A method and system for obtaining a linear model

Abstract

The present invention discloses a method and system for obtaining a linear model, and belongs to the field of simulation modeling. In view of the problem that the existing simulation model is inaccurate and inefficient, the present invention provides a method for obtaining a linear model, comprising the following steps: obtaining an original image; determining whether the original curve in the original image includes both an amplitude-frequency curve and a phase-frequency curve; if the original curve includes both an amplitude-frequency curve and a phase-frequency curve, sampling the points on the original curve, and then creating a complete model; if the original curve lacks a phase-frequency curve, restoring the phase-frequency curve to obtain a complete curve model, sampling the points on the complete curve, and then creating a complete model; generating the generated complete model into a model format for recognition by mainstream simulation tools. The present invention is easy to operate, and the added detection steps enable different situations to correspond to different operations, thereby ensuring the accuracy of model establishment and improving work efficiency. The system of the present invention improves the overall degree of automation and greatly improves work efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of software simulation modeling, and more specifically, relates to a method and system for obtaining a linear model. Background Art

[0002] Nowadays, early prototype verification of electronic circuit systems is increasingly inseparable from software simulation. Simulation can discover deficiencies in the design at an early stage and control the overall trend of the design object. However, obtaining some simulation models has always been a problem. Some chips (mainly RF devices, operational amplifiers and other devices) do not provide complete model curves, such as only providing amplitude-frequency response curves but no phase-frequency response curves, or there is a contradiction between the amplitude-frequency and phase-frequency response curves provided, or the model may contain unnecessary noise. The above incomplete models or models containing erroneous information will bring unreasonable results to the simulation, thereby misleading the design direction.

[0003] For example, Chinese patent application number CN202110842163.4, published on October 8, 2021, discloses a simulation modeling method based on graphic matching recognition, including: obtaining configuration information of drawings in a source file of a first format, the configuration information including line segment information and text information; generating a redrawn picture and a mapping table according to the obtained configuration information, wherein the redrawn picture includes a gallery composed of line segments, connecting line segments between galleries, and the mapping table includes a mapping relationship between text content and line segments and their position coordinates; identifying the gallery category, the attributes of the connecting line segments, and the parameter attributes in the mapping table in the redrawn picture according to a preset feature information library; converting the identified gallery category, the attributes of the connecting line segments, and the parameter attributes into a simulation model file; importing the simulation model file into the simulation modeling software to generate an executable simulation model. The disadvantage of this patent is that although it can effectively provide modeling efficiency, the overall modeling accuracy is average.

[0004] Another example is Chinese patent application number CN202110115835.1, which was published on June 18, 2021. The patent discloses a packaging model for improving DDR simulation accuracy, including a resistor R and a transmission line model with impedance Z and time delay TD. The resistor R and the transmission line model are connected in series. The present invention also relates to a packaging model modeling method for improving DDR simulation accuracy. In the packaging link of high-frequency signals, the RLC parameters are extracted and corrected to obtain a corrected packaging model to characterize the packaging link. The present invention uses the RLC parameters of the existing main chip simulation model to make corrections, and connects the original resistors in series to obtain a corrected packaging model. A set of data signals of the DDR system are simulated by the corrected packaging model to obtain simulation results consistent with the test results. The present invention improves the DDR simulation accuracy, ensures the consistency of the simulation results and the test results, and provides accurate data for the design and debugging of PCB boards. The shortcomings of this patent are: although the accuracy is improved, the process is complicated and the cost is high. Summary of the invention

[0005] 1. Problems to be solved

[0006] In view of the problem that existing simulation modeling is inaccurate and inefficient, the present invention provides a method and system for obtaining a linear model. The method of the present invention determines whether the phase-frequency curve in the original image is missing by detecting the original image. If it is missing, the phase-frequency curve is restored in a reasonable manner. If it is not missing, the curve is sampled, thereby ensuring the integrity and accuracy of the final model establishment, avoiding the problem of inaccurate design direction caused by unreasonable results in the subsequent simulation. The system of the present invention performs corresponding functions through each module, has a simple structure, is easy to build, improves the overall automation level, and greatly improves work efficiency.

[0007] 2. Technical solution

[0008] To solve the above problems, the present invention adopts the following technical solutions.

[0009] A method for obtaining a linear model comprises the following steps:

[0010] S1: Get the original image with the original curve of the model;

[0011] S2: Detect the original image to determine whether the original curve in the original image includes both the amplitude-frequency curve and the phase-frequency curve;

[0012] S3: If the original curve includes both the amplitude-frequency curve and the phase-frequency curve, the original image is identified to obtain the original curve, and the points on the original curve are sampled, and then a complete model is created;

[0013] S4: If the original curve only includes the amplitude-frequency curve and lacks the phase-frequency curve, the phase-frequency curve is restored to obtain the complete curve, the points on the complete curve are sampled, and then the complete model is created;

[0014] S5: Generate the generated complete model into a model format that can be recognized by mainstream simulation tools for subsequent simulation operations.

[0015] Furthermore, before step S2, a preprocessing operation is also performed on the original image, and the preprocessing operation includes removing interference, reducing noise and sharpening the original image.

[0016] Furthermore, when sampling is performed in step S3 or step S4, the sampling density is determined by using the error between the sampling curve composed of the sampling points and the original curve.

[0017] Furthermore, in step S4, when more than 90% of the energy of the device is concentrated in the low frequency, the Hilbert transform method is used to restore the phase-frequency curve; and when the energy distributed in the high-frequency part becomes non-negligible for the simulation of the system, such as accounting for more than 10%, the rational polynomial method is used to restore the phase-frequency curve.

[0018] Furthermore, the specific formula of rational polynomial is as follows:

[0019]

[0020] Where: Sij represents the energy transfer coefficient from port j to port i, and dij represents the constant term;

[0021] With P ij,n are mutually conjugate poles, With r ij,n are mutually conjugated numerator coefficients, Tij is the delay constant, and s is the Laplace character.

[0022] Furthermore, when the amplitude-frequency curve and the phase-frequency curve exist at the same time, the two curves are tested to see whether they satisfy the causal correspondence relationship. If they do not satisfy the causal correspondence relationship, correction processing is performed.

[0023] A system using a linear model acquisition method as described in any one of the above, comprising:

[0024] Image acquisition module: used to obtain original images;

[0025] Detection module: used to determine whether the original image includes both the amplitude-frequency curve and the phase-frequency curve;

[0026] Restoration module: used to restore the phase-frequency curve;

[0027] Identification module: used to identify curves;

[0028] Sampling module: used for sampling on the curve;

[0029] Creation module: used to create a complete model;

[0030] Format conversion module: used to convert the format of the complete model.

[0031] Furthermore, it also includes an alarm module: used to monitor the working status of each module and issue timely warnings.

[0032] 3. Beneficial effects

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) The present invention determines whether the phase-frequency curve in the original image is missing by detecting the original image. If it is missing, the phase-frequency curve is restored in a reasonable way. If it is not missing, the curve is sampled, thereby ensuring the integrity and accuracy of the final model establishment, avoiding the problem of unreasonable results in the subsequent simulation and inaccurate design direction. The whole method is easy to operate, and the added detection steps make different situations correspond to different operations, ensuring the accuracy of model establishment while improving work efficiency, effectively solving the inaccuracy of modeling when the original curve is incomplete in the past;

[0035] (2) The present invention pre-processes the original image and performs noise reduction on the original image, thereby reducing the amount of subsequent calculations on the original image, effectively improving work efficiency, reducing error interference, and thereby improving the accuracy of subsequent operations; and when sampling curve points, the sampling density is judged by error to avoid excessive sampling that increases the amount of calculation and causes waste of resources, and too little sampling that easily loses details and causes inaccurate results. Therefore, the density judgment is performed to reasonably ensure the amount of sampling while ensuring accuracy;

[0036] (3) The present invention performs different phase-frequency curve restoration on different devices to ensure the accuracy of phase-frequency curve restoration, thereby providing an accurate basis for subsequent modeling. At the same time, when there are amplitude-frequency curves and phase-frequency curves, it is added to detect whether the two curves meet the causal correspondence operation, and if they do not meet the requirements, correction is performed. This further avoids the situation where the amplitude-frequency curve and the phase-frequency curve contain unnecessary noise or contradiction, thereby further improving the accuracy of subsequent modeling and ensuring the smoothness and accuracy of the subsequent simulation process.

[0037] (4) The system of the present invention performs corresponding functions through each module, works stably and does not affect each other, has a simple structure and is easy to build. The setting of each module greatly improves the overall automation level, greatly improves work efficiency, and reduces the investment in labor costs; and through the introduction of the alarm module, the working status of each other module is monitored and warned in real time, thereby improving the timeliness of the entire process and ensuring that the staff can make corresponding adjustments in time when the system is damaged, thereby ensuring the safety and stability of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a schematic diagram of the coordinate system;

[0039] Figure 2 A schematic diagram of a process of the present invention;

[0040] Figure 3 Another schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0041] The present invention is further described below in conjunction with specific embodiments and drawings.

[0042] It is explained here that, generally speaking, the establishment of a device model in an electronic circuit generally relies on the device manufacturer to provide a model curve. When the device manufacturer does not provide the model curve or the model curve is incomplete, the subsequent simulation process cannot be performed accurately. This application is based on the explanation of this situation.

[0043] Example 1

[0044] like Figure 1 , Figure 2 and Figure 3 As shown, a method for obtaining a linear model comprises the following steps:

[0045] S1: Obtain an original image with an original curve of the model; specifically, the original image can be obtained from a device manufacturer or other data manuals, and the original image contains the original curve of the device model.

[0046] S2: Detect the original image to determine whether the original curve in the original image includes both the amplitude-frequency curve and the phase-frequency curve. This step is to detect the integrity of the original image, so as to facilitate different subsequent operations according to different situations of the original image, thereby greatly improving work efficiency.

[0047] S3: If the original curve includes both the amplitude-frequency curve and the phase-frequency curve, it means that the original curve is complete. At this time, the original image is identified to obtain the original curve. The identification here includes identifying the coordinates of the original curve, which is generally a rectangular coordinate system, that is, identifying the horizontal and vertical coordinates of the original curve to obtain the physical quantities represented by the horizontal and vertical coordinates respectively; when the coordinate system identification here is incorrect, it can be identified by adding manual assistance to ensure the accuracy of the identification; at the same time, the points on the original curve are sampled to obtain a number of data points, and these data points are saved in a certain format for subsequent creation of a complete model; specifically, each data point includes horizontal and vertical coordinate values, and the sampling method can be to select a point on the horizontal or vertical coordinate after identifying the horizontal and vertical coordinates, and read the corresponding vertical or horizontal coordinate value on the original curve; or it can be by pixel reading of the original curve, that is, defining the position of the original curve in the original image as the target area, and the original curve in the target area is used to reflect the parameter information of the specified type, and then determining the coordinates of the selected pixel points in the target area, and determining the parameter values ​​corresponding to the selected pixel point coordinates based on the association between the values ​​of the specified type parameters and the pixel point coordinates, that is, the horizontal and vertical coordinate values. Of course, since there are many ways to sample image data, which are also common in the prior art and do not involve the core creative point of this application, they will not be described in detail.

[0048] S4: If the original curve only includes the amplitude-frequency curve and lacks the phase-frequency curve, the phase-frequency curve is restored to obtain the complete curve, and the points on the complete curve are sampled. The specific sampling steps are consistent with step S3, and then a complete model is created. Specifically, using the model of a naturally existing device must not violate the principle of causality. There are two main methods for restoring the phase-frequency curve. One is to obtain the phase-frequency curve from the amplitude-frequency curve through the Hilbert transform based on the principle that the model must satisfy the causal property; the other is to fit the amplitude-frequency using a polynomial that naturally satisfies causality. In this case, the byproduct is the phase-frequency curve. Of course, different methods should be used to restore the phase-frequency curve for different situations to ensure the accuracy of the restoration. In this embodiment, for devices in which more than 90% of the energy is concentrated in low frequencies, the Hilbert transform is used to restore the phase-frequency curve; the specific formula of the Hilbert transform is as follows:

[0049]

[0050] in: represents the transformed function, x(t) represents the original function, τ is the integral variable, and [H] represents the Hilbert transform operation function; when the energy distributed in the high-frequency part becomes non-negligible for the simulation of the system, such as when the energy distributed in the high-frequency part accounts for more than 10%, the phase-frequency curve is restored by a rational polynomial; the specific formula of the rational polynomial is as follows:

[0051]

[0052] Where: Sij represents the energy transfer coefficient from port j to port i, and dij represents the constant term;

[0053] With P ij,n are mutually conjugate poles, With r ij,n are mutually conjugated numerator coefficients, Tij is the delay constant, and s is the Laplace character. This step ensures the accuracy of phase-frequency curve recovery by recovering different phase-frequency curves for different devices, providing an accurate basis for subsequent modeling.

[0054] S5: Generate the generated complete model into a model format that can be recognized by mainstream simulation tools for subsequent simulation operations; specifically, generate different complete models according to the fitting object, and generate the complete model into model formats of different formats, such as discrete data points, S parameters, SPICE netlist and other model formats. If it is a RF device, S parameters and SPICE netlist can be selected. If it is an operational amplifier or some other linear device system, SPICE netlist can be selected to describe the transfer function. The specific situation can be analyzed specifically.

[0055] The present invention determines whether the phase-frequency curve in the original image is missing by detecting the original image. If it is missing, the phase-frequency curve is restored in a reasonable way. If it is not missing, the curve is sampled, thereby ensuring the integrity and accuracy of the final model establishment, avoiding the problem of unreasonable results in the subsequent simulation leading to inaccurate design direction; the whole method is easy to operate, and the added detection steps make different situations correspond to different operations, ensuring the accuracy of model establishment while improving work efficiency, effectively solving the inaccuracy of modeling when the original curve is incomplete in the past. And it effectively avoids the coping plan when the manufacturer does not provide the model, reduces the dependence on the outside world, and has a wide range of application prospects.

[0056] Example 2

[0057] Basically the same as Example 1, specifically, in this embodiment, before step S2, a preprocessing operation is also included for the original image, and the preprocessing operation includes removing interference, reducing noise and sharpening the original image. Because the original image may be affected when it is uploaded, or there are shadows or stains in the original image, and the coordinate axis undergoes various changes during the shooting process, a preprocessing process is performed on the original image to reduce noise, thereby reducing the subsequent calculation amount of the original image, effectively improving work efficiency, and reducing error interference to improve the accuracy of subsequent operations.

[0058] At the same time, in this embodiment, when sampling is performed in step S3 or step S4, the error between the sampling curve composed of the sampling points and the original curve is used to judge the sampling density. When sampling, in order to avoid the problem that too much sampling increases the amount of calculation and causes waste of resources, and too little sampling easily loses details and causes inaccurate results, a density judgment index is introduced, which is specifically expressed by the cumulative error between the curve composed of the sampling points and the original curve, written as When the cumulative error is maintained within a certain range, it means that the sampling quantity meets the requirement, and no more sampling is performed, so as to reasonably ensure the sampling quantity and accuracy. The specific range value can be obtained by experience and is determined according to different situations, and is not specifically limited in this embodiment.

[0059] Furthermore, a correction process is added in this embodiment, that is, when the amplitude-frequency curve and the phase-frequency curve exist at the same time, the two curves are detected to detect whether the two curves satisfy the causal correspondence relationship. If the causal correspondence relationship is not satisfied, a correction process is performed. At present, there are mainly two methods for causal correction of models: (1) using the real part and the imaginary part of the model to satisfy the Hilbert transform to check and correct. The disadvantage is that the model has a limited bandwidth and cannot strictly satisfy the Hilbert transform; (2) using rational compression technology to fit the existing curve with a polynomial that satisfies causality. Corresponding detection of the amplitude-frequency curve and the phase-frequency curve that exist at the same time further avoids the situation where the amplitude-frequency curve and the phase-frequency curve contain unnecessary noise or contradictions, thereby further improving the accuracy of subsequent modeling and ensuring the smoothness and accuracy of the subsequent simulation process.

[0060] Example 3

[0061] A system using a method for obtaining a linear model as described in any one of the above embodiments 1-2, comprising: an image acquisition module: used to obtain an original image, and can perform a preprocessing process on the original image to reduce the error caused by the original image to subsequent operations; a detection module: used to determine whether the original image includes both an amplitude-frequency curve and a phase-frequency curve; a recovery module: used to restore the phase-frequency curve; an identification module: used to identify a curve, where the identification curve is to identify a complete curve, i.e., including an amplitude-frequency curve and a phase-frequency curve; a sampling module: used to perform sampling on the curve, and similarly, sampling is also performed on the complete curve; a creation module: used to create a complete model; a format conversion module: used to convert the format of the complete model. The system of the present invention performs corresponding functions through each module, works stably and does not affect each other, has a simple structure, is easy to build, and the setting of each module greatly improves the overall automation level, greatly improves work efficiency, and reduces the investment in labor costs; at the same time, the integrity and accuracy of the final model establishment are guaranteed to avoid the problem of inaccurate design direction caused by unreasonable results in the subsequent simulation.

[0062] At the same time, the system also includes an alarm module: it is used to monitor the working status of each module and issue timely warnings. Through the introduction of the alarm module, the working status of other modules can be monitored and warned in real time, thereby improving the timeliness of the entire process and ensuring that the staff can make timely corresponding adjustments when the system is damaged, thereby ensuring the safety and stability of the entire system.

[0063] The examples described in the present invention are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made to the technical solutions of the present invention by engineers and technicians in this field should all fall within the protection scope of the present invention.

Claims

1. A method for obtaining a linear model, Features: The following steps are involved: S1: Get the original image with the original curve of the model; S2: Detect the original image to determine whether the original curve in the original image includes both the amplitude-frequency curve and the phase-frequency curve; S3: If the original curve includes both the amplitude-frequency curve and the phase-frequency curve, the original image is identified to obtain the original curve, and the points on the original curve are sampled, and then a complete model is created; S4: If the original curve only includes the amplitude-frequency curve and lacks the phase-frequency curve, the phase-frequency curve is restored to obtain the complete curve, the points on the complete curve are sampled, and then the complete model is created; In step S4, when more than 90% of the energy of the device is concentrated in the low frequency, the Hilbert transform method is used to restore the phase-frequency curve; and when the energy distributed in the high frequency part becomes non-negligible for the simulation of the system, the rational polynomial method is used to restore the phase-frequency curve, and the non-negligible means that the energy distributed in the high frequency part accounts for more than 10%; S5: Generate the generated complete model into a model format that can be recognized by the simulation tool for subsequent simulation operations.

2. The method for obtaining a linear model according to claim 1, Features: Before step S2, the original image is also preprocessed, and the preprocessing operation includes removing interference, reducing noise and sharpening the original image.

3. The method for obtaining a linear model according to claim 1, Features: When sampling is performed in step S3 or step S4, the sampling density is determined by using the error between the sampling curve composed of the sampling points and the original curve.

4. The method for obtaining a linear model according to claim 1, Features: The specific formula of rational polynomial is as follows: Where: S ij represents the energy transfer coefficient from port j to port i, d ij represents a constant term; With P ij,n are mutually conjugate poles, With r ij,n are the molecular coefficients that are conjugated to each other, T ij is the delay constant and s is the Laplace character.

5. The method for obtaining a linear model according to claim 1, Features: When the amplitude-frequency curve and the phase-frequency curve exist at the same time, the two curves are tested to see whether they satisfy the causal correspondence relationship. If they do not satisfy the causal correspondence relationship, correction processing is performed.

6. A system using a method for obtaining a linear model as claimed in any one of claims 1 to 5, Features: include: Image acquisition module: used to obtain original images; Detection module: used to determine whether the original image includes both the amplitude-frequency curve and the phase-frequency curve; Restoration module: used to restore the phase-frequency curve; Identification module: used to identify curves; Sampling module: used for sampling on the curve; Creation module: used to create a complete model; Format conversion module: used to convert the format of the complete model.

7. A linear model acquisition system according to claim 6, Features: It also includes an alarm module: used to monitor the working status of each module and provide timely warnings.

Images