CN112883670B - Inductance automatic design comprehensive model and method based on artificial neural network - Google Patents
Inductance automatic design comprehensive model and method based on artificial neural network Download PDFInfo
- Publication number
- CN112883670B CN112883670B CN202110285875.0A CN202110285875A CN112883670B CN 112883670 B CN112883670 B CN 112883670B CN 202110285875 A CN202110285875 A CN 202110285875A CN 112883670 B CN112883670 B CN 112883670B
- Authority
- CN
- China
- Prior art keywords
- inductor
- inductance
- module
- self
- parameters
- 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.)
- Expired - Fee Related
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 46
- 238000013461 design Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims description 28
- 238000004458 analytical method Methods 0.000 claims description 44
- 238000012549 training Methods 0.000 claims description 27
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000015572 biosynthetic process Effects 0.000 claims description 7
- 238000003786 synthesis reaction Methods 0.000 claims description 7
- 239000002184 metal Substances 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims 4
- 238000012795 verification Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 8
- 238000004088 simulation Methods 0.000 description 7
- 238000010606 normalization Methods 0.000 description 5
- 239000013598 vector Substances 0.000 description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000012938 design process Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 210000004205 output neuron Anatomy 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000012351 Integrated analysis Methods 0.000 description 2
- 230000005672 electromagnetic field Effects 0.000 description 2
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/32—Circuit design at the digital level
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2115/00—Details relating to the type of the circuit
- G06F2115/10—Processors
Abstract
The invention provides a comprehensive model for on-chip inductance automatic design based on an artificial neural network, and belongs to the field of electronic communication. The comprehensive model judges whether the given required parameters are within the design capability range of the model aiming at the inductance value, the quality factor, the self-resonant frequency, the outer diameter and the frequency point needing to work of the given required inductor, if the given required parameters are judged to be 'yes', an inductor layout can be generated, the inductance value, the quality factor and the scattering parameters of the generated inductor can be predicted at the same time, finally, a self-checking module in the model compares the predicted inductance value and the quality factor with the required inductance value and the required quality factor, and if the inductance value and the quality factor accord with the design requirements, the current inductor layout and the predicted S parameters are output; otherwise, outputting error prompt information and error percentages of all indexes.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence, and particularly relates to an on-chip inductance automatic design comprehensive model and method based on an artificial neural network.
Background
On-chip inductors are a significant concern in the design of radio frequency integrated circuits. From the perspective of the integrated circuit designer, several issues of most concern in the on-chip inductor design process are: the accuracy of the inductance and quality factor (Q value), the model accuracy, and the time it takes to obtain the proper layout dimensions and layout. The above problems will seriously affect the design cycle and even the success rate of chip flow. There is a great deal of effort to achieve a suitable on-chip inductance for a particular circuit portion (e.g., matching network, load, etc.). In most cases, the process design library provided by the foundry does not include the inductor model, and the inductor model in the process design library of the inductor model is usually applicable to a limited frequency band and limited model accuracy, which increases the difficulty of inductor design.
A design flow chart of an existing inductor is shown in fig. 1, a chip designer needs to draw a preliminary inductor layout, perform electromagnetic simulation (EM simulation) on the inductor, and extract characteristic parameters of the drawn inductor, such as: sensing value (L), quality factor (Q value) and S parameter, and judging whether the characteristic parameter of the drawn inductor meets the expectation, if not, re-optimizing the inductor layout; and if the expectation is met, finishing the inductor layout design. However, the metal layer of the silicon-based process is complex, and performing the EM simulation consumes a lot of time; meanwhile, the inductor design usually needs to be adjusted continuously to meet the circuit requirements, which means that multiple rounds of EM simulation are required, which consumes a lot of effort and time of designers.
Therefore, how to rapidly and accurately design the on-chip inductor automatically becomes one of the important directions of current research.
Disclosure of Invention
Aiming at the problems in the background art, the invention aims to provide an inductance automatic design comprehensive model and method based on an artificial neural network. The comprehensive model generates an inductance layout aiming at the inductance value, the quality factor, the Self-Resonance Frequency (SRF), the outer diameter (D) and the Frequency point needing to work of a given required inductance, and simultaneously predicts the inductance value, the quality factor and the scattering parameter (S parameter) of the generated inductance, a Self-checking module in the model compares the predicted inductance value and the quality factor with the required inductance value and the required quality factor, and if the inductance value and the quality factor accord with the design requirement, the generated inductance layout and the predicted S parameter are output; otherwise, outputting error prompt information and error percentages of all indexes.
In order to achieve the purpose, the technical scheme of the invention is as follows:
an inductance automatic design comprehensive model based on an artificial neural network comprises a comprehensive module, an analysis module, a self-checking module and an output selection module;
the comprehensive module processes the input required inductor working frequency range, inductor outer diameter and performance parameters to generate an inductor layout, transmits the inductor layout to the output selection module, and respectively transmits the predicted geometric parameters of the layout to the analysis module;
The analysis module generates performance parameters of the predicted inductor, such as inductance value L ', quality factor Q ', self-resonant frequency SRF ' and scattering parameters, according to the input required frequency range and the predicted geometric parameters input by the synthesis module, transmits the predicted inductance value L ', quality factor Q ' and scattering parameters of the inductor to the self-checking module, and inputs the predicted performance parameters into the output selection module;
the self-checking module judges according to the input inductance performance parameters and the predicted inductance performance parameters transmitted by the analysis module to determine whether the generated inductance layout meets the design requirements;
the output selection module is used for selectively outputting the layout and the performance parameters or error prompt information according to the judgment result of the self-checking module.
Further, the performance parameters of the inductor are specifically: inductance (L), quality factor (Q), self-resonant frequency (SRF), and scattering parameters, where L affects the matching network, Q affects circuit loss and noise, and SRF affects stability.
Further, the geometric parameters of the inductor are specifically: the inductor comprises a metal wire width W, a metal wire edge distance S, an inductor inner diameter d and an inductor turn number N.
Further, the method for designing the inductor based on the comprehensive model of the on-chip inductor automatic design comprises the following steps of:
step 1, inputting a required working frequency range, an inductor outer diameter and performance parameters into a comprehensive module and a self-checking module, and simultaneously inputting the required working frequency range into an analysis module;
step 3, the analysis module obtains the inductance value L ', the quality factor Q ' and the self-resonant frequency SRF ' of the inductor generated in the step 2 by using the trained first analysis artificial neural network according to the required working frequency range and the geometric parameters input in the step 2, obtains the real part and the imaginary part of the scattering parameters of the generated inductor layout by using the trained second analysis artificial neural network, further obtains the amplitude and the phase of the scattering parameters by using a calculation formula, simultaneously transmits the amplitude and the phase of the scattering parameters, the inductance value L ' and the quality factor Q ' of the generated inductor to the output selection module, and transmits the inductance value, the quality factor and the self-resonant frequency of the generated inductor layout to the self-detection module;
step 5, the output selection module outputs the current inductance layout and the performance parameters thereof according to the judgment result of the step 4 if the judgment result meets the requirements; if not, outputting error prompt information and error percentages of each index.
Further, the calculation formula in step 3 is specifically:
real is the real part of the scattering parameter, image is the imaginary part of the scattering parameter, A is the amplitude of the scattering parameter obtained by calculation, and theta is the phase of the scattering parameter obtained by calculation.
Furthermore, the training process of the artificial neural network of the synthesis module and the training process of the artificial neural network analyzed by the modeling module are respectively trained in advance, no mutual connection exists, and no mutual influence exists in the training effect.
Further, the specified value in step 4 is preferably 10%.
Further, the designer can adjust the design requirement according to the error percentage of the index output in step 4.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. according to the invention, an artificial neural network is innovatively introduced in the traditional on-chip inductor design process, so that the required inductor parameters are directly input, the final inductor layout meeting the requirements and the S parameters of the inductor layout are automatically generated, the inductor layout does not need to be continuously optimized and adjusted, the time of a chip designer is greatly saved, and the design process is rapidly promoted.
2. According to the comprehensive analysis model for the automatic design of the on-chip inductor, the output error layout is avoided through the setting of the self-checking module, and the accuracy of the final output on-chip inductor is realized.
Drawings
Fig. 1 is a flow chart of a prior art on-chip inductor design.
Fig. 2 is a schematic structural diagram of a comprehensive analysis model of the on-chip inductor automation design of the present invention.
Fig. 3 is a schematic structural diagram of the spiral inductor obtained by a comprehensive analysis model based on the on-chip inductor automation design in embodiment 1.
Fig. 4 is a verification result of an analysis module of the comprehensive analysis model of the on-chip inductor automation design of the present invention.
Fig. 5 is a comprehensive module verification result of the comprehensive analysis model of the on-chip inductor automation design of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
Fig. 2 is a schematic structural diagram of an integrated analysis model of the on-chip inductor automation design of the present invention. A comprehensive model for on-chip inductance automatic design based on an artificial neural network comprises a comprehensive module, an analysis module, a self-checking module and an output selection module;
the comprehensive module is connected with the analysis module and the output selection module, and predicts geometric parameters of the inductance by utilizing an Artificial Neural Network (ANN);
the analysis module is connected with the self-checking module and the output selection module and comprises two analysis artificial neural networks, wherein one analysis artificial neural network is used for predicting the inductance value and the quality factor of the inductor, and the other analysis artificial neural network is used for predicting the scattering parameter of the inductor;
the self-checking module is connected with the output selection module and is used for checking the generated inductance layout and judging whether the design requirements are met;
the output selection module is used for selectively outputting the layout and the parameters or error prompt information according to the judgment result of the self-checking module.
Example 1
The spiral inductor is designed by adopting the comprehensive model for the automatic design of the on-chip inductor, and the specific process is as follows:
step 1, inputting a required working frequency range, an inductor outer diameter (D) and performance parameters to a comprehensive module and a self-checking module, and simultaneously transmitting the required working frequency range to an analysis module, wherein the performance parameters comprise an inductance value (L), a quality factor (Q) and a self-resonant frequency (SRF); l affects the matching network, Q affects the circuit loss and noise, D determines the area of the inductor, and SRF affects the stability of the inductor;
step A: performing electromagnetic field simulation on different inductor layouts to obtain S parameters and Y parameters of the inductor;
and B: extracting inductance value and Q value of the inductor from the Y parameter as training sample and test sample, and performing normalization processing to obtain training sample normalization data and test sample normalization data, wherein the extraction process is as follows,
Wherein, L is inductance of the inductor, Q is quality factor of the inductor, Im represents imaginary part, Re represents real part, Y 11 Is the first element of the Y parameter matrix, and f is the frequency;
and C: and B, establishing a primary artificial neural network model and training by using a neural network tool according to the training sample normalized data obtained in the step B, wherein the specific process is as follows:
step C1, training sample normalization data is imported, and input vectors (inductance characteristic parameters) and expected output vectors (inductance layout geometric parameters) of the artificial neural network are obtained
Step C2, setting a training error allowable value e, and initializing each weight value and threshold value;
step C3, calculating the output of each hidden layer and output layer node;
step C4, calculating an error index function E between the actual output and the expected output of the primary artificial neural network, and comparing the error index function E with the training error allowable value E;
step C5, if E is less than or equal to E, finishing training and outputting a training result;
step C6, if E is larger than E, updating the weight value and the threshold value, judging whether the preset training times are reached, and if the preset training times are reached, stopping training;
c7, if the preset training times are not reached, jumping to the step C3 to continue training;
Step D: verifying the primary artificial neural network model in the step C according to the verification sample normalization data obtained in the step B to obtain an output result of the primary artificial neural network model, comparing a verification error between the output result and an expected output vector, and judging whether the preset expected precision is met; if the verification result meets the preset expected precision, determining the primary artificial neural network model as a final artificial neural network model, namely the comprehensive artificial neural network required to be trained; if the verification result does not accord with the preset expected precision, correcting the primary artificial neural network model, and skipping to the step C;
when the artificial neural network is trained, 4 layers of hidden layers, 4 output neurons, a ReLU activation function, an MSE Loss function and an Adam optimizer are adopted for training, the learning rate is 0.0001, and the iteration is carried out for 2000 rounds;
step 3, the analysis module obtains the inductance value L ', the quality factor Q ' and the self-resonant frequency SRF ' of the inductor generated in the step 2 by using the trained first analysis artificial neural network (I) according to the required working frequency range and the geometric parameters input in the step 2, obtains the real part and the imaginary part of the scattering parameter of the generated inductor by using the trained second analysis artificial neural network (II), further obtains the amplitude and the phase of the scattering parameter by using a calculation formula, and simultaneously transmits the amplitude and the phase of the scattering parameter, the inductance value L ' and the quality factor Q ' of the generated inductor to the output selection module;
The first analysis artificial neural network and the second analysis artificial neural network adopt the same training process, the training process is basically the same as that of the comprehensive artificial neural network, only the input vector and the expected output vector are adjusted into the geometric parameters and the performance parameters of the inductance layout, the artificial neural network I is analyzed to obtain L and Q, and the artificial neural network II is analyzed to obtain S parameters;
when the artificial neural network I training is analyzed, training is carried out by adopting 4 layers of hidden layers, 2 output neurons, a Tanh activation function, an MSE Loss function and an Adam optimizer, the learning rate is 0.0001, and 3000 iterations are carried out; when training of the artificial neural network II is analyzed, training is carried out by adopting 4 layers of hidden layers, 4 output neurons, a Tanh activation function, an MSE Loss function and an Adam optimizer, the learning rate is 0.0001, and 3000 iterations are carried out;
Step 5, the output selection module outputs the current inductance layout and the performance parameters (including L, Q, SRF, D and S parameters) thereof if the judgment result in the step 4 meets the requirements; if not, outputting error prompt information and error percentages of each index.
The final structure of the spiral inductor designed in this embodiment is shown in fig. 3, and by using the comprehensive analysis model provided by the present invention, an inductor layout similar to that in fig. 3 can be finally obtained.
Table 1 shows inductance test verification data, which includes 7 sets of data, named as a, b, c, d, e, f, and g. The left data is the inductor design requirement; the right data is the corresponding inductance geometric parameters (W, S, T, D) obtained by inputting each set of data (L, Q, SRF, D) of the design requirements into the comprehensive analysis model.
TABLE 1
FIG. 4 shows the verification result of the analysis module of the integrated model of the present invention. Wherein, the solid line is the result of EM simulation of the inductor layout drawn according to the geometric parameters of the inductor given by the comprehensive analysis model; the dotted line is the inductance characteristic predicted by the integrated analysis model at the frequency point with the frequency interval of 0.5GHz within the frequency band of 1GHz to 10GHz, such as inductance value, quality factor and S parameter (including S11 amplitude, S21 amplitude, S11 phase and S21 phase). From the contact ratio of the solid line and the dotted line in the figure, the comprehensive analysis model can accurately analyze the characteristics of the inductor.
FIG. 5 is a verification result of the synthesis module. Where the square symbols are the inductance design requirements on the left side of the table. The circular mark is the characteristic of the inductance layout output by the synthesis module, the inductance layout is given by the synthesis module according to the design requirement, and the characteristic parameter is the result given by the electromagnetic field simulation. As can be seen from fig. 5, the three performance parameter indexes of the inductor (inductance value, quality factor and self-resonant frequency) and the fitting degree of the circular mark and the square mark of the layout parameter inductor outer diameter are high, which indicates that the synthesis module can relatively provide the inductor layout.
While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.
Claims (7)
1. A method for designing an inductance based on an inductance automatic design comprehensive model of an artificial neural network is characterized in that the inductance automatic design comprehensive model comprises a comprehensive module, an analysis module, a self-checking module and an output selection module;
The comprehensive module processes the input required inductor working frequency range, the inductor outer diameter and the performance parameters to obtain the geometric parameters of the inductor, generates an inductor layout according to the geometric parameters, and simultaneously transmits the inductor layout to the output selection module and transmits the geometric parameters to the analysis module;
the analysis module generates a predicted performance parameter of the inductor according to the input required inductor working frequency range and the geometric parameters input by the synthesis module, transmits the inductance value, the quality factor and the self-resonance frequency of the inductor in the predicted performance parameter to the self-checking module, and transmits the predicted performance parameter to the output selection module;
the self-checking module judges according to the input inductance performance parameters and the predicted inductance performance parameters obtained by the analysis module to determine whether the generated inductance layout meets the design requirements;
the output selection module selects an output inductance layout and a predicted performance parameter or error prompt message according to the judgment result of the self-checking module;
the method for designing the inductor comprises the following steps:
step 1, inputting a required working frequency range, an inductance outer diameter and performance parameters to a comprehensive module and a self-checking module, and simultaneously inputting the required working frequency range to an analysis module;
Step 2, the comprehensive module predicts the geometric parameters of the inductor by using the trained comprehensive artificial neural network, generates an inductor layout by using a generating tool according to the predicted geometric parameters, transmits the geometric parameters to the analysis module, and transmits the inductor layout to the output selection module;
step 3, the analysis module obtains the inductance value, the quality factor and the self-resonant frequency of the inductance layout generated in the step 2 by utilizing the trained first analysis artificial neural network according to the required working frequency range and the geometric parameters input in the step 2, obtains the real part and the imaginary part of the scattering parameters of the generated inductance layout by utilizing the trained second analysis artificial neural network, further obtains the amplitude and the phase of the scattering parameters by utilizing a calculation formula, simultaneously transmits the amplitude and the phase of the scattering parameters and the inductance value and the quality factor of the generated inductance to the output selection module, and transmits the inductance value, the quality factor and the self-resonant frequency of the generated inductance layout to the self-detection module;
step 4, the self-checking module compares the inductance value, the quality factor and the self-resonant frequency of the generated inductance layout obtained in the step 3 with the inductance value, the quality factor and the self-resonant frequency input in the step 1, judges the error between each characteristic of the generated inductance and the required characteristic of the inductance, and judges that the generated inductance meets the requirement if the error is smaller than a specified value; if not, the judgment result is transmitted to the output selection module;
Step 5, the output selection module outputs the current inductance layout and the performance parameters thereof according to the judgment result of the step 4 if the judgment result meets the requirements; if not, outputting error prompt information and error percentages of all indexes.
2. The method for designing the inductor by using the integrated model for the automated inductor design as claimed in claim 1, wherein the performance parameters of the inductor are specifically as follows: inductance, quality factor, self-resonant frequency, and scattering parameters.
3. The method for designing the inductor by using the integrated model for the automated inductor design as claimed in claim 1, wherein the geometric parameters of the inductor are specifically as follows: the inductor comprises a metal wire width W, a metal wire edge distance S, an inductor inner diameter d and an inductor turn number N.
4. The method for designing the inductor by using the integrated model for the automated inductor design as claimed in claim 1, wherein the calculation formula in the step 3 is specifically as follows:
real is the real part of the scattering parameter, image is the imaginary part of the scattering parameter, A is the amplitude of the scattering parameter obtained by calculation, and theta is the phase of the scattering parameter obtained by calculation.
5. The method for designing an inductor by using the automatic inductor design integration model as claimed in claim 1, wherein the training processes of the integration artificial neural network of the integration module and the analysis artificial neural network of the analysis module are respectively trained in advance, and there is no mutual connection and no mutual influence on the training effect.
6. The method for designing an inductor according to the automated inductor design integration model of claim 1, wherein the specified value in step 4 is 10%.
7. The method of claim 1, wherein the designer adjusts the design requirement according to the error percentage of each index outputted in step 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110285875.0A CN112883670B (en) | 2021-03-17 | 2021-03-17 | Inductance automatic design comprehensive model and method based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110285875.0A CN112883670B (en) | 2021-03-17 | 2021-03-17 | Inductance automatic design comprehensive model and method based on artificial neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112883670A CN112883670A (en) | 2021-06-01 |
| CN112883670B true CN112883670B (en) | 2022-07-29 |
Family
ID=76041058
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110285875.0A Expired - Fee Related CN112883670B (en) | 2021-03-17 | 2021-03-17 | Inductance automatic design comprehensive model and method based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112883670B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113687995B (en) * | 2021-10-27 | 2022-01-04 | 成都嘉纳海威科技有限责任公司 | Chip screening method based on neural network |
| CN116205195B (en) * | 2023-05-05 | 2023-07-07 | 南京邮电大学 | Method, device, equipment and medium for acquiring parameters of planar spiral inductor |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354611B (en) * | 2015-10-08 | 2018-01-09 | 程涛 | A kind of best quality image scan method and system based on artificial neural network |
| US10255909B2 (en) * | 2017-06-29 | 2019-04-09 | Intel IP Corporation | Statistical-analysis-based reset of recurrent neural networks for automatic speech recognition |
| CN108170950B (en) * | 2017-12-27 | 2021-08-06 | 电子科技大学 | Multilayer frequency selection surface wave-absorbing material modeling optimization method based on neural network |
| CN108959728B (en) * | 2018-06-12 | 2023-04-07 | 杭州法动科技有限公司 | Radio frequency device parameter optimization method based on deep learning |
| CN109117938A (en) * | 2018-10-25 | 2019-01-01 | 西南林业大学 | A kind of image scan method and system based on artificial neural network |
| CN110069805A (en) * | 2019-01-18 | 2019-07-30 | 湖南大学 | Spiral inductance optimization method based on HPSO and GA-BP algorithm |
| CN112072735A (en) * | 2020-08-28 | 2020-12-11 | 广东电网有限责任公司广州供电局 | Equalization management system of liquid metal battery pack based on convolutional neural network |
-
2021
- 2021-03-17 CN CN202110285875.0A patent/CN112883670B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN112883670A (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112883670B (en) | Inductance automatic design comprehensive model and method based on artificial neural network | |
| Crombecq | Surrogate modeling of computer experiments with sequential experimental design | |
| CN104122796B (en) | A kind of Intelligent assembly Sequence Planning method | |
| Canelas et al. | FUZYE: A Fuzzy ${c} $-Means Analog IC Yield Optimization Using Evolutionary-Based Algorithms | |
| CN113987984A (en) | On-chip inductor automatic design system and design method | |
| CN108009481A (en) | A kind of training method and device of CNN models, face identification method and device | |
| CN115809624B (en) | Automatic analysis design method for integrated circuit microstrip line transmission line | |
| CN108182316B (en) | Electromagnetic simulation method based on artificial intelligence and electromagnetic brain thereof | |
| CN107564007B (en) | Scene segmentation correction method and system fusing global information | |
| CN102508977B (en) | Circuit optimization method for artificial circuit transplant and circuit optimization device for artificial circuit transplant | |
| CN110765704A (en) | Novel automatic deep neural network modeling method applied to microwave device | |
| CN113988558B (en) | Power grid dynamic security assessment method based on blind area identification and electric coordinate system expansion | |
| CN110688723B (en) | Rapid design method for clock distribution network | |
| CN115906303A (en) | Planar microwave filter design method and device based on machine learning | |
| CN115358182A (en) | Chip wiring system and method based on generative deep learning | |
| Fan et al. | From specification to topology: Automatic power converter design via reinforcement learning | |
| CN113962163A (en) | Optimization method, device and equipment for realizing efficient design of passive microwave device | |
| CN112199918B (en) | Method for reconstructing physical connection relation of general EDA model layout | |
| CN105302968B (en) | A kind of distributed power amplifier optimum design method | |
| CN101477582B (en) | Model modification method for a semiconductor device | |
| CN111027642A (en) | Electric energy meter quality monitoring method | |
| CN111651846B (en) | Automatic optimization method for pipeline design of refrigeration equipment | |
| Wang et al. | A synthesis-analysis machine with self-inspection mechanism for automatic design of on-chip inductors based on artificial neural networks | |
| CN117669300A (en) | Multi-objective optimization design method and device for wireless power transmission system and storage medium | |
| CN103164565A (en) | Method for automatically forming antenna regular test vectors |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220729 |