Adaptive thermal control heat sink control method for radio frequency microwave front end

Abstract

The present application relates to the field of automatic control technology, and further relates to an adaptive thermal control radiator control method for a radio frequency microwave front end, comprising the following steps: registering a digital pre-distortion tap and an envelope tracking voltage rail as a soft actuator, and registering a controllable non-condensing gas plug heat pipe gas plug driving unit as a hard actuator; during continuous operation of a service waveform, a system controller performs the following closed-loop operation based on frame data collection within a control window of each service frame: within the control window, a synthesized baseband sequence, a rail voltage sequence and an operation sequence are respectively sent to a transmitting link, an envelope tracking power supply and a gas plug driving unit for execution. The present application can actively intervene before thermal stress is formed, quickly reduce the peak envelope ratio and output power through the soft actuator, reduce instantaneous power consumption from the signal domain, and simultaneously reconstruct a heat passage through the hard actuator to guide heat to an area with higher heat dissipation efficiency, thereby avoiding local overheating.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology, specifically relating to an adaptive thermal control heat sink control method for radio frequency microwave front-ends. Background Technology

[0002] As a critical component of modern communications, radar, and satellite links, the power amplification, signal linearization, and energy efficiency management technologies of radio frequency (RF) microwave front-ends have undergone significant development over the past two decades. Existing technologies typically employ digital predistortion, envelope tracking, and various forms of thermal management structures to maintain stable operation of devices at high power densities. Digital predistortion compensates for signal nonlinearity in the baseband domain to counteract the nonlinear distortion of the power amplifier; envelope tracking improves overall energy efficiency by adjusting the power amplifier's supply voltage rails in real time to match the amplifier's power supply with its output envelope. Meanwhile, in terms of thermal management, traditional heat dissipation methods primarily rely on passive heat sinks, heat pipes, phase change materials, or liquid cooling plates to dissipate the heat generated by the devices. However, with the increasing frequency of RF systems, the growing complexity of modulation, and the increasing integration of power amplifiers, the transient thermal load of the system has significantly increased, and the response speed and spatial adjustment capabilities of traditional thermal management methods are no longer sufficient to meet the rapidly changing thermal dynamics.

[0003] In existing RF front-ends, envelope tracking and digital predistortion typically operate independently, with their control objectives focused on signal quality and energy efficiency, lacking a feedback mechanism linked to thermal conditions. When a power amplifier operates continuously under high peak-to-envelope ratio signals (e.g., 5G NR or WLAN high-order modulation waveforms), its hotspot areas accumulate heat rapidly, causing junction temperatures to rise too quickly. Existing systems often rely solely on temperature threshold-triggered protection mechanisms, such as gain reduction or channel shutdown. This approach suffers from response lag and coarse regulation, often intervening only after temperature rises, failing to proactively suppress heat in the early stages of thermal stress. Furthermore, traditional heat pipes or phase-change cooling structures are mostly passive, with fixed heat transfer paths, unable to adjust the thermal resistance distribution between the evaporation and condensation sections according to different load conditions. This results in localized hotspots remaining in high heat flux density regions for extended periods, impacting device reliability. In recent years, some studies have proposed using controllable non-condensing plug heat pipes for active thermal path regulation, controlling the effective heat transfer length of the heat pipe by adjusting the position or volume of the non-condensing gas. However, most of these methods rely primarily on temperature control or heating band power control, without dynamically coordinating with the behavioral characteristics of the radio frequency signal itself. When the radio frequency output power changes rapidly, the migration of the gas plug lags behind the instantaneous change in heat, resulting in insufficient response speed of thermal control. Furthermore, gas plug regulation is usually independent of the radio frequency electrical control system, lacking unified control logic, which leads to difficulties in system coordination and low response efficiency. Summary of the Invention

[0004] The main objective of this invention is to provide an adaptive thermal control method for RF microwave front-ends. This method proactively intervenes before thermal stress occurs, rapidly reducing the peak envelope ratio and output power through soft actuators to cut instantaneous power consumption in the signal domain. Simultaneously, it reconstructs the thermal path through hard actuators, guiding heat to areas with higher heat dissipation efficiency and avoiding localized overheating. Compared to traditional passive cooling and delay protection mechanisms, this invention achieves millisecond-level thermal control response and structural-level thermal conductivity optimization, enabling the RF power amplifier to maintain output linearity and thermal stability under high peak envelope ratio signals. This significantly improves system reliability, heat dissipation uniformity, and energy efficiency, providing a real-time adaptive thermal control solution for high-power-density RF equipment.

[0005] To solve the above problems, the technical solution of the present invention is implemented as follows:

[0006] An adaptive thermal control radiator control method for RF microwave front-ends includes the following steps: registering digital predistortion taps and envelope tracking voltage rails as soft actuators, and registering the plug drive unit of a controllable non-condensing plug heat pipe as a hard actuator; during continuous operation of the service waveform, the system controller performs the following closed-loop operation based on the data acquired in each service frame within the control window: acquiring and aligning baseband sampling, rail pressure sampling, temperature and heat flux sampling, and plug position status to generate a homogeneous frame containing a baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and plug state sequence; based on the same... The system constructs a frame and derives the thermal stress level representing the current state. Based on the thermal stress level and the waveform feature detection results of the baseband sequence, it retrieves the soft actuator action combination and hard actuator action combination from the preset thermal control mapping table. Based on the soft actuator action combination, it synthesizes the baseband sequence and rail pressure sequence used for shaping the peak envelope ratio and implementing output backoff. Based on the hard actuator action combination, it synthesizes the operation sequence used for migrating the gas plug to reconstruct the thermal path. Within the control window, the synthesized baseband sequence, rail pressure sequence, and operation sequence are sent to the transmit link, envelope tracking power supply, and gas plug drive unit for execution, respectively.

[0007] Furthermore, the waveform feature detection results of the baseband sequence include at least one of the following: a peak density sequence obtained by sliding detection of the baseband sequence, a frequency of change sequence obtained by statistically analyzing the number of amplitude changes, and a high-frequency occupancy indication sequence obtained by transforming to the frequency domain and extracting edge frequency energy.

[0008] Furthermore, the steps for generating isomorphic frames specifically include: within the control window, buffering baseband sampling, rail pressure sampling, temperature and heat flux sampling, and the position status of the gas plug; and using linear insertion or nearest neighbor methods to complete or dilute each buffered sequence so that each sequence has a corresponding value at each sampling position.

[0009] Furthermore, the steps for deriving the thermal stress level specifically include: mapping thermal path status labels indicating the combination of dominant, bypass, and restrictive paths between the evaporation and condensation sections based on the temperature sequence, heat flow sequence, and gas lock state sequence of the isomorphic frame; and obtaining the thermal stress level by looking up a table based on the thermal path status labels and the level combination of temperature and heat flow.

[0010] Furthermore, the step of synthesizing the baseband sequence for shaping the peak envelope ratio specifically includes: performing sliding detection on the baseband sequence to locate the peak position; for each located peak, selecting a window sequence that matches the peak width from a preset window table, and replacing the amplitude of the coverage area on both sides of the peak position point by point according to the window sequence, keeping the phase unchanged during replacement, thereby forming a suppressed amplitude sequence.

[0011] Furthermore, when the high-frequency occupancy indication in the waveform feature detection result is significant, the step of synthesizing the baseband sequence also includes a phase rearrangement gate operation, specifically: converting the suppressed amplitude sequence and the original phase information into amplitude-to-phase pairs in the frequency domain; reading the target phase sequence from a preset phase table; replacing the phase values ​​in the marked frequency point set with the values ​​in the target phase sequence; and then inversely transforming back to the time domain.

[0012] Furthermore, the step of synthesizing the rail pressure sequence for implementing output rollback specifically includes: reading the upper limit value, lower limit value, shortest continuous sampling number of the plateau segment, and maximum single-step change amount corresponding to the output rollback intention from a preset rail pressure table; scanning the rail pressure sequence, replacing sampling points that are higher than the upper limit value or lower than the lower limit value with the corresponding limit value, and limiting the difference between adjacent samples to not exceed the maximum single-step change amount; and inserting a plateau segment that is not shorter than the shortest continuous sampling number of the plateau segment after the change ends, thereby obtaining the shaped rail pressure sequence.

[0013] Furthermore, the step of synthesizing the operation sequence for migrating the gas plug specifically includes: reading the operation sequence corresponding to the gas plug migration intention from a preset gas plug migration table; when the intention is to extend the gas plug, the operation sequence controls the opening of the heating belt on the outside of the evaporation section and introduces non-condensable gas into the heat pipe; when the intention is to shorten the gas plug, the operation sequence controls the opening of the heating belt on the outside of the condensation section and discharges non-condensable gas; when the intention is to migrate the gas plug direction, the operation sequence controls the on and off of the heating belt according to a preset combination to form a temperature gradient.

[0014] Furthermore, the issuance steps follow interlocking and sequential control, and the sequence is as follows: first complete the rail pressure trajectory shaping, then complete the peak envelope ratio shaping and phase rearrangement gate, and finally trigger the gas plug migration.

[0015] Furthermore, after all actions in a frame are completed, a closed-loop evaluation and online resetting process is also included: comparing the samples from the frame tail and frame head to evaluate the thermal stress improvement conclusion and waveform fidelity conclusion; if either conclusion fails to meet the preset pass condition, an online resetting process is performed before the frame ends. The online resetting process includes increasing the output backoff intention. Figure 1 One of the following: select a gear, switch to a stronger window table entry, or invoke an acceleration entry to perform a higher-intensity plug fine-tuning.

[0016] The adaptive thermal control radiator control method for RF microwave front-ends of the present invention has the following beneficial effects: By establishing a cooperative closed loop between soft and hard actuators within the same service frame, bidirectional real-time coupling between the signal domain and the thermal domain is achieved, thereby significantly improving the thermal stability and linear output performance of the RF system under high power density. By registering the digital predistortion tap and envelope tracking voltage rail as soft actuators, the system can automatically adjust the peak envelope ratio and output backoff amplitude with a millisecond-level response when an increase in thermal stress level is detected, thereby reducing the instantaneous power consumption and junction temperature rise rate of the power amplifier without affecting signal integrity. At the same time, by using the plug drive unit of the controllable non-condensing plug heat pipe as a hard actuator, the dynamic reconstruction of the thermal path can be achieved within tens of milliseconds by migrating the plug position and adjusting its length, so that heat is guided from the hot spot area to the condensation section area with higher heat dissipation efficiency, improving the heat conduction uniformity of the heat dissipation path. Soft actuators rapidly suppress peak envelope ratio and smooth envelope trajectories in the waveform domain, while hard actuators redistribute thermal resistance in the structural domain. Both types of actuators achieve timing synchronization and command interlocking through a unified data structure of isomorphic frames, avoiding response deviations and interference under independent control. Compared to traditional passive cooling and post-derating mechanisms, this invention proactively intervenes before thermal stress occurs, reducing peak temperature rise, shortening thermal recovery time, and maintaining stable output linearity and adjacent channel suppression performance. The overall effect is reflected in more uniform temperature distribution, lower waveform distortion, higher average power amplifier efficiency, and significantly improved overall equipment reliability and lifespan during long-term continuous high-load operation. By introducing real-time closed-loop mapping between signal control and thermal control, this invention achieves a functional leap from "passive protection" to "adaptive thermal control" in the RF microwave front-end, making the heat dissipation process controllable at both the physical and signal levels, providing a new technical path for the thermal design of high-frequency, high-power communication equipment. Attached Figure Description

[0017] Figure 1 A schematic diagram of the controllable non-condensing plug heat pipe structure provided in an embodiment of the present invention;

[0018] Figure 2 This is a schematic diagram of the peak coverage window sequence provided in an embodiment of the present invention;

[0019] Figure 3This is a timing diagram of the actions of the soft and hard actuators within the control window provided in an embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of the temperature distribution curves before and after the gas plug migration provided in an embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0022] Example 1: An adaptive thermal control radiator control method for RF microwave front-ends. First, the digital predistortion tap and envelope tracking voltage rail are registered as soft actuators, and the plug drive unit of the controllable non-condensing plug heat pipe is registered as a hard actuator. In specific implementation, before entering the service, the system controller first identifies and registers the soft and hard actuators. Identification actions include reading the baseband input and feedback path of the digital predistortion tap, reading the rail voltage control and rail voltage sampling path of the envelope tracking voltage rail, and reading the on / off status and position of the heating element, metering microvalve, discharge microvalve, and position sensor of the plug drive unit. Registration actions are implemented using action tables: a window table, phase table, and rail voltage table are created for soft actuators, and a plug migration table is created for hard actuators, with action combination indexes created for both. The criterion for successful registration is that the soft actuator can complete the replacement and shaping of the baseband sequence and rail voltage sequence according to the action table entries within the control window, and the hard actuator can receive the operation sequence and initiate one or more fine-tuning actions within the control window.

[0023] The registration process for soft actuators employs a reusable three-part description. The first part is capability enumeration, specifying the types and intensity levels of actions the soft actuator can perform. For example, the window length of the window table is set to five levels, corresponding to 64, 96, 128, 160, and 192 sampling points per peak coverage area; the phase adjustment range of the phase table corresponds to five discrete sets, acting on the marked edge frequency point sets with set sizes of 16, 24, 32, 40, and 48 frequencies; the maximum single-step change of the rail pressure table is set to five levels, corresponding to 1, 2, 3, 4, and 5 rail pressure units; and the minimum continuous sampling number for the platform segment is set to five levels, corresponding to 256, 384, 512, 640, and 768 sampling points. The second part is the instruction format, using record-style entries. Each record includes the action type, starting sampling index, target coverage length or target frequency point set number, and intensity level number. The third section outlines the effective rules, specifying that within the same control window, the rail voltage table entries should be applied first, followed by the window shape table entries, and finally the phase table entries. This order ensures that the changes in the envelope tracking voltage rail are first fixed within the allowable range and rhythm, and then the peak values ​​are covered and the phases are rearranged. This prevents the initial baseband processing from being disrupted by subsequent rail voltage changes, thereby reducing the delay caused by repeated processing and lowering the risk of waveform tearing at the service frame boundaries.

[0024] The registration process for hard actuators uses the same three-stage description. The first stage is capability enumeration, specifying the types and intensity levels of actions that the hard actuator can perform. For example, the gas plug migration table sets four types of actions: extending the gas plug, shortening the gas plug, migrating towards the evaporation section, and migrating towards the condensation section; each type of action has five intensity levels, corresponding to different heating band on / off durations and microvalve opening durations. Taking extending the gas plug as an example, the on / off durations of the heating band on the outer side of the evaporation section for the five levels are 50 ms, 80 ms, 120 ms, 160 ms, and 200 ms, respectively, and the metering microvalve opening durations are 50 ms, 80 ms, 120 ms, 160 ms, and 200 ms, respectively; taking shortening the gas plug as an example, the on / off durations of the heating band on the outer side of the condensation section and the opening duration of the discharge microvalve use the same five-level setting. The second stage is the operation sequence format, where each record includes the action type, target partition number or target direction, intensity level number, and the minimum interval time between the previous migration. The third section defines the effective rules, which stipulate that the operation sequence must be started at the end of the control window to avoid short-term disturbances to sensor readings caused by soft actuator commands. It also stipulates that after each fine-tuning action, the position sensor status must be read and the next action can only proceed after a status change marker appears. If no status change marker appears, the same type of action is repeated according to the next duration level in the migration table until a status change marker appears or the end position of the control window for this frame is reached.

[0025] refer to Figure 3 , Figure 3This diagram illustrates the timing relationship between soft and hard actuators within a control window in one embodiment of the present invention. The horizontal axis represents time, and the vertical axis represents the action sequence of different actuators, demonstrating the start time, execution order, and mutual time constraints of each actuator within a 6ms control window in a 10ms service frame. Figure 3As shown, the service frame length is set to 10ms, and the control window is located in the middle to later part of the service frame, lasting for 6ms. The start time of the control window corresponds to the 2ms position on the time axis, and the end time corresponds to the 8ms position. The control window is marked in the figure with a dashed boundary to distinguish it from other time periods of the service frame. Within the control window, the actions of the soft actuator and the hard actuator are executed sequentially according to a preset time order, as follows: First, the rail voltage trajectory shaping action. This action starts no earlier than 0.5ms after the start of the control window, i.e., at the 2.5ms position on the time axis. At this moment, the system controller reads the rail voltage sequence in the isomorphic frame and shapes the rail voltage sequence according to the preset rail voltage table, including limiting the upper and lower limits of the rail voltage, constraining the maximum change in a single step, and inserting platform segments. The generation and submission of the rail voltage sequence are represented by a solid rectangle in the figure, and the width of the rectangle corresponds to the time window required for rail voltage processing. The purpose of rail voltage trajectory shaping is to limit the upper and lower boundaries and the change rhythm of the envelope tracking voltage rail, providing stable power supply conditions for subsequent baseband processing. Second, peak coverage. This action is initiated no earlier than 0.3ms after rail voltage submission. At this moment, the system controller reads the baseband sequence in the isomorphic frame and performs coverage processing on the detected peaks according to a preset window table. The coverage and submission of the baseband sequence are represented by a solid rectangle in the figure, which is located to the right of the rail voltage processing rectangle and connected by a diagonal arrow. The arrow direction indicates the timing relationship from rail voltage processing to peak coverage. The purpose of peak coverage is to weaken the peak amplitude, reduce the instantaneous conduction pressure of the power amplifier, and thus reduce the instantaneous heat input of device hotspots. Third, phase rearrangement. This action is performed after peak coverage is completed. The system controller calls the frequency domain transformation unit to convert the baseband sequence to the frequency domain, performs phase replacement on the marked edge frequencies according to a preset phase table, and then returns to the time domain through inverse transformation. The phase table replacement is represented by a solid rectangle in the figure, which is located to the right of the peak coverage rectangle and connected by a diagonal arrow. The purpose of phase rearrangement is to mitigate energy leakage at the frequency band edges and reduce additional heat generation and adjacent channel interference caused by out-of-band energy. Fourth, the plug migration action. This action is initiated no earlier than 2ms after baseband submission to reduce disturbance to soft actuator submission. At this moment, the system controller generates an operation sequence based on a preset plug migration table and sends the operation sequence to the plug drive unit. The initiation of the operation sequence is represented by a solid rectangle in the figure, which is located below the phase rearrangement rectangle. The two are connected by a vertical arrow, the direction of which indicates the timing relationship from phase rearrangement to plug migration. The purpose of plug migration is to reconstruct the thermal path by changing the length and position of the plug, allowing heat to bypass the congested path inside the device and flow to a more efficient area of ​​the heat sink, thereby achieving structural optimization of the thermal path. After the plug migration action is initiated, the system controller sequentially controls the on / off state of the heating band outside the evaporation section and the metering microvalve.The on / off state of the heating element is indicated by a gray rectangle in the diagram, while the opening / closing state of the micro-valve is indicated by a dark gray rectangle. They are arranged sequentially on the time axis. The on / off duration of the heating element and micro-valve is set according to the gear position in the airlock migration table, with a commonly used on / off duration of 80ms. After a fine-tuning action is completed, the system controller reads the position sensor status. If a status change marker is detected, the next action is initiated; otherwise, the same action is repeated after a minimum interval of 100ms. This minimum interval is indicated by text labels in the diagram. Figure 3 The diagram also illustrates the partitioned layout of the soft and hard actuators within the control window. The soft actuator action zone covers three actions: rail pressure trajectory shaping, peak coverage, and phase rearrangement, indicated by a light blue semi-transparent background in the diagram. The hard actuator action zone covers slug migration and its subsequent heating band and microvalve on / off actions, indicated by a light red semi-transparent background in the diagram. This partitioned layout visually demonstrates the complementary relationship between the soft and hard actuators in terms of time scale and target: the soft actuator completes baseband and rail pressure shaping within a millisecond-level control window, while the hard actuator completes stable slug migration within a time scale of tens to hundreds of milliseconds. At the end of the control window, the system controller sets a checkpoint to compare the temperature and heat flux samples at the end of the frame with those at the beginning of the frame to evaluate the synergistic effect of the soft and hard actuators within this frame. The checkpoint is indicated by a dashed arrow in the diagram, located at the end of the control window.

[0026] The system controller creates a unified control window and alignment reference for both soft and hard actuators. The service frame length is set to 10 milliseconds, the control window is located in the middle to later part of the service frame, and the duration is set to 6 milliseconds. At the beginning of the control window, the system controller uses the baseband start flag, the envelope tracking voltage rail control edge flag, the temperature and heat flux sensor time stamps, and the gas plug position state change flag as the alignment reference, and performs one-time delay compensation to ensure that the baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and gas plug state sequence in the isomorphic frame correspond point-to-point at the sampling position. This alignment method is adopted because the baseband peak, envelope power supply, and thermal path state at the same moment must be aligned in time before they can be included in the same judgment and command synthesis. Otherwise, expired peaks may be covered, or the next migration may be triggered before the gas plug has stabilized, resulting in a decrease in peak envelope ratio shaping effect or hindering thermal path reconstruction.

[0027] The immediate activation process after soft actuator registration is as follows. At the start of the control window, the system controller reads the isomorphic frame. Based on the thermal stress level and waveform characteristic detection results, it retrieves the soft actuator action combination from the thermal control mapping table and generates three types of records: rail pressure trajectory shaping record, peak coverage record, and phase rearrangement record. Taking a typical peak coverage as an example, the system controller detects the peak in the baseband sequence. The peak coverage area uses a window table entry with 128 sampling points. The initial sampling index is located 64 sampling points to the left of the peak, and the intensity level is selected as the third level. This selection has two considerations: first, matching the coverage area width with the peak width can reduce the peak value without expanding the peak's influence range, avoiding excessive stretching of adjacent symbols; second, placing the initial sampling index at the midpoint to the left of the peak allows for a gradual reduction in amplitude before the peak arrives, thus slowing the upward slope to the peak position, reducing the instantaneous conduction pressure of the power amplifier, and helping the thermal stress level to decrease rapidly within the same frame. For rail pressure trajectory shaping, the system controller selects the third level of maximum single-step change and the third level of shortest continuous sampling number for the plateau segment, and performs point-by-point scanning and replacement of the rail pressure sequence within the entire control window. For phase rearrangement, the system controller selects phase table entries acting on 32 frequency points, focusing on the set of frequency points marked by high-frequency occupancy indicators to reduce energy leakage from edge frequency points to adjacent bandwidths, thereby reducing unnecessary out-of-band energy and additional heat generation.

[0028] Figure 2This is a schematic diagram illustrating the effect of the peak-coverage window sequence described in this invention. The diagram uses the sampling points of the baseband signal as the horizontal axis and the signal amplitude as the vertical axis, comparing the waveform changes before and after peak coverage. The horizontal axis marks the sampling point positions from 0 to 320, with evenly distributed sampling point intervals. The vertical axis marks the normalized amplitude values, ranging from 0 to 1.0, with a scale interval of 0.2. The dashed curve in the diagram represents the original peak waveform. This waveform exhibits a significant peak near sampling point 128, with the highest amplitude approaching 1.0. The amplitude rapidly increases from sampling point 64 on the left side of the peak, reaching its peak value at sampling points 96 and 128, then gradually decreases at sampling points 160 and 192, returning to the baseline level of approximately 0.2 at sampling point 256. This peak waveform causes the power amplifier to experience significant instantaneous conduction pressure at the peak arrival time, resulting in concentrated heat input. The solid curve in the diagram represents the waveform after being covered by the window sequence. After coverage, the peak amplitude of the waveform is significantly reduced within the same sampling point range, with a maximum amplitude of approximately 0.6, a reduction of about 40% compared to the original peak. The coverage area starts at sampling point 64 and ends at sampling point 192, with a total length of 128 sampling points. Within the coverage area, the waveform exhibits a smooth parabolic shape without sharp abrupt changes. The starting point of the coverage area is chosen at sampling point 64 to the left of the peak. This design aims to begin reducing the amplitude before reaching the peak, gradually easing the conduction pressure on the power amplifier, rather than abruptly reducing it at the peak. The extent of the coverage area is marked with a double-headed arrow in the figure, clearly indicating that the coverage area length is equal to 128 sampling points. This length matches the width of the original peak, effectively covering the peak area to achieve peak reduction without over-extending to adjacent symbols and causing unnecessary waveform distortion. Figure 2 A table is plotted on the right side, listing the five settings for the window shape table. Setting 1 corresponds to a coverage length of 64 sampling points, suitable for narrower peaks; setting 2 corresponds to 96 sampling points, setting 3 to 128 sampling points, setting 4 to 160 sampling points, and setting 5 to 192 sampling points. The window shape sequences at different settings differ in center weight and edge smoothness. The system controller selects the appropriate setting based on the detected peak width to achieve optimal peak reduction and minimal waveform distortion. Figure 2 It is clear that the peak coverage technology modulates the baseband signal through a pre-designed window sequence, significantly reducing the peak envelope ratio while keeping the phase information unchanged. This reduces the instantaneous heat input of the power amplifier and creates favorable conditions for subsequent thermal control measures.

[0029] The immediate activation process after hard actuator registration is as follows. The system controller generates an operation sequence for gas plug migration at the end of the same control window, and controls the on / off states of the heating bands on the outer sides of the evaporation and condensation sections, as well as the metering and discharge microvalves, item by item according to the sequence. Taking a typical extended gas plug as an example, the system controller selects the third setting for on / off and on / off duration. First, it activates the heating band on the outer side of the evaporation section for 120 milliseconds, making it easier for the working fluid near the evaporation section to transfer heat and allowing the leading edge of the non-condensable gas to advance towards the condensation section. Then, it activates the metering microvalves for 120 milliseconds to introduce a small amount of non-condensable gas, increasing the gas plug length. Immediately after the action is completed, the position sensor status is read. When a status change marker for the next zone is detected, it confirms that the leading edge of the gas plug has crossed the zone boundary, and then closes the heating bands and metering microvalves. The advantage of this sequence is that by first creating a temperature difference through the heating bands and then introducing non-condensable gas, the leading edge of the gas plug can advance stably with a smaller gas volume, reducing disturbances to the pipe pressure. This helps maintain a stable switching between the dominant and bypass pathways between the evaporation and condensation sections, thus more quickly forming a combination of thermal bypass and thermal limiting channels that are beneficial to the hot spot.

[0030] The registration process includes explicit interlocks and safety constraints. Constraints on the soft actuator side include: rail pressure trajectory shaping must be performed before peak coverage and phase rearrangement; peak coverage and phase rearrangement must not overlap the same sampling position; if overlapping areas occur, the earlier record prevails; the maximum single-step change in rail pressure must not be exceeded at any two adjacent sampling points to avoid overshoot in the envelope tracking power supply. Constraints on the hard actuator side include: the metering microvalve and the discharge microvalve must not be opened simultaneously at any time; the outer heating belts of the evaporation and condensation sections must not be opened simultaneously in the same action; after each fine-tuning action, the position sensor status must be read and the status change marker confirmed; the next action cannot proceed without confirmation; the minimum interval between one fine-tuning action and the next is no less than 100 milliseconds to ensure the length and position of the gas plug remain stable within the fluid. Constraints across actuators include: the hard actuator's action start time within the same control window must be no less than 2 milliseconds later than the soft actuator's baseband and rail pressure submission time to avoid electromagnetic interference caused by heating belt switching affecting the immediate submission of baseband processing results.

[0031] After registration, the system controller establishes independent checkpoints for soft and hard actuators. The checkpoint for soft actuators is located at the end of the control window, where real-time temperature and heat flux samples are read and compared with the samples at the beginning of the control window. If the thermal stress level drops to a lower level or remains at an acceptable level, the peak coverage record, phase rearrangement record, and rail pressure trajectory shaping record generated in this frame are marked as reusable files for rapid reuse when similar waveform characteristics appear in the next frame, reducing command synthesis latency. The checkpoint for hard actuators is located after each fine-tuning action, reading the position sensor status. If a status change marker is detected, the next operation sequence is executed; otherwise, the next duration setting of the same action is called and repeated. When the control window of the current frame ends, if no status change marker appears, the remaining operation sequence is executed in the next frame, and the temperature and heat flux samples are checked again at the beginning of the control window of the next frame to confirm whether the thermal path has moved towards the target combination.

[0032] The tables upon which the registration depends provide directly usable numerical default configurations. The window table contains 5 window sequences by default, corresponding to coverage lengths of 64, 96, 128, 160, and 192 sampling points. The center weight of each sequence increases smoothly from weak to strong to achieve similar peak reduction effects under different peak widths, while avoiding abrupt changes at the edges of the coverage area. The phase table contains 5 target phase sequences by default, applied to 16, 24, 32, 40, and 48 marked frequency points respectively. The phase difference between adjacent frequency points is constrained by a fixed set of differences between entries, ensuring that the replaced spectrum gradually decreases at the frequency band edges in a stepwise manner, without generating new sharp spikes. The rail pressure table contains 5 levels by default, with maximum single-step changes of 1, 2, 3, 4, and 5 rail pressure units respectively, and minimum continuous sampling numbers for the plateau segment of 256, 384, 512, 640, and 768 sampling points respectively, used to balance output back-off speed and waveform fidelity. The gas plug migration table divides the heat pipe into 8 zones by default, with the zone numbers increasing sequentially from the evaporation section to the condensation section. Each of the four types of actions contains 5 intensity levels, and the corresponding heating band on / off time and micro-valve on / off time are as described above. The minimum interval time is 100 milliseconds by default.

[0033] The digital predistortion tap was chosen as a soft actuator because an excessively high peak envelope ratio would cause high-density instantaneous heat input during the power amplifier's conduction period. Peak coverage can weaken the peak in advance within the same frame, allowing the power amplifier's conduction pressure to drop before the peak arrives, directly reducing the instantaneous temperature rise of hot spots. The envelope tracking voltage rail was chosen as a soft actuator because rail voltage trajectory shaping can limit the upper and lower boundaries and variation rhythm of the power rail without changing the baseband sign decision result, making the output back-off execution more controllable, thereby reducing continuous heat input. The plug drive unit of the controllable non-condensing plug heat pipe was chosen as a hard actuator because by changing the plug length and position, the dominant and bypass paths between the evaporation and condensation sections can be rearranged, allowing heat to bypass congested paths in the device body and flow to areas of higher heat sink efficiency, achieving structural optimization of the thermal path. Soft actuators can complete the shaping of baseband and rail pressure within a millisecond-level control window, while hard actuators can complete the stable migration of gas plugs within a timescale of tens to hundreds of milliseconds. The two complement each other in terms of time scale and target, resulting in a perceptible improvement in thermal stress level within the same service frame and further stabilization in subsequent frames.

[0034] In an optional implementation, the service frame length can be set to 5 milliseconds, and the control window duration can be set to 3 milliseconds, suitable for applications more sensitive to latency. The window table can be configured with five settings covering 48, 72, 96, 120, and 144 sampling points to achieve more compact peak coverage at shorter frame lengths. The phase table can be configured with five settings operating on 12, 20, 28, 36, and 44 frequency points to reduce overall changes to the frequency domain edges. The maximum single-step change of the rail pressure table can be set to 1, 2, 2, 3, and 3 rail pressure units to reduce rail pressure jitter within short frames. The number of partitions in the gas plug migration table can be adjusted to 6 partitions to shorten the necessary time for a single cross-partition migration. Combined with the shortened on / off time of each heating band and the micro-valve opening time, a fine-tuning can be completed within 80 milliseconds.

[0035] In another optional implementation, the service frame length remains 10 milliseconds, but the control window is divided into two segments, each lasting 3 milliseconds, with a 2-millisecond interval between them. The first segment is used to generate and submit the rail pressure trajectory shaping record for the soft actuator, and the second segment is used to generate and submit the peak coverage record and phase rearrangement record, while simultaneously initiating the hard actuator operation sequence at the end of the second segment. This segmented approach allows for fixing the rail pressure within the same frame first, and then making more precise adjustments to the peak coverage and phase rearrangement based on the fixed rail pressure, while also providing a clean start-up time for the hard actuator and reducing interference with the soft actuator. When using this approach, the window shape table and phase table can remain unchanged with the aforementioned five settings. It is recommended to select the second or third setting for the maximum single-step change in the rail pressure table to avoid abrupt changes in rail pressure between the first and second segments.

[0036] In one embodiment, the registration process also includes error detection and rollback strategies. On the soft actuator side, if an out-of-bounds alarm for the envelope tracking power supply occurs after rail pressure trajectory shaping is completed, the system controller automatically lowers the maximum single-step change by one level within the same control window and resubmits the rail pressure record. If, after peak coverage is completed, an overlap of adjacent peak coverage areas is detected, causing the coverage area length to exceed the maximum value of the window table, the system controller merges two adjacent peaks into a wider coverage area and selects the next longer window table entry for re-coverage. If, after phase rearrangement is completed, an energy mutation at the frequency domain edge is detected exceeding a preset threshold, the system controller removes the replacements of several outermost marked frequency points within the same control window, retaining the replacements of the inner frequency points. On the hard actuator side, if no position sensor state change marker is detected after a fine-tuning action, the system controller increases the duration of the same action by one level and executes it again after the minimum interval time is reached. If no state change marker is detected even after increasing to the highest level, the system controller initiates a small-amplitude reverse action in the next frame to release the stress within the tube, then returns to the original direction to continue migration until a state change marker is detected.

[0037] In one embodiment, after registration, the system controller establishes a one-to-one correspondence between soft actuator action combinations, hard actuator action combinations, and thermal stress levels. For example, when the thermal stress level is level one, only the first-level platform segment limitation of the rail pressure gauge is activated, without peak coverage, phase rearrangement, or gas slug migration; when the thermal stress level is level two, the second level of the rail pressure gauge is activated, and the second level of the window table is applied to the detected peaks, without phase rearrangement, and gas slug migration is selected to migrate towards the evaporation section in level one; when the thermal stress level is level three, the third level of the rail pressure gauge is activated, the third level of the window table is applied to the peaks, the third level of the phase table is applied to the marked frequency point set, and gas slug migration is selected to extend the gas slug. The second setting is as follows: When the thermal stress level is at the fourth setting, the fourth setting of the rail pressure gauge is activated, the fourth setting of the window table is applied to the peak, and the fourth setting of the phase table is applied to the marked frequency set. For the gas plug migration, the third setting of the gas plug is extended, and a first setting of migration towards the condensation section is added in the next frame. When the thermal stress level is at the fifth setting, the fifth setting of the rail pressure gauge is activated, the fifth setting of the window table is applied to the peak, and the fifth setting of the phase table is applied to the marked frequency set. For the gas plug migration, the fourth setting of the gas plug is extended, and a second setting of migration towards the condensation section is added at the end of the same frame. This correspondence allows the system controller to retrieve a clear action combination within the same frame based on the isomorphic frame and waveform feature detection results, reducing the complexity of ad-hoc judgments and shortening the time from sensing to execution.

[0038] Figure 4This is a comparative curve of the heat pipe temperature distribution before and after the gas plug migration described in this invention. The graph uses the heat pipe's partition location as the horizontal axis and temperature as the vertical axis to demonstrate the improvement effect of the gas plug migration operation on the heat pipe temperature distribution. The horizontal axis marks the seven partitions of the heat pipe, from left to right: partition 1 to partition 7, corresponding to different locations in the evaporation section, adiabatic section, and condensation section, respectively. The vertical axis marks the temperature values ​​in degrees Celsius (°C), ranging from 20°C to 80°C, with increments of 10°C. The dashed curve in the graph represents the temperature distribution before the gas plug migration. The curve shows that the temperature in partition 1 (the starting position of the evaporation section) is as high as approximately 70°C, because this area is in direct contact with the heating element, resulting in continuous heat input. As the position moves towards the condensation section, the temperature shows a slow decreasing trend: partition 2 is approximately 72°C, partition 3 is approximately 55°C, partition 4 is approximately 45°C, partition 5 is approximately 38°C, partition 6 is approximately 34°C, and partition 7 is approximately 32°C. This temperature distribution indicates that the hotspot temperature in the evaporation section is very high, while the heat transfer efficiency to the condensation section is low, indicating a significant temperature gradient and heat transfer bottleneck within the heat pipe. The location of the gas plug before migration is marked with a gray rectangle in the figure; this gas plug covers the area from zone 3 to zone 4. The two vertical dashed lines show that the leading edge of the gas plug is near the boundary between zone 3 and zone 4, and the trailing edge is near the boundary between zone 2 and zone 3. The presence of the gas plug obstructs the circulation of the working fluid in this area, making it difficult for heat from the evaporation section to be quickly transferred to the condensation section, resulting in a persistently high temperature in the evaporation section. The solid curve in the figure represents the temperature distribution after the gas plug migration. The migration operation involves activating the heating band on the outside of the evaporation section and the metering micro-valve, pushing the gas plug towards the condensation section, moving its position from the original zone 3-4 area to the zone 4-5 area. The location of the gas plug after migration is marked with a gray rectangle at the bottom of the figure; the leading edge of the gas plug has moved to the vicinity of the boundary between zone 5 and zone 6, and the trailing edge has moved to the vicinity of the boundary between zone 4 and zone 5. The temperature curves after migration show that the temperature in zone 1 decreased from 70℃ to approximately 50℃, a drop of 20℃. This significant temperature reduction is marked with a double-headed arrow "ΔT↓". The reason for the temperature decrease is that after the gas plug moved towards the condensation section, a new heat transfer path was established between the evaporation and condensation sections. The working fluid in the evaporation section can more smoothly carry heat to the condensation section, reducing heat accumulation in the evaporation section. The temperatures at other locations on the curves after migration are: zone 2 approximately 52℃, zone 3 approximately 54℃, zone 4 approximately 58℃, zone 5 approximately 62℃, zone 6 approximately 65℃, and zone 7 approximately 68℃. The overall temperature distribution is more uniform, and the temperature gradient between the evaporation and condensation sections is reduced. The three functional regions of the heat pipe are labeled below the figure: zones 1-2 belong to the evaporation section, zones 3-5 belong to the adiabatic section, and zones 6-7 belong to the condensation section.By observing the two curves, it can be found that the gas plug migration operation changes the dominant heat transfer path. The hot spot temperature of the evaporation section is effectively controlled, and the temperature of the condensation section rises slightly, indicating that more heat is transferred to the condensation section and dissipated to the outside, thus improving the heat transfer efficiency of the entire heat pipe system. Figure 4 The invention visually demonstrates the technical effect of reconstructing the thermal path by actively controlling the position of the gas plug, and verifies that the gas plug migration operation can significantly improve the temperature distribution of the heat pipe and reduce the hot spot temperature of the device within a time scale of tens to hundreds of milliseconds, thus achieving the goal of adaptive thermal control.

[0039] Example 2: During continuous operation of the service waveform, the system controller performs the following closed-loop operation based on the data acquired in each service frame within the control window: It acquires and aligns baseband sampling, rail pressure sampling, temperature and heat flux sampling, and slug position status to generate a homogeneous frame containing a baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and slug state sequence; based on the homogeneous frame, it derives the thermal stress level characterizing the current state; based on the thermal stress level and combined with the waveform feature detection results of the baseband sequence, it retrieves the soft actuator action combination and hard actuator action combination from the preset thermal control mapping table; based on the soft actuator action combination, it synthesizes the baseband sequence and rail pressure sequence used for shaping the peak envelope ratio and implementing output backoff; based on the hard actuator action combination, it synthesizes the operation sequence used for migrating the slug to reconstruct the thermal path; within the control window, it sends the synthesized baseband sequence, rail pressure sequence, and operation sequence to the transmit link, envelope tracking power supply, and slug drive unit for execution, respectively.

[0040] refer to Figure 1 , Figure 1 This is a schematic diagram illustrating the structural principle of the controllable non-condensing plug heat pipe described in this invention. Figure 1As shown, the heat pipe is divided into three functional regions along its length: an evaporation section, an adiabatic section, and a condensation section. The evaporation section, located on the left side of the heat pipe, receives heat from the heating element; the adiabatic section, located in the middle of the heat pipe, acts as a transition for heat transfer; and the condensation section, located on the right side of the heat pipe, releases heat to the radiator. Inside the heat pipe, the working fluid is distributed in the evaporation and condensation sections. In the adiabatic section region between the evaporation and condensation sections, there is a gas plug formed by non-condensable gas. This gas plug is elliptical in shape, with its leading edge pointing towards the condensation section and its trailing edge pointing towards the evaporation section. The presence of the gas plug divides the working fluid into two discontinuous regions, thus allowing the heat conduction path between the evaporation and condensation sections to be altered by adjusting the length and position of the gas plug. To achieve active control of the gas plug, this invention provides an evaporation section heating strip on the outside of the evaporation section and a condensation section heating strip on the outside of the condensation section. The evaporation section heating strip is in close contact with the outer wall of the evaporation section and can create a temperature gradient near the evaporation section by applying electricity, pushing the leading edge of the gas plug towards the condensation section. The heating band in the condensing section is in close contact with the outer wall of the condensing section. By applying electricity, a reverse temperature gradient can be created near the condensing section, pushing the gas plug backward towards the evaporating section. A gas control device, including a metering microvalve and a discharge microvalve, is installed above the heat pipe. The metering microvalve is located above the adiabatic section, with its outlet connected to the inner cavity of the heat pipe. It is used to introduce a small amount of non-condensable gas into the heat pipe to increase the gas plug length. The discharge microvalve is also located above the adiabatic section, with its inlet connected to the inner cavity of the heat pipe. It is used to discharge some non-condensable gas from the heat pipe to reduce the gas plug length. Both the metering and discharge microvalve are electromagnetically driven, enabling precise control of the on / off time. Multiple position sensors are evenly distributed along the length of the bottom of the heat pipe to detect the zone position of the leading and trailing edges of the gas plug. In this embodiment, the heat pipe is divided into 8 zones, with zone numbers increasing sequentially from the evaporating section to the condensing section. Each position sensor corresponds to a zone boundary. When the leading or trailing edge of the gas plug crosses a zone boundary, the corresponding position sensor outputs a status change marker. By reading the status of multiple position sensors, the system controller can accurately determine the current zone range covered by the gas plug. The diagram also indicates the heat flow direction and heat dissipation direction. In the evaporation section, heat flows from an external heat source into the working fluid, with the heat flow direction from left to right towards the inside of the heat pipe. In the condensation section, heat flows from the working fluid to the radiator, with the heat dissipation direction from the inside of the heat pipe to the right towards the external environment. By adjusting the position and length of the gas plug, the effective heat transfer path between the evaporation and condensation sections can be altered, allowing heat to flow along the path of least resistance, thereby achieving adaptive reconfiguration of the heat path.

[0041] In one specific implementation, the service frame and control window are set using a fixed-length frame and a fixed window. Taking a common implementation as an example, the service frame length is set to 10 milliseconds, the control window is located in the middle to late part of the frame's time axis, and its duration is set to 6 milliseconds. The start and end points of the window are jointly determined by the baseband start flag and the envelope tracking voltage rail control edge flag. This location can cover the main active range of the peak envelope ratio and the key modulation range of the envelope tracking voltage rail within the same frame, while reserving a buffer time at the end for the start of the hard actuator, reducing the risk of interference to baseband and rail voltage submission. In an optional implementation, the service frame length can be set to 5 milliseconds, and the control window duration can be set to 3 milliseconds for scenarios more sensitive to latency; in another optional implementation, a 6-millisecond control window is split into two 3-millisecond windows with a 2-millisecond interval in between, so as to decouple rail voltage submission from peak processing.

[0042] The process of acquiring and aligning baseband sampling, rail pressure sampling, temperature and heat flux sampling, and plug position status to generate isomorphic frames containing baseband, rail pressure, temperature, heat flux, and plug state sequences is executed in the following order. First, at the same time reference at the start of the control window, four types of acquisition are triggered simultaneously: baseband sampling comes from the baseband input and feedback paths of the digital predistortion tap, with a sampling resolution of 2048 sampling points per frame; rail pressure sampling comes from the rail pressure detection points of the envelope tracking voltage rail, with a sampling resolution of 512 sampling points per frame; temperature and heat flux sampling comes from several temperature and heat flux sensors deployed at device hotspots, heat sink inlets, and heat sink outlets, with a sampling resolution of 64 sampling points per frame for each temperature and heat flux; and plug position status comes from the leading and trailing edge position sensors of the controllable non-condensing plug heat pipe, with the position status represented by partition numbers and state change markers. Secondly, due to the different resolutions of the four types of sampling, the system controller uses linear insertion or nearest neighbor methods to pad or dilute the sequences without changing the total length within the frame. This ensures that each of the five sequences has a corresponding value at every sampling position, and the padded or diluted trajectories are stored in a buffer along with the original timestamps. Thirdly, after alignment, the system controller assembles the five sequences into a homogeneous frame with a fixed format: baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and gas plug state sequence, arranged from front to back according to sampling position. The reason for using this homogeneous frame format is that the baseband peak, envelope power supply, instantaneous states of temperature and heat flux, and gas plug position state need to be strictly aligned at the sampling positions. Only in this way can the decisions at the same sampling position simultaneously reflect waveform behavior and thermal path behavior, thereby avoiding the overwriting of outdated peaks or triggering the next migration before the gas plug has stabilized.

[0043] Based on the isomorphic frame, the step of deriving the thermal stress level characterizing the current state is completed using a combination of label mapping and table lookup. The system controller first reads the relative magnitudes of different measurement points in the temperature sequence, assigning a direction marker indicating whether temperature rise or fall is dominant at each sampling location. Then, it reads the heat flow direction and intensity level in the heat flow sequence, determining at each sampling location whether the heat flow is closer to the evaporation section or the condensation section. Simultaneously, it reads the partition number and state change marker in the gas plug state sequence, obtaining the partition combination of the gas plug's leading and trailing edges. The data from these three dimensions are combined at each sampling location to form a thermal path state label, indicating the combination of dominant, bypass, and limiting paths between the evaporation and condensation sections. Subsequently, the system controller inputs the thermal path state label, combined with the temperature and heat flow level, into the thermal control mapping table, where it looks up the thermal stress level for the current frame. Taking a common implementation as an example, the thermal stress level is divided into five levels. The first level indicates that heat mainly flows smoothly to the condensation section through the main path and the airlock is stable. The third level indicates that the main path is congested and the bypass has not been fully established. The fifth level indicates that neither the main path nor the bypass is sufficient to handle the current heat and the proportion of the restricted channel is high. The reason for adopting this derivation method is that temperature and heat flow are direct reflections of the thermal state, and the location of the airlock partition determines the effective channel structure of the heat pipe. Combining the three at the same sampling location can accurately reflect the thermal stress state of the current frame.

[0044] Based on the thermal stress level and combined with the waveform feature detection results of the baseband sequence, the specific process for retrieving the soft actuator action combination and hard actuator action combination from the preset thermal control mapping table is as follows. The system controller performs three types of directly achievable detections on the baseband sequence of the isomorphic frame: First, sliding detection to locate peaks. The sliding window length is related to the coverage length of the window table, with a commonly used length of 128 sampling points. The judgment condition is that a peak is marked when the center position of the window is continuously higher than the positions on both sides reaching a preset number of samples. Second, counting the number of amplitude changes. Within a fixed-length buffer, the number of amplitude increases and decreases is counted and summed to obtain the frequency of change. Third, the baseband sequence and the original phase information are passed to a frequency domain transformation unit to obtain the amplitude of several edge frequency points. If the amplitude of the edge frequency points is above a preset threshold, it is marked as high-frequency occupancy. The system controller uses the thermal stress level, peak density, frequency of change, and high-frequency occupancy as search keys to obtain the soft actuator action combination and hard actuator action combination from the thermal control mapping table. Taking a fast-response configuration as an example, when the thermal stress level is at level three and the peak density and frequency of change are both moderate, with high-frequency occupancy present, the soft actuator action combination is level three for rail pressure trajectory shaping, level three for peak coverage, and level three for phase rearrangement; the hard actuator action combination is level two for extending the gas plug, and in the next frame, level one for migrating towards the condensation section is added. The consideration for adopting this retrieval method is that the thermal stress level determines the baseline of processing intensity, and the waveform characteristics determine the focus of processing; using both together can avoid over-processing or omission of processing.

[0045] Based on the soft actuator action combination, the steps of synthesizing the baseband sequence and rail voltage sequence used for shaping the peak envelope ratio and implementing output backoff are completed in stages within the control window. The system controller first performs rail voltage trajectory shaping to ensure that subsequent peak coverage and phase rearrangement are performed under stable rail voltage conditions. Taking the third gear as an example, the upper and lower limits given in the rail voltage table define the allowable range of rail voltage, with a maximum single-step change of 3 rail voltage units and a minimum continuous sampling number of 512 sampling points for the plateau segment. The system controller scans point by point from the first sampling position in the control window, replacing sampling values ​​higher than the upper limit or lower than the lower limit with the corresponding limit, and limiting the change between two adjacent sampling positions to no more than 3 rail voltage units. After any rise or fall, a plateau segment of no less than 512 sampling points is inserted. The direct effect of this is to allow the envelope tracking power supply to implement output backoff within a controllable rhythm, avoiding overshoot and oscillation. Subsequently, the system controller performs peak coverage on the baseband sequence. Taking the third level as an example, the window table provides a window sequence with a coverage length of 128 sampling points. For each located peak, the system controller starts replacing the amplitude of the coverage area point by point according to the window sequence, beginning at the 64 sampling points to the left of the peak. The phase remains unchanged during the replacement process. If the coverage areas of adjacent peaks overlap, they are handled according to the principle of first-come, first-served coverage, with later coverage only effective at the uncovered sampling positions. The direct significance of this approach is that matching the coverage area width with the peak width can reduce the peak envelope ratio without expanding the peak's influence range, reducing the instantaneous heat input at device hotspots to a level controllable by the soft actuator. If a high-frequency occupancy is marked, the system controller then performs phase rearrangement. Taking the third level as an example, this applies to 32 marked frequency points, replacing them according to the target phase value of the corresponding entry in the phase table. After the replacement is completed, the frequency domain transformation unit returns to the time domain. The purpose of rearrangement is to mitigate energy leakage at the frequency band edges, thereby reducing additional heat generation and adjacent channel interference caused by out-of-band energy.

[0046] Based on the hard actuator action combination, the steps of synthesizing the operation sequence for migrating the gas plug to reconstruct the thermal path are initiated at the end of the control window to reduce disturbances to the soft actuator submission. Taking the second setting of extending the gas plug as an example, the operation sequence includes three instructions: first, open the heating strip on the outside of the evaporation section for 80 milliseconds to create a temperature difference near the evaporation section that is conducive to the advancement of the leading edge; then, open the metering microvalve for 80 milliseconds to introduce an appropriate amount of non-condensable gas, thereby increasing the length of the gas plug; finally, read the position sensor status. If a status change marker for the next zone is detected, close the heating strip and the metering microvalve and record the new zone number; if no status change marker is detected, repeat the operation once according to the next setting of the same action after the minimum interval time of 100 milliseconds is reached. If the hard actuator action combination requires migration towards the condensation section, the operation sequence will switch the heating strip on and off on the outside of the condensation section according to a preset combination to create a temperature gradient that is conducive to the movement of the trailing edge. The advantage of this sequence is that by first establishing the driving force for migration through thermal means, and then changing the gas plug geometry with a small amount of non-condensable gas, stable leading or trailing edge displacement can be obtained with less in-tube disturbance. This is beneficial for reconstructing the dominant path and bypass in a short time, allowing heat to flow to the more efficient area of ​​the radiator more quickly.

[0047] Within the control window, the synthesized baseband sequence, rail pressure sequence, and operation sequence are respectively sent to the transmit link, envelope tracking power supply, and plug drive unit, with the execution steps following interlocking and sequential control. The system controller first submits the shaped rail pressure sequence within the control window, with the submission point no earlier than 0.5 milliseconds after the window begins. Subsequently, it submits the baseband sequence after peak coverage and phase rearrangement processing, with the submission point no earlier than 0.3 milliseconds after the rail pressure submission point, to ensure that rail pressure changes have entered the plateau phase. Finally, at the end of the control window, it submits the operation sequence to trigger plug migration, with the start point no earlier than 2 milliseconds after the baseband submission point. The position sensor status is immediately read for confirmation after each fine-tuning action. This sequential arrangement reduces interference from power rail transients and heating strip on / off states on the baseband submission, while ensuring stable output of peak and frequency domain edge processing results.

[0048] To complete the closed loop within the same service frame, the system controller sets a checkpoint at the end of the control window, comparing the temperature and heat flux samples at the end and beginning of the frame. If the thermal stress level drops to a lower level or remains at an acceptable level, the system controller marks the soft actuator action combination and hard actuator action combination used in this frame as reusable, so that they can be directly called when similar waveform characteristics and thermal path states appear in the next frame, reducing command synthesis latency. If the thermal stress level does not meet the target, the system controller adds a shortest-duration airlock fine-tuning action at the end of this frame, or extends the rail pressure platform segment by a fixed length, or adds a coverage to the most prominent peak. In common implementations, the added rail pressure platform segment length is 256 sampling points, and the added peak coverage length is 64 sampling points.

[0049] In an optional implementation, waveform feature detection and command synthesis can be performed in segments. In a dual-window configuration, the first window is used only for rail voltage trajectory shaping and submission, while the second window is used for peak coverage, phase rearrangement, and hard actuator activation. This approach first fixes the allowable range of the envelope tracking voltage rail within the same frame, then performs fine-tuning of the baseband on the stable rail voltage, and finally triggers plug migration, reducing the mutual influence between the three types of actions. In another optional implementation, the number of temperature and heat flux sampling points is increased to 128 sampling points per frame to improve the spatial resolution of the thermal path status label; this is more advantageous on devices with larger heat sink areas and more complex hotspot distributions, enabling faster determination of the switching timing between the dominant path and the bypass, and making the selection of hard actuator action combinations closer to the actual heat dissipation pattern.

[0050] Through the above implementation, the isomorphic frames of the current frame truly reflect the baseband behavior, envelope power supply, temperature and heat flux status, and gas plug geometry. The derivation of the thermal stress level and the retrieval of the thermal control mapping table can provide clear and executable soft actuator action combinations and hard actuator action combinations within the same control window. The baseband sequence and rail pressure sequence, constructed through rail pressure trajectory shaping, peak coverage, and phase rearrangement, take effect immediately within the current frame. The operation sequence, constructed by extending the gas plug, shortening the gas plug, or shifting the gas plug direction, is initiated within the current frame and continues across frames if necessary. Soft actuators reduce the peak envelope ratio and implement output backoff within milliseconds, while hard actuators reconstruct the thermal path within tens to hundreds of milliseconds, adaptively guiding heat to areas with higher radiator efficiency. The complementarity of the two types of actuators in terms of time and object ensures that the improvement in the thermal stress level can be observed within the same service frame and further stabilized in subsequent service frames.

[0051] Example 3: The system controller performs three types of detection on the baseband sequence within the same service frame. These three types of detection can be enabled individually or in combination. The first type is sliding detection to obtain the peak density sequence. Taking a common setting as an example, the sliding window length is 128 sampling points, and the step size is 8 sampling points. If the center position of the window is continuously higher than three adjacent sampling points on each side of the window, and the magnitude of the increase meets a threshold (the threshold can be set to the value in the top 10 percentile after sorting the amplitude within the window), then a peak is marked at that center position. The system controller counts the number of peaks within a fixed interval of 256 sampling points and outputs them in interval order, thus obtaining the peak density sequence. This approach has two direct benefits: first, the small step size allows for the capture of adjacent peaks without omission; second, counting within a fixed interval preserves both the concentration and location of the peak distribution, facilitating subsequent targeted selection of window table entries. The second type is counting the number of amplitude changes to obtain the frequency of change sequence. The system controller compares the amplitude relationships of adjacent samples point by point within a fixed-length buffer of 256 sampling points. An increase is counted once, a decrease once, and a flat value is not counted. The sum of these two counts yields the number of changes in the buffer, which is then output in buffer order as the frequency of change sequence. A high frequency of change interval indicates fast symbol transitions and numerous transients; suppressing peaks and extending plateau segments beforehand can effectively reduce the power amplifier's conduction pressure. The third type involves transforming to the frequency domain to extract edge frequency energy to obtain a high-frequency occupancy indication sequence. The system controller calls the frequency domain transformation unit to convert the baseband sequence and phase information into amplitude ratios of several discrete frequency points. A set of frequency points near the bandwidth edge is selected as the marked frequency point set, with a maximum size of 32 frequency points. When the amplitude of any frequency point in the set exceeds 1.5 times the median amplitude within the set, high-frequency occupancy is marked within the corresponding time-domain sampling range. Aligning the time-domain intervals with the high-frequency occupancy marks yields the high-frequency occupancy indication sequence. This approach transforms the risk of energy leakage at the frequency band edge into a time-domain location marker without altering the data structure. Consequently, only the marked interval is processed in the subsequent phase rearrangement gate, resulting in a smaller processing range and less disturbance to the main waveform.

[0052] Example 4: The system controller simultaneously triggers and buffers four types of data acquisition at the start of the control window using the same time base: baseband sampling from digital predistortion taps, with a common resolution of 2048 sampling points per frame; rail pressure sampling from the envelope tracking voltage rail, with a common resolution of 512 sampling points per frame; temperature and heat flow sampling from sensors at the device hotspot, radiator inlet, and radiator outlet positions, with a common resolution of 64 sampling points each per frame; and plug position status from the leading and trailing edge position sensors of the controllable non-condensing plug heat pipe, represented by partition numbers and state change markers. Since the four types of sampling have different resolutions, the system controller performs a two-step alignment to achieve point-to-point correspondence at the same sampling location. The first step is length alignment: using the 2048 sampling points of the baseband sampling as the target length, rail pressure sampling is linearly interpolated to fill to 2048 sampling points, temperature and heat flow sampling is repeated in a nearest-neighbor manner to fill to 2048 sampling points, and plug position status is filled to fill to 2048 sampling points based on the most recent state. The second step is time alignment: The time difference between the control edge flag and the baseband start flag of the envelope tracking voltage rail is read and applied as a uniform delay compensation to the sampling index of the rail pressure sequence. Simultaneously, the time stamps for temperature and heat flux sampling are aligned to the start of the control window. After completion, the five sequences are concatenated into a homogeneous frame according to the sampling position order, with a fixed format: baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and gas plug state sequence. The core consideration in this approach is that only by allowing the same sampling position to simultaneously describe the waveform, power supply, and thermal path states can subsequent judgments and executions avoid overwriting expired peaks or re-migrating before the gas plug has stabilized, thus preventing invalid operations and additional thermal disturbances. In an optional implementation, the target length can be set to 1024 or 4096 sampling points per frame. Choosing 4096 sampling points improves the precision of peak location when the device's baseband sampling rate is higher; choosing 1024 sampling points reduces buffer pressure when device resources are limited.

[0053] Example 5: The system controller takes the temperature sequence, heat flow sequence, and gas plug state sequence from the isomorphic frame as input, and provides a thermal path state label for each sampling location and obtains the thermal stress level by looking up a table. Specifically, at each sampling location, the temperature relationship between the device hotspot and the radiator inlet is compared. If the device hotspot temperature is more than 2 degrees Celsius higher than the radiator inlet temperature, it is marked as hotspot dominant; otherwise, it is marked as balanced. The heat flow direction indication is read. If the heat mainly points to the condensation section, it is marked as condensation-oriented dominant; otherwise, it is marked as evaporation-oriented dominant. The partition numbers of the leading and trailing edges of the gas plug are read. For example, if the heat pipe is divided into 8 partitions, a combination of two partition numbers is obtained. The system controller merges the three types of information into a thermal path state label. For example, "hotspot dominant, condensation-oriented dominant, partitions 3 to 5" indicates that the dominant path is operating in the condensation-oriented direction and the gas plug covers the area from the end to the middle of the evaporation section. The thermal control mapping table is then accessed. A possible example is: when "hot spot dominant, condensation section dominant, zones 3 to 5" appears and the heat flow intensity is moderate, the thermal stress level is level three; when "hot spot dominant, evaporation section dominant, zones 2 to 4" appears and the heat flow intensity is low, the thermal stress level is level four; when "balanced, condensation section dominant, zones 5 to 7" appears, the thermal stress level is level two. In an optional implementation, the temperature difference threshold can be set to 1 degree Celsius or 3 degrees Celsius, the number of heat pipe zones can be 6 or 10, and the mapping entries can be adjusted accordingly.

[0054] Example 6: The system controller performs sliding detection to locate the peak position on the baseband sequence and selects a window sequence matching the peak width for coverage replacement for each peak. Taking a common setting as an example, the window table contains five window sequences with coverage lengths of 64, 96, 128, 160, and 192 sampling points. The peak width can be obtained by the position difference between the first descent of the left and right sides of the peak to half the height. If the width falls between 96 and 160, a window sequence with a coverage length of 128 is selected. The coverage starts at half the coverage length on the left side of the peak, for example, starting at 64 sampling points on the left side of the peak. Then, the baseband amplitude is replaced point by point along the coverage area with the replacement value in the window sequence, keeping the phase unchanged during the replacement process. When adjacent peak coverage areas overlap, the replacement of the area that arrives first takes priority, and the replacement of the area that arrives later is only performed at the unreplaced positions. The consideration for setting the coverage start point to the left of the peak is that the amplitude can be gradually reduced before the peak is reached, allowing the power amplifier to reduce conduction pressure in the rising phase, thereby reducing instantaneous heat input. Matching the coverage length to the peak width ensures that the coverage area is neither too short, resulting in insufficient peak attenuation, nor too long, causing unnecessary stretching of adjacent symbols. In an optional implementation, the coverage start point can be set at the 40th or 60th percentile of the coverage length to the left of the peak to accommodate waveforms with different rising slopes.

[0055] After completing peak coverage, the system controller executes a phase rearrangement gate. The process is as follows: The frequency domain transformation unit is invoked to convert the suppressed amplitude sequence and the original phase information to the frequency domain. The target phase sequence matching the high-frequency occupancy intensity is read from the phase table. A set of marked frequency points is selected as the replacement target, for example, replacing 32 marked frequency points at a time. The system controller replaces the current phase value with the phase value in the target phase sequence according to the frequency point order, and applies the difference set listed in the phase table between adjacent frequency points. The difference set is progressively stepped to ensure smooth phase changes in the spectrum. After replacement, an inverse transform is invoked to return to the time domain. The consideration for replacing the phase only on the marked frequency point set is that energy leakage at edge frequencies usually originates from phase abrupt changes near the edge. Limiting the replacement range concentrates the disturbance at the problematic location, avoiding impact on the symbol decision of the center frequency band. The consideration for using a stepped difference set is that it can smooth out phase changes at the frequency band edge, eliminating sharp spikes, thereby reducing out-of-band energy and additional heat generation. In an optional implementation, the set of marked frequency points can be 24 or 40 frequency points, and the target phase sequence can be set to two or three groups, each configured for different bandwidths.

[0056] Example 7: The system controller reads the upper and lower limits, the shortest continuous sampling number of the plateau segment, and the maximum change in a single step from the rail pressure table, and shapes the rail pressure sequence using a point-by-point scanning method. Taking the third level as an example, the upper and lower limits define the allowable range of rail pressure, the shortest continuous sampling number of the plateau segment is 512 sampling points, and the maximum change in a single step is 3 rail pressure units. The system controller scans backward from the starting point of the control window, replacing any sampling value higher than the upper limit with the upper limit value and lower than the lower limit with the lower limit value; if the difference between any two adjacent sampling points exceeds 3 rail pressure units, the latter is replaced with a value that meets the condition that the difference does not exceed 3 rail pressure units; after the rise or fall ends, a plateau segment of no less than 512 sampling points is inserted. The direct effect of this is to fix the output backoff rhythm within the power supply's allowable conversion rate, preventing overshoot, while leaving a stable range for baseband submission, reducing waveform tearing and noise propagation. In optional implementations, the minimum continuous sampling number for the platform segment can be set to 384 or 640 sampling points, and the maximum single-step change can be set to 2 or 4 rail pressure units to achieve balance under different power amplifier and power supply combinations. The system controller reads the operation sequence corresponding to the gas plug migration intention from the gas plug migration table and starts execution at the end of the control window. Taking extending the gas plug as an example, the operation sequence for the second level includes three actions: opening the heating strip on the outside of the evaporation section for 80 milliseconds to create a temperature difference that is conducive to the advancement of the leading edge; opening the metering microvalve for 80 milliseconds to introduce non-condensable gas to increase the gas plug length; reading the position sensor status, and closing the heating strip and metering microvalve and recording the new section number when the status change mark of the next section appears. Taking shortening the gas plug as an example, the operation sequence for the second level is to open the heating strip on the outside of the condensation section for 80 milliseconds to create a temperature difference that is conducive to the retreat of the trailing edge, then open the exhaust microvalve for 80 milliseconds to release non-condensable gas, and finally read the position sensor status to confirm the status recovery mark of the previous section. Taking the migration direction of the gas plug as an example, when migrating towards the evaporation section, the heating band on the outside of the evaporation section is intermittently switched on and off according to a preset combination, with each on / off cycle lasting 40 milliseconds, for a total of 160 milliseconds, driving the leading edge to move along the evaporation section direction; when migrating towards the condensation section, the heating band on the outside of the condensation section is intermittently switched on and off according to a preset combination, with the same rhythm. After any fine-tuning action, the system controller checks the position sensor status again after the minimum interval time of 100 milliseconds is reached. If no status change marker is found, it automatically increases to the next duration and repeats the same action. The sequential arrangement of first establishing the migration driving force with heat and then changing the gas plug geometry with a small amount of non-condensable gas can achieve stable displacement with less internal pipe disturbance, reduce the impact of bubble breakage and liquid phase backflow, and facilitate the reconstruction of the dominant path and bypass in a short time, allowing heat to quickly bypass the congested path inside the device. In optional embodiments, the single on / off duration can be 50 milliseconds or 120 milliseconds, the minimum interval time can be 80 milliseconds or 120 milliseconds, and the number of partitions can be adjusted from 8 to 6 or 10 to adapt to different heat pipe lengths.

[0057] Example 8: The system controller follows a specific sequence within the control window: first, it completes rail pressure trajectory shaping; then, it completes peak envelope ratio shaping and phase rearrangement; finally, it triggers plug migration. To ensure strict execution, the system controller sets minimum intervals for submission times: the rail pressure submission point is no earlier than 0.5 milliseconds after the start of the control window; the baseband submission point is no earlier than 0.3 milliseconds after the rail pressure submission point; and the operation sequence start point is no earlier than 2 milliseconds after the baseband submission point. Furthermore, cross-actuator interlocking includes three constraints: the metering microvalve and the discharge microvalve cannot open simultaneously at any time; the outer heating strips of the evaporator section and the condenser section cannot open simultaneously in the same action; and the next action is not initiated until a plug fine-tuning action is completed and the position sensor status has been read. Prioritizing rail voltage submission is crucial for fixing the allowable range of the envelope tracking voltage rail, ensuring stable power supply for baseband coverage and phase replacement. Delaying plug migration is important because the thermal actions of hard actuators and microvalve movements can introduce short-term disturbances, which should be avoided from overlapping with baseband and rail voltage submissions to minimize real-time impact on the waveform. In an optional implementation, the minimum interval between rail voltage and baseband can be set to 0.5 milliseconds, and the minimum interval between baseband and the operation sequence can be set to 3 milliseconds to accommodate more sensitive wireless links.

[0058] After all actions in a frame are completed, the system controller reads the frame tail samples of temperature and heat flux at the end of the control window and compares them with the frame header samples at the beginning of the control window. The evaluation pass condition can be set to meet any of the following: the device hotspot temperature drops by at least 1 degree Celsius; the heat flux intensity level pointing to the condensation section increases by one level; the thermal stress level decreases by one level or more. If the evaluation passes, the system controller marks the soft actuator action combination and hard actuator action combination used in this frame as reusable for quick retrieval in the next frame. If the evaluation fails, the system controller performs an online resetting process before the end of the current frame. Online resetting includes at least one of three operations: increasing the output rollback intention by one level and adding a plateau segment of 256 sampling points at the end of the rail pressure sequence; switching the peak envelope ratio shaping to a stronger window table entry and adding a coverage of the most prominent peak, with an additional coverage length of 64 sampling points; or invoking the acceleration entry of the slug migration table to perform a more intense slug fine-tuning, such as increasing the on / off duration of the outer heating band of the evaporator section from 80 milliseconds to 120 milliseconds and the on-time of the metering microvalve from 80 milliseconds to 120 milliseconds. In an optional implementation, the evaluation pass condition can be supplemented with a constraint, such as the amplitude at the out-of-band energy monitoring point being below a set threshold, to ensure that spectral leakage after the phase rearrangement gate is synchronously controlled.

[0059] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling an adaptive thermally controlled heat sink for an RF microwave front-end, characterized in that, Includes the following steps: The digital predistortion tap and envelope tracking voltage rail are registered as soft actuators, and the plug drive unit of the controllable non-condensing plug heat pipe is registered as a hard actuator. During the continuous operation of the service waveform, the system controller performs the following closed-loop operation based on the data acquired in the current frame within the control window of each service frame: acquire and align baseband sampling, rail pressure sampling, temperature and heat flux sampling, and plug position status to generate a homogeneous frame containing baseband sequence, rail pressure sequence, temperature sequence, heat flux sequence, and plug state sequence; based on the homogeneous frame, derive the thermal stress level characterizing the current state. Based on the thermal stress level and combined with the waveform feature detection results of the baseband sequence, the soft actuator action combination and hard actuator action combination are retrieved from the preset thermal control mapping table; based on the soft actuator action combination, the baseband sequence and rail pressure sequence used for shaping the peak envelope ratio and implementing output back-off are synthesized; based on the hard actuator action combination, the operation sequence used for migrating the air plug to reconstruct the thermal path is synthesized. Within the control window, the synthesized baseband sequence, rail pressure sequence, and operation sequence are respectively sent to the transmit link, envelope tracking power supply, and plug drive unit for execution. Specifically, the system controller first submits the shaped rail pressure sequence within the control window, with the submission point no earlier than 0.5 milliseconds after the start of the window; then, it submits the baseband sequence after peak coverage and phase rearrangement processing, with the submission point no earlier than 0.3 milliseconds after the rail pressure submission point, to ensure that the rail pressure change has entered the plateau segment; finally, at the end of the control window, it submits the operation sequence to trigger plug migration, with the start point no earlier than 2 milliseconds after the baseband submission point, and immediately reads the position sensor status for confirmation after each fine-tuning action is completed.

2. The method according to claim 1, characterized in that, The waveform feature detection results of the baseband sequence include at least one of the following: a peak density sequence obtained by sliding detection of the baseband sequence, a frequency of change sequence obtained by statistically analyzing the number of amplitude changes, and a high-frequency occupancy indication sequence obtained by transforming to the frequency domain and extracting edge frequency energy.

3. The method according to claim 1, characterized in that, The steps for generating isomorphic frames specifically include: within the control window, buffering baseband sampling, rail pressure sampling, temperature and heat flux sampling, and the position status of the gas plug; and using linear insertion or nearest neighbor methods to complete or dilute each buffered sequence so that each sequence has a corresponding value at each sampling position.

4. The method according to claim 1, characterized in that, The steps for deriving the thermal stress level specifically include: mapping thermal path status labels indicating the combination of dominant, bypass, and limiting paths between the evaporation and condensation sections based on the temperature sequence, heat flow sequence, and gas lock state sequence of the isomorphic frame; and looking up the thermal stress level in a table based on the thermal path status labels and the level combination of temperature and heat flow.

5. The method according to claim 1, characterized in that, The steps for synthesizing a baseband sequence for shaping the peak envelope ratio specifically include: performing sliding detection on the baseband sequence to locate the peak position; for each located peak, selecting a window sequence that matches the peak width from a preset window table, and replacing the amplitude of the coverage area on both sides of the peak position point by point according to the window sequence, keeping the phase unchanged during replacement, thereby forming a suppressed amplitude sequence.

6. The method according to claim 5, characterized in that, When the high-frequency occupancy indication in the waveform feature detection result is significant, the steps of synthesizing the baseband sequence also include a phase rearrangement gate operation, specifically: converting the suppressed amplitude sequence and the original phase information into amplitude-to-phase information in the frequency domain; reading the target phase sequence from the preset phase table; replacing the phase values ​​in the marked frequency point set with the values ​​in the target phase sequence; and then inversely transforming back to the time domain.

7. The method according to claim 1, characterized in that, The steps for synthesizing the rail pressure sequence used to implement output rollback specifically include: reading the upper limit value, lower limit value, shortest continuous sampling number of the plateau segment, and maximum single-step change amount corresponding to the output rollback intention from a preset rail pressure table; scanning the rail pressure sequence, replacing sampling points that are higher than the upper limit value or lower than the lower limit value with the corresponding limit value, limiting the difference between adjacent samples to not exceed the maximum single-step change amount, and inserting a plateau segment that is not shorter than the shortest continuous sampling number of the plateau segment after the change ends, thereby obtaining the shaped rail pressure sequence.

8. The method according to claim 1, characterized in that, The steps of synthesizing the operation sequence for migrating gas plugs specifically include: reading the operation sequence corresponding to the gas plug migration intention from a preset gas plug migration table; when the intention is to extend the gas plug, the operation sequence controls the opening of the heating belt on the outside of the evaporation section and introduces non-condensable gas into the heat pipe; when the intention is to shorten the gas plug, the operation sequence controls the opening of the heating belt on the outside of the condensation section and discharges non-condensable gas; when the intention is to migrate the gas plug direction, the operation sequence controls the on and off of the heating belt according to a preset combination to form a temperature gradient.

9. The method according to claim 1, characterized in that, The issuance process follows interlocking and sequential control, and the sequence is as follows: first, rail pressure trajectory shaping is completed, then peak envelope ratio shaping and phase rearrangement are completed, and finally, gas plug migration is triggered.

10. The method according to claim 1, characterized in that, After all actions in a frame are completed, a closed-loop evaluation and online resetting step is also included: comparing the sampling at the end of the frame and the beginning of the frame to evaluate the thermal stress improvement conclusion and the waveform fidelity conclusion; if either conclusion fails to meet the preset pass condition, an online resetting process is performed before the end of the frame. The online resetting process includes at least one of the following: increasing the output rollback intention by one level, switching to a stronger window table entry, or calling the acceleration entry to perform a higher intensity plug fine-tuning.

Images