Frequency-adjustable power source with calibration function

Abstract

The invention discloses a frequency-adjustable power source with a calibration function, and relates to the field of power sources, and the power source comprises a frequency adjustment module which is used for generating a radio frequency signal according to a frequency control signal; the power amplification module is used for amplifying and outputting the radio frequency signal; the calibration module is used for acquiring output signal characteristics and calibrating the frequency adjusting module; the control module is used for generating the frequency control signal and adjusting according to calibration feedback; a pre-distortion and analog calibration channel; and a digital twin optimization engine. Transmission parameters of the power amplification module are dynamically configured through the pre-distortion and analog calibration channel according to non-linear characteristics actually measured by the power amplification module, feedforward amplitude-phase pre-compensation is carried out on signals, and the influence of non-linear distortion on the quality of output signals is restrained from the source.

Description

Technical Field

[0001] This invention relates to the field of power sources, specifically to a frequency-tunable power source with calibration function. Background Technology

[0002] A frequency-tunable power source is an electronic device capable of generating a specific frequency radio frequency signal and providing a certain power output. It is widely used in radar, communications, medical instruments (such as linear accelerators), industrial heating, and scientific experiments. Its core performance indicators include frequency accuracy, output power stability, spectral purity, and long-term operational reliability. With the increasing demands for frequency agility, power control precision, and adaptive capabilities in systems, frequency-tunable power sources with high-performance calibration functions have become key components in related high-end equipment.

[0003] In existing technologies, frequency-adjustable power sources with calibration functions are typically implemented based on analog or digital control loops. A typical solution includes a voltage-controlled oscillator (VCO) or direct digital frequency synthesizer (DDS) as the core for frequency adjustment, followed by a power amplifier for signal amplification. To maintain stable output frequency and power, the system samples the output signal using a detector, coupler, or phase detector, compares the sampled signal with a reference standard, and processes the resulting error signal before feeding it back to the frequency adjustment unit or power control unit, forming a closed-loop control to correct deviations. Some solutions incorporate temperature compensation circuits or lookup tables to address the effects of environmental changes, ensuring that performance requirements are met across the entire temperature range.

[0004] However, the aforementioned existing technical solutions still have several limitations in practical applications. First, traditional feedback calibration loops rely heavily on the direct processing of real-time error signals. During calibration initiation or sudden environmental changes, they are prone to transient overshoot, oscillations, or even lockouts in the output signal due to loop response lag or overcompensation, affecting system stability and response speed. Second, conventional calibration strategies are often singular and passive, making it difficult to predict and optimize the system's dynamic behavior before calibration, and unable to provide feedforward compensation for the inherent nonlinear distortion of power amplifiers. Furthermore, existing systems lack the ability to predict and proactively manage the performance degradation of their key components, failing to implement "soft" parameter adjustments before performance decline due to component aging, potentially leading to sudden changes or interruptions in output performance. In addition, traditional systems typically lack the ability to intelligently identify and classify complex error sources (such as thermal drift, power supply noise, and load disturbances), resulting in a lack of targeted calibration strategies and impacting the final calibration effect and efficiency. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a frequency-adjustable power source with calibration function to solve the technical problems in the prior art, such as output transient instability caused by lag or overcompensation in the calibration loop response, signal quality affected by nonlinear distortion of the power amplifier, sudden changes in output performance caused by lack of predictive management of device performance degradation, and lack of targeted calibration strategies due to the inability to intelligently identify the source of error.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a frequency-tunable power source with calibration function, comprising: a frequency adjustment module for generating a radio frequency signal according to a frequency control signal; a power amplification module for amplifying and outputting the radio frequency signal; a calibration module for acquiring output signal characteristics and calibrating the frequency adjustment module; and a control module for generating the frequency control signal and adjusting it according to calibration feedback; wherein, it further comprises: a predistortion and analog calibration channel, and a digital twin optimization engine; the predistortion and analog calibration channel is connected in parallel between the frequency adjustment module and the power amplification module, and includes an analog circuit network with adjustable transmission characteristics; the control module... The control module can dynamically configure the transmission parameters of the analog circuit network according to the instructions of the digital twin optimization engine, so that it generates a predetermined amplitude and phase predistortion for the input radio frequency signal; the digital twin optimization engine has a built-in digital twin model of the power source; the engine is configured to: receive real-time monitoring data from the calibration module and the operating status parameters of the power amplifier module, and drive the digital twin model to perform synchronous simulation; based on the simulation results, predict the transient overshoot or oscillation risks that may occur when directly performing closed-loop calibration under the current configuration, and generate a calibration strategy that includes the predistortion parameter pilot adjustment instructions for the predistortion and analog calibration channels and the progressive frequency control signal adjustment sequence.

[0007] The present invention is further configured such that the transmission parameters of the analog circuit network of the pre-distortion and analog calibration channel are configured based on the measurement results of the nonlinear amplitude and phase characteristics of the power amplification module at the current operating frequency and power, so as to pre-compensate for the nonlinear distortion introduced by the power amplification module.

[0008] The present invention is further configured such that the digital twin optimization engine is also configured to perform virtual calibration iteration: before the physical power source actually performs a calibration adjustment, multiple rapid simulation iterations are performed on the digital twin model according to the proposed calibration strategy to evaluate the stability speed and overshoot of different strategies, and the optimal one from multiple candidate strategies is selected and output to the physical system for execution.

[0009] The invention is further configured to include: a lifetime prediction and parameter mitigation unit; the lifetime prediction and parameter mitigation unit is connected to the temperature and current monitoring points of the key components of the power amplifier module, and the digital twin optimization engine; the lifetime prediction and parameter mitigation unit is configured to: predict the performance degradation trend and remaining reliable lifetime of the key components based on real-time stress data and historical cumulative operating time; when it is predicted that performance degradation may affect the output indicators in a specific period of the future, issue an early warning to the digital twin optimization engine; after receiving the early warning, the digital twin optimization engine actively and gradually introduces compensation parameters that match the predicted performance degradation value in the subsequently generated optimization calibration strategy, so that the output performance of the power source remains stable during the actual aging process of the device, and achieves a smooth transition of output performance.

[0010] The present invention is further configured such that the lifetime prediction and parameter mitigation unit is also configured to: record the correspondence between historical warnings and actual performance degradation, and use the correspondence to continuously correct its prediction model.

[0011] The present invention is further configured such that, when generating the progressive frequency control signal adjustment sequence, the digital twin optimization engine adopts a non-uniform time step, using a smaller step size and a denser period in the early stage of adjustment, and a larger step size and a sparser period in the later stage of adjustment.

[0012] The present invention is further configured to include a dynamic reference generation unit; the dynamic reference generation unit receives instructions from the digital twin optimization engine and is capable of generating a non-fixed value dynamic reference signal to replace the traditional fixed value reference source provided to the calibration module; wherein, the value of the dynamic reference signal is calculated in real time by the digital twin optimization engine based on the current target output frequency, power and ambient temperature by querying a pre-stored optimal operating point mapping table.

[0013] The present invention is further configured such that the calibration module includes a self-learning error classifier; the self-learning error classifier performs real-time feature extraction and pattern recognition on the monitored error signals, distinguishes whether the error mainly originates from thermal effects, power supply noise or load changes, and sends the classification results and confidence levels to the digital twin optimization engine; the digital twin optimization engine calls different twin model sub-modules or adjusts model parameters for different types of error-dominant modes to generate more targeted calibration strategies.

[0014] The present invention is further configured such that the digital twin model is a model containing multi-physics coupling of electrical, thermal, and stress fields, and the key node temperature data calculated by its simulation is used to correct the compensation coefficient of the temperature-sensitive element in the calibration module in real time.

[0015] This invention also discloses a power source calibration method based on digital twins, applied to the aforementioned frequency-tunable power source with calibration function, comprising the following steps: establishing and maintaining a digital twin model of the power source; real-time acquisition of operating status data and calibration monitoring data of the physical power source, driving the digital twin model to run synchronously; when calibration is required, performing virtual calibration iteration in the digital twin model to evaluate the transient and steady-state effects of different calibration strategies; generating an optimized physical calibration strategy based on the virtual iteration results, the physical calibration strategy including at least a pilot adjustment command for the predistortion channel and a progressive frequency adjustment sequence; controlling the physical power source to perform calibration operations according to the optimized physical calibration strategy; during execution, dynamically introducing compensation parameters matching the predicted performance degradation into the optimized physical calibration strategy based on lifetime prediction information.

[0016] In summary, the present invention has the following main beneficial effects: This invention effectively improves the calibration performance and long-term stability of a power source by introducing a digital twin optimization engine and a pre-distortion simulation calibration channel. Specifically, the digital twin optimization engine, through real-time synchronous simulation and virtual iteration, can predict and avoid the transient overshoot and oscillation risks that are prone to occur in traditional closed-loop calibration before actual calibration, thereby significantly improving the stability and response speed of the calibration process. At the same time, the pre-distortion and simulation calibration channels dynamically configure their transmission parameters according to the measured nonlinear characteristics of the power amplifier module, and perform feedforward amplitude and phase pre-compensation on the signal, suppressing the impact of nonlinear distortion on the output signal quality from the source. Furthermore, the integrated lifetime prediction and parameter mitigation unit predicts the performance degradation trend of key components by monitoring their stress and accumulated operating data, and drives the digital twin engine to proactively and progressively introduce compensation parameters in subsequent calibration strategies, achieving a "soft landing" of output performance during the device aging process and avoiding system interruptions caused by sudden performance changes. In addition, the self-learning error classifier in the calibration module can identify in real time that the error mainly comes from thermal effects, power supply noise or load changes, enabling the digital twin engine to call targeted models and strategies for calibration, which greatly improves the accuracy and adaptability of calibration. Attached Figure Description

[0017] Figure 1 This is a diagram of the overall architecture of the present invention; Figure 2 This is a flowchart of the calibration method of the present invention; Figure 3 This is a virtual iterative logic diagram of the digital twin engine of the present invention; Figure 4 This is a schematic diagram illustrating the linkage between lifetime prediction and calibration compensation in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0019] The embodiments of the present invention will now be described.

[0020] This invention provides a frequency-tunable power source with calibration functionality, aiming to solve the technical problems in existing technologies, such as transient instability caused by calibration loop response lag, signal quality issues due to power amplifier nonlinear distortion, sudden output performance changes due to lack of predictive management of device performance degradation, and a lack of targeted calibration strategies due to the inability to intelligently identify error sources. The core of this invention lies in introducing a digital twin optimization engine and pre-distortion and analog calibration channels to construct an intelligent calibration system with predictive, adaptive, and highly stable capabilities. The following detailed description, using a preferred embodiment, will cover everything from underlying technical details to the upper-level solution.

[0021] System overall architecture and working principle: refer to Figure 1 The frequency-adjustable power source with calibration function provided in this embodiment of the invention mainly includes: a frequency adjustment module, a power amplification module, a calibration module, a control module, a predistortion and analog calibration channel, a digital twin optimization engine, a lifetime prediction and parameter mitigation unit, and a dynamic reference generation unit. All modules work collaboratively under the coordination of the control module through electrical connections or a data bus.

[0022] Its basic workflow is as follows: The control module generates an initial frequency control signal based on external instructions (such as target frequency and target power) and sends it to the frequency adjustment module. The frequency adjustment module then generates an initial radio frequency (RF) signal. This RF signal is not directly sent to the power amplifier module, but first enters the pre-distortion and analog calibration channel. According to the instructions of the digital twin optimization engine, the amplitude and phase of the signal are pre-adjusted (i.e., pre-distorted). The pre-distorted signal is then sent to the power amplifier module for amplification and finally output from the output port.

[0023] Meanwhile, the calibration module acquires the signal characteristics of the power amplifier module's output signal in real time through sampling devices such as directional couplers (e.g., obtaining the phase difference with the reference signal through a phase detector, and obtaining the power value through a power detector). This real-time monitoring data, along with the power amplifier module's operating status parameters (such as power amplifier die temperature, supply voltage, and current), is fed into the digital twin optimization engine. The digital twin optimization engine drives its internal digital twin model to perform synchronous simulation, predicting the risks that might arise from direct feedback calibration in the traditional way, and generating an optimized calibration strategy. This strategy includes not only a progressive frequency control signal adjustment sequence with specific timing and amplitude sent to the control module, but also adjustment instructions for the pre-distortion channel parameters issued before frequency adjustment. Based on this strategy, the control module coordinates the control of the frequency adjustment module and the pre-distortion channel to achieve accurate, rapid, and stable calibration of the output frequency and signal quality.

[0024] Detailed implementation methods for each module: Frequency modulation module: Its core can be a direct digital frequency synthesizer (DDS) or a phase-locked loop (PLL) frequency synthesizer. For example, an integrated DDS chip can be used, with its reference clock provided by a high-stability temperature-controlled crystal oscillator. The control module writes frequency control words to the DDS chip via the SPI bus, thereby precisely controlling the frequency of its output signal. Its frequency modulation range covers the S-band (e.g., 2.8 GHz to 2.9 GHz), with a minimum frequency step of 1 kHz or less to meet high-precision frequency modulation requirements. The RF signal output by this module is a low-power signal and requires subsequent amplification.

[0025] Predistortion and Analog Calibration Channel: This channel is connected in parallel between the output of the frequency adjustment module and the input of the power amplifier module. Specifically, it consists of a cascaded numerically controlled attenuator and a numerically controlled phase shifter. The numerically controlled attenuator can be a GaAs MMIC digital attenuator chip with an attenuation range of, for example, 0-31.5 dB in 0.5 dB steps. The numerically controlled phase shifter can be a similar MMIC digital phase shifter chip with a phase shift range of 0-360° in 5.625° steps. The states (attenuation value, phase shift value) of these two devices are set by the control module via parallel control lines or a serial bus according to instructions from the digital twin optimization engine.

[0026] Transmission parameter configuration principle: Before or periodically, the nonlinear characteristics of the power amplifier module need to be measured. The specific method is as follows: A swept or single-frequency signal is output through the frequency adjustment module. With the predistortion channel closed (attenuation set to 0dB, phase shift set to 0°), the gain compression (AM-AM characteristics) and phase shift (AM-PM characteristics) of the power amplifier module under different input powers are measured, forming a set of nonlinear characteristic data tables. After obtaining this table, the digital twin optimization engine calculates, through algorithms (such as lookup table inversion or polynomial fitting inversion), the amount of amplitude and phase predistortion required to be injected before the power amplifier to compensate for the nonlinear distortion at a specific operating point (specific frequency, specific target output power), i.e., the attenuation and phase shift values ​​that should be set in the predistortion channel. This achieves feedforward pre-compensation for the nonlinear distortion introduced by the power amplifier module, rather than traditional feedback compensation.

[0027] Power amplifier module: Composed of multi-stage RF amplifiers, the final stage uses S-band power transistors (such as GaN HEMTs) to achieve pulsed peak power output from 50W to 300W. The module integrates temperature sensors (such as thermistors or thermocouples) and current sampling circuits for real-time monitoring of the temperature and current of critical components; this data is crucial input for lifetime prediction. An isolator is connected in series at the module output to improve the output VSWR and protect the power amplifier transistors from reflected power damage.

[0028] Calibration module: Its core function is to acquire the characteristics of the output signal. In this embodiment, it specifically includes: Reference signal channel: A signal is coupled out from the local oscillator or output terminal of the frequency adjustment module as a reference signal.

[0029] Feedback signal channel: A signal is coupled out from the output of the power amplifier module through a directional coupler as a feedback signal.

[0030] Phase detection and detection unit: A vector phase detector (e.g., I / Q phase detector) is used to compare the phase and amplitude of the reference signal and the feedback signal, and outputs DC voltage signals (I signal and Q signal) that are proportional to the ratio of their phase difference and amplitude. These DC voltage signals characterize the signal characteristics of the output signal.

[0031] Self-learning error classifier: This is a software algorithm module implemented in a field-programmable gate array (FPGA) or digital signal processor (DSP). It continuously receives the I and Q signal sequences output from the phase detector. By performing real-time feature extraction on this sequence (such as calculating the slope of long-term drift, analyzing the spectral components of short-term disturbances, and statistically analyzing noise variance), and running a pattern recognition algorithm (such as a threshold-based decision tree or a simple neural network model), it distinguishes whether the main characteristic pattern of the current error signal is slow temperature drift, power frequency-related power ripple noise, or sudden load impedance changes. The classification results and their confidence scores are sent to the digital twin optimization engine.

[0032] Control Module: As the system's central control unit, it can be implemented using a microcontroller (MCU) or a microprocessor (MPU). It is responsible for receiving all external commands and internal status data, running the main control program, and specifically executing the calibration strategy issued by the digital twin optimization engine. Its key tasks include: achieving remote numerical control via communication interfaces (such as RS-485, Ethernet); generating precise timing frequency control signals and predistortion channel control signals; and managing the dynamic reference generation unit.

[0033] Dynamic Reference Generation Unit: This unit consists of a high-precision digital-to-analog converter (DAC). In traditional calibration, the phase detector is typically compared to a reference source with a fixed voltage. In this invention, this reference voltage is no longer fixed. The digital twin optimization engine, based on the currently set target output frequency, target power, and ambient temperature, queries a pre-stored optimal operating point mapping table, calculates an optimal reference voltage value in real time, and commands the DAC to output this value as a dynamic reference signal to the phase detector in the calibration module via the control module. This allows the system to automatically adjust the calibration "zero point" under different operating conditions, improving calibration accuracy and adaptability.

[0034] Digital Twin Optimization Engine: This is the intelligent core of the invention and can be deployed on the same processor as the control module, or on a more powerful coprocessor. It includes: Digital Twin Model: This is a virtual model of a power source that incorporates multi-physics coupling of electrical, thermal, and stress fields. The electrical components are built based on the S-parameter models and nonlinear behavior models of the frequency adjustment module, predistortion channel, and power amplification module; the thermal components are built based on the encapsulated thermal resistance model and heat sink model of the power amplification module; and the stress model is correlated with electrothermal parameters. This model can be exported using model reduction techniques in commercial simulation software (such as ADS and ANSYS) or embedded as empirical formulas.

[0035] Model synchronization and virtual calibration iteration: such as Figure 3As shown, after receiving monitoring data from the calibration module, the engine drives the twin model to run synchronously, ensuring that the "output" of the virtual model is consistent with the current actual output of the physical entity. When calibration is required (e.g., when a frequency error is detected), the engine does not immediately command the physical system to act. Instead, it first performs virtual calibration iterations on the twin model: simulating the injection of the same error into the model, and then trying various different calibration strategies (e.g., different predistortion parameter adjustments, different frequency adjustment step sizes, and adjustments by combination). The engine rapidly (much faster than in real time) evaluates the stabilization process of the model output under each strategy in the simulation environment, recording metrics such as stabilization time and overshoot. Finally, it selects the optimal strategy from multiple candidate strategies (e.g., short stabilization time and minimal overshoot) as the optimized physical calibration strategy to be executed on the physical system. This strategy explicitly includes the steps of first adjusting the predistortion channel parameters and then adjusting the frequency according to a specific asymptotic sequence.

[0036] Policy generation and coordination: For different error patterns reported by the self-learning error classifier, the engine can call different model sub-modules or adjust the model parameter weights. For example, if the error is classified as "thermal effect", the engine will focus more on the response of the thermal model in simulation and policy generation; if it is classified as "power supply noise", it may generate a slow adjustment sequence with filtering characteristics to avoid over-responding to noise.

[0037] Lifetime prediction and parameter mitigation unit: This is a module implemented based on software algorithms. It continuously reads the real-time junction temperature (estimated through a thermal model) and collector current data of key components of the power amplifier module (such as the final stage power transistor) and accumulates their operating time.

[0038] Prediction Model: A temperature-accelerated lifetime model based on the Arrhenius equation or a fatigue accumulation model based on rainflow counting is used to predict the performance degradation trend (such as gain reduction and saturated output power reduction) and remaining reliable lifetime of the device.

[0039] Early warning and mitigation: such as Figure 4 As shown, when the prediction model indicates that, based on the current degradation trend, a key performance characteristic of the device (such as gain) will decrease by more than a preset threshold (e.g., 1 dB) within the next N hours, this unit issues a warning to the digital twin optimization engine. When generating subsequent calibration strategies, the engine does not wait for actual performance degradation to react passively, but actively and progressively introduces a compensation amount into the strategy. For example, in each calibration, a small offset matching the predicted degradation curve is added to the instruction parameters used to calculate the pre-distortion channel attenuation value. In this way, during the actual aging process of the device, the system's output performance remains stable through continuous fine-tuning, achieving a smooth performance transition. This unit also records each warning and the actual observed performance changes to subsequently revise its prediction model and improve prediction accuracy.

[0040] Complete workflow example, such as Figure 2 As shown: The system is powered on, and the target frequency is set to F0 and the target power to P0.

[0041] Initial setup: The control module controls the frequency adjustment module to output F0. The digital twin optimization engine queries the internal mapping table based on F0 and P0 to generate and configure the initial predistortion channel parameters (A0, Φ0), while setting the dynamic reference voltage Vref0.

[0042] Monitoring and Simulation: The power amplifier module outputs a signal, and the calibration module measures the actual signal characteristics (phase difference Δφ, amplitude ratio ΔA). These data are then fed into the digital twin optimization engine.

[0043] Virtual Iteration: The engine injects the same error (Δφ, ΔA) into the digital twin model and then performs rapid virtual iterations to test multiple (e.g., 5) paths from (A0, Φ0) to the new state (A1, Φ1), as well as the corresponding frequency fine-tuning sequences.

[0044] Strategy Execution: The engine selects the optimal strategy, for example: "Step 1: Gradually change the predistortion channel parameters to (A1, Φ1) within 100μs; Step 2: Wait 50μs; Step 3: Fine-tune the frequency control signal in 4 steps, with each step at a 20μs interval, adjusting ΔF / 4 in each step." The control module strictly follows this sequence.

[0045] Lifetime compensation integration: Simultaneously, the lifetime prediction unit predicts that the power amplifier tube gain will decrease by 0.2dB after 24 hours. Therefore, in this and subsequent strategy calculations, the digital twin optimization engine will pre-increase the target value by 0.002dB (a very gradual amount) when calculating the target amplitude, thus beginning compensation before the actual performance degradation.

[0046] Cyclic Adaptation: The system continuously monitors and repeats the above process based on the error classification results. For example, if the classification is rapid load change, a dedicated strategy that emphasizes rapid response and allows for some overshoot may be triggered.

[0047] This invention constructs a complete predictive calibration system by building an intelligent optimization engine containing a digital twin model, a pre-distortion calibration channel that complements the nonlinear characteristics of the power amplifier, an error classifier with self-learning capabilities, and a forward-looking lifetime prediction and parameter mitigation unit. This system not only avoids transient risks through virtual iteration and suppresses nonlinear distortion using feedforward compensation before actual calibration, but also intelligently identifies error sources and performs forward-looking parameter mitigation for device aging. This comprehensively solves the key technical problems of response lag, insufficient stability, lack of adaptability and predictive management in traditional power source calibration, and ultimately realizes a highly stable, high-precision, and long-cycle reliable intelligent frequency-tunable power source.

[0048] Although embodiments of the present invention have been shown and described, these specific embodiments are merely explanations of the invention and are not intended to limit it. The specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. After reading this specification, those skilled in the art may make modifications, substitutions, and variations to the embodiments as needed without departing from the principles and spirit of the invention, but such modifications, substitutions, and variations are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A frequency adjustable power source with calibration function, characterized in that, include: The frequency adjustment module is used to generate radio frequency signals based on the frequency control signal; The power amplifier module is used to amplify and output radio frequency signals; A calibration module is used to acquire the characteristics of the output signal and calibrate the frequency adjustment module; The control module is used to generate the frequency control signal and adjust it according to calibration feedback; This also includes: predistortion and analog calibration channels, and a digital twin optimization engine; The predistortion and analog calibration channel is connected in parallel between the frequency adjustment module and the power amplification module, and includes an analog circuit network with adjustable transmission characteristics; the control module can dynamically configure the transmission parameters of the analog circuit network according to the instructions of the digital twin optimization engine, so that it generates a predetermined amplitude and phase predistortion for the input radio frequency signal. The digital twin optimization engine has a built-in digital twin model of the power source. The digital twin optimization engine is configured to: receive real-time monitoring data from the calibration module and the operating status parameters of the power amplification module, and drive the digital twin model to perform synchronous simulation; based on the simulation results, predict the transient overshoot or oscillation risks that may occur when performing closed-loop calibration directly under the current configuration, and generate a calibration strategy that includes the predistortion parameter pilot adjustment command for the predistortion and analog calibration channels and the progressive frequency control signal adjustment sequence.

2. The frequency tunable power source with calibration function according to claim 1, characterized in that, The transmission parameters of the analog circuit network of the pre-distortion and analog calibration channel are configured based on the measurement results of the nonlinear amplitude and phase characteristics of the power amplifier module at the current operating frequency and power, so as to pre-compensate for the nonlinear distortion introduced by the power amplifier module.

3. The frequency tunable power source with calibration function according to claim 1, characterized in that, The digital twin optimization engine is also configured to perform virtual calibration iterations: before the physical power source actually performs a calibration adjustment, multiple rapid simulation iterations are performed on the digital twin model according to the proposed calibration strategy to evaluate the stability speed and overshoot of different strategies, and the optimal one from multiple candidate strategies is selected and output to the physical system for execution.

4. The frequency tunable power source with calibration function according to claim 1, characterized in that, Also includes: Lifetime prediction and parameter release unit; The lifetime prediction and parameter mitigation unit is connected to the temperature and current monitoring points of key components in the power amplifier module, as well as the digital twin optimization engine; the lifetime prediction and parameter mitigation unit is configured as follows: Based on real-time stress data and historical cumulative operating time of key components, predict their performance degradation trend and remaining reliable life. When it is predicted that performance degradation may affect output metrics in a specific future period, an early warning is issued to the digital twin optimization engine. After receiving the warning, the digital twin optimization engine actively and gradually introduces compensation parameters that match the predicted performance degradation value in the subsequently generated optimization calibration strategy, so that the output performance of the power source remains stable during the actual aging process of the device and achieves a smooth transition of output performance.

5. The frequency tunable power source with calibration function according to claim 4, characterized in that, The lifetime prediction and parameter mitigation unit is also configured to record the correspondence between historical warnings and actual performance degradation, and to continuously revise its prediction model using this correspondence.

6. The frequency-adjustable power source with calibration function according to claim 1, characterized in that, When generating the progressive frequency control signal adjustment sequence, the digital twin optimization engine uses a non-uniform time step, employing a smaller step size and denser period in the early stage of adjustment, and a larger step size and sparser period in the later stage of adjustment.

7. The frequency-adjustable power source with calibration function according to claim 1, characterized in that, It also includes a dynamic benchmark generation unit; The dynamic reference generation unit receives instructions from the digital twin optimization engine and is able to generate a non-fixed value dynamic reference signal to replace the traditional fixed value reference source and provide it to the calibration module. The value of the dynamic reference signal is calculated in real time by the digital twin optimization engine based on the current target output frequency, power, and ambient temperature by querying a pre-stored optimal operating point mapping table.

8. The frequency-adjustable power source with calibration function according to claim 1, characterized in that, The calibration module includes a self-learning error classifier; The self-learning error classifier performs real-time feature extraction and pattern recognition on the monitored error signals, distinguishes whether the error mainly comes from thermal effects, power supply noise or load changes, and sends the classification results and confidence scores to the digital twin optimization engine. The digital twin optimization engine calls different twin model sub-modules or adjusts model parameters to generate more targeted calibration strategies for different types of error-dominant modes.

9. The frequency-adjustable power source with calibration function according to claim 1, characterized in that, The digital twin model is a model that includes multi-physics coupling of electrical, thermal, and stress fields. The temperature data of key nodes calculated by its simulation is used to correct the compensation coefficient of the temperature-sensitive element in the calibration module in real time.

10. A power source calibration method based on digital twins, applied to a frequency-tunable power source with calibration function as described in any one of claims 1-9, characterized in that, Including the following steps: Establish and maintain a digital twin model of the power source; Real-time acquisition of operating status data and calibration monitoring data of physical power sources drives the synchronous operation of digital twin models; When calibration is required, virtual calibration iterations are performed in the digital twin model to evaluate the transient and steady-state effects of different calibration strategies. Based on the virtual iteration results, an optimized physical calibration strategy is generated, which includes at least a pilot adjustment command for the predistortion channel and a progressive frequency adjustment sequence. Control the physical power source to perform calibration operations according to the optimized physical calibration strategy; During execution, compensation parameters that match the predicted performance degradation are dynamically introduced into the optimized physical calibration strategy based on lifetime prediction information.

Images