A kind of optical-electric dual feedback based on Mach-Zehnder modulator optical computing device

Abstract

The application discloses a kind of optical-electric dual feedback based on Mach-Zehnder modulator optical reserve pool computing device, by input module, optical-electric dual feedback reserve pool module and data acquisition module composition.Input module is composed of signal generator and combiner, optical-electric dual feedback reserve pool module is composed of continuous light laser, coupler 1, Mach-Zehnder modulator, adjustable attenuator, coupler 2, by delay optical fiber 1, optical isolator and optical amplifier constitute optical feedback branch and by delay optical fiber 2, photoelectric detector and electric amplifier constitute optical-electric feedback branch, feedback light is together as the optical input of Mach-Zehnder modulator with the output light of continuous light laser by coupler 1, feedback electric signal is as the radio frequency input of Mach-Zehnder modulator, the output light of modulator passes through coupler 2 and enters optical feedback and optical-electric feedback branch respectively;Data acquisition module is composed of power divider and data acquisition unit.The application is dynamic and rich, and has good network performance.

Description

TECHNICAL FIELD

[0001] The application relates to a reservoir computing device based on Mach-Zehnder modulator optoelectronic double feedback, and belongs to the technical field of optoelectronic information processing. BACKGROUND

[0002] An artificial neural network is a machine learning model that simulates the brain to process information. Reservoir computing is a simplified recurrent neural network that is easy to implement in hardware to improve its processing speed. The hardware implementation of traditional reservoir computing requires a large number of physical devices to form a reservoir. The time-delay reservoir computing reduces the hardware requirement of reservoir computing to the extreme by using a time-delay dynamic system and setting a plurality of virtual nodes to replace a plurality of nodes in the reservoir. Only a nonlinear node and a delay feedback loop are needed to implement the time-delay reservoir computing. Especially, the implementation mode of the optical time-delay system not only takes advantage of the characteristics of fast light propagation, but also has the advantage of low power consumption.

[0003] Photon reservoir computing is divided into all-optical reservoir computing and optoelectronic reservoir computing. The all-optical reservoir computing needs to be strictly controlled in terms of temperature and the like, and the sampling rate of signal generation and reception is required to be high. In comparison, the optoelectronic reservoir computing not only takes advantage of the high speed and low power consumption of light, but also has lower requirements for the hardware system and is easy to implement. However, the dynamic characteristics of the existing optoelectronic reservoir still need to be improved. The existing improvement schemes, such as double-node double-optoelectronic feedback and optoelectronic reservoir computing system cascade, obviously increase the complexity of the system. Therefore, it is an urgent technical problem to design a new type of low-cost optoelectronic reservoir computing device to improve the processing performance of the network. SUMMARY

[0004] The application aims to overcome the deficiencies of the prior art and provide an optoelectronic reservoir computing device based on Mach-Zehnder modulator optoelectronic double feedback. The application uses an existing optoelectronic oscillation system based on Mach-Zehnder modulator optoelectronic double feedback that can generate more complex chaotic signals. The optoelectronic oscillation system can realize high-performance reservoir computing without increasing the cost of the system too much. That is, the application uses the structure of combining optoelectronic feedback and optical feedback of the Mach-Zehnder modulator to generate complex dynamic responses, so as to realize the optoelectronic reservoir computing with good network performance.

[0005] To achieve the above-mentioned application purposes, the application adopts the following technical scheme:

[0006] An optoelectronic reservoir computing device based on Mach-Zehnder modulator optoelectronic double feedback comprises three modules: an input module, an optoelectronic double feedback reservoir module and a data acquisition module.

[0007] The input module is composed of a signal generator and a combiner;

[0008] The optoelectronic double-feedback reservoir module comprises a continuous light laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator, a coupler 2 connected in sequence, and further comprises an optical feedback branch composed of a delay optical fiber 1, an optical isolator and an optical amplifier connected in sequence, and an optoelectronic feedback branch composed of a delay optical fiber 2, an optoelectronic detector and an electrical amplifier connected in sequence; the feedback light output by the optical feedback branch and the output light of the continuous light laser are coupled by the coupler 1 to serve as the optical input of the Mach-Zehnder modulator, the feedback electrical signal output by the optoelectronic feedback branch serves as the radio frequency input of the Mach-Zehnder modulator, and the output light of the Mach-Zehnder modulator passes through the adjustable attenuator and then enters the input end of the delay optical fiber 1 of the optical feedback branch and the receiving end of the optoelectronic detector of the optoelectronic feedback branch through the coupler 2;

[0009] The data acquisition module is composed of a data acquisition unit connected with a power divider;

[0010] The combiner of the input module is connected on the connection circuit between the optoelectronic detector and the electrical amplifier in the optoelectronic double-feedback reservoir module, and the power divider of the data acquisition module is connected on the connection circuit between the electrical amplifier and the radio frequency input end of the Mach-Zehnder modulator.

[0011] Preferably, the continuous light laser is a semiconductor laser or other types of continuous light laser.

[0012] Preferably, the optical amplifier is an erbium-doped optical fiber amplifier, a semiconductor optical amplifier or other types of optical amplifier.

[0013] Compared with the prior art, the present application has the following obvious and essential features and advantages:

[0014] 1. The reservoir computing device uses the combination of optical feedback and optoelectronic feedback to make the system produce more complex dynamic response and have good network performance.

[0015] 2. The reservoir computing device has simple structure, low power consumption and simple and easy signal input and data acquisition.

[0016] 3. The optoelectronic reservoir computing device can improve the processing performance of the network; the present application uses the existing optoelectronic oscillation system based on the Mach-Zehnder modulator optoelectronic double feedback which can produce more complex chaotic signals to develop a new type of optoelectronic reservoir computing system, and the system can realize high-performance reservoir computing without increasing the cost of the system too much. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The figure is a structural schematic diagram of the reservoir computing device.

[0018] Figure 2Test input signal, test output signal and target output signal graph of a classification task of a reservoir computing device for a preferred embodiment of the present application.

[0019] Figure 3 Test output signal, target output signal and error between the predicted output signal and the target output signal of a prediction task of a reservoir computing device for a preferred embodiment of the present application.

[0020] Figure 4 Test output signal and target output signal graph of a modeling task of a reservoir computing device for a preferred embodiment of the present application. DETAILED DESCRIPTION

[0021] The above scheme is further described below in combination with specific examples, and the preferred embodiments of the present application are described in detail as follows:

[0022] Embodiment One:

[0023] In this embodiment, referring to Figure 1 A reservoir computing device based on Mach-Zehnder modulator optoelectronic double feedback, characterized in that it comprises three modules: an input module, an optoelectronic double feedback reservoir module and a data acquisition module;

[0024] The input module is composed of a signal generator connected to a combiner;

[0025] The optoelectronic double feedback reservoir module comprises a continuous light laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator and a coupler 2 connected in sequence, and further comprises an optical feedback branch composed of a delay optical fiber 1, an optical isolator and an optical amplifier connected in sequence, and an optoelectronic feedback branch composed of a delay optical fiber 2, an optoelectronic detector and an electrical amplifier connected in sequence; the feedback light output by the optical feedback branch is used as the optical input of the Mach-Zehnder modulator together with the output light of the continuous light laser through the coupler 1, the feedback electrical signal output by the optoelectronic feedback branch is used as the radio frequency input of the Mach-Zehnder modulator, and the output light of the Mach-Zehnder modulator passes through the adjustable attenuator and then enters the input end of the delay optical fiber 1 of the optical feedback branch and the receiving end of the optoelectronic detector of the optoelectronic feedback branch through the coupler 2 respectively;

[0026] The data acquisition module is composed of a data acquisition unit connected to a power divider;

[0027] The combiner of the input module is connected to the connection circuit between the optoelectronic detector and the electrical amplifier in the optoelectronic double feedback reservoir module, and the power divider of the data acquisition module is connected to the connection circuit between the electrical amplifier and the radio frequency input end of the Mach-Zehnder modulator.

[0028] The embodiment is based on the optical-electric double feedback reserve pool computing device of Mach-Zehnder modulator, which uses the combination of optical feedback and optical-electric feedback to make the system produce more complex dynamic response and good network performance.

[0029] Embodiment two:

[0030] The embodiment is basically the same as embodiment one, and the difference is that:

[0031] In the embodiment, the continuous light laser is a semiconductor laser or other types of continuous light laser.

[0032] In the embodiment, the optical amplifier is an erbium-doped fiber amplifier or a semiconductor optical amplifier or other types of optical amplifier.

[0033] The continuous light laser and the optical amplifier in the embodiment can be different devices to realize alternative technical solutions.

[0034] Embodiment three:

[0035] The embodiment is basically the same as the above-mentioned embodiments, and the difference is that:

[0036] In the embodiment, the system model of the reserve pool computing device based on the optical-electric double feedback of Mach-Zehnder modulator is built by numerical simulation. Wherein, t represents time, x(t) represents the output time sequence of the system, t H and t L respectively represent the high and low frequency cutoff time of the system, β1 represents the optical-electric feedback gain, T1 represents the optical-electric feedback delay time, β2 represents the optical feedback gain, T2 represents the optical feedback delay time, represents the bias phase of the Mach-Zehnder modulator, S(t) represents the input signal, g represents the input gain, and n(t) represents the additive white Gaussian noise introduced by the optical-electric feedback branch. In the absence of input signal, according to the parameters β1, T1, β2, T2, The influence of the nonlinear state change of the system on the selection of the parameter range of the reserve pool calculation, numerical simulation and signal processing are realized by MATLAB software on a general microcomputer with Intel(R) Core(TM) i5-7200U CPU@2.50GHz, 8GB RAM and Windows10 system.

[0037] The binary classification task, chaotic time series prediction task and modeling benchmark task are processed respectively. The input signal of the binary classification task is a plurality of square wave and sine wave signals connected randomly, the labels of the square wave and the sine wave are "1" and "0" respectively, the number of sampling points in a period is set to 12, 1500 data points participate in training, 500 data points participate in testing, the mask signal adopts a random signal uniformly distributed in the range of [-1, 1], the number of virtual nodes is set to 50, the delay time T1 of the optoelectronic feedback branch is 0.255 ns, other parameters are adjusted randomly in the parameter value range of the reservoir calculation, the test classification accuracy obtained is 100%, and the test result is shown in Figure 2 . It is shown that the device has excellent classification performance.

[0038] The input signal of the chaotic time series prediction task is a SantaFe chaotic time series signal, 3000 data points participate in training, 1000 data points participate in testing, the mask signal adopts a random signal uniformly distributed in the range of [0, 1], the number of virtual nodes is also set to 50, the delay time T2 of the optical feedback branch is 0.1 ns, the delay time T1 of the optoelectronic feedback branch is 0.255 ns, the input gain g is 2.225, the optoelectronic feedback gain β1 is 0.785, the optical feedback gain β2 is 0.685, and the bias phase is 5π / 32, the test normalized mean square error of single-step prediction obtained is 0.0097, and the test result is shown in Figure 3 . It is shown that the device has excellent prediction performance.

[0039] The input signal of the modeling task is a real number randomly generated in [0, 0.5] according to uniform distribution, the target output is the result of the NARMA10 equation, 1000 data points participate in training, 1000 data points participate in testing, the mask signal also adopts a random signal uniformly distributed in the range of [-1, 1], the number of virtual nodes is also set to 50, the delay time T2 of the optical feedback branch is 2.2 n s , the delay time T1 of the optoelectronic feedback branch is 0.255 ns, the input gain g is 0.425, the optoelectronic feedback gain β1 is 0.68, the optical feedback gain β2 is 0.73, and the bias phase is 11π / 64, the test normalized mean square error obtained is 0.046, and the test result is shown in Figure 4 . The device has good modeling capability.

[0040] The above embodiment is based on the optical-electric double feedback reserve pool computing device of Mach-Zehnder modulator, which is composed of an input module, an optical-electric double feedback reserve pool module and a data acquisition module. The input module is composed of a signal generator and a combiner, the optical-electric double feedback reserve pool module is composed of a continuous light laser, a coupler 1, a Mach-Zehnder modulator, an adjustable attenuator, a coupler 2, an optical feedback branch composed of a delay optical fiber 1, an optical isolator and an optical amplifier, and an optical-electric feedback branch composed of a delay optical fiber 2, an optical-electric detector and an electric amplifier, feedback light is taken together with the output light of the continuous light laser as the optical input of the Mach-Zehnder modulator through the coupler 1, feedback electric signals are taken as the radio frequency input of the Mach-Zehnder modulator, the output light of the modulator enters the optical feedback and optical-electric feedback branches through the coupler 2 respectively, and the data acquisition module is composed of a power divider and a data acquisition unit. The application is dynamic and rich, and has good network performance.

[0041] The above describes the embodiments of the application in combination with the drawings, but the application is not limited to the above embodiments, and can be changed in many ways according to the purpose of the application. Any change, modification, replacement, combination or simplification made according to the spirit and principle of the application should be an equivalent replacement, as long as it meets the purpose of the application and does not deviate from the technical principle and concept of the application.

Claims

1. A reservoir computing device based on Mach-Zehnder modulator photoelectric dual feedback, characterized in that, It includes three modules: an input module, a photoelectric dual feedback reservoir module, and a data acquisition module; The input module consists of a signal generator connected to a combiner; The optoelectronic dual feedback reservoir module includes a continuous-wave laser, coupler 1, Mach-Zehnder modulator, adjustable attenuator, and coupler 2 connected in sequence. It also includes an optical feedback branch consisting of a delay fiber 1, an optical isolator, and an optical amplifier connected in sequence, and an optoelectronic feedback branch consisting of a delay fiber 2, a photodetector, and an electrical amplifier connected in sequence. The feedback light output from the optical feedback branch is used as the optical input of the Mach-Zehnder modulator together with the output light of the continuous-wave laser through coupler 1. The feedback electrical signal output from the optoelectronic feedback branch is used as the radio frequency input of the Mach-Zehnder modulator. The output light of the Mach-Zehnder modulator passes through the adjustable attenuator and then enters the input end of the delay fiber 1 of the optical feedback branch and the receiving end of the photodetector of the optoelectronic feedback branch through coupler 2. The data acquisition module consists of a data acquisition unit connected to a power divider; The photoelectric dual feedback reservoir module has a combiner connected to the input module on the connection circuit between the photodetector and the amplifier, and a power divider connected to the data acquisition module on the connection circuit between the amplifier and the radio frequency input terminal of the Mach-Zehnder modulator.

2. The reservoir computing device based on Mach-Zehnder modulator photoelectric dual feedback according to claim 1, characterized in that: The continuous light laser is a semiconductor laser.

3. The reservoir computing device based on Mach-Zehnder modulator photoelectric dual feedback according to claim 1, characterized in that: The optical amplifier is an erbium-doped fiber amplifier or a semiconductor optical amplifier.

Images