set of carrier motion characteristic parameters is a one-dimensional ordered vector containing the values ​​of the above five characteristic parameters, arranged in a fixed order of root mean square value of angular velocity, peak value of angular velocity change rate, root mean square value of acceleration, peak value of acceleration change rate, and angular velocity energy concentration.

[0098] Specifically, the multi-scale motion feature extraction method is adopted in S12 because the determination of the carrier's motion state cannot rely solely on a single-dimensional feature. In practical inertial navigation applications, the carrier's motion patterns are complex. For example, when a drone is hovering and subjected to gusts of wind, the root mean square value of its instantaneous angular velocity may briefly increase. If only the root mean square value of angular velocity is relied upon for state determination, the hovering state may be misjudged as a maneuvering state, leading to an unnecessary increase in power supply. By simultaneously introducing the peak value of the rate of change of angular velocity as a supplementary criterion for the intensity of motion, and the concentration of angular velocity energy as an auxiliary criterion for the validity of motion, the carrier's motion state can be cross-validated at both the time domain and frequency domain scales, reducing the probability of misjudgment under a single feature dimension. The set of carrier motion feature parameters extracted in step S12 is the sole input for motion state determination in step S13. Without the multi-scale feature extraction in step S12, step S13 will not be able to obtain sufficient dimensional information for determination, resulting in a decrease in the accuracy of the motion state classification label, which in turn affects the effectiveness of the power classification configuration table in step S14.

[0099] S13: Based on the set of carrier motion feature parameters and combined with the task priority parameter, motion state is determined and a carrier motion state hierarchical label is generated;

[0100] Further, step S13 includes:

[0101] S131: Determine the carrier's motion state sequentially according to priority from high to low. First, determine the core working mode: The determination criteria for the core working mode are: the root mean square value of angular velocity in the carrier motion characteristic parameter set is greater than or equal to the first angular velocity threshold, or the root mean square value of acceleration is greater than or equal to the first acceleration threshold, or the task priority parameter is marked as a precise positioning task identifier. If the core working mode determination criteria are met, it is directly determined to be in the core working mode, and no further determination is made. If the core working mode determination criteria are not met, continue to determine the standby mode: The determination criteria for the standby mode are: the root mean square value of angular velocity is less than the second angular velocity threshold, and the root mean square value of acceleration is less than the second acceleration threshold, and the peak value of the rate of change of angular velocity is less than the first rate of change of angular velocity threshold, and the peak value of the rate of change of acceleration is less than the first rate of change of acceleration threshold, and the angular velocity energy concentration is less than the first energy concentration threshold. Among them, the first angular velocity threshold is greater than the second angular velocity threshold, and the first acceleration threshold is greater than the second acceleration threshold. The first angular velocity threshold is determined based on the lower limit of the typical angular velocity range of the carrier type under high-speed maneuvering conditions. For example, for a UAV carrier, the first angular velocity threshold can be set to 30 degrees per second; for a vehicle-mounted carrier, the first angular velocity threshold can be set to 15 degrees per second. The second angular velocity threshold is determined based on three times the standard deviation of the gyroscope noise level of the carrier type under stationary conditions. For example, for a MEMS gyroscope, the second angular velocity threshold can be set to 0.5 degrees per second. The determination of the first acceleration threshold and the second acceleration threshold is similar to that of the angular velocity threshold, respectively determined based on the lower limit of the typical acceleration range of the carrier under high-speed maneuvering conditions and the accelerometer noise level under stationary conditions. The first angular velocity rate of change threshold and the first acceleration rate of change threshold are determined based on the upper limit of the angular velocity rate of change and the upper limit of the acceleration rate of change of the carrier under low-speed steady motion conditions, respectively. The first energy concentration threshold is determined based on the statistical mean of the energy concentration of the gyroscope noise signal under stationary conditions.

[0102] S132: If neither the core operating mode determination condition nor the standby mode determination condition is met, then it is determined to be in normal operating mode. The state that meets the core operating mode determination condition is marked as the core operating mode label, the state that meets the standby mode determination condition is marked as the standby mode label, and the normal operating mode is marked as the normal operating mode label. The core operating mode label, normal operating mode label, and standby mode label are collectively referred to as the carrier motion state hierarchical label. The carrier motion state hierarchical label is an enumerated type variable, and its value space includes three discrete values: the core operating mode label, the normal operating mode label, and the standby mode label.

[0103] S133: Introducing a state switching hysteresis protection mechanism. Specifically, when a parameter in the carrier motion characteristic parameter set crosses the boundary from one mode determination interval to an adjacent mode determination interval, the carrier motion state classification label is only updated after the parameter has remained continuously within the boundary interval for a duration exceeding a preset hysteresis confirmation duration. Specifically, a fast response hysteresis confirmation duration is used when switching from a low-power mode to a high-power mode, and a normal hysteresis confirmation duration is used when switching from a high-power mode to a low-power mode. The fast response hysteresis confirmation duration is shorter than the normal hysteresis confirmation duration. For example, the fast response hysteresis confirmation duration can be set to 50 milliseconds to 200 milliseconds, and the normal hysteresis confirmation duration can be set to 200 milliseconds to 1000 milliseconds. The normal hysteresis confirmation duration is determined based on the typical switching transition time of the carrier motion state and the response time of the power supply circuit. The normal hysteresis confirmation duration should be greater than the stabilization time required for the power supply circuit to switch from one power level to another, to avoid the power supply circuit receiving a new switching command before the previous switching is completed. During the hysteresis confirmation period, the carrier motion status classification label retains the value of the previous cycle.

[0104] See Figure 2 This is a schematic diagram of the state switching hysteresis protection provided in an embodiment of this application. For example... Figure 2 As shown, the upper part is a continuous curve showing the change of the carrier's motion characteristic parameters (taking the root mean square value of angular velocity as an example) over time. The vertical axis is marked with the first and second angular velocity thresholds, dividing the vertical axis space into three regions: the core working mode determination interval, the normal working mode determination interval, and the standby mode determination interval. The curve shows several brief out-of-bounds jitters near the thresholds. Figure 2 The dashed box in the middle marks the "filtering high-frequency jitter" area, indicating that these instantaneous out-of-bounds events will not immediately trigger a state switch. The double-headed arrow indicating "hysteresis confirmation time" means that the parameters must remain within the out-of-bounds range for a preset time before the carrier motion state classification label is updated. The lower half shows the step transition curve of the corresponding carrier motion state classification label over time. It can be seen that the classification label transitions sequentially from the standby mode label to the normal working mode label, and then to the core working mode label. Each transition occurs after the parameters in the upper half have continuously met the out-of-bounds conditions and exceeded the hysteresis confirmation time. In actual inertial navigation applications such as UAV flight, the angular velocity parameters of the carrier are prone to high-frequency jitter near the state boundary due to airflow disturbances or body vibration. If the motion state is directly determined based on instantaneous parameters, the power supply will frequently jump between adjacent levels, increasing the switching losses of power switching devices and causing transient impacts in the power supply circuit. This hysteresis protection mechanism effectively filters out false state switching caused by short-term jitter by setting an acknowledgment threshold in the time dimension. It only triggers power level adjustment when the carrier's motion state undergoes a continuous change, ensuring the smooth operation of the power supply system and the reliability of the power supply optimization strategy.

[0105] Specifically, the introduction of the state switching hysteresis protection mechanism in S13 addresses the practical engineering problem of frequent fluctuations in the motion state of mobile inertial navigation vehicles. During UAV flight, the vehicle's motion parameters may experience high-frequency jitter near the state boundaries due to airflow disturbances or vibrations. If the motion state is determined directly based on instantaneous parameters, the vehicle's motion state classification label will rapidly and repeatedly switch between adjacent modes, causing frequent power jumps between different power levels. Frequent power jumps not only increase the switching losses of power switching devices in the power supply circuit but also generate unnecessary transient responses in inductors and capacitors, thus increasing system energy consumption. The hysteresis protection mechanism filters out short-term parameter jitter by setting a confirmation threshold in the time dimension. Power level switching is only triggered when the motion state has indeed undergone a sustained change, ensuring the stable operation of the power supply system. A shorter fast-response hysteresis confirmation time is used when switching from a low-power mode to a high-power mode, ensuring that the system can respond quickly when the vehicle suddenly starts moving, avoiding navigation accuracy degradation due to response delays in the initial stage of maneuvering.

[0106] S14: Construct a power classification configuration table based on the carrier motion state classification label;

[0107] Further, step S14 includes:

[0108] S141: Establish a component list for each core component in the inertial navigation system, including gyroscopes, accelerometers, and a navigation computer. For each core component in the component list, set its rated power supply value under the core operating mode label, its reduced power supply value under the normal operating mode label, and its minimum hold power supply value under the standby mode label. The rated power supply value is the nominal power of the core component under full-performance operation, provided by the component manufacturer's technical manual. The reduced power supply value is determined based on the actual power consumption test results of the core component after reducing the sampling frequency or shutting down some functional channels, and the reduced power supply value is less than the rated power supply value. For the navigation computer, the reduced power supply value is the power supply value corresponding to the navigation computer under a preset reference load rate, which is set to 50%. The minimum hold power supply value is the minimum power required for the core component to maintain a basic power supply state without continuous measurement. This power must ensure the normal operation of the internal reference circuit and temperature control circuit of the component so that a complete power-on preheating process is not required when full power operation is needed. For example, for a MEMS gyroscope, the rated power supply value can be 0.5 watts, the downshifted power supply value can be 0.2 watts, and the minimum hold power supply value can be 0.05 watts.

[0109] S142: For each core device in the device list, set its rated sampling frequency under the core operating mode label, its down-frequency sampling frequency under the normal operating mode label, and its intermittent sampling frequency under the standby mode label. The down-frequency sampling frequency is determined based on the highest effective frequency component of the motion signal of the carrier in the normal operating mode, and the down-frequency sampling frequency should not be less than twice that highest effective frequency component. The intermittent sampling frequency corresponds to an intermittent power supply cycle, which includes a power supply activation period and a power supply sleep period. During the power supply activation period, the device collects data at the down-frequency sampling frequency; during the power supply sleep period, the device enters a low-power hold state, maintaining only the minimum hold power value. The total duration of the intermittent power supply cycle and the duty cycle of the power supply activation period are determined based on the minimum refresh rate requirement for carrier motion status monitoring in standby mode. For example, the intermittent power supply cycle can be set to 1 second, and the duty cycle of the power supply activation period can be set to 10% to 30%. It should be noted that the aforementioned down-frequency sampling frequency and intermittent sampling frequency only apply to the navigation data output channel of the sensor. The monitoring channel used for motion state discrimination always operates independently according to the monitoring sampling frequency defined in step S111, and is not affected by the sampling frequency adjustment of the navigation data output channel.

[0110] S143: Arrange the power supply values ​​and sampling frequencies of each core device under each carrier motion state classification label, with rows representing core devices and columns representing carrier motion state classification labels, to construct a power classification configuration table. The power classification configuration table has a two-dimensional data structure, with the number of rows equal to the number of core devices in the device list, and three columns corresponding to the core operating mode label, the normal operating mode label, and the standby mode label. Each entry in the power classification configuration table includes a device identifier, the corresponding carrier motion state classification label, the power supply value, the sampling frequency, and the intermittent power supply cycle parameter. For entries under the normal operating mode label for the navigation computer, the power supply value field records the downgraded power supply value under the reference load rate, and includes a dynamic frequency modulation flag. The dynamic frequency modulation flag indicates that the actual power supply of the device in the normal operating mode needs to be calculated in real time according to the dynamic frequency modulation logic in step S223. The downgraded power supply value recorded in the power classification configuration table serves as the reference benchmark value for dynamic frequency modulation calculation. The device identifier is a unique number for each core device in the device list. The power classification configuration table is stored in the non-volatile memory of the power supply optimization control system. It is written once during the system initialization phase and repeatedly queried by the power supply scheduling logic in step S2 during the operation phase.

[0111] Specifically, step S1 acquires motion data from the inertial navigation system's own sensors, uses multi-dimensional feature extraction and threshold discrimination mechanisms to divide the carrier's motion state into three power levels, and solidifies the power supply parameters of each core component at each level into a power grading configuration table. This step solves the problem that existing inertial navigation systems cannot distinguish the differences in carrier motion states under constant full-power power supply mode. By introducing cross-discrimination of time-domain features such as the root mean square value of angular velocity and the peak value of the rate of change of angular velocity, and frequency-domain features such as the concentration of angular velocity energy, misjudgment of state due to sensor noise or environmental disturbances under a single feature dimension is avoided. The state switching hysteresis protection mechanism ensures the smoothness of power level switching and prevents the power supply circuit from generating additional switching losses and transient impacts due to frequent switching. The power grading configuration table, as the core variable passed from step S1 to step S2, carries the complete power supply parameters of all core components in all motion states in the form of a structured two-dimensional table. It provides a deterministic query basis for time-sharing on-demand power supply scheduling in the subsequent step S2, so that the power supply scheduling logic does not need to repeatedly perform state discrimination and parameter calculation, reducing the computational power consumption of the power supply optimization control system itself.

[0112] S2: Based on the power level configuration table, perform time-sharing on-demand power supply scheduling for each core component of the inertial navigation system, generate a real-time power command sequence for each core component, and construct a circular buffer queue of device power status based on the real-time power command sequence;

[0113] Further, step S2 includes:

[0114] S21: Read the power grading configuration table. Based on the current carrier motion state grading label, query the power supply value and sampling frequency of each core device in the power grading configuration table that match the current carrier motion state grading label, and generate the power allocation instruction group for the current cycle. The current carrier motion state grading label is updated and output at the end of each sliding time window in step S13. The power allocation instruction group is a data set containing the power supply value and sampling frequency that all core devices should execute in the current control cycle. Each record in the power allocation instruction group corresponds to one core device. For core devices in the power grading configuration table where the dynamic frequency modulation flag is true, the power supply value in the power allocation instruction group is a placeholder mark to be dynamically calculated and determined in step S223.

[0115] S22: Execute time-sharing on-demand control logic for power supply scheduling of each core device in the power allocation instruction group to generate a real-time power instruction sequence;

[0116] Further, step S22 includes:

[0117] S221: For the gyroscope, when the carrier motion status classification label is the core operating mode label, a power supply command is output to the gyroscope according to the rated power value and rated sampling frequency corresponding to the core operating mode label in the power classification configuration table. When the carrier motion status classification label is the standby mode label, a power supply command corresponding to the reduced power value is output to the gyroscope during the power supply activation period, and a zero power command is output to the gyroscope during the power supply sleep period, according to the intermittent power supply cycle parameters in the power classification configuration table. When the carrier motion status classification label is the normal operating mode label, differentiated control is performed according to the gyroscope's device type: if the gyroscope type is a MEMS gyroscope, the corresponding reduced power value command is output according to the reduced sampling frequency; if the gyroscope type is a fiber optic gyroscope or laser gyroscope, the power supply of redundant detection channels is turned off, only the power supply of the core measurement channel is retained, and the power supply value command corresponding to the core measurement channel is output. The redundant detection channel refers to the backup optical path or backup detector channel set up in the fiber optic gyroscope or laser gyroscope to improve reliability. In normal working mode, the output accuracy of a single core measurement channel is sufficient for navigation requirements. Turning off the redundant detection channel can reduce the power consumption of the light source drive and photoelectric detection circuit.

[0118] S222: For the accelerometer, when the carrier motion status classification label is the core operating mode label, a power supply command is output according to the rated power supply value and rated sampling frequency. When the carrier motion status classification label is the standby mode label, the same intermittent power supply control logic as the gyroscope is executed according to the intermittent power supply cycle parameters, outputting a power supply command corresponding to the reduced power supply value during the power supply activation period and a zero power command during the power supply sleep period. When the carrier motion status classification label is the normal operating mode label, a reduced power supply value command is output according to the reduced sampling frequency.

[0119] S223: For the navigation computer, dynamic frequency modulation power supply control is adopted. Specifically, the current processing load rate of the navigation computer's processor is acquired in real time. The processing load rate is the ratio of the number of processing cycles currently occupied by the processor to the total number of processing cycles at the current actual operating frequency. When the carrier motion status classification label is the core working mode label, a power supply command is output according to the rated power value and rated operating frequency. When the carrier motion status classification label is the standby mode label and the processing load rate is lower than the preset low load threshold, the processor clock frequency is reduced to the preset minimum operating frequency, and a minimum holding power value command corresponding to the minimum operating frequency is output. The low load threshold is determined based on the minimum amount of computation required for the navigation computer to maintain basic status monitoring and communication functions. For example, the low load threshold can be set to 15%. When the carrier motion status classification label is the normal working mode label, a stepped frequency modulation strategy is adopted to determine the target operating frequency: the frequency range between the minimum operating frequency and the rated operating frequency is divided into N frequency levels, and the corresponding frequency level is selected as the target operating frequency according to the range in which the processing load rate is located. Specifically, when the computing load rate is in the k-th interval, the target operating frequency is set to the k-th frequency level value, where k is equal to the computing load rate multiplied by N and rounded up, and the value of k is at least 1. For example, N can be set to 4 to 8. Frequency level switching is only performed when the computing load rate remains within the same interval for M consecutive control cycles. M can be set to 3 to 5 control cycles to avoid frequent frequency switching due to instantaneous fluctuations in the load rate. The power supply value command corresponding to the target operating frequency is output. The power supply value corresponding to the target operating frequency is determined by looking up the frequency-power mapping table of the navigation computer. The dynamic frequency adjustment power supply control of the navigation computer utilizes the positive correlation between processor power consumption and clock frequency. When the processor computing load is low, reducing the clock frequency can significantly reduce the processor's dynamic power consumption.

[0120] S224: Arrange the power supply value commands generated by each core device in each control cycle according to the timestamp to form a real-time power command sequence. The data structure of the real-time power command sequence is an ordered list. Each command in the list includes a timestamp, device identifier, target power supply value, target sampling frequency, and channel control flag. The channel control flag is a Boolean variable used to indicate whether the current command involves closing the redundant detection channel. When the channel control flag is true, it indicates that the redundant detection channel needs to be closed; when the channel control flag is false, it indicates that the channel state does not need to be changed.

[0121] See Figure 3 This is an example diagram of time-sharing on-demand power supply scheduling provided in the embodiments of this application. Figure 3As shown, three sub-graphs illustrate the changes in power supply waveforms of the three core components—gyroscope, accelerometer, and navigation computer—over time under different carrier motion state grading labels. The bottom time axis is divided into three stages: standby mode label, normal operating mode label, and core operating mode label, using vertical dashed lines. In the gyroscope and accelerometer sub-graphs, the standby mode label stage exhibits periodic pulse waveforms. The high-level pulse segment corresponds to the power activation period in the intermittent power supply cycle, during which the device briefly operates at a reduced power supply value and collects data. The low-level pulse segment corresponds to the power sleep period, during which the device outputs a zero-power command to maximize battery energy conservation. After entering the normal operating mode label stage, the waveform jumps to the reduced power supply value and remains continuous. The accelerometer sub-graph is marked with the explanation "Powered at a reduced sampling frequency." After entering the core operating mode label stage, the waveform further jumps to the rated power supply value. The navigation computer's sub-map maintains a constant minimum power supply during the standby mode phase, exhibits irregular line fluctuations during the normal operating mode phase marked "dynamic frequency adjustment power supply," reflecting the process of linear interpolation adjustment of the processor clock frequency based on the computational load rate, and jumps to the rated power supply value during the core operating mode phase. In battery-powered mobile inertial navigation systems, different types of core components have differentiated power consumption characteristics. This time-sharing on-demand power scheduling mechanism adopts differentiated control strategies such as intermittent power supply, frequency reduction power supply, and dynamic frequency adjustment power supply according to the physical characteristics of each component. Compared with the traditional constant full-power power supply mode, it can fully explore the energy-saving potential of various components under non-full-load conditions while ensuring that each component meets the navigation accuracy requirements under the current motion state, effectively extending the battery's runtime.

[0122] Specifically, step S22 employs differentiated time-sharing on-demand control logic for different types of core devices based on the varying power consumption characteristics of various inertial navigation devices. The power consumption of MEMS gyroscopes is approximately linearly related to the sampling frequency; reducing the sampling frequency directly reduces power consumption. The power consumption of fiber optic gyroscopes and laser gyroscopes mainly originates from the light source drive and photoelectric detection circuits; disabling redundant detection channels is more energy-efficient than simply reducing the sampling frequency. The power consumption of the navigation computer is primarily driven by the processor's dynamic power consumption, which is approximately proportional to the square of the clock frequency; dynamic frequency modulation is the most effective means of reducing processor power consumption. The navigation computer employs a stepped frequency modulation strategy combined with a confirmation mechanism for M consecutive cycles, avoiding the positive feedback oscillation problem between the computational load rate and the clock frequency under linear interpolation. If a uniform frequency reduction strategy is applied to all types of devices, the energy-saving potential of each device cannot be fully exploited. The real-time power command sequence generated in step S22 is the data source for constructing the device power status circular buffer queue in step S23. If the differentiated control logic in step S22 is missing, the power values ​​in the real-time power command sequence will not accurately reflect the actual power supply requirements of each device, resulting in distortion of the historical power change trend analysis in the subsequent step S3.

[0123] S23: Construct a circular buffer queue for device power status based on real-time power command sequence;

[0124] Further, step S23 includes:

[0125] S231: A circular buffer space of fixed length L is allocated in the memory of the power supply optimization control system, and the write pointer is initialized to point to the beginning of the circular buffer space. The fixed length L is the preset length of the historical power state recording window. The value of L is determined according to the number of historical control cycles that the power supply optimization control system needs to trace back. L should cover the number of continuous cycles in which the carrier transitions completely from one typical motion state to another. For example, L can be set to 100 to 500. The circular buffer space is a fixed-size storage area that is continuously allocated in physical memory and logically connected end to end to form a circular structure. At the same time, a write counter is maintained and initialized to zero to record the number of power state recording nodes that have been written.

[0126] S232: After generating a real-time power command sequence in each control cycle, the current target power supply value, current carrier motion state classification label, and current timestamp of each core device in the real-time power command sequence are encapsulated into a power state record node and written to the position pointed to by the write pointer in the circular buffer space. The write pointer moves one position forward, and the written counter is incremented by one but not exceeding L. When the write pointer reaches the end of the circular buffer space, it wraps back to the first position and overwrites the earliest power state record node. The data structure of the power state record node is a fixed-length record body. Each power state record node contains the following fields: timestamp field, carrier motion state classification label field, and target power supply value field corresponding to each core device in the device list.

[0127] S233: The device power status circular buffer queue is the circular buffer space where power status record nodes have been written. The device power status circular buffer queue stores power status record nodes from the most recent L control cycles. Each power status record node contains the target power supply value for all core devices, a carrier motion status classification label, and a timestamp. The device power status circular buffer queue supports traversal reading in timestamp order. The starting position of the read operation is determined based on the written counter: when the written counter is less than L, the read operation starts from the first position of the circular buffer space and traverses the power status record nodes indicated by the written counter; when the written counter is greater than or equal to L, the read operation starts from the current position of the write pointer and traverses L power status record nodes in circular order. At this time, the current position of the write pointer is the position of the oldest record node. It also supports indexing and querying the target power supply value of the corresponding device within any power status record node by device identifier.

[0128] See Figure 4 This is a schematic diagram of the device power state ring buffer queue provided in the embodiments of this application. Figure 4As shown, the central element is a ring structure composed of multiple sectors, representing a ring buffer space of fixed length L. Each sector is labeled clockwise with time markers such as period T, period T-1, period T-2, and period T-3, indicating that each power state recording node is written sequentially according to the control cycle. The dark gray sector labeled "Period T" represents the latest power state recording node currently being written; the light gray sector labeled "Period T-L+1" represents the oldest power state recording node in the ring buffer, which will be overwritten by the next write operation; the remaining white sectors represent historical nodes that have already been written. The "Write Pointer" arrow in the upper left corner points to the current write position, and the "Read Pointer" arrow in the upper right corner points to the earliest recorded position. The inertial navigation system carrier icon on the left is connected to the ring buffer via an arrow indicating "Real-time Encapsulation of Target Power Value," signifying that the power supply execution results for each control cycle are written to the queue in real time. The details box on the right displays the internal data structure of a single power state recording node, including a timestamp field, a carrier motion state classification label field, and target power value fields for the gyroscope, accelerometer, and navigation computer, respectively. The dotted arc arrow and text at the bottom explain the circular overwrite mechanism: "Write clockwise according to the control cycle, and wrap around to overwrite the old record at the end." In resource-constrained embedded inertial navigation power supply optimization control systems, the circular cache data organization method can continuously record the complete power state history of the most recent L control cycles with a fixed memory footprint and constant time complexity. This avoids the risk of memory exhaustion caused by the infinite growth of linear lists, and retains timely recent historical data through the automatic overwrite mechanism. This provides efficient and reliable data support for subsequent energy budget equalization algorithms to extract trend characteristics such as the power change slope and power fluctuation standard deviation of each core device.

[0129] Specifically, step S2 transforms the static power supply parameters in the power grading configuration table into a dynamic real-time power command sequence through time-sharing on-demand power supply scheduling, and organizes the historical records of the real-time power command sequence into a device power status circular cache queue. The device power status circular cache queue, as the core data structure passed from step S2 to step S3, uses a circular cache organization instead of a linear list because the power supply optimization control system runs on a resource-constrained embedded platform. The fixed memory usage and constant-time write operations of the circular cache are suitable for the deterministic timing requirements of the real-time control system. The circular cache's overwrite mechanism automatically discards the oldest power status record node, retaining historical data from the most recent L control cycles. The data within this window is sufficient to reflect the recent trends in the carrier's motion state, providing timely historical data for the energy budget balancing algorithm in step S3. Using an infinitely growing linear list to store historical data would lead to a continuous increase in memory usage, eventually exhausting the limited memory resources of the embedded system. The ability to query by device identifier in the device power status circular cache queue allows step S3 to quickly locate the power data of a specific device when analyzing the historical power change trends of each core device, avoiding the need to traverse and filter all records.

[0130] S3: Read the device power status circular buffer queue, extract the historical power change trend of each core device, combine it with the state of charge data of the power supply battery, generate a forward-looking power scheduling plan through the energy budget balancing algorithm, and construct a voltage regulation control and energy recovery instruction topology based on the forward-looking power scheduling plan;

[0131] Further, step S3 includes:

[0132] S31: Read all power status record nodes of the most recent L control cycles from the device power status circular buffer queue in timestamp order, and extract the historical power change trend of each core device.

[0133] Further, step S31 includes:

[0134] S311: For each power status record node stored in the device power status circular buffer queue, group them according to device identifier to obtain the target power supply value time series of each core device in the most recent L control cycles. The target power supply value time series is a one-dimensional ordered numerical sequence, the sequence length of which is equal to the smaller value of the counter written in the device power status circular buffer queue and L. The i-th element in the sequence is the target power supply value of the core device in the i-th power status record node arranged in timestamp order.

[0135] S312: Perform linear regression fitting on the time series of the target power supply value for each core device, and calculate the power change slope and the standard deviation of power fluctuation. The linear regression fitting uses the least squares method, with the control period number as the independent variable and the target power supply value as the dependent variable, to fit a linear function. The coefficient of the first-order term is the power change slope. The standard deviation of power fluctuation is the standard deviation of the residuals of the target power supply value time series relative to the linear regression fitted value. The power change slope and the standard deviation of power fluctuation are used as parameters describing the historical power change trend of the core device. The physical meaning of the power change slope is: a positive value indicates that the power supply of the device has recently shown an upward trend, indicating that the carrier's motion intensity is increasing; a negative value indicates that the power supply has recently shown a downward trend, indicating that the carrier is tending to stabilize or remain stationary. The physical meaning of the standard deviation of power fluctuation is: the larger the value, the more drastic the recent fluctuation of the power supply of the device, indicating frequent switching of the carrier's motion state; the smaller the value, the more stable the recent power supply.

[0136] S313: Calculate the overall system power consumption rate based on the historical power change trend description parameters of each core component. The overall system power consumption rate is calculated by summing the power change slopes of all core components. The overall system power consumption rate characterizes the trend of the total power consumption of the inertial navigation system over time. When the overall system power consumption rate is positive, it indicates that the total system power consumption is increasing and the discharge rate of the power supply battery is accelerating.

[0137] S32: Collect the state of charge data of the power supply battery, combine it with the overall power consumption rate of the system, and generate a forward-looking power scheduling plan through the energy budget balancing algorithm;

[0138] Further, step S32 includes:

[0139] S321: The battery management chip in the power supply module collects the current state of charge percentage, current output voltage, and current discharge current of the power supply battery. Based on the current state of charge percentage and the nominal capacity of the power supply battery, the remaining usable energy of the power supply battery is calculated. The remaining usable energy is calculated as follows: remaining usable energy equals the nominal capacity of the power supply battery multiplied by the current state of charge percentage multiplied by the nominal voltage of the power supply battery. Based on the overall system power consumption rate and the current total system power, the estimated remaining runtime under the current power consumption mode is estimated. The current total system power is the sum of the target power values ​​of all core devices in the latest power state record node in the device power state circular cache queue. The estimated remaining runtime is calculated as follows: when the overall system power consumption rate is less than or equal to zero, the estimated remaining runtime equals the remaining usable energy divided by the current total system power; when the overall system power consumption rate is greater than zero, the estimated remaining runtime equals the remaining usable energy divided by the current total system power and the sum of the overall system power consumption rate multiplied by half of a preset trend extrapolation period. ,in To estimate the remaining battery life, For the remaining usable energy, This represents the current total power of the system. The preset trend extrapolation period is the overall power consumption rate of the system. The fixed length L of the device power state ring buffer queue is equal to the preset trend extrapolation period number, which is equal to the fixed length L of the device power state ring buffer queue.

[0140] S322: Compare the estimated remaining battery life with the preset remaining mission duration. The remaining mission duration is calculated and provided by the inertial navigation main controller's mission scheduling module based on the current mission plan. If the estimated remaining battery life is greater than or equal to a preset safety multiple of the remaining mission duration, the power supply value in the current power grading configuration table remains unchanged, and the planned power supply value sequence of each core device in the generated forward power scheduling plan is the same as the target power supply value in the current real-time power command sequence. The preset safety multiple is determined based on the battery capacity estimation error and the uncertainty of mission execution time; for example, the preset safety multiple can be set to 1.2 to 1.5. If the estimated remaining battery life is less than the preset safety multiple of the remaining mission duration, the energy budget equalization algorithm is activated.

[0141] S323: The energy budget balancing algorithm includes the following steps: First, calculate the target average power based on the remaining task duration and a preset safety factor. The target average power is equal to the remaining available energy divided by the product of the remaining task duration and the preset safety factor. Then, calculate the total power to be reduced. The total power to be reduced is equal to the current system total power minus the target average power. When the total power to be reduced is less than or equal to zero, no power reduction operation is performed, and the current power supply value remains unchanged. When the total power to be reduced is greater than zero, sort the core devices according to the power fluctuation standard deviation in the historical power change trend description parameters of each core device, in descending order of power fluctuation standard deviation. Allocate power reductions to each core device according to the sorting results: for core devices ranked higher (i.e., those with larger power fluctuation standard deviations), allocate larger power reductions first. For each core device to be reduced, calculate its power reduction amount, which does not exceed the difference between the current target power supply value of the core device and the minimum maintained power supply value of the core device under the standby mode label in the power grading configuration table. Specifically, the power reduction amount for each core device to be reduced is equal to the total power to be reduced multiplied by the ratio of the power fluctuation standard deviation of that core device to the sum of the power fluctuation standard deviations of all core devices. If the power reduction amount calculated by this ratio exceeds the maximum reduction amount of that device (i.e., the current target power supply value minus the minimum maintainable power supply value), then the maximum reduction amount is taken as the power reduction amount for that device, and the excess is allocated to the next core device in the sequence. The basis for prioritizing power reduction in descending order of power fluctuation standard deviation is as follows: devices with larger power fluctuation standard deviations indicate that their power supply has changed frequently recently, and the corresponding carrier motion channel has experienced many switching of operating modes recently. Their instantaneous power contains more non-steady-state energy consumption components, and reducing the power of such devices has a relatively small steady-state impact on navigation accuracy.

[0142] S324: The target power supply value, adjustment effective time interval, and power transition slope limit of each core device after adjustment by the energy budget equalization algorithm are encapsulated into a forward-looking power scheduling plan. The adjusted target power supply value is equal to the current target power supply value of the core device minus the allocated power reduction. The adjustment effective time interval is a series of consecutive control cycles starting from the current control cycle, and the number of control cycles is equal to the fixed length L of the device power state circular buffer queue. The power transition slope limit is the maximum power change allowed in each control cycle when each core device transitions from the current target power supply value to the adjusted target power supply value. The power transition slope limit is set based on the maximum current change rate limit of the power converter in the power supply loop and the power step response characteristics of the device. For example, the power transition slope limit can be set to not exceed 5% of the rated power supply value in each control cycle. The data structure of the forward-looking power scheduling plan is a two-dimensional table with the number of rows equal to the number of core devices. Each row contains the device identifier, the adjusted target power supply value, the start control cycle number of the adjustment effective time interval, the end control cycle number of the adjustment effective time interval, and the power transition slope limit. In the forward-looking power scheduling plan, the sequence of planned power supply values ​​for each core device in each future control cycle is determined by decreasing the current target power supply value cycle by cycle according to the power transition slope limit until the adjusted target power supply value is reached. After that, the adjusted target power supply value remains unchanged for the remaining control cycle within the adjustment effective time interval.

[0143] Specifically, the energy budget balancing algorithm in step S32 is designed to treat the limited energy of the power supply battery as a budgetary resource that needs to be rationally allocated throughout the mission's entire lifecycle. Existing inertial navigation systems' power supply strategies only focus on the current power demand, neglecting the matching relationship between remaining battery energy and the remaining mission duration. This can lead to situations where the battery energy is depleted before the mission is completed. The energy budget balancing algorithm compares the expected remaining flight time with the remaining mission duration to identify the risk of insufficient energy in advance. While ensuring all core components maintain a minimum power supply value, it selectively reduces power for components with large power fluctuations. The setting of a power transition slope limit prevents sudden power drops from impacting the device's operating state, ensuring the smoothness of the power adjustment process. The historical power change trend description parameters extracted in step S31 provide a sorting basis for power reduction allocation in step S32. Without the analysis of the device power state circular buffer queue in step S31, step S32 would be unable to distinguish the differences in power stability among components, and would only adopt a uniform reduction strategy. This could lead to unnecessary power reduction for key measurement devices with small power fluctuations and relatively stable operating states, affecting navigation accuracy.

[0144] S33: Based on a forward-looking power scheduling plan, construct a topology diagram for voltage regulation control and energy recovery commands;

[0145] Further, step S33 includes:

[0146] S331: Based on the planned power supply value sequence of each core device in the forward-looking power scheduling plan, calculate the target output voltage value of each core device in each control cycle. The target output voltage value is determined by looking up a power-voltage mapping table. This table is pre-established based on the correspondence between rated power and rated supply voltage in the device specification manuals provided by each core device manufacturer. After offline calibration verification and correction of each core device during the system initialization phase, it is stored in the non-volatile memory of the power optimization control system, recording the required supply voltage value of each core device under different power supply values. When looking up the table, the planned power supply value of each core device is used as the index to find the entry in the power-voltage mapping table that is closest to that power value, and the corresponding supply voltage value is taken as the target output voltage value. When the planned power supply value is between two adjacent entries, a linear interpolation method is used to calculate the corresponding target output voltage value. For each core device, the difference between the target output voltage values ​​of adjacent control cycles is recorded as the voltage jump variable. The voltage jump variable is a signed value; a positive value indicates that a boost operation is required, a negative value indicates that a buck operation is required, and a zero value indicates that the current voltage remains unchanged.

[0147] S332: Construct the node set of the voltage regulation control and energy recovery command topology graph. The voltage regulation control and energy recovery command topology graph is a directed acyclic graph, and its data structure consists of a node set and a directed edge set. Each node in the node set represents a power supply operation for a core device within a control cycle. The total number of nodes is equal to the number of core devices multiplied by the number of control cycles spanned by the adjustment effective time interval in the forward-looking power scheduling plan. Each node contains the following attributes: device identifier, control cycle number, target output voltage value, voltage jump variable, and device type identifier. The device type identifier is used to distinguish between gyroscopes, accelerometers, and navigation computers, so that the corresponding power supply channel can be selected according to the device type in step S4.

[0148] S333: Construct a set of directed edges for the voltage regulation control and energy recovery command topology graph. Directed edges represent execution dependencies and energy transfer relationships between nodes. Specifically, within the same control cycle, traverse all nodes corresponding to core devices. If the voltage jump variable of a core device is negative (meaning the node performs a buck operation), and the estimated recoverable energy of this buck operation is greater than a preset minimum energy recovery threshold, and the voltage jump variable of another core device is positive (meaning the node performs a boost operation), then add a directed energy recovery edge from the buck operation node to the boost operation node. The minimum energy recovery threshold is determined based on the control power consumption of the energy recovery circuit itself. An energy recovery directed edge is only established when the estimated recoverable energy is greater than the control power consumption of the energy recovery circuit within one control cycle, to avoid invalid recovery operations where the recovered energy is less than the power consumption of the recovery circuit. For example, the minimum energy recovery threshold can be set to 0.1 millijoules to 1 millijoule. The attribute of this directed energy recovery edge includes the estimated recoverable energy value. The estimated value of recyclable energy is calculated as follows: the estimated value of recyclable energy equals the sum of the inductor energy and capacitor energy stored in the power supply circuit of the device corresponding to the buck operation node, multiplied by the energy recovery efficiency coefficient. The inductor energy is equal to half the inductance value of the power supply circuit multiplied by the square of the current before bucking. The capacitor energy is equal to half the difference between the square of the voltage before bucking and the square of the voltage after bucking, multiplied by the capacitance value of the power supply circuit. The energy recovery efficiency coefficient is the end-to-end conversion efficiency of energy released from the buck channel, transferred through the intermediate energy storage capacitor, and then to the boost channel. It is determined based on the measured conversion efficiency of the cascaded energy recovery circuit and the boost channel. For example, the energy recovery efficiency coefficient can be set to 0.4 to 0.7. When there are no matching buck and boost operation nodes that meet the conditions within the same control cycle, the energy released by the buck operation is directly fed back to the power supply battery or intermediate energy storage capacitor through the energy recovery circuit, without establishing a directed edge for energy recovery. Between adjacent control cycles, a time-dependent directed edge is added from the preceding control cycle node to the following control cycle node for the same core device, indicating the timing execution order of the device's power supply operation. Temporally dependent directed edges do not carry energy transfer properties.

[0149] S334: Perform topology sorting on the voltage regulation control and energy recovery command topology diagram to determine the execution priority sequence of all nodes. The topology sorting determines the execution order of nodes according to the following rules: First, sort by control cycle number from smallest to largest to ensure the correctness of the time sequence; within the same control cycle, buck operation nodes with negative voltage jump variables are placed before boost operation nodes with positive voltage jump variables, and maintenance operation nodes with zero voltage jump variables are placed after boost operation nodes. This sorting rule ensures that buck operations are executed before boost operations in each control cycle, so that the energy released during the buck process can be transferred to the device that needs boosting through the directed edge of energy recovery, realizing energy allocation between devices. The execution priority sequence is an ordered list of all nodes in the voltage regulation control and energy recovery command topology diagram arranged according to the topology sorting result.

[0150] See Figure 5 This is a topology diagram of voltage regulation control and energy recovery commands provided in the embodiments of this application. For example... Figure 5 As shown, the topology is a directed acyclic graph (DAG). Horizontally, it is arranged according to control cycles T, T+1, and T+2; vertically, it is arranged according to the three core devices: gyroscope, accelerometer, and navigation computer. A total of nine rectangular nodes form the node set. Each node is labeled with the voltage regulation operation type and voltage jump variable of the device within the corresponding control cycle. For example, in cycle T, the gyroscope node is labeled "Buck -0.2V," the accelerometer node is labeled "Boost +0.1V," and the navigation computer node is labeled "Maintain." There are two types of directed edges between nodes: solid arrows represent time-dependent directed edges, connecting nodes of the same device between adjacent control cycles, indicating the timing sequence of power supply operations; dashed arrows represent energy recovery directed edges, connecting nodes performing buck operations and nodes performing boost operations within the same control cycle, indicating that the transient energy released by the inductors and capacitors in the device's power supply circuit during bucking can be directionally transferred to the device requiring boosting. For example, in period T, the gyroscope's buck node points to the accelerometer's boost node via an energy recovery directed edge; in period T+1, the accelerometer's buck node points to the navigation computer's boost node via the same energy recovery directed edge. In traditional inertial navigation power supply systems, the power supply channels for each device are independent. When a device bucks, the energy stored in the inductors and capacitors in the power supply circuit is dissipated as heat and wasted. At the same time, another device may be boosting and drawing additional power from the battery, resulting in a double energy loss. This topology graph establishes a connection between the buck and boost operations of different devices within the same control cycle through a directed acyclic graph structure. Topological sorting ensures that the buck operation precedes the boost operation, allowing the energy released during bucking to be directionally fed back to the device requiring boosting through the energy recovery circuit. This achieves energy redistribution between different power supply channels within the inertial navigation system, improving the effective utilization rate of battery energy at the power supply circuit level.

[0151] Specifically, the construction of the voltage regulation control and energy recovery command topology diagram in step S33 is the core innovation of this invention in the field of power supply optimization. Existing inertial navigation systems' power supply modules supply power to each device independently, with no energy interaction between power supply channels. When one device steps down, the inductor and capacitor energy released is dissipated as heat, while at the same time, another device may be boosting and drawing power from the battery, resulting in double energy waste. The voltage regulation control and energy recovery command topology diagram, through a directed acyclic graph structure, establishes a connection between the buck and boost operations of different devices within the same control cycle, allowing the energy released during bucking to be directionally transferred to the device requiring boosting, forming an energy redistribution channel between devices. Introducing a minimum energy recovery threshold ensures that the recovery operation is only executed when the recovered energy is greater than the power consumption of the recovery circuit itself, avoiding the problem of increasing system power consumption due to minute energy recovery. The topological ordering of the directed acyclic graph guarantees the timing constraint that the buck operation executes before the boost operation, ensuring the feasibility of the energy recovery operation from a data structure perspective. The voltage regulation control and energy recovery command topology diagram, as the core variable passed from step S3 to step S4, encodes the voltage regulation operation sequence of all core devices, the energy transfer paths between devices, and the estimated recoverable energy in the form of a graph structure. This provides a complete operation plan for the cycle-by-cycle execution in step S4. Without the global planning of the voltage regulation control and energy recovery command topology diagram, step S4 would only be able to execute voltage regulation operations independently for each device, making it impossible to achieve energy recovery and reuse between devices. The overall energy utilization rate of the power supply system would be significantly lower than that of the present invention.

[0152] S4: Read the voltage regulation control and energy recovery instruction topology diagram, drive the wide voltage adaptive power supply chip and energy recovery circuit to perform cycle-by-cycle voltage regulation power supply operation and energy recovery operation according to the execution priority sequence of the topology sorting, and at the same time feed the execution results back to the inertial navigation main controller for navigation accuracy verification and power supply strategy closed-loop correction.

[0153] Further, step S4 includes:

[0154] S41: According to the execution priority sequence determined by the topological sorting in the voltage regulation control and energy recovery command topology diagram, within each control cycle, all nodes and their attributes corresponding to the current control cycle are read sequentially. The reading process involves filtering all nodes whose control cycle number is equal to the current control cycle number from the execution priority sequence and taking them out sequentially according to the arrangement order in the execution priority sequence.

[0155] S42: Perform voltage reduction and energy recovery operations on nodes with negative voltage jump variables in the current control cycle;

[0156] Further, step S42 includes:

[0157] S421: Based on the device identifier of the node, locate the corresponding power supply channel, drive the wide-voltage adaptive power supply chip on that channel to perform a step-down operation, adjusting the power supply voltage of the corresponding core device from the current voltage to the target output voltage value in the node's attributes. The wide-voltage adaptive power supply chip is a DC-DC converter with a wide input voltage range and programmable output voltage. Its output voltage is set in real-time by the power supply optimization control system via a digital control interface. During the step-down process, the wide-voltage adaptive power supply chip gradually reduces the output voltage from the current value to the target output voltage value by adjusting the duty cycle of its internal power switches. The rate of the step-down process is constrained by the power transition slope corresponding to the node.

[0158] S422: During the transient process of buck operation, check whether there is a directed edge for energy recovery originating from the buck operation node in the voltage regulation control and energy recovery command topology diagram. If a directed edge for energy recovery exists, activate the energy recovery circuit on the power supply channel. The energy recovery circuit consists of an energy storage inductor, an energy storage capacitor, a rectifier diode, and a feedback charging control switch. When the wide-voltage adaptive power supply chip reduces the output voltage, the energy stored in the inductor and capacitor in the device's power supply circuit needs to be released due to the voltage reduction. The energy recovery circuit converts the transient current released by the inductor and the transient current discharged by the capacitor into a unidirectional DC feedback current through the rectifier diode. The feedback charging control switch is controlled by the power supply optimization control system. During the buck transient process, the feedback charging control switch is closed to guide the DC feedback current to the intermediate energy storage capacitor in the power supply module. If there is no directed edge for energy recovery, the energy released during buck operation is fed back to the power supply battery through the energy recovery circuit. Record the actual energy value recovered by the energy recovery circuit during this buck operation as the actual recovered energy value. The actual recovered energy value is calculated by integrating the feedback current through the current sensor in the energy recovery circuit.

[0159] S423: If there is a directed edge for energy recovery, the actual recovered energy value is written into the attribute of the directed edge for energy recovery, replacing the original estimated recoverable energy value, for the boost operation node pointed to by the directed edge for reading and use in step S43.

[0160] S43: Perform a boost operation on nodes where the voltage jump variable is positive in the current control cycle;

[0161] Further, step S43 includes:

[0162] S431: Check if there is a directed edge for energy recovery pointing to the boost operation node in the voltage regulation control and energy recovery command topology diagram. If there is a directed edge for energy recovery, read the actual recovered energy value from the directed edge. Calculate the total boost energy required for the boost operation based on the target output voltage value and the current voltage of the boost operation node. The total boost energy is calculated as follows: the total boost energy equals half the difference between the square of the target output voltage value and the square of the current voltage, multiplied by the capacitance value in the power supply circuit of the core device, plus the energy required for the inductor to build up current during the boost transition. Compare the actual recovered energy value with the total boost energy: if the actual recovered energy value is greater than or equal to the total boost energy, the entire boost operation is completed using recovered energy, without drawing power from the power supply battery; if the actual recovered energy value is less than the total boost energy, the recovered energy is used to provide part of the boost energy, and the difference is supplemented from the power supply battery. If there is no directed edge for energy recovery pointing to the boost operation node, the entire boost energy is obtained from the power supply battery.

[0163] S432: Drives the wide-voltage adaptive power supply chip on this power supply channel to perform a boost operation, adjusting the supply voltage of the corresponding core device from the current voltage to the target output voltage value in the node attributes. During the boost process, when there is an energy recovery directed edge, the input terminal of the wide-voltage adaptive power supply chip preferentially obtains energy from the intermediate energy storage capacitor. When the recovered energy in the intermediate energy storage capacitor is exhausted, it switches to obtaining energy from the power supply battery.

[0164] S44: For nodes where the voltage jump variable is zero within the current control cycle, maintain the output voltage of the wide-voltage adaptive power supply chip unchanged, and only check whether the deviation between the supply voltage and the target output voltage value is within the preset voltage stability tolerance range. The voltage stability tolerance range is determined based on the allowable deviation range of the supply voltage of the core device. For example, the voltage stability tolerance range can be set to ±2% of the target output voltage value. If the deviation between the supply voltage and the target output voltage value exceeds the voltage stability tolerance range, drive the wide-voltage adaptive power supply chip to perform fine-tuning compensation to adjust the supply voltage back to the voltage stability tolerance range.

[0165] S45: Summarize the actual supply voltage, actual supply power, and actual recovered energy value of all core devices after the completion of the current control cycle into the current cycle power supply execution result. The data structure of the current cycle power supply execution result is a fixed-length record, including the timestamp of the current control cycle, the actual supply voltage field and actual supply power field corresponding to each core device in the device list, and the sum field of the actual recovered energy value generated by all step-down operation nodes in this control cycle. The current cycle power supply execution result is fed back to the inertial navigation main controller through the data communication interface between the power supply optimization control system and the inertial navigation main controller.

[0166] S46: After receiving the power supply execution result of the current cycle, the inertial navigation main controller performs navigation accuracy verification and power supply strategy closed-loop correction.

[0167] Further, step S46 includes:

[0168] S461: The inertial navigation main controller determines the sampling frequency corresponding to each core device under the current actual power supply based on the actual power supply power of each core device in the current cycle power supply execution result. For core devices with a false dynamic frequency modulation flag in the power grading configuration table, the sampling frequency corresponding to the mode closest to the actual power supply power is selected by matching the actual power supply power with the power supply power values ​​of each mode in the power grading configuration table. For core devices with a true dynamic frequency modulation flag in the power grading configuration table, the current actual operating frequency is looked up in the frequency-power mapping table of the navigation computer based on the actual power supply, and the effective sampling processing capability is determined in combination with the current computing load rate of the navigation computer. The inertial navigation main controller evaluates the positioning accuracy index and attitude accuracy index of the current navigation calculation output based on the sampling frequency of each core device and the sensor output data in the most recent navigation calculation cycle. The positioning accuracy index is characterized by the root mean square error between the position solution output by the navigation calculation and the position reference value of the external auxiliary positioning source. The attitude accuracy index is characterized by the maximum deviation angle between the attitude angle output by the navigation calculation and the attitude reference value. When the inertial navigation system cannot acquire an external auxiliary positioning source, the positioning accuracy index is estimated using the square root of the diagonal elements of the inertial navigation system's own covariance matrix. The positioning accuracy index is compared with a preset lower limit for positioning accuracy, and the attitude accuracy index is compared with a preset lower limit for attitude accuracy. These lower limits are determined based on the accuracy requirements of the current navigation task and are provided by the accuracy requirement configuration table corresponding to the task priority parameters.

[0169] S462: If the positioning accuracy index is lower than the lower limit of positioning accuracy or the attitude accuracy index is lower than the lower limit of attitude accuracy, it is determined that the current power supply strategy is causing insufficient navigation accuracy. The inertial navigation main controller sends an accuracy protection command to the state discrimination step in step S1. The function of the accuracy protection command is to forcibly upgrade the current carrier motion state classification label by one level, that is, upgrade the standby mode label to the normal working mode label, upgrade the normal working mode label to the core working mode label, and keep the core working mode label unchanged. After the carrier motion state classification label is upgraded, step S21 will re-query the power classification configuration table to obtain the power supply power value and sampling frequency corresponding to the higher power level, triggering step S22 to generate a higher power real-time power command sequence to ensure that the navigation accuracy is restored to above the lower limit of positioning accuracy and attitude accuracy. The triggering condition of the accuracy protection command has the characteristic of higher priority than the motion state discrimination result in step S13, that is, when the accuracy protection command is effective, even if the discrimination result of step S13 is the standby mode label, the system still performs power supply scheduling according to the forced upgrade of the normal working mode label. The accuracy protection command is effective for a preset accuracy recovery observation period. During this period, the system operates according to the upgraded carrier motion state classification label. After the accuracy recovery observation period ends, the accuracy protection command automatically expires, and the system recovers the carrier motion state classification label determined by the judgment result in step S13. However, this recovery process must follow the state switching hysteresis protection mechanism in step S133. That is, when switching from the higher power mode forcibly set by the accuracy protection command to the lower power mode judged in step S13, the switching must be performed only after confirmation for a normal hysteresis confirmation period. The accuracy recovery observation period is determined based on the typical time required for the inertial navigation system to recover from a low-precision state to normal accuracy. For example, the accuracy recovery observation period can be set to 5 seconds to 20 seconds.

[0170] S463: If the positioning accuracy index is higher than or equal to the lower limit of positioning accuracy and the attitude accuracy index is higher than or equal to the lower limit of attitude accuracy, then the current power supply strategy is determined to meet the navigation accuracy requirements, and the current power supply strategy remains unchanged. Simultaneously, the sum of the actual recovered energy values ​​in the current cycle's power supply execution results is added to the system's cumulative energy recovery value. The cumulative system energy recovery value is a global counter recording the total energy recovered by the energy recovery circuit since the system started. The remaining available energy in the battery's state of charge data is updated based on the sum of the actual recovered energy values. The updated remaining available energy equals the remaining available energy of the previous cycle minus the product of the current cycle's total system power consumption and control cycle duration, plus the sum of the actual recovered energy values. ,in The updated remaining available energy. This represents the remaining usable energy from the previous cycle. This represents the total system power consumption for the current cycle. To control the cycle duration, This represents the total actual energy recovered in the current cycle. The updated remaining available energy is used in step S321 of the next cycle to calculate the expected remaining runtime, forming a closed loop for energy management of the power supply strategy.

[0171] Specifically, step S4 executes the operation instructions in the voltage regulation control and energy recovery instruction topology diagram cycle by cycle, transforming the forward-looking power scheduling plan generated in step S3 into actual hardware power supply control actions. Through navigation accuracy verification and power supply strategy closed-loop correction mechanisms, a dynamic balance is established between power consumption optimization and navigation accuracy assurance. The accuracy protection instruction mechanism in step S46 ensures that reducing power consumption does not come at the expense of navigation accuracy. When the accuracy index reaches the lower limit, the system automatically increases the power level to restore accuracy, reflecting the design principle that navigation accuracy takes precedence over power consumption optimization. When reverting to a lower power mode after the accuracy protection instruction fails, a hysteresis confirmation period is required to prevent oscillations caused by a sudden drop in power leading to insufficient accuracy again. The operation in step S463, which updates the remaining available energy based on the actual recovered energy value, enables the energy budget balancing algorithm in step S3 to make power scheduling decisions based on more accurate battery state data in the next cycle, forming a complete closed loop from power scheduling to hardware execution to energy state update. This closed-loop structure enables the power supply optimization system to adaptively adjust the power supply strategy based on the real-time consumption and recovery of battery energy, avoiding the risks of excessive energy saving or energy depletion caused by energy estimation errors in traditional open-loop control methods. The voltage regulation control and energy recovery command topology is executed cycle by cycle in step S4. By arranging the buck operation before the boost operation, energy reuse between different power supply channels within the inertial navigation system is realized. The reactive energy that was originally dissipated as heat during the buck transient is converted into usable power supply energy, improving the effective utilization rate of battery energy at the power supply circuit level. Such energy allocation between devices is impossible to achieve in a constant full-power power supply mode.

[0172] Example 2:

[0173] This embodiment, based on Embodiment 1, provides an optimized low-power power supply system for inertial navigation, such as... Figure 6 As shown, it includes:

[0174] Data acquisition and state discrimination module: Acquires raw angular velocity sampling data and raw acceleration sampling data from gyroscopes and accelerometers in the inertial navigation system, performs multi-scale motion feature extraction and state discrimination, obtains carrier motion state classification labels, and constructs a power classification configuration table;

[0175] Time-sharing power supply scheduling module: Based on the power classification configuration table, it performs time-sharing on-demand power supply scheduling for each core component of the inertial navigation system, generates a real-time power command sequence for each core component, and constructs a circular buffer queue for the power status of the components.

[0176] Energy budget and scheduling planning module: reads the power status ring cache queue of the device, extracts historical power change trends, combines the state of charge of the power supply battery to generate a forward-looking power scheduling plan, and constructs a voltage regulation control and energy recovery instruction topology diagram;

[0177] Voltage regulation execution and closed-loop correction module: reads the voltage regulation control and energy recovery instruction topology diagram, drives the wide voltage adaptive power supply chip and energy recovery circuit to perform operations according to the topology sorting priority, and feeds back the execution results to the inertial navigation main controller for closed-loop correction.

Claims

1. A low-power power supply optimization method for inertial navigation, characterized in that, The method includes: Raw angular velocity and raw acceleration sampling data from gyroscopes and accelerometers in the inertial navigation system are collected. Multi-scale motion feature extraction and state discrimination are performed on the raw angular velocity and raw acceleration sampling data to obtain a carrier motion state classification label. A power classification configuration table is constructed based on the carrier motion state classification label. Based on the power classification configuration table, time-sharing on-demand power supply scheduling is performed on each core component of the inertial navigation system to generate a real-time power command sequence for each core component, and a circular buffer queue of device power status is constructed based on the real-time power command sequence. The device power status circular buffer queue is read, the historical power change trend of each core device is extracted, and combined with the state of charge data of the power supply battery, a forward-looking power scheduling plan is generated through the energy budget balancing algorithm. Based on the forward-looking power scheduling plan, a voltage regulation control and energy recovery instruction topology is constructed. The voltage regulation control and energy recovery command topology diagram is read, and the wide-voltage adaptive power supply chip and energy recovery circuit are driven to perform cycle-by-cycle voltage regulation power supply operation and energy recovery operation according to the execution priority sequence of the topology sorting. At the same time, the execution results are fed back to the inertial navigation main controller for navigation accuracy verification and power supply strategy closed-loop correction.

2. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The multi-scale motion feature extraction includes: The root mean square value of angular velocity and the peak value of angular velocity change rate are calculated for the raw angular velocity sampling data of the three axes within a preset sliding time window. The root mean square value of acceleration and the peak value of acceleration change rate are calculated for the raw acceleration sampling data of the three axes within the same sliding time window. Short-time energy spectral density is calculated for the original angular velocity sampling data and original acceleration sampling data within the sliding time window, and angular velocity energy concentration is extracted. The angular velocity energy concentration is the ratio of the power spectral density integral value within the preset motion-related frequency band to the power spectral density integral value across the entire frequency band. The root mean square value of angular velocity, peak value of angular velocity change rate, root mean square value of acceleration, peak value of acceleration change rate, and angular velocity energy concentration are combined to form a set of carrier motion characteristic parameters.

3. The inertial navigation low-power power supply optimization method according to claim 2, characterized in that, The state determination includes: Based on the set of carrier motion characteristic parameters and combined with the task priority parameters, the carrier motion state is determined sequentially in descending order of priority. When the root mean square value of angular velocity is greater than or equal to the first angular velocity threshold, or the root mean square value of acceleration is greater than or equal to the first acceleration threshold, or the task priority parameter is marked as a precise positioning task identifier, it is determined to be in core working mode. When the root mean square value of angular velocity is less than the second angular velocity threshold, the root mean square value of acceleration is less than the second acceleration threshold, the peak value of the rate of change of angular velocity is less than the first rate of change of angular velocity threshold, the peak value of the rate of change of acceleration is less than the first rate of change of acceleration threshold, and the angular velocity energy concentration is less than the first energy concentration threshold, it is determined to be in standby mode, wherein the first angular velocity threshold is greater than the second angular velocity threshold. If the conditions for determining the core working mode and the conditions for determining the standby mode are not met, it is determined to be the normal working mode.

4. The inertial navigation low-power power supply optimization method according to claim 3, characterized in that, The state determination also includes a state transition hysteresis protection mechanism: When a parameter in the carrier motion characteristic parameter set crosses the boundary from one mode determination interval to an adjacent mode determination interval, the carrier motion state classification label is only updated after the parameter has been continuously kept in the boundary interval for a duration exceeding the preset hysteresis confirmation duration. When switching from low power mode to high power mode, a fast response hysteresis confirmation time is used, and when switching from high power mode to low power mode, a normal hysteresis confirmation time is used. The fast response hysteresis confirmation time is shorter than the normal hysteresis confirmation time.

5. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The construction of the power tier configuration table includes: Establish a component list for each core component in the inertial navigation system, including gyroscopes, accelerometers, and navigation computers; For each core device, set its rated power supply value under the core operating mode label, its downgraded power supply value under the normal operating mode label, its minimum hold power supply value under the standby mode label, as well as its corresponding rated sampling frequency, downgraded sampling frequency, and intermittent sampling frequency. The power supply power value and sampling frequency of each core device under each carrier motion state classification label are arranged by row as core device and by column as carrier motion state classification label to construct the power classification configuration table.

6. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The time-sharing on-demand power supply scheduling includes: For gyroscopes, when the carrier motion state classification label is the normal working mode label, differentiated control is performed according to the device type of the gyroscope. If the gyroscope type is a MEMS gyroscope, the power supply value instruction is output according to the down-frequency sampling frequency. If the gyroscope type is a fiber optic gyroscope or laser gyroscope, the power supply of redundant detection channels is turned off and only the power supply of the core measurement channel is retained. For the navigation computer, dynamic frequency modulation power supply control is adopted to obtain the current computing load rate of the navigation computer's processor in real time. When the carrier motion status classification label is the normal working mode label, a stepped frequency modulation strategy is adopted to divide the frequency range between the minimum working frequency and the rated working frequency into N frequency levels. The corresponding frequency level is selected as the target working frequency according to the interval where the computing load rate is located. Frequency gear switching is only performed when the computing load rate remains within the same range for M consecutive control cycles.

7. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The construction of the device power state circular buffer queue includes: In the memory of the power supply optimization control system, a circular cache space of fixed length L is allocated, and the write pointer and the written counter are initialized. After generating the real-time power instruction sequence in each control cycle, the current target power supply value of each core device, the current carrier motion state classification label and the current timestamp are encapsulated into a power state record node and written into the circular buffer space at the position pointed to by the write pointer. When the write pointer reaches the end of the circular buffer space, it wraps back to the beginning position to overwrite the earliest power state record node. The device power state circular buffer queue stores the power state record nodes of the most recent L control cycles.

8. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The extraction of the historical power change trend includes: The power status record nodes in the device power status circular buffer queue are grouped by device identifier to obtain the time series of the target power supply value of each core device. Linear regression fitting is performed on the time series of the target power supply value of each core device to calculate the power change slope and power fluctuation standard deviation, which are used as descriptive parameters of the historical power change trend of the core device. The overall power consumption rate of the system is obtained by summing the power change slopes of all core components.

9. The inertial navigation low-power power supply optimization method according to claim 8, characterized in that, The energy budget equalization algorithm includes: The remaining available energy is calculated based on the current state of charge percentage and nominal capacity of the power supply battery, and the estimated remaining driving time is estimated by combining the overall power consumption rate of the system and the current total power of the system. The estimated remaining battery life is compared with a preset safety multiple of the remaining mission time. If the estimated remaining battery life is less than the preset safety multiple of the remaining mission time, the target average power is calculated based on the remaining mission time and the preset safety multiple, and the total power to be reduced is calculated. Based on the power fluctuation standard deviation of each core device, the core devices are sorted in descending order of power fluctuation standard deviation, and power reduction is allocated to each core device in turn. The power reduction of each core device shall not exceed the difference between its current target power supply value and the minimum hold power supply value.

10. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The voltage regulation control and energy recovery command topology is a directed acyclic graph, and its construction includes: Each node in the node set represents a power supply operation for a core device within a control cycle. Each node includes a device identifier, control cycle number, target output voltage value, and voltage jump variable. Within the same control cycle, if the voltage jump variable of one core device is negative and its estimated recoverable energy is greater than the preset minimum energy recovery threshold, and the voltage jump variable of another core device is positive, then an energy recovery directed edge is added from the buck operation node to the boost operation node. The voltage regulation control and energy recovery command topology graph is sorted, and the buck operation node is executed before the boost operation node within the same control cycle, so that the energy released during the buck process is transferred to the device that needs to be boosted through the directed edge of the energy recovery.

11. The inertial navigation low-power power supply optimization method according to claim 1, characterized in that, The navigation accuracy verification and power supply strategy closed-loop correction include: The inertial navigation main controller evaluates the positioning accuracy and attitude accuracy indicators of the current navigation solution output based on the actual power supply of each core component; If the positioning accuracy index is lower than the lower limit of positioning accuracy or the attitude accuracy index is lower than the lower limit of attitude accuracy, an accuracy protection command is sent to the state discrimination stage to forcibly upgrade the current carrier motion state classification label by one level. The duration of the accuracy protection command is the preset accuracy recovery observation period. After the accuracy recovery observation period ends, the accuracy protection command automatically becomes invalid. When switching from the higher power mode forcibly set by the accuracy protection command to the lower power mode, the switching is only performed after confirmation of the normal hysteresis confirmation period.

12. An inertial navigation low-power power supply optimization system, used to implement the inertial navigation low-power power supply optimization method according to any one of claims 1-11, characterized in that, The system includes: Data acquisition and state discrimination module: used to acquire raw angular velocity sampling data and raw acceleration sampling data of gyroscope and accelerometer in inertial navigation system, perform multi-scale motion feature extraction and state discrimination on the raw angular velocity sampling data and raw acceleration sampling data, obtain carrier motion state classification labels, and construct power classification configuration table based on the carrier motion state classification labels; Time-sharing power supply scheduling module: Based on the power classification configuration table, it performs time-sharing on-demand power supply scheduling for each core component of the inertial navigation system, generates a real-time power command sequence for each core component, and constructs a circular buffer queue of device power status based on the real-time power command sequence; Energy Budget and Scheduling Planning Module: This module is used to read the power status circular buffer queue of the devices, extract the historical power change trends of each core device, combine the state of charge data of the power supply battery, generate a forward-looking power scheduling plan through the energy budget balancing algorithm, and construct a voltage regulation control and energy recovery instruction topology based on the forward-looking power scheduling plan. Voltage regulation execution and closed-loop correction module: It is used to read the voltage regulation control and energy recovery command topology diagram, drive the wide voltage adaptive power supply chip and energy recovery circuit to perform cycle-by-cycle voltage regulation power supply operation and energy recovery operation according to the execution priority sequence of the topology sorting, and at the same time feed back the execution results to the inertial navigation main controller for navigation accuracy verification and power supply strategy closed-loop correction.

Images