CN116611489A - Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback - Google Patents
Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback Download PDFInfo
- Publication number
- CN116611489A CN116611489A CN202310501914.5A CN202310501914A CN116611489A CN 116611489 A CN116611489 A CN 116611489A CN 202310501914 A CN202310501914 A CN 202310501914A CN 116611489 A CN116611489 A CN 116611489A
- Authority
- CN
- China
- Prior art keywords
- feedback
- optical
- photoelectric
- mach
- zehnder modulator
- 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
- 230000003287 optical effect Effects 0.000 claims abstract description 60
- 239000013307 optical fiber Substances 0.000 claims abstract description 13
- 239000004065 semiconductor Substances 0.000 claims description 6
- 229910052691 Erbium Inorganic materials 0.000 claims description 3
- UYAHIZSMUZPPFV-UHFFFAOYSA-N erbium Chemical group [Er] UYAHIZSMUZPPFV-UHFFFAOYSA-N 0.000 claims description 3
- 239000000835 fiber Substances 0.000 claims description 3
- 238000012360 testing method Methods 0.000 description 13
- 238000004364 calculation method Methods 0.000 description 11
- 230000000739 chaotic effect Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 210000004027 cell Anatomy 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 238000006467 substitution reaction Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000010365 information processing Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000000034 method Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 210000000352 storage cell Anatomy 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/067—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using optical means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Neurology (AREA)
- Optical Communication System (AREA)
Abstract
The invention discloses a storage pool computing device based on photoelectric double feedback of a Mach-Zehnder modulator, which consists of an input module, a photoelectric double feedback storage pool module and a data acquisition module. The input module consists of a signal generator and a combiner, the photoelectric dual-feedback reservoir module consists of a continuous optical laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator, a coupler 2, an optical feedback branch consisting of a delay optical fiber 1, an optical isolator and an optical amplifier and a photoelectric feedback branch consisting of the delay optical fiber 2, a photoelectric detector and an electric amplifier, feedback light is taken as the optical input of the Mach-Zehnder modulator together with the output light of the continuous optical laser through the coupler 1, feedback electric signals are taken as the radio frequency input of the Mach-Zehnder modulator, and the output light of the modulator enters the optical feedback branch and the photoelectric feedback branch respectively through the coupler 2; the data acquisition module is composed of a power divider and a data acquisition unit. The invention has rich dynamic state and good network performance.
Description
Technical Field
The invention relates to a reserve tank computing device based on photoelectric double feedback of a Mach-Zehnder modulator, and belongs to the technical field of photoelectric information processing.
Background
An artificial neural network is a machine learning model that mimics the brain for information processing. Reservoir computation is a simplified recurrent neural network, which is conveniently implemented in hardware to increase its processing speed. Hardware implementations of traditional pool computing require a large number of physical devices to construct the pool. The time delay reserve pool calculation is realized by adopting a time delay dynamic system and arranging a plurality of virtual nodes to replace a plurality of nodes in the reserve pool, so that the demand of the reserve pool calculation on hardware is reduced to the greatest extent, and the time delay reserve pool calculation can be realized by only one nonlinear node and one delay feedback loop. Especially, the implementation mode of the optical time delay system not only exerts the characteristic of quick light propagation, but also has the advantage of low power consumption.
Photon reservoir calculations are divided into full light reservoir calculations and photoelectric reservoir calculations. The calculation of the all-optical reserve pool needs to strictly control conditions such as temperature, and the sampling rate of signal generation and receiving is high. In contrast, the photoelectric reserve cell calculation not only utilizes the advantages of high speed, low power consumption and the like of light, but also has lower requirements on a hardware system and is easy to realize. However, the dynamic characteristics of the inside of the photovoltaic storage cells currently existing are still to be improved. The existing improvement schemes, such as double-node double-photoelectric feedback, photoelectric reserve cell calculation system cascading and the like, obviously increase the complexity of the system. Therefore, designing a new, low-cost photovoltaic reservoir computing device to improve the processing performance of the network is an urgent technical problem to be solved.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a reserve pool computing device based on photoelectric double feedback of a Mach-Zehnder modulator. The invention can realize high-performance reserve pool calculation on the premise of not increasing the system cost by utilizing the existing photoelectric oscillation system based on the Mach-Zehnder modulator photoelectric double feedback which can generate more complex chaotic signals. The invention utilizes the structure of combining photoelectric feedback and optical feedback of the Mach-Zehnder modulator to generate complex dynamic response so as to realize the calculation of the photoelectric reserve pool with good network performance.
In order to achieve the aim of the invention, the invention adopts the following technical scheme:
a reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback comprises three modules: the system comprises an input module, a photoelectric dual-feedback reserve tank module and a data acquisition module;
the input module is formed by connecting a signal generator with a combiner;
the photoelectric dual-feedback reserve pool module comprises a continuous optical laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator and a coupler 2 which are sequentially connected, and also comprises an optical feedback branch formed by sequentially connecting a delay optical fiber 1, an optical isolator and an optical amplifier and a photoelectric feedback branch formed by sequentially connecting the delay optical fiber 2, a photoelectric detector and an electric amplifier; the feedback light output by the optical feedback branch is taken as the optical input of the Mach-Zehnder modulator together with the output light of the continuous optical laser through the coupler 1, the feedback electric signal output by the photoelectric feedback branch is taken as the radio frequency input of the Mach-Zehnder modulator, and the output light of the Mach-Zehnder modulator respectively enters the input end of the delay optical fiber 1 of the optical feedback branch and the receiving end of the photoelectric detector of the photoelectric feedback branch through the coupler 2 after passing through the adjustable attenuator;
the data acquisition module is formed by connecting a data acquisition unit with a power divider;
and a combiner of the input module is connected to a connecting circuit between a photoelectric detector in the photoelectric double-feedback reserve tank module and an electric amplifier, and a power divider of the data acquisition module is connected to a connecting circuit between the electric amplifier and a radio frequency input end of the Mach-Zehnder modulator.
Preferably, the continuous-light laser is a semiconductor laser, or other type of continuous-light laser.
Preferably, the optical amplifier is an erbium doped fiber amplifier, a semiconductor optical amplifier or other type of optical amplifier.
Compared with the prior art, the invention has the following obvious prominent substantive features and obvious advantages:
1. the storage pool computing device uses a mode of combining optical feedback and photoelectric feedback, so that the system generates more complex dynamic response, and the network performance is good;
2. the reserve pool computing device has the advantages of simple structure, low power consumption, and simple and easy signal input and data acquisition;
3. the photoelectric reserve cell computing device can improve the processing performance of a network; the invention innovatively develops a novel photoelectric reserve pool computing system by utilizing the existing photoelectric oscillation system based on the Mach-Zehnder modulator photoelectric double feedback, which can generate more complex chaotic signals, and can realize the reserve pool computing with high performance on the premise of not increasing the system cost.
Drawings
FIG. 1 is a schematic diagram of a pool computing device according to the present invention.
FIG. 2 is a graph of test input signals, test output signals, and target output signals for a classification task of a pool computing device in accordance with a preferred embodiment of the present invention.
FIG. 3 is a graph of test output signals, target output signals, and errors between predicted output signals and target output signals for a predictive task of a pool computing device in accordance with a preferred embodiment of the present invention.
FIG. 4 is a graph of test output signals and target output signals for a modeling task of a pool computing device in accordance with a preferred embodiment of the present invention.
Detailed Description
The foregoing aspects are further described in conjunction with specific embodiments, and the following detailed description of preferred embodiments of the present invention is provided:
embodiment one:
in this embodiment, referring to fig. 1, a pool computing device based on mach-zehnder modulator photoelectric dual feedback is characterized by comprising three modules: the system comprises an input module, a photoelectric dual-feedback reserve tank module and a data acquisition module;
the input module is formed by connecting a signal generator with a combiner;
the photoelectric dual-feedback reserve pool module comprises a continuous optical laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator and a coupler 2 which are sequentially connected, and also comprises an optical feedback branch formed by sequentially connecting a delay optical fiber 1, an optical isolator and an optical amplifier and a photoelectric feedback branch formed by sequentially connecting the delay optical fiber 2, a photoelectric detector and an electric amplifier; the feedback light output by the optical feedback branch is taken as the optical input of the Mach-Zehnder modulator together with the output light of the continuous optical laser through the coupler 1, the feedback electric signal output by the photoelectric feedback branch is taken as the radio frequency input of the Mach-Zehnder modulator, and the output light of the Mach-Zehnder modulator respectively enters the input end of the delay optical fiber 1 of the optical feedback branch and the receiving end of the photoelectric detector of the photoelectric feedback branch through the coupler 2 after passing through the adjustable attenuator;
the data acquisition module is formed by connecting a data acquisition unit with a power divider;
and a combiner of the input module is connected to a connecting circuit between a photoelectric detector in the photoelectric double-feedback reserve tank module and an electric amplifier, and a power divider of the data acquisition module is connected to a connecting circuit between the electric amplifier and a radio frequency input end of the Mach-Zehnder modulator.
According to the storage pool computing device based on the Mach-Zehnder modulator photoelectric double feedback, a mode of combining optical feedback and photoelectric feedback is used, so that a system generates more complex dynamic response, and network performance is good. The device of the embodiment has simple structure, low power consumption, and simple and easy signal input and data acquisition.
Embodiment two:
this embodiment is substantially the same as the first embodiment, and is characterized in that:
in this embodiment, the continuous light laser is a semiconductor laser, or other type of continuous light laser;
in this embodiment, the optical amplifier is an erbium doped fiber amplifier, or a semiconductor optical amplifier, or other type of optical amplifier.
The continuous optical laser and the optical amplifier can adopt different devices to realize alternative technical schemes.
Embodiment III:
this embodiment is substantially the same as the above embodiment, and is characterized in that:
in this embodiment, a system model of a pool computing device based on mach-zehnder modulator photoelectric double feedback is built by numerical simulation: wherein t represents time, x (t) represents the output time series of the system, t H And t L Respectively represent the high and low frequency cut-off time of the system, beta 1 Representing the photoelectric feedback gain, T 1 Indicating the photoelectric feedback delay time beta 2 Indicating the optical feedback gain, T 2 Indicating optical feedback delay time,/->Representing the offset phase of the Mach-Zehnder modulator, S (t) representing the input signal, g representing the input gain, and n (t) representing the additive Gaussian white noise introduced by the optical feedback branch. In the absence of input signals, according to parameter beta 1 ，T 1 ，β 2 ，T 2 ，/>The method is characterized in that the influence on the nonlinear state change of the system is achieved by selecting a parameter range calculated by a reserve pool, and numerical simulation and signal processing are realized by MATLAB software on a common microcomputer with an Intel (R) Core (TM) i5-7200U CPU@2.50GHz, an 8GB RAM and a Windows 10 system.
And respectively processing a classification task, a chaotic time sequence prediction task and a modeling reference task. The input signals of the two classification tasks are a plurality of square wave and sine wave signals which are randomly connected, the labels of the square wave and the sine wave are respectively 1 and 0, the sampling point number in one period is set to 12, 1500 data points participate in training, 500 data points participate in testing, and the mask signals adopt [ -1,1]Random distributed uniformly over a rangeThe number of the virtual nodes is set to be 50, and the delay time T of the photoelectric feedback branch circuit 1 For 0.255ns, other parameters are randomly regulated within the range of the calculated parameter values of the reserve pool, the obtained test classification accuracy is 100%, and the test result is shown in fig. 2. The device of the embodiment of the invention has excellent classification performance.
The input signal of the chaotic time series prediction task is SantaFe chaotic time series signal, 3000 data points participate in training, 1000 data points participate in testing, and the mask signal adopts [0,1 ]]Random signals uniformly distributed in the range, the number of virtual nodes is also set to be 50, and the delay time T of the optical feedback branch circuit 2 0.1ns, photoelectric feedback branch delay time T 1 0.255ns, an input gain g of 2.225, and a photo feedback gain beta 1 An optical feedback gain beta of 0.785 2 0.685, offset phaseIn the case of 5 pi/32, the single-step prediction obtained had a test normalized mean square error of 0.0097, and the test results are shown in FIG. 3. The above-described embodiment of the present invention is described as having a first-class predictive performance.
The input signal of the modeling task is at [0,0.5]The real numbers generated randomly according to uniform distribution are output as the result of NARMA10 equation, 1000 data points participate in training, 1000 data points participate in testing, and the mask signal adopts [ -1,1]Random signals uniformly distributed in the range, the number of virtual nodes is also set to be 50, and the delay time T of the optical feedback branch circuit 2 Is 2.2n s Photoelectric feedback branch delay time T 1 0.255ns, an input gain g of 0.425, and a photo feedback gain beta 1 An optical feedback gain beta of 0.68 2 0.73 offset phaseFor 11 pi/64, the normalized mean square error of the test was 0.046, and the test results are shown in FIG. 4. The device has good modeling capability.
The pool computing device based on photoelectric double feedback of the Mach-Zehnder modulator in the embodiment consists of an input module, a photoelectric double feedback pool module and a data acquisition module. The input module consists of a signal generator and a combiner, the photoelectric dual-feedback reservoir module consists of a continuous optical laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator, a coupler 2, an optical feedback branch consisting of a delay optical fiber 1, an optical isolator and an optical amplifier and a photoelectric feedback branch consisting of the delay optical fiber 2, a photoelectric detector and an electric amplifier, feedback light is taken as the optical input of the Mach-Zehnder modulator together with the output light of the continuous optical laser through the coupler 1, feedback electric signals are taken as the radio frequency input of the Mach-Zehnder modulator, and the output light of the modulator enters the optical feedback branch and the photoelectric feedback branch respectively through the coupler 2; the data acquisition module is composed of a power divider and a data acquisition unit. The invention has rich dynamic state and good network performance.
The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the embodiments described above, and various changes, modifications, substitutions, combinations or simplifications made under the spirit and principles of the technical solution of the present invention can be made according to the purpose of the present invention, and all the changes, modifications, substitutions, combinations or simplifications should be equivalent to the substitution, so long as the purpose of the present invention is met, and all the changes are within the scope of the present invention without departing from the technical principles and the inventive concept of the present invention.
Claims (3)
1. A pool computing device based on mach-zehnder modulator photoelectric double feedback, comprising three modules: the system comprises an input module, a photoelectric dual-feedback reserve tank module and a data acquisition module;
the input module is formed by connecting a signal generator with a combiner;
the photoelectric dual-feedback reserve pool module comprises a continuous optical laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator and a coupler 2 which are sequentially connected, and also comprises an optical feedback branch formed by sequentially connecting a delay optical fiber 1, an optical isolator and an optical amplifier and a photoelectric feedback branch formed by sequentially connecting the delay optical fiber 2, a photoelectric detector and an electric amplifier; the feedback light output by the optical feedback branch is taken as the optical input of the Mach-Zehnder modulator together with the output light of the continuous optical laser through the coupler 1, the feedback electric signal output by the photoelectric feedback branch is taken as the radio frequency input of the Mach-Zehnder modulator, and the output light of the Mach-Zehnder modulator respectively enters the input end of the delay optical fiber 1 of the optical feedback branch and the receiving end of the photoelectric detector of the photoelectric feedback branch through the coupler 2 after passing through the adjustable attenuator;
the data acquisition module is formed by connecting a data acquisition unit with a power divider;
and a combiner of the input module is connected to a connecting circuit between a photoelectric detector in the photoelectric double-feedback reserve tank module and an electric amplifier, and a power divider of the data acquisition module is connected to a connecting circuit between the electric amplifier and a radio frequency input end of the Mach-Zehnder modulator.
2. A pool computing device based on mach-zehnder modulator optical-electrical double feedback as defined in claim 1, wherein: the continuous-light laser is a semiconductor laser, or other type of continuous-light laser.
3. A pool computing device based on mach-zehnder modulator optical-electrical double feedback as defined in claim 1, wherein: the optical amplifier is an erbium doped fiber amplifier, a semiconductor optical amplifier, or other type of optical amplifier.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310501914.5A CN116611489A (en) | 2023-05-06 | 2023-05-06 | Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310501914.5A CN116611489A (en) | 2023-05-06 | 2023-05-06 | Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116611489A true CN116611489A (en) | 2023-08-18 |
Family
ID=87675714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310501914.5A Pending CN116611489A (en) | 2023-05-06 | 2023-05-06 | Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116611489A (en) |
-
2023
- 2023-05-06 CN CN202310501914.5A patent/CN116611489A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110247299B (en) | Parallel double-optical feedback semiconductor laser storage pool computing system | |
| CN114358271A (en) | Time-wavelength interweaving photon neural network convolution acceleration chip | |
| CN112269226B (en) | Hybrid modulation method and system | |
| CN103236833A (en) | Photoelectric feedback chaotic laser reservoir parallel calculating device | |
| US11934943B1 (en) | Two-dimensional photonic neural network convolutional acceleration chip based on series connection structure | |
| CN212086198U (en) | Self-adaptive high-precision optical fiber delay system | |
| CN116611489A (en) | Reservoir computing device based on Mach-Zehnder modulator photoelectric double feedback | |
| CN109711612A (en) | Wind power generation power prediction method and device for optimizing echo state network | |
| CN102244551A (en) | Device and method for generating optical code signal | |
| Cai et al. | Research on short-term wind speed prediction based on EEMD-GRU prediction model | |
| CN115579871B (en) | Distribution network with distributed power source penetration and photovoltaic digestion method | |
| CN115021247B (en) | Static voltage stability partition assessment method for large-scale photovoltaic power generation grid-connected system | |
| Sun | Power instability prediction method for wind turbine based on fuzzy decision tree | |
| CN114781636B (en) | Integrated photon storage pool computing device | |
| CN107918920B (en) | Output correlation analysis method for multiple photovoltaic power stations | |
| CN206489177U (en) | System when a kind of multi-channel high-speed is surveyed | |
| CN113904338A (en) | Wind power grid-connected system and frequency characteristic probability load flow calculation method and system | |
| CN116257289A (en) | All-optical reserve pool parallel computing device | |
| Zhang et al. | Design of “Internet+”-Based Intelligent Greenhouse Control System | |
| Sun et al. | Research on Carbon Traceability of Power System Based on Blockchain and Power Flow Calculation under Carbon Peaking and Carbon Neutrality Goals | |
| Lu et al. | Signal Location Method Based on GRU Regression Model for Sagnac Optical Fiber Perimeter Sensing | |
| CN201252548Y (en) | Microwave/millimeter wave generator based on an optical fiber coupler and fiber grating | |
| CN104956260B (en) | A kind of all-optical logic gate | |
| Huang | Line Loss Prediction of Distribution Network Based on BP Neural Network | |
| CN213814216U (en) | All-optical pulse stretcher |
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 |