Intelligent fusion terminal power consumption dynamic adjustment method and system based on environment perception

Abstract

The application provides an environment perception-based intelligent fusion terminal power consumption dynamic adjustment method and system, and relates to the technical field of intelligent fusion terminals.The method comprises the following steps: connecting an environment sensor module to collect environment information; determining a current environment information entropy level based on a time domain variation rate and an amplitude threshold of an original environment perception signal; calling an impact event detection unit to identify an impact event; inputting the current environment information entropy level and an impact event trigger flag into an asymmetric adjustment control unit to control the fusion terminal to perform stepwise migration between multiple predefined power consumption levels, and the power consumption levels change step by step or by leaps and bounds according to discrete steps. The application solves the technical problem that the power consumption adjustment strategy in the prior art has excessive response or insufficient response due to the single perception dimension and lack of hierarchical division of the environment situation, and ensures the response capability of the intelligent fusion terminal by introducing environment information entropy and performing stepwise migration based on the asymmetric adjustment control unit.

Description

Technical Field

[0001] This application relates to the field of intelligent converged terminal technology, and in particular to a method and system for dynamic power consumption adjustment of intelligent converged terminals based on environmental perception. Background Technology

[0002] In the field of intelligent converged terminals, existing dynamic power consumption adjustment methods mostly rely on single-dimensional static threshold control, typically employing continuous linear frequency modulation strategies, which are ill-suited to cope with complex and ever-changing external environments. In practical applications, when the terminal is in a stable but high-information-entropy environment, existing methods may frequently trigger power consumption level increases and decreases due to load fluctuations, leading to unnecessary mode switching overhead. Conversely, in the event of a brief, strong impact in a low-entropy context, the response may be delayed due to the statistical window averaging effect, failing to promptly improve performance to ensure task real-time performance. Furthermore, existing adjustment strategies generally employ symmetrical increase and decrease paths, meaning that performance improvement and recovery reduction use the same step size and speed, lacking asymmetric management of the system state after the impact event, which easily leads to repeated oscillations in power consumption levels.

[0003] In summary, existing technologies suffer from a problem where the perception of environmental conditions is limited by a single dimension and lacks hierarchical differentiation, leading to over- or under-response in power consumption regulation strategies. This further impacts the overall energy efficiency of intelligent converged terminals in real, complex environments. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for dynamic power consumption adjustment of intelligent fusion terminals based on environmental perception, in order to solve the technical problem in the prior art that the power consumption adjustment strategy over-responds or under-responds due to the single perception dimension of the environmental situation and the lack of hierarchical differentiation, which further affects the overall energy efficiency ratio of intelligent fusion terminals in real complex environments.

[0005] In view of the above problems, this application provides a method and system for dynamic power consumption adjustment of intelligent fusion terminals based on environmental perception.

[0006] Firstly, this application provides a method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception. This method is implemented through an intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception. The method includes: connecting an environmental sensor module to collect environmental information and outputting a raw environmental perception signal; receiving the raw environmental perception signal through an information entropy calculation unit deployed within a processor, and determining the current environmental information entropy level based on the time-domain change rate and amplitude threshold of the raw environmental perception signal; calling an impact event detection unit to identify impact events and generate an impact event trigger flag based on the change rate of the raw environmental perception signal; inputting the current environmental information entropy level and the impact event trigger flag into an asymmetric adjustment control unit, and controlling the fusion terminal to perform a stepped migration between multiple predefined power consumption levels according to the combination of the environmental information entropy level and the existence state of the impact event. The stepped migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

[0007] Optionally, the average rate of change of the original environmental sensing signal within the time window and the statistical mean of the signal amplitude are calculated to obtain the average rate of change and the statistical mean of the amplitude; when the average rate of change is lower than the first rate threshold and the statistical mean of the amplitude is lower than the first amplitude threshold, it is determined to be a low information entropy level, otherwise it is determined to be a high information entropy level.

[0008] Optionally, the average rate of change is the integral of the absolute value of the first derivative of the original environmental sensing signal over the time window divided by the window length; both the first rate threshold and the first amplitude threshold are dynamically adjustable, and the adjustment is based on: the terminal's current remaining battery power and historical statistical values ​​of environmental fluctuations within a preset time period.

[0009] Optionally, the derivative of information entropy with respect to time is calculated to obtain the entropy change rate. If the entropy change rate is negative and its absolute value exceeds the first derivative threshold, even if the average change rate and amplitude statistical mean at the current moment do not meet the low information entropy condition, the environmental state is marked as an entropy reduction warning state, and the entry conditions for the low information entropy level are relaxed. If the entropy change rate is positive and exceeds the second derivative threshold, it is marked as an entropy increase warning state, and the entry conditions for the low information entropy level are tightened. The way to relax or tighten the entry conditions includes: temporarily adjusting the values ​​of the first rate threshold and / or the first amplitude threshold, with the adjustment magnitude being positively correlated with the absolute value of the entropy change rate.

[0010] Optionally, the original environmental sensing signal is compared with multiple preset incremental impact rate thresholds to determine impact events of different intensities. When the rate of change first exceeds the impact threshold, a minimum dwell timer is started. Before the timer expires, the impact event trigger flag remains valid even if the rate of change falls back below the threshold. At the same time, a stabilization counter is started. If the rate of change is continuously lower than the fall-off threshold within a preset stabilization time window, the impact event trigger flag is canceled. Different impact weights are configured according to different signal types, and the weighted comprehensive impact intensity is calculated. The weighted impact intensity is positively correlated with the response amplitude of the asymmetric adjustment control.

[0011] Optionally, when an impact event trigger flag is present, the target power consumption level is determined based on the intensity of the impact event. The higher the impact intensity, the higher the target power consumption level. Among them, severe impact corresponds to the highest power consumption level and is not limited by the current information entropy level. When no impact event trigger flag is present, the baseline power consumption level is selected based on the environmental information entropy level. After the impact event trigger flag disappears, the power consumption is restored from the current power consumption level to the baseline level by gradually reducing the power consumption level. The dwell time at each level is not less than the preset minimum dwell time in seconds.

[0012] Optionally, the power consumption levels include at least: full-speed mode, for all peripherals and processors to operate at the highest frequency voltage; balanced mode, for some non-critical peripherals to be frequency-reduced or shut down; and preemptive sleep mode, for cutting off power to at least one high-power peripheral while retaining at least a low-power wake-up channel; wherein the preemptive sleep mode further includes: mild preemptive sleep and deep preemptive sleep, wherein mild preemptive sleep disables the image signal processor and AI accelerator while retaining the wireless communication module with a low heartbeat rate; and deep preemptive sleep disables the wireless communication module while retaining only the wake-up monitoring circuit and real-time clock.

[0013] Optionally, the asymmetric adjustment control unit dynamically adjusts the migration path according to the changing trend of the real-time information entropy level, including: when descending from full-speed mode to a lower level, if the current information entropy level is low and the entropy change rate is negative and the absolute value exceeds the preset shortcut threshold, skip the intermediate power consumption level and directly switch to the deprivation sleep mode; if the current information entropy level is low but the frequency of alternating positive and negative entropy change rates exceeds the fluctuation threshold, then a transient monitoring mode is forcibly inserted between the balanced mode and the deprivation sleep mode. In the transient monitoring mode, only the key sensors are maintained to work at the first sampling rate, while other high-power peripherals are kept off, and the first sampling rate is higher than the second sampling rate in the deprivation sleep mode.

[0014] Optionally, the extra energy consumed in each impact event is recorded and accumulated to determine the energy debt balance. The extra energy is calculated based on the impact event's intensity, duration, and the difference between the actual power consumption and the baseline power consumption during the impact. During the gradual reduction of power consumption after the impact event ends, the dwell time at each level is dynamically extended based on the energy debt balance. The higher the energy debt balance, the longer the dwell time at each level. After entering the preemptive sleep mode, the system operates at a power level lower than the normal preemptive sleep power consumption to deduct the energy debt balance until the balance reaches zero, at which point the normal preemptive sleep power consumption is restored. Operating at a power level lower than the normal preemptive sleep power consumption means reducing the sampling rate or communication frequency of non-critical peripherals.

[0015] Secondly, this application also provides a dynamic power consumption adjustment system for an intelligent fusion terminal based on environmental perception, used to execute the dynamic power consumption adjustment method for an intelligent fusion terminal based on environmental perception as described in the first aspect. The system includes: an environmental information acquisition module for connecting to an environmental sensor module to acquire environmental information and outputting a raw environmental perception signal; an information entropy calculation module for receiving the raw environmental perception signal through an information entropy calculation unit deployed within a processor, and determining the current environmental information entropy level based on the time-domain change rate and amplitude threshold of the raw environmental perception signal; an impact event identification module for calling an impact event detection unit to identify impact events and generate impact event trigger flags based on the change rate of the raw environmental perception signal; and a power consumption adjustment control module for inputting the current environmental information entropy level and the impact event trigger flags into an asymmetric adjustment control unit, and controlling the fusion terminal to perform a stepped migration between multiple predefined power consumption levels according to the combination of the environmental information entropy level and the existence state of the impact event. The stepped migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

[0016] One or more technical solutions provided in this application have at least the following beneficial effects:

[0017] Environmental information is collected by connecting to an environmental sensor module, outputting raw environmental perception signals. These signals are received by an information entropy calculation unit deployed within the processor. Based on the time-domain change rate and amplitude threshold of the raw environmental perception signals, the current environmental information entropy level is determined. An impact event detection unit is invoked to identify impact events based on the change rate of the raw environmental perception signals, generating an impact event trigger flag. The current environmental information entropy level and the impact event trigger flag are input to an asymmetric adjustment control unit. Based on the combination of the environmental information entropy level and the existence state of the impact event, the fusion terminal is controlled to perform a stepped migration between multiple predefined power consumption levels. This stepped migration refers to the power consumption level changing stepwise or skipping steps in a discrete manner. In other words, by utilizing the dual characteristics of environmental information entropy and impact events, the power consumption level is guided to implement asymmetric stepped migration based on the asymmetric adjustment control unit. This ensures low-latency response to sudden events while effectively suppressing ineffective power consumption switching caused by environmental disturbances, thereby improving the overall energy efficiency ratio in real-world complex environments.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application 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 merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception, as described in this application.

[0021] Figure 2 This is a schematic diagram of the power consumption dynamic adjustment system for the intelligent fusion terminal based on environmental perception, as described in this application.

[0022] Figure labeling: 11 Environmental information acquisition module, 12 Information entropy calculation module, 13 Impact event recognition module, 14 Power consumption regulation and control module. Detailed Implementation

[0023] This application provides a method and system for dynamic power consumption adjustment of intelligent fusion terminals based on environmental perception. This addresses the technical problem in existing technologies where the single dimension and lack of hierarchical differentiation in environmental situation perception leads to over- or under-response in power consumption adjustment strategies, further impacting the overall energy efficiency of intelligent fusion terminals in real-world complex environments. By utilizing the dual characteristics of environmental information entropy and impact events, the method guides power consumption levels to implement asymmetric stepwise migration based on an asymmetric adjustment control unit. This ensures low-latency response to sudden events while effectively suppressing ineffective power consumption switching caused by environmental disturbances, thereby improving the overall energy efficiency in real-world complex environments.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides a method for dynamic power consumption adjustment of intelligent fusion terminals based on environmental perception. The method is executed by an intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception, and specifically includes the following steps: Connect to the environmental sensor module to collect environmental information and output raw environmental perception signals.

[0026] Specifically, after the intelligent fusion terminal is started, the environmental sensor module is initialized and configured through a general input / output interface, including setting the sampling rate, measurement range, resolution, and operating mode for each sensor. After configuration, the sensor module is powered on and begins data acquisition. The sensor module converts the sensed environmental physical quantities into electrical signals according to a set cycle, and then quantizes them into digital values ​​through an analog-to-digital converter. The processor periodically reads these values ​​from the sensor data registers and parses them according to the data format specified by the sensors. After removing device identifiers and status bits, a series of raw data points arranged in chronological order are obtained. The raw environmental sensing signal is a voltage, current, or digital quantization sequence directly output by the sensor module without any filtering, amplification, or feature extraction. It usually exists in the form of a continuous analog voltage waveform or discrete digital sampling points, such as the voltage value across a photoresistor, the pulse-code modulation stream output by a microphone, or a 16-bit integer value output by an accelerometer. The raw environmental sensing signal retains all the details of environmental changes, including noise and transient spikes.

[0027] The original environmental perception signal is received by an information entropy calculation unit deployed within the processor. Based on the time-domain change rate and amplitude threshold of the original environmental perception signal, the current environmental information entropy level is determined.

[0028] Furthermore, this application also includes the following steps: calculating the average rate of change of the original environmental perception signal within a time window and the statistical mean of the signal amplitude to obtain the average rate of change and the statistical mean of the amplitude; when the average rate of change is lower than a first rate threshold and the statistical mean of the amplitude is lower than a first amplitude threshold, it is determined to be a low information entropy level; otherwise, it is determined to be a high information entropy level.

[0029] Furthermore, this application also includes the following steps: the average rate of change is the integral of the absolute value of the first derivative of the original environmental sensing signal over the time window divided by the window length; both the first rate threshold and the first amplitude threshold are dynamically adjustable, and the adjustment basis includes: the current remaining battery power of the terminal and the historical statistical values ​​of environmental fluctuations within a preset time period.

[0030] Specifically, the information entropy calculation unit is a software functional module or fixed logic circuit deployed inside the processor. It is dedicated to extracting quantitative indicators reflecting the complexity and intensity of environmental changes from the raw environmental sensing signals. It does not alter the original signal data, but only performs mathematical transformations and statistical analysis on the signals. The information entropy calculation unit continuously reads the time-series data of the raw environmental sensing signals from the memory buffer and processes each sensor signal.

[0031] The information entropy calculation unit sets a fixed-length time window, the length of which can be pre-configured according to the terminal application scenario, such as 0.5s for rapidly changing environments and 5s for slowly changing environments. The window moves forward in a sliding manner, such as moving backward a fixed step size every 0.1s or after receiving a sampling point.

[0032] Within each time window, the information entropy calculation unit performs two parallel computation tasks. First, it calculates the first derivative of the original signal. For discrete sampling points, it approximates the derivative by dividing the difference between adjacent sample values ​​by the sampling interval. Then, it takes the absolute value of each derivative and sums the product of all absolute values ​​with the product of the sampling interval, effectively performing a time integral on the absolute derivative curve. The integral result is divided by the window length to obtain the average rate of change within that window. Finally, it directly sums the amplitudes (sensor readings) of all original sampling points within that window and divides this sum by the total number of sampling points to obtain the statistical mean of the amplitudes.

[0033] The system reads the dynamically adjusted first rate threshold and first amplitude threshold. Both the first rate threshold and the first amplitude threshold are dynamically adjustable, and the adjustment is based on factors including the terminal's current remaining battery power and historical environmental fluctuation statistics over a preset time period. For example, the system reads the remaining battery percentage from the battery management chip in real time; if the battery level is below 20%, the threshold is increased proportionally. At the same time, the system reviews the environmental fluctuation statistics over the past hour; if historical fluctuations are extremely large, the threshold is relaxed accordingly to avoid oversensitivity.

[0034] The first rate threshold is a dynamically adjustable benchmark value used to determine whether the average rate of change is low enough to indicate slow environmental change. When it is above the first rate threshold, it means frequent environmental fluctuations. The first amplitude threshold is a dynamically adjustable benchmark value used to determine whether the statistical mean of the amplitude is low enough to indicate that the overall environmental signal is weak. When it is below the first amplitude threshold, it means the environment is at a low energy level. If the average rate of change in the current window is less than the first rate threshold and the statistical mean of the amplitude is less than the first amplitude threshold, the environmental entropy level corresponding to the sensor is determined to be low; otherwise, it is determined to be high.

[0035] The information entropy calculation unit can weightedly combine the judgment results from different sensors. This process is repeated in each time window, continuously outputting an updated environmental information entropy level. The current environmental information entropy level indicates how much truly useful information is contained within these data.

[0036] For example, the terminal integrates a light sensor with a fixed sampling rate of 10Hz, collecting a value every 0.1s. The time window length of the information entropy calculation unit is set to 2.0s, meaning the window contains 20 sampling points. The first rate threshold is initially set to 15.0lx / s, and the first amplitude threshold is initially set to 50.0lx. To demonstrate dynamic adjustability, assuming the terminal currently has 23% remaining battery power, for more aggressive energy saving, the processor temporarily adjusts the first rate threshold to 20.0lx / s and the first amplitude threshold to 40.0lx. Within a 2.0s time window, the acquired raw light intensity sequence (in lx) is as follows: 32.5, 33.0, 32.8, 33.1, 32.9, 33.2, 33.0, 32.7, 33.1, 32.8, 33.0, 32.9, 33.1, 32.8, 33.0, 32.7, 33.2, 32.9, 33.1, 33.0. The absolute values ​​of the first derivatives (adjacent differences divided by the sampling interval of 0.1s) are calculated: the first difference is 5.0 lx / s, the second is 2.0 lx / s, the third is 3.0 lx / s, and so on. Due to the small fluctuations, the absolute values ​​of all derivatives are between 2.0 lx / s and 6.0 lx / s. Multiplying these absolute derivative values ​​by the sampling interval of 0.1s is equivalent to multiplying each absolute derivative by 0.1 and then summing them. Integrating (accumulating) yields an area under the absolute derivative curve of 8.5 lx. Dividing this by the window length of 2.0 s gives an average rate of change of 4.25 lx / s. Adding these 20 values ​​together gives 658.2 lx, which, divided by 20, yields 32.91 lx. Comparing this calculated value with the current dynamic threshold, the average rate of change of 4.25 lx / s is less than the first rate threshold of 20.0 lx / s, and the average amplitude statistical value of 32.91 lx is less than the first amplitude threshold of 40.0 lx. Both conditions are met, therefore the environmental information entropy level within this time window is determined to be low. If the terminal is in a different scenario, with an average rate of change of 85.0 lx / s within the time window (e.g., frequent switching of lights), or an average amplitude statistical value of 560.0 lx (e.g., a very bright environment), then it is determined to be a high information entropy level.

[0037] The information entropy calculation unit combines the time-domain change rate and amplitude statistical mean of the original environmental sensing signal to classify the environmental state into low entropy and high entropy, effectively filtering out instantaneous noise and occasional spikes, and avoiding the erroneous triggering of power consumption level switching due to brief abnormal fluctuations.

[0038] Furthermore, this application also includes the following steps: calculating the derivative of information entropy with respect to time to obtain the entropy change rate; if the entropy change rate is negative and its absolute value exceeds the first derivative threshold, even if the average change rate and amplitude statistical mean at the current moment do not meet the conditions for low information entropy, the environmental state is marked as an entropy reduction warning state, and the entry conditions for the low information entropy level are relaxed; if the entropy change rate is positive and exceeds the second derivative threshold, it is marked as an entropy increase warning state, and the entry conditions for the low information entropy level are tightened; wherein, the way to relax or tighten the entry conditions includes: temporarily adjusting the values ​​of the first rate threshold and / or the first amplitude threshold, the adjustment magnitude being positively correlated with the absolute value of the entropy change rate.

[0039] Specifically, the information entropy calculation unit needs to maintain a continuous information entropy value, not just the final high / low level. This is obtained by weighting and fusing the average rate of change and the statistical mean of the amplitude. For example, the entropy value equals the average rate of change divided by the rate reference value plus the average amplitude divided by the amplitude reference value, resulting in a dimensionless real number. The information entropy calculation unit records the current entropy value at fixed time intervals and divides the difference between the current and previous entropy values ​​by the time interval to obtain an approximate value of the entropy change rate.

[0040] The information entropy calculation unit compares the entropy change rate with two preset derivative thresholds. The first derivative threshold is used to detect negative changes, and the second derivative threshold is used to detect positive changes. The first derivative threshold is a preset positive value used to determine whether the absolute value of the entropy change rate reaches a level worthy of triggering an alert. The first derivative threshold is independent of the aforementioned rate threshold and amplitude threshold, and is usually set as an empirical value, such as an entropy change of 0.3 or more per second being considered significant.

[0041] If a negative entropy change rate is detected, and its absolute value is greater than the first derivative threshold, the unit triggers an entropy reduction warning state. Even if the average change rate and amplitude statistical mean calculated in the current window do not yet meet the low entropy condition of being below their respective thresholds, the environment is still considered to be rapidly stabilizing, therefore the entry conditions for the low entropy level need to be actively relaxed. The specific relaxation operation is to temporarily increase the first rate threshold and the first amplitude threshold. The increase is proportional to the absolute value of the entropy change rate, such as increase = proportional coefficient × |entropy change rate|. The increased thresholds will be used for low entropy judgment in subsequent windows until the warning state is lifted. Relaxing the entry conditions for the low information entropy level temporarily reduces the strictness of the judgment, making it easier for the terminal to be judged as having a low information entropy level, including increasing the first rate threshold and increasing the first amplitude threshold.

[0042] If a positive entropy change rate is detected, and its value exceeds the second derivative threshold, the unit triggers an entropy increase warning state. At this time, the environment is rapidly deteriorating, and the low-entropy entry conditions should be tightened in advance to avoid prematurely classifying the terminal as low-entropy and causing it to be unable to cope with the upcoming high load after frequency reduction. Tightening the conditions involves temporarily lowering the first rate threshold and the first amplitude threshold; the amount of reduction is also positively correlated with the absolute value of the entropy change rate. Tightening the entry conditions for low information entropy levels is the opposite of relaxing them; it temporarily increases the strictness of the judgment, making it more difficult for the terminal to be classified as low information entropy. Specific methods include lowering the first rate threshold and the first amplitude threshold. The magnitude of the threshold change is proportional to the absolute value of the entropy change rate. The larger the absolute value of the entropy change rate, the greater the temporary adjustment of the threshold, thus producing a stronger warning effect.

[0043] The warning state is typically maintained for a fixed time window, or until the entropy change rate falls below a threshold. Once the warning disappears, the temporarily adjusted threshold gradually reverts to its original base value. By calculating the entropy change rate and introducing entropy decrease and increase warning mechanisms, the information entropy calculation unit gains the ability to predict environmental trends, rather than simply sensing static states. When the environment is rapidly stabilizing, the low-entropy entry conditions are relaxed in advance, allowing the terminal to switch to low-power mode earlier, thus saving more energy. When the environment is rapidly deteriorating, the low-entropy entry conditions are tightened in advance to prevent the terminal from prematurely reducing performance, which could lead to delayed response or timeouts in subsequent tasks.

[0044] The impact event detection unit is invoked to identify impact events and generate impact event trigger flags based on the rate of change of the original environmental perception signal.

[0045] Furthermore, this application also includes the following steps: comparing the original environmental sensing signal with multiple preset incremental impact rate thresholds to determine impact events of different intensities; wherein, when the rate of change first exceeds the impact threshold, a minimum dwell timer is started, and the impact event trigger flag remains valid even if the rate of change falls back below the threshold before the timer expires; simultaneously, a stabilization counter is started, and the impact event trigger flag is only deactivated if the rate of change is continuously lower than the fall-off threshold within a preset stabilization time window; different impact weights are configured according to different signal types, and the weighted comprehensive impact intensity is calculated, wherein the weighted impact intensity is positively correlated with the response amplitude of the asymmetric adjustment control.

[0046] Specifically, the impact event detection unit is a dedicated software module deployed within the processor. It continuously monitors the rate of change of the raw environmental sensing signals to identify sudden, brief, and dramatically large physical changes, generating corresponding trigger flags and intensity levels. The impact event detection unit continuously receives raw sensing signals from the environmental sensor module and independently calculates the rate of change for each signal. A set of preset impact rate thresholds is used, arranged in ascending order of value to correspond to different levels of impact intensity, such as a low impact threshold of 50 lx / s, a medium impact threshold of 200 lx / s, and a high impact threshold of 800 lx / s. Simultaneously, each sensor is configured with an independent minimum dwell timer and a stabilization counter.

[0047] When the real-time rate of change of a signal first exceeds the lowest-level impact threshold, the impact event trigger flag of the sensor is immediately set, and the original intensity level of the impact is determined based on the threshold level exceeded. Simultaneously, the corresponding minimum dwell timer for the sensor is started, with a preset duration. During the timer's operation, regardless of subsequent fluctuations or declines in the rate of change, the impact flag remains valid, ensuring that the brief spike of the impact is not immediately ignored. The minimum dwell timer is a software timer that starts when the rate of change first exceeds any impact threshold, with a timing duration of 100ms. Before this timer expires, even if the rate of change has fallen back below the threshold, the impact event trigger flag remains valid to prevent accidental flag cancellation due to the brief spike of the impact.

[0048] When the rate of change falls below a preset fall-off threshold, the flag is not immediately revoked; instead, a stabilization counter is activated to eliminate signal jitter. The stabilization counter continuously monitors the rate of change within a preset stabilization time window. Only when the number of consecutive samplings with the rate of change below the fall-off threshold within this window reaches the set stabilization count value is the impact considered truly over, at which point the impact event trigger flag for that sensor is revoked. If the rate of change exceeds the impact threshold again within the stabilization window, the counter resets and detection restarts.

[0049] For terminals using multi-sensor fusion, the impact intensity levels of all sensors are collected and converted into numerical values ​​(e.g., Level 1 = 1, Level 2 = 2, Level 3 = 3). Each value is then multiplied by its preset impact weight, and all products are summed to obtain a weighted composite impact intensity. A higher composite impact intensity results in a larger response amplitude, meaning a jump to a higher power consumption level. The composite impact intensity is used to ultimately determine the extent to which the asymmetric adjustment control unit should increase the power consumption level. A minimum dwell timer prevents premature flag cancellation due to brief dips in the impact waveform, ensuring the continuity of performance improvement during the impact period. A stabilization counter eliminates repeated flag jumps caused by signal noise, preventing unnecessary oscillations in the power consumption level. The multi-sensor weighted fusion mechanism allows the terminal to assign different sensitivities to different types of impacts based on the application scenario; for example, wearable devices prioritize acceleration impacts, while voice control devices prioritize acoustic impacts.

[0050] The current environmental information entropy level and the impact event trigger flag are input into the asymmetric adjustment and control unit. Based on the combination of the environmental information entropy level and the existence state of the impact event, the fusion terminal is controlled to perform a step-by-step migration between multiple predefined power consumption levels. The step-by-step migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

[0051] Furthermore, this application also includes the following steps: when there is an impact event triggering flag, the target power consumption level is determined according to the level and intensity of the impact event. The higher the impact intensity, the higher the target power consumption level. Among them, severe impact corresponds to the highest power consumption level, which is not limited by the current information entropy level; when there is no impact event triggering flag, the baseline power consumption level is selected according to the environmental information entropy level; after the impact event triggering flag disappears, the power consumption is restored from the current power consumption level to the baseline level in a step-by-step manner, and the dwell time of each level is not less than the preset minimum dwell time in seconds.

[0052] Specifically, the asymmetric regulation control unit is a decision-making module deployed within the processor. It receives the information entropy level and the impact event trigger flag as input, and outputs power level switching instructions according to preset rules. Asymmetry indicates that the increase in power level (responding to the impact) and the decrease in power level (recovery after the impact) adopt different paths, speeds, and strategies. The asymmetric regulation control unit continuously monitors the current environmental information entropy level output by the information entropy calculation unit, as well as the impact event trigger flag and impact intensity level output by the impact event detection unit.

[0053] The first priority is handling impact events. When the impact event trigger flag becomes valid, the current information entropy level is ignored, and the target power consumption level is determined directly based on the intensity of the impact event. Typically, a mild impact corresponds to balanced mode, a moderate impact to full-speed mode, and a severe impact to the highest power consumption level. The current power consumption level is compared with the target power consumption level: if the target level is higher than the current level, a jump-level upgrade operation is immediately performed, switching the terminal configuration to the target level in one go; if the target level is not higher than the current level, it remains unchanged. During impact event handling, the intensity level and occurrence time of this impact are recorded for use in the subsequent recovery phase.

[0054] The second priority is baseline adjustment in the absence of impact events. When the impact event trigger flag is invalid, the corresponding baseline power consumption level is selected based on the current environmental entropy level. For example, the preset rule is that the baseline level is full-speed mode when the entropy is high, and a mild preemptive sleep mode when the entropy is low. The baseline level is used as the desired target state, but it will not be forcibly switched immediately. Instead, it will gradually approach the target state based on the current actual power consumption level.

[0055] The third priority is asymmetric recovery after the impact event ends. When the impact event trigger flag changes from valid to invalid, it is recognized that the impact has ended, and a step-by-step degradation process is initiated. Starting from the current power consumption level, the power consumption level is reduced by one step every preset minimum dwell time (on a second) until the baseline level corresponding to the current environmental entropy is reached. During each dwell time, even if a new impact event occurs, the degradation process will be immediately interrupted and the ascending logic will be re-executed. If the environmental entropy level changes during the degradation process, such as from high entropy to low entropy, the baseline level will also be updated accordingly, and the terminal continues to degradation step-by-step until the new baseline level is reached. Skipping levels is not allowed in the step-by-step degradation path to ensure sufficient time to stabilize in the new state and avoid task jitter or peripheral abnormalities caused by excessively rapid power changes.

[0056] In the absence of impact events, the baseline power consumption levels corresponding to different information entropy levels; in the presence of impact events, the target power consumption levels corresponding to different impact intensities; and the allowed jump directions and minimum dwell time for each progressively decreasing level. Step-by-step migration refers to a power consumption level transition that is not a continuous, smooth change, but rather a direct jump from one discrete level to another, forming a step-like trajectory. Progressive transitions refer to moving only to adjacent power consumption levels at a time, while jump transitions refer to skipping multiple levels at once.

[0057] For example, four power consumption levels are predefined, numbered and configured as follows: Level 0 is deep preemptive sleep, the CPU stops, only the wake-up circuit works, power consumption 0.05W; Level 1 is mild preemptive sleep, CPU frequency 200MHz, Wi-Fi heartbeat interval 5s, power consumption 0.18W; Level 2 is balanced mode, CPU frequency 1.0GHz, graphics processor is off, power consumption 0.65W; Level 3 is full-speed mode, CPU frequency 1.8GHz, all peripherals are on, power consumption 1.35W. The mapping rules are as follows: without impact, high information entropy corresponds to baseline Level 2; low information entropy corresponds to baseline Level 1. With impact, mild impact (intensity 1) corresponds to Level 2, moderate impact (intensity 2) corresponds to Level 3, and severe impact (intensity 3) corresponds to Level 3 with extended full-speed duration. The minimum dwell time for each progressively decreasing level is set to 2.0s. At t=0.0s, the terminal is in a low information entropy environment, with a current level of 1 (light sleep), no impact, and a baseline level of 1, requiring no switching. At t=0.5s, the impact event detection unit outputs a moderate impact trigger flag. The control unit reads that the flag is valid, and the target level is 3. Since the current level 1 < target 3, an immediate level switch occurs: within 2ms, the CPU frequency is increased from 200MHz to 1.8GHz, the image processor and all peripherals are activated, power consumption increases to 1.35W, and the current level becomes 3. At t=0.9s, the impact flag disappears, the detected flag becomes invalid, the baseline level remains 1, and the current level 3 is higher than the baseline, initiating a gradual decrease in power consumption. Starting from t=0.9s, it needs to stay at level 3 for 2.0s. At t=2.9s, the level 3 dwell time is complete, and the control unit outputs a command to reduce the power consumption level to level 2, the CPU frequency drops to 1.0GHz, the image processor is shut down, power consumption drops to 0.65W, the current level becomes 2, and the level 2 dwell timer is initiated. At t=4.9s, Level 2 is at its maximum (2.0), and the system downgrades to Level 1. The CPU frequency drops to 200MHz, Wi-Fi enters heartbeat mode, and power consumption drops to 0.18W. The current level reaches the baseline Level 1, and the downgrade stops. If, at t=3.5s, during the Level 2 period, the environmental entropy level suddenly becomes high, the baseline level is updated to 2. At this point, the current Level 2 is equal to the baseline level, and the control unit no longer performs further downgrades to Level 1; the terminal stabilizes at Level 2. If a new impact event occurs at any time during the downgrade process, the control unit will immediately interrupt the downgrade and re-execute the jump-up logic.

[0058] The asymmetric adjustment control unit enables joint decision-making on environmental conditions and sudden impact events, and adopts discrete step migration as the basic form of power consumption level change.

[0059] Furthermore, this application also includes the following steps: the power consumption levels include at least: full-speed mode, for all peripherals and processors to operate at the highest frequency voltage; balanced mode, for some non-critical peripherals to be frequency-reduced or turned off; preemptive sleep mode, for cutting off power supply to at least one high-power peripheral, while retaining at least a low-power wake-up channel; wherein, the preemptive sleep mode further includes: mild preemptive sleep and deep preemptive sleep, wherein mild preemptive sleep is to turn off the image signal processor and AI accelerator, while retaining the wireless communication module to heartbeat at a low rate; deep preemptive sleep is to turn off the wireless communication module, while retaining only the wake-up monitoring circuit and real-time clock.

[0060] Specifically, power consumption levels are a set of discrete hardware operating states that can be configured for a smart converged terminal during operation. Each level corresponds to different processor core voltage and frequency, peripheral power supply switch status, and communication module operating modes. Higher levels result in stronger system performance but higher instantaneous power consumption; lower levels limit performance but significantly reduce standby power consumption. Power consumption levels include at least full-speed mode, balanced mode, and preemptive sleep mode.

[0061] In full-speed mode, the CPU clock frequency is set to the highest frequency supported by the chip by calling the processor power management interface, while the core voltage regulator output voltage is set to the minimum stable voltage required for that frequency. The clock gating registers of all peripherals are iterated to ensure that the clocks of all modules are enabled; for peripherals that support independent power domains, such as cameras and AI chips, their power switches must be turned on. Full-speed mode does not impose any power-saving restrictions.

[0062] In balanced mode, the processor frequency is first reduced to approximately 60% of its maximum frequency, and the core voltage is correspondingly lowered to the minimum value corresponding to the new frequency. Then, non-critical peripherals that can operate intermittently in the current application scenario are identified. For example, if the terminal is not currently running an image recognition task, the power supply to the image signal processor is cut off via the power management unit, and the AI ​​accelerator's clock is turned off. For wireless communication modules, normal connectivity is maintained, but transmit power can be reduced or switched to power-saving mode. Other necessary peripherals remain operational.

[0063] For preemptive sleep mode, more granular power domain management is required. The processor is placed in a deep sleep state, such as with the WFE or WFI instructions in ARM architecture combined with a deep sleep flag. In this state, the processor core clock stops, but some debugging or wake-up logic within the CPU can still function. Power supply to peripherals is controlled according to the sub-mode requirements: Mild preemptive sleep disconnects the power supply to the image signal processor and AI accelerator via a power switch chip or load switch. The wireless communication module is not powered off, but it is configured in a low-rate heartbeat mode, such as extending the beacon listening interval of the Wi-Fi module from the standard 100ms to 5s, or setting the heartbeat interval of the Bluetooth module to 2s. A general-purpose input / output interrupt is retained as a wake-up source to resume operation upon receiving network data or external interrupts. Deep preemptive sleep further disconnects the power supply to the wireless communication module, interrupting all communication connections between the terminal and the outside world, retaining only a very low-power wake-up monitoring circuit and a real-time clock module. The real-time clock can be set to generate a wake-up signal every 10s, allowing the terminal to briefly resume operation to check for important events. The wake-up monitoring circuit is used to detect external physical signals, such as button presses or accelerometer interrupt pin level changes.

[0064] Mild deprivation hibernation is a sub-mode of deprivation hibernation where power to the high-power image signal processor and AI accelerator is completely cut off. Wireless communication modules such as Wi-Fi, Bluetooth, or cellular modules retain power but operate in a low-rate heartbeat mode, sending a keep-alive beacon or listening for a wake-up signal every few seconds to maintain network connectivity. The processor cores can enter light sleep or run at low frequency. Deep deprivation hibernation is another sub-mode of deprivation hibernation, even more energy-efficient than mild deprivation hibernation. In addition to shutting down the image processor and AI accelerator, it also cuts off power to the wireless communication modules. Only a separate low-power wake-up monitoring circuit and a real-time clock module are retained. The real-time clock generates periodic wake-up interrupts, and the wake-up monitoring circuit detects wake-up events from external buttons, sensor interrupts, or specific physical signals.

[0065] Switching between all power consumption levels is accomplished by calling pre-written configuration functions. The switching process must ensure timing coordination between peripherals and the processor, such as reducing frequency before reducing voltage, or restoring voltage before increasing frequency. By clearly defining and subdividing power consumption levels, a tiered coverage from high performance to extremely low power consumption is achieved, providing a clear migration target for the asymmetric adjustment control unit. The distinction between mild preemptive sleep and deep preemptive sleep is particularly important: the former disconnects high-performance computing peripherals while retaining low-speed communication capabilities, suitable for scenarios requiring network connectivity but not fast response; the latter further disconnects the communication module, used for long-term standby that can be resumed via a local wake-up signal.

[0066] Furthermore, this application also includes the following steps: the asymmetric adjustment control unit dynamically adjusts the migration path according to the changing trend of the real-time information entropy level, including: when descending from full-speed mode to a lower level, if the current information entropy level is low and the entropy change rate is negative and the absolute value exceeds the preset shortcut threshold, skip the intermediate power consumption level and directly switch to the deprivation sleep mode; if the current information entropy level is low but the frequency of alternating positive and negative entropy change rates exceeds the fluctuation threshold, then a transient monitoring mode is forcibly inserted between the balanced mode and the deprivation sleep mode. In the transient monitoring mode, only the key sensors are maintained to work at the first sampling rate, while other high-power peripherals remain disconnected, and the first sampling rate is higher than the second sampling rate in the deprivation sleep mode.

[0067] Specifically, in addition to receiving the information entropy level and the impact flag, the asymmetric adjustment control unit continuously receives the entropy change rate and alternating positive and negative statistical data of the entropy change rate from the information entropy calculation unit. When the intelligent fusion terminal is currently in full-speed mode and is executing the gradual descent process after the impact disappears (i.e., the impact flag has disappeared and the current level is higher than the baseline level), the asymmetric adjustment control unit first checks whether the current environmental information entropy level is low. If so, it further checks whether the entropy change rate is negative and whether its absolute value is greater than a preset shortcut threshold. If both conditions are met simultaneously, it indicates that the environment is not only currently low entropy but is also becoming more stable at a very rapid pace. At this point, the asymmetric adjustment control unit believes that there is no need to remain in the equilibrium mode and can directly switch from full-speed mode to the deprivation sleep mode. In specific operation, the asymmetric adjustment control unit will abort the original gradual descent plan and directly call the switching function from full-speed mode to deprivation sleep mode, skipping the equilibrium mode.

[0068] When environmental entropy fluctuates frequently, a transient monitoring mode is forcibly inserted. When the current environmental entropy level of the intelligent fusion terminal is low, but the asymmetric adjustment control unit detects that the frequency of alternating positive and negative entropy changes over a period of time exceeds a preset fluctuation threshold, it means that although the environment is generally in a low-entropy state, it is unstable and repeatedly jumps between two states, like a weak but frequent wind blowing leaves causing slight fluctuations in light. In this scenario, directly putting the terminal into a sleep-deprivation mode may cause frequent wake-ups due to slight environmental disturbances, resulting in additional energy consumption; remaining in the balanced mode would lead to unnecessary high power consumption. To solve this dilemma, the control unit forcibly inserts a temporary transient monitoring mode between the balanced mode and the sleep-deprivation mode, with power consumption between the two. Specifically, this involves cutting off all high-power peripherals, such as image processors, AI accelerators, and wireless communication modules, while keeping key sensors operating at a high first sampling rate, typically more than an order of magnitude higher than the second sampling rate in sleep-deprivation mode. The purpose of the transient monitoring mode is to avoid entering deep sleep and being unable to respond in a timely manner when the environment is unstable and fluctuating, while also avoiding the waste caused by remaining in the high-power balanced mode. The asymmetric regulation control unit instructs the processor to switch to transient monitoring mode and run in this mode for a preset period of time, during which it continuously monitors the alternation of positive and negative values ​​in the rate of change of environmental entropy. If the environment stabilizes after this period, it can further degrade to preemptive hibernation mode; if the environment becomes more complex, it should revert to balanced mode or full-speed mode. The transient monitoring mode acts as a buffer, avoiding frequent oscillations between two extreme power consumption levels.

[0069] The first sampling rate is the operating frequency of key sensors in transient monitoring mode. For example, an accelerometer might collect data 100 times per second. This sampling rate is high enough to capture rapid changes in the environment, allowing for timely detection of trends towards stabilization or increasing complexity. The second sampling rate is the operating frequency of key sensors in sleep-deprivation mode, typically much lower than the first sampling rate. For example, an accelerometer might collect data only 5 times per second. This is mainly used to detect large movements or impacts to determine whether to wake the sensor. The second sampling rate only maintains a minimum level of environmental awareness.

[0070] Furthermore, this application also includes the following steps: recording the extra energy consumed in each impact event, accumulating and determining the energy debt balance, wherein the extra energy is calculated based on the level intensity, duration, and difference between the actual power consumption and the baseline power consumption during the impact event; during the gradual reduction of power consumption after the impact event ends, dynamically extending the dwell time of each level according to the energy debt balance, the higher the energy debt balance, the longer the dwell time of each level; after entering the preemptive sleep mode, operating at a power level lower than the normal preemptive sleep power consumption to deduct the energy debt balance, until the balance is zero before restoring the normal preemptive sleep power consumption, wherein operating at a power level lower than the normal preemptive sleep power consumption refers to reducing the sampling rate or communication frequency of non-critical peripherals.

[0071] Specifically, the asymmetric regulation control unit has a pre-set energy debt balance variable, initially set to 0. Each time an impact event occurs, the control unit records the impact's intensity and duration. During the effective period of the impact event trigger flag, the current actual power consumption level and the baseline power consumption level that should exist without an impact are determined. The instantaneous power values ​​corresponding to the two levels are read from a pre-stored power consumption level table, and the power difference is calculated by subtracting the baseline power from the actual power. The power difference is multiplied by the impact duration to obtain the extra energy from this impact. This extra energy is added to the energy debt balance variable. If a second impact occurs shortly after the initial impact, the accumulation continues, and the energy debt balance continuously increases. The energy debt balance is a cumulative value obtained by adding up the extra energy caused by each impact event, representing how much energy the terminal has overdrawn from the overall energy budget, which needs to be repaid through lower power consumption in subsequent energy-saving phases.

[0072] After the impact event, the intelligent converged terminal begins to gradually descend from its current high power consumption level to the baseline level. Before pausing at each level, the asymmetric adjustment control unit reads the current energy debt balance and calculates the additional dwell time required at that level based on a preset mapping function, such as extra dwell time = proportional coefficient × energy debt balance. The original minimum dwell time for each level plus the extra time equals the actual dwell time for that level. The higher the energy debt balance, the longer the extra time, allowing the terminal to remain at higher power consumption levels for a longer period. The dwell time is not fixed but dynamically extended based on the energy debt balance. Because energy was previously overdrawn, the recovery process should not enter a low power consumption state too quickly. Instead, by extending the time at the medium power consumption level, the intelligent converged terminal can repay its energy debt in a relatively smooth manner, avoiding even greater energy overdrawn due to another impact after a rapid degradation.

[0073] When the intelligent converged terminal finally degrades to sleep deprivation mode (light or deep), if the energy debt balance is still greater than zero, the asymmetric adjustment control unit will enter the debt deduction phase. In the normal sleep deprivation mode, there is a preset power consumption value. The asymmetric adjustment control unit temporarily reduces non-critical peripherals to a level lower than the normal setting, thereby further reducing instantaneous power consumption below the normal value, such as 1.8mW. This low-power state is maintained for a predetermined period, or continues until the debt is deducted. The deduction rate = normal power consumption - current low power consumption. Every second, the energy debt balance is reduced by this difference. When the energy debt balance reaches zero, the control unit restores the peripherals to the normal sleep deprivation configuration, and the intelligent converged terminal returns to normal energy-saving state. If the intelligent converged terminal detects a new impact event during the debt deduction period, it will immediately interrupt the deduction process, increase the power consumption level to respond to the impact, and the additional energy from the new impact event will continue to accumulate on the energy debt balance, further increasing the debt. After the intelligent converged terminal enters the deprivation sleep mode, it operates at a lower power. The difference between the actual power consumed and the normal deprivation sleep power is multiplied by the operating time, which is the amount deducted from the energy debt balance. The intelligent converged terminal only resumes the normal deprivation sleep power consumption when the balance reaches zero.

[0074] By accumulating the extra energy consumption of each impact, the cost of short-term performance gains is recorded and compensated for with lower power consumption during subsequent idle periods, preventing a significant reduction in battery life due to multiple impacts. After an impact, the dwell time is dynamically extended based on the outstanding debt, preventing the intelligent fusion terminal from returning to the baseline level too quickly. This mitigates the risk of repeated energy overdrafts in the event of another impact within a short period. Even if the intelligent fusion terminal experiences frequent impacts, the energy debt balance remains at a high level, forcing the system to operate at extremely low power consumption for a longer period, thus preventing uncontrolled total energy consumption due to multiple impacts. Simultaneously, wake-up capability is not completely lost during the debt deduction period, ensuring basic event response capabilities.

[0075] In summary, the environmental perception-based intelligent fusion terminal power dynamic adjustment method provided in this application has the following beneficial effects: Environmental information is collected by connecting an environmental sensor module, and a raw environmental perception signal is output; the raw environmental perception signal is received by an information entropy calculation unit deployed within the processor, and the current environmental information entropy level is determined based on the time-domain change rate and amplitude threshold of the raw environmental perception signal; an impact event detection unit is invoked to identify impact events and generate impact event trigger flags based on the change rate of the raw environmental perception signal; the current environmental information entropy level and the impact event trigger flag are input into an asymmetric adjustment control unit, and the fusion terminal is controlled to perform a stepped migration between multiple predefined power consumption levels according to the combination of the environmental information entropy level and the existence state of the impact event. The stepped migration refers to the power consumption level changing step by step or skipping steps in a discrete manner. In other words, by utilizing the dual characteristics of environmental information entropy and impact events, the power consumption level is guided to implement asymmetric stepped migration based on the asymmetric adjustment control unit, ensuring low-latency response to sudden events while effectively suppressing ineffective power consumption switching caused by environmental disturbances, thereby improving the overall energy efficiency ratio in real complex environments.

[0076] Example 2: Based on the same inventive concept as the environment-aware intelligent fusion terminal power dynamic adjustment method in Example 1, this application also provides an environment-aware intelligent fusion terminal power dynamic adjustment system. Please refer to the appendix. Figure 2 The intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception includes: an environmental information acquisition module 11, used to connect to an environmental sensor module to collect environmental information and output raw environmental perception signals; an information entropy calculation module 12, used to receive the raw environmental perception signals through an information entropy calculation unit deployed in the processor, and determine the current environmental information entropy level based on the time-domain change rate and amplitude threshold of the raw environmental perception signals; an impact event identification module 13, used to call an impact event detection unit to identify impact events and generate impact event trigger flags based on the change rate of the raw environmental perception signals; and a power consumption adjustment control module 14, used to input the current environmental information entropy level and the impact event trigger flags into an asymmetric adjustment control unit, and control the fusion terminal to perform a step-by-step migration between multiple predefined power consumption levels according to the combination of the environmental information entropy level and the existence state of the impact event, wherein the step-by-step migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

[0077] Furthermore, the information entropy calculation module 12 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used to: calculate the average rate of change of the original environmental perception signal within a time window and the statistical mean of the signal amplitude to obtain the average rate of change and the statistical mean of the amplitude; when the average rate of change is lower than a first rate threshold and the statistical mean of the amplitude is lower than a first amplitude threshold, it is determined to be a low information entropy level, otherwise it is determined to be a high information entropy level.

[0078] Furthermore, the information entropy calculation module 12 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used for: the average rate of change is the integral of the absolute value of the first derivative of the original environmental perception signal over the time window divided by the window length; the first rate threshold and the first amplitude threshold are both dynamically adjustable, and the adjustment basis includes: the current remaining power of the terminal and the historical statistical value of environmental fluctuations within the past preset time period.

[0079] Furthermore, the information entropy calculation module 12 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used to: calculate the derivative of information entropy with respect to time to obtain the entropy change rate; if the entropy change rate is negative and the absolute value exceeds the first derivative threshold, even if the average change rate and amplitude statistical mean at the current moment do not meet the conditions for low information entropy, the environmental state will be marked as an entropy reduction warning state, and the entry conditions for the low information entropy level will be relaxed; if the entropy change rate is positive and exceeds the second derivative threshold, it will be marked as an entropy increase warning state, and the entry conditions for the low information entropy level will be tightened; wherein, the way to relax or tighten the entry conditions includes: temporarily adjusting the values ​​of the first rate threshold and / or the first amplitude threshold, the adjustment magnitude being positively correlated with the absolute value of the entropy change rate.

[0080] Furthermore, the impact event identification module 13 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used to: compare the original environmental perception signal with thresholds according to a plurality of preset incremental impact rate thresholds to determine impact events of different intensities; wherein, when the rate of change first exceeds the impact threshold, a minimum dwell timer is started, and the impact event trigger flag remains valid even if the rate of change falls back below the threshold before the timer expires; at the same time, a stabilization counter is started, and the impact event trigger flag is revoked only if the rate of change is continuously lower than the fall-off threshold within a preset stabilization time window; different impact weights are configured according to different signal types, and the weighted comprehensive impact intensity is calculated, wherein the weighted impact intensity is positively correlated with the response amplitude of the asymmetric adjustment control.

[0081] Furthermore, the power consumption adjustment control module 14 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used to: when there is an impact event triggering flag, determine the target power consumption level according to the level and intensity of the impact event. The higher the impact intensity, the higher the target power consumption level. Among them, severe impact corresponds to the highest power consumption level, which is not limited by the current information entropy level; when there is no impact event triggering flag, select the baseline power consumption level according to the environmental information entropy level; after the impact event triggering flag disappears, restore the power consumption from the current power consumption level to the baseline level in a step-by-step manner, with the dwell time of each level not less than the preset minimum dwell time in seconds.

[0082] Furthermore, the power consumption adjustment control module 14 in the environmentally aware intelligent fusion terminal power consumption dynamic adjustment system is also used to: the power consumption levels include at least: full-speed mode, for all peripherals and processors to operate at the highest frequency voltage; balanced mode, for some non-critical peripherals to reduce frequency or shut down; and preemptive sleep mode, for cutting off power supply to at least one high-power peripheral while retaining at least a low-power wake-up channel; wherein, the preemptive sleep mode further includes: mild preemptive sleep and deep preemptive sleep, wherein mild preemptive sleep is to turn off the image signal processor and AI accelerator while retaining the wireless communication module to heartbeat at a low rate; and deep preemptive sleep is to turn off the wireless communication module while retaining only the wake-up monitoring circuit and real-time clock.

[0083] Furthermore, the power consumption adjustment control module 14 in the intelligent fusion terminal power consumption dynamic adjustment system based on environmental perception is also used for: the asymmetric adjustment control unit dynamically adjusts the migration path according to the changing trend of the real-time information entropy level, including: when descending from full-speed mode to a lower level, if the current information entropy level is low and the entropy change rate is negative and the absolute value exceeds the preset shortcut threshold, skip the intermediate power consumption level and directly switch to the deprivation sleep mode; if the current information entropy level is low but the frequency of alternating positive and negative entropy change rates exceeds the fluctuation threshold, then a transient monitoring mode is forcibly inserted between the balanced mode and the deprivation sleep mode. In the transient monitoring mode, only the key sensors are maintained to work at the first sampling rate, and other high-power peripherals are kept off, and the first sampling rate is higher than the second sampling rate in the deprivation sleep mode.

[0084] Furthermore, the power consumption adjustment control module 14 in the environmental perception-based intelligent fusion terminal power consumption dynamic adjustment system is also used to: record the extra energy consumed by each impact event, accumulate and determine the energy debt balance, wherein the extra energy is calculated based on the level intensity, duration of the impact event, and the difference between the actual power consumption and the baseline power consumption during the impact; during the gradual reduction of power consumption after the impact event ends, the dwell time of each level is dynamically extended according to the energy debt balance, the higher the energy debt balance, the longer the dwell time of each level; after entering the deprivation sleep mode, it operates at a power level lower than the normal deprivation sleep power consumption to deduct the energy debt balance, until the balance is zero before restoring the normal deprivation sleep power consumption, wherein operating at a power level lower than the normal deprivation sleep power consumption refers to reducing the sampling rate or communication frequency of non-critical peripherals.

[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The method and specific examples of dynamic power consumption adjustment for intelligent fusion terminals based on environmental awareness in Example 1 are also applicable to the dynamic power consumption adjustment system for intelligent fusion terminals based on environmental awareness in this embodiment. Through the foregoing detailed description of the dynamic power consumption adjustment method for intelligent fusion terminals based on environmental awareness, those skilled in the art can clearly understand the dynamic power consumption adjustment system for intelligent fusion terminals based on environmental awareness in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for dynamic power consumption adjustment of intelligent fusion terminals based on environmental perception, characterized in that, include: Connect to the environmental sensor module to collect environmental information and output raw environmental perception signals; The original environmental perception signal is received by the information entropy calculation unit deployed in the processor, and the current environmental information entropy level is determined based on the time domain change rate and amplitude threshold of the original environmental perception signal. The impact event detection unit is invoked to identify impact events and generate impact event trigger flags based on the rate of change of the original environmental perception signal. The current environmental information entropy level and the impact event trigger flag are input into the asymmetric adjustment and control unit. Based on the combination of the environmental information entropy level and the existence state of the impact event, the fusion terminal is controlled to perform a step-by-step migration between multiple predefined power consumption levels. The step-by-step migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

2. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 1, characterized in that, Based on the combination of environmental information entropy level and the existence state of the impact event, the fusion terminal is controlled to perform a stepped migration between multiple predefined power consumption levels, including: When an impact event trigger flag is present, the target power consumption level is determined based on the level and intensity of the impact event. The higher the impact intensity, the higher the target power consumption level. Among them, severe impact corresponds to the highest power consumption level, which is not limited by the current information entropy level. When there is no impact event trigger flag, the baseline power consumption level is selected based on the environmental information entropy level. After the impact event trigger flag disappears, the power consumption level is restored from the current power consumption level to the baseline level by gradually reducing the power consumption, with the dwell time at each level not less than the preset minimum dwell time in seconds.

3. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 1, characterized in that, The power consumption levels include at least: Full-speed mode, used for all peripherals and processors to operate at the highest frequency voltage; Equalization mode is used to reduce the frequency or shut down some non-critical peripherals. Preemptive sleep mode is used to cut off power to at least one high-power peripheral while retaining at least a low-power wake-up channel; The sleep deprivation mode includes two types: mild sleep deprivation and deep sleep deprivation. Mild sleep deprivation disables the image signal processor and AI accelerator, while retaining the wireless communication module with a low heart rate. Deep sleep deprivation disables the wireless communication module, while retaining only the wake-up monitoring circuit and real-time clock.

4. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 1, characterized in that, Based on the time-domain rate of change and amplitude threshold of the original environmental sensing signal, the current environmental information entropy level is determined, including: Calculate the average rate of change of the original environmental sensing signal within the time window, and the statistical mean of the signal amplitude, to obtain the average rate of change and the statistical mean of the amplitude; When the average rate of change is lower than the first rate threshold and the average amplitude is lower than the first amplitude threshold, it is determined to be a low information entropy level; otherwise, it is determined to be a high information entropy level.

5. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 4, characterized in that, The average rate of change is the integral of the absolute value of the first derivative of the original environmental sensing signal over the time window, divided by the window length. Both the first rate threshold and the first amplitude threshold are dynamically adjustable, and the adjustment is based on: the terminal's current remaining battery power and historical statistical values ​​of environmental fluctuations within a preset time period.

6. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 5, characterized in that, Determining the current environmental entropy level also includes: Calculate the derivative of information entropy with respect to time to obtain the rate of change of entropy; If the rate of change of entropy is negative and the absolute value exceeds the first derivative threshold, even if the average rate of change and the average value of the magnitude at the current moment do not meet the conditions for low information entropy, the environmental state will be marked as an entropy reduction warning state, and the entry conditions for the low information entropy level will be relaxed. If the rate of change of entropy is positive and exceeds the second derivative threshold, it is marked as an entropy rise warning state, and the entry conditions for low information entropy levels are tightened. The methods for relaxing or tightening entry conditions include: temporarily adjusting the values ​​of the first rate threshold and / or the first amplitude threshold, wherein the adjustment magnitude is positively correlated with the absolute value of the entropy change rate.

7. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 1, characterized in that, Identify impact events and generate impact event trigger flags, including: Based on multiple preset, incremental impact rate thresholds, the original environmental perception signals are compared to determine impact events of different intensities. When the rate of change exceeds the impact threshold for the first time, a minimum dwell timer is started. Before the timer expires, even if the rate of change falls back below the threshold, the impact event trigger flag remains valid. At the same time, the image stabilization counter is activated. If the rate of change is continuously lower than the fallback threshold within the preset image stabilization time window, the impact event trigger flag is canceled. Different impact weights are configured according to different signal types, and the comprehensive impact intensity is calculated by weighting. The weighted impact intensity is positively correlated with the response amplitude of the asymmetric regulation control.

8. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 2, characterized in that, The asymmetric adjustment and control unit dynamically adjusts the migration path based on the changing trend of the real-time information entropy level, including: When descending from full-speed mode to a lower level, if the current information entropy level is low and the entropy change rate is negative and the absolute value exceeds the preset shortcut threshold, skip the intermediate power consumption level and directly switch to preemptive sleep mode. If the current information entropy level is low but the frequency of alternating positive and negative entropy change rates exceeds the fluctuation threshold, a transient monitoring mode is forcibly inserted between the balanced mode and the deprivation sleep mode. In the transient monitoring mode, only the key sensors are kept working at the first sampling rate, while other high-power peripherals are kept off, and the first sampling rate is higher than the second sampling rate in the deprivation sleep mode.

9. The method for dynamic power consumption adjustment of an intelligent fusion terminal based on environmental perception according to claim 3, characterized in that, Also includes: Record the extra energy consumed in each impact event, and accumulate it to determine the energy debt balance. The extra energy is calculated based on the level and intensity of the impact event, the duration, and the difference between the actual power consumption and the baseline power consumption during the impact. During the gradual reduction of power consumption after the impact event, the dwell time at each stage is dynamically extended according to the energy debt balance. The higher the energy debt balance, the longer the dwell time at each stage. After entering sleep deprivation mode, it operates at a power level lower than the normal sleep deprivation power consumption to deduct the energy debt balance until the balance is zero, at which point it resumes the normal sleep deprivation power consumption. The power level lower than the normal sleep deprivation power consumption refers to reducing the sampling rate or communication frequency of non-critical peripherals.

10. A power consumption dynamic adjustment system for intelligent fusion terminals based on environmental perception, characterized in that, The steps for implementing the environment-aware intelligent fusion terminal power consumption dynamic adjustment method according to any one of claims 1 to 9, wherein the environment-aware intelligent fusion terminal power consumption dynamic adjustment system comprises: The environmental information acquisition module is used to connect to the environmental sensor module to collect environmental information and output raw environmental sensing signals; The information entropy calculation module is used to receive the original environmental perception signal through the information entropy calculation unit deployed in the processor, and determine the current environmental information entropy level based on the time domain change rate and amplitude threshold of the original environmental perception signal. The impact event identification module is used to call the impact event detection unit to identify impact events and generate impact event trigger flags based on the rate of change of the original environmental perception signal. The power consumption regulation control module is used to input the current environmental information entropy level and the impact event trigger flag into the asymmetric regulation control unit. Based on the combination of the environmental information entropy level and the existence state of the impact event, the module controls the fusion terminal to perform a step-by-step migration between multiple predefined power consumption levels. The step-by-step migration refers to the power consumption level changing step by step or skipping steps in a discrete manner.

Images