Broadband microwave circuit matching performance characterization method and microwave circuit design method
By plotting the bounding rectangle of the return loss frequency response curve on the Smith chart, and combining the weighted sum method to characterize the matching performance of microwave circuits, and by using reinforcement learning algorithms to optimize design parameters, the problem of universality of broadband microwave circuit matching performance characterization methods is solved, and design efficiency and accuracy are improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA ELECTRONIC TECH GRP CORP NO 38 RES INST
- Filing Date
- 2026-04-08
- Publication Date
- 2026-07-10
AI Technical Summary
Existing methods for characterizing the matching performance of broadband microwave circuits have poor universality and cannot effectively characterize the matching performance in the broadband frequency range, resulting in insufficient design efficiency and accuracy.
A Smith chart is used to plot the bounding rectangle of the return loss frequency response curve. The circuit matching performance is characterized by the distance from the center point of the bounding rectangle and the area. A weighted sum method is used for comprehensive evaluation, and the design parameters are optimized by combining reinforcement learning algorithm.
The physical meaning of the matching performance of broadband microwave circuits is clear, the applicability is strong, and the modeling is simple, which improves the design efficiency and accuracy and is suitable for microwave circuit optimization design based on machine learning.
Smart Images
Figure CN122088424B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer-aided microwave circuit design technology, specifically to a method for characterizing the matching performance of broadband microwave circuits for computer-aided optimization design and a microwave circuit design method. Background Technology
[0002] Computer-aided microwave circuit design can significantly improve design efficiency and accuracy. In recent years, with the rapid development and widespread application of artificial intelligence algorithms, computer-aided microwave circuit design has become more intelligent. A representative example is microwave circuit matching design based on machine learning. This method uses intelligent machine learning algorithms to guide the selection of microwave circuit design parameters based on the mapping relationship between the return loss frequency response curve (S11-frequency curve) obtained from electromagnetic simulation software and the matching performance of the microwave circuit, thereby significantly shortening the design cycle and reducing simulation calculation costs.
[0003] One of the challenges in implementing the aforementioned intelligent microwave matching design is how to characterize and model the return loss frequency response curve, so that the optimization algorithm can more accurately and autonomously determine the matching performance of the microwave circuit. For microwave circuits operating at a single frequency point, their matching performance can be directly determined by the value of the return loss S11. For example, a simulation of S11 ≤ -25dB (VSWR ≤ 1.12) indicates good matching performance. However, for broadband microwave circuits, characterizing the matching performance at every frequency point within a continuous broadband frequency range is very complex and difficult. Simply characterizing the circuit matching performance using the maximum, minimum, or average value of S11 across the entire frequency band is rather one-sided and will lose a lot of crucial information. Currently, a common method for characterizing the matching performance of broadband microwave circuits is to extract circuit performance from the S-parameter frequency response curve. This method uses the idea of frequency sampling and segmented weighting of S-parameter values, assigning different fitness values to S-parameters located in different value intervals. By summing the fitness values of all sampled frequency points, a fitness function characterizing the circuit performance across the entire frequency band can be obtained, which has significant engineering application value.
[0004] However, how to sample and determine the weighted interval and weighted value to avoid information loss and the impact of extreme points requires specific consideration and verification, and its universality is poor. Summary of the Invention
[0005] (a) Technical problems to be solved
[0006] To address the shortcomings of existing technologies, this invention provides a method for characterizing the matching performance of broadband microwave circuits and a method for designing microwave circuits, which solves the technical problem of poor universality of existing methods for characterizing the matching performance of broadband microwave circuits.
[0007] (II) Technical Solution
[0008] To achieve the above objectives, the present invention provides the following technical solution:
[0009] In a first aspect, the present invention provides a method for characterizing the matching performance of broadband microwave circuits, comprising:
[0010] Plot the return loss frequency response curve of the microwave circuit on the Smith chart.
[0011] Draw a bounding rectangle around the return loss frequency response curve on the Smith chart, where each side of the bounding rectangle intersects the return loss frequency response curve at least once, and ensure that all points of the return loss frequency response curve are located inside the bounding rectangle or on its four sides.
[0012] The matching performance of the microwave circuit is characterized by the distance L from the center point N inside the enclosing rectangle to the center point O of the Smith chart and the area W of the enclosing rectangle.
[0013] A dimensionless value obtained by dividing the distance L by the unit distance. A dimensionless value obtained by dividing the area W by the unit area. ,based on and The weighted sum of these values comprehensively characterizes the matching performance of microwave circuits.
[0014] Preferably, the enclosing rectangle must satisfy the condition that the shape of the enclosing rectangle does not change when the return loss frequency response curve rotates around the matching point on the Smith chart.
[0015] Preferably, the distance L represents the average matching performance within the operating frequency band of the microwave circuit. The smaller L is, the closer the return loss frequency response curve is to the matching point, and the better the average matching performance of the circuit.
[0016] The area W of the enclosing rectangle characterizes the broadband performance of the microwave circuit matching. The smaller the value of W, the smaller the difference in distance between different frequency points and the matching point on the return loss frequency response curve, and the better the broadband performance of the circuit matching.
[0017] Preferably, the method of plotting a bounding rectangle of the return loss frequency response curve on the Smith chart includes:
[0018] Find the point a on the return loss frequency response curve that is closest to the matching point O (0, 0), and the distance between points a and O is L1;
[0019] Draw a circle with point O as the center and L1 as the radius;
[0020] Draw a tangent line be to the circle at point a, where a is the point of tangency.
[0021] Find the point c on the return loss frequency response curve that is farthest from the straight line aO. Draw a perpendicular line from point c to the tangent line be of the circle, with the foot of the perpendicular at point b. The perpendicular line bc is the straight line containing one side of the circumscribed rectangle.
[0022] Find the point d on the return loss frequency response curve that is farthest from the straight line bc. Draw a perpendicular line from point d to the tangent line be of the circle, with the foot of the perpendicular at point e. The perpendicular line de is the straight line containing one side of the circumscribed rectangle.
[0023] Find the point f on the return loss frequency response curve that is farthest from the tangent line be. Draw a line parallel to the tangent line be through point f. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the lines bc and de is the two vertices m1 and m2 of the rectangle.
[0024] Find the point g on the return loss frequency response curve that is farthest from the parallel line m1m2 to the tangent line be. Draw a line parallel to the line m1m2 through point g. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the lines bc and de is the two vertices m4 and m3 of the rectangle.
[0025] The rectangle m1m2m3m4 is an enclosing rectangle in the Smith chart for the return loss frequency response curve that meets the conditions.
[0026] Preferably, the step of plotting the return loss frequency response curve of the microwave circuit on a Smith chart includes:
[0027] Place the Smith chart in the xoy rectangular coordinate system, with the origin of the coordinate system coinciding with the center point of the Smith chart. The range of the Smith chart is a unit circle with a radius of 1.
[0028] At this point, point O1 (-1, 0) represents the short-circuit point in the Smith chart, point O (0, 0) represents the matching point, and point O2 (1, 0) represents the open-circuit point;
[0029] According to physical principles, the return loss at each frequency point is a complex number. The return loss frequency response curve is plotted on the Smith chart with the real part of the complex number corresponding to the x-axis coordinate and the imaginary part corresponding to the y-axis coordinate.
[0030] Preferably, the one based on and The expression for the weighted sum is: because and The smaller the value, the better the circuit's matching performance; therefore, the weighted parameter... and The value is taken as a negative value; when it is necessary to convert the weighted sum to a positive value, a positive constant is added. The expression is changed to .
[0031] Secondly, the present invention provides a microwave circuit design method, comprising:
[0032] Based on the initial values of several design parameters of the microwave circuit, and following the broadband microwave circuit matching performance characterization method described above, the return loss frequency response curve and its enclosing rectangle are plotted on the Smith chart.
[0033] by and The weighted sum is used as the reward function for the reinforcement learning algorithm to autonomously find the optimal values of microwave circuit design parameters; where, , Weighted parameters, It is a positive constant.
[0034] Thirdly, the present invention provides a microwave circuit design system for performing the microwave circuit design method described above.
[0035] Fourthly, the present invention provides a computer-readable storage medium storing a computer program for microwave circuit design, wherein the computer program causes a computer to perform the microwave circuit design method as described above.
[0036] Fifthly, the present invention provides an electronic device, comprising:
[0037] One or more processors;
[0038] Memory; and
[0039] One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing microwave circuit design as described above.
[0040] (III) Beneficial Effects
[0041] This invention provides a method for characterizing the matching performance of broadband microwave circuits and a method for designing microwave circuits. Compared with the prior art, it has the following advantages:
[0042] This invention provides a method for characterizing the matching performance of broadband microwave circuits and a method for designing microwave circuits for computer-aided optimization. Compared with currently disclosed characterization methods, this method has advantages such as clear physical meaning, strong universality, simple modeling, and convenient application. This invention can be widely applied to microwave circuit optimization design scenarios based on artificial intelligence algorithms such as machine learning, meeting the application requirements of artificial intelligence optimization algorithms to autonomously judge the broadband matching performance of microwave circuits, and significantly improving the design efficiency and accuracy of broadband microwave matching circuits. Attached Figure Description
[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0044] Figure 1 This is a block diagram of a broadband microwave circuit matching performance characterization method in Example 1;
[0045] Figure 2 This is a specific application example of plotting the return loss (S11) frequency response curve on a Smith chart;
[0046] Figure 3 A specific application example of the method for plotting the bounding rectangle m1m2m3m4 of the return loss (S11) frequency response curve C0 on a Smith chart;
[0047] Figure 4 To complete the characterization of broadband microwave circuit matching performance according to the method in Example 1 Figure 2 The rectangle enclosed by the two return loss frequency response curves is used for plotting.
[0048] Figure 5 This is a specific application example in the design of microstrip line gold wire bonding matching circuits;
[0049] Figure 6 In order to be in Figure 5 The diagram illustrates the application effect of the reinforcement learning algorithm guided by the broadband microwave circuit matching performance characterization method provided in Example 1 to autonomously find the optimal values of design parameters aa, bb, cc, and dd in the design of gold wire bonding circuits (i.e., the microwave circuit design method in Example 2); where, Figure 6 Part (a) in the figure refers to the return loss frequency response curve and its enclosing rectangle in the Smith chart with initial values of aa, bb, cc, dd. Figure 6 Part (b) describes the convergence process of the reinforcement learning algorithm that autonomously finds the optimal values of parameters aa, bb, cc, and dd.
[0050] Figure 7 In order to be in Figure 5 A schematic diagram showing the design results of the gold wire bonding circuit design, in which the broadband microwave circuit matching performance characterization method provided in Example 1 guides the reinforcement learning algorithm to autonomously complete the selection of matching circuit parameters. Detailed Implementation
[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. 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.
[0052] This application provides a broadband microwave circuit matching performance characterization method and a microwave circuit design method, which solves the technical problem of poor universality of existing broadband microwave circuit matching performance characterization methods. It realizes the geometric expression of the physical meaning of circuit matching performance and can be widely used in the autonomous optimization design of microwave circuits based on artificial intelligence algorithms such as machine learning, which greatly improves the design efficiency and design accuracy of broadband microwave matching circuits.
[0053] The technical solution in this application is to solve the above-mentioned technical problems, and the general idea is as follows:
[0054] Existing research on modeling return loss frequency response curves to characterize the matching performance of broadband microwave circuits generally employs sampling and segmentation approaches, which suffer from shortcomings such as unclear physical meaning, complex modeling calculations, weak universality, and limited ease of application. To better meet the needs of microwave circuit optimization design based on machine learning intelligent algorithms, this invention presents a method for characterizing the matching performance of broadband microwave circuits based on the Smith chart. Since the characterization process is modeled using the Smith chart tool, it has clear physical meaning and strong universality. The characterization method utilizes the geometric characteristics of the frequency response curve on the Smith chart, making modeling simple, intuitive, and convenient to apply.
[0055] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.
[0056] Example 1:
[0057] This embodiment provides a method for characterizing the matching performance of broadband microwave circuits for computer-aided optimization design, the method comprising:
[0058] S1. Plot the return loss frequency response curve of the microwave circuit on the Smith chart.
[0059] S2. Draw a bounding rectangle for the return loss frequency response curve on the Smith chart. Each side of the bounding rectangle intersects the return loss frequency response curve at least once, and ensure that all points of the return loss frequency response curve are located inside the bounding rectangle or on its four sides.
[0060] S3. The matching performance of the microwave circuit is characterized by the distance L from the center point N inside the enclosing rectangle to the center point O of the Smith chart and the area W of the enclosing rectangle.
[0061] S4. A dimensionless numerical value obtained by dividing the distance L by the unit distance. A dimensionless value obtained by dividing the area W by the unit area. ,based on and The weighted sum of these values comprehensively characterizes the matching performance of microwave circuits.
[0062] The following is a further detailed description of each step:
[0063] In step S1, the return loss frequency response curve of the microwave circuit is plotted on a Smith chart. The specific implementation process is as follows:
[0064] The return loss (S11) frequency response curve, which characterizes the matching performance of microwave circuits, is plotted on a Smith chart. The Smith chart can be placed in an xoy rectangular coordinate system, with the origin coinciding with the center of the Smith chart. The Smith chart is a unit circle with a radius of 1. Point O1 (-1, 0) represents a short-circuit point on the Smith chart, point O (0, 0) represents a matching point, and point O2 (1, 0) represents an open-circuit point. According to physical principles, S11 at each frequency is a complex number, with its real part corresponding to the x-axis and its imaginary part corresponding to the y-axis. Thus, the S11 frequency response curve can be plotted on the Smith chart. The distance L between each point on the curve corresponding to a frequency and the origin O is the magnitude of the S11 value at that frequency, which is also the reflection coefficient. According to physical principles, the smaller this distance L, the smaller the reflection coefficient, and the better the circuit matching performance, thus achieving a geometric expression of the physical meaning of circuit matching performance. The Smith chart is a calculation graph plotted on the reflective plane, showing a family of circles representing the normalized input impedance (or admittance). It is a chart used in electrical and electronic engineering.
[0065] In step S2, a bounding rectangle is drawn on the Smith chart to enclose the return loss frequency response curve. Each side of the bounding rectangle (including the two endpoints) intersects the return loss frequency response curve at least once, and it is ensured that all points of the return loss frequency response curve lie inside the bounding rectangle or on its four sides. The specific implementation process is as follows:
[0066] To plot the return loss frequency response curve within a bounding rectangle on the Smith chart, the following condition must be met: the shape of the rectangle must remain unchanged when the return loss frequency response curve rotates around the matching point on the Smith chart (i.e., point O(0,0)). This is because, from a physical perspective, when the return loss frequency response curve rotates around point O(0,0), only the transmission phase changes, and the circuit matching performance remains unchanged.
[0067] The specific process of plotting a bounding rectangle for the return loss frequency response curve on a Smith chart is as follows:
[0068] (1) First, find the point a on the return loss frequency response curve that is closest to the matching point O (0, 0), and the distance between points a and O is L1;
[0069] (2) Draw a circle with point O as the center and L1 as the radius;
[0070] (3) Draw a tangent line be to the circle at point a, where a is the point of tangency;
[0071] (4) Find the point c on the return loss frequency response curve that is farthest from the straight line aO. Draw a perpendicular line from point c to the tangent line be of the circle, with the foot of the perpendicular at point b. The perpendicular line bc is the straight line containing one side of the circumscribed rectangle.
[0072] (5) Find the point d on the return loss frequency response curve that is farthest from the straight line bc. Draw a perpendicular line from point d to the tangent line be of the circle, with the foot of the perpendicular at point e. The perpendicular line de is the straight line containing one side of the circumscribed rectangle.
[0073] (6) Find the point f on the return loss frequency response curve that is farthest from the tangent be. Draw a line parallel to the tangent be through point f. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the line bc and the line de is the two vertices m1 and m2 of the rectangle.
[0074] (7) Find the point g on the return loss frequency response curve that is farthest from the parallel line m1m2 of the tangent be. Draw a line parallel to the line m1m2 through point g. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the line bc and the line de is the two vertices m4 and m3 of the rectangle.
[0075] (8) The rectangle m1m2m3m4 is a bounding rectangle of the return loss frequency response curve that meets the conditions in the Smith chart.
[0076] In step S3, the matching performance of the microwave circuit is characterized by the distance L from the center point N inside the enclosing rectangle to the center point O (0, 0) of the Smith chart (the matching point) and the area W of the enclosing rectangle. The specific implementation process is as follows:
[0077] The matching performance of the microwave circuit is characterized by the distance L from the center point N inside the enclosing rectangle to the center point O(0,0) (matching point) of the Smith chart in step S2, and the area W of the enclosing rectangle. The distance L represents the average matching performance within the operating frequency band of the microwave circuit. The smaller L is, the closer the return loss frequency response curve is to the matching point, and the better the average matching performance of the circuit. The area W represents the broadband matching performance of the microwave circuit. The smaller the value of W, the smaller the difference in distance between different frequency points on the return loss frequency response curve and the matching point, and the better the broadband matching performance of the circuit.
[0078] In step S4, a dimensionless value is obtained by dividing the distance L by the unit distance. A dimensionless value obtained by dividing the area W by the unit area. ,based on and weighted sum The matching performance of microwave circuits is comprehensively characterized, among which... , All parameters are weighted. The specific implementation process is as follows:
[0079] Based on the specific design requirements of the microwave circuit, the weighting parameters are adjusted. and A balance is struck between the average matching performance and broadband performance of the circuit matching to characterize the matching performance of broadband microwave circuits.
[0080] because and The smaller the value, the better the circuit's matching performance; therefore, the weighted parameter... and The value can be negative. If you need to add a weighted sum... Converting the value to a positive value can increase the positive constant. Weighted sum change to .
[0081] Example 2:
[0082] This invention provides a microwave circuit design method, the method comprising:
[0083] Step 1: Based on the initial values of several design parameters of the microwave circuit, and following the broadband microwave circuit matching performance characterization method described in Example 1, draw the return loss frequency response curve and its enclosing rectangle on the Smith chart; wherein, the setting of the design parameters depends on the specific design scenario of the circuit.
[0084] Step Two, with or As and The weighted sum is used as the reward function for the reinforcement learning algorithm to autonomously find the optimal values for each design parameter; where, , Weighted parameters, It is a positive constant.
[0085] It is understood that Embodiment 2 is based on Embodiment 1 in terms of setup and implementation, and its related technical settings can be referred to Embodiment 1.
[0086] The following specific application example further illustrates the above-mentioned broadband microwave circuit matching performance characterization method and microwave circuit design method for computer-aided optimization design:
[0087] Please see Figure 2 The return loss (S11) frequency response curve is plotted on a Smith chart. The Smith chart is an important graphical tool for microwave circuit design, and its basic structure consists of a series of equal-resistance and equal-reactance clusters. Point O1 on the far left of the Smith chart represents the short-circuit point, point O2 on the far right represents the open-circuit point, and point O in the center represents the matching point. In the figure, C1 and C2 are the return loss frequency response curves of two circuits plotted on the Smith chart. According to the physical meaning of the Smith chart, if the return loss value at a certain frequency is closer to the matching point O, it indicates better circuit matching performance. Intuitively, curve C1 is closer to the matching point O than curve C2. However, how to quantitatively compare the broadband matching performance represented by the return loss curves of the two circuits as a whole, and how to apply this to return loss frequency response curves of various shapes, is the main problem solved by this embodiment of the invention.
[0088] Please see Figure 3 Taking the return loss frequency response curve C0 of a certain circuit as an example, the bounding rectangle of C0 is drawn according to the broadband microwave circuit matching performance characterization method for computer-aided optimization design provided in this embodiment of the invention. The steps are as follows:
[0089] (1) Find the point a on the C0 curve that is closest to the matching point O(0,0), and the length of aO is L1. Draw a circle with O as the center and L1 as the radius. Draw a tangent line be through point a to the circle, with the point of tangency being point a.
[0090] (2) Find the point c on the return loss frequency response curve C0 that is farthest from the straight line aO. Draw a perpendicular line from point c to the tangent line be of the circle, with the foot of the perpendicular at point b. The perpendicular line bc is the straight line containing one side of the circumscribed rectangle.
[0091] (3) Find the point d on the return loss frequency response curve C0 that is farthest from the straight line bc. Draw a perpendicular line from point d to the tangent line be of the circle, with the foot of the perpendicular at point e. The perpendicular line de is the straight line containing one side of the circumscribed rectangle.
[0092] (4) Find the point f on the return loss frequency response curve C0 that is farthest from the tangent be. It is found that point f coincides with point c. Draw a line parallel to the tangent be through point c. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the line bc and the line de is the two vertices m1 and m2 of the rectangle. It is found that m1 coincides with point c.
[0093] (5) Find the point g on the return loss frequency response curve C0 that is farthest from the parallel line m1m2 of the tangent be. It is found that point g coincides with point a. The parallel line of the line m1m2 passing through point a is also the line be. The intersection of the line be with the line bc and the line de is the two vertices m4 and m3 of the rectangle. At this time, m4 coincides with point b and m3 coincides with point e.
[0094] (6) Rectangle m1m2m3m4 is an outer rectangle of the return loss frequency response curve in the Smith chart that meets the following conditions: all points of the return loss curve C0 are located inside the bounding rectangle, and each side of the rectangle (including the two endpoints) intersects the return loss frequency response curve at least once; when the return loss frequency response curve C0 rotates around the matching point (0,0) of the Smith chart, the shape of the bounding rectangle does not change; starting from any point on the return loss frequency response curve C0, along the four sides of the rectangle or in a direction parallel to the four sides of the rectangle, the point a on the curve that is closest to the matching point (0,0) can be reached in the fastest way from two orthogonal directions.
[0095] Please see Figure 4 According to an embodiment of the present invention, a method for characterizing the matching performance of a broadband microwave circuit for computer-aided optimization design is provided, and a plot is drawn. Figure 2 The return loss frequency response curves C1 and C2 of the two circuits are enclosing rectangles. The distances from the center point of the two enclosing rectangles to the matching point are L1 and L2, respectively, and the areas of the two enclosing rectangles are W1 and W2, respectively. According to the method provided in the embodiment of the present invention, the matching performance of curve C1 can be represented by the weighted sum of L1 and W1; the matching performance of curve C2 can be represented by the weighted sum of L2 and W2.
[0096] Please see Figure 5 In the figure, 50-ohm microstrip lines M1 and M2 are connected by a bonding wire BW, and the gold wire structure causes circuit mismatch. High-resistance lines HRL1 and HRL2 are introduced at the bonding points at both ends of the bonding wire BW in the circuit to match pads Pad1 and Pad2, thereby improving circuit matching performance. Optimal matching performance is obtained by adjusting the dimensional parameters aa, bb, cc, and dd of the matching structure. The broadband microwave circuit matching performance characterization method for computer-aided optimization design provided in this embodiment can be used to guide reinforcement learning algorithms to autonomously select design parameters for the gold wire bonding matching circuit.
[0097] Please see Figure 6 In part (a) of the diagram, with initial values of aa, bb, cc, and dd, according to the broadband microwave circuit matching performance characterization method for computer-aided optimization design provided in this embodiment of the invention, the return loss frequency response curve and its enclosing rectangle are plotted on a Smith chart; in the diagram, L is the distance from the center point of the rectangle enclosed by curve C to the matching point, W is the area of the rectangle enclosed by curve C, and a dimensionless value is obtained by dividing the distance L by the unit distance. A dimensionless value obtained by dividing the area W by the unit area. .
[0098] Please see Figure 6 Part (b) of the text, with R=(1- )+3×(1- As and The weighted sum (i.e. , , The weighted sum is used as the reward function for the Q-learning algorithm to guide the algorithm to autonomously find the optimal values of design parameters a, b, c, and d. The return loss frequency response curves and their circumscribed rectangles corresponding to all process states during the algorithm's convergence process are plotted on the same Smith chart. It can be seen that as the algorithm converges, the R value gradually decreases, the return loss frequency response curve gradually approaches the matching point, and the broadband matching performance of the gold wire bonded matching circuit gradually improves. With initial values of aa, bb, cc, and dd, the reward function value R0 = 3.12. Through the autonomous optimization of the values of aa, bb, cc, and dd by the Q-learning algorithm, the reward function value can be increased to R = 3.95, achieving excellent broadband matching performance.
[0099] Please see Figure 7 In the Cartesian coordinate system, curve 1 represents the frequency response curve of the return loss of the circuit without the introduction of the gold wire bonding matching circuit; curve 2 represents the frequency response curve of the return loss of the circuit after the Q-learning algorithm autonomously completes the selection of the design parameters of the gold wire bonding matching circuit using the broadband microwave circuit matching performance characterization method for computer-aided optimization design provided by this invention; the Q-learning algorithm reward function corresponding to curve 2 has a value of R=3.95, and the return loss S11 is less than -30dB in the frequency range of 6-18GHz, which shows very good broadband matching performance.
[0100] Example 3:
[0101] This invention also provides a microwave circuit design system for performing the microwave circuit design method described above.
[0102] Example 4:
[0103] This invention also provides a computer-readable storage medium storing a computer program for microwave circuit design, wherein the computer program causes a computer to perform the microwave circuit design method as described above.
[0104] Example 5:
[0105] This invention also provides an electronic device, including: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing microwave circuit design as described above.
[0106] In summary, compared with existing technologies, it has the following beneficial effects:
[0107] 1. The present invention provides a method for characterizing the matching performance of broadband microwave circuits and a method for designing microwave circuits for computer-aided optimization design. Compared with the currently disclosed characterization methods, it has advantages such as clear physical meaning, strong universality, simple modeling, and convenient application.
[0108] 2. The embodiments of the present invention can be widely applied to microwave circuit optimization design scenarios based on artificial intelligence algorithms such as machine learning, meeting the application requirements of artificial intelligence optimization algorithms to autonomously judge the broadband matching performance of microwave circuits, and greatly improving the design efficiency and design accuracy of broadband microwave matching circuits.
[0109] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for characterizing the matching performance of broadband microwave circuits, characterized in that, include: Plot the return loss frequency response curve of the microwave circuit on the Smith chart. Draw a bounding rectangle around the return loss frequency response curve on the Smith chart. Each side of the bounding rectangle intersects the return loss frequency response curve at least once, and ensures that all points of the return loss frequency response curve are located inside the bounding rectangle or on its four sides. The bounding rectangle must satisfy the condition that its shape does not change when the return loss frequency response curve is rotated around the matching point on the Smith chart. The matching performance of the microwave circuit is characterized by the distance L from the center point N inside the enclosing rectangle to the center point O of the Smith chart and the area W of the enclosing rectangle. The distance L represents the average matching performance within the operating frequency band of the microwave circuit. The smaller L is, the closer the return loss frequency response curve is to the matching point, and the better the average matching performance of the circuit. The area W of the enclosing rectangle represents the matching broadband performance of the microwave circuit. The smaller the value of W is, the smaller the difference in distance between different frequency points on the return loss frequency response curve and the matching point, and the better the broadband matching performance of the circuit. A dimensionless value obtained by dividing the distance L by the unit distance. A dimensionless value obtained by dividing the area W by the unit area. ,based on and The weighted sum of these values comprehensively characterizes the matching performance of microwave circuits.
2. The broadband microwave circuit matching performance characterization method as described in claim 1, characterized in that, The method of plotting a bounding rectangle for the return loss frequency response curve on a Smith chart includes: Find the point a on the return loss frequency response curve that is closest to the matching point O (0, 0), and the distance between points a and O is L1; Draw a circle with point O as the center and L1 as the radius; Draw a tangent line be to the circle at point a, where a is the point of tangency. Find the point c on the return loss frequency response curve that is farthest from the straight line aO. Draw a perpendicular line from point c to the tangent line be of the circle, with the foot of the perpendicular at point b. The perpendicular line bc is the straight line containing one side of the circumscribed rectangle. Find the point d on the return loss frequency response curve that is farthest from the straight line bc. Draw a perpendicular line from point d to the tangent line be of the circle, with the foot of the perpendicular at point e. The perpendicular line de is the straight line containing one side of the circumscribed rectangle. Find the point f on the return loss frequency response curve that is farthest from the tangent line be. Draw a line parallel to the tangent line be through point f. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the lines bc and de is the two vertices m1 and m2 of the rectangle. Find the point g on the return loss frequency response curve that is farthest from the parallel line m1m2 to the tangent line be. Draw a line parallel to the line m1m2 through point g. The parallel line is the line containing one side of the circumscribed rectangle. The intersection of the parallel line with the lines bc and de is the two vertices m4 and m3 of the rectangle. The rectangle m1m2m3m4 is an enclosing rectangle in the Smith chart for the return loss frequency response curve that meets the conditions.
3. The broadband microwave circuit matching performance characterization method as described in claim 1, characterized in that, The step of plotting the return loss frequency response curve of the microwave circuit on a Smith chart includes: Place the Smith chart in the xoy rectangular coordinate system, with the origin of the coordinate system coinciding with the center point of the Smith chart. The range of the Smith chart is a unit circle with a radius of 1. At this point, point O1 (-1, 0) represents the short-circuit point in the Smith chart, point O (0, 0) represents the matching point, and point O2 (1, 0) represents the open-circuit point; According to physical principles, the return loss at each frequency point is a complex number. The return loss frequency response curve is plotted on the Smith chart with the real part of the complex number corresponding to the x-axis coordinate and the imaginary part corresponding to the y-axis coordinate.
4. The broadband microwave circuit matching performance characterization method as described in claim 1, characterized in that, The basis and The expression for the weighted sum is: ,because and The smaller the value, the better the circuit's matching performance; therefore, the weighted parameter... and The value is taken as a negative value; when it is necessary to convert the weighted sum to a positive value, a positive constant is added. The expression is changed to .
5. A microwave circuit design method, characterized in that, include: Based on the initial values of several design parameters of the microwave circuit, and in accordance with the broadband microwave circuit matching performance characterization method as described in any one of claims 1 to 4, the return loss frequency response curve and its enclosing rectangle are plotted on the Smith chart. by or As and The weighted sum is used as the reward function for the reinforcement learning algorithm to autonomously find the optimal values of microwave circuit design parameters; where, , Weighted parameters, It is a positive constant.
6. A microwave circuit design system, characterized in that, The microwave circuit design system is used to perform the microwave circuit design method as described in claim 5.
7. A computer-readable storage medium, characterized in that, It stores a computer program for microwave circuit design, wherein the computer program causes the computer to execute the microwave circuit design method as described in claim 5.
8. An electronic device, characterized in that, include: One or more processors; Memory; as well as One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including methods for performing the microwave circuit design method as described in claim 5.