40 results about "Large-signal model" patented technology

The present invention discloses a transistor modeling method based on an artificial neural network. The method comprises: constructing a corresponding artificial neural network topological structure for a current source, a charge source and a non-linear element separately; according to the artificial neural network topological structure and a value of an internal parameter, training the current source, the charge source and the non-linear element separately by using an artificial neural network technology; and importing the well-trained current source, charge source and non-linear element into circuit simulation software, adding an external parasitic inductor, capacitor and resistor, and carrying out encapsulation to form a large signal model of a transistor. The method can be adapted to transistor devices under various processes, and by introducing a channel temperature variable and an ambient temperature variable into an input layer of the artificial neural network topological structure during construction of the current source, the charge source and the non-linear element, memory effects of the transistor, such as self-heating, can be effectively modeled. The present invention further discloses a system modeling method based on an artificial neural network.

The invention discloses a parameter extraction method for InP HBT small-signal models, which mainly solves the problems of the prior art, i.e. complex extraction process and inaccurate extraction results. The technical scheme of the invention is as follows: an open circuit soldering point structure and a short circuit soldering point structure are adopted to analyze an equivalent circuit to extract parasitic parameters; a collector open circuit state is adopted to analyze the equivalent circuit to extract external resistance parameters; a circuit network theory is utilized to analyze the scattering parameter S, impedance parameter Z and admittance parameter Y of the intrinsic part of an InP HBT device; and a layer-by-layer stripping method is adopted to determine a fixed expression for each intrinsic model parameter to extract intrinsic parameters. The parameter extraction method has the advantage of accurate, rapid and visual parameter extraction. A simulation result shows that a small-signal model result extracted by the method can perfectly fit the scattering parameter S of an actual device-testing result and can be used for directing circuit design and the determination of a large-signal model peripheral circuit.

The invention discloses a method for constructing a nonlinear scalable GaN HEMT model. The problems of incapability of an existing GaN HEMT large-signal model in accurately fitting various effects in output current and difficulty in simulation of devices with different sizes are mainly solved. According to the technical scheme, the method comprises the following steps: 1, measuring a used device, and calculating a parameter of an EEHEMT large-signal model; 2, constructing a current source including direct current output curve data of the used device, scaling the current source by virtue of a normalization factor, and constructing an active compensation sub-circuit; 3, constructing a source potential adjustable active compensation sub-circuit on the basis of the active compensation sub-circuit, and connecting the source potential adjustable active compensation sub-circuit in parallel with an EEHEMT to finish construction of the nonlinear scalable GaN HEMT model. According to the method, various effects in a current direct output curve can be accurately fitted, a grid width can be effectively scaled, and the method can be used for GaN circuit design.

The invention relates to an airborne power system stability analysis method based on a unified large signal model. The method comprises the steps that first, the unified large signal model of an airplane power system is established; and second, a stability criterion of the airplane power system under the unified large signal model is solved, wherein a large signal stability criterion under the unified large signal model is obtained through solving by the adoption of a mixed potential function theory in combination with a Lyapunov energy function, and the stability criterion is also a small signal stability criterion of the airplane power system.

The invention relates to the field of power system transformer simulation, in particular to a large-signal simulation model of a single-phase power electronic transformer, aiming at improving the simulation speed. The simulation model of the invention comprises an equivalent large signal model of a cascaded H-bridge converter and an equivalent large signal model of a dual active bridge converter.The cascaded H-bridge converter equivalent large signal model is composed of a cascaded H-bridge converter module, a first controlled voltage source H1, a first controlled current source S1 and a support capacitor C1. The cascaded H-bridge converter module includes four input signals and four output terminals. The equivalent large signal model of double active bridge converters is composed of double active bridge converter modules, an equivalent resistor Req, an equivalent inductor Leq, an output capacitor C2, a second controlled voltage source H2 and a second controlled current source S2. Thedual active bridge converter module consists of four input signals and four output terminals. The model can accelerate the simulation speed of the single-phase power electronic transformer when the single-phase power electronic transformer is disturbed by large signals such as load switching or fault.

The invention discloses an establishing method of a multi-physics-field coupling large signal model of a GaN HEMT component. The method comprises the steps that S1, mapping relationships among physical parameters of the component and thermal parameters, electric parameters and stress parameters of the component are established respectively; S2, an analytical expression of the inner core of the large signal model is induced based on a large signal model modeling principle; S3, thermal-electric coupling, thermal-dynamic coupling and dynamic-electric coupling are quantized and embedded in the inner core of the large signal model to obtain a corrected inner core of the large signal model; S4, the corrected inner core of the large signal model is brought into a model equivalent circuit topologyto construct a complete electric-thermal-dynamic multi-physics-field coupling large signal model. According to the method, the influence of stress of a GaN outward extending layer is introduced, so that the electric-thermal-dynamic multi-physics-field coupling effect of the component is comprehensively described to improve the accuracy of a GaN HEMT component model, and the model can be used forguiding the designing and processing improvement of a novel component.

The invention provides a GaN technological parameter statistical analysis method based on large signal equivalent circuit model. The analysis method comprises the following steps: a step 1: establishing a GaN device small signal equivalent circuit model, and extracting small signal model parameters; a step 2: establishing a GaN device large signal equivalent circuit model, and extracting large signal model parameters, that is non-linear current source model parameter and non-linear capacitance model parameter; a step 3: actually measured microwave characteristics of the device are used as objects in order to tune and optimize large signal model parameters; a step 4: extracting multiple-batch GaN device technological parameters based on established large signal model, and carrying out statistical analysis of the technological parameters. The GaN device model technological parameter statistical analysis method firstly establishes the GaN device small signal equivalent circuit model, andthen establishes GaN device large signal equivalent circuit model related to technological parameters, finally technological parameter statistical distribution is obtained by multiple-bath device modeling, and the method is good for device yield analysis and technological parameter optimization.

A decentralized voltage control method for a microgrid based on nonlinear state observers, comprising the steps of step 10), establishing a large-signal model of distributed generations, a connection network and impedance-type loads in the microgrid; step 20), establishing a Luenberger-like nonlinear state observer for each distributed generation; step 30), estimating the dynamic characteristics of other distributed generations in real time based on the local measured values of each distributed generation; and step 40), implementing the decentralized voltage control based on the control requirements of reactive power sharing and voltage restoration. The control method realizes the voltage control of microgrid based on the decentralized state observers, which does not rely on communication transmission or remote measurement and avoids the adverse effects of communication latency and data drop-out on the control performance.

The invention discloses a large-signal model method for a germanium-silicon heterojunction transistor. The large-signal model method includes the steps that a Gummel-Poon model is established; a large-signal model composed of the Gummel-Poon model, second base resistance, second collector resistance, second emitter resistance, second base collecting capacitance and second base transmitting capacitance is established; parasitic resistance of electrodes of the transistor is tested, acquired third base resistance is the sum of corresponding first base resistance and the corresponding second base resistance, acquired third collector resistance is the sum of corresponding first collector resistance and the corresponding second collector resistance, and acquired third emitter resistance is the sum of corresponding first emitter resistance and the corresponding second emitter resistance. An S parameter test is conducted on the transistor, third base collecting resistance and third base transmitting capacitance are calculated, the third base collecting resistance is the sum of corresponding first base collecting capacitance and the corresponding second base collecting capacitance, and the third base transmitting resistance is the sum of corresponding first base transmitting resistance and the corresponding second base transmitting resistance. The large-signal model is simulated, a large-signal test is conducted on the transistor, and parameter fitting of the large-signal model is conducted. The simulation accuracy of the model can be improved.

The invention discloses a universal nonlinear modeling module applied to a switching power supply and a modeling method thereof. A universal ITIS module corresponds to three different modeling models,namely, a universal ITIS module steady-state model, a universal ITIS module large-signal model and a universal ITIS module small-signal model. A universal ITIS module is used to carry out unified expression on a nonlinear part including part of inductive elements and a switching device in a switching converter to complete the modeling of the complex nonlinear part in the switching converter in advance. A steady-state model, a large-signal model or a small-signal model of a target converter can be established by only establishing an external linear network state equation of an equivalent ITISmodule of the target converter and substituting the equation into a universal ITIS module. According to the invention, the difficulty of user modeling is greatly simplified, and the efficiency of modeling analysis is improved.

The invention discloses a method and system for measuring the frequency dispersion characteristic of transistor output resistance. The method includes the steps that a negative bias voltage is provided for the grid electrode of a transistor; a forward bias voltage and an AC signal are provided for the drain electrode of the transistor; the output resistance of the transistor is acquired when the AC signal is under a selected frequency; the corresponding relation between the output resistance and the selected frequency of the AC signal is acquired so as to obtain the frequency dispersion characteristic of the output resistance. By means of the method and system, the frequency dispersion characteristic of the transistor output resistance can be measured, the advantages of being simple to measure and high in accuracy are achieved, and the output resistance frequency dispersion characteristic obtained through the method and system can be used for evaluating material growth quality, monitoring technological processes, and analyzing device physical characteristics; in addition, technicians in the field can further use the frequency dispersion characteristic to correct the correlation of the output resistance of a frequency dispersion sub circuit in a large-signal model so as to accurately establish the large-signal model.

Provided is a method of configuring a large signal model of an active device. The method may include configuring a large signal model of a first active device, preparing a first measured value on a first characteristic of a second active device, the second active device being larger than the first active device, processing the large signal model of the first active device using a circuit simulator to configure a large signal model of the second active device, simulating the large signal model of the second active device to obtain a calculated value on the first characteristic, comparing the measured and calculated values on the first characteristic to each other, and establishing the large signal model of the second active device, if a difference between the measured and calculated values on the first characteristic may be smaller than a predetermined error margin. Further, if the difference between the measured and calculated values on the first characteristic may be greater than the predetermined error margin, the large signal model of the second active device may be configured by modifying parameters of passive devices.

The invention discloses a physical-base large-signal modeling method and system for microwave gallium nitride devices. The method includes the steps of obtaining test result data of each gallium nitride device on the same processing line under different input powers; calculating the mean value of the test result data under each input power; based on the mean value, building a physical-base large-signal model, setting physical parameters as nominal values, and then obtaining the initial values of fitting parameters; selecting devices corresponding to an upper edge curve and a lower edge curve;building physical-base large-signal models of a upper edge device and a lower edge device; substituting the fitting parameters into the physical-base large-signal models of the two edge devices, adjusting the fitting parameters and the physical parameters to ensure that the two models share the same set of fitting parameters and use different physical parameters, and therefore determining the fitting parameters and physical parameters of the physical-base large-signal models. According to the scheme, the extraction reliability of the fitting parameters and the accuracy of the physical parameters can be improved, and therefore the precision of the models is improved.

The invention discloses an LADRC control method based on a DAB converter, and the method comprises the steps: carrying out the LADRC control through a DAB converter linear large signal model with a unit value of transmission power as a control quantity; wherein the DAB converter linear large signal model is based on a single-side asymmetric duty ratio modulation method, and the output voltage is controlled by directly controlling the scaled transmission power. According to the control method provided by the invention, on the premise of ensuring the system performance, the traditional linear active interference suppression controller is improved, and the parameter setting process is optimized. And finally, verification is carried out through a simulation platform, the overall efficiency in the full power range is remarkably improved, the proposed control has better robustness than the traditional PI control, and the problems of excessive parameters and difficult value taking in the design of the traditional ADRC are greatly simplified.

The invention discloses a neural network space mapping method for large signal modeling of a power transistor. The method can be used for establishing a large signal model of the power transistor which accurately considers a room temperature effect and a self-heating effect. The method comprises the steps of 1, selecting an electric heating rough model according to the type of a to-be-modeled power transistor; 2, initializing a dynamic mapping neural network; 3, respectively establishing a direct current simulation model, a small signal simulation model and a large signal simulation model of the dynamic neural network space mapping electric heating initial model; 4, obtaining a final weight w1 of the dynamic mapping neural network; step 5, establishing a dynamic mapping neural network in commercial circuit simulation software; and step 6, establishing a dynamic neural network space mapping electric heating model in commercial circuit simulation software. The method has the beneficial effects that the large-signal electric heating characteristics of the power transistor considering the room temperature effect and the self-heating effect are reflected more accurately; the method is embedded into commercial circuit simulation software for high-level microwave circuit and system simulation, design and optimization.

The invention discloses a modeling method of a field effect transistor, which comprises the following steps: establishing a small signal intrinsic part equivalent circuit of an FET (Field Effect Transistor), and obtaining a relationship between an internal intrinsic parameter and an external bias; constructing a large signal model of the FET, wherein the large signal model comprises a gate charge source, a drain charge source, a gate current source, a drain current source and an NQS sub-circuit; performing path integration on the port voltage to obtain a relation among the nonlinear current source, the charge source and the port voltage; and then a neural network analysis model can be obtained through storage in a lookup table mode or through neural network training. In the integration process of the current source and the charge source, the NQS effect is completely eliminated, the modeling mode is consistent with the physical mechanism of the NQS, unification of a small signal model and a large signal model is ensured, the precision of the model is not influenced by a frequency band, a high-order source does not need to be used, and in terms of robustness and precision of the model and the difficulty of model extraction, the modeling efficiency is greatly improved. And the model framework is obviously superior to the existing model framework.

The invention discloses a full-order dynamic modeling method for a universal switching converter in a current interrupted mode, and the method comprises the steps: constructing a to-be-modeled targetconverter topology, and expressing a target converter through a TIS module and a linear network; determining internal parameters of the TIS module and an external linear network state equation, and constructing a TIS module equivalent model; constructing a TIS module average model in a DCM mode, wherein the TIS module average model comprises a steady-state model, a large signal model and a full-order small signal model; and substituting the TIS module average model in the DCM mode into the TIS module equivalent model to obtain a steady-state model, a large-signal model and a full-order small-signal model of the target converter in the DCM mode. Modeling of a complex nonlinear part in the switching converter is completed in advance. The difficulty of user modeling can be greatly simplified,the efficiency of modeling analysis is improved, the method is suitable for steady-state and dynamic modeling of the switching power supply comprising a transformer and a coupling inductor, the frequency bandwidth of the finally obtained model can reach 1 / 2 switching frequency or above, and compared with a traditional modeling method, the method has higher modeling precision and effective frequency bandwidth.

The invention discloses a modeling method of a microwave GaN power device. The modeling method of the microwave GaN power device comprises the following steps: creating a small signal equivalent circuit model of the GaN power device, obtaining the parameters of the small signal model; according to actually-measured multi-bias scattering parameters, optimizing the parameters of the small signal model, creating a large signal equivalent circuit model of the GaN power device, obtaining the parameters of the large signal model, by using actually-measured multi-bias scattering parameters and large signal characteristic parameters as the goals, tuning and optimizing the parameters of the large signal model; carrying out the modeling for multiple batches of GaN power devices, obtaining the scattering parameters of a process line, and a large signal statistical model. According to the modeling method of the microwave GaN power device, the small signal equivalent circuit model and large signal equivalent circuit model of the GaN power device are created, according to the model parameter statistical characteristics, the scattering parameters of the GaN process line are thus obtained and the large signal characteristic statistical model is created at the same time, therefore, the accurate modeling for the small signal characteristics and the large signal characteristics of a specific process line can be achieved, and the degree of accuracy of the model is improved.

The invention discloses a Boost converter large signal modeling method. Constraint conditions during modeling are increased, a large signal modeling method based on joint constraint of the state matrix and the input matrix is provided, and the large signal modeling method is suitable for large signal modeling of Boost converters adopting different modulation modes. The large signal model of the Boost circuit is more accurate and is not limited by a modulation mode, so that subsequent equivalent circuit derivation is facilitated.

The invention discloses a transistor modeling method based on narrow-pulse small signal measurement in the field of transistors. The method comprises the following steps of 1, de-embedding: de-embedding an S parameter on the transistor by using an open-circuit\short-circuit structure corresponding to the transistor; 2, extracting an external parameter: extracting an external parasitic parameter from the S parameter of the open-circuit\short-circuit structure; 3, building a constitutive relation: processing the de-embedded S parameter of the transistor, and determining each order of current source and each order of charge source needed by a transistor model; 4, training and fitting: training or fitting the current source and the charge source; and 5, modeling: importing the trained or fit each order of current source and charge source into circuit simulation software, and packaging to form a big signal model of the transistor. According to the method provided by the invention, a small signal equivalent circuit in the transistor is not needed, narrow-pulse small signal measured data is directly used for extracting the constitutive relation between the voltage and current of a port, and thus a complete model of the transistor is built; the method can be used for modeling the transistor.

The invention relates to a parameter extraction modeling method suitable for a discrete device after scribing. The method comprises the steps that a left side bonding pad, a to-be-tested device after scribing and a right side bonding pad are connected into a GSG on-chip testing structure in a gold wire bonding mode; capacitive compensation parameters of the left side compensation network and the right side compensation network caused by the left side bonding pad and the right side bonding pad are obtained through measurement; respectively determining inductive compensation parameters of left and right side compensation networks caused by gold wires bonded by gold wires based on the straight-through distances between the left and right side bonding pads and the to-be-tested device; determining S parameters of the left and right side compensation networks according to the inductive compensation parameters and the capacitive compensation parameters of the left and right side compensation networks; and respectively adding the S parameters of the compensation networks on the left side and the right side to the input end and the output end of test calibration compensation data of the vector network analyzer so as to remove the wiring influence of the GSG on-chip test structure. According to the method, the accurate large signal model can be quickly and effectively obtained, and model change caused by inter-chip non-uniformity of a device process is avoided.

The invention discloses a Buck type converter cascade system stability analysis method based on a gyrator model, and belongs to the technical field of power electronics and control. The method comprises the following steps: 1, selecting an inductive current and an output voltage as state variables, establishing a large signal model of the converter in a continuous conduction mode, and obtaining a differential equation about the inductive current and the output voltage; 2, deriving a state-space equation according to the differential equation in combination with PI control, voltage feedback and a gyrator model, and performing derivation; 3, popularizing the matrix to a converter cascade system to derive a state space matrix of the cascade system, and deriving; and 4, observing the stability of the cascade system through the complex plane characteristic value trajectory diagram and the characteristic value discrimination theorem, and obtaining the influence of the change of each parameter on the stability. On the basis of a traditional large-signal model, a gyrator model and a complex plane characteristic value trajectory diagram are adopted, a transfer function does not need to be calculated, a Baud diagram does not need to be drawn, and the stability change when a certain parameter changes can be directly observed.

The invention discloses a microwave transistor quasi-physical basis statistical model parameter extraction method, and relates to the technical field of electronic information-information. In order tosolve the problems in the prior art, the invention provides an implementation method for efficient parameter extraction of a device statistical model of a microwave gallium nitride high-electron-mobility transistor quasi-physical-base large-signal model. The method comprises the following steps: acquiring a large signal model parameter data set corresponding to a plurality of different GaN devicedies with the same size; and in the parameter data set, performing statistical analysis on the plurality of physical parameters and the sub-model parameters thereof, accurately characterizing association characteristics among the parameters in combination with a statistical theory of factor analysis, and finally realizing prediction of statistical distribution of device output characteristics.

The invention discloses a high-order switching power supply converter modeling method based on a signal flow diagram method. The method comprises: dividing a nonlinear converter circuit system into two linear subsystems (an ON state and an OFF state); then obtaining signal flow graphs of the two subsystems according to the flow direction and the form of the signals; associating the signal flow graphs of the two subsystems with each other; and finally, obtaining a large signal model and a small signal model of topologies (such as Sepic, Cuk, Zeta and other high-order circuits) of the switchingpower supply converter through replacement operation. The whole modeling process takes a signal flow graph as an operation object, is visual and concise, does not need complex mathematical calculation, is simple in operation, and lays a foundation for further circuit design.