Radio frequency spiral inductor scalable model construction method and device based on neural network

Abstract

The invention relates to a radio frequency spiral inductor scalable model construction method and device based on a neural network, and the method comprises the steps: obtaining the physical size parameters of different spiral inductors; establishing a pi-type equivalent circuit model for each spiral inductor, and obtaining parameters of each element in the pi-type equivalent circuit model; taking the physical size parameter of each spiral inductor and the parameter of each element in the corresponding pi-type equivalent circuit model as a group of data to form a data set, and dividing the data set into a training set and a test set; constructing a neural network model, and training the neural network model by adopting the training set to obtain a relational expression between the physical size parameters of the spiral inductor and the parameters of each element in the pi-type equivalent circuit model; and testing the obtained relational expression by adopting a test set, and substituting the obtained relational expression into the model file after the test is passed, so as to obtain the radio frequency spiral inductor scalable model. According to the method, the parameter acquisition difficulty of the lumped element can be reduced, and the device simulation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of integrated circuit technology, and in particular to a method and device for constructing a scalable model of a radio frequency spiral inductor based on a neural network. Background Art

[0002] Planar spiral inductors are one of the most critical passive components in radio frequency integrated circuits. They play a variety of important roles in various radio frequency unit circuits, such as impedance matching, filtering, and forming oscillation networks. Their performance directly affects the performance of the entire circuit. Establishing an electromagnetic model of a device through electromagnetic simulation technology can accurately reflect the performance of the inductor. However, before modeling, it is necessary to have an in-depth understanding of the device's process parameters and geometric structure, and the model does not have continuity. In addition, electromagnetic models often require a lot of simulation time, and every time the layout size changes, the simulation needs to be re-performed, which is a very cumbersome and time-consuming process.

[0003] Establishing a device equivalent circuit model to obtain initial values ​​and then performing device scalability processing can effectively save simulation time and improve simulation efficiency. The existing technology for implementing device scalable models is mostly achieved by establishing scalable relationships and fitting polynomial coefficients. However, due to the complex structure and large number of physical parameters of planar spiral inductor devices, it is very complicated to establish equations that relate lumped components to physical dimensions, which is difficult to implement. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and device for constructing a scalable model of a radio frequency spiral inductor based on a neural network, which can reduce the difficulty of obtaining parameters of lumped elements.

[0005] The technical solution adopted by the present invention to solve the technical problem is: to provide a method for constructing a scalable model of a radio frequency spiral inductor based on a neural network, comprising the following steps:

[0006] Obtain the physical size parameters of different spiral inductors;

[0007] Establishing a π-type equivalent circuit model for each spiral inductor, and obtaining the parameters of each component in the π-type equivalent circuit model;

[0008] The physical size parameters of each spiral inductor and the parameters of each component in the corresponding π-type equivalent circuit model are used as a set of data to form a data set, and the data set is divided into a training set and a test set;

[0009] Constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model, and training the neural network model using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model;

[0010] The obtained relationship is tested using a test set. If the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the radio frequency spiral inductor; if the test fails, the neural network model continues to be trained.

[0011] The physical size parameters of the spiral inductor include: line width, spacing, number of turns and inner diameter.

[0012] The elements of the π-type equivalent circuit model include: π-type network inductance, π-type network series parasitic resistance, RL network device, substrate parasitic capacitance, substrate parasitic resistance, microstrip connecting line parasitic inductance, and overlapping and edge parasitic capacitance between coils.

[0013] The neural network model is an artificial neural network model.

[0014] The number of the neural network models is the same as and corresponds to the number of elements in the π-type equivalent circuit model. The number of hidden layers, the number of neurons and the number of iterations in each neural network model are adjusted according to the parameters of the elements in the corresponding π-type equivalent circuit model.

[0015] The technical solution adopted by the present invention to solve the technical problem is: to provide a device for building a scalable model of a radio frequency spiral inductor based on a neural network, comprising:

[0016] A first acquisition module is used to acquire physical size parameters of different spiral inductors;

[0017] A second acquisition module is used to establish a π-type equivalent circuit model for each spiral inductor and obtain parameters of each element in the π-type equivalent circuit model;

[0018] A data set establishment module is used to take the physical size parameters of each spiral inductor and the parameters of each element in the corresponding π-type equivalent circuit model as a set of data to form a data set, and divide the data set into a training set and a test set;

[0019] Constructing a training module for constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model, and the neural network model is trained using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model;

[0020] The test module is used to test the obtained relationship using a test set; if the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the radio frequency spiral inductor; if the test fails, the neural network model continues to be trained.

[0021] The technical solution adopted by the present invention to solve its technical problem is: to provide an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for constructing a scalable model of a RF spiral inductor based on a neural network when executing the computer program.

[0022] The technical solution adopted by the present invention to solve its technical problem is: providing a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method for constructing a scalable model of a RF spiral inductor based on a neural network are implemented.

[0023] Beneficial Effects

[0024] Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages and positive effects compared with the prior art: the present invention constructs a scalable model of a spiral inductor through an artificial neural network, and the scalable inductor model can directly generate equivalent model parameters of the inductor device based on the physical size parameters of the spiral inductor, thereby reducing the difficulty of obtaining parameters of lumped components, thereby improving the device simulation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of a method for constructing a scalable model of a radio frequency spiral inductor according to a first embodiment of the present invention;

[0026] Figure 2 is a schematic diagram of spiral inductor structures of different shapes and physical parameters in the first embodiment of the present invention;

[0027] Figure 3 is a schematic diagram of a π-type equivalent circuit model of a spiral inductor in the first embodiment of the present invention;

[0028] Figure 4 is a corresponding relationship diagram between parameters predicted by the scalable model of the radio frequency spiral inductor and each inductor according to the first embodiment of the present invention;

[0029] Figure 5 This is a comparison chart of electromagnetic simulation and prediction results when the inductor size is W = 10um, S = 10um, R = 100um, T = 3.5;

[0030] Figure 6 This is a comparison chart of electromagnetic simulation and prediction results when the inductor size is W=10um, S=10um, R=100um, T=5.5. DETAILED DESCRIPTION

[0031] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0032] The first embodiment of the present invention relates to a method for constructing a scalable model of a RF spiral inductor based on a neural network, such as Figure 1 As shown, the following steps are included:

[0033] Step 1, obtain the physical size parameters of different spiral inductors. Figure 2 As shown, in this step, the physical size parameters of spiral inductors of different shapes are extracted, including: line width W, spacing S, number of turns T and inner diameter R. In this step, electromagnetic simulation tools can also be used to obtain S' parameters of spiral inductors of different shapes within the required frequency band. The S' parameters are small signal scattering parameters.

[0034] Step 2: Establish a π-type equivalent circuit model for each spiral inductor and obtain the parameters of each component in the π-type equivalent circuit model. Figure 3 What is shown is a π-type equivalent circuit model, which includes the following elements: π-type network inductance Lsp, π-type network series parasitic resistance Rsp, RL network devices (i.e., resistance Rsh and inductance Lsh), substrate parasitic capacitance Csub1 and substrate parasitic capacitance Csub2, substrate parasitic resistance Rsub1 and substrate parasitic resistance Rsub2, microstrip connecting line parasitic inductance Lt1 and microstrip connecting line parasitic inductance Lt2, and overlapping and edge parasitic capacitance Cp between coils.

[0035] Step 3, taking the physical size parameters of each spiral inductor and the parameters of each element in the corresponding π-type equivalent circuit model as a set of data to form a data set, and dividing the data set into a training set and a test set; Figure 2 The line width W, spacing S, number of turns T and inner diameter R of the square spiral inductor on the left and the parameters of each component in its corresponding π-type equivalent circuit model form a set of data. Figure 2 The line width W, spacing S, number of turns T and inner diameter R of the hexagonal spiral inductor in the middle and the parameters of each component in its corresponding π-type equivalent circuit model form a set of data. Figure 2 The line width W, spacing S, number of turns T and inner diameter R of the circular spiral inductor on the right and the corresponding component parameters in the π-type equivalent circuit model form a set of data until the data of all spiral inductors are sorted. Then, multiple sets of data are obtained to form a data set, which is divided into a training set and a test set in a ratio of 8:2.

[0036] Step 4, constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model. The neural network model is trained using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model.

[0037] The neural network model constructed in this step is an artificial neural network model, and its number is the same as and corresponds to the number of elements in the π-type equivalent circuit model, that is, 11 neural network models are constructed in this embodiment, and the input of these 11 neural network models is the physical size parameters of the spiral inductor, and the output is the parameters of the 11 elements in the π-type equivalent circuit model. The number of hidden layers, the number of neurons and the number of iterations in each neural network model are adjusted according to the parameters of the elements in the corresponding π-type equivalent circuit model. When training, the physical size parameters of the spiral inductor in the training set are used as the input of the neural network model, and the parameters of each element in the π-type equivalent circuit model in the training set are used as labels for training, and finally the relationship between the physical size parameters of the 11 spiral inductors and the parameters of the elements in the π-type equivalent circuit model is obtained.

[0038] Step 5, using a test set to test the obtained relationship, if the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the RF spiral inductor; if the test fails, continue to train the neural network model.

[0039] In this step, the physical size parameters of the spiral inductor in the test set are substituted into the obtained relationship to obtain the parameters of the components in the 11 π-type equivalent circuit models. The obtained parameters are compared with the parameters of each component in the π-type equivalent circuit model in the test set. If the errors of the 11 parameters are all within 5%, it means that the test has passed, and the obtained relationship is substituted into the model file to obtain a scalable model of the RF spiral inductor; if one of the errors of the 11 parameters is not within 5%, it means that the test has failed, and the neural network model continues to be trained.

[0040] Figure 3 The equivalent circuit model output by the scalable model of the RF spiral inductor includes the π-type network inductance Lsp, the π-type network series parasitic resistance Rsp, the RL network devices (i.e., the resistance Rsh and the inductance Lsh), the substrate parasitic capacitance Csub1 and the substrate parasitic capacitance Csub2, the substrate parasitic resistance Rsub1 and the substrate parasitic resistance Rsub2, and the overlap between the coils and the correspondence between the edge parasitic capacitance Cp and each inductor. Figure 4 and Figure 5The figures are a comparison of the inductance, Q value, parasitic resistance, and S' parameter electromagnetic simulation results of inductors of two different sizes with the prediction results of the scalable RF spiral inductor model. The results show a good fit, which means that the scalable RF spiral inductor model can accurately predict the parameters of the components in the π-type equivalent circuit model.

[0041] A second embodiment of the present invention relates to a device for constructing a scalable model of a radio frequency spiral inductor based on a neural network, comprising:

[0042] A first acquisition module is used to acquire physical size parameters of different spiral inductors;

[0043] A second acquisition module is used to establish a π-type equivalent circuit model for each spiral inductor and obtain parameters of each element in the π-type equivalent circuit model;

[0044] A data set establishment module is used to take the physical size parameters of each spiral inductor and the parameters of each element in the corresponding π-type equivalent circuit model as a set of data to form a data set, and divide the data set into a training set and a test set;

[0045] Constructing a training module for constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model, and the neural network model is trained using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model;

[0046] The test module is used to test the obtained relationship using a test set; if the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the radio frequency spiral inductor; if the test fails, the neural network model continues to be trained.

[0047] The physical size parameters of the spiral inductor include: line width, spacing, number of turns and inner diameter.

[0048] The elements of the π-type equivalent circuit model include: π-type network inductance, π-type network series parasitic resistance, RL network device, substrate parasitic capacitance, substrate parasitic resistance, microstrip connecting line parasitic inductance, and overlapping and edge parasitic capacitance between coils.

[0049] The neural network model constructed by the training module is an artificial neural network model.

[0050] The number of neural network models constructed by the training module is the same as and corresponds to the number of elements in the π-type equivalent circuit model. The number of hidden layers, the number of neurons and the number of iterations in each neural network model are adjusted according to the parameters of the elements in the corresponding π-type equivalent circuit model.

[0051] A third embodiment of the present invention relates to an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the neural network-based scalable model construction method of the RF spiral inductor of the first embodiment when executing the computer program.

[0052] A fourth embodiment of the present invention relates to a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for constructing a scalable model of a RF spiral inductor based on a neural network of the first embodiment are implemented.

[0053] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0054] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0055] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction method, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0056] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0057] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for constructing a scalable model of a radio frequency spiral inductor based on a neural network, characterized in that: The following steps are involved: Obtain the physical size parameters of different spiral inductors; A π-type equivalent circuit model is established for each spiral inductor, and the parameters of each component in the π-type equivalent circuit model are obtained; the physical size parameters of each spiral inductor and the parameters of each component in the corresponding π-type equivalent circuit model are used as a set of data to form a data set, and the data set is divided into a training set and a test set; Constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model, and training the neural network model using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model; The obtained relationship is tested using a test set. If the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the radio frequency spiral inductor; if the test fails, the neural network model continues to be trained.

2. The method for constructing a scalable model of a RF spiral inductor based on a neural network according to claim 1, characterized in that: The physical dimension parameters of the spiral inductor include: line width, spacing, number of turns and inner diameter.

3. The method for constructing a scalable model of a RF spiral inductor based on a neural network according to claim 1, characterized in that: The elements of the π-type equivalent circuit model include: π-type network inductance, π-type network series parasitic resistance, RL network device, substrate parasitic capacitance, substrate parasitic resistance, microstrip connecting line parasitic inductance, and overlapping and edge parasitic capacitance between coils.

4. The method for constructing a scalable model of a RF spiral inductor based on a neural network according to claim 1, characterized in that: The neural network model is an artificial neural network model.

5. The method for constructing a scalable model of a RF spiral inductor based on a neural network according to claim 1, characterized in that: The number of the neural network models is the same as and corresponds to the number of elements in the π-type equivalent circuit model. The number of hidden layers, the number of neurons and the number of iterations in each neural network model are adjusted according to the parameters of the elements in the corresponding π-type equivalent circuit model.

6. A device for constructing a scalable model of a radio frequency spiral inductor based on a neural network, characterized in that: include: A first acquisition module is used to acquire physical size parameters of different spiral inductors; A second acquisition module is used to establish a π-type equivalent circuit model for each spiral inductor and obtain parameters of each element in the π-type equivalent circuit model; A data set establishment module is used to take the physical size parameters of each spiral inductor and the parameters of each element in the corresponding π-type equivalent circuit model as a set of data to form a data set, and divide the data set into a training set and a test set; Constructing a training module for constructing a neural network model, wherein the input of the neural network model is the physical size parameters of the spiral inductor, and the output is the parameters of each component in the π-type equivalent circuit model, and the neural network model is trained using a training set to obtain a relationship between the physical size parameters of the spiral inductor and the parameters of each component in the π-type equivalent circuit model; The test module is used to test the obtained relationship using a test set; if the test passes, the obtained relationship is brought into the model file to obtain a scalable model of the radio frequency spiral inductor; if the test fails, the neural network model continues to be trained.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for constructing a scalable model of a RF spiral inductor based on a neural network as described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for constructing a scalable model of a RF spiral inductor based on a neural network as claimed in any one of claims 1 to 5 are implemented.

Images