An amplifier dynamic performance optimization control method and system

Abstract

The application provides an amplifier dynamic performance optimization control method and system, and belongs to the technical field of amplifier dynamic control. The method comprises the following steps: inputting working condition data of a power amplifier into a pre-trained time sequence prediction neural network in real time to obtain a predicted transient response overshoot feature; generating a pre-distortion compensation signal according to the transient response overshoot feature; superimposing the pre-distortion compensation signal on an input signal of the power amplifier to obtain a compensated input signal; adjusting equivalent electromagnetic parameters of an adjustable metamaterial structure integrated in an output matching network of the power amplifier based on the predicted transient response overshoot feature to obtain a target output transmission characteristic; monitoring an output waveform generated by the compensated input signal and the target output transmission characteristic, extracting a residual error feature of the output waveform; judging whether the residual error feature is abnormal to obtain a judgment result; and optimizing the time sequence prediction neural network, the pre-distortion compensation signal and the equivalent electromagnetic parameters based on the judgment result.

Description

Technical Field

[0001] This application relates to the field of amplifier dynamic control technology, and in particular to an amplifier dynamic performance optimization control method and system. Background Technology

[0002] Power amplifiers are crucial components in systems such as radio frequency communication and power electronics. Their dynamic transient performance significantly impacts operational stability and signal transmission quality. Existing power amplifier control technologies largely employ fixed-parameter compensation and static matching network schemes, only suitable for steady-state conditions. However, in complex real-world operating conditions, frequent changes in load impedance and transmitted power can lead to transient overshoot, waveform oscillations, and signal distortion in the power amplifier output. Current post-error correction control methods for power amplifiers suffer from response lag and low compensation accuracy, failing to proactively suppress transient errors. Furthermore, fixed matching network electromagnetic parameters cannot dynamically adapt to fluctuations in operating conditions and cannot optimize output transmission characteristics at the hardware level. The lack of a linkage mechanism between predictive compensation and dynamic hardware tuning in existing technologies results in poor dynamic performance and weak adaptability of power amplifiers, failing to meet the demands for high precision and high stability in dynamic operation. Summary of the Invention

[0003] To address the aforementioned technical problems, this application provides a method and system for optimizing the dynamic performance control of an amplifier.

[0004] A first aspect of this application provides an amplifier dynamic performance optimization control method, comprising: The operating condition data of the power amplifier is collected and input into a pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier. Based on the transient response overshoot characteristics, a pre-distortion compensation signal with the opposite trend to the transient change is generated; the pre-distortion compensation signal is superimposed on the input signal of the power amplifier in advance to obtain the compensated input signal; Based on the predicted transient response overshoot characteristics, the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network are adjusted to obtain the target output transmission characteristics. Monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and extract the residual error characteristics of the output waveform; Determine whether the residual error characteristics are abnormal, and obtain the determination result; Based on the judgment result, the time-series prediction neural network, the pre-distortion compensation signal, and the equivalent electromagnetic parameters are optimized. The operating condition data acquisition and prediction module is used to acquire the operating condition data of the power amplifier and input the operating condition data into the pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier. A second aspect of this application provides an amplifier dynamic performance optimization control system, comprising: The predistortion signal generation and superposition module is used to generate a predistortion compensation signal that is opposite to the transient change trend based on the transient response overshoot characteristics; and to superimpose the predistortion compensation signal onto the input signal of the power amplifier in advance to obtain the compensated input signal. The predistortion signal generation and superposition module is used to adjust the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network based on the predicted transient response overshoot characteristics, so as to obtain the target output transmission characteristics. The output waveform error extraction module is used to monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and to extract the residual error characteristics of the output waveform. An error state anomaly determination module is used to determine whether the residual error characteristics are abnormal and to obtain a determination result; The global parameter iterative optimization module is used to optimize the time-series predictive neural network, the pre-distortion compensation signal, and the equivalent electromagnetic parameters based on the judgment result.

[0005] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described amplifier dynamic performance optimization control method.

[0006] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described amplifier dynamic performance optimization control method.

[0007] The beneficial effects of the amplifier dynamic performance optimization control method and system provided in this application are as follows: This application uses a time-series predictive neural network to predict transient response overshoot characteristics, thereby enabling early prediction of transient faults in the power amplifier and reducing the lag problem in power amplifier control. Furthermore, by predicting the transient response overshoot characteristics and generating a reverse pre-distortion compensation signal, which is then superimposed onto the input signal in advance, the accuracy of offsetting the transient overshoot trend from the signal input end is improved, suppressing waveform distortion in the power amplifier. Simultaneously, by dynamically adjusting the equivalent electromagnetic parameters of the adjustable metamaterial in the output matching network based on the predicted transient response overshoot characteristics, the transient oscillation suppression capability is strengthened at the hardware level. In addition, monitoring the output waveform and extracting residual error characteristics improves the accuracy of residual error feature location and control. Subsequently, by judging the anomalies of the residual error characteristics, the type of waveform failure problem can be distinguished. The neural network, pre-distortion signal, and metamaterial parameters are optimized based on the judgment results, effectively improving the stability and dynamic response accuracy of the power amplifier under load abrupt changes and power jump conditions. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating an amplifier dynamic performance optimization control method provided in an embodiment of this application. Figure 2 This is a structural block diagram of an amplifier dynamic performance optimization control system provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0009] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0010] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1 - Appendix Figure 3 The following is an explanation using specific examples.

[0011] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an amplifier dynamic performance optimization control method according to an embodiment of this application. The method includes: S101: Collect the operating condition data of the power amplifier and input the operating condition data into the pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier.

[0012] In this embodiment, operating condition data during the operation of the power amplifier is collected. This data includes operating parameters such as voltage, current, input signal amplitude, load status, output power, and RF transmission status. The operating condition data is acquired using sensing and detection modules deployed throughout the power amplifier and signal link. Specifically, voltage detection units, current detection units, signal amplitude detection units, load detection units, power detection units, and RF status detection units are integrated at the power amplifier's input port, output port, power supply circuit, and load connection point, respectively. These detection devices are integrated and adapted to the amplifier's main circuitry. Operating condition data is continuously acquired using synchronous sampling, with each detection unit operating in parallel. The voltage and current detection units collect voltage and current data from the power supply circuit and signal path; the signal amplitude detection unit samples and analyzes the original input signal to obtain the input signal amplitude; the load detection unit identifies load information such as backend load impedance and connection status; the power detection unit calculates the amplifier's real-time output power; and the RF status detection unit monitors RF transmission status such as link reflection and transmission loss.

[0013] The operating condition data is filtered and denoised to remove interference noise generated during the acquisition process. Finally, the processed complete operating condition data is input into a pre-trained time-series prediction neural network to obtain the predicted transient response overshoot characteristics.

[0014] The time-series prediction neural network adopts a multi-input single-output feature-based time-series network hierarchical architecture, consisting of four sequentially flowing layers: a data preprocessing layer, a time-series feature extraction layer, a deep fusion layer, and a feature decoding output layer. Specifically, the data preprocessing layer receives standard operating condition data after hardware sampling and software filtering and denoising, performing data normalization, dimensionality regularization, and time-series sequence grouping, stitching discrete sampling points into continuous time-series samples. The time-series feature extraction layer, as the core layer of the network, specifically mines the dynamic patterns of various operating condition data such as voltage, current, load, and power over time, capturing transient precursor information such as power surges and load fluctuations. The deep fusion layer performs cross-feature fusion and nonlinear mapping on the multi-dimensional time-series features extracted by the previous layer, strengthening the correlation features of transient changes. The feature decoding output layer analyzes and maps the fused high-level features into concrete transient response overshoot features, completing the final output.

[0015] The temporal prediction neural network is constructed primarily using a temporal recurrent network (TRRN) and a fully connected network, adaptable to the prediction of continuous timing data from power amplifiers. The basic structure employs a lightweight Long Short-Term Memory (LSTM) network as the backbone for temporal feature extraction, coupled with multiple fully connected layers for feature fusion and decoding. The temporal prediction neural network adopts an end-to-end training mode. Initially, offline pre-training is performed based on historical operating data and corresponding measured overshoot features. After convergence, fixed base weights are used for real-time operation, while also supporting subsequent online incremental learning to update parameters. The input to this temporal prediction neural network is a normalized multivariate time series, with the input dimension corresponding to the collected operating parameters. The output of the temporal prediction neural network is a predicted transient response overshoot feature vector, with the output dimension corresponding to the overshoot index. The predicted transient response overshoot features include the start time, magnitude, overshoot slope, and peak occurrence time of power demand mutations.

[0016] The dataset for the time-series predictive neural network is derived from historical operating data of power amplifiers. Each sample is a continuous time-series sampling segment, including the continuous trend of operating parameters and information on the precursors of transient changes. The dataset label is the overshoot feature of the transient response obtained from the actual measurement of the corresponding time-series segment, which marks the start time, magnitude, overshoot slope, and time label of the overshoot peak occurrence of the power demand change.

[0017] The dataset is divided into a training set, a validation set, and a test set. The training set is used for iterative updates of the network's basic weights, the validation set is used for hyperparameter tuning during training to prevent overfitting and underfitting, and the test set is used to verify the generalization ability and prediction accuracy of the final model.

[0018] S102: Based on the transient response overshoot characteristics, generate a pre-distortion compensation signal that is opposite to the transient change trend; superimpose the pre-distortion compensation signal onto the input signal of the power amplifier in advance to obtain the compensated input signal.

[0019] In this embodiment, the predicted transient response overshoot characteristics include the start time, amplitude, overshoot slope, and peak time of the power demand mutation. Specifically, based on the amplitude, slope, and total transmission delay of the power demand mutation in the predicted transient response overshoot characteristics, the timing advance of the predistortion compensation signal is calculated; using the start time and timing advance of the power demand mutation, the injection start time and duration of the predistortion compensation signal are determined, obtaining the predistortion compensation timing parameters; the phase of the predistortion compensation signal is adjusted according to the load impedance and reflected power in the operating data, obtaining the phase-calibrated predistortion compensation signal; the input baseband signal is processed based on the predistortion compensation timing parameters and the phase-calibrated predistortion compensation signal to obtain the baseband predistortion component; the baseband predistortion component is amplitude-scaled and envelope-shaped based on the overshoot slope and amplitude mutation to obtain the predistortion compensation signal with the opposite trend to the transient change.

[0020] The predistortion compensation signal is pre-superimposed onto the input signal of the power amplifier to obtain the compensated input signal. The superposition process is initiated precisely according to the time advance, the injection start time of the predistortion compensation signal, and its duration. During the signal superposition stage, the predistortion compensation signal is connected to the input signal path of the power amplifier. A linear superposition method is used to superimpose the predistortion compensation signal onto the input signal of the power amplifier. The two types of signals are fused in the same signal link to obtain the compensated input signal.

[0021] S103: Based on the predicted transient response overshoot characteristics, the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network are adjusted to obtain the target output transmission characteristics.

[0022] In this embodiment, the overshoot energy value is calculated based on the abrupt change amplitude, the overshoot rise time, and the load impedance and reflected power in the operating data. Based on the overshoot energy value, the target energy ratio that the tunable metamaterial structure needs to absorb is determined. Based on the acquired RF impedance loss characteristics, resonance characteristics, and current load state, the target equivalent resistance and target equivalent inductance values ​​of the tunable metamaterial structure are calculated using the target energy ratio. According to the start time of the power demand abrupt change, the time of the overshoot peak occurrence, and the recovery time of the overshoot oscillation, a holding time window for the target equivalent resistance and target equivalent inductance values ​​is determined. The holding time window includes the time from the start of the power demand abrupt change to the recovery time of the overshoot oscillation. Based on the holding time window and the target equivalent resistance and target equivalent inductance values, the equivalent resistance and equivalent inductance values ​​of the tunable metamaterial structure are adjusted synchronously to obtain the target output transmission characteristics.

[0023] Specifically, overshoot energy is the extra radio frequency (RF) energy accumulated when the signal exceeds its steady-state rated value during a transient power surge. The tunable metamaterial structure is an artificial electromagnetic structure integrated into the power amplifier's output matching network. It possesses adjustable electromagnetic parameters, actively absorbing and dissipating excess RF energy to suppress waveform overshoot and oscillation. The target energy ratio is the percentage of overshoot energy absorbed by the tunable metamaterial structure, calculated based on the total overshoot energy. The target equivalent resistance is the optimal equivalent resistance of the metamaterial calculated to achieve the specified energy absorption ratio. The target equivalent inductance is the optimal equivalent inductance of the metamaterial determined to match RF resonance and impedance network requirements, used to adjust the RF phase and resonant point, suppressing high-frequency oscillations and waveform distortion. The hold time window is the continuous time period from the start of the power surge to the complete recovery of the waveform to steady state. The target output transmission characteristics are the comprehensive RF performance achieved by the power amplifier's output link, including impedance matching, energy transmission, and oscillation suppression.

[0024] S104: Monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and extract the residual error characteristics of the output waveform.

[0025] In this embodiment, the compensated input signal is time-series sampled to capture the amplitude, phase, envelope changes, and timing fluctuations of the compensated input signal, and the state of the compensated input signal is recorded. Simultaneously, the output waveform generated by the power amplifier under the target output transmission characteristics is sampled and monitored, including information such as the peak value, rising edge, falling edge, oscillation state, and steady-state deviation of the output waveform.

[0026] After completing the synchronous monitoring and data acquisition of the input signal and output waveform, the output waveform data is subjected to in-depth analysis and comparative analysis to extract the residual error characteristics of the output waveform. Specifically, using a preset ideal standard output waveform as a reference, the monitored output waveform is compared with the reference waveform point by point in time to identify waveform deviations that remain after pre-distortion compensation and electromagnetic parameter adjustment, effectively eliminating steady-state normal fluctuation interference. Through this monitoring and feature extraction process, residual error characteristics such as subtle waveform distortion, residual overshoot, small oscillations, and phase shifts remaining after the power amplifier's dynamic optimization are captured. The preset ideal standard is the rated normal output waveform of the power amplifier without distortion, overshoot, or oscillation. This ideal standard is pre-calibrated and determined based on the power amplifier's rated operating parameters, design specifications, and steady-state ideal output state.

[0027] S105: Determine whether the residual error characteristics are abnormal and obtain the judgment result.

[0028] In this embodiment, residual error characteristics are verified to distinguish between normal operating condition fluctuations and abnormal waveform defects. Specifically, a multi-dimensional, hierarchical anomaly judgment threshold system is pre-established based on the rated operating parameters of the power amplifier, dynamic response requirements, and different load and power change scenarios. This threshold system includes multiple evaluation indicators such as residual overshoot peak value, oscillation duration, waveform distortion degree, phase offset, and error change rate, which can adapt to various complex transient operating conditions.

[0029] The residual error characteristics are compared and verified one by one with the standard parameters in the threshold system, checking the actual state of various characteristics such as residual overshoot, small oscillations, waveform distortion, and phase shift. During the judgment process, priority is given to distinguishing between steady-state normal fluctuations and transient abnormal deviations. For residual errors generated by the power amplifier under standard operating conditions that are stable and always within the threshold range, the operating condition is judged to be normal and will not trigger subsequent optimization actions. If any indicator in the residual error characteristics exceeds the corresponding preset threshold, or if the error change rate fluctuates drastically or the distortion pattern deviates from the normal range, then the current waveform is judged to have an abnormal problem.

[0030] In this embodiment, auxiliary verification is also performed based on the historical operating data and working conditions of the power amplifier to eliminate false anomalies caused by non-equipment-related issues such as environmental electromagnetic interference, sampling noise, and short-term occasional fluctuations. The resulting judgment marks the abnormal state, type, and severity of the residual error characteristics, providing feedback on shortcomings and deficiencies in the front-end signal compensation and hardware parameter control processes.

[0031] S106: Based on the judgment results, optimize the time-series prediction neural network, the predistortion compensation signal, and the equivalent electromagnetic parameters.

[0032] In this embodiment, if the determination result indicates an abnormality in the residual error characteristics, the abnormality type is extracted. If the abnormality type is that the residual overshoot peak value is greater than a preset first threshold or the residual oscillation duration is greater than a preset second threshold, a first optimization instruction is generated. The first optimization instruction is used to trigger the online incremental learning of the time-series prediction neural network to update the network weights and re-extract the transient response overshoot characteristics. If the abnormality type is that the waveform distortion morphology exhibits asymmetric overshoot or undercompensation, a second optimization instruction is generated. The second optimization instruction is used to adjust the scaling factor and shaping time constant of the predistortion compensation signal. If the abnormality type is that the waveform distortion morphology exhibits high-frequency ripple or damped oscillation, a third optimization instruction is generated. The third optimization instruction is used to adjust the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure to change the damping coefficient of the output matching network.

[0033] As can be seen from the above, this application achieves early prediction of transient faults in power amplifiers by using a time-series predictive neural network to predict transient response overshoot characteristics, thus reducing the lag problem in power amplifier control. Furthermore, the generation of a reverse pre-distortion compensation signal based on the predicted transient response overshoot characteristics and its pre-superposition onto the input signal improves the accuracy of offsetting transient overshoot trends at the signal input end and suppresses waveform distortion in the power amplifier. Simultaneously, the equivalent electromagnetic parameters of the adjustable metamaterial in the output matching network are dynamically adjusted based on the predicted transient response overshoot characteristics, strengthening the transient oscillation suppression capability at the hardware level. In addition, monitoring the output waveform and extracting residual error characteristics improves the accuracy of residual error feature localization and control. Subsequently, the anomaly detection of residual error characteristics allows for the differentiation of waveform failure types. Optimization of the neural network, pre-distortion signal, and metamaterial parameters based on the judgment results effectively improves the stability and dynamic response accuracy of the power amplifier under conditions of sudden load changes and power jumps.

[0034] In one embodiment of this application, the predicted transient response overshoot characteristics include the start time of the power demand mutation, the mutation magnitude, the overshoot slope, and the time when the overshoot peak occurs. Based on the predicted transient response overshoot characteristics, a predistortion compensation signal with the opposite transient change trend is generated, including: Based on the amplitude of the power demand mutation, the overshoot slope, and the total transmission delay in the predicted transient response overshoot characteristics, the timing advance of the predistortion compensation signal is calculated. By utilizing the start time and timing advance of the power demand mutation, the injection start time and duration of the predistortion compensation signal are determined, and the predistortion compensation timing parameters are obtained. Based on the load impedance and reflected power in the operating data, the phase of the predistortion compensation signal is adjusted to obtain the phase-calibrated predistortion compensation signal. The input baseband signal is processed based on the predistortion compensation timing parameters and the predistortion compensation signal after phase calibration to obtain the baseband predistortion component; The baseband predistortion component is amplitude scaled and envelope shaped based on the overshoot slope and abrupt change amplitude to obtain a predistortion compensation signal that is opposite to the transient change trend.

[0035] In this embodiment, the timing advance of the predistortion compensation signal is first calculated based on the amplitude, overshoot slope, and total transmission delay of the power demand mutation in the predicted transient response overshoot characteristics. The mutation amplitude determines the required compensation strength for the transient error, the overshoot slope indicates the rate of transient change, and the total transmission delay is the inherent delay of the power amplifier signal transmission. By matching the mutation amplitude, overshoot slope, and total transmission delay of the power demand mutation in the predicted transient response overshoot characteristics to the timing of transient overshoot occurrence, the required lead-off duration of the compensation signal can be calculated, ensuring that the compensation action occurs before the transient distortion. Based on this, the injection start time and duration of the predistortion compensation signal are locked using the start time of the power demand mutation and the timing advance, thereby obtaining the predistortion compensation timing parameters. This achieves alignment between the compensation timing and the transient change timing, avoiding compensation failure and secondary waveform distortion caused by compensation being too early or too late.

[0036] After determining the predistortion compensation timing parameters, the phase of the predistortion compensation signal is adjusted based on the load impedance and reflected power in the operating condition data to obtain a phase-calibrated predistortion compensation signal. Specifically, load impedance and reflected power affect the transmission phase offset and link loss of the RF signal, and phase calibration is completed by adapting to the operating condition parameters. Subsequently, the input baseband signal is processed in conjunction with the predistortion compensation timing parameters and the phase-calibrated predistortion compensation signal to remove transient distortion components from the baseband signal, resulting in a baseband predistortion component adapted to the current dynamic operating condition. The baseband predistortion component is the basic compensation signal obtained after processing the original input baseband signal according to the compensation timing and calibrated phase, and it serves as the compensation carrier to compensate for transient distortion.

[0037] Finally, based on the overshoot slope and abrupt change amplitude, the baseband predistortion component is amplitude-scaled and envelope-shaped to obtain a predistortion compensation signal that has the opposite trend to the transient change. The predistortion compensation signal refers to the complete compensation signal generated after amplitude scaling and envelope shaping of the baseband predistortion component, which has the opposite trend to the transient change of the power amplifier. It is directly superimposed on the power amplifier input to suppress overshoot.

[0038] As can be seen from the above, in this embodiment, the compensation signal timing advance is first calculated based on the abrupt change amplitude, overshoot slope, and transmission delay. The compensation timing is then set and phase calibration is completed. The baseband predistortion component is then obtained through processing. Finally, the inverse predistortion compensation signal is generated through amplitude scaling and envelope shaping. This enables the compensation action to precede the waveform distortion, thereby achieving full-range matching of timing, phase, and amplitude. This effectively reduces the secondary distortion caused by compensation timing deviation and phase mismatch, weakens the transient overshoot caused by power amplifier power abrupt changes, and thus improves the signal output quality and the dynamic working stability of the power amplifier.

[0039] In one embodiment of this application, amplitude scaling and envelope shaping of the baseband predistortion component are performed based on the overshoot slope and abrupt change amplitude to obtain a predistortion compensation signal with a trend opposite to the transient change, including: The scaling factor is determined based on the mutation amplitude; the shaping time constant is determined based on the overshoot slope. The larger the mutation amplitude, the larger the scaling factor, and the larger the overshoot slope, the smaller the shaping time constant. Multiply the amplitude of the baseband predistortion component by a scaling factor to obtain the amplitude-scaled predistortion component; The envelope of the amplitude-scaled predistortion component is passed through a low-pass filter to obtain the envelope-shaped predistortion component, where the time constant of the low-pass filter is equal to the shaping time constant. The envelope-shaped predistortion component is output as a predistortion compensation signal that is opposite to the transient change trend.

[0040] In this embodiment, the scaling factor is first determined based on the abrupt change amplitude, and the shaping time constant is determined based on the overshoot slope. A larger abrupt change amplitude corresponds to a larger scaling factor, while a larger overshoot slope corresponds to a smaller shaping time constant. Specifically, the scaling factor is determined as follows: First, the abrupt change amplitude value corresponding to the predicted transient response overshoot characteristic under the current operating condition is read. This abrupt change amplitude is compared and matched with a preset abrupt change amplitude benchmark interval. Based on a linear or piecewise proportional mapping relationship, a scaling factor of the corresponding magnitude is adaptively output. When the abrupt change amplitude is small and the power transient fluctuation is weak, a smaller scaling factor is matched to avoid overcompensation distortion caused by an excessively large compensation amplitude. When the abrupt change amplitude increases and the power transient jump is severe, the scaling factor value is increased synchronously, and the amplitude of the baseband predistortion component is amplified proportionally to match the predistortion compensation intensity with the severity of the current transient overshoot. The method for determining the shaping time constant is as follows: The overshoot slope value corresponding to the current transient operating condition is collected to characterize the steepness of the rise and fall of the power amplifier's transient signal. The overshoot slope is then substituted into a preset inverse mapping relationship to obtain the shaping time constant. Specifically, when the overshoot slope is small and the transient change process is gentle, a larger shaping time constant is matched to enhance the smoothing effect of the low-pass filter, ensuring a stable and fluctuation-free envelope for the compensation signal. When the overshoot slope is large and the transient fall is rapid and steep, a smaller shaping time constant is matched to shorten the filtering response time of the low-pass filter. The preset abrupt change amplitude reference range is a continuous numerical range divided based on the power amplifier's rated operating parameters, the full-range power fluctuation range, and the measured transient response data of the equipment. This range is used to classify the abrupt change amplitude. The preset inverse mapping relationship is a mathematical correspondence between the overshoot slope and the shaping time constant, satisfying the negative correlation characteristic that a larger overshoot slope corresponds to a smaller shaping time constant, specifically adapted to the parameter configuration requirements of the low-pass filter.

[0041] Secondly, the amplitude of the baseband predistortion component is multiplied by a scaling factor to obtain the amplitude-scaled predistortion component. The scaling factor is a dimensionless coefficient dynamically calculated based on the amplitude of the sudden change. The larger the amplitude of the sudden change, the larger the value of this coefficient. Its function is to proportionally amplify or reduce the signal amplitude to match the compensation intensity with the transient overshoot intensity. The amplitude-scaled predistortion component is a new signal obtained by adjusting the amplitude of the baseband predistortion component through proportional calculation. The signal waveform, phase, and timing of the amplitude-scaled predistortion component remain unchanged; only the signal strength changes. It serves as an intermediate compensation signal after amplitude adaptation.

[0042] Subsequently, the envelope of the amplitude-scaled predistortion component is passed through a low-pass filter to obtain the envelope-shaped predistortion component, where the time constant of the low-pass filter is equal to the shaping time constant. The low-pass filter smooths the signal envelope, and the adaptive shaping time constant matches transient overshoot variations of different steepness, ensuring smooth and timely changes in the compensation signal envelope. Finally, the envelope-shaped predistortion component is output as a predistortion compensation signal with the opposite trend to the transient change, providing high-quality signal support for accurate predistortion compensation of the power amplifier input signal.

[0043] As can be seen from the above, this embodiment dynamically matches the scaling factor and shaping time constant according to the abrupt change amplitude and overshoot slope, and sequentially performs amplitude scaling and low-pass filter envelope shaping on the baseband predistortion component. It can adjust the compensation intensity and signal change rate as needed, improve the accuracy of the final output compensation signal matching transient distortion characteristics, and can not only fully offset the overshoot caused by power abrupt changes of different intensities, but also avoid new waveform distortion caused by abrupt changes in the compensation signal, thus ensuring that the compensation effect of transient overshoot suppression of the power amplifier is stable and effective.

[0044] In one embodiment of this application, after passing the envelope of the amplitude-scaled predistortion component through a low-pass filter to obtain the envelope-shaped predistortion component, the method further includes: The envelope-shaped predistortion component is compared with the amplitude-scaled predistortion component, and the envelope difference is calculated. When the envelope difference is greater than the preset difference threshold, the shaping time constant is adjusted according to the relationship between the direction and magnitude of the envelope difference to obtain the target shaping time constant. The target shaping time constant is applied to the low-pass filter to perform secondary filtering on the envelope of the amplitude-scaled pre-distortion component until the envelope difference is reduced to within the difference threshold or the preset maximum number of filtering times is reached.

[0045] In this embodiment, the pre-distortion component after envelope shaping and the pre-distortion component after amplitude scaling are first compared point-by-point in time to capture the envelope shape deviation between the two signals, thus obtaining the envelope difference degree. The envelope difference degree characterizes the degree of envelope distortion and waveform offset before and after the low-pass filter shaping process, and represents the adaptation performance of the current shaping time constant. It provides quantitative data for subsequent parameter correction and effectively avoids subjective adaptation errors caused by relying on a single parameter setting.

[0046] The envelope difference is compared with a preset difference threshold to determine whether the current envelope shaping effect meets the standard. When the envelope difference is less than or equal to the preset difference threshold, it means that the current shaping time constant is suitable for the current transient conditions, the envelope shaping effect is good, and no secondary optimization is needed. When the envelope difference is greater than the preset difference threshold, it indicates that there is a significant deviation in a single filtering shaping. Based on the correspondence between the direction and magnitude of the envelope difference, the shaping time constant is dynamically corrected to calculate the target shaping time constant suitable for the current conditions. The direction of the difference is used to determine whether the envelope is over-smoothed or under-smoothed, and the magnitude of the difference is used to precisely control the correction magnitude of the time constant. Specifically, the shaping deviation type of the current low-pass filter is determined based on the direction of the envelope difference. If the envelope difference is positive, it indicates that the low-pass filter's smoothing effect is too strong, resulting in over-smoothing of the envelope. This causes the edge response of the shaped compensation signal to be slow and unable to quickly follow steep transient changes. If the envelope difference is negative, it indicates that the low-pass filter's smoothing effect is insufficient, resulting in inadequate envelope shaping. This leaves high-frequency fluctuations and fine ripples in the shaped compensation signal. The severity of the shaping deviation is quantified based on the magnitude of the envelope difference, determining the correction step size of the shaping time constant. A larger envelope difference value indicates a higher degree of deviation in a single filtering and shaping operation, requiring a larger parameter correction amplitude. A smaller envelope difference value allows for small-scale fine-tuning to ensure the stability and accuracy of parameter iteration. Based on this, adaptive correction is performed by combining the direction and magnitude of the difference, calculating the target shaping time constant. A larger envelope difference value indicates a higher degree of deviation in a single filtering and shaping operation, requiring a larger parameter correction amplitude. A smaller envelope difference value allows for a smaller parameter correction amplitude.

[0047] The preset difference threshold is a quantitative criterion for judging whether the envelope shaping effect is acceptable. It is set by combining power amplifier design specifications, transient response requirements, and measured waveform data. Specifically, firstly, standard samples under all operating conditions of the power amplifier are selected, and the pre-distortion component after amplitude scaling and the pre-distortion component after envelope shaping under ideal conditions are collected. The baseline envelope difference under standard operating conditions is calculated by comparing them point by point. Secondly, based on the maximum allowable waveform distortion of the power amplifier and the compensation signal response speed requirements, a reasonable error tolerance is defined, and the upper limit of the tolerance is determined as the basic threshold. Simultaneously, normal operating conditions and extreme transient operating conditions are distinguished. For scenarios with drastic power changes and extremely high overshoot slopes, the basic threshold is adaptively increased to obtain the difference threshold.

[0048] After obtaining the target shaping time constant, the system reapplies it to the low-pass filter, performing secondary filtering and envelope shaping iterative processing on the envelope of the amplitude-scaled predistortion component. This iterative process is continuously executed in a loop, constantly updating the operating parameters of the low-pass filter and continuously optimizing the envelope shape of the predistortion component until the envelope difference is reduced to within the difference threshold, or the number of iterations reaches the preset maximum number of filtering iterations, at which point the process terminates. Through this closed-loop iterative optimization mechanism, the system can adaptively correct the adaptation deviation of the shaping time constant, completely solving the defect that fixed parameter filtering cannot adapt to complex transient conditions, and ultimately outputting a predistortion compensation signal with smooth waveform, accurate timing, and extremely low distortion. The preset maximum number of filtering iterations is the termination protection condition for the iterative filtering stage. Specifically, the basic range is determined based on the power amplifier signal transmission delay and transient response timing requirements. The computation time required for a single filtering, envelope comparison, and parameter adjustment is statistically analyzed, and based on the time advance of the predistortion compensation signal, the maximum time that the iterative operation can occupy is calculated to obtain the maximum number of iterations.

[0049] As can be seen from the above, this embodiment, by comparing the envelopes of the pre-distortion components before and after filtering and calculating the difference, dynamically iteratively adjusting the shaping time constant for the over-threshold deviation and performing secondary filtering optimization, can effectively correct the adaptation deviation existing in single envelope shaping, adaptively calibrate the filtering characteristics of the low-pass filter, reduce the problems of over-compensation, under-compensation, or envelope distortion caused by fixed shaping parameters, ensure that the envelope shape of the pre-distortion compensation signal closely matches the transient distortion suppression requirements of the power amplifier, further improve the dynamic compensation accuracy and adaptability, and ensure the stability and accuracy of the transient overshoot suppression effect of the power amplifier under different power change conditions.

[0050] In one embodiment of this application, based on the predicted transient response overshoot characteristics, the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network are adjusted to obtain the target output transmission characteristics, including: The overshoot energy value is calculated based on the abrupt change amplitude, the overshoot rise time, the load impedance and reflected power in the operating data. Based on the overshoot energy value, determine the target energy ratio that the tunable metamaterial structure needs to absorb; Based on the acquired RF impedance loss characteristics, resonance characteristics, and current load state, the target equivalent resistance and target equivalent inductance values ​​of the tunable metamaterial structure are calculated using the target energy ratio. Based on the start time of the power demand surge, the time of the overshoot peak, and the recovery time of the overshoot oscillation, determine the holding time window for the target equivalent resistance value and the target equivalent inductance value. The holding time window includes the start time of the power demand surge to the recovery time of the overshoot oscillation. Based on the time window and the target equivalent resistance and inductance values, the equivalent resistance and inductance values ​​of the tunable metamaterial structure are adjusted synchronously to obtain the target output transmission characteristics.

[0051] In this embodiment, firstly, the overshoot energy value during the transient process is calculated based on the abrupt change amplitude, the overshoot rise time, and the load impedance and reflected power in the operating data. The abrupt change amplitude represents the power fluctuation intensity, the overshoot rise time represents the transient change rate, and the load impedance and reflected power determine the energy loss and reflection state of the RF link. The formula for the overshoot energy value is: E=(AP)·t² / (2·Z), where A is the abrupt change amplitude in W (power amplitude change); t is the overshoot rise time in s; Z is the load impedance in the operating data in Ω; P is the reflected power in W; and E is the overshoot energy value in J (joules).

[0052] Secondly, the target energy ratio that the tunable metamaterial structure needs to absorb is determined based on the impulse energy value. This target energy ratio is dynamically matched according to the overall vibration suppression requirements of the power amplifier and the link carrying capacity to determine the energy absorption task that the tunable metamaterial structure needs to undertake, so as to avoid insufficient energy absorption leading to overshoot suppression failure, or excessive absorption ratio affecting the normal power transmission of the equipment. Subsequently, based on the collected RF impedance loss characteristics, resonance characteristics, and current load state, the target energy ratio is substituted into the calculation to obtain the target equivalent resistance value and target equivalent inductance value of the tunable metamaterial structure adapted to the current operating conditions.

[0053] Subsequently, a dedicated hold-up time window is defined based on the start time of the power demand surge, the time of the overshoot peak, and the recovery time of the overshoot oscillation. This hold-up time window includes the period from the start time of the power demand surge to the recovery time of the overshoot oscillation. Finally, based on the hold-up time window and the target equivalent resistance and equivalent inductance values, the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure are adjusted synchronously to obtain the target output transmission characteristics.

[0054] As can be seen from the above, this embodiment first calculates the overshoot energy value based on multiple operating parameters to determine the energy absorption ratio of the metamaterial. Then, it solves for the corresponding equivalent resistance and inductance parameters according to the RF characteristics and load conditions, and defines the holding time window for the effective action of the parameters. Within the holding time window, the electromagnetic parameters of the metamaterial are dynamically adjusted, which can improve the accuracy of absorbing excess energy generated by transient changes in the power amplifier, suppressing signal oscillation and reflection problems, realizing dynamic optimization of output link impedance and transmission characteristics, and improving the stability and transmission quality of the power amplifier output waveform under different transient operating conditions.

[0055] In one embodiment of this application, based on a holding time window and a target equivalent resistance value and a target equivalent inductance value, the equivalent resistance value and equivalent inductance value of the tunable metamaterial structure are adjusted synchronously to obtain the target output transmission characteristics, including: Obtain the current equivalent resistance and maximum adjustable resistance of the tunable metamaterial structure; Input the target energy ratio, the current equivalent resistance value, and the load impedance into the preset RF energy loss model to obtain the initial equivalent resistance value; Compare the initial equivalent resistance value with the maximum adjustable resistance value; If the initial equivalent resistance value is less than or equal to the maximum adjustable resistance value, then the initial equivalent resistance value is used as the target equivalent resistance value. If the initial equivalent resistance value is greater than the maximum adjustable resistance value, the maximum adjustable resistance value is used as the target equivalent resistance value, and a resistance saturation flag is generated.

[0056] In this embodiment, the current equivalent resistance value and the maximum adjustable resistance value of the adjustable metamaterial structure are first obtained. The current equivalent resistance value is the real-time electromagnetic parameter of the metamaterial structure under the current steady-state operating condition, representing the real-time operating state of the power amplifier; the maximum adjustable resistance value is the upper limit of resistance adjustment that the hardware structure of the adjustable metamaterial structure can support, which is a fixed hardware limit parameter and provides a benchmark for subsequent parameter limiting determination.

[0057] Subsequently, the target energy ratio, current equivalent resistance value, and load impedance are input into a preset RF energy loss model to solve for the initial equivalent resistance value adapted to the current transient operating conditions. The RF energy loss model is a mechanistic physical calculation model based on the principles of RF transmission loss and the electromagnetic loss characteristics of metamaterials. The RF energy loss model has three layers: an input parameter layer, an energy loss calculation layer, and a parameter output layer. The input parameter layer receives external real-time operating condition parameters, providing the data foundation for the RF energy loss model's calculations. The energy loss calculation layer, as the core layer of the RF energy loss model, completes energy matching and resistance conversion through a built-in RF loss mechanism formula. The parameter output layer outputs a compliant and usable initial equivalent resistance value, achieving a precise mapping from operating condition parameters to metamaterial electromagnetic parameters. The input parameters of the RF energy loss model include three items: the target energy ratio, the current equivalent resistance value, and the load impedance. The output parameter of the RF energy loss model is the initial equivalent resistance value, which is the theoretically optimal resistance parameter that meets the current target energy absorption requirements.

[0058] Secondly, the initial equivalent resistance value is compared with the maximum adjustable resistance value to complete parameter validity screening and correction. Specifically, if the initial equivalent resistance value is less than or equal to the maximum adjustable resistance value, it indicates that the theoretically solved parameters are within the hardware's adjustable range, and the hardware can respond normally to the control commands. Therefore, the initial equivalent resistance value is used as the target equivalent resistance value to ensure energy absorption accuracy and optimal control. If the initial equivalent resistance value is greater than the maximum adjustable resistance value, it indicates that the theoretical requirement exceeds the hardware's adjustment limit, and the target energy absorption effect cannot be achieved through conventional parameter adjustments. In this case, the maximum adjustable resistance value is forcibly used as the target equivalent resistance value to fully utilize the hardware's maximum control capability. Simultaneously, a resistance saturation flag is generated to record the hardware's saturation state, providing a state basis for subsequent compensation strategy linkage optimization. Finally, in conjunction with the target equivalent inductance value and the hold time window, the metamaterial parameters are coordinated to achieve stable target output transmission characteristics.

[0059] As can be seen from the above, this embodiment obtains the current equivalent resistance value and the maximum adjustable resistance value of the tunable metamaterial structure, calculates the initial equivalent resistance value using the radio frequency energy loss model, and compares the initial equivalent resistance value with the maximum adjustable resistance value. When the initial equivalent resistance value exceeds the limit, the maximum adjustable resistance value is activated and a resistance saturation flag is generated. This can prevent parameter adjustment from exceeding the physical adjustable range of the device, ensuring the safe and stable operation of the tunable metamaterial structure. At the same time, it determines a compliant target equivalent resistance value, providing reliable support for optimizing the target output transmission characteristics.

[0060] In one embodiment of this application, after using the maximum adjustable resistance value as the target equivalent resistance value and generating a resistance saturation flag, the method further includes: The inherent response delay time of the tunable metamaterial structure is obtained, which includes the time required from receiving a control command to the actual establishment of the equivalent resistance value. Based on the resistance saturation flag, a resistance saturation notification is sent to the generation path of the predistortion compensation signal; The effective window for scaling factor adjustment is determined by superimposing the inherent response delay time with the generation time of the resistance saturation notification. Within the effective window, the preset additional compensation value is temporarily added to the scaling factor; After the effective window ends, restore the scaling factor to the level before the additional compensation value was added.

[0061] In this embodiment, the inherent response delay time of the adjustable metamaterial structure is first obtained. This inherent response delay time refers to the time required from receiving the control command to the actual establishment of the equivalent resistance value. It is a fixed timing parameter determined by the hardware response characteristics of the metamaterial structure and represents the hysteresis characteristics of hardware parameter adjustment.

[0062] Secondly, based on the resistance saturation flag, a resistance saturation notification is sent to the generation path of the pre-distortion compensation signal, enabling cross-module linkage and interoperability between the output hardware control state and the input signal compensation strategy. Through synchronous transmission of status signals, the pre-distortion compensation generation module can sense the saturation and confinement state of the metamaterial structure and promptly trigger targeted auxiliary compensation mechanisms.

[0063] Subsequently, the inherent response delay time is superimposed with the generation time of the resistance saturation notification to determine the effective window for scaling factor adjustment. The effective window includes all abnormal timing intervals where hardware saturation and response lag occur, ensuring that the compensation action perfectly matches the timing of hardware defects. Within the defined effective window, a preset additional compensation value is temporarily superimposed on the scaling factor. By instantaneously increasing the predistortion compensation intensity, the insufficient overshoot energy absorption caused by the resistance saturation of the adjustable metamaterial structure is compensated for. This addresses the hardware control shortcomings at the signal predistortion level, effectively suppressing transient overshoot and waveform distortion.

[0064] The preset additional compensation value is a fixed correction amount temporarily superimposed on the scaling factor. It primarily compensates for the energy absorption gap caused by resistance saturation and inherent response delay in the adjustable metamaterial structure. Its setting is determined comprehensively based on the power amplifier's rated parameters, hardware limiting characteristics, and historical operating data. For example, firstly, based on the maximum adjustable resistance value of the adjustable metamaterial structure, the overshoot energy absorption gap under resistance saturation is determined. Then, the dynamic compensation deviation caused by hardware hysteresis is calculated based on the inherent response delay time. Next, saturation error data corresponding to different abrupt change amplitudes and overshoot slopes in historical operating conditions are statistically analyzed, and a standard correction reference amount is fitted to obtain the corrective value. Finally, using the hardware saturation missing energy as the core and historical error data as the calibration basis, a fixed correction amount is determined as the additional compensation value.

[0065] Simultaneously, after the effective operating window ends, the scaling factor is restored to the level before the additional compensation value was added, and the temporary compensation effect is canceled. This dynamic temporary compensation mechanism can accurately adapt to the special operating conditions of metamaterial hardware saturation and response lag. Without affecting the steady-state output quality of the amplifier, it can achieve precise fallback correction of transient defects, greatly improving the adaptability and stability of the entire optimized control scheme under extreme operating conditions.

[0066] As can be seen from the above, after generating the resistance saturation flag, this embodiment obtains the inherent response delay time of the adjustable metamaterial structure and sends a resistance saturation notification. Based on the time parameter, it defines the effective window for scaling factor adjustment, superimposes additional compensation values ​​within the effective window, and restores the original value of the scaling factor after the window ends. This can compensate for the compensation gap caused by the equivalent resistance value reaching the upper limit and the inherent response delay, and realize the coordinated cooperation between hardware and signal compensation, ensuring the continuous and stable transient overshoot suppression effect of the power amplifier.

[0067] In one embodiment of this application, based on the judgment result, the timing prediction neural network, the predistortion compensation signal, and the equivalent electromagnetic parameters are optimized, including: If the judgment result indicates that the residual error characteristics are abnormal, then the abnormality type is extracted; If the anomaly type is that the residual overshoot peak value is greater than the preset first threshold or the residual oscillation duration is greater than the preset second threshold, then a first optimization instruction is generated; the first optimization instruction is used to trigger the online incremental learning of the time-series prediction neural network to update the network weights and re-extract the transient response overshoot features; If the anomaly type is a waveform distortion pattern exhibiting asymmetric overshoot or undercompensation, a second optimization instruction is generated; the second optimization instruction is used to adjust the scaling factor and shaping time constant of the predistortion compensation signal. If the anomaly type is a waveform distortion pattern exhibiting high-frequency ripple or damped oscillation, a third optimization instruction is generated. The third optimization instruction is used to adjust the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure to change the damping coefficient of the output matching network.

[0068] In this embodiment, when the judgment result indicates an abnormality in the residual error characteristics, the abnormality type corresponding to the current output waveform is first extracted. The source of error is distinguished through differentiated distortion features, providing a basis for subsequent path optimization processing. Different residual error abnormality patterns correspond to performance deviations of different modules in the system. Insufficient prediction accuracy, mismatch of compensation parameters, and improper adaptation of hardware electromagnetic parameters can all cause differentiated waveform distortion. By classifying and identifying abnormality types, resource waste and optimization failure caused by blind optimization can be effectively avoided.

[0069] Specifically, if the anomaly type is that the residual overshoot peak value is greater than the preset first threshold or the residual oscillation duration is greater than the preset second threshold, the system generates a first optimization instruction. This type of anomaly mainly stems from the insufficient accuracy of transient feature prediction in the time-series predictive neural network, which cannot capture the precursor information of drastic power changes, leading to pre-prediction bias and delayed compensation control. The first optimization instruction is used to trigger online incremental learning of the time-series predictive neural network, update network weights in real time, correct model prediction bias, and re-extract transient response overshoot features, thereby improving the accuracy of transient feature recognition from the source of prediction and suppressing persistent overshoot and long-term oscillation problems at the root. Among them, the preset first threshold is a quantitative critical value used to judge whether the residual overshoot peak value exceeds the standard, and is the criterion for distinguishing between normal small-amplitude waveform fluctuations and abnormal overshoot distortion. The specific setting method is as follows: first, based on the power amplifier's factory design specifications and rated output waveform parameters, determine the maximum allowable normal overshoot peak value under distortion-free conditions; then, based on the actual measured data of the equipment under full operating conditions, superimpose the reasonable error margin caused by environmental interference and normal operating condition fluctuations, and comprehensively define the benchmark value. Simultaneously, based on different loads and power surge scenarios, tiered adaptation is implemented, and this fixed value is ultimately set as the first threshold. When the residual overshoot peak value exceeds the first threshold, it is judged as abnormal. The preset second threshold is used as the time critical value for judging whether the residual oscillation duration exceeds the standard, and is used to distinguish between short-term normal waveform jitter and continuous abnormal oscillation. The specific setting method is as follows: according to the transient response standard requirements of the power amplifier, the longest duration of waveform oscillation when the power amplifier is working normally is statistically analyzed, and based on this, a fault tolerance time is reserved according to the allowable fluctuation range under different operating conditions. Short-term oscillation data caused by transient interference is eliminated, and a reasonable upper limit of time is determined as the second threshold. Once the measured residual oscillation duration exceeds the second threshold, the oscillation problem is judged as abnormal, and the corresponding optimization process is triggered.

[0070] If the anomaly type is a waveform distortion exhibiting asymmetric overshoot or undercompensation, a second optimization instruction is generated. This type of waveform defect is mainly caused by insufficient amplitude matching of the predistortion compensation signal and an envelope shaping rate that is not adapted to the current transient conditions, belonging to parameter deviations at the signal compensation level. The second optimization instruction is used to selectively adjust the scaling factor and shaping time constant of the predistortion compensation signal, adaptively correcting the amplitude intensity and envelope change rate of the compensation signal, adapting to the current transient change characteristics, reducing asymmetric distortion and undercompensation problems, and improving the dynamic matching accuracy of signal predistortion.

[0071] If the anomaly is a waveform distortion exhibiting high-frequency ripple or damped oscillation, a third optimization instruction is generated. This type of anomaly is often caused by poor damping characteristics of the output matching network or an imbalance in the electromagnetic parameters of the metamaterial. The third optimization instruction adjusts the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure. By optimizing the hardware equivalent electromagnetic parameters, it changes the damping coefficient of the output matching network, optimizes the RF link resonance and loss characteristics, and effectively suppresses high-frequency ripple and continuous damped oscillation.

[0072] As can be seen from the above, this embodiment distinguishes different anomaly types and generates three types of optimization instructions when residual error characteristics are abnormal. It triggers online incremental learning and updates the network weights of the timing prediction neural network for abnormal residual overshoot peak value and residual oscillation duration, and re-extracts transient response overshoot features. It adjusts the scaling factor and shaping time constant of the pre-distortion compensation signal for asymmetric overshoot and undercompensation morphology. It adjusts the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure to change the damping coefficient of the output matching network for high-frequency ripple and damped oscillation morphology. This allows for more accurate location of the root cause of the problem and targeted optimization, forming a complete closed-loop control mechanism, continuously correcting system deviations, effectively reducing various waveform anomalies, and stabilizing the overall dynamic operation performance of the power amplifier.

[0073] In one embodiment of this application, before determining whether the residual error characteristics are abnormal and obtaining the determination result, the method further includes: The residual error features are input into a preset event detector, which is used to identify whether the rate of change of the residual error features is greater than a preset event threshold. If the rate of change is less than the event threshold, the current state is determined to be a steady-state maintenance phase, and no optimization instructions are triggered. If the rate of change is greater than the event threshold, it is determined that a transient event has occurred, and the timestamp and intensity of the event are recorded. The event intensity is matched with similar events in the preset historical event database. If a similar event is matched, the corresponding historical optimized instruction sequence for that similar event is invoked for open-loop execution. If no similar event is found, optimization is performed based on the first optimization instruction, the second optimization instruction, or the third optimization instruction.

[0074] In this embodiment, residual error characteristics are input into a preset event detector. The event detector monitors the dynamic fluctuation trend of the residual error characteristics and continuously identifies whether the rate of change of the residual error characteristics is greater than a preset event threshold. The event threshold is used as a criterion to distinguish between steady-state operating conditions and transient abnormal events. The preset event detector is a detection module integrated into the system. The function of the event detector is to calculate the rate of change of the residual error characteristics and to complete the operating condition judgment according to the preset event threshold, distinguishing between the steady-state operating state and transient abnormal events of the power amplifier. The preset event threshold is a quantitative critical judgment standard, using the rate of change of the residual error characteristics as the evaluation object, used to divide the boundary between steady-state micro-fluctuations and transient abnormal changes. When the rate of change is less than the event threshold, it is judged as normal steady state; when the rate of change is greater than the event threshold, it is judged as a transient event. The setting method includes statistically analyzing the residual error characteristic data under long-term steady-state operation of the power amplifier and calculating the maximum value of all rates of change within the normal fluctuation range, which is used as a basic reference value. Then, based on the allowable normal waveform jitter range of the power amplifier, the error fluctuation caused by external environmental interference, and a reasonable fault tolerance margin, the event threshold is obtained.

[0075] Specifically, when the rate of change of the residual error characteristic is less than the event threshold, it indicates that the current power amplifier output waveform error changes smoothly and the operating condition is stable. The residual error is only a small fluctuation in normal steady state, with no new transient distortion or abnormal disturbance. Based on this, it is determined that the current state is in a steady-state maintenance phase, and no optimization instructions are triggered to avoid meaningless iterative corrections to the timing prediction neural network, predistortion compensation signal, and equivalent electromagnetic parameters.

[0076] When the rate of change of the residual error characteristics exceeds the event threshold, a transient event is determined to have occurred. This indicates an abnormal situation such as sudden power fluctuations or load disturbances in the current operating condition, with the waveform residual error showing a rapid deterioration trend. The timestamp and intensity of the event are recorded. The intensity of the current transient event is then iterated and matched against similar events in a preset historical event database.

[0077] Specifically, if a similar event is matched, the corresponding historical optimization instruction sequence is invoked for open-loop execution, without the need for repeated calculations and iterative optimization of parameters. If no similar event is matched, it indicates that the current transient condition is a completely new type of disturbance, and there is no mature historical optimization experience to reuse. In this case, closed-loop optimization is carried out based on the first, second, or third optimization instructions, by correcting the weights of the time-series prediction neural network, the pre-distortion compensation signal parameters, and the equivalent electromagnetic parameters of the tunable metamaterial structure for the anomaly type.

[0078] As can be seen from the above, this embodiment effectively distinguishes between the steady-state operation and transient abnormal conditions of the power amplifier by inputting residual error features into a preset event detector and using event threshold discrimination, transient event recording, historical event matching, and differentiated optimization execution mechanisms. This avoids invalid iterative optimization in the steady-state stage, reducing algorithm computational overhead and frequent parameter jitter. Simultaneously, by identifying the rate of change of residual error features, the occurrence time and disturbance intensity of transient events are captured. Based on the historical event database, rapid matching of similar operating conditions is achieved, enabling the reuse of mature historical optimization instruction sequences for rapid open-loop correction, effectively improving the response speed of transient faults. For novel error anomalies without historical matching samples, switching to the first, second, or third optimization instruction completes precise closed-loop optimization, improving the dynamic response speed and adaptability of the power amplifier to unknown operating conditions, effectively enhancing the stability and adaptive optimization performance of the power amplifier under all operating conditions.

[0079] Corresponding to the amplifier dynamic performance optimization control method in the above embodiment, Figure 2 This is a structural block diagram of an amplifier dynamic performance optimization control system provided in one embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The amplifier dynamic performance optimization control system 20 includes: a working condition data acquisition and prediction module 21, a predistortion signal generation and superposition module 22, a predistortion signal generation and superposition module 23, an output waveform error extraction module 24, an error state anomaly determination module 25, and a global parameter iterative optimization module 26.

[0080] Among them, the operating condition data acquisition and prediction module 21 is used to acquire the operating condition data of the power amplifier and input the operating condition data into the pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier. The predistortion signal generation and superposition module 22 is used to generate a predistortion compensation signal that is opposite to the transient change trend based on the transient response overshoot characteristics; and to superimpose the predistortion compensation signal into the input signal of the power amplifier in advance to obtain the compensated input signal. The predistortion signal generation and superposition module 23 is used to adjust the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network based on the predicted transient response overshoot characteristics to obtain the target output transmission characteristics. The output waveform error extraction module 24 is used to monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and extract the residual error characteristics of the output waveform. Error state anomaly determination module 25 is used to determine whether the residual error characteristics are abnormal and obtain the determination result; The global parameter iterative optimization module 26 is used to optimize the time-series prediction neural network, predistortion compensation signal and equivalent electromagnetic parameters based on the judgment results.

[0081] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the working condition data acquisition and prediction module 21, the predistortion signal generation and superposition module 22, the predistortion signal generation and superposition module 23, the output waveform error extraction module 24, the error state anomaly judgment module 25, and the global parameter iterative optimization module 26 are shown.

[0082] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0083] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0084] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0085] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the amplifier dynamic performance optimization control method provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0086] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0087] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0088] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0089] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0090] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0091] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0092] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0093] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An amplifier dynamic performance optimization control method, characterized by, include: The operating condition data of the power amplifier is collected and input into a pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier. Based on the transient response overshoot characteristics, a pre-distortion compensation signal with the opposite trend to the transient change is generated; the pre-distortion compensation signal is superimposed on the input signal of the power amplifier in advance to obtain the compensated input signal; Based on the predicted transient response overshoot characteristics, the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network are adjusted to obtain the target output transmission characteristics. Monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and extract the residual error characteristics of the output waveform; Determine whether the residual error characteristics are abnormal, and obtain the determination result; Based on the judgment result, the time-series prediction neural network, the predistortion compensation signal, and the equivalent electromagnetic parameters are optimized.

2. The method of claim 1, wherein The predicted transient response overshoot characteristics include the start time, magnitude, overshoot slope, and time of overshoot peak occurrence of the power demand mutation. The step of generating a predistortion compensation signal that is opposite to the transient change trend based on the predicted transient response overshoot characteristics includes: Based on the magnitude of the power demand mutation, the overshoot slope, and the total transmission delay in the predicted transient response overshoot characteristics, the timing advance of the predistortion compensation signal is calculated. By using the start time of the power demand mutation and the timing advance, the injection start time and duration of the predistortion compensation signal are determined, and the predistortion compensation timing parameters are obtained. Based on the load impedance and reflected power in the operating condition data, the phase of the predistortion compensation signal is adjusted to obtain the phase-calibrated predistortion compensation signal. The input baseband signal is processed based on the predistortion compensation timing parameters and the phase-calibrated predistortion compensation signal to obtain the baseband predistortion component; Based on the overshoot slope and the abrupt change amplitude, the baseband predistortion component is subjected to amplitude scaling and envelope shaping to obtain a predistortion compensation signal that is opposite to the transient change trend.

3. The amplifier dynamic performance optimization control method according to claim 2, characterized in that, The step of scaling and envelope shaping the baseband predistortion component based on the overshoot slope and the abrupt change amplitude to obtain a predistortion compensation signal with a trend opposite to the transient change includes: Based on the mutation amplitude, a scaling factor is determined; based on the overshoot slope, a shaping time constant is determined, wherein a larger mutation amplitude results in a larger scaling factor, and a larger overshoot slope results in a smaller shaping time constant. Multiply the amplitude of the baseband predistortion component by the scaling factor to obtain the amplitude-scaled predistortion component; The envelope of the amplitude-scaled predistortion component is passed through a low-pass filter to obtain the envelope-shaped predistortion component, wherein the time constant of the low-pass filter is equal to the shaping time constant. The envelope-shaped predistortion component is output as a predistortion compensation signal that is opposite to the transient change trend.

4. The amplifier dynamic performance optimization control method according to claim 3, characterized in that, After passing the envelope of the amplitude-scaled predistortion component through a low-pass filter to obtain the envelope-shaped predistortion component, the method further includes: The envelope-shaped pre-distortion component is compared with the amplitude-scaled pre-distortion component, and the envelope difference is calculated. When the envelope difference is greater than the preset difference threshold, the shaping time constant is adjusted according to the relationship between the direction and magnitude of the envelope difference to obtain the target shaping time constant. The target shaping time constant is applied to the low-pass filter to perform secondary filtering on the envelope of the amplitude-scaled pre-distortion component until the envelope difference is reduced to within the difference threshold or the preset maximum number of filtering times is reached.

5. The amplifier dynamic performance optimization control method according to claim 2, characterized in that, The step of adjusting the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network based on the predicted transient response overshoot characteristics to obtain the target output transmission characteristics includes: Based on the abrupt change amplitude, the duration of the overshoot rise edge, and the load impedance and reflected power in the operating data, the overshoot energy value is calculated. Based on the overshoot energy value, determine the target energy ratio that the tunable metamaterial structure needs to absorb; Based on the acquired RF impedance loss characteristics, resonance characteristics, and current load state, the target equivalent resistance and target equivalent inductance values ​​of the tunable metamaterial structure are calculated using the target energy ratio. Based on the start time of the power demand mutation, the time of the overshoot peak occurrence, and the recovery time of the overshoot oscillation, a holding time window for the target equivalent resistance value and the target equivalent inductance value is determined. The holding time window includes the start time of the power demand mutation to the recovery time of the overshoot oscillation. Based on the holding time window and the target equivalent resistance and the target equivalent inductance, the equivalent resistance and equivalent inductance of the adjustable metamaterial structure are adjusted synchronously to obtain the target output transmission characteristics.

6. The amplifier dynamic performance optimization control method according to claim 5, characterized in that, The step of simultaneously adjusting the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure based on the holding time window, the target equivalent resistance value, and the target equivalent inductance value to obtain the target output transmission characteristics includes: Obtain the current equivalent resistance and maximum adjustable resistance of the tunable metamaterial structure; The target energy ratio, the current equivalent resistance value, and the load impedance are input into a preset RF energy loss model to obtain the initial equivalent resistance value. Compare the initial equivalent resistance value with the maximum adjustable resistance value; If the initial equivalent resistance value is less than or equal to the maximum adjustable resistance value, then the initial equivalent resistance value is taken as the target equivalent resistance value. If the initial equivalent resistance value is greater than the maximum adjustable resistance value, then the maximum adjustable resistance value is used as the target equivalent resistance value, and a resistance saturation flag is generated.

7. The amplifier dynamic performance optimization control method according to claim 6, characterized in that, After setting the maximum adjustable resistance value as the target equivalent resistance value and generating a resistance saturation flag, the method further includes: The inherent response delay time of the tunable metamaterial structure is obtained, and the inherent response delay time includes the time required from receiving a control command to the actual establishment of the equivalent resistance value. Based on the resistance saturation flag, a resistance saturation notification is sent to the generation path of the pre-distortion compensation signal; The inherent response delay time is superimposed with the generation time of the resistance saturation notification to determine the effective window for scaling factor adjustment; Within the effective operating window, a preset additional compensation value is temporarily superimposed on the scaling factor; After the effective window ends, the scaling factor is restored to the level before the additional compensation value was added.

8. The amplifier dynamic performance optimization control method according to claim 1, characterized in that, The optimization of the time-series prediction neural network, the predistortion compensation signal, and the equivalent electromagnetic parameters based on the judgment result includes: If the judgment result indicates that the residual error characteristics are abnormal, then the abnormality type is extracted; If the anomaly type is that the residual overshoot peak value is greater than a preset first threshold or the residual oscillation duration is greater than a preset second threshold, then a first optimization instruction is generated; the first optimization instruction is used to trigger online incremental learning of the time-series prediction neural network to update the network weights and re-extract the transient response overshoot features; If the anomaly type is a waveform distortion pattern exhibiting asymmetric overshoot or undercompensation, a second optimization instruction is generated; the second optimization instruction is used to adjust the scaling factor and shaping time constant of the pre-distortion compensation signal. If the anomaly type is a waveform distortion pattern exhibiting high-frequency ripple or damped oscillation, a third optimization instruction is generated; the third optimization instruction is used to adjust the equivalent resistance and equivalent inductance values ​​of the adjustable metamaterial structure to change the damping coefficient of the output matching network.

9. The amplifier dynamic performance optimization control method according to claim 8, characterized in that, Before determining whether the residual error characteristic is abnormal and obtaining the determination result, the method further includes: The residual error feature is input into a preset event detector, which is used to identify whether the rate of change of the residual error feature is greater than a preset event threshold. If the rate of change is less than the event threshold, the current state is determined to be a steady-state maintenance phase, and no optimization instructions are triggered. If the rate of change is greater than the event threshold, it is determined that a transient event has occurred, and the timestamp and intensity of the event are recorded. The event intensity is matched with similar events in the preset historical event database. If a similar event is matched, the corresponding historical optimized instruction sequence for that similar event is invoked for open-loop execution. If no similar event is matched, optimization processing is performed based on the first optimization instruction, the second optimization instruction, or the third optimization instruction.

10. An amplifier dynamic performance optimization control system, characterized in that, include: The operating condition data acquisition and prediction module is used to acquire the operating condition data of the power amplifier and input the operating condition data into a pre-trained time-series prediction neural network in real time to obtain the predicted transient response overshoot characteristics of the power amplifier. The predistortion signal generation and superposition module is used to generate a predistortion compensation signal that is opposite to the transient change trend based on the transient response overshoot characteristics; and to superimpose the predistortion compensation signal onto the input signal of the power amplifier in advance to obtain the compensated input signal. The predistortion signal generation and superposition module is used to adjust the equivalent electromagnetic parameters of the tunable metamaterial structure integrated in the power amplifier output matching network based on the predicted transient response overshoot characteristics, so as to obtain the target output transmission characteristics. The output waveform error extraction module is used to monitor the output waveform generated by the compensated input signal and the target output transmission characteristics, and to extract the residual error characteristics of the output waveform. An error state anomaly determination module is used to determine whether the residual error characteristics are abnormal and to obtain a determination result; The global parameter iterative optimization module is used to optimize the time-series predictive neural network, the pre-distortion compensation signal, and the equivalent electromagnetic parameters based on the judgment result.

Images