CN110070170A - PSO-BP neural network sensor calibrating system and method based on MCU - Google Patents
PSO-BP neural network sensor calibrating system and method based on MCU Download PDFInfo
- Publication number
- CN110070170A CN110070170A CN201910432693.4A CN201910432693A CN110070170A CN 110070170 A CN110070170 A CN 110070170A CN 201910432693 A CN201910432693 A CN 201910432693A CN 110070170 A CN110070170 A CN 110070170A
- Authority
- CN
- China
- Prior art keywords
- neural network
- pso
- mcu
- calibration
- module
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims description 20
- 238000004891 communication Methods 0.000 claims abstract description 21
- 230000006870 function Effects 0.000 claims description 18
- 230000004913 activation Effects 0.000 claims description 10
- 238000012549 training Methods 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000012886 linear function Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 2
- 238000012937 correction Methods 0.000 abstract description 4
- 230000007547 defect Effects 0.000 abstract description 2
- 230000014509 gene expression Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 241000208340 Araliaceae Species 0.000 description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 description 3
- 235000008434 ginseng Nutrition 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 229910002056 binary alloy Inorganic materials 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D18/00—Testing or calibrating apparatus or arrangements provided for in groups G01D1/00 - G01D15/00
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
- G06F13/38—Information transfer, e.g. on bus
- G06F13/42—Bus transfer protocol, e.g. handshake; Synchronisation
- G06F13/4282—Bus transfer protocol, e.g. handshake; Synchronisation on a serial bus, e.g. I2C bus, SPI bus
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
The present invention relates to a kind of PSO-BP neural network sensor calibrating system based on MCU, it is characterised in that: the system comprises microprocessor, communication module, calibration module, digital tube display module and storage units；The microprocessor is separately connected by AHB-Lite bus and communication module, calibration module, digital tube display module and storage unit；The communication module is connect with sensor to be measured.The present invention can read the correction of data to sensor in real time, can effectively eliminate the nonlinear characteristic of sensor, solve the defects of existing software calibration mode real-time is poor, hardware calibration mode precision is limited.
Description
Technical field
The present invention relates to sensing datas to calibrate field, and in particular to a kind of PSO-BP neural network sensing based on MCU
Device calibration system and method.
Background technique
Current sensor calibrating system implementation method is broadly divided into software calibration and two kinds of hardware calibration.
Traditional software calibration generally uses the methods of fitting of a polynomial, interpolation method, and calibration accuracy is than hardware calibration method
Height, but real-time is poor.The calibration accuracy of fitting of a polynomial is vulnerable to the quantity of nominal data and the shadow of polynomial fitting order
It rings, and error is larger outside input data section.Curve of the Newton interpolating method in endpoint junction may rough, cubic spline
Although interpolation method precision is high but calculating process is excessively complicated.Neural network algorithm nonlinear function approximation ability with higher,
Its fitting effect is preferable.The realization of Neural Network Calibration algorithm at this stage is mostly based on software, and subsequent needs are artificially to sensor
Data are acquired and handle, and are unfavorable for real-time, the automatic measurement of sensor.
Hardware calibration real-time is high, but precision is lower, and that there are precision is related to circuit complexity, is not easy to debug and the offices such as integrates
It is sex-limited.Traditional hardware calibration method has the methods of resistor network, feedback adjustment amplifier, feedback modification supply voltage.Using electricity
It is smaller to hinder network progress calibration correction range, and correction accuracy is not high.It is not easy using the calibrating mode that feedback adjustment amplifier inputs
In integrating, the debugging of integrated circuit is also complex, needs the ratio of strict control resistance.Feedback modification supply voltage carries out school
Quasi- circuit is excessively complicated, is not easy to integrate, debugging process is cumbersome.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of PSO-BP neural network pick up calibration system based on MCU
System and method, calibration algorithm neural network based read the correction of data to sensor in real time, can effectively eliminate sensor
Nonlinear characteristic solves the defects of existing software calibration mode real-time is poor, hardware calibration mode precision is limited.
To achieve the above object, the present invention adopts the following technical scheme:
A kind of PSO-BP neural network sensor calibrating system based on MCU, the system comprises microprocessor, communication module,
Calibration module, digital tube display module and storage unit；The microprocessor passes through AHB-Lite bus and communication module, calibration
Module, digital tube display module and storage unit are separately connected；The communication module is connect with sensor to be measured.
Further, the communication module uses I2C bus, and bus is made of data line and clock line.
Further, the calibration algorithm of the calibration module uses PSO-BP neural network, the PSO-BP neural network
Structure is single hidden layer, input layer, hidden layer, output layer neuron number be 1:3:1, input layer and hidden layer use
Sigmoid function uses linear function as activation primitive, the activation primitive of output layer.
Further, the Sigmoid function is fitted using polynomial segmentation fitting mode.
Further, the BP neural network building specifically:
Step 1:BP neural network carries out model training using the learning rules of error back propagation；
Step 2: after having collected enough multisensor reading data, importing the study that data carry out weight and biasing；
Step 3: in the error retrospectively calculate stage, according to chain rule, optimizing secondary cost function using stochastic gradient descent method；
Step 4: the initialization using the model error of BP neural network as fitness function, using PSO algorithm to BP neural network
Initial parameter optimizes, and obtains the lesser initial parameter of error;
Step 5: the lesser initial parameter of error will be obtained, be trained by BP neural network, finally complete model training, obtained
To PSO-BP neural network model.
Further, the display module uses digital tube display module.
A kind of control method of the PSO-BP neural network sensor calibrating system based on MCU, comprising the following steps:
Step S1: microprocessor reads sensing data by communication module；
Step S2: the microprocessor is standardized sensing data, obtains pretreated reading data；
Step S3: pretreated reading data being exported to calibration module and are calibrated, and by the reading data after calibration into
It exports after the anti-standardization of row to microprocessor；
Step S4: microprocessor shows data transmission to the display module after anti-standardization.
Compared with the prior art, the invention has the following beneficial effects:
1, present invention employs the stronger BP neural networks of capability of fitting as calibration algorithm, and using PSO algorithm to initial ginseng
Number optimizes, and improves the precision and convergence rate of traditional BP neural network.
2, the present invention migrates to Neural Network Calibration system on hardware end MCU from software end, completes calibration system
It builds, and the data communication that I2C protocol communication module carries out sensor chip and calibration system is added in systems, solve soft
The deficiency of part calibration real-time difference.
3, the present invention realizes Sigmoid function, and piecewise polynomial fitting mode is utilized to the activation primitive of neural network
It is designed, while module reuse technology is introduced to the activation primitive of neural network, ensure that the same of hardware calibration precision
When reduce the area and power consumption of system.
Detailed description of the invention
Fig. 1 is calibration system block diagram of the present invention；
Fig. 2 is neural network hardware structure chart of the present invention；
Fig. 3 is Sigmoid activation primitive hardware realization block diagram of the present invention；
Fig. 4 is present system workflow block diagram.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
Fig. 1 is please referred to, the present invention provides a kind of PSO-BP neural network sensor calibrating system based on MCU, the system
System includes microprocessor, communication module, calibration module, digital tube display module and storage unit；The microprocessor passes through
AHB-Lite bus is separately connected with communication module, calibration module, digital tube display module and storage unit；The communication module
It is connect with sensor to be measured.
In the present embodiment, microprocessor is mainly responsible for the node state scheduling of data prediction and system.Microprocessor passes through logical
Believe that module reads sensing data, will then read data and be standardized, it, will in the case where not changing data discrete degree
Sensor reads data uniform range to improve the convergence rate of model.Standardized data are transmitted to calibration module and carry out school
Standard then standardizes the readout after calibration is counter again, transfers data to display module by microprocessor.Microprocessor master
The read-write operation to slave is realized by the cooperation of multiplexer and address decoder.
In the present embodiment, for storing HEX file, the main machine code for storing C language and generating, code will lead to storage unit
It crosses AHB-Lite bus and is sent to microprocessor and it is enabled to be written and read slave.
In the present embodiment, the communication module uses I2C bus, and bus is made of data line and clock line.By sensor
As slave, the timing of microprocessor and C language simulation I2C host is utilized.Host sends commencing signal, is sought according to address signal
Slave to be looked for, and sends the instruction read or write, slave generates response, corresponding data transmission work is then carried out between slave,
Each time operation terminate, slave can all generate an answer signal with judge communication whether succeed, after the data transfer ends, host
End signal is sent, I2C bus is discharged.
In the present embodiment, BP neural network carries out model training using the learning rules of error back propagation.Foot is collected
After more than enough sensor reads data, the study of weight and biasing is carried out in software end.In the error retrospectively calculate stage, according to chain type
Rule optimizes secondary cost function using stochastic gradient descent method.Population (PSO) algorithm is introduced to the initial ginseng of neural network
Number is rationally arranged.Using the model error of BP neural network as fitness function, using PSO algorithm to the first of BP neural network
Beginningization initial parameter optimizes, and obtains the lesser initial parameter of error, is then trained again by BP neural network, finally
Complete model training.
In this implementation, the integrally-built hardware circuit mapping of PSO-BP neural network is as shown in Figure 2.Its structure is set to single hidden
Hide layer, input layer, hidden layer, output layer neuron number be set to 1:3:1.In view of sensor read data in negative data compared with
To be common, ReLu activation primitive is difficult to be fitted negative data, so using Sigmoid function as sharp in input layer and hidden layer
Function living.To guarantee finally can normally show positive negative value, the activation primitive of output layer uses linear function.
For the computational accuracy and hardware resource consumption of equilibrium nerve network model, software end is not used in hardware realization
In usually used floating number, and calculated using fixed-point number.In order to guarantee precision, the input of model and output number in hardware
According to 16 fixed-point numbers are used, data, which calculate, uses 32 fixed-point numbers.From left to right it is followed successively by sign bit in 16 fixed-point numbers, 3
Integer-bit, 12 decimal places；Sign bit, 7 integer-bits, 24 decimal places are from left to right followed successively by 32 fixed-point numbers.
Adder and multiplier carries out shifter-adder to binary system multiplier according to by binary system multiplier, is by a succession of of clock control
" displacement-addition " operation.
In the present embodiment, Sigmoid activation primitive hardware realization block diagram is as shown in Figure 3.Due to being difficult to realize refer in hardware
Number function, so being fitted Sigmoid function using polynomial segmentation fitting mode.Under the premise of ensuring the accuracy requirement, will swash
Function living is divided into 12 sections, and the order of fitting function is up to two ranks.The ginseng of different segmentations is stored and searched using look-up table
Number realizes the multinomial in different sections.The calculating of one second order polynomial can substantially be divided into Four processes: firstly, according to input
Data x searches corresponding second order polynomial coefficient a, b, c；Then, second stage uses two multiplier parallel computation x2With bx；
Later, ax is calculated simultaneously using a multiplier and adder2With bx+c；Finally, being added output y using adder.In order to protect
Certain calculating speed is demonstrate,proved, assembly line is not used in the design, while introducing parallel computation.Its adder and multiplication module
Using module reuse to reduce resource consumption.
In the present embodiment, the display module uses digital tube display module.Since data within hardware are stored as two
System or hexadecimal, it is therefore desirable to which carrying out certain conversion could be shown with the decimal system.Pass through simple division modulus
Remainder can obtain corresponding decimal value.It is built due to division with hardware circuit and relatively occupies resource area, converter section
Divide and is completed by microprocessor.Highest order is sign bit for indicating positive and negative, rear 5 expressions numerical value, wherein preceding 3 expressions integer portion
Point, rear 2 expressions fractional part.It is displayed data using six Digital sum pipes, highest order is sign bit, 2-4 expression integers
Part, 5-6 expression fractional parts.
Referring to Fig. 4, in the present embodiment, a kind of controlling party of the PSO-BP neural network sensor calibrating system based on MCU
Method, comprising the following steps:
Step S1: microprocessor reads sensing data by communication module；
Step S2: the microprocessor is standardized sensing data, obtains pretreated reading data；
Step S3: pretreated reading data being exported to calibration module and are calibrated, and by the reading data after calibration into
It exports after the anti-standardization of row to microprocessor；
Step S4: microprocessor shows data transmission to the display module after anti-standardization.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (7)
1. a kind of PSO-BP neural network sensor calibrating system based on MCU, it is characterised in that: the system comprises micro processs
Device, communication module, calibration module, digital tube display module and storage unit；The microprocessor by AHB-Lite bus with
Communication module, calibration module, digital tube display module and storage unit are separately connected；The communication module and sensor to be measured connect
It connects.
2. the PSO-BP neural network sensor calibrating system according to claim 1 based on MCU, it is characterised in that: institute
Communication module is stated using I2C bus, bus is made of data line and clock line.
3. the PSO-BP neural network sensor calibrating system according to claim 1 based on MCU, it is characterised in that: institute
The calibration algorithm of calibration module is stated using PSO-BP neural network, the PSO-BP neural network structure is single hidden layer, input
Layer, hidden layer, output layer neuron number be 1:3:1, input layer and hidden layer use Sigmoid function as activation primitive,
The activation primitive of output layer uses linear function.
4. the PSO-BP neural network sensor calibrating system according to claim 3 based on MCU, it is characterised in that: institute
State BP neural network building specifically:
Step 1:BP neural network carries out model training using the learning rules of error back propagation；
Step 2: after having collected enough multisensor reading data, carrying out the study of weight and biasing；
Step 3: in the error retrospectively calculate stage, according to chain rule, optimizing secondary cost function using stochastic gradient descent method；
Step 4: the initialization using the model error of BP neural network as fitness function, using PSO algorithm to BP neural network
Initial parameter optimizes, and obtains the lesser initial parameter of error;
Step 5: the lesser initial parameter of error will be obtained, be trained by BP neural network, finally complete model training, obtained
To PSO-BP neural network model.
5. the PSO-BP neural network sensor calibrating system according to claim 3 based on MCU, it is characterised in that: institute
Sigmoid function is stated to be fitted using polynomial segmentation fitting mode.
6. the PSO-BP neural network sensor calibrating system according to claim 1 based on MCU, it is characterised in that: institute
Display module is stated using digital tube display module.
7. a kind of control method of the PSO-BP neural network sensor calibrating system based on MCU, which is characterized in that including following
Step:
Step S1: microprocessor reads sensing data by communication module；
Step S2: the microprocessor is standardized sensing data, obtains pretreated reading data；
Step S3: pretreated reading data being exported to calibration module and are calibrated, and by the reading data after calibration into
It exports after the anti-standardization of row to microprocessor；
Step S4: microprocessor shows data transmission to the display module after anti-standardization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910432693.4A CN110070170A (en) | 2019-05-23 | 2019-05-23 | PSO-BP neural network sensor calibrating system and method based on MCU |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910432693.4A CN110070170A (en) | 2019-05-23 | 2019-05-23 | PSO-BP neural network sensor calibrating system and method based on MCU |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110070170A true CN110070170A (en) | 2019-07-30 |
Family
ID=67371358
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910432693.4A Pending CN110070170A (en) | 2019-05-23 | 2019-05-23 | PSO-BP neural network sensor calibrating system and method based on MCU |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110070170A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5554273A (en) * | 1995-07-26 | 1996-09-10 | Praxair Technology, Inc. | Neural network compensation for sensors |
| RU2199089C1 (en) * | 2002-04-11 | 2003-02-20 | ООО "НТ-Центр" | Method for correcting static characteristics of measuring transducers |
| US20100169254A1 (en) * | 2008-12-29 | 2010-07-01 | Chi Mei Communication Systems, Inc. | System and method for calibrating radio frequency power of communication devices |
| CN103983332A (en) * | 2014-05-31 | 2014-08-13 | 福州大学 | Method for error compensation of sensor based on HGSA-BP algorithm |
| CN104484703A (en) * | 2014-12-30 | 2015-04-01 | 合肥工业大学 | Sigmoid function fitting hardware circuit based on Remez approximating algorithm |
| CN104875577A (en) * | 2014-02-27 | 2015-09-02 | 惠州市德赛西威汽车电子有限公司 | All-condition double-area temperature calibration method and automobile air conditioner control system |
| CN105258839A (en) * | 2015-10-30 | 2016-01-20 | 南京信息工程大学 | Array type air pressure measurement compensation device and method based on quantum particle swarm wavelet neural network |
| CN106153567A (en) * | 2016-06-22 | 2016-11-23 | 南京信息工程大学 | Based on the pressure compensated infrared gas sensor of BP neutral net and detection method |
| CN108073075A (en) * | 2017-12-21 | 2018-05-25 | 苏州大学 | Silicon micro accerometer temperature-compensation method, system based on GA Optimized BP Neural Networks |
| CN108537332A (en) * | 2018-04-12 | 2018-09-14 | 合肥工业大学 | A kind of Sigmoid function hardware-efficient rate implementation methods based on Remez algorithms |
| CN210488594U (en) * | 2019-05-23 | 2020-05-08 | 福州大学 | PSO-BP neural network sensor calibration system based on MCU |
-
2019
- 2019-05-23 CN CN201910432693.4A patent/CN110070170A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5554273A (en) * | 1995-07-26 | 1996-09-10 | Praxair Technology, Inc. | Neural network compensation for sensors |
| RU2199089C1 (en) * | 2002-04-11 | 2003-02-20 | ООО "НТ-Центр" | Method for correcting static characteristics of measuring transducers |
| US20100169254A1 (en) * | 2008-12-29 | 2010-07-01 | Chi Mei Communication Systems, Inc. | System and method for calibrating radio frequency power of communication devices |
| CN104875577A (en) * | 2014-02-27 | 2015-09-02 | 惠州市德赛西威汽车电子有限公司 | All-condition double-area temperature calibration method and automobile air conditioner control system |
| CN103983332A (en) * | 2014-05-31 | 2014-08-13 | 福州大学 | Method for error compensation of sensor based on HGSA-BP algorithm |
| CN104484703A (en) * | 2014-12-30 | 2015-04-01 | 合肥工业大学 | Sigmoid function fitting hardware circuit based on Remez approximating algorithm |
| CN105258839A (en) * | 2015-10-30 | 2016-01-20 | 南京信息工程大学 | Array type air pressure measurement compensation device and method based on quantum particle swarm wavelet neural network |
| CN106153567A (en) * | 2016-06-22 | 2016-11-23 | 南京信息工程大学 | Based on the pressure compensated infrared gas sensor of BP neutral net and detection method |
| CN108073075A (en) * | 2017-12-21 | 2018-05-25 | 苏州大学 | Silicon micro accerometer temperature-compensation method, system based on GA Optimized BP Neural Networks |
| CN108537332A (en) * | 2018-04-12 | 2018-09-14 | 合肥工业大学 | A kind of Sigmoid function hardware-efficient rate implementation methods based on Remez algorithms |
| CN210488594U (en) * | 2019-05-23 | 2020-05-08 | 福州大学 | PSO-BP neural network sensor calibration system based on MCU |
Non-Patent Citations (1)
| Title |
|---|
| 袁东明 等著: "《现代数字电路与逻辑设计实验教程》", 北京：北京邮电大学出版社, pages: 210 - 212 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Judd et al. | Asymptotic methods for aggregate growth models | |
| CN107169563B (en) | Processing system and method applied to two-value weight convolutional network | |
| CN104238993A (en) | Vector matrix product accelerator for microprocessor integration | |
| CN109146067B (en) | Policy convolution neural network accelerator based on FPGA | |
| KR20220097961A (en) | Recurrent neural networks and systems for decoding encoded data | |
| CN107423816A (en) | A kind of more computational accuracy Processing with Neural Network method and systems | |
| CN107133699A (en) | Box office receipts trend forecasting method and device, equipment, storage medium | |
| CN109934336A (en) | Neural network dynamic based on optimum structure search accelerates platform designing method and neural network dynamic to accelerate platform | |
| CN104142892B (en) | A kind of data read-write method, apparatus and system | |
| WO2022111002A1 (en) | Method and apparatus for training neural network, and computer readable storage medium | |
| CN115425976A (en) | ADC sampling data calibration method and system based on FPGA | |
| CN210488594U (en) | PSO-BP neural network sensor calibration system based on MCU | |
| CN115796338A (en) | Photovoltaic power generation power prediction model construction and photovoltaic power generation power prediction method | |
| CN110070170A (en) | PSO-BP neural network sensor calibrating system and method based on MCU | |
| CN114970357A (en) | Energy-saving effect evaluation method, system, device and storage medium | |
| CN116720554B (en) | Method for realizing multi-section linear fitting neuron circuit based on FPGA technology | |
| CN118410849A (en) | Method, apparatus and medium for determining quantization cut-off value | |
| CN112183744A (en) | Neural network pruning method and device | |
| JP6973651B2 (en) | Arithmetic optimizers, methods and programs | |
| CN102314215A (en) | Low power consumption optimization method of decimal multiplier in integrated circuit system | |
| CN114997386B (en) | CNN neural network acceleration design method based on multi-FPGA heterogeneous architecture | |
| US20220358183A1 (en) | Matrix multiplier and operation method thereof | |
| CN115826911A (en) | Variable bit width adder tree generating system based on multi-type approximate computing unit | |
| CN111198714B (en) | Retraining method and related product | |
| CN111260070B (en) | Operation method, device and related product |
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 |