A microwave input and output matching drive circuit based on CMOS

Abstract

The present invention discloses a microwave input and output matching drive circuit based on CMOS, which relates to the field of artificial intelligence control circuits, including a signal input terminal, a frequency detection sensor, an AI control module, a multi-mode matching network module, a switch control circuit and a signal output terminal, wherein the input of the frequency detection sensor is connected to the signal input terminal, and the output of the frequency detection sensor includes two paths, one path is connected to the input of the AI ​​control module, and the other path is connected to the input of the multi-mode matching network module, and the output of the AI ​​control module is connected to the input of the switch element of the multi-mode matching network module through the switch control circuit, and the output of the multi-mode matching network module is connected to the signal output terminal. The present invention integrates the original AI model, and can automatically adjust the configuration of the matching network according to the current operating frequency and environmental conditions to achieve the best impedance matching, so that the matching network can adaptively optimize the performance at different frequency points without manual adjustment.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence control circuits, in particular to a microwave input and output matching drive circuit based on CMOS. Background Art

[0002] CMOS microwave input and output matching circuits are circuit designs that are designed to achieve impedance matching of signal inputs and outputs in radio frequency (RF) and microwave systems while providing appropriate drive signals to meet system performance requirements. In this circuit, matching networks play an important role in ensuring that signal transmission and reflection are optimized to improve performance and efficiency. The role of matching networks in this circuit is critical because they help minimize signal reflections, thereby improving the efficiency of signal transmission. The design of matching networks needs to consider factors such as operating frequency, characteristics of circuit components, and target impedance to ensure the effectiveness of impedance matching. In CMOS circuits, because signal frequencies may be as high as the microwave range, impedance matching is particularly important to reduce reflections and signal losses. Matching networks usually include components such as capacitors, inductors, and transformers, which are configured according to operating frequency and impedance requirements to maximize signal quality.

[0003] As for the shortcomings of the above-mentioned existing solutions: in complex systems such as multi-mode matching networks, when the frequency point changes, the matching network may not be able to be optimized in real time to adapt to the new frequency point, resulting in performance degradation, such as increased reflection loss or reduced transmission efficiency; and with the increase of frequency points and changes in matching configurations, manual maintenance of the matching network may become cumbersome and expensive. In a dynamic communication environment, the frequency points and signal conditions may change rapidly, and the manually configured matching network may not be able to adapt to these rapid changes, resulting in performance fluctuations. How to automatically adjust the configuration of the matching network to achieve optimal impedance matching is a problem that needs to be solved at present. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a microwave input and output matching drive circuit based on CMOS to solve the problems that the existing matching network has difficulty in switching frequency points and is not suitable for dynamic environments.

[0005] Among them, a CMOS-based microwave input and output matching drive circuit includes a signal input terminal, a frequency detection sensor, an AI control module, a multi-mode matching network module, a switch control circuit and a signal output terminal, wherein the input of the frequency detection sensor is connected to the signal input terminal, the output of the frequency detection sensor includes two paths, one path is connected to the input of the AI ​​control module, and the other path is connected to the input of the multi-mode matching network module, the output of the AI ​​control module is connected to the input of the switch element of the multi-mode matching network module through the switch control circuit, and the output of the multi-mode matching network module is connected to the signal output terminal;

[0006] Among them, the signal input end is used to receive a signal source, and the signal source enters the frequency detection sensor through the signal input end. The frequency detection sensor collects the operating frequency of the signal source and sends the operating frequency to the AI ​​control module. The AI ​​control module receives the operating frequency and makes an intelligent decision to select the matching network configuration. The multi-mode matching network module includes multiple matching network configurations for impedance matching. The switch control circuit is used to control the switching elements in the multi-mode matching network module so that the matching network configuration in the multi-mode matching network module performs switch selection according to the decision of the AI ​​control module. The signal output end is used to output the matched signal.

[0007] Furthermore, the multi-mode matching network module includes multiple matching circuits, which include capacitors C1, C2, C3, inductors L1, L2, transformer T1 and switch element K1, the input of the switch element K1 is connected to the signal input end of the matching circuit, the output of the switch element is connected to the input of the inductor L1, the output of the inductor L1 is connected to the input of the capacitor C2, the output of the capacitor C2 is connected to the input of the transformer T1, the output of the transformer T1 is connected to the input of the inductor L2, the output of the inductor L2 is connected to the input of the capacitor C3, the output of the capacitor C3 is connected to the signal output end of the matching circuit, the capacitor C1 is connected in parallel to both ends of the inductor L1, and the switch element K1 is controlled by the switch control circuit.

[0008] Furthermore, the coverage range of the matching network configuration is 2 GHz-6 GHz, and the frequency points supported by the matching network configuration include at least: 2.4 GHz, 3.5 GHz, 4.5 GHz, 5.2 GHz and 5.8 GHz.

[0009] Furthermore, the AI ​​control module performs model training through the DQN training algorithm, establishes a mapping between frequency points and the best matching network configuration, and predicts monitoring results through a reward function to optimize the target.

[0010] Furthermore, the DQN training algorithm updates the Q value by using Q-learning to select the best matching configuration at different frequency points, and the specific method is as follows:

[0011] ;

[0012] Among them, the Indicates status Take action of value, represents the learning rate, Indicates in status Take action After receiving the instant reward, Represents a discount factor, which is used to consider the importance of future rewards. Indicates the current operating frequency point and matching network configuration status. Represents the matching network configuration, the Indicates taking action After the next state, the Indicates that in the next state Best move for you.

[0013] Furthermore, the Q value is calculated as follows:

[0014] ;

[0015] Among them, the represents the target Q value, Represents the Q-value function of the target network.

[0016] Furthermore, the DQN uses mean square error as the loss function to update the neural network parameters. The calculation method is:

[0017] .

[0018] The beneficial effects of the present invention are:

[0019] (1) The present invention integrates an AI model that can automatically adjust the configuration of the matching network according to the current operating frequency and environmental conditions to achieve optimal impedance matching, so that the matching network can adaptively optimize performance at different frequency points without manual adjustment;

[0020] (2) The present invention uses the DQN algorithm to optimize the impedance matching configuration through large-scale data and deep learning algorithms. Find more complex and sophisticated matching configurations to minimize reflection loss and improve transmission efficiency and performance; in complex systems such as multi-mode matching networks, the AI ​​model can handle multiple frequency points and multiple configurations to better adapt to complexity and improve system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A system structure block diagram of a CMOS-based microwave input and output matching drive circuit provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited to the following.

[0023] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative work are within the scope of protection of the present invention. It should be noted that relational terms such as the terms "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0025] Moreover, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or apparatus that includes a list of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or apparatus. In the absence of more restrictions, an element defined by the phrase "comprises a ..." does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0026] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0027] like Figure 1 As shown, a CMOS-based microwave input and output matching driving circuit includes a signal input terminal, a frequency detection sensor, an AI control module, a multi-mode matching network module, a switch control circuit and a signal output terminal, wherein the input of the frequency detection sensor is connected to the signal input terminal, the output of the frequency detection sensor includes two paths, one path is connected to the input of the AI ​​control module, and the other path is connected to the input of the multi-mode matching network module, the output of the AI ​​control module is connected to the input of the switch element of the multi-mode matching network module through the switch control circuit, and the output of the multi-mode matching network module is connected to the signal output terminal;

[0028] Among them, the signal input end is used to receive a signal source, and the signal source enters the frequency detection sensor through the signal input end. The frequency detection sensor collects the operating frequency of the signal source and sends the operating frequency to the AI ​​control module. The AI ​​control module receives the operating frequency and makes an intelligent decision to select the matching network configuration. The multi-mode matching network module includes multiple matching network configurations for impedance matching. The switch control circuit is used to control the switching elements in the multi-mode matching network module so that the matching network configuration in the multi-mode matching network module performs switch selection according to the decision of the AI ​​control module. The signal output end is used to output the matched signal.

[0029] Specifically, a specific example of the elements used in the above embodiment is proposed. In the above embodiment, the switch element uses a switch circuit, such as an RF switch, a PIN diode or other adjustable elements to control the switch state of the matching network; the AI ​​control module is an integrated AI chip and a control system, which monitors the frequency and automatically selects a suitable switch configuration; in the signal input end, the RF signal source is connected to the input port and enters the multi-mode matching network, and the frequency detection sensor is installed between the input end and the multi-mode matching network to monitor the current operating frequency. The sensor transmits the frequency information to the AI ​​control system, and the AI ​​control system receives the frequency information and determines which matching network configuration should be used according to the pre-trained AI model. The AI ​​model includes the optimization of matching parameters and switch states at different frequencies. In the multi-mode matching network, for each operating frequency point, there is a matching network configuration, which includes adjustable capacitors, inductors, transformers and switches for impedance matching, and the AI ​​control system indicates the corresponding configuration. Specific parameters need to be designed and optimized according to frequency and application requirements; the switch control circuit controls the switch state in the matching network to switch different matching configurations, and the switch control circuit is controlled by the AI ​​control system; the output of the multi-mode matching network is connected to a radio frequency receiver or a downstream circuit.

[0030] Further, the multi-mode matching network module includes a plurality of matching circuits, the matching circuits include capacitors C1, C2, C3, inductors L1, L2, transformer T1 and switch element K1, the input of the switch element K1 is connected to the signal input end of the matching circuit, the output of the switch element is connected to the input of the inductor L1, the output of the inductor L1 is connected to the input of the capacitor C2, the output of the capacitor C2 is connected to the input of the transformer T1, the output of the transformer T1 is connected to the input of the inductor L2, the output of the inductor L2 is connected to the input of the capacitor C3, the output of the capacitor C3 is connected to the signal output end of the matching circuit, the capacitor C1 is connected in parallel to both ends of the inductor L1, and the switch element K1 is controlled by the switch control circuit. Specifically, the matching circuit is a specific circuit part used to achieve impedance matching, including various electronic components, such as capacitors, inductors, transformers, diodes, etc. Matching network configuration refers to the parameter settings and connection methods of each matching circuit element in the matching network. The goal of matching network configuration is to achieve impedance matching at different operating frequency points of the multi-mode matching network. The configuration includes the values ​​of capacitors and inductors, switch states, number of turns of the transformer, etc.

[0031] Furthermore, the coverage range of the matching network configuration is 2 GHz-6 GHz, and the frequency points supported by the matching network configuration include at least: 2.4 GHz, 3.5 GHz, 4.5 GHz, 5.2 GHz and 5.8 GHz.

[0032] Furthermore, the AI ​​control module performs model training through the DQN training algorithm, establishes a mapping between frequency points and the best matching network configuration, and predicts the monitoring results through the reward function to optimize the target. Specifically, in DQN, it is necessary to define the representation of the state. In this case, the state may include the current operating frequency point and the configuration state of the matching network. The state representation may be a vector, such as [freq, config], where freq is the current frequency point and config is the configuration information of the matching network. In addition, it is also necessary to define the action space, that is, the different actions or configurations that can be taken. In the case of the matching network, the action can be represented as different matching configurations, including capacitance, inductance, switch state, etc. The action space can be a discrete set, such as [action1, action2, ...], each action corresponds to a different matching configuration. Use DQN to train the matching network to automatically select the best matching configuration at different frequency points. Through repeated training and updating, the best impedance matching is calculated at different frequency points, thereby improving the performance of the multi-mode matching network.

[0033] Furthermore, the DQN training algorithm updates the Q value by using Q-learning to select the best matching configuration at different frequency points, and the specific method is as follows:

[0034] ;

[0035] Among them, the Indicates status Take action of value, represents the learning rate, Indicates in status Take action After receiving the instant reward, Represents a discount factor, which is used to consider the importance of future rewards. Indicates the current operating frequency point and matching network configuration status. Represents the matching network configuration, the Indicates taking action After the next state, the Indicates that in the next state Specifically, the DQN algorithm establishes a Q-value function that maps state-action pairs to expected cumulative rewards. By continuously learning the Q-value function, the algorithm can select the best action under different states to maximize the cumulative reward, which helps to automate the configuration selection of the matching network to optimize performance.

[0036] Furthermore, the Q value is calculated as follows:

[0037] ;

[0038] Among them, the represents the target Q value, Represents the Q-value function of the target network.

[0039] Furthermore, the DQN uses mean square error as the loss function to update the neural network parameters. The calculation method is:

[0040] .

[0041] Further, as a preferred specific implementation of this embodiment, a matching method for a microwave input and output matching driving circuit based on CMOS is proposed, which includes the following steps:

[0042] S1. The external RF signal source transmits the signal to the input end of the multi-mode matching network, and the sensor detects the operating frequency of the current input signal. The sensor transmits the frequency information to the AI ​​control system;

[0043] S2. The AI ​​control system receives the frequency information and uses the pre-trained AI model to make a decision on the matching network configuration to be used;

[0044] S3. The AI ​​control system controls the switch control circuit in the matching network to adjust the switch state according to the decision to switch to the appropriate matching configuration;

[0045] S4. The matched signal is transmitted from the output of the multimode matching network to the RF receiver or downstream circuit for subsequent processing or transmission.

[0046] Furthermore, the step S2 specifically includes the following sub-steps:

[0047] S201. Collect information on frequency, matching network configuration and performance indicators, and initialize the matching network state and AI model;

[0048] S202. Train the AI ​​model through the DQN reinforcement learning model to establish a mapping between frequency points and the best matching configuration;

[0049] S203. For each working frequency point, obtain the current frequency point, use the trained AI model, input the current frequency point, obtain matching configuration suggestions, switch the matching network configuration according to the AI ​​model's suggestions, and monitor the matching network's performance, such as S parameters and reflection loss; calculate rewards or penalties based on the performance monitoring results, and the reward value is defined according to the optimization target of the performance indicator; and update the model;

[0050] S204. When the target performance requirement is reached, the control loop ends.

[0051] Furthermore, in step S202, the AI ​​model is trained by the DQN reinforcement learning model as follows: the DQN training algorithm updates the Q value by using Q-learning and selects the best matching configuration at different frequency points, and the specific method is as follows:

[0052] ;

[0053] Among them, the Indicates status Take action of value, represents the learning rate, Indicates in status Take action After receiving the instant reward, Represents a discount factor, which is used to consider the importance of future rewards. Indicates the current operating frequency point and matching network configuration status. Represents the matching network configuration, the Indicates taking action After the next state, the Indicates that in the next state Best move for you.

[0054] Furthermore, the Q value is calculated as follows:

[0055] ;

[0056] Among them, the represents the target Q value, Represents the Q-value function of the target network.

[0057] Furthermore, the DQN uses mean square error as the loss function to update the neural network parameters. The calculation method is:

[0058] .

[0059] Furthermore, in step S203, rewards or penalties are calculated based on the performance monitoring results, and the reward value is defined according to the optimization target of the performance indicator as follows:

[0060] ;

[0061] Among them, the reward function can be defined based on performance indicators and frequency points. For example, minimizing reflection loss or maximizing transmission power as performance indicators, the reward can be defined based on different goals, such as:

[0062] Reward for minimizing reflection loss: ;

[0063] Rewards for maximizing transmission power: .

[0064] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art shall not deviate from the spirit and scope of the present invention, and shall be within the scope of protection of the claims attached to the present invention.

Claims

1. A CMOS-based microwave input and output matching drive circuit, characterized in that: It includes a signal input terminal, a frequency detection sensor, an AI control module, a multi-mode matching network module, a switch control circuit and a signal output terminal, wherein the input of the frequency detection sensor is connected to the signal input terminal, the output of the frequency detection sensor includes two paths, one path is connected to the input of the AI ​​control module, and the other path is connected to the input of the multi-mode matching network module, the output of the AI ​​control module is connected to the input of the switch element of the multi-mode matching network module through the switch control circuit, and the output of the multi-mode matching network module is connected to the signal output terminal; The signal input terminal is used to receive a signal source, and the signal source enters a frequency detection sensor through the signal input terminal. The frequency detection sensor collects the operating frequency of the signal source and sends the operating frequency to the AI ​​control module. The AI ​​control module receives the operating frequency, obtains the current frequency point, uses the trained AI model, inputs the current frequency point, obtains a matching configuration suggestion, switches the matching network configuration according to the suggestion of the AI ​​model, monitors the performance of the matching network, and calculates a reward or penalty based on the performance monitoring result; The multi-mode matching network module includes a plurality of matching network configurations for impedance matching. The switch control circuit is used to control the switch elements in the multi-mode matching network module so that the matching network configurations in the multi-mode matching network module perform switch selection according to the decision of the AI ​​control module. The signal output terminal is used to output the matched signal. The AI ​​control module performs model training through the DQN training algorithm, establishes a mapping between frequency points and the best matching network configuration, and predicts the monitoring results and optimizes the target through the reward function. The reward function is defined according to minimizing reflection loss or maximizing transmission power. The reward for minimizing reflection loss is: ; The maximum transmission power reward is: .

2. A CMOS-based microwave input and output matching driving circuit as claimed in claim 1, characterized in that: The multi-mode matching network module includes a plurality of matching circuits, which include capacitors C1, C2, C3, inductors L1, L2, transformer T1 and switch element K1. The input of the switch element K1 is connected to the signal input end of the matching circuit, the output of the switch element is connected to the input of the inductor L1, the output of the inductor L1 is connected to the input of the capacitor C2, the output of the capacitor C2 is connected to the input of the transformer T1, the output of the transformer T1 is connected to the input of the inductor L2, the output of the inductor L2 is connected to the input of the capacitor C3, the output of the capacitor C3 is connected to the signal output end of the matching circuit, the capacitor C1 is connected in parallel to both ends of the inductor L1, and the switch element K1 is controlled by a switch control circuit.

3. A CMOS-based microwave input and output matching driving circuit as claimed in claim 1, characterized in that: The coverage range of the matching network configuration is 2 GHz-6 GHz, and the frequency points supported by the matching network configuration include at least: 2.4 GHz, 3.5 GHz, 4.5 GHz, 5.2 GHz and 5.8 GHz.

4. A CMOS-based microwave input and output matching driving circuit as claimed in claim 1, characterized in that: The DQN training algorithm updates the Q value using Q-learning and selects the best matching configuration at different frequency points. The specific method is as follows: ; Among them, the Indicates status Take action of value, represents the learning rate, Indicates in status Take action After receiving the instant reward, Represents a discount factor, which is used to consider the importance of future rewards. Indicates the current operating frequency point and matching network configuration status. Represents the matching network configuration, the Indicates taking action After the next state, the Indicates that in the next state Best move for you.

5. A CMOS-based microwave input and output matching driving circuit as claimed in claim 4, characterized in that: The Q value is calculated as follows: ; Among them, the represents the target Q value, Represents the Q-value function of the target network.

6. A CMOS-based microwave input and output matching driving circuit as claimed in claim 5, characterized in that: The DQN uses mean square error as the loss function to update the neural network parameters. The calculation method is: .

Images