CN208889147U - Electronic synapse experiment porch based on FPGA - Google Patents
Electronic synapse experiment porch based on FPGA Download PDFInfo
- Publication number
- CN208889147U CN208889147U CN201821000736.9U CN201821000736U CN208889147U CN 208889147 U CN208889147 U CN 208889147U CN 201821000736 U CN201821000736 U CN 201821000736U CN 208889147 U CN208889147 U CN 208889147U
- Authority
- CN
- China
- Prior art keywords
- fpga
- synapse
- electronic
- experiment porch
- neuron
- 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.)
- Active
Links
Abstract
The utility model provides a kind of electronic synapse experiment porch based on FPGA, including computer, computer is connect with four FPGA signals respectively, it both connect with the first FPGA signal with the 2nd FPGA and the 3rd FPGA or is connect with the 4th FPGA signal, four FPGA are connected with digital analog converter, and digital analog converter is connected with oscillograph;First FPGA generates action potential as No. 1 electronic nerve cell;2nd FPGA generates excitatory synapse electric current;3rd FPGA generates inhibitory synapse electric current;4th FPGA does not apply any environmental stimuli as No. 2 electronic nerve cells, stimulates from synaptic currents.The experiment porch can be observed under different stiffness of couplings, the similarities and differences of nerve information transmitting;Different type cynapse, that is, inhibition and excitory chemical synapse are respectively to the influence under the effect and two types chemical synapse collective effect of information transmitting to information transmitting.
Description
Technical field
The utility model belongs to biomedical engineering technology field, is related to a kind of experiment porch for imitating electronic synapse, special
It is not related to a kind of experiment porch of electronic synapse based on FPGA.
Background technique
1894, the famous physiologist Charles Sherrington of Britain was put forward for the first time the concept of " cynapse ".?
In biological nervous system, a large amount of neurons are interconnected to form neural network by cynapse.The key effect of cynapse is processing and passes
Pass the information between neuron.Famous physiologist professor Feng Depei in China has one section of elaboration to the effect of cynapse, i.e., " entire
In Neurobiology, cynapse and its related research can be described as holding the central place.Because nervous system is substantially information
System of processing, and Information procession requires neuron and neuron to talk with, this is carried out by cynapse ".Scholars pass through biology
The physiological property of process and chemical synapse that experimental exploring electric pulse is transmitted in chemical synapse, establishes based on different mechanism
Chemical synapse mathematical model.1992, Hansel et al. proposed a kind of most simple in research chaotic neural network synchronizing process
Single mathematical model for portraying neuron chemical synapse;1996, Sharp et al. was based on neurotransmitter and discharges in synaptic cleft
With the absorption characteristics in post-synapse neuron, a kind of chemical synapse mathematical model is proposed;1996, Xiao-Jing Wang
Et al. research hippocampal neuron cynapse inhibit network model γ Oscillating Mechanism when propose a kind of chemical synapse model;1997
Year, Rabinovich et al. is established in view of the excitement of synaptic delay and neuron is with after inhibition with time lag
Cynapse mathematical model;2007, Leonid proposed a kind of kinetic model of complexity of cynapse transmitting.
Hardware realization biological nervous system is always a research hotspot 1989 in computational neuroscience field, Carver
Mead etc. constructs analog circuit, realizes neural network model;2009, Theodore Yu etc. was in an analogue VLSI silicon
The biophysics cynapse kinetic model of four coupling HH neuroids is realized on chip;2010, Theodore Yu etc.
Propose the analogue VLSI network of 4 neurons and 12 cynapses based on conductance, and realize biophysics film dynamics and
The exaflex injection type of detailed Ion channel kinetics;2011, Marko Noack etc. was based on the quantum model from neurotransmitter regulator
Circuit analysis method propose and a kind of realize cynapse kinetic mechanism with more the actual VLSI of biology;2015,
Pradyumna S. G. etc. is realized using CMOS technology with different concentration of neurotransmitters, intake time and propagation delay
Inhibitory synapse;2016, Sadia Alam etc. proposed the VLSI circuit of the chemical synapse process based on bio-molecular level
Emulation.
In recent years, compared with the disadvantages of consuming of analog circuit implementation is big, time-consuming, field programmable gate array
(Field-Programmable Gate Array), i.e. FPGA has the characteristics that parallel processing manner, while having the development cycle concurrently
The advantages such as short, integrated level is high, calculating speed is fast, repeatable configuration, low cost.Therefore, biological neuron, mind are realized using FPGA
There is very wide application prospect in terms of through cynapse and neural network, and for bionics, artificial intelligence, neurotherapeutic
Etc. have very big application value.
The prior art is also in foundation phase, therefore there are still following disadvantages: 1) there is no the special electronic based on FPGA prominent
Touch experiment porch;2) the electronic synapse experiment porch that can replace the zoopery of biological neural Synaptic junction does not apply to reality
In.
Summary of the invention
The purpose of the utility model is to provide a kind of experiment porch of electronic synapse based on FPGA, it can be observed that not
With under stiffness of coupling, the similarities and differences of nerve information transmitting, different type cynapse, that is, inhibition and excitory chemical synapse are respectively to letter
Cease the influence under the effect and two types chemical synapse collective effect of transmitting to information transmitting.
To achieve the above object, the technical scheme adopted by the utility model is a kind of electronic synapse based on FPGA is real
Platform, including computer are tested, computer is connect with the first FPGA, the 2nd FPGA, the 3rd FPGA and the 4th FPGA signal respectively, the
One FPGA is connect with the 2nd FPGA and the 3rd FPGA signal respectively, and the 2nd FPGA and the 3rd FPGA connect with the 4th FPGA signal respectively
It connects, the first FPGA, the 2nd FPGA, the 3rd FPGA and the 4th FPGA are connected with digital analog converter, digital analog converter and oscillograph
It is connected;
First FPGA is used as No. 1 electronic nerve cell in the experiment porch, for being used as presynaptic neuron, receives electricity thorn
Swash, generates action potential;
2nd FPGA is for generating excitatory synapse electric current;
3rd FPGA is for generating inhibitory synapse electric current;
4th FPGA is used as No. 2 electronic nerve cells in the experiment porch, for being used as postsynaptic neuron, the neuron
Do not apply any environmental stimuli, stimulates from synaptic currents.
Biological neural cynapse is simulated with electronic synapse in the utility model experiment porch, can allow the professions such as biology, medicine
Student observe Electrophysiology experiment when, replace zoopery with electronic synapse experiment.It is established according to electrophysiological characteristics
The mathematical model of cynapse be entirely encapsulated in the core chips of FPGA, can be seen by changing the input of stimulated current
Examine the lag characteristic of transmitting and the cynapse of the action potential of presynaptic caudacoria neuron.Electronic synapse can be small for hardware realization
Type neuroid plays an important role, or the synchronous discharge of the observation of students neuron of the profession such as biology, medicine
Journey provides experimental method.It can be observed under different stiffness of couplings using the experiment porch, the similarities and differences of nerve information transmitting;It is different
Type cynapse, that is, inhibition and excitory chemical synapse are common to the effect and two types chemical synapse of information transmitting respectively
To the influence of information transmitting under effect.
Detailed description of the invention
Fig. 1 is the schematic diagram of the utility model experiment porch.
Fig. 2 is digital integration implementation model.
Fig. 3 is the calculating pipeline organization model that variable S is gated in chemical synapse electric current.
Fig. 4 is No. 2 HH neuron models pipeline organization models.
Fig. 5 is the hardware realization procedure chart for realizing electronics nerve synapse.
Fig. 6 is No. 1 electronic nerve cell action potential waveform, No. 2 electronic nerve cells pass through excitory chemical synapse and inhibition
Property chemical synapse transmitting when action potential waveform and excitory chemical synapse response and inhibition chemical synapse response wave
Shape figure.
In Fig. 1: 1. computers, 2. the oneth FPGA, 3. the 2nd FPGA, 4. the 3rd FPGA, 5. the 4th FPGA, 6. digital-to-analogue conversions
Device, 7. oscillographs.
Specific embodiment
The utility model is described in detail with reference to the accompanying drawings and detailed description.
As shown in Figure 1, the utility model experiment porch, including computer 1, computer 1 respectively with the first FPGA2, second
FPGA3, the 3rd FPGA4 and the connection of the 4th FPGA5 signal, the first FPGA2 connect with the 2nd FPGA3 and the 3rd FPGA4 signal respectively
It connects, the 2nd FPGA3 and the 3rd FPGA4 are connect with the 4th FPGA5 signal respectively, the first FPGA2, the 2nd FPGA3, the 3rd FPGA4
It is connected with digital analog converter 6 with the 4th FPGA5, digital analog converter 6 is connected with oscillograph 10.
Computer 1 by USB control downloading wire respectively with the first FPGA2, the 2nd FPGA3, the 3rd FPGA4 and the 4th
The connection of FPGA5 signal.First FPGA2, the 2nd FPGA3, the 3rd FPGA4 and the 4th FPGA5 are all made of altera corp's production
II series EP2C8Q208C8 chip of Cyclone.
First FPGA2 is used as No. 1 electronic nerve cell in the utility model experiment porch, for being used as presynaptic nerve
Member receives certain electro photoluminescence, to generate action potential i.e. nerve information.
2nd FPGA3 is for generating excitatory synapse electric current.
3rd FPGA4 is for generating inhibitory synapse electric current.
Synaptic currents are pierced for the output par, c as cynapse, and to the electric current of another neuron (No. 2 neurons)
Swash part.
4th FPGA5 is used as No. 2 electronic nerve cells in the utility model experiment porch, for being used as postsynaptic neuronal
Member, the neuron do not apply any environmental stimuli, stimulate from synaptic currents.
Digital analog converter 6 uses chip DAC900E;For the nerve letter of No. 2 electronic nerve cells will to be transmitted to by cynapse
Breath is that action potential digital signal is converted to the action potential analog signal that can be shown on oscillograph.
Oscillograph 10 is used for synaptic currents and nerve information the analog signal real-time display in the form of waveform that will be obtained,
Convenient for observation analysis.
FPGA has that flexible design, repeatable configuration, development cycle be short, the advantages such as at low cost.It is as dedicated integrated
One of circuit field semi-custom circuit and occur, not only solved the deficiency of custom circuit, but also overcome original programmable
The limited disadvantage of device gate circuit number.
The mathematical model used in first FPGA2 and the 4th FPGA5 is for describing neuron action potential granting and passing
The Hodgkin-Huxley neuron models led.Hodgkin-Huxley neuron models, i.e. HH neuron models.It is by
The mathematical model for being used to describe neuron action potential granting and conduction that Hodgkin and Huxley is proposed in nineteen fifty-two, number
Learn nonlinear differential equation and six algebraic equations descriptions of the model by four single orders.The neuron models are closest to really
The mathematical model of biological neuron electrophysiological characteristics.
It is prominent in research hippocampal neuron in 1996 with Xiao-Jing Wang et al. in 2nd FPGA3 and the 3rd FPGA4
Touching inhibits a kind of chemical synapse mathematical model proposed when the γ Oscillating Mechanism of network model to generate synaptic currents, the mathematical model
In gate variable S described by a First-Order Nonlinear Differential Equation.
During realizing electronic synapse, electronic nerve cell selects HH neuron models to model, and realization is with neuron
The simulation of No. 1 electronic nerve cell and No. 2 electronic nerve cells of importation.Synaptic currents selection Xiao-Jing Wang et al. is mentioned
A kind of chemical synapse mathematical model out models, and realizes using synaptic currents as the simulation of output par, c.
Using FPGA to the hardware realization of HH neuron models and chemical synapse mathematical model, actually to HH neuron
Model equation and chemical synapse mathematical model equation carry out hardware mode parsing.HH neuron models and chemical synapse mathematical model
In all include is First-Order Nonlinear Differential Equation, illustrated below with a typical differential equation:
In formula,fFunction is the differential equation,y 0It is variable in the differential equationyInitial value.It is carried out using Euler's method from
It dissipates and solves, then obtain its corresponding difference equation:
y(n+1)=y n +ΔT×f(t n ,y n ) (n=0,1,2 ...) (2)
Wherein,y n It is the sequence of iterations of variable, ΔTIt is the time of integration.
According to above-mentioned basic principle (Fig. 2 is the functional block diagram for realizing digital integration function), to chemical synapse mathematical model
In gate variable S using Euler's method carry out it is discrete, obtain its difference equation:
S(k+1)=S(k)+ ΔT{αF(V 1) ﹝ 1-S(k) ﹞-βS(k) (3)
Realize electronic nerve cell and chemical synapse mathematical model with FPGA, be substantially to the differential equation of given model into
The mode of the difference equation Digital Signal Processing obtained after row discretization is calculated on FPGA.
It is directed to and is related to the system design of the class of algorithms, altera corp is proposed the design aids of a algorithm level
Case DSP Builder.The whole design that system is completed with DSP Builder, depends primarily on tool of mathematical analysis
MATLAB/Simulink, DSP Builder are embedded in MATLAB, are called in Simulink each in the library DSP Builder
A module for realizing calculating, can be patterned design and emulation in Simulink, to complete the whole of the system
Body design process.The DSP Builder model of foundation is downloaded on fpga chip, and reaches final purpose, then needs one
Serial procedures.The development language of FPGA is Hardware Description Language VHDL (Very-High-Speed Integrated Circuit
Hardware Description Language) or Verilog HDL.Therefore, it is necessary to pass through Signal Compiler module
The model file of MATLAB/Simulink is converted into corresponding hardware description language.It, will finally under II environment of Quartus
Obtained neuron and synaptic currents module with frequency division module mixed compiling, downloads to FPGA core by USB control respectively respectively
Piece.
For exterior design, such as frequency division module, program code realization can be write by VHDL.Mould can be divided by adjusting
Block parameter keeps hardware output result consistent with the time scale of numerical simulation.
On the one hand the service condition of FPGA resource, especially multiplier are considered in the modeling process of DSP Builder
It uses, multiplying is more in HH neuron models and chemical synapse model, if calling multiplication module always in modeling
Meeting causes fpga chip to be unable to complete operation so that limited multiplier resources are inadequate in FPGA.It therefore, can in modeling process
To consider that the use of other modules, such as look-up table (LUT) module can realize complicated multiplying, look-up table is comprehensive
The resource consumed when conjunction is Memory Bits and M4ks.On the other hand it is also contemplated that the timing of each variable is asked in calculating process
Topic, only ensure that timing synchronization just in calculating process can guarantee that operation is correct.Based on sequence problem, Weinstein et al.
A kind of patterned design scheme of use is proposed, the sequential relationship of each variable is showed by pipeline organization figure.Its
In, the computing module for introducing timing is mainly multiplication module, which is provided with 3 clock cycle.Delay register mould
Block for be delayed operation and achieve the purpose that adjust timing.The main function of time delay module is adjustment timing, guarantees that timing is same
Step, obtains correct operation result in calculating process.
The calculating pipeline organization model of variable S, the stream of the sequence problem of gate variable S are gated in chemical synapse electric current
Line structure figure, as shown in figure 3, the Z marked in Fig. 3-3Module indicates is time delay module in assembly line, be labelled with LUT
Module indicate be to look for table module, be labelled with Mult (Z-3) module indicate be multiplication module, be labelled with the module of β
What is indicated is ion channel shutdown rate module, and the module expression for being labelled with △ T is integration step module.Fig. 3 is shown firmly
Part is how to solve the problems, such as timing synchronization during realizing nerve synapse.
The stream of the sequence problem of postsynaptic neuron i.e. No. 2 neuron in No. 2 HH neuron models pipeline organization models
Line structure figure, as shown in figure 4, the Z wherein marked-16、Z-3、Z-10、Z-6、Z-1Module indicates is delay in assembly line
Module is labelled with Mult (Z-3) module indicate be multiplication module, be labelled with S (Z-10) module indicate be gate become
Module is measured, the module expression for being labelled with △ T is integration step module.Fig. 4 equally also shows hardware realization nerve synapse mistake
It is how to solve the problems, such as timing synchronization in journey.
No. 1 electronic nerve cell is both the importation in presynaptic neuron and the subsystem, and synaptic currents are the sons
The output par, c of system, and No. 2 neurons are both an input in postsynaptic neuron and synaptic currents;That is No. 1 electronics
Neuron, No. 2 there are relationships between electronic nerve cell and cynapse, from the mathematics model analysis of cynapse, during constituting cynapse
Input of the film potential of No. 1 electronic nerve cell and No. 2 electronic nerve cells as cynapse is needed, then cynapse is with the shape of synaptic currents
Formula stimulates No. 2 electronic nerve cells, to finally realize the effect of cynapse transmission information.
When using the utility model experiment porch, process is as shown in figure 5, established to neuron and nerve synapse first
DSP Builder model, the model are the cores of hardware realization electronic nerve cell and nerve synapse.Followed by under software environment
(software used is (II 11.0 version of Quartus) of altera corp to the corresponding hardware description language file of the model
Mixed editorial is carried out with frequency dividing file, that is, carries out establishing the processes such as metafile, configuration pin.A series of editors will finally be passed through
Programming file download generated is into fpga core chip later, to finally realize electronic nerve cell and neural process with FPGA
Touching.
The waveform diagram shown on oscillograph 7, as shown in Figure 6.What Fig. 6 (a) and Fig. 6 (b) was indicated is in excitability
Waveform when cynapse transmission information is learned, wherein 1 channel in (a) is the action potential waveform of No. 1 electronic nerve cell, 2 is logical
Road is the waveform of excitory chemical synapse response.(b) 1 channel in is the action potential waveform of No. 1 electronic nerve cell, 2 channels
It is No. 2 electronic nerve cell action potential waveforms acted on by the transmitting of excitory chemical synapse.(c) it is indicated with (d) figure
Be waveform in inhibition chemical synapse transmission information, wherein 1 channel in (c) is the dynamic of No. 1 electronic nerve cell
Make potential waveform, 2 channels are the waveforms of inhibition chemical synapse response.(d) 1 channel in is the movement of No. 1 electronic nerve cell
Potential waveform, 2 channels are the waveforms of No. 2 electronic nerve cells acted on by the transmitting of inhibition chemical synapse.
The experiment porch of electronic synapse based on FPGA, it can be observed that under different stiffness of couplings, nerve signal transmitting
The similarities and differences, the effect and two types to information transmitting respectively of different type cynapse, that is, inhibition and excitory chemical synapse
To the influence of information transmitting under chemical synapse collective effect.The experiment porch is real for replacing the animal of biological neural Synaptic junction
Testing has certain application value.
Claims (2)
1. a kind of electronic synapse experiment porch based on FPGA, which is characterized in that including computer (1), computer (1) respectively with
First FPGA(2), the 2nd FPGA(3), the 3rd FPGA(4) connected with the 4th FPGA(5) signal, the first FPGA(2) respectively with
Two FPGA(3) connected with the 3rd FPGA(4) signal, the 2nd FPGA(3) and the 3rd FPGA(4) respectively with the 4th FPGA(5) signal
Connection, the first FPGA(2), the 2nd FPGA(3), the 3rd FPGA(4) and the 4th FPGA(5) be connected with digital analog converter (6), it is several
Mode converter (6) is connected with oscillograph (10);
First FPGA(2) No. 1 electronic nerve cell is used as in the experiment porch, for being used as presynaptic neuron, receive electricity thorn
Swash, generates action potential;
2nd FPGA(3) for generating excitatory synapse electric current;
3rd FPGA(4) for generating inhibitory synapse electric current;
4th FPGA(5) No. 2 electronic nerve cells are used as in the experiment porch, for being used as postsynaptic neuron, the neuron
Do not apply any environmental stimuli, stimulates from synaptic currents.
2. the electronic synapse experiment porch according to claim 1 based on FPGA, which is characterized in that the computer
(1) downloading wire is controlled by USB to believe with the first FPGA(2), the 2nd FPGA(3), the 3rd FPGA(4) and the 4th FPGA(5) respectively
Number connection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201821000736.9U CN208889147U (en) | 2018-06-27 | 2018-06-27 | Electronic synapse experiment porch based on FPGA |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201821000736.9U CN208889147U (en) | 2018-06-27 | 2018-06-27 | Electronic synapse experiment porch based on FPGA |
Publications (1)
Publication Number | Publication Date |
---|---|
CN208889147U true CN208889147U (en) | 2019-05-21 |
Family
ID=66505984
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201821000736.9U Active CN208889147U (en) | 2018-06-27 | 2018-06-27 | Electronic synapse experiment porch based on FPGA |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN208889147U (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108711351A (en) * | 2018-06-27 | 2018-10-26 | 兰州交通大学 | Electronic synapse experiment porch based on FPGA |
-
2018
- 2018-06-27 CN CN201821000736.9U patent/CN208889147U/en active Active
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108711351A (en) * | 2018-06-27 | 2018-10-26 | 兰州交通大学 | Electronic synapse experiment porch based on FPGA |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Pearson et al. | Implementing spiking neural networks for real-time signal-processing and control applications: A model-validated FPGA approach | |
Heidarpur et al. | CORDIC-SNN: On-FPGA STDP learning with izhikevich neurons | |
Yang et al. | CerebelluMorphic: large-scale neuromorphic model and architecture for supervised motor learning | |
CN103691058B (en) | The deep brain stimulation FPGA experiment porch of parkinson disease basal nuclei-thalamus network | |
Yang et al. | Cost-efficient FPGA implementation of basal ganglia and their Parkinsonian analysis | |
Pani et al. | An FPGA platform for real-time simulation of spiking neuronal networks | |
CN104615909B (en) | Izhikevich neuroid synchronous discharge emulation platforms based on FPGA | |
CN104112066A (en) | Epilepsy state closed-loop control experiment platform based on FPGA | |
Yang et al. | Digital implementations of thalamocortical neuron models and its application in thalamocortical control using FPGA for Parkinson׳ s disease | |
CN104375878A (en) | FPGA simulation system for modulating hippocampus network rhythm through weak electromagnetic stimulation | |
Smaragdos et al. | FPGA-based biophysically-meaningful modeling of olivocerebellar neurons | |
CN104689473B (en) | Neuron random response and resonance laboratory platform under electricity irritation based on FPGA | |
Yu et al. | Energy balance via memristor synapse in Morris-Lecar two-neuron network with FPGA implementation | |
Grassia et al. | Digital hardware implementation of a stochastic two-dimensional neuron model | |
CN107169184A (en) | Substrate nuclear network beta rhythm and pace of moving things analogue systems based on FPGA | |
CN105631222B (en) | STDP synaptic plasticity experiment porch under feedforward neural network based on FPGA | |
Niu et al. | Emulated muscle spindle and spiking afferents validates VLSI neuromorphic hardware as a testbed for sensorimotor function and disease | |
CN208889147U (en) | Electronic synapse experiment porch based on FPGA | |
Stradmann et al. | Demonstrating analog inference on the brainscales-2 mobile system | |
CN108711351A (en) | Electronic synapse experiment porch based on FPGA | |
CN106407568A (en) | A Parkinson-state basal nuclei stimulation real-time simulation platform based on finite element analysis | |
CN105845001B (en) | Multiple electrode array Simulation Experimental Platform based on FPGA | |
Alfaro-Badilla et al. | Improving the simulation of biologically accurate neural networks using data flow HLS transformations on heterogeneous SoC-FPGA platforms | |
CN105631223B (en) | Cardiopulmonary rhythm and pace of moving things synchronous control experiment porch based on CPG | |
Smaragdos et al. | Real-time olivary neuron simulations on dataflow computing machines |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
GR01 | Patent grant | ||
GR01 | Patent grant |