Dynamic optical phase calculation method based on mode control

Abstract

The invention discloses a dynamic optical phase calculation method based on mode control. The method comprises the following steps: carrying out beam combination and shunt processing on reference light and signal light from a sensing system to obtain a first mixed optical signal and a second mixed optical signal; in an optical signal demodulation module, the first mixed optical signal is input into an intermediate layer to be converted and amplified into a first electric signal, and the second mixed optical signal is input into an input layer to be modulated into a second optical signal; inputting the second optical signal into a multi-mode optical waveguide coupled by an electric signal control mode of a middle layer, and applying the first electric signal to an electrode of the multi-mode optical waveguide to dynamically regulate and control coupling among a plurality of guided wave modes in the multi-mode optical waveguide, so that the energy of the second optical signal is redistributed among different modes, and an output optical signal is generated; and the output optical signal is detected, quantized and calculated in the readout layer to obtain a demodulated dynamic phase signal. According to the method, optical phase dynamic change information demodulation can be realized, and the system integration level and the processing efficiency are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of fiber optic sensing and optical computing technology, specifically relating to a dynamic optical phase calculation method based on mode control. Background Technology

[0002] Optical computing technology, especially optical neural networks, has shown great application potential in fields with high demands for information processing speed and throughput due to its advantages such as high speed and low power consumption. Currently, the two mainstream optical computing system architectures—optical neural networks based on optoelectronic hybrid integrated devices and diffraction deep neural networks based on spatial light modulation—have been widely validated in multiple fields such as communication, image processing, data recognition, and classification. The former implements the nonlinear activation function required by optical neural networks in active optical devices with nonlinear responses, completing the optical computing architecture by implementing multiple neural network nodes in an integrated optical chip. Thanks to the maturity of optical integration technology, scalable optical neural networks can already be realized based on this architecture. The latter achieves the coupling of light between multiple diffraction layers, controlling the connection weights between equivalent nodes in the optical network by spatially modulating the amplitude and phase of the light to complete the design of the optical computing network.

[0003] In the field of optical sensing, signal analysis and recognition are crucial steps, forming a vital link between sensing and applications. Especially in high-speed distributed fiber optic sensing systems, it is typically necessary to acquire various characteristics of the probe light in real time, such as amplitude, phase, and frequency. These systems usually involve high measurement speeds, large data volumes, and complex signal demodulation and processing. Traditional methods using photoelectric detection and digital signal processing demodulation place high demands on the electrical components related to detection, analog-to-digital conversion, and digital signal processing. If the traditional approach of first converting the optical signal into an electrical signal using a photodetector and then relying on an analog-to-digital converter and digital signal processor for subsequent demodulation is adopted, extremely stringent requirements are placed on the performance of the electrical components. For example, in high-performance equipment requiring high sensitivity, high spatial resolution, or large dynamic range measurements, the subsequent point signal processing modules are typically large, power-consuming, and costly. Furthermore, optical and electrical components have different characteristics in design, fabrication, and packaging, making the integrated development of a hybrid optoelectronic sensing system challenging, and the current technology is still immature.

[0004] Therefore, implementing more signal processing functions in the optical domain and increasing system integration within the optical domain can significantly reduce system size, weight, and power consumption. It also avoids signal distortion and bandwidth bottlenecks that may be introduced during photoelectric conversion, laying a technological foundation for system applications in resource-constrained or highly mobile scenarios. Considering that signal sensing and transmission in fiber optic sensing systems both rely on optical fiber as a natural carrier, directly embedding optical computing functions into the fiber to achieve, for example, real-time demodulation of phase signals has significant technical compatibility and engineering advantages. This integrated "sensing-transmission-processing" all-fiber architecture not only helps simplify system structure and reduce latency but also further improves the robustness and environmental adaptability of the sensing system. Summary of the Invention

[0005] In view of the above, the purpose of this invention is to provide a dynamic optical phase calculation method based on mode control. This method uses multimode optical fiber with special structure or functional materials as a carrier, constructs a reconfigurable optical neural network inside it, and dynamically adjusts the mode distribution and coupling relationship of the optical field by applying external control signals, thereby optimizing network parameters and realizing real-time and accurate demodulation of dynamic optical phase change information in the optical fiber.

[0006] To achieve the above-mentioned objectives, the present invention provides the following technical solution: This invention provides a dynamic optical phase calculation method based on mode control, comprising the following steps: A reference light from a coherent light source and a signal light returned by an optical fiber sensor are provided. The reference light and the signal light are combined and split to form a first mixed optical signal and a second mixed optical signal. In the optical signal demodulation module, the first mixed optical signal is input to the intermediate layer for conversion and amplification into a first electrical signal, and the second mixed optical signal is input to the input layer to be modulated into a second optical signal; The second optical signal is input into the multimode optical waveguide in the middle layer, which is coupled by an electrical signal control mode, and the first electrical signal is applied to the electrodes of the multimode optical waveguide to dynamically regulate the coupling between multiple waveguide modes inside, so that the energy of the second optical signal is redistributed between different modes to generate an output optical signal. In the readout layer of the optical signal demodulation module, the output optical signal is converted into an electrical signal sequence by multiple detectors, the electrical signal sequence is quantized and calculated, and finally the demodulated dynamic phase signal is output.

[0007] Preferably, the process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: directly inputting the signal light and the reference light into a beam combiner, which outputs the first mixed optical signal and the second mixed optical signal.

[0008] Preferably, the process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: firstly splitting the signal light into two parts by a beam splitter, one part serving as the second mixed optical signal, and the other part entering the beam combiner with the reference light to generate the first mixed optical signal.

[0009] Preferably, the process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: firstly splitting the reference light into two parts by a beam splitter, one part serving as the second mixed optical signal, and the other part entering the beam combiner with the signal light to generate the first mixed optical signal.

[0010] Preferably, the multimode optical waveguide coupled by the electrical signal control mode includes a section of multimode optical fiber, the side of which is processed with through holes, the through holes are filled with a functional material whose refractive index is affected by voltage, and electrodes are provided on both sides of the through holes.

[0011] Preferably, the multimode optical waveguide coupled by the electrical signal control mode is composed of two sections of multimode optical fiber and a middle section of glass capillary. The glass capillary has through holes on its side, which are filled with a functional material whose refractive index is affected by voltage, and electrodes are provided on both sides of the through holes.

[0012] Preferably, the multimode optical waveguide coupled by the electrical signal control mode has a structure of a heterogeneous multimode optical fiber with metal wires embedded inside, and electrodes are processed on the side of the optical fiber at the corresponding positions of the metal wires.

[0013] Preferably, the optical signal demodulation module operates based on the theoretical framework of reservoir calculation; wherein, the input layer implements fixed input weights to map the optical signal to the mode space; the multimode optical waveguide of the intermediate layer constitutes a dynamic reservoir, and nonlinearity is introduced through electronically controlled feedback, and the dynamic evolution process of its internal mode coupling constitutes the state iteration of the reservoir; the readout layer uses the output weights obtained through training to linearly map the output state of the reservoir to calculate the dynamic phase signal.

[0014] Preferably, the training process of the reservoir calculation only trains the weights of the output layer, while the internal parameters of the reservoir remain fixed. During training, the state sequence output by the optical signal demodulation module and the target dynamic phase signal obtained by the traditional demodulation method are collected simultaneously, and the output weights are obtained by solving the linear regression problem. Regularization optimization is used to prevent overfitting, or a recursive least squares algorithm is used to achieve online adaptive weight updates.

[0015] Preferably, the performance of the reservoir is optimized through at least one of the following strategies: real-time monitoring of the reservoir status and adjustment of electronic control parameters to ensure system dynamic stability; establishing a mapping database between electronic control parameters and system performance indicators to determine the parameter combination that allows the reservoir to operate in the optimal dynamic range; adjusting waveguide parameters, cascading multiple electronically controlled multimode waveguides with different parameters, or adding waveguides without electronic control functions between cascaded electronically controlled multimode waveguides to enable the propagation delay between different modes to cover different scales, thereby achieving parallel processing of the input signal on multiple time scales and enhancing the reservoir's ability to process signals with different frequency components.

[0016] Compared with the prior art, the beneficial effects of the present invention include at least the following: (1) This invention directly completes phase demodulation in an optical neural network with optical fiber as the carrier, avoiding the multiple photoelectric conversion steps in the traditional scheme. This not only significantly reduces signal processing delay and provides key support for high-speed sensing systems, but also effectively reduces the additional noise and bandwidth limitations introduced by optoelectronic devices, and improves the throughput and fidelity of signal processing.

[0017] (2) The core of this invention lies in using electrical control signals to dynamically regulate the mode coupling in a multimode optical waveguide, which essentially constructs a dynamic optical reservoir computing system with a feedback mechanism. The inherent nonlinear characteristics of this structure enable it to process complex dynamic phase signals, while its reconfigurability gives the system a strong environmental adaptability. The demodulation performance can be optimized by adjusting the electrical signals without changing the hardware, overcoming the shortcomings of traditional fixed circuits or poor optical path flexibility.

[0018] (3) This invention lays the technical foundation for building a highly integrated all-fiber sensing and processing system by integrating signal processing functions into the optical fiber itself or a compact optical module. This significantly reduces the system size, weight, and power consumption, solves the problems of bulky and power-consuming traditional high-performance demodulation equipment, and expands its application prospects in aerospace, distributed monitoring, and other fields that are sensitive to size and power consumption. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic flowchart of the dynamic optical phase calculation method based on mode control provided in the embodiment; Figure 2This is a schematic diagram of the sensing system to which the mode-controlled dynamic optical phase calculation method provided in the embodiment is applied; Figure 3 This is a schematic diagram of the optical signal demodulation module provided in method 1 of the embodiments; Figure 4 This is a schematic diagram of the optical signal demodulation module provided in method 2 of the embodiment; Figure 5 This is a schematic diagram of the optical signal demodulation module provided in method 3 of the embodiment; Figure 6 This is a schematic diagram of the electrically controlled mode coupled waveguide structure provided in Design 1 of the embodiments; Figure 7 This is the fabrication process of the electrically controlled mode coupled waveguide provided in Design 1 of the embodiment; Figure 8 This is a schematic diagram of the electrically controlled mode coupled waveguide structure provided in Design 2 of the embodiment; Figure 9 This is the fabrication process of the electrically controlled mode coupled waveguide provided in Design 2 of the embodiment; Figure 10 This is a schematic diagram of the electrically controlled mode coupled waveguide structure provided in Design 3 of the embodiment; Figure 11 This is the fabrication process of the electrically controlled mode coupled waveguide provided in Design 3 of the embodiment; Figure 12 This is the system architecture for sensor data calibration and training data acquisition provided in the embodiment. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of this invention.

[0022] The inventive concept of this invention is as follows: Addressing the technical problems of low system integration, high power consumption, and large and costly subsequent signal processing modules caused by the use of optoelectronic hybrid architectures in existing technologies, this invention provides a dynamic optical phase calculation method based on mode control. By processing multimode and few-mode optical waveguides, it enables them to control mode coupling via electrical signals. This function is used to achieve electrically controlled nonlinear mode coupling, which serves as the physical implementation of the nonlinear activation function in an optical neural network. The optical neural network based on this structure is then used for calculating dynamic optical phase signals. This effectively avoids the signal distortion problem caused by multiple optoelectronic conversions in traditional solutions, significantly improving the system integration and energy efficiency ratio, and providing a new technical path for developing miniaturized, low-power, high-speed distributed optical fiber sensing systems.

[0023] like Figure 1 As shown in the figure, the embodiment provides a dynamic optical phase calculation method based on mode control, including the following steps: S1 provides reference light from a coherent light source and signal light returned by an optical fiber sensor, and performs beam combining and splitting of the reference light and signal light to form a first mixed optical signal and a second mixed optical signal.

[0024] In the embodiments, the sensing system used is as follows: Figure 2 As shown, the frequency is Continuous coherent light exits from the coherent source and enters port 1 of the fiber coupler. The fiber coupler splits the input light into two parts, which are then output through ports 2 and 3. Port 1 of the coupler connects to the modulator, and the light output from port 2 serves as the reference input for the optical signal demodulation module. The modulator modulates the continuous coherent light into pulses with a pulse width of [missing value]. The repetition frequency is The optical pulses output from the modulator are amplified by the optical amplifier, then pass through a bandpass filter and enter port 1 of the circulator, and finally enter the fiber optic sensor through port 2. The reflected light from the fiber optic sensor returns to port 2 of the circulator and is then output from port 3, serving as the signal input for the optical signal demodulation module. The optical signal demodulation module demodulates the dynamic phase changes of the optical signal from the fiber optic sensor.

[0025] The modulator includes, but is not limited to, electro-optic modulators, acousto-optic modulators, and microelectromechanical system (MEMS) modulators. The optical amplifier includes, but is not limited to, fiber optic amplifiers and semiconductor optical amplifiers. The bandpass filter can be used with different bandwidths or removed depending on system requirements. If the sensor under test has high reflectivity, both the optical amplifier and the bandpass filter can be omitted.

[0026] Fiber optic sensors can be single sensors, such as a single fiber Bragg grating or a single Fabry-Perot interferometer, or multiple sensors, such as an array of fiber Bragg gratings or Fabry-Perot interferometers. They can also be fully distributed sensors, such as optical fibers and specially treated optical fibers with enhanced scattering signals. If the fiber optic sensor is a single sensor, the system may not need a modulator. If the fiber optic sensor is a discrete sensor array or optical fiber, a modulator is required.

[0027] If the fiber optic sensor is an array of multiple sensors or a single optical fiber, then the length of the sensor... and the repetition frequency of the light pulse Relationships must be satisfied: , in, The speed of light in a vacuum. To detect the equivalent refractive index of light propagating in an optical fiber, This refers to the repetition frequency of the aforementioned probe light pulse.

[0028] If the fiber optic sensor is a discrete sensor array, the width of the optical pulse output by the modulator is... Minimum spacing between sensors in the sensor array Relationships must be satisfied: .

[0029] In this embodiment, one implementation of the optical signal demodulation module is shown in Figure 3 (Method 1), wherein the signal light and the reference light enter a beam combiner, and after passing through the beam combiner, two mixed light paths are formed: a first mixed light signal... Second mixed optical signal Both contain signal light and reference light components. The splitting ratio of the signal light and reference light can be equal or unequal.

[0030] In another embodiment, one implementation of the optical signal demodulation module is shown in Figure 4 (Method 2), wherein the signal light is split into two parts by a beam splitter, one part serving as... The first part enters the input layer, and the second part enters the beam combiner, where it is combined with the reference beam to obtain the final product. .

[0031] In yet another embodiment, one implementation of the optical signal demodulation module is shown in Figure 5 (Method 3), wherein the reference light is split into two parts by a beam splitter, one part serving as... The other part enters the input layer and enters the beam combiner, where it is combined with the signal light to obtain... .

[0032] S2, in the optical signal demodulation module, the first mixed optical signal is input to the intermediate layer for conversion and amplification into the first electrical signal, and the second mixed optical signal is input to the input layer to be modulated into the second optical signal.

[0033] In this embodiment, using the signal light and reference light as two inputs, the dynamic change in the phase difference between the signal light and the reference light is calculated through the following structure and steps: i, the first mixed optical signal After entering the detector, it is converted into a preliminary electrical signal. ,at this time The signal is an interference signal between the reference light and the signal light, and its amplitude is affected by the phase difference between the signal light and the reference light. The impact, namely: ; ii. The initial electrical signal output by the detector The signal then enters an electrical amplifier, where it undergoes linear or nonlinear amplification to obtain the first output electrical signal. , This serves as the electrical control signal for the intermediate layer. The amplifier's gain is denoted as... ,but It can be written as: ;

[0034] iii. Another optical signal passes through the reflective surface and enters the input layer of the optical signal demodulation module. The optical field modulation module of the input layer modulates the input second mixed optical signal. A second optical signal is obtained by spatially modulating the amplitude, phase, or both amplitude and phase to change the transverse field distribution of light. ,Then As the optical input signal of the intermediate layer; iv. The intermediate layer of the optical signal demodulation module is a multimode optical waveguide with mode coupling controlled by electrical signals. The energy of multiple waveguide modes can be transferred through mode coupling. The mode coupling can be dynamically controlled by applying electrical signals to change the optical parameters of the waveguide. After passing through this waveguide, the distribution ratio of total energy among different waveguide modes changes, and the transverse energy field distribution at the output end will change accordingly. , It is the electrical control signal for the input field distribution and intermediate layer. The function, therefore Simultaneously dependent on and ; v, After entering the readout layer, the optical signal is expanded and simultaneously detected by multiple detectors, then converted into an electrical signal sequence. This electrical signal sequence is quantized and multiplied in the readout calculation module to output a dynamic phase signal. .

[0035] S3, the second optical signal is input into the multimode optical waveguide of the intermediate layer, which is coupled by the electrical signal control mode, and the first electrical signal is applied to the electrodes of the multimode optical waveguide to dynamically regulate the coupling between multiple waveguide modes inside, so that the energy of the second optical signal is redistributed between different modes to generate the output optical signal.

[0036] In the embodiment, a structure of a multimode optical waveguide coupled by an electrical signal control mode is as follows: Figure 6 As shown in Design 1, it consists of a multimode optical fiber with through holes, a filler with electrically controllable refractive index, and two electrodes. Its fabrication steps are shown in Figure 7. i. Remove the coating layer from the multimode optical fiber. The methods for removing the coating layer include, but are not limited to, mechanical stripping, etching, and post-irradiation etching. ii. Process through holes from the side of the optical fiber, and the processing methods include, but are not limited to, etching, electron beam etching, etc. iii. Injecting or inserting a material whose optical parameters are affected by voltage into the through-hole, including but not limited to liquid crystal materials, electro-absorbing materials, etc. iv. Electrodes are machined on both sides of the through hole. The machining methods include, but are not limited to, post-sputtering etching.

[0037] In another embodiment, a structure of a multimode optical waveguide coupled by an electrically controlled mode is shown in Figure 8 (Design 2), consisting of two multimode optical waveguide segments, a glass capillary, a filler, and two electrodes. Its fabrication steps are as follows: Figure 9 As shown: i. Take two multimode optical fibers and one glass capillary tube, ensuring that the outer diameters of the three materials are close, and remove the coating. The methods for removing the coating include, but are not limited to, mechanical stripping, etching, and post-irradiation etching. ii. The three materials are fused together to form a three-layer structure of "multimode fiber-capillary-multimode fiber"; iii. Process through holes from the side of the capillary, and the processing methods include, but are not limited to, etching, electron beam etching, etc. iv. Injecting or inserting a material whose optical parameters are affected by voltage into the through-hole, including but not limited to liquid crystal materials, electro-absorbing materials, etc. v. Electrodes are machined on both sides of the through hole. The machining methods include, but are not limited to, etching after sputtering.

[0038] In another embodiment, a structure of a multimode optical waveguide coupled by an electrically controlled mode is shown in Figure 10 (Design 3), consisting of a multimode optical fiber with an implanted metal wire and two electrodes, and its fabrication steps are as follows. Figure 11 As shown: i. Take a heterogeneous multimode fiber and remove the coating layer. Two metal microfilaments are pre-implanted in the heterogeneous fiber. The coating layer can be removed by means including but not limited to mechanical stripping, corrosion, and post-irradiation corrosion. ii. From the holes on both sides of the multimode fiber to the depth where the metal wire is located; iii. Machining electrodes at the location of the holes, including but not limited to welding, sputtering followed by etching, etc.

[0039] S4, in the readout layer of the optical signal demodulation module, converts the output optical signal into an electrical signal sequence through multiple detectors, quantizes and calculates the electrical signal sequence, and finally outputs the demodulated dynamic phase signal.

[0040] In this embodiment, the optical signal demodulation module, based on the theoretical framework of reservoir computing, realizes electrically controlled dynamic nonlinear optical phase demodulation. This module employs the coupled feedback nonlinearity of the electrically controlled signal and the optical mode, as well as a training mechanism based on reservoir computing, as detailed below: i. Feedback nonlinear model of coupling between electronic control signal and optical mode: Electronic control signals Controlled by the input optical signal, forming Dependencies: , in, For the first Control function for each electrode The complex amplitude vectors for each mode represent the electrical signals acting on the intermediate multimode optical waveguide, which are controlled by the interference signal between the input signal light and the reference light. This interference signal, in turn, acts on the optical signals input to the intermediate layer. This creates feedback coupling. This feedback coupling mechanism introduces significant nonlinear effects: , An intermediate-layer electrically controlled mode-coupled multimode optical waveguide constitutes a dynamically reconfigurable reservoir computing system. Considering the dependency between the electrical control signal and the mode state, the mode propagation equation is: , in, For the first The complex amplitude of each mode, For the first The propagation dispersion coefficients of each mode, , The first r、q Complex amplitude of each mode Let be the axial coordinate of the optical waveguide. The linear coupling coefficients for voltage control that depend on the mode state. The equation represents a nonlinear second-order mode coupling tensor. This equation realizes a dynamically controlled first-order and / or second-order nonlinearly coordinated dynamic reservoir with a certain number of adjustable degrees of freedom.

[0041] ii. Physical implementation of the storage pool computing architecture: The optical signal demodulation module constructed in this invention has an input layer that is an optical field modulation module, which modulates the input signal. Mapping to mode space: , in, The input weight matrix is ​​a fixed value, determined by the physical parameters of the light field control module. It is a vector representing the composition of modes excited by the input signal in the intermediate layer.

[0042] The reservoir layer is a dynamic reservoir composed of electrically controlled mode-coupled multimode optical waveguides, and its state evolution vector is: , Where R is the nonlinear dynamic operator of the reserve pool, The recursive connection weight matrix within the reserve pool. To delay the spread, is the waveguide length.

[0043] The output layer readout calculation module implements linear output mapping: , in, This is the observation vector of the reservoir state. To output the weight matrix, the superscript... This is a transpose.

[0044] iii. Characteristics and physical constraints of the storage pool: The system ensures that the core characteristics of the reservoir calculation are met by adjusting the electrical control parameters: by adjusting the voltage range, the spectral radius of the reservoir connection matrix is ​​ensured to meet the requirements. This ensures the system's dynamic stability and echo state characteristics. Second-order nonlinear coupling of the electronic control system. The system's memory capacity is significantly enhanced, enabling the reservoir to retain short-term historical information of the input signal; the coupling loss and propagation loss between modes ensure the gradually decaying memory characteristics, meeting the basic requirements of reservoir computing. The system generates a high-dimensional state space in each mode, and with the electronic control feedback nonlinearity, it achieves rich nonlinear dynamic characteristics.

[0045] iv. Training algorithm based on reservoir computation: Strictly following the training principles of the reservoir calculation, only the output weights are trained, while the reservoir parameters remain fixed: a) Training data collection and usage Figure 12 The synchronous acquisition architecture shown collects the state sequence of the storage pool. and the corresponding target output The acquisition structure simultaneously obtains two sets of data, one of which is... The training input for the model is obtained from another set of data after demodulation using traditional methods. This serves as the fitting target for model training. It should be noted that, because the signal to be tested is a dynamic signal, to maintain a strict correspondence between the loaded signal and the original collected data, we use... Figure 12 The system shown must collect two sets of data simultaneously.

[0046] b) Linear Regression Solution: The core advantage of the reservoir calculation lies in the fact that it only requires solving linear regression problems. , The closed-form solution is: , in, The output weights to be optimized It is the identity matrix. For regularization parameters, , .

[0047] c) Regularization optimization uses ridge regression to prevent overfitting and selects the optimal regularization parameter through cross-validation. We evaluated the generalization performance and optimized the operating point of the reservoir.

[0048] d) Online Adaptation: For dynamic environments, a recursive least squares algorithm is used to implement online weight updates. , in, Let P be the prediction error, and P be the covariance matrix. This represents the number of iterations.

[0049] v. Performance optimization strategies for the reserve pool: To ensure the optimal performance of the reservoir computing system, the method employs the following technical measures for system optimization: (1) By real-time monitoring of the linear independence and dynamic range of the reservoir state vector, the modulation signal is controlled by automatically adjusting the electronic control parameters, i.e., the gain of the electronic amplifier, when the system condition number exceeds the threshold. (2) To improve matrix stability, further adjust the nonlinear parameters of mode coupling; The voltage amplitude, frequency and phase of the signal under test are stored in the electronic control parameter database. The memory capacity, nonlinearity and dynamic range of the reservoir are measured under different parameter configurations through systematic experiments. The parameter settings that make the reservoir work in the optimal dynamic range are determined. These include the number of modes of the multimode waveguide, the magnitude of waveguide dispersion, the length of the waveguide, the rated gain of the electric amplifier, etc. (3) By utilizing the natural time delay distribution generated by different propagation paths in the optical waveguide, by adjusting the waveguide parameters, cascading multiple electronically controlled multimode waveguides with different parameters, and adding waveguides without electronic control function between the cascaded electronically controlled multimode waveguides, the propagation delay between different modes covers different scales, thereby realizing the parallel processing of the input signal on multiple time scales and enhancing the reservoir's ability to process signals with different frequency components.

[0050] The specific embodiments described above illustrate the technical solution and beneficial effects of the present invention in detail. It should be understood that the above description is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic optical phase calculation method based on mode control, characterized in that, Includes the following steps: A reference light from a coherent light source and a signal light returned by an optical fiber sensor are provided. The reference light and the signal light are combined and split to form a first mixed optical signal and a second mixed optical signal. In the optical signal demodulation module, the first mixed optical signal is input to the intermediate layer for conversion and amplification into a first electrical signal, and the second mixed optical signal is input to the input layer to be modulated into a second optical signal; The second optical signal is input into the multimode optical waveguide in the middle layer, which is coupled by an electrical signal control mode, and the first electrical signal is applied to the electrodes of the multimode optical waveguide to dynamically regulate the coupling between multiple waveguide modes inside, so that the energy of the second optical signal is redistributed between different modes to generate an output optical signal. In the readout layer of the optical signal demodulation module, the output optical signal is converted into an electrical signal sequence by multiple detectors, the electrical signal sequence is quantized and calculated, and finally the demodulated dynamic phase signal is output.

2. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: directly inputting the signal light and the reference light into a beam combiner, which outputs the first mixed optical signal and the second mixed optical signal.

3. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: firstly splitting the signal light into two parts via a beam splitter, one part serving as the second mixed optical signal, and the other part entering the beam combiner with the reference light to generate the first mixed optical signal.

4. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The process of combining and splitting the reference light and the signal light to form a first mixed optical signal and a second mixed optical signal includes: firstly splitting the reference light into two parts by a beam splitter, one part serving as the second mixed optical signal, and the other part entering the beam combiner with the signal light to generate the first mixed optical signal.

5. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The multimode optical waveguide coupled by the electrical signal control mode includes a section of multimode optical fiber with through holes processed on the side. The through holes are filled with a functional material whose refractive index is affected by voltage, and electrodes are provided on both sides of the through holes.

6. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The multimode optical waveguide coupled by the electrical signal control mode is composed of two sections of multimode optical fiber and a middle section of glass capillary. The glass capillary has through holes on its side, which are filled with a functional material whose refractive index is affected by voltage. Electrodes are provided on both sides of the through holes.

7. The dynamic optical phase calculation method based on mode control according to claim 1, characterized in that, The multimode optical waveguide coupled by the electrical signal control mode has a structure of a heterogeneous multimode optical fiber with metal wires embedded inside, and electrodes are processed on the side of the optical fiber at the corresponding positions of the metal wires.

8. The dynamic optical phase calculation method based on mode control according to any one of claims 1 to 7, characterized in that, The optical signal demodulation module operates based on the theoretical framework of reservoir calculation. The input layer implements fixed input weights to map the optical signal to the mode space. The multimode optical waveguide in the intermediate layer constitutes a dynamic reservoir. Nonlinearity is introduced through electronically controlled feedback, and the dynamic evolution process of the internal mode coupling constitutes the state iteration of the reservoir. The readout layer uses the output weights obtained through training to linearly map the output state of the reservoir to calculate the dynamic phase signal.

9. The dynamic optical phase calculation method based on mode control according to claim 8, characterized in that, The training process of the reservoir calculation only trains the weights of the output layer, while the internal parameters of the reservoir remain fixed. During training, the state sequence output by the optical signal demodulation module and the target dynamic phase signal obtained by the traditional demodulation method are collected simultaneously. The output weights are obtained by solving the linear regression problem. Regularization optimization is used to prevent overfitting, or a recursive least squares algorithm is used to achieve online adaptive weight updates.

10. The dynamic optical phase calculation method based on mode control according to claim 9, characterized in that, The performance of the storage tank can be optimized by at least one of the following strategies: real-time monitoring of the storage tank status and adjustment of electrical control parameters to ensure the dynamic stability of the system; A mapping database between electronic control parameters and system performance indicators is established to determine the parameter combination that allows the reservoir to operate in the optimal dynamic range. By adjusting waveguide parameters, cascading multiple electronically controlled multimode waveguides with different parameters, or adding waveguides without electronic control functions between cascaded electronically controlled multimode waveguides, the propagation delay between different modes can cover different scales, thereby enabling parallel processing of input signals on multiple time scales and enhancing the reservoir's ability to process signals with different frequency components.

Images