A method for harmonic tuning optimization of a microwave solid-state power amplifier with harmonic impedance

Abstract

The application discloses a kind of harmonic impedance microwave solid-state power amplifier harmonic tuning optimization methods, belong to microwave solid-state power amplifier technical field.The method includes: obtaining the input signal spectrum and output signal spectrum of microwave solid-state power amplifier, determines fundamental frequency component and each harmonic frequency component;According to fundamental frequency component and each harmonic frequency component, construct harmonic impedance state vector;Harmonic impedance state vector is input into pre-trained nonlinear dynamic impedance mapping model, and the target harmonic impedance tuning parameter set is obtained;According to target harmonic impedance tuning parameter set, adjustable impedance matching network at output end is adjusted, and harmonic tuning optimization is completed.The application constructs impedance vector by representing the joint state of multiple harmonics, and directly maps out tuning parameters using nonlinear dynamic model, which realizes efficient optimization of power amplifier harmonic characteristics.

Description

Technical Field

[0001] This invention relates to the field of microwave solid-state power amplifier technology, specifically to a method for optimizing harmonic tuning of microwave solid-state power amplifiers based on harmonic impedance. Background Technology

[0002] When microwave solid-state power amplifiers are operating, the nonlinear characteristics of transistors result in a large number of harmonic components in the output signal. These harmonic components not only reduce energy conversion efficiency but also cause electromagnetic interference to adjacent channels. To suppress harmonics and improve efficiency, a tunable impedance matching network needs to be constructed at the amplifier output. By adjusting the adjustable components in the network, the fundamental frequency and each harmonic can be matched to their respective optimal impedance states.

[0003] Existing harmonic impedance tuning methods mostly employ a successive approximation scanning strategy, that is, independently tuning each harmonic frequency sequentially while keeping the impedance state at other frequencies constant. This approach completely ignores the complex and close nonlinear coupling relationship between the fundamental impedance and harmonic impedances, as well as between different harmonic impedances. Since the impedance network exhibits a holistic response across all frequencies, adjusting any component will simultaneously affect the impedance of the fundamental and all harmonic frequencies. Independent tuning inevitably leads to a re-mismatch of the optimized frequencies, making the tuning process iterative, slow to converge, and difficult to find the globally optimal impedance combination under real-time changing conditions. Furthermore, existing tuning processes typically rely on simple feedback control or lookup table methods, lacking the ability to accurately model the mapping relationship between the current complex spectral state and the optimal tuning parameters.

[0004] Faced with impedance shifts caused by real-time changes in amplifier operating conditions such as temperature, supply voltage, and input power, existing solutions struggle to quickly and accurately adjust the output network to an ideal state that balances fundamental output power and harmonic suppression in a single operation. A key challenge is how to directly and accurately determine the optimal parameters of all adjustable components from the combined spectral state of the fundamental and harmonic frequencies, and how to achieve coordinated tuning across multiple frequency targets in a single operation. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a harmonic tuning optimization method for microwave solid-state power amplifiers with harmonic impedance. This method can start from the joint spectrum state containing the fundamental wave and harmonics, and directly determine the tuning parameters of all adjustable components at one time through a nonlinear mapping model that can dynamically adapt to different signal characteristics, thereby achieving rapid global collaborative optimization in complex coupled impedance networks.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a harmonic tuning optimization method for microwave solid-state power amplifiers with harmonic impedance. This method constructs a harmonic impedance state vector characterizing the impedance state at multiple frequency points and processes the state vector using a pre-trained nonlinear dynamic impedance mapping model. This enables the rapid and accurate acquisition of the target harmonic impedance tuning parameter set, thereby driving a tunable impedance matching network to complete harmonic tuning optimization and effectively improving the overall performance of the amplifier between high efficiency and low distortion.

[0007] In the technical solution of this invention, the input signal spectrum and output signal spectrum of a microwave solid-state power amplifier are obtained, and the fundamental frequency component and harmonic frequency components are determined accordingly. Preferably, peak frequency detection is performed on the input signal spectrum and the output signal spectrum to determine the fundamental and harmonic frequency distributions on the input side and the output side. The input fundamental frequency and the output fundamental frequency are checked for consistency. After the check passes, the output fundamental frequency is taken as a reliable fundamental frequency component. After determining the fundamental frequency component, based on its integer harmonic position, the frequency components that coexist at both the input and output harmonic frequency distributions are extracted as harmonic frequency components. This frequency component determination method based on dual verification and commonality verification significantly improves the accuracy of fundamental and harmonic frequency identification and avoids frequency misjudgment caused by measurement noise or interference.

[0008] After obtaining accurate fundamental and harmonic frequency components, the harmonic impedance state vector of the microwave solid-state power amplifier is constructed. As a preferred implementation, the first reflection coefficient at the amplifier output is measured at the fundamental frequency component, and the corresponding second reflection coefficient is measured at each harmonic frequency point. These reflection coefficients are arranged in ascending order of frequency to form an initial impedance sequence. This initial impedance sequence is normalized to obtain a normalized impedance sequence, which is then converted into a complex impedance vector to obtain the harmonic impedance state vector. Each element in this vector corresponds to a complex impedance value at a specific frequency point. This construction method integrates broadband impedance information into structured vector data, completely preserving the amplitude and phase information of the impedance, providing a high-quality input foundation for the accurate processing of subsequent models.

[0009] The aforementioned harmonic impedance state vector is input into a pre-trained nonlinear dynamic impedance mapping model to obtain the target harmonic impedance tuning parameter set. This model possesses dynamic nonlinear processing capabilities, containing multiple hidden layers, each employing a dynamic activation function. The parameters of these activation functions are dynamically adjusted based on the magnitude of the input harmonic impedance state vector. The harmonic impedance state vector undergoes layer-by-layer nonlinear transformations within the model's multiple hidden layers, and the model outputs an initial tuning parameter vector from the output layer. This dynamic adjustment mechanism enables the model to adaptively adjust its nonlinear mapping characteristics according to the strength of the current impedance state, significantly enhancing its ability to represent and generalize complex impedance relationships of power amplifiers under different operating conditions.

[0010] The initial tuning parameter vector output by the model needs to be mapped to the physical adjustable range of the tunable impedance matching network to ensure the engineering feasibility of the output parameters. As a preferred implementation, the upper and lower physical adjustment limits of each adjustable element in the tunable impedance matching network are obtained, and their respective physical adjustable ranges are determined accordingly. A one-to-one mapping relationship is established between each parameter in the initial tuning parameter vector and the physical adjustable range of each adjustable element, according to the parameter type. For each parameter, when its value is within the corresponding physical adjustable range, it is directly used as the target tuning parameter for that adjustable element; when the parameter value exceeds the range, it is truncated to the nearest upper or lower physical adjustment limit, and this truncated value is used as the target tuning parameter. Finally, the target tuning parameters of all adjustable elements are combined to form a target harmonic impedance tuning parameter set containing the fundamental tuning parameter and the tuning parameters of each harmonic. This process ensures that the theoretically optimal parameters output by the model can be converted into hardware-executable physical control quantities without errors, eliminating the risk of control failure or system mismatch caused by parameters going out of bounds.

[0011] Based on the obtained target harmonic impedance tuning parameter set, the tunable impedance matching network at the output of the microwave solid-state power amplifier is adjusted to complete harmonic tuning optimization. Specifically, the parameter set is analyzed to obtain the fundamental tuning control quantity and the tuning control quantities for each harmonic. The fundamental tuning control quantity is converted into a driving voltage or driving current signal to drive the fundamental tuning branch, adjusting the fundamental tuning branch to the target fundamental impedance state. Simultaneously, the tuning control quantities for each harmonic are converted into driving signals for their respective harmonic tuning branches, and each harmonic tuning branch is adjusted to its corresponding target harmonic impedance state. During the adjustment process, the actual impedance values ​​of the fundamental and each harmonic tuning branch are continuously monitored. When the monitored actual impedance value matches the target value in the parameter set, the current adjustment state of the tunable impedance matching network is locked. This closed-loop verification and state locking mechanism ensures the precise and long-term stability of the impedance tuning operation, enabling the amplifier to operate precisely in the preset optimal impedance environment, thereby achieving the expected harmonic control and energy efficiency optimization.

[0012] The nonlinear dynamic impedance mapping model relied upon in this invention is obtained through training with large-scale, multi-dimensional measured data. Preferably, the model training process includes collecting input and output spectrum samples of a microwave solid-state power amplifier under various operating conditions, including combinations of different input power levels, ambient temperatures, and supply voltages, as well as corresponding optimal harmonic impedance tuning parameter labels. Each training data point is generated through an automated scanning and filtering process: for each combination of operating conditions, each tunable element in the tunable impedance matching network is traversed and scanned according to a preset step sequence, and the current input signal spectrum, output signal spectrum, and current tuning parameter combination are recorded synchronously after each step. Harmonic distortion and power gain indices are calculated based on the current input and output signal spectra. These indices are compared with preset distortion and gain thresholds, respectively, to filter out candidate tuning parameter combinations that simultaneously meet both indices. From these candidate combinations, the set of parameters with the highest power gain index is further selected as the optimal harmonic impedance tuning parameter label under that operating condition. This label generation strategy, with gain maximization as the objective and distortion compliance as the constraint, ensures the global optimality of the training data. After obtaining all samples, the input and output spectrum samples are analyzed to extract the fundamental frequency and the frequencies of each harmonic sample. A sample harmonic impedance state vector is constructed as the training input, and the optimal tuning parameter label is used as the training objective. The model is iteratively trained using the backpropagation algorithm until the loss function converges. This training mechanism enables the model to accurately predict the optimal tuning scheme under different operating conditions.

[0013] As another preferred aspect of the invention, before formally inputting the currently acquired harmonic impedance state vector into the nonlinear dynamic impedance mapping model, real-time compensation correction for temperature and voltage factors is introduced. By real-time monitoring of the power amplifier's current operating temperature and supply voltage parameters, a pre-established temperature and voltage compensation table is consulted to obtain the corresponding temperature and voltage compensation coefficients. For each complex impedance value in the harmonic impedance state vector, its real and imaginary parts are extracted. The real part is added to the product of the temperature and voltage compensation coefficients to obtain the compensated real part; the imaginary part is added to the product of the temperature and voltage compensation coefficients to obtain the compensated imaginary part. The compensated real and imaginary parts are then recombine to generate the compensated complex impedance value, and all compensated impedance values ​​are arranged in their original order to obtain the compensated harmonic impedance state vector, which serves as the final input to the model. This compensation correction step effectively offsets the impedance measurement errors introduced by fluctuations in ambient temperature and supply voltage, ensuring that the state vector input to the model accurately reflects the true impedance characteristics of the power amplifier device under the current environment. This guarantees the consistency and reliability of the model's predictions in complex and ever-changing real-world application scenarios.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By extracting the fundamental frequency component and harmonic frequency components from the input and output spectra, a harmonic impedance state vector is constructed. This vector combines and normalizes the reflection coefficient measurements scattered across the fundamental and multiple harmonic frequencies into a complex impedance vector in frequency order. This vector fully encodes the complete impedance characteristics of the fundamental and harmonics in the current state of the amplifier, as well as their relative relationships. Instead of treating each harmonic as an isolated tuning object, it treats them as an inherently interconnected overall state input to subsequent processing stages. Decisions based on this overall state vector allow the subsequent mapping model to simultaneously perceive the complete impedance profile of both the fundamental and harmonics. This provides a unified input basis for solving the co-optimal solution of all frequency matching network parameters in one go, avoiding the iterative processes caused by neglecting coupling effects in traditional frequency-by-frequency scanning tuning.

[0015] By inputting the harmonic impedance state vector into a pre-trained nonlinear dynamic impedance mapping model, the hidden layers of which employ dynamic activation functions whose parameters are dynamically adjusted based on the magnitude of the input harmonic impedance state vector. The magnitude of the input vector reflects the overall strength of the current signal and the degree of impedance shift. The activation function thus alters its nonlinear transformation characteristics, enabling the model to adaptively transform to different impedance states. After multiple nonlinear transformations by this model, the target harmonic impedance tuning parameter set, including the fundamental tuning parameters and the tuning parameters of each harmonic, is directly obtained from the output layer. This direct mapping mechanism from the overall impedance state to a multi-parameter set bypasses a complex point-by-point optimization process, providing a single, coordinated adjustment of each adjustable element within its physically adjustable range. This allows for the rapid achievement of a global matching state that balances fundamental transmission efficiency and multiple harmonic suppression in a highly coupled tunable matching network of fundamental and harmonic impedances. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of a harmonic tuning optimization method for microwave solid-state power amplifiers; Figure 2 This is a flowchart of the process for constructing the harmonic impedance state vector of a microwave solid-state power amplifier; Figure 3 This is a flowchart of the harmonic impedance tuning optimization method; Figure 4 This is a flowchart of the training process for a nonlinear dynamic impedance mapping model. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1A method for harmonic tuning optimization of a microwave solid-state power amplifier includes: acquiring the input signal spectrum and output signal spectrum of the microwave solid-state power amplifier; determining the fundamental frequency component and each harmonic frequency component based on the input signal spectrum and output signal spectrum; constructing a harmonic impedance state vector of the microwave solid-state power amplifier based on the fundamental frequency component and each harmonic frequency component; inputting the harmonic impedance state vector into a pre-trained nonlinear dynamic impedance mapping model to obtain a target harmonic impedance tuning parameter set; and adjusting the tunable impedance matching network at the output of the microwave solid-state power amplifier according to the target harmonic impedance tuning parameter set to complete the harmonic tuning optimization.

[0020] Example 1: In specific implementation, please refer to Figure 2 When performing peak frequency detection on the input signal spectrum, the input signal is coupled from the input terminal of the microwave solid-state power amplifier. A spectrum analysis device is used to perform a Fourier transform on the input signal to obtain its spectrum. In the input signal spectrum, frequency points whose amplitude exceeds a preset amplitude threshold are marked as input peak frequencies by amplitude comparison. The frequency with the largest amplitude among the input peak frequencies is extracted as the input fundamental frequency, and the remaining input peak frequencies are used as the input harmonic frequency distribution. Similarly, when performing peak frequency detection on the output signal spectrum, the output signal is coupled from the output terminal of the microwave solid-state power amplifier. A spectrum analysis device is used to perform a Fourier transform on the output signal to obtain its spectrum. In the output signal spectrum, frequency points whose amplitude exceeds a preset amplitude threshold are marked as output peak frequencies by amplitude comparison. The frequency with the largest amplitude among the output peak frequencies is extracted as the output fundamental frequency, and the remaining output peak frequencies are used as the output harmonic frequency distribution.

[0021] Optionally, the preset amplitude threshold is set to -40dB of the largest amplitude in the input signal spectrum or the output signal spectrum. In specific implementations, when performing consistency verification between the input fundamental frequency and the output fundamental frequency, the absolute value of the frequency difference between the two frequencies is calculated, and compared with a preset frequency deviation tolerance. The preset frequency deviation tolerance is set to the larger of 1% of the input fundamental frequency and 1% of the output fundamental frequency. When the absolute value of the frequency difference is less than or equal to the preset frequency deviation tolerance, the verification is considered successful, and the output fundamental frequency is used as the fundamental frequency component; when the absolute value of the frequency difference is greater than the preset frequency deviation tolerance, the verification is considered unsuccessful, and the input signal spectrum and the output signal spectrum are reacquired.

[0022] In practical implementation, when extracting common frequency components from the input and output harmonic frequency distributions based on the integer harmonic positions of the fundamental frequency components, for a preset harmonic order P, the P-harmonic frequency of the fundamental frequency component is calculated to form a P-harmonic reference frequency. A frequency search window is set with the P-harmonic reference frequency as the center, and the half-width of the frequency search window is selected as 2% of the fundamental frequency component. Input harmonic frequency components falling within the P-harmonic reference frequency search window in the input harmonic frequency distribution are matched with output harmonic frequency components falling within the same P-harmonic reference frequency search window in the output harmonic frequency distribution. When both exist and the absolute value of the frequency difference is less than the half-width of the frequency search window, the corresponding frequency component in the output harmonic frequency distribution is taken as the harmonic frequency component of that order. This process is repeated for all preset harmonic orders P to obtain the frequency components of each harmonic.

[0023] In practical implementation, when measuring the first reflection coefficient at the output of the microwave solid-state power amplifier at the fundamental frequency component, one port of the vector network analyzer is connected to the output of the microwave solid-state power amplifier. The vector network analyzer performs a single-point frequency scan on the fundamental frequency component and reads the S11 parameter as the first reflection coefficient. When measuring the corresponding second reflection coefficient at each harmonic frequency component, for each harmonic frequency component, the vector network analyzer performs a single-point frequency scan on that harmonic frequency and reads the S11 parameter as the second reflection coefficient corresponding to that harmonic frequency. Before measuring the reflection coefficient, the vector network analyzer undergoes open-circuit, short-circuit, and load calibration.

[0024] It can be understood that when the first reflection coefficient and all second reflection coefficients are arranged in ascending order of frequency, the fundamental frequency component has the lowest frequency value, and the frequency values ​​of each harmonic frequency component increase sequentially. The first reflection coefficient corresponding to the fundamental frequency component is placed at the beginning of the sequence, and then the second reflection coefficients corresponding to each harmonic frequency component are arranged sequentially in ascending order of harmonic order, forming the initial impedance sequence. The initial impedance sequence is a complex sequence, where each element contains both a real part and an imaginary part.

[0025] In practical implementation, when normalizing the initial impedance sequence, the system characteristic impedance is used to convert the impedance of each reflection coefficient in the initial impedance sequence. The conversion relationship is as follows: in, This represents the normalized impedance value corresponding to the m-th frequency point in the initial impedance sequence. This represents the reflection coefficient corresponding to the m-th frequency point in the initial impedance sequence, with the system characteristic impedance set to 50 ohms. After normalization, a normalized impedance sequence is obtained, where each element is a dimensionless complex number.

[0026] In practical implementation, when converting the normalized impedance sequence into a complex impedance vector, the order of elements in the normalized impedance sequence is directly used as the order of vector elements to construct the harmonic impedance state vector. The dimension of the harmonic impedance state vector is equal to the total number of fundamental frequency components and all harmonic frequency components. Each element of the harmonic impedance state vector corresponds to a complex impedance value at a frequency point, which consists of a real part and an imaginary part.

[0027] Example 2: In practical implementation, when the harmonic impedance state vector is used as the input layer data of the nonlinear dynamic impedance mapping model, the harmonic impedance state vector is denoted as... , The dimension is , It equals the sum of the number of fundamental frequency components and the number of harmonic frequency components. The nonlinear dynamic impedance mapping model employs a multi-layer feedforward neural network structure, including an input layer, One hidden layer and one output layer The value is 3. The input layer contains... The nth input node, the nth Each input node receives the harmonic impedance state vector. The first in There are 10 elements, which are complex impedance values. During input, the real and imaginary parts of the complex impedance value are treated as two independent real inputs. The total number of input layer nodes is expanded to 100. Hidden layers are fully connected, and each hidden layer contains... One neuron, The value is 128. The output layer contains... One output node, It equals the total number of adjustable elements in the tunable impedance matching network, and each node in the output layer outputs a real parameter value.

[0028] In some embodiments, when performing layer-by-layer nonlinear transformation through multiple hidden layers of the nonlinear dynamic impedance mapping model, for the first... The first hidden layer For each neuron, calculate the weighted sum of its inputs: in, Indicates the first The number of neurons in the layer Indicates the connection of the first Layer The first neuron to the second Layer The weight coefficients of each neuron, Indicates the first Layer The output value of each neuron Indicates the first Layer The bias of each neuron. Hidden layer Output value of each neuron The expression for the dynamic activation function, calculated using the dynamic activation function, is: in, For the first The dynamic scaling factor of the hidden layer Represents the hyperbolic tangent function. Dynamic scaling factor. Based on the harmonic impedance state vector The modulus is dynamically adjusted. The calculation method is as follows: , Represents the harmonic impedance state vector of Norm, For the first The base scaling factor of the hidden layer, For the first The modulus sensitivity coefficient of the hidden layer. The range of values ​​is , The range of values ​​is , and All parameters were optimized using the backpropagation algorithm during the training of the nonlinear dynamic impedance mapping model. When the harmonic impedance state vector... As the modulus increases, the dynamic scaling factor... As the harmonic impedance state vector increases accordingly, the slope of the dynamic activation function also increases. When the modulus decreases, the dynamic scaling factor Consequently, as the activation value decreases, the slope of the dynamic activation function also decreases. After... The transformation of the hidden layer by layer, the th... The output value vector of the hidden layer is passed to the output layer.

[0029] In practical implementation, when obtaining the initial tuning parameter vector from the output layer of the nonlinear dynamic impedance mapping model, the output layer uses a linear activation function, and the output layer's first... Output value of each node The result is obtained through weighted summation: ,in Indicates the output layer number The node and the first Hidden layer Weight coefficients between neurons Indicates the first Hidden layer The output value of each neuron Indicates the output layer number The bias of each node. The output values ​​of all nodes in the output layer are combined to form the initial tuning parameter vector. , The dimension is , Each element in the array corresponds to an initial tuning parameter value for an adjustable element, which is a continuous real number that is not physically constrained.

[0030] It can be understood that when mapping each parameter in the initial tuning parameter vector to the corresponding physically adjustable range of the tunable impedance matching network, the upper and lower limits of the physical adjustment for each adjustable element in the tunable impedance matching network are obtained. The tunable impedance matching network consists of a fundamental tuning branch and harmonic tuning branches. The fundamental tuning branch includes a fundamental adjustable capacitor and a fundamental adjustable inductor, and each harmonic tuning branch includes harmonic adjustable capacitors and harmonic adjustable inductors for the corresponding order. Each adjustable capacitor and each adjustable inductor is considered a tunable element. For the ... For each adjustable element, the upper limit of its physical adjustment can be obtained by consulting its datasheet or factory calibration data. and physical adjustment lower limit And the upper limit is adjusted by physical adjustment. and physical adjustment lower limit Determine the first The physical adjustable range of each adjustable element Physical adjustment upper limit and physical adjustment lower limit The unit is consistent with the control parameter type of the adjustable element. When the adjustable element is a voltage-controlled varactor diode, the unit is volt; when the adjustable element is a current-controlled variable inductor, the unit is milliampere.

[0031] In practical implementation, the initial tuning parameter vector When establishing a one-to-one correspondence between each parameter in the tuning parameters and the physical adjustable range of the adjustable element, an initial tuning parameter vector is predefined. The order of the elements in the middle is consistent with the fixed arrangement order of the adjustable components. The fixed arrangement order of the adjustable components is: fundamental frequency adjustable capacitor, fundamental frequency adjustable inductor, second harmonic adjustable capacitor, second harmonic adjustable inductor, up to the highest harmonic adjustable inductor. For the initial tuning parameter vector... The first in Parameters , For real numbers, establish With the The physical adjustable range of each adjustable element The correspondence.

[0032] In some embodiments, for the initial tuning parameter vector Each parameter in, when In the range Inside, directly As the first Target tuning parameters of each adjustable element ,Right now .when Less than the lower limit of physical regulation At that time, Cut off to the lower limit of physical regulation ,Right now .when Greater than the physical adjustment limit At that time, Cut off to the physical adjustment limit ,Right now For the initial tuning parameter vector All of them Each parameter is truncated one by one as described above, resulting in... One target tuning parameter.

[0033] In practical implementation, when combining the target tuning parameters of all adjustable elements, the first to the second... The target tuning parameters are arranged in order to form a set of target harmonic impedance tuning parameters. , Target harmonic impedance tuning parameter set It includes fundamental tuning parameters for the fundamental frequency component and harmonic tuning parameters for each harmonic frequency component. The fundamental tuning parameters specifically correspond to the target tuning parameters of the fundamental adjustable capacitor and the fundamental adjustable inductor, and the harmonic tuning parameters specifically correspond to the target tuning parameters of each harmonic adjustable capacitor and the harmonic adjustable inductor.

[0034] Example 3: In specific implementation, please refer to Figure 3When analyzing the target harmonic impedance tuning parameter set, the target harmonic impedance tuning parameter set is denoted as... , Includes One target tuning parameter This equals the total number of adjustable elements in the tunable impedance matching network. Following a fixed arrangement of the adjustable elements, from... The target tuning parameters corresponding to the fundamental frequency adjustable capacitor, the fundamental frequency adjustable inductor, and the target tuning parameters corresponding to the adjustable capacitors and inductors of each harmonic are read sequentially. The target tuning parameters corresponding to the fundamental frequency adjustable capacitor and the fundamental frequency adjustable inductor are combined to form a fundamental frequency tuning control quantity. This fundamental frequency tuning control quantity is a binary tuple, where the first element is the fundamental frequency capacitor control value and the second element is the fundamental frequency inductor control value. Similarly, the target tuning parameters of the harmonic adjustable capacitor and the harmonic adjustable inductor corresponding to each harmonic are combined to form the harmonic tuning control quantity for each harmonic. Each harmonic tuning control quantity is a binary tuple, where the first element is the harmonic capacitor control value and the second element is the harmonic inductor control value.

[0035] It can be understood that when the fundamental tuning control value is converted into a driving voltage or driving current signal for the fundamental tuning branch in a tunable impedance matching network, the fundamental tuning branch includes a fundamental adjustable capacitor and a fundamental adjustable inductor. When the fundamental adjustable capacitor is implemented using a voltage-controlled varactor diode, the fundamental capacitor control value is converted into a DC bias voltage signal via a digital-to-analog converter (DAC). The reference voltage of the DAC is selected as the physical upper limit of the fundamental adjustable capacitor's adjustment, and the DAC's output resolution is no less than 12 bits. The DC bias voltage signal output by the DAC is connected to the control terminal of the fundamental adjustable capacitor to adjust its capacitance value. When the fundamental adjustable inductor is implemented using a current-controlled variable inductor, the fundamental inductor control value is converted into a driving current signal via a voltage-controlled current source circuit. The transconductance of the voltage-controlled current source circuit ensures that the output current at full scale equals the physical upper limit of the fundamental adjustable inductor's adjustment. The driving current signal is injected into the control winding of the fundamental adjustable inductor to adjust its inductance value. By simultaneously adjusting the DC bias voltage signal of the fundamental frequency adjustable capacitor and the drive current signal of the fundamental frequency adjustable inductor, the fundamental frequency tuning branch is adjusted to the target fundamental frequency impedance state, which is jointly determined by the fundamental frequency capacitor control value and the fundamental frequency inductor control value.

[0036] In some embodiments, when the harmonic tuning control quantity is converted into the driving signal of the corresponding harmonic tuning branch in the tunable impedance matching network, for the first harmonic tuning control quantity... Second harmonic For integers greater than or equal to 2, the first... The subharmonic tuning branch includes the first The adjustable capacitor for the second harmonic and the first harmonic The first harmonic adjustable inductor. Extracting the first harmonic tuning control quantity from each harmonic tuning control quantity. The control value of the second harmonic capacitor and the first harmonic capacitor control value The control value of the second harmonic inductor. When the first harmonic inductor... When the adjustable capacitor for the second harmonic is implemented using a voltage-controlled varactor diode, the second harmonic... The control value of the second harmonic capacitor is obtained through the corresponding first harmonic capacitor. Number of times analog-to-digital converter to the first The second DC bias voltage signal, the first The reference voltage for the multiple-order analog-to-digital converter is selected as the first... The physical adjustment upper limit of the subharmonic adjustable capacitor, the first The output resolution of the first-order analog-to-digital converter is no less than 12 bits. The secondary DC bias voltage signal is connected to the first The control terminal of the adjustable capacitor for the second harmonic. When the... When the second harmonic adjustable inductor is implemented using a current-controlled variable inductor, the second harmonic adjustable inductor will be... The control value of the second harmonic inductor is obtained through the corresponding first harmonic inductor control value. The secondary voltage-controlled current source circuit is converted to the first The next drive current signal, the first The transconductance of the secondary voltage-controlled current source circuit makes the second voltage control current source circuit... When the control value of the second harmonic inductor is at full scale, the output current is equal to the first harmonic inductor. The physical adjustment upper limit of the second harmonic adjustable inductor will be the first harmonic... The second drive current signal is injected into the first The control winding of the second harmonic adjustable inductor. By simultaneously adjusting the first harmonic... The DC bias voltage signal of the adjustable capacitor for the second harmonic and the first harmonic The drive current signal of the second harmonic adjustable inductor will be the first harmonic. The second harmonic tuning branch is adjusted to the corresponding first harmonic tuning branch. The target harmonic impedance state is then determined. The above adjustment process is performed on each harmonic tuning branch one by one, adjusting all harmonic tuning branches to their corresponding target harmonic impedance states.

[0037] In practical implementation, when monitoring the actual impedance values ​​of the fundamental frequency tuning branch and each harmonic tuning branch after adjustment, a port of a vector network analyzer is connected to the input port of the fundamental frequency tuning branch. The vector network analyzer performs a single-point frequency scan at the fundamental frequency component to measure the actual reflection coefficient of the fundamental frequency tuning branch. The actual reflection coefficient of the fundamental frequency tuning branch is then converted to impedance to obtain the actual impedance value of the fundamental frequency tuning branch. In the... The input port of the second harmonic tuning branch is connected to a port of the vector network analyzer. The vector network analyzer is in the... A single-point frequency scan was performed at the second harmonic frequency component to measure the first harmonic frequency component. The actual reflection coefficient of the second harmonic tuned branch will be the first harmonic tuning branch. The actual reflection coefficient of the second harmonic tuned branch is obtained after impedance conversion. The actual impedance value of the subharmonic tuned branch.

[0038] It is understandable that when determining whether the actual impedance value matches the target value in the target harmonic impedance tuning parameter set, the target impedance value of the fundamental tuning branch is calculated from the fundamental capacitor control value and the fundamental inductor control value using the impedance synthesis formula, which is: in, This represents the target impedance value of the fundamental frequency tuned branch. Represents the imaginary unit. This represents the angular frequency corresponding to the fundamental frequency component. , The frequency value of the fundamental frequency component. This represents the inductance value of the fundamental frequency adjustable inductor. There is a monotonic mapping relationship between the inductance value and the control value of the fundamental frequency adjustable inductor, as defined in the fundamental frequency adjustable inductor datasheet. This indicates the capacitance value of the fundamental frequency adjustable capacitor. A monotonic mapping relationship exists between the capacitance value and the control value of the fundamental frequency adjustable capacitor, as defined in the fundamental frequency adjustable capacitor's datasheet. This compares the actual impedance value of the fundamental frequency tuning branch with the target impedance value. Compare and calculate the actual impedance value with the target impedance value. The Euclidean distance between them is considered. When the Euclidean distance is less than the preset impedance deviation tolerance, the actual impedance value of the fundamental tuning branch is determined to be consistent with the target value, and the preset impedance deviation tolerance is set to 5% of the magnitude of the target impedance value. For the first... The second harmonic tuning branch is calculated. When determining the target impedance value of the subharmonic tuning branch, the angular frequency corresponding to the fundamental frequency component will be used. Replace with the first The angular frequency corresponding to the subharmonic frequency component , For the first The frequency value of the second harmonic component is used to replace the inductance value of the fundamental adjustable inductor with the value of the third harmonic component. The inductance value of the second harmonic adjustable inductor is used to replace the capacitance value of the fundamental adjustable capacitor. The capacitance value of the adjustable capacitor for the second harmonic is calculated using the same formula. The target impedance value of the second harmonic tuning branch, and the first harmonic tuning branch target impedance value, and the second ... The actual impedance value of the second harmonic tuning branch is related to the first harmonic tuning branch. The target impedance values ​​of the secondary harmonic tuning branches are compared, and the determination method is the same as that of the fundamental tuning branch. When the actual impedance value of the fundamental tuning branch is consistent with the target value, and the actual impedance values ​​of all harmonic tuning branches are consistent with their respective target values, the current adjustment state of the tunable impedance matching network is locked. The locking method is to write the output values ​​of all digital-to-analog converters into non-volatile memory and keep the digital-to-analog converters continuously outputting the current voltage signal. At the same time, the control terminal level of the voltage-controlled current source circuit is latched, and it no longer responds to subsequent changes in tuning parameters, thus completing the harmonic tuning optimization process.

[0039] Example 4: In specific implementation, please refer to Figure 4 The nonlinear dynamic impedance mapping model was pre-trained as follows: Input and output spectrum samples of the microwave solid-state power amplifier under different operating conditions and corresponding optimal harmonic impedance tuning parameter labels were collected. Different operating conditions consisted of combinations of three dimensions: input power level, ambient temperature, and supply voltage. The input power level was set at five points uniformly between the lower limit of the linear operating region and the 1dB compression point of the microwave solid-state power amplifier: 25%, 50%, 75% of the lower limit of the linear operating region, and the 1dB compression point value. The ambient temperature was controlled using a temperature test chamber, with set points of -40℃, -20℃, 0℃, 20℃, 40℃, 60℃, and 80℃. The supply voltage was set at 0.9 times, 1.0 times, and 1.1 times the rated supply voltage. All permutations of input power level, ambient temperature, and supply voltage resulted in a total of 5 × 7 × 3 = 105 combinations of operating conditions.

[0040] For each combination of operating conditions, the microwave solid-state power amplifier is placed under the corresponding stable operating conditions, and each adjustable element in the tunable impedance matching network is traversed and scanned according to a preset step sequence. The adjustable elements include a fundamental frequency adjustable capacitor and a fundamental frequency adjustable inductor, as well as adjustable capacitors and inductors for each harmonic. The total number of adjustable elements is denoted as […]. , The value is 2 plus twice the highest harmonic order. The preset step sequence is: for each adjustable element, its physical adjustable range is uniformly quantized to... A discrete value, The value is 8, and the quantization step size is the upper limit of physical adjustment minus the lower limit of physical adjustment, divided by 8. The discrete values ​​of all adjustable elements are combined to form... Each tuning parameter combination is output sequentially in lexicographical order. During each scan step, the controller loads the current tuning parameter combination onto each adjustable element of the tunable impedance matching network. After the RF circuit stabilizes, the current input signal spectrum, the current output signal spectrum, and the current tuning parameter combination are recorded.

[0041] Calculate the harmonic distortion and power gain indices under the current operating conditions based on the current input and output signal spectra. The harmonic distortion index uses the total harmonic distortion power ratio (THW / HW), calculated by extracting the fundamental power from the current output signal spectrum. and the second to highest harmonic order Subharmonic power , The value is 5, representing the total harmonic distortion power ratio. Defined as: in, This represents the power value corresponding to the fundamental frequency component in the output signal spectrum. Indicates the first element in the output signal spectrum. The power values ​​corresponding to the subharmonic frequency components. This is an index for harmonic orders. The power gain specification uses the fundamental power gain. The calculation method is as follows: extract the fundamental power from the spectrum of the current input signal. , .

[0042] Harmonic distortion index With preset distortion threshold Compare and compare the power gain metrics With preset gain threshold Compare them. Set to -30dBc, Set the gain to 70% of the nominal gain value specified in the microwave solid-state power amplifier datasheet. Filter for those that simultaneously meet the following conditions. and The current combination of tuning parameters is used as a candidate combination of tuning parameters.

[0043] Selecting power gain parameters from candidate tuning parameter combinations The highest combination of tuning parameters is used as the label for the optimal harmonic impedance tuning parameters under the corresponding operating conditions, and the input signal spectrum and output signal spectrum under these conditions are used as the corresponding input and output spectrum samples. If the set of candidate tuning parameter combinations is empty, the preset distortion threshold is relaxed. The value is increased in 1dB steps. The power gain index is selected from at least one candidate combination of tuning parameters until at least one such combination is found. The highest combination of tuning parameters is labeled as the optimal harmonic impedance tuning parameter.

[0044] Spectral analysis is performed on each input and output spectrum sample to extract the fundamental frequency and harmonic frequency samples. A peak frequency detection method is used to determine the input fundamental frequency and harmonic frequency distribution for the input signal spectrum, and the output fundamental frequency and harmonic frequency distribution for the output signal spectrum. After consistency verification, the fundamental frequency sample is obtained. Common frequency components are extracted from the input and output harmonic frequency distributions as the harmonic frequency samples. The first reflection coefficient at the output is measured at the fundamental frequency sample, and the corresponding second reflection coefficient is measured at each harmonic frequency sample. The first reflection coefficient and all second reflection coefficients are arranged from low to high frequency to form an initial impedance sequence. The initial impedance sequence is normalized and converted into a complex impedance vector form to obtain the sample harmonic impedance state vector. The dimension and construction rules of the sample harmonic impedance state vector are the same as those of the harmonic impedance state vector.

[0045] The sample harmonic impedance state vector is used as the training input data, and the corresponding optimal harmonic impedance tuning parameter label is used as the training output target. The nonlinear dynamic impedance mapping model employs a multi-layer feedforward neural network, including an input layer, One hidden layer and one output layer The value is 3. The number of input layer nodes is... , Let be the dimension of the sample harmonic impedance state vector. Each complex impedance value has a separate input node for its real and imaginary parts. Hidden layers are fully connected, and each hidden layer contains... One neuron, The value is set to 128. Each hidden layer uses a dynamic activation function, the parameters of which are dynamically adjusted based on the magnitude of the sample harmonic impedance state vector input to that hidden layer. The output layer contains... There are 1 output node, and the output layer uses a linear activation function.

[0046] The weights and biases of the initial nonlinear dynamic impedance mapping model were initialized using a Xavier uniform distribution. The initial nonlinear dynamic impedance mapping model was iteratively trained using the backpropagation algorithm. The Adam optimizer was used during training, with an initial learning rate of 0.001, an exponential decay rate of 0.9 for the first-order moment estimate, an exponential decay rate of 0.999 for the second-order moment estimate, and a numerical stability constant of [missing value]. The loss function uses the mean squared error function, and the batch size is set to 32. In each iteration, a batch of sample harmonic impedance state vectors and corresponding optimal harmonic impedance tuning parameter labels are randomly selected from the samples corresponding to all working condition combinations. The mean squared error between the predicted tuning parameter vector output by the model and the optimal harmonic impedance tuning parameter labels is calculated. The gradient of the loss function with respect to all weight coefficients and biases is calculated through backpropagation, and the weight coefficients and biases are updated using the Adam optimizer. When the relative change in the loss function value over 20 consecutive iterations is less than... When the loss function converges, training is stopped, and all weight coefficients, biases, and relevant parameters of the dynamic activation function of the trained nonlinear dynamic impedance mapping model are saved.

[0047] Example 5: In the specific implementation, before inputting the harmonic impedance state vector into the pre-trained nonlinear dynamic impedance mapping model, the current operating temperature and supply voltage parameters of the microwave solid-state power amplifier are monitored in real time. A platinum resistance temperature sensor mounted on the metal base of the microwave solid-state power amplifier is used to acquire the simulated temperature signal. The platinum resistance temperature sensor has a measurement accuracy of ±0.5℃ within the range of -40℃ to 80℃. The voltage signal output by the platinum resistance temperature sensor is converted into a digital value of the current operating temperature parameter through a 24-bit analog-to-digital converter (ADC), with the ADC sampling rate set to 10 times per second. Simultaneously, a high-precision resistor voltage divider network is used to attenuate the supply voltage of the microwave solid-state power amplifier at an attenuation ratio of 10:1. The attenuated voltage signal is then converted into a digital value of the current supply voltage parameter through a 24-bit ADC, with the ADC sampling rate set to 10 times per second. Both the current operating temperature and current supply voltage parameters are then subjected to sliding window averaging filtering, with a sliding window length of 10 sampling points, to filter out transient noise interference.

[0048] Based on the current operating temperature and supply voltage parameters, a pre-established temperature and voltage compensation table is consulted. This table comprises two independent sub-tables: a temperature compensation coefficient sub-table and a voltage compensation coefficient sub-table. The temperature compensation coefficient sub-table is indexed by Celsius temperature, with index temperature points spaced 5°C apart, covering a temperature range of -45°C to 85°C. Each index temperature point corresponds to a temperature compensation coefficient. The voltage compensation coefficient sub-table is indexed by supply voltage value, with index voltage points spaced 0.05 times the rated supply voltage, covering a voltage range of 0.85 times to 1.15 times the rated supply voltage. Each index voltage point corresponds to a voltage compensation coefficient. The temperature and voltage compensation coefficient sub-tables are established by connecting a known complex impedance value to the output of the microwave solid-state power amplifier under different combinations of temperature and voltage points. Standard impedance load, The real part is 50 ohms and the imaginary part is 0 ohms. The harmonic impedance state vector at the output of the microwave solid-state power amplifier is measured. The measured harmonic impedance state vector is compared with the theoretical harmonic impedance state vector of a standard impedance load. The complex impedance deviation value at each frequency point is calculated, and the average real part of the complex impedance deviation values ​​at all frequency points is denoted as . The average value of the imaginary part is denoted as ;Will and Based on the effects of temperature and voltage, a least-squares decomposition is performed, and the decomposition model is as follows: and ,in The real part is the amplitude factor. For the imaginary part amplitude factor, and These are constants obtained through least squares fitting; the temperature compensation coefficients for each temperature point are obtained through decomposition. and the voltage compensation coefficient corresponding to each voltage point Temperature compensation coefficient The value range is [-0.6, 0.6], voltage compensation coefficient. The value range is [-0.4, 0.4]. During the query, based on the current operating temperature parameter value, the temperature compensation coefficient is obtained from the temperature compensation coefficient sub-table through linear interpolation, denoted as . Based on the current power supply voltage parameters, the voltage compensation coefficient is obtained through linear interpolation in the voltage compensation coefficient sub-table, and denoted as . When the current operating temperature parameter exceeds the temperature range covered by the temperature compensation coefficient sub-table, the temperature compensation coefficient corresponding to the nearest boundary temperature point is used. When the current supply voltage parameter exceeds the voltage range covered by the voltage compensation coefficient sub-table, the voltage compensation coefficient corresponding to the nearest boundary voltage point is used.

[0049] Using temperature compensation coefficient and voltage compensation coefficient Compensation and correction are performed on each complex impedance value in the harmonic impedance state vector. For the first complex impedance value in the harmonic impedance state vector... Complex impedance values , For frequency point index, The value range is 1 to , Extracting complex impedance values ​​from the dimension of the harmonic impedance state vector. real part value and imaginary part values The real part value With temperature compensation coefficient and voltage compensation coefficient Add the products together to obtain the compensated real part value. Then add the imaginary part value. With temperature compensation coefficient and voltage compensation coefficient The products are added together to obtain the compensated imaginary part value. The calculation formula is: in, The first harmonic impedance state vector after compensation represents the first harmonic impedance state vector. The compensated complex impedance values ​​corresponding to each frequency point The first harmonic impedance state vector before compensation represents the first harmonic impedance state vector. Each frequency point corresponds to the real part of the complex impedance value. The first harmonic impedance state vector before compensation represents the first harmonic impedance state vector. Each frequency point corresponds to the imaginary part of the complex impedance value. This represents the temperature compensation coefficient obtained by interpolation from the temperature compensation coefficient sub-table based on the current operating temperature parameters. This represents the voltage compensation coefficient obtained by interpolation from the voltage compensation coefficient sub-table based on the current supply voltage parameters. This represents the imaginary unit. According to the compensation formula above, for all... After compensating each complex impedance value one by one, the generated The compensated complex impedance values ​​are combined in their original order to obtain the compensated harmonic impedance state vector. This compensated harmonic impedance state vector is then used as the input layer data for the nonlinear dynamic impedance mapping model.

[0050] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance, characterized in that, include: Obtain the input signal spectrum and output signal spectrum of a microwave solid-state power amplifier, and determine the fundamental frequency component and each harmonic frequency component based on the input signal spectrum and the output signal spectrum; Based on the fundamental frequency component and the harmonic frequency components, construct the harmonic impedance state vector of the microwave solid-state power amplifier; The harmonic impedance state vector is input into a pre-trained nonlinear dynamic impedance mapping model to obtain the target harmonic impedance tuning parameter set. Adjust the tunable impedance matching network at the output of the microwave solid-state power amplifier according to the target harmonic impedance tuning parameter set to complete the harmonic tuning optimization.

2. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 1, characterized in that, The process of acquiring the input signal spectrum and output signal spectrum of the microwave solid-state power amplifier, and determining the fundamental frequency component and each harmonic frequency component based on the input signal spectrum and the output signal spectrum, includes: Peak frequency detection is performed on the input signal spectrum to determine the input fundamental frequency and input harmonic frequency distribution; Peak frequency detection is performed on the spectrum of the output signal to determine the fundamental frequency and harmonic frequency distribution of the output; The input fundamental frequency and the output fundamental frequency are checked for consistency. If the check passes, the output fundamental frequency is used as the fundamental frequency component. Based on the integer multiples of the fundamental frequency components, common frequency components are extracted from the input harmonic frequency distribution and the output harmonic frequency distribution, and used as the harmonic frequency components of each order.

3. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 2, characterized in that, The step of constructing the harmonic impedance state vector of the microwave solid-state power amplifier based on the fundamental frequency component and the harmonic frequency components includes: The first reflection coefficient of the output terminal of the microwave solid-state power amplifier is measured at the fundamental frequency component, and the corresponding second reflection coefficient is measured at each harmonic frequency component. The first reflection coefficient and all the second reflection coefficients are arranged in order of frequency from low to high to form an initial impedance sequence; The initial impedance sequence is normalized to obtain a normalized impedance sequence. The normalized impedance sequence is converted into a complex impedance vector form to obtain the harmonic impedance state vector, where each element of the harmonic impedance state vector corresponds to a complex impedance value at a frequency point.

4. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 1, characterized in that, The step of inputting the harmonic impedance state vector into a pre-trained nonlinear dynamic impedance mapping model to obtain the target harmonic impedance tuning parameter set includes: The harmonic impedance state vector is used as the input layer data of the nonlinear dynamic impedance mapping model. The nonlinear dynamic impedance mapping model is transformed layer by layer through multiple hidden layers. Each hidden layer adopts a dynamic activation function, and the parameters of the dynamic activation function are dynamically adjusted according to the magnitude of the harmonic impedance state vector. After the layer-by-layer transformation of the multiple hidden layers, the initial tuning parameter vector is obtained from the output layer of the nonlinear dynamic impedance mapping model. Each parameter in the initial tuning parameter vector is mapped to the physical adjustable range of the corresponding tunable impedance matching network to obtain the target harmonic impedance tuning parameter set, which includes fundamental tuning parameters for the fundamental frequency component and harmonic tuning parameters for each harmonic frequency component.

5. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 4, characterized in that, The step of mapping each parameter in the initial tuning parameter vector to the physical tunable range of the corresponding tunable impedance matching network to obtain the target harmonic impedance tuning parameter set includes: Obtain the upper and lower limits of the physical adjustment for each adjustable element in the tunable impedance matching network, and determine the physical adjustable range of each adjustable element based on the upper and lower limits of the physical adjustment. Establish a one-to-one correspondence between each parameter in the initial tuning parameter vector and the physical adjustable range of the adjustable element according to the parameter type; For each parameter in the initial tuning parameter vector, when the parameter is within the corresponding physical adjustable range, the parameter is directly used as the target tuning parameter of the adjustable element. When the parameter exceeds the corresponding physical adjustable range, the parameter is truncated to the lower limit of the physical adjustment or the upper limit of the physical adjustment, and the truncated value is used as the target tuning parameter of the adjustable element. The target tuning parameters of all the adjustable elements are combined to obtain the target harmonic impedance tuning parameter set.

6. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 1, characterized in that, The step of adjusting the tunable impedance matching network at the output of the microwave solid-state power amplifier according to the target harmonic impedance tuning parameter set to complete harmonic tuning optimization includes: By analyzing the target harmonic impedance tuning parameter set, the fundamental tuning control quantity and the tuning control quantity of each harmonic are obtained. The fundamental tuning control quantity is converted into a driving voltage or driving current signal for the fundamental tuning branch in the tunable impedance matching network, and the fundamental tuning branch is adjusted to the target fundamental impedance state. The harmonic tuning control quantities are converted into driving signals for the corresponding harmonic tuning branches in the tunable impedance matching network, and the harmonic tuning branches are adjusted to the corresponding target harmonic impedance state. Monitor the actual impedance values ​​of the fundamental tuning branch and each harmonic tuning branch after adjustment. When the actual impedance value is consistent with the target value in the target harmonic impedance tuning parameter set, lock the current adjustment state of the tunable impedance matching network.

7. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 1, characterized in that, The nonlinear dynamic impedance mapping model is pre-trained through the following steps: The input and output spectrum samples of the microwave solid-state power amplifier under different operating conditions and the corresponding optimal harmonic impedance tuning parameter labels are collected. The different operating conditions include different input power levels, different ambient temperatures and different supply voltages. Spectral analysis is performed on each of the input and output spectrum samples to extract the fundamental frequency and the frequencies of each harmonic sample, and a sample harmonic impedance state vector is constructed based on the fundamental frequency and the frequencies of each harmonic sample. The sample harmonic impedance state vector is used as the training input data, and the corresponding optimal harmonic impedance tuning parameter label is used as the training output target. The backpropagation algorithm is used to iteratively train the initial nonlinear dynamic impedance mapping model until the loss function converges, thus obtaining the trained nonlinear dynamic impedance mapping model.

8. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 7, characterized in that, The acquisition of input and output spectrum samples of the microwave solid-state power amplifier under different operating conditions and the corresponding optimal harmonic impedance tuning parameter labels include: For each combination of operating conditions, each adjustable element in the tunable impedance matching network is traversed and scanned according to a preset step sequence, and the current input signal spectrum, the current output signal spectrum and the current tuning parameter combination are recorded after each scan step. Calculate the harmonic distortion index and power gain index under the current operating conditions based on the current input signal spectrum and the current output signal spectrum; The harmonic distortion index and the power gain index are compared with preset distortion thresholds and preset gain thresholds, respectively, and candidate tuning parameter combinations that satisfy the preset distortion thresholds and preset gain thresholds are selected. The tuning parameter combination with the highest power gain index is selected from the candidate tuning parameter combinations as the optimal harmonic impedance tuning parameter label under the operating condition, and the input signal spectrum and output signal spectrum under the operating condition are used as the corresponding input and output spectrum samples.

9. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 1, characterized in that, Before inputting the harmonic impedance state vector into the pre-trained nonlinear dynamic impedance mapping model, the method further includes: Real-time monitoring of the current operating temperature and current power supply voltage parameters of the microwave solid-state power amplifier; Based on the current operating temperature parameters and the current power supply voltage parameters, a pre-established temperature and voltage compensation table is consulted to obtain the temperature compensation coefficient and voltage compensation coefficient. The temperature compensation coefficient and the voltage compensation coefficient are used to compensate and correct each complex impedance value in the harmonic impedance state vector to obtain the compensated harmonic impedance state vector. The compensated harmonic impedance state vector is used as the input to the nonlinear dynamic impedance mapping model.

10. The method for optimizing harmonic tuning of a microwave solid-state power amplifier based on harmonic impedance according to claim 9, characterized in that, The step of compensating and correcting each complex impedance value in the harmonic impedance state vector using the temperature compensation coefficient and the voltage compensation coefficient to obtain the compensated harmonic impedance state vector includes: For each complex impedance value in the harmonic impedance state vector, extract the real part and imaginary part of the complex impedance value; The real part value is added to the product of the temperature compensation coefficient and the voltage compensation coefficient to obtain the compensated real part value. The imaginary part value is added to the product of the temperature compensation coefficient and the voltage compensation coefficient to obtain the compensated imaginary part value. The compensated complex impedance value is generated by recombining the real and imaginary parts of the compensated value, and all the compensated complex impedance values ​​are arranged in their original order to obtain the compensated harmonic impedance state vector.

Images