CN110598849A - HMET scattering parameter extraction method and system based on neural network and storage medium - Google Patents
HMET scattering parameter extraction method and system based on neural network and storage medium Download PDFInfo
- Publication number
- CN110598849A CN110598849A CN201910788925.XA CN201910788925A CN110598849A CN 110598849 A CN110598849 A CN 110598849A CN 201910788925 A CN201910788925 A CN 201910788925A CN 110598849 A CN110598849 A CN 110598849A
- Authority
- CN
- China
- Prior art keywords
- neural network
- sample data
- test
- scattering parameter
- scattering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
Abstract
The invention discloses a neural network-based HMET scattering parameter extraction method, a system and a storage medium, wherein the extraction method comprises the following steps: acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor; extracting partial groups of training neural networks in a plurality of groups of scattering parameter sample data, wherein the neural networks comprise a first neural network and a second neural network which are separated from each other; testing the rest groups in the plurality of groups of scattering parameter sample data into a neural network; when the neural network passes the test, obtaining the test parameters of the high electron mobility transistor to be tested and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested; otherwise, continuing to train the neural network until the neural network test is passed. The invention adopts two independent neural networks to extract the scattering parameters of the high electron mobility transistor, and has the advantages of high speed, high accuracy and low cost. The method is widely applied to extraction of the scattering parameters of the high electron mobility transistor.
Description
Technical Field
The invention relates to extraction of scattering parameters of a high electron mobility transistor, in particular to a neural network-based HMET scattering parameter extraction method, system and storage medium.
Background
High electron mobility transistor (HMET): a field effect transistor utilizing two-dimensional electron gas high mobility characteristics in a heterojunction or modulated doped structure. The HMET structure herein is a modulation-doped heterojunction, and forms an electron potential well (approximately triangular) with an intrinsic semiconductor at its interface, and electrons in the potential well are two-dimensional electron gas (2-DEG) with high mobility. The electrons do not suffer from ionized impurity scattering in the potential well and, therefore, the electron mobility is high. HMET differs from a general transistor in that it has a two-dimensional electron gas characteristic that makes it have high mobility characteristics.
Two-port network: a circuit or device having 2 ports connected to a circuit internal network, capable of representing the whole or a part of the circuit by their corresponding external characteristic parameters.
Scattering parameter (S parameter): an important set of parameters in microwave transmission is used to characterize the electrical properties or performance of a radio frequency component or network.
When studying devices of the HMET structure, it is generally necessary to extract the S parameter value in order to calculate the intrinsic parameters of the semiconductor device: parasitic parameters, which are important for the design of semiconductor devices. The existing S parameter test is obtained by a network analyzer under two ports, the network analyzer inputs power or voltage to a GaAs HMET two-port device, and then tests reflected power and incident power and voltage of the two-port device, so as to calculate the S parameter value. The two-port network structure can be simply represented by the structure shown in fig. 1, and in the two-port network, a Y parameter (admittance parameter), a Z parameter (impedance parameter) are also involved, and the parameters are all useful for obtaining internal parameters of the device with the HMET structure, and Y, Z parameters can be obtained from the transformation of the S parameter, so that the S parameter is also used for measuring.
The S parameter is divided into S according to the difference of the input port and the output port corresponding to the subscript11、S22、S21And S12Four different coefficients, S11Representative is the input reflection coefficient, S22Representing the output reflection coefficient, note that at S11And S22The port is generally connected with a 50 omega resistor for port matching, S21Representing the forward propagation coefficient，S12Representing the back propagation coefficient, the S parameter is expressed in matrix form as follows:
wherein the content of the first and second substances,it is the input port that reflects the voltage,it is the output port that reflects the voltage,is the voltage incident on the input port and,is the output port incident voltage. The calculation formula of the four S parameter values is as follows:
wherein, gamma isiRepresenting the input reflection coefficient, ΓoRepresenting the output reflection coefficient.
The existing S parameter extraction method adopts a two-port network analyzer for testing, the process is complicated and complicated, the testing time is long, and the error of the testing result is possibly caused by improper operation in the experimental testing process. Meanwhile, the network analyzer is expensive and can be used only after being trained and learned by professionals, and the material cost and the labor cost are too high.
Disclosure of Invention
In view of the above, an object of the embodiments of the present invention is to provide a method, a system and a storage medium for extracting HMET scattering parameters based on a neural network. The invention adopts two independent neural networks to extract the scattering parameters of the high electron mobility transistor, and has the advantages of high speed, high accuracy and low cost.
In a first aspect, an embodiment of the present invention provides a method for extracting an HMET scattering parameter based on a neural network, including the following steps:
acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
extracting partial groups in the plurality of groups of scattering parameter sample data as training sample data to train a neural network, wherein the neural network comprises a first neural network and a second neural network which are separated from each other;
testing the neural network by taking the rest groups in the plurality of groups of scattering parameter sample data as test sample data;
when the neural network passes the test, obtaining the test parameters of the high electron mobility transistor to be tested and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
Preferably, the input parameters of the first and second neural networks are the same.
Preferably, the outputs of the first and second neural networks are the reflection coefficient and the propagation coefficient of the scattering parameter, respectively.
Preferably, the first neural network and the second neural network may vary in the number of layers and the number of neurons in the training process.
Preferably, the sample data comprises frequency band, voltage, current, frequency and scattering parameters, and the test parameters comprise frequency band, voltage, current and frequency.
Preferably, the judgment criteria that the neural network test passes are as follows: the average relative error of the output result of the test sample data and the actual value is not more than 5% and/or the mean square error is not more than 5%.
In a second aspect, an embodiment of the present invention provides a system for extracting HMET scattering parameters based on a neural network, including:
the sample data acquisition module is used for acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
the neural network training module is used for extracting partial groups in the plurality of groups of scattering parameter sample data to serve as training sample data to train a neural network, and the neural network comprises a first neural network and a second neural network which are separated from each other;
the neural network testing module is used for testing the neural network by taking the rest groups in the groups of scattering parameter sample data as test sample data;
the neural network prediction module is used for acquiring the test parameters of the high electron mobility transistor to be tested when the neural network passes the test and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
In a third aspect, an embodiment of the present invention provides an apparatus for extracting an HMET scattering parameter based on a neural network, including:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, the at least one program causes the at least one processor to implement the neural network-based HMET scattering parameter extraction method.
In a fourth aspect, the embodiment of the present invention provides a storage medium, in which processor-executable instructions are stored, and when the processor-executable instructions are executed by a processor, the method for extracting HMET scattering parameters based on a neural network is performed.
In a fifth aspect, an embodiment of the present invention provides a neural network-based HMET scattering parameter extraction system, including a server device and a computer device connected to the server device; wherein the content of the first and second substances,
the server device is used for training and testing the neural network through the data samples;
the computer device includes:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, the at least one program causes the at least one processor to implement the neural network-based HMET scattering parameter extraction method.
The implementation of the invention comprises the following beneficial effects: in the embodiment of the invention, sample data is adopted to train and test the first neural network and the second neural network which are separated from each other, and if the test result passes, the trained neural network is adopted to predict the scattering parameters of the high electron mobility transistor to be measured. The first neural network and the second neural network are adopted to separate the two components of the scattering parameter from each other, so that mutual interference and influence are avoided, the precision is improved, the calculation speed is high, and the cost is relatively low.
Drawings
FIG. 1 is a schematic diagram of a two-port network of the prior art;
FIG. 2 is a schematic diagram of a neural network according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a neural network according to an embodiment of the present invention;
fig. 4 is a schematic flowchart illustrating steps of a method for extracting HMET scattering parameters based on a neural network according to an embodiment of the present invention;
fig. 5 is a schematic diagram illustrating a principle of a neural network-based HMET scattering parameter extraction method according to an embodiment of the present invention;
FIG. 6 is a diagram illustrating the results of testing a first reflection coefficient by a first neural network according to an embodiment of the present invention;
FIG. 7 is a graph illustrating the results of testing a first propagation coefficient by a second neural network according to an embodiment of the present invention;
FIG. 8 is a diagram illustrating the results of testing a second propagation coefficient by a second neural network according to an embodiment of the present invention;
FIG. 9 is a graph illustrating the results of testing a second reflection coefficient by a first neural network according to an embodiment of the present invention;
fig. 10 is a block diagram of a system for extracting HMET scattering parameters based on a neural network according to an embodiment of the present invention;
fig. 11 is a block diagram of a structure of an HMET scattering parameter extraction apparatus based on a neural network according to an embodiment of the present invention;
fig. 12 is a block diagram of another HMET scattering parameter extraction system based on a neural network according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
The artificial neuron is made by simulating human brain nerve cells, and the neural network is formed by connecting a large number of processing units, namely neurons. Artificial neural networks are actually some kind of abstraction, simplification and simulation of brain function. To some extent he reflects some of the basic properties of the human brain. An artificial neural network is a computer algorithm that mimics the structure and function of the biological brain. The structure of the neural network is shown in fig. 2, in which the output of one neuron may be the input of another neuron, a plurality of neurons are connected to form a neural network, and the output of one neuron is determined by the input, weight, bias and excitation function. The expression is as follows:
ok＝f(netk+bk)
wherein, wkyThe weight of the y input to the k neuron; a isyIs the y input component; netkIs the weighted sum of the kth neural network, bkIs the bias of the kth neuron, f is the excitation function, okIs the calculated output of unit k. Sigmoid function commonly used for activation functions.
In the neural network, there are actually a plurality of upper basic neurons, and these neurons are connected to each other to form a layer with a plurality of neurons, and the network has a multilayer structure, that is, a multilayer perceptron network MLP (Muti-layer persistence), and a simple network structure as shown in fig. 3, in this network, the neurons are arranged in layers, which may be referred to as an input layer, a hidden layer, and an output layer from bottom to top, respectively, where a circle in the figure represents a neuron, the lowest of the neural network is input layers a1, a2, and a3, the highest is output layers b1 and b2, and the middle of the two layers is hidden layers o1, o2, and o3, and the hidden layer may be a multilayer, instead of a layer, and sometimes the number of layers and the number of neurons in the middle hidden layer are increased to improve the precision. FIG. 3 shows a 2-layer neural network comprising 3 input cells and 2 output cells.
The neural network can be widely applied because the value that people want can be obtained only by training and learning, and the neural network obtains a group of networks by learning the relation of input and output training sets. The learning process of brain is the process of the connection strength between neurons changing adaptively with the external excitation information, and the result of the brain processing information is expressed by the state of neurons. The same is true in the artificial neural network learning and training process, where a sample is input to an input neuron, a value is calculated forward, and a value is output at the final output layer, which is a forward propagation process, after the value is output, the output value is compared with a desired value and then an error is found, the error is obtained to update weights in the neural network, and the function of the neural network is mainly determined by the weights between neurons. In updating the weights, derivatives of the errors are calculated, and the weights and biases in the neural network are updated by the derivative errors, which is a back-propagation process. And in the training process of the input samples, continuously adjusting the weight of the neural network until the error between the output numerical value and the expected value meets the requirement or the maximum training number is reached. Such an algorithm in a neural network is called BP neural network. In the training process, a sample is input into an input layer, an error is calculated between an output value and an expected value, in order to reduce the error, the error is reversely propagated from the output layer to an intermediate hidden layer and then to the input layer, the weight is updated layer by layer, and therefore the error is reversely propagated. Once the neural network training is completed, the neural network can be tested, and a value different from the training sample is input to see whether the neural network function can be completed or not.
As shown in fig. 4, an embodiment of the present invention provides a method for extracting HMET scattering parameters based on a neural network, including the following steps:
s1, acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
s2, extracting partial groups in the plurality of groups of scattering parameter sample data as training sample data to train a neural network, wherein the neural network comprises a first neural network and a second neural network which are separated from each other;
s3, testing the neural network by taking the rest groups in the plurality of groups of scattering parameter sample data as test sample data;
s4, when the neural network passes the test, obtaining the test parameters of the high electron mobility transistor to be tested and inputting the test parameters into the neural network to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
Preferably, the input parameters of the first and second neural networks are the same.
Preferably, the outputs of the first and second neural networks are the reflection coefficient and the propagation coefficient of the scattering parameter, respectively.
Specifically, a plurality of sets of scattering parameter sample data of the high electron mobility transistor are acquired, wherein the scattering parameter sample data are known and are more accurate data after multiple tests and verifications, and the data comprise test parameters and test results. A large portion of the sample data is selected for training, e.g., 80% of the sample data is used for training the neural network, and a small portion is used for testing the trained neural network, e.g., 20%. The more sample data used for training and testing, the more accurate the neural network prediction result is. If the result of the sample data used for testing the neural network can meet the requirement of practical application, namely within an allowable error range, the trained neural network can be used for the scattering parameters of the high electron mobility transistor to be tested. The scattering parameters of the high electron mobility transistor are composed of two parts, namely a reflection coefficient and a propagation coefficient, and the reflection coefficient and the propagation coefficient are respectively predicted by adopting a first neural network and a second neural network which are separated from each other.
The implementation of the invention comprises the following beneficial effects: in the embodiment of the invention, sample data is adopted to train and test the first neural network and the second neural network which are separated from each other, and if the test result passes, the trained neural network is adopted to predict the scattering parameters of the high electron mobility transistor to be measured. The first neural network and the second neural network are adopted to separate the two components of the scattering parameter from each other, so that mutual interference and influence are avoided, the precision is improved, the calculation speed is high, and the cost is relatively low.
Preferably, the first neural network and the second neural network may vary in the number of layers and the number of neurons in the training process. In the training process of the neural network, the number of layers of the neural network and the number of neurons are properly adjusted according to the training result condition, so that the training and testing results are more ideal, and the number of layers of the first neural network and the number of neurons of the second neural network can be the same or different.
Preferably, the sample data comprises frequency band, voltage, current, frequency and scattering parameters, and the test parameters comprise frequency band, voltage, current and frequency. In the embodiment of the invention, the input parameters of the first neural network and the second neural network are as follows: frequency band, voltage, current and frequency, the output parameter is the scattering parameter, is used for testing the scattering parameter under different frequency bands.
Preferably, the judgment criteria that the neural network test passes are as follows: the average relative error of the output result of the test sample data and the actual value is not more than 5% and/or the mean square error is not more than 5%. One of the average relative error and the mean square error can meet the requirement, and if the two requirements are met at the same time, the more accurate the prediction result of the neural network is.
In an alternative embodiment, shown in FIG. 5, the first neural network and the second neural network are separated from each other, and the input parameters are all frequency bands FbMains voltage VddSupply current IddThe frequency f is output as the amplitude Mag and the phase angle Ang of the scattering parameter, and the output of the first neural network is the amplitude and the phase angle of the reflection coefficient: mag (S)11)、Ang(S11)、Mag(S22) And Ang (S)22) The second neural network output is of propagation coefficients: mag (S)21)、Ang(S21)、Mag(S12) And Ang (S)12). The training, testing and prediction of the neural network are completed in a remalab development environment, and are shown as follows:
Mag(sij)＝F1[Fb，Vdd，Idd，f]
Ang(sij)＝F1[Fb，Vdd，Idd，f]
wherein, the values of i and j are both 1 or 2. The method comprises the steps of selecting datasheet data of 3 series products of an HEMT of a certain company as sample data to train and test a neural network, wherein the frequency band data of the training samples of the 3 series products are respectively 450 plus 1450MHz, 1440 plus 2350MHz and 1950 plus 2700MHz, the power supply voltage is 4.8V, the power supply current is respectively 35mA, 55mA and 60mA, and the frequency values corresponding to each group of frequency bands, power supply voltage and power supply current are unequally spaced values between 0.1GHz and 19 GHz. Predicting samples as bands 1950-2700MHz, 4.8V of power supply voltage, 75mA of power supply current and unequal interval values between 0.1GHz and 19GHz frequency. In the training process, the number of layers of the first neural network and the second neural network and the number of neurons are tested for multiple times, the neural network structure of the three hidden layers is found to be the best, and after test parameters are input from the input layer, the test parameters pass through the three hidden layers with the number of neurons being 8-6-6 and are output from the output layer. The output result of the test sample is shown in fig. 6, where the solid line is actual datasheet data of the product, the asterisks and circles are data obtained by fitting the trained neural network test, and as can be seen from fig. 6, 7, 8 and 9, the data amplitude and phase angle calculated by fitting the neural network are as follows: mag (S)11)、Ang(S11)、Mag(S22)、Ang(S22)、Mag(S21)、Ang(S21)、Mag(S12) And Ang (S)12) The method is basically consistent with datasheet data, is well consistent at the corner point and has small error.
As shown in fig. 10, an embodiment of the present invention provides a system for extracting HMET scattering parameters based on a neural network, including:
the sample data acquisition module is used for acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
the neural network training module is used for extracting partial groups in the plurality of groups of scattering parameter sample data to serve as training sample data to train a neural network, and the neural network comprises a first neural network and a second neural network which are separated from each other;
the neural network testing module is used for testing the neural network by taking the rest groups in the groups of scattering parameter sample data as test sample data;
the neural network prediction module is used for acquiring the test parameters of the high electron mobility transistor to be tested when the neural network passes the test and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
It can be seen that the contents in the foregoing method embodiments are all applicable to this system embodiment, the functions specifically implemented by this system embodiment are the same as those in the foregoing method embodiment, and the advantageous effects achieved by this system embodiment are also the same as those achieved by the foregoing method embodiment.
As shown in fig. 11, an embodiment of the present invention provides a neural network-based HMET scattering parameter extraction apparatus, including:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, the at least one program causes the at least one processor to implement the neural network-based HMET scattering parameter extraction method.
It can be seen that the contents in the foregoing method embodiments are all applicable to this apparatus embodiment, the functions specifically implemented by this apparatus embodiment are the same as those in the foregoing method embodiment, and the advantageous effects achieved by this apparatus embodiment are also the same as those achieved by the foregoing method embodiment.
Furthermore, a storage medium is provided, in which processor-executable instructions are stored, and when executed by a processor, the processor-executable instructions are configured to perform the method for extracting HMET scattering parameters based on a neural network. Likewise, the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as those of the above method embodiments, and the advantageous effects achieved by the present storage medium embodiment are also the same as those achieved by the above method embodiments.
As shown in fig. 12, an embodiment of the present invention provides a neural network-based HMET scattering parameter extraction system, which includes a server device and a computer device connected to the server device; wherein the content of the first and second substances,
the server device is used for training and testing the neural network through the data samples;
the computer device includes:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, the at least one program causes the at least one processor to implement the neural network-based HMET scattering parameter extraction method.
Specifically, the computer device may be different types of electronic devices, including but not limited to a desktop computer, a laptop computer, and other terminals.
It can be seen that the contents in the foregoing method embodiments are all applicable to this system embodiment, the functions specifically implemented by this system embodiment are the same as those in the foregoing method embodiment, and the advantageous effects achieved by this system embodiment are also the same as those achieved by the foregoing method embodiment.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A HMET scattering parameter extraction method based on a neural network is characterized by comprising the following steps:
acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
extracting partial groups in the plurality of groups of scattering parameter sample data as training sample data to train a neural network, wherein the neural network comprises a first neural network and a second neural network which are separated from each other;
testing the neural network by taking the rest groups in the plurality of groups of scattering parameter sample data as test sample data;
when the neural network passes the test, obtaining the test parameters of the high electron mobility transistor to be tested and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
2. The method of claim 1, wherein the input parameters of the first neural network and the second neural network are the same.
3. The method of claim 2, wherein the outputs of the first and second neural networks are the reflection coefficient and propagation coefficient of the scattering parameter, respectively.
4. The method of claim 3, wherein the number of layers and the number of neurons in the training process of the first neural network and the second neural network can be changed.
5. The method according to any one of claims 1-4, wherein the sample data comprises frequency band, voltage, current, frequency and scattering parameters, and the test parameters comprise frequency band, voltage, current and frequency.
6. The method for extracting HMET scattering parameters based on neural network as claimed in claim 5, wherein the judgment criteria that the neural network test passes are: the average relative error of the output result of the test sample data and the actual value is not more than 5% and/or the mean square error is not more than 5%.
7. A HMET scattering parameter extraction system based on a neural network is characterized by comprising:
the sample data acquisition module is used for acquiring a plurality of groups of scattering parameter sample data of the high electron mobility transistor;
the neural network training module is used for extracting partial groups in the plurality of groups of scattering parameter sample data to serve as training sample data to train a neural network, and the neural network comprises a first neural network and a second neural network which are separated from each other;
the neural network testing module is used for testing the neural network by taking the rest groups in the groups of scattering parameter sample data as test sample data;
the neural network prediction module is used for acquiring the test parameters of the high electron mobility transistor to be tested when the neural network passes the test and inputting the test parameters into the neural network so as to output the scattering parameters of the high electron mobility transistor to be tested;
otherwise, continuing to train the neural network until the neural network passes the test.
8. An HMET scattering parameter extraction device based on a neural network is characterized by comprising:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to implement the neural network-based HMET scattering parameter extraction method of any of claims 1-6.
9. A storage medium having stored therein processor-executable instructions, which when executed by a processor, are configured to perform the neural network-based HMET scattering parameter extraction method of any one of claims 1-6.
10. The HMET scattering parameter extraction system based on the neural network is characterized by comprising server equipment and computer equipment connected with the server equipment; wherein the content of the first and second substances,
the server device is used for training and testing the neural network through the data samples;
the computer device includes:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to implement the neural network-based HMET scattering parameter extraction method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910788925.XA CN110598849A (en) | 2019-08-26 | 2019-08-26 | HMET scattering parameter extraction method and system based on neural network and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910788925.XA CN110598849A (en) | 2019-08-26 | 2019-08-26 | HMET scattering parameter extraction method and system based on neural network and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110598849A true CN110598849A (en) | 2019-12-20 |
Family
ID=68855594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910788925.XA Pending CN110598849A (en) | 2019-08-26 | 2019-08-26 | HMET scattering parameter extraction method and system based on neural network and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110598849A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111880003A (en) * | 2020-08-06 | 2020-11-03 | 广州大学 | Radio frequency power amplifier harmonic balance parameter extraction method based on neural network |
| CN111898320A (en) * | 2020-08-06 | 2020-11-06 | 广州大学 | Radio frequency power amplifier scattering parameter extraction method and device based on neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004186258A (en) * | 2002-11-29 | 2004-07-02 | Fujitsu Ltd | Method and program for extracting parameter |
| CN108664924A (en) * | 2018-05-10 | 2018-10-16 | 东南大学 | A kind of multi-tag object identification method based on convolutional neural networks |
-
2019
- 2019-08-26 CN CN201910788925.XA patent/CN110598849A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004186258A (en) * | 2002-11-29 | 2004-07-02 | Fujitsu Ltd | Method and program for extracting parameter |
| CN108664924A (en) * | 2018-05-10 | 2018-10-16 | 东南大学 | A kind of multi-tag object identification method based on convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| 王庆娜: "《GaN HEMT毫米波器件及建模》", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111880003A (en) * | 2020-08-06 | 2020-11-03 | 广州大学 | Radio frequency power amplifier harmonic balance parameter extraction method based on neural network |
| CN111898320A (en) * | 2020-08-06 | 2020-11-06 | 广州大学 | Radio frequency power amplifier scattering parameter extraction method and device based on neural network |
| CN111880003B (en) * | 2020-08-06 | 2023-08-11 | 广州大学 | Method for extracting harmonic balance parameters of radio frequency power amplifier based on neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Devabhaktuni et al. | Neural networks for microwave modeling: Model development issues and nonlinear modeling techniques | |
| Patnaik et al. | ANN techniques in microwave engineering | |
| Marinković et al. | Temperature‐dependent models of low‐noise microwave transistors based on neural networks | |
| CN101299055B (en) | Simulation integrated switch current circuit testing method based on wavelet-neural net | |
| CN112485652B (en) | Analog circuit single fault diagnosis method based on improved sine and cosine algorithm | |
| Tan et al. | Analog fault diagnosis using S-transform preprocessor and a QNN classifier | |
| CN110598849A (en) | HMET scattering parameter extraction method and system based on neural network and storage medium | |
| Baumann et al. | Impulse test fault diagnosis on power transformers using Kohonen's self-organizing neural network | |
| Goay et al. | Eye-height/width prediction using artificial neural networks from S-parameters with vector fitting | |
| CN113533945A (en) | Analog circuit fault diagnosis method based on two-dimensional convolutional neural network | |
| Litovski et al. | Analogue electronic circuit diagnosis based on ANNs | |
| CN114462292A (en) | Circuit parameter optimization method based on SQP algorithm | |
| Geng et al. | Accurate and effective nonlinear behavioral modeling of a 10-w gan hemt based on lstm neural networks | |
| CN111898320A (en) | Radio frequency power amplifier scattering parameter extraction method and device based on neural network | |
| Marković et al. | HEMT noise neural model based on bias conditions | |
| Shirong et al. | Wavelet neural network based fault diagnosis in nonlinear analog circuits | |
| CN115308552A (en) | Method and system for calculating wave head and wave tail resistance values of impulse voltage test device | |
| Werle et al. | Partial discharge localisation on power transformers using neural networks combined with sectional winding transfer functions as knowledge base | |
| Kashyap et al. | A novel use of deep learning to optimize solution space exploration for signal integrity analysis | |
| Zhang et al. | LSSVM parameters optimizing and non-linear system prediction based on cross validation | |
| Guerreiro et al. | Fault list compression for efficient analogue and mixed-signal production test preparation | |
| Shirong | Application of BP neural network in analog circuits diagnosis | |
| Buhry et al. | Automated tuning of analog neuromimetic integrated circuits | |
| Shuai et al. | A hybrid electronic components detection parameter sequence optimization method based on random forest and mean decrease accuracy | |
| Litovski et al. | Electronic modelling using ANNs for analogue and mixed-mode behavioural simulation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191220 |
|
| RJ01 | Rejection of invention patent application after publication |