Hybrid spectrometer and spectrum reconstruction method

Abstract

The invention discloses a hybrid spectrometer. The hybrid spectrometer comprises an adjustable weighting unit and a processing unit which are cascaded, the adjustable weighting unit performs weighted stacking on discrete spectral components with uniform stepping characteristics in time sequence to generate a composite optical signal; the processing unit calculates the light intensity based on the composite signal and the stepping characteristic to reconstruct the spectrum. According to the invention, through frequency step scanning and a linear weighting mechanism, the physical bottleneck that wide bandwidth and high resolution are difficult to consider at the same time is broken through. Spectral reconstruction is converted into low-calculation-power linear equation solution, calculation consumption is greatly reduced, meanwhile, algorithm errors are eliminated, and high fidelity is guaranteed. In addition, multi-wavelength analysis is achieved through single-point detection, the system cost and the integration complexity are remarkably reduced, and the system is suitable for high-performance portable spectral analysis.

Description

Technical Field

[0001] This invention relates to the field of spectral analysis, and more specifically to a hybrid spectrometer and a spectral reconstruction method. Background Technology

[0002] As a core tool for analyzing the composition and physical properties of substances, spectrometers have broad application prospects in portable biochemical detection, environmental monitoring, food safety, and biomedicine. With the development of micro-nano fabrication technology, on-chip micro-spectroscopy based on photonic integrated circuits (PICs) has gradually become a research hotspot due to its advantages such as small size, low power consumption, and low cost.

[0003] Current on-chip spectrometers are mainly divided into two categories: dispersive or narrowband filtered structures. The first category is direct-reading spectrometers based on dispersive or narrowband filtering. These spectrometers utilize devices such as arrayed waveguide gratings (AWGs), etched diffraction gratings (EDGs), or microring resonators (MRRs) to spatially disperse incident light or select frequencies through narrowband filtering, and use linear array detectors to receive light intensities at different wavelengths. The principle is similar to the prism dispersion of traditional benchtop spectrometers. However, this type has significant physical limitations: on-chip filtering devices (such as microring resonators) typically have periodic filtering characteristics, limited by the free spectral range. To improve spectral resolution, it is usually necessary to improve the device's quality factor, but this leads to a reduction in the free spectral range, thus significantly limiting the spectrometer's unambiguous operating bandwidth (typically below 100 nm). Achieving high resolution while covering a wide wavelength range is difficult, making a compromise impossible. Secondly, in order to directly analyze the spectrum, a dense array of detectors is usually required to receive each wavelength component separately. This not only increases the integration complexity of the photoelectric detection and readout circuits, but also significantly increases the manufacturing cost.

[0004] Computational spectrometers typically use filtering structures with complex spectral responses (such as speckle or broadband random responses) to sample incident light, acquire the compressed signal through a single-point detector, and then reconstruct the spectrum using algorithms. While this approach reduces the number of hardware channels, it essentially solves a highly underdetermined system of equations (i.e., the number of unknowns far exceeds the number of measurements). To recover the spectrum, compressed sensing or deep learning algorithms are usually required, involving complex optimization processes. As the required spectral resolution increases, the consumption of computational resources grows exponentially, resulting in large data processing delays and high power consumption, making it difficult to meet the real-time requirements of applications. Because the underdetermined solution process is prone to introducing reconstruction errors and artifacts, leading to distortion of the reconstructed spectrum, the reliability and accuracy of the system face challenges in complex spectral environments.

[0005] In summary, existing on-chip spectrometer technologies struggle to simultaneously meet the combined requirements of wide operating bandwidth, high spectral resolution, low computational complexity, and high reconstruction fidelity. Therefore, a novel spectrometer and spectral reconstruction method are urgently needed. Summary of the Invention

[0006] The purpose of this invention is to provide a hybrid spectrometer and spectral reconstruction method to address the problems of existing on-chip spectrometers being limited by the free spectral range of filtering devices, thus failing to achieve both wide operating bandwidth and high spectral resolution, and the high system integration complexity resulting from the reliance on dense detector arrays in traditional spectrometers. Simultaneously, this invention aims to solve the problems of existing computational spectrometers, which suffer from high computational resource consumption, poor real-time performance, and insufficient spectral reconstruction accuracy and stability due to the need to solve underdetermined equations.

[0007] This invention provides an adjustable weighting unit configured to receive multiple sets of discrete spectral components in a time sequence, apply intensity weights to each set of discrete spectral components, and superimpose them to form a composite optical signal; wherein the multiple sets of discrete spectral components correspond to the same incident broadband light, and the discrete spectral components of adjacent sets have a preset uniform step displacement feature in the spectral domain; a processing unit configured to acquire the electrical signal corresponding to the composite optical signal, and calculate the intensity of the discrete spectral components according to the preset uniform step displacement feature and the applied weights, thereby reconstructing the spectrum of the incident broadband light.

[0008] In some embodiments, a discrete sampling unit is further included, configured to filter the incident broadband light to generate the plurality of discrete spectral components; the frequency range of the broadband light is divided into several detection ranges on an average basis; the discrete sampling unit is configured to acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, thereby forming an initial set of discrete spectral components.

[0009] In some embodiments, the discrete sampling unit is configured to synchronously acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, thereby forming an initial set of discrete spectral components.

[0010] In some embodiments, the discrete sampling unit is further configured to perform frequency step scanning starting from the starting frequency position corresponding to each detection range with a preset frequency step size, so that at each frequency position, a set of several discrete spectral components are synchronously acquired from the several detection ranges.

[0011] In some embodiments, the scanning at preset frequency steps covers the bandwidth of a single detection range.

[0012] In some embodiments, the adjustable weighting unit is configured to receive a plurality of discrete spectral components synchronously output by the discrete sampling unit for each frequency position, and to perform transmittance modulation on the plurality of discrete spectral components a plurality of times to perform intensity weighting and superposition on the plurality of discrete spectral components to form a composite optical signal, wherein the number of discrete spectral components output by the discrete sampling unit each time is the same as the number of transmittance modulations corresponding to the performance of transmittance modulation on that set of discrete spectral components.

[0013] In some embodiments, each transmittance modulation employs a set of transmittance combinations that are linearly independent of other transmittance modulations for each of the plurality of discrete spectral components.

[0014] In some embodiments, the processing unit is configured to, for each frequency position and for each transmittance modulation, use the light intensity corresponding to the plurality of discrete spectral components as unknowns, multiply each unknown with the corresponding transmittance in the transmittance modulation, and add them to make it equal to the value of the composite optical signal converted into an electrical signal, thereby establishing a linear equation; obtain the plurality of linear equations established by the plurality of transmittance modulations to form a system of linear equations, and obtain the light intensity of the plurality of discrete spectral components at each frequency position according to the system of linear equations.

[0015] In some embodiments, a photodetector unit is further included, configured to receive the composite optical signal output by the adjustable weighting unit and convert the composite optical signal into an electrical signal and output it to the processing unit.

[0016] In some embodiments, the photodetector unit employs a single broadband photodetector or a photodetector array.

[0017] In some embodiments, the discrete sampling unit is configured with a tuning mechanism to perform the preset frequency step scan. The tuning mechanism causes the transmission spectrum of the discrete sampling unit to be translated on the frequency axis, thereby changing the filter center frequency according to the preset uniform frequency step.

[0018] In some embodiments, the preset frequency step size is less than or equal to half the full width at half maximum (FWHM) of the filtered resonant peak of the discrete sampling unit.

[0019] In some embodiments, the processing unit is adapted to be configured in the form of The matrix equation is as follows: I is an n×1 measurement vector composed of electrical signal values ​​corresponding to n transmittance modulations; X is an n×1 unknown vector composed of light intensities of the n discrete spectral components; A is an n×n transmission matrix, where the k-th row element corresponds to the transmittance of the adjustable weighting unit for the n discrete spectral components during the k-th transmittance modulation.

[0020] In some embodiments, the modulation scheme of the adjustable weighting unit is configured to ensure that the transmission matrix A is a full-rank matrix or a non-singular matrix; the processing unit uses equations The light intensity of the several discrete spectral components is obtained.

[0021] In some embodiments, the processing unit reconstructs the spectrum by sorting and splicing the light intensity data calculated from all frequency positions according to frequency magnitude to form a continuous spectral distribution.

[0022] The present invention also provides a spectral reconstruction method for a hybrid spectrometer, comprising the following steps: Multiple sets of discrete spectral components corresponding to the same incident broadband light are obtained in a time sequence, wherein adjacent sets of discrete spectral components have a preset uniform step displacement feature in the spectral domain; light intensity weights are applied to each set of discrete spectral components and they are superimposed to form a composite optical signal; the electrical signal corresponding to the composite optical signal is obtained; based on the preset uniform step displacement feature and the applied weights, the light intensity of the discrete spectral components is calculated by processing the electrical signals corresponding to the multiple sets of discrete spectral components, thereby reconstructing the spectrum of the incident broadband light.

[0023] In some embodiments, obtaining multiple sets of discrete spectral components includes: controlling the discrete sampling unit of the hybrid spectrometer to be in the initial filtering position, dividing the frequency range of the incident broadband light into several detection ranges on average, and simultaneously obtaining one discrete spectral component from each detection range to obtain a set of initial discrete spectral components composed of several discrete spectral components.

[0024] In some embodiments, forming a composite optical signal and a corresponding electrical signal includes: keeping the filtering center frequency of the discrete sampling unit unchanged, controlling the adjustable weighting unit of the hybrid spectrometer to perform several transmittance modulations on a current set of several discrete spectral components, wherein the transmittance combinations used in the several modulations are linearly independent of each other; during each modulation, detecting the composite optical signal formed by the weighted superposition of the several discrete spectral components, and converting the composite optical signal into an electrical signal.

[0025] In some embodiments, acquiring and processing multiple sets of discrete spectral components to reconstruct the spectrum includes: constructing a system of linear equations based on the electrical signal values ​​corresponding to the several modulations and the known transmittance combination, and solving for the light intensity corresponding to each of the several discrete spectral components in the current set; controlling the discrete sampling unit to change the filter center frequency with a preset frequency step size, entering the next frequency position, and repeatedly performing the formation of a composite optical signal, acquiring its corresponding electrical signal, and calculating the light intensity of the discrete spectral components for each of the several discrete spectral components, until the accumulated scanning range covers the bandwidth of a single detection range; and stitching together the light intensity data corresponding to the multiple sets of discrete spectral components calculated at all frequency positions in order of frequency magnitude to form a continuous spectral distribution.

[0026] In some embodiments, the process of constructing and solving a system of linear equations includes: constructing a system of equations in the form of... The matrix equation; where I is an n×1 measurement vector, whose elements Let X represent the electrical signal value measured during the k-th modulation; X is an n×1 unknown vector, whose elements... The light intensity to be measured is represented by the j-th discrete spectral component; A is an n×n transfer matrix, and the element in the k-th row and j-th column is... This represents the transmittance of the adjustable weighting unit for the j-th discrete spectral component at the k-th modulation; it is calculated... Solving for the light intensity vector X of the n discrete spectral components at the current step position yields the result.

[0027] In some embodiments, the transmission matrix A is obtained through system pre-calibration, including: pre-measuring the spectral response curves of the adjustable weighting unit under different driving conditions, and extracting the transmittance values ​​at the n discrete spectral components corresponding to the current step position.

[0028] As described above, the hybrid spectrometer and spectral reconstruction method of the present invention have the following beneficial effects: By adopting a structure of cascaded discrete sampling units and adjustable weighting units, and combining frequency step scanning and linear equation solving methods, the present invention effectively overcomes the physical bottleneck of traditional on-chip spectrometers, which are limited by the free spectral range and cannot simultaneously achieve wide bandwidth and high resolution, thus realizing continuous high-resolution detection over a wide spectral range. Simultaneously, the present invention transforms the spectral reconstruction process from a complex underdetermined solution into a low-computational-power solution, significantly reducing computational resource consumption and improving real-time performance while avoiding reconstruction errors introduced by the algorithm, ensuring extremely high spectral fidelity. Furthermore, the present invention can achieve multi-wavelength resolution using single-point detection, greatly reducing the integration complexity and manufacturing cost of photoelectric detection arrays, providing a high-performance, low-power solution for portable and highly integrated spectral analysis. Attached Figure Description

[0029] Figure 1 The diagram shown is a schematic representation of the hybrid spectrometer structure of the present invention. Figure 2 The diagram shows a schematic flow chart of the spectral reconstruction method of the hybrid spectrometer of the present invention. Detailed Implementation

[0030] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0031] Please see Figures 1 to 2 It should be noted that the illustrations provided in this embodiment are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0032] This invention provides a hybrid spectrometer, comprising at least: an adjustable weighting unit configured to receive multiple sets of discrete spectral components in a time sequence, apply intensity weights to each set of discrete spectral components and superimpose them to form a composite optical signal; wherein the multiple sets of discrete spectral components correspond to the same incident broadband light, and adjacent sets of discrete spectral components have a preset uniform step displacement feature in the spectral domain; and a processing unit configured to acquire the electrical signal corresponding to the composite optical signal, and calculate the intensity of the discrete spectral components according to the preset uniform step displacement feature and the applied weights, thereby reconstructing the spectrum of the incident broadband light.

[0033] In embodiments of the present invention, the hybrid spectrometer further includes a discrete sampling unit configured to filter the incident broadband light to generate the plurality of discrete spectral components; the frequency range of the broadband light is evenly divided into several detection ranges; the discrete sampling unit is configured to acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, forming an initial set of discrete spectral components. The discrete sampling unit is also configured to synchronously acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, forming an initial set of discrete spectral components. The discrete sampling unit is further configured to perform a frequency-stepped scan starting from the start frequency position corresponding to each detection range with a preset frequency step size, such that at each frequency position, a plurality of discrete spectral components are synchronously acquired from the plurality of detection ranges. The scan with the preset frequency step size covers the bandwidth of a single detection range. The discrete sampling unit is equipped with a tuning mechanism to perform the preset frequency step scanning. The tuning mechanism causes the transmission spectrum of the discrete sampling unit to be translated on the frequency axis, thereby changing the filter center frequency according to the preset uniform frequency step.

[0034] like Figure 1 As shown, Figure 1 The diagram shows a schematic of the hybrid spectrometer structure of the present invention. In this embodiment, the hybrid spectrometer further includes a discrete sampling unit 01, configured to filter the incident broadband light to generate the multiple sets of discrete spectral components. Specifically, the discrete sampling unit 01 preferably employs an optical resonant cavity structure with periodic filtering characteristics, such as a micro-ring resonator (MRR). The input of the micro-ring resonator is coupled to the input waveguide, thereby enabling the discrete sampling unit 01 to naturally form multiple resonant peaks with spacing within the frequency range of the broadband light, thus simultaneously filtering out n discrete wavelength channels from the n detection ranges.

[0035] To achieve the frequency step scanning, the discrete sampling unit is equipped with a tuning mechanism to perform the frequency step scanning. The tuning mechanism includes one of a thermo-optic modulator, an electro-optic modulator, or a mechanical tuning structure, and is configured to change the refractive index, equivalent optical path, or physical shape of the discrete sampling unit by changing temperature, injected current, or mechanical deformation, causing a translation of the transmission spectrum of the discrete sampling unit on the frequency axis, thereby changing the filter center frequency according to the preset uniform frequency step size. A phase modulation structure 03 (i.e., the tuning mechanism) is integrated on the discrete sampling unit 01. The phase modulation structure 03 (e.g., a thermo-optic heating electrode or an electro-optic modulated PN junction) is coupled to the waveguide of the microring resonator and is configured to adjust the effective refractive index of the waveguide material by changing the applied driving signal (e.g., voltage or current). According to the resonance condition, a small change in the effective refractive index will cause a synchronous linear drift of all resonant center frequencies of the microring resonator. Therefore, by precisely controlling the driving signal of the phase modulation structure 03, the center frequencies of the n discrete spectral components can be driven to move finely within their respective detection ranges with a preset uniform step size, thereby completing the scanning coverage of the spectral gaps. The optical signal filtered by the discrete sampling unit 01 is then transmitted to the adjustable weighting unit 02 (and its internal phase modulation structure 04) for subsequent weighting processing.

[0036] The discrete sampling unit 01 preferably employs a microring resonator as its core filtering device, utilizing its inherent periodic resonance characteristics to achieve comb filtering of broadband light. When broadband light passes through the discrete sampling unit 01, the microring resonator generates a series of uniformly spaced resonance peaks in the spectrum. These resonance peaks naturally divide the total frequency range of the broadband light into n consecutive detection intervals, simultaneously filtering out a discrete spectral component of a specific frequency from each interval, thus forming an initial set of discrete spectral components. To achieve continuous detection across the entire spectrum, the discrete sampling unit 01 integrates a phase modulation structure 03.

[0037] like Figure 1As shown, the phase modulation structure 03 is typically a thermo-optical heating electrode or an integrated electro-optical modulation junction covering a micro-ring waveguide. Its working principle is as follows: by controlling the external driving signal (such as voltage or current) applied to the phase modulation structure 03, the temperature or carrier concentration in the micro-ring waveguide region is changed, thereby utilizing the thermo-optical or electro-optical effects of the material to finely adjust the effective refractive index of the waveguide. According to the resonance condition of the micro-ring, the change in the effective refractive index will cause a synchronous drift of the resonant center frequencies of all orders within the resonant cavity. Based on this, the frequency step scanning is achieved by applying a preset stepped driving signal to the phase modulation structure 03. At each step position, the phase modulation structure 03 drives the center frequencies of the aforementioned n discrete spectral components to move by a fine, uniform frequency step within their respective detection ranges. As the scanning continues, when the accumulated frequency shift covers the bandwidth of a single detection range, it means that the i-th resonance peak has been scanned to the position of the (i+1)-th resonance peak at the start of the scan. In this way, by utilizing periodic resonance peaks to scan in parallel within n sub-intervals, seamless coverage and sampling of the entire broadband optical frequency range is achieved. The discrete spectral components with time-series characteristics obtained after being processed by the discrete sampling unit 01 are then transmitted to the adjustable weighting unit 02 (and its internal phase modulation structure 04) for intensity weighting and superposition processing.

[0038] In this embodiment, the preset frequency step size is less than or equal to half the full width at half maximum (FWHM) of the filtered resonant peak of the discrete sampling unit. This parameter setting follows the Nyquist sampling theorem. Specifically, in order to fully capture all high-frequency details in the spectral signal and prevent signal distortion, the sampling frequency must be at least twice the signal bandwidth. Here, the frequency step size is set to be less than or equal to half the FWHM. Based on this setting, although physical spectral overlap leads to high correlation between adjacent sampled data, the adjustable weighting unit introduced in this embodiment can construct a dataset containing redundant information by applying controlled transmittance weights at each step position. The processing unit can utilize this oversampled data to not only construct a full-rank linear equation system but also further suppress random noise through averaging effects using data redundancy, thereby mathematically accurately solving for the high-resolution original spectrum.

[0039] In this embodiment, the adjustable weighting unit is configured to receive several discrete spectral components synchronously output by the discrete sampling unit for each frequency position, and perform several transmittance modulations on the several discrete spectral components to perform intensity weighting and superposition on the several discrete spectral components to form a composite optical signal. The number of discrete spectral components output by the discrete sampling unit each time is the same as the number of transmittance modulations corresponding to the performance of the transmittance modulation of that set of discrete spectral components.

[0040] In this embodiment, each transmittance modulation employs a set of transmittance combinations linearly independent of other transmittance modulations for each of the discrete spectral components. The adjustable weighting unit is configured to receive several discrete spectral components simultaneously output by the discrete sampling unit at each step position, and perform several transmittance modulations on these discrete spectral components. Each transmittance modulation employs a set of transmittances linearly independent of other modulations for each of the discrete spectral components, thereby weighting and superimposing the light intensity of the discrete spectral components to output a composite optical signal. This is the time-domain modulation mechanism of the adjustable weighting unit at a single frequency step position. At each step position of the frequency scan, the input discrete spectral components are temporarily fixed in frequency. At this time, the adjustable weighting unit does not simply allow light to pass through, but enters a rapid modulation cycle. It performs several consecutive sampling operations on these light components. The weighting refers to the attenuation control of the light intensity at a specific wavelength. Each modulation assigns a specific transmittance value (weight) to each of the spectral components. For example, the first component transmits 50%, the second 80%, and so on. This means that the weighting combination used in the first modulation cannot be a simple repetition or linear superposition (e.g., a multiple relationship) of the second or third combination. This ensures that in each of these modulations, the transmittance combination vector points to a different direction in mathematical space, ensuring that each measurement obtains unique information about these spectral components and avoiding information redundancy. The n optical signals after transmittance weighting are mixed (superimposed) together in the physical waveguide, ultimately forming a composite optical signal containing the energy of all components, which is emitted from the output port.

[0041] In this embodiment, the modulation scheme of the adjustable weighting unit is configured to ensure that the transmission matrix A is a full-rank matrix or a non-singular matrix, so that the linear equation system has a unique solution; the processing unit calculates... Solve for the light intensity of the several discrete spectral components.

[0042] Here, several modulation processes are abstracted as a mathematical transmission matrix A. Each row of matrix A represents a modulation operation, and each column represents the transmittance change of a spectral component. Full rank or non-singularity means that several sets of uncorrelated weighted states must be generated. If the matrix is ​​singular (i.e., the determinant is 0), it means that some dimensional information is lost, making it impossible to reconstruct the spectrum. Only when the transmission matrix A is full rank is it invertible. This ensures that for any measured signal, there exists only one set of original spectral intensities corresponding to it, eliminating the uncertainty of multiple solutions or no solution.

[0043] formula This indicates that the solution process is a standard inverse problem, where the measurement results and modulation method are known, and the unknown that needs to be derived in reverse is the original spectral input. Through matrix inversion (or an equivalent linear solution algorithm), the intensity of each independent component can be directly solved from the mixed signal.

[0044] In this embodiment, a photodetector unit is also included, configured to receive the composite optical signal output by the adjustable weighting unit and convert the composite optical signal into an electrical signal for output to the processing unit. The photodetector unit employs a single broadband photodetector or a photodetector array. The photodetector unit linearly converts the energy information (light intensity of the composite optical signal) in the optical domain into an electrical signal (such as voltage or current amplitude), enabling the processing unit to read and calculate it. The detector itself does not possess the ability to split light or distinguish colors; it merely records the total incident photon energy over a period of time. Spectral resolution is achieved through optical modulation in the preceding stage and mathematical operations in the subsequent stage; the detector acts only as an energy integrator. The detector's response band must cover the entire frequency range of the spectrum under test to ensure a response to all spectral components.

[0045] In this embodiment, the processing unit is configured to, for each frequency position and for each transmittance modulation, use the light intensity corresponding to the plurality of discrete spectral components as unknowns, multiply each unknown by the corresponding transmittance in that transmittance modulation, and then add them to make them equal to the value of the composite optical signal converted into an electrical signal, thus establishing a linear equation; obtain the plurality of linear equations established by several transmittance modulations, forming a system of linear equations, and obtain the light intensity of the plurality of discrete spectral components at each frequency position according to the system of linear equations. The processing unit maps the physical process into algebraic equations. The intensity of the plurality of spectral components at different frequencies (denoted as...) , ,..., The transmittance applied in this modulation (let's say it's denoted as 'transmittance') , ,..., The total voltage or current value (denoted as I) read by the detector under this modulation. The equation is: Because there are several unknowns, a single equation based on a single measurement cannot be solved. Therefore, the processing unit collects data from several modulations and establishes a system of simultaneous equations containing the aforementioned linear equations. By mathematically calculating this system of equations, the processing unit can determine the light intensity at several specific frequency points at the current specific frequency scanning step.

[0046] The processing unit is adapted to be configured in the form of The matrix equation is as follows: I is an n×1 measurement vector composed of electrical signal values ​​corresponding to n transmittance modulations; X is an n×1 unknown vector composed of light intensities of the n discrete spectral components; A is an n×n transmission matrix, where the k-th row element corresponds to the transmittance of the adjustable weighting unit for the n discrete spectral components during the k-th transmittance modulation.

[0047] The n voltage or current values ​​measured sequentially in time are arranged vertically by the processing unit to form vector I. This vector represents the system output. The unknown vector X (n×1) represents the physical quantity to be measured, namely the light intensity values ​​of the n discrete spectral components currently entering the system, arranged into vector X, representing the system input. The transfer matrix A (n×n) is the transfer function. The k-th row of the matrix completely records the weighted state of the system for all light components at the k-th time moment (k-th modulation). The elements in the matrix... This represents the transmittance to the j-th spectral component at the k-th modulation.

[0048] The processing unit, when reconstructing the spectrum, is configured to: cache the intensity data of the n discrete spectral components calculated at the current step position; when the accumulated scan range covers a complete detection range, sort and stitch the intensity data calculated at all step positions according to frequency to form a continuous spectral distribution. Since frequency scanning is performed step-by-step, the X vector (containing n intensity values) calculated at each step is only a small part of the final full spectrum. The processing unit temporarily saves these intermediate results. When the total stroke of the frequency step scan reaches a preset detection bandwidth (i.e., covering a detection range), it means that sampling has covered all bands. The processing unit needs to rearrange all cached intensity data points according to the actual physical frequency corresponding to each step position. Through sorting and stitching, the discrete, time-divisionally calculated data points are connected, ultimately reconstructing a complete, high-resolution continuous spectral curve.

[0049] In other embodiments, the calculated light intensity data can be sorted and stitched according to wavelength to finally reconstruct a complete, high-resolution continuous spectral curve.

[0050] like Figure 2 As shown, Figure 2The diagram shows a schematic flow of the spectral reconstruction method for the hybrid spectrometer of the present invention. The present invention also provides a spectral reconstruction method for a hybrid spectrometer, which combines discrete optical sampling and computational reconstruction techniques. This method utilizes a single-point detector to achieve high-throughput multi-wavelength parallel detection. The method includes the following steps: acquiring multiple sets of discrete spectral components corresponding to the same incident broadband light in a temporal sequence, wherein adjacent sets of discrete spectral components have a preset uniform step displacement characteristic in the spectral domain; applying intensity weights to each set of discrete spectral components and superimposing them to form a composite optical signal; acquiring the electrical signal corresponding to the composite optical signal; and calculating the intensity of the discrete spectral components by processing the electrical signals corresponding to the multiple sets of discrete spectral components based on the preset uniform step displacement characteristic and the applied weights, thereby reconstructing the spectrum of the incident broadband light. Multiple sets of discrete spectral components corresponding to the same incident broadband light are acquired in time sequence. Adjacent sets of discrete spectral components have a preset uniform step displacement characteristic in the spectral domain. The system uses minute frequency step scanning to continuously move the discrete sampling window along the spectral axis, enabling high-resolution traversal of the entire measured spectral range. Intensity weights are applied to each set of discrete spectral components and superimposed to form a composite optical signal. The electrical signal corresponding to the composite optical signal is acquired, and a photoelectric conversion device converts the sum of light intensities containing information into voltage or current values ​​readable by a digital processor. Based on the preset uniform step displacement characteristic and applied weights, the intensity of the discrete spectral components is calculated by processing the electrical signals corresponding to the multiple sets of discrete spectral components, thereby reconstructing the spectrum of the incident broadband light. In other words, through a mathematical inversion algorithm, the original intensity information of each wavelength is extracted from the mixed signal, completing the reverse reconstruction from measurement data to spectral distribution.

[0051] In this embodiment, acquiring multiple sets of discrete spectral components includes: controlling the discrete sampling unit of the hybrid spectrometer to be in the initial filtering position, at which point the resonant wavelength of the discrete sampling unit (such as a micro-ring resonator array) is aligned with a preset starting frequency; dividing the frequency range of the incident broadband light into several detection ranges on an average basis, and utilizing the periodic filtering characteristics to logically divide the entire broadband spectrum into several sub-bands, and simultaneously acquiring one discrete spectral component from each detection range to obtain an initial set of discrete spectral components composed of several discrete spectral components. This parallel sampling mechanism allows the system to capture several narrowband signals distributed in different frequency bands at the same time, greatly improving the efficiency of spectral information acquisition compared to traditional single-wavelength scanning.

[0052] In this embodiment, the formation of the composite optical signal and the corresponding electrical signal includes: maintaining the filter center frequency of the discrete sampling unit unchanged, i.e., temporarily fixing the wavelength position of the spectral sampling during the interval of a frequency step scan to construct a stable measurement time window; controlling the adjustable weighting unit of the hybrid spectrometer to perform several transmittance modulations on the current set of discrete spectral components, wherein the transmittance combinations used in the several modulations are linearly independent of each other, and by changing the weighting unit, ensuring that each modulation provides independent constraints for the equation system; during each modulation, detecting the composite optical signal formed by the weighted superposition of the several discrete spectral components, and converting the composite optical signal into an electrical signal. The photodetector acts as an integrator here, recording the linear sum of the light intensities of all wavelengths after specific weighting, and collecting enough information to solve several unknowns through several rapid time series measurements.

[0053] In this embodiment, the acquisition and processing of multiple sets of discrete spectral components to reconstruct the spectrum includes: constructing a system of linear equations based on the electrical signal values ​​corresponding to the several modulations and the known transmittance combination, and solving for the light intensity corresponding to the several discrete spectral components in the current set, thereby achieving multi-wavelength parallel demodulation at a single step position; controlling the discrete sampling unit to change the filter center frequency with a preset uniform frequency step size, entering the next step position, and fine-tuning the resonant cavity through thermo-optic or electro-optic effects, so that several sampling peaks are shifted as a whole by a small step size in the spectrum; and repeatedly performing the process of forming a composite optical signal and acquiring its corresponding frequency points for several new discrete spectral components. The electrical signal and the intensity of the discrete spectral components are calculated until the cumulative scanning range covers the bandwidth of a single detection range. Since several sampling peaks move simultaneously, only the width of one sub-band needs to be scanned to complete the sampling of the entire broadband spectrum through the relay coverage of several peaks, significantly shortening the total scanning time. The intensity data corresponding to the multiple sets of discrete spectral components calculated at all step positions are spliced ​​together in order of frequency magnitude to form a continuous spectral distribution. The processing unit reorders and fuses the data segments calculated sequentially in the time domain and interleaved in the frequency domain according to their physical frequency coordinates, and finally outputs a complete high-resolution spectral curve.

[0054] In this embodiment, the process of constructing and solving the linear equation system includes: constructing the system in the form of... The matrix equation; where I is an n×1 measurement vector, whose elements Let X represent the electrical signal value measured during the k-th modulation; X is an n×1 unknown vector, whose elements... Let represent the intensity of the j-th discrete spectral component to be measured, and represent the unknown input of the system; A is an n×n transfer matrix, and the element in the k-th row and j-th column is denoted as . This represents the transmittance of the adjustable weighting unit for the j-th discrete spectral component at the k-th modulation; it is calculated... The light intensity vector X of the n discrete spectral components at the current step position is obtained by solving. The original spectral signal X is then accurately separated from the mixed observations I using matrix inversion, provided that the modulation strategy ensures that matrix A is full-rank and well-state.

[0055] In this embodiment, the transfer matrix A is obtained through pre-calibration of the system, ensuring reconstruction accuracy and eliminating the influence of hardware manufacturing errors and environmental factors. This process includes: pre-measuring the spectral response curves of the adjustable weighting unit under different driving conditions; that is, during the factory or calibration stage, using a tunable laser to scan and establish a database of precise mapping relationships between driving voltage and transmittance, and extracting the transmittance values ​​corresponding to several discrete spectral components at the current step position. During actual measurement, precise transfer coefficients are obtained by looking up tables or interpolation from this database, thereby constructing a high-precision transfer matrix A in real time, ensuring the accuracy of the equation solution.

[0056] This invention utilizes a cascaded structure of discrete sampling units and adjustable weighting units, combined with frequency step scanning and linear equation solving methods. This effectively overcomes the physical bottleneck of traditional on-chip spectrometers, which are limited by the free spectral range and cannot simultaneously achieve wide bandwidth and high resolution, thus realizing continuous high-resolution detection over a wide spectral range. Simultaneously, this invention transforms the spectral reconstruction process from a complex underdetermined solution into a low-computational-power solution. This significantly reduces computational resource consumption and improves real-time performance while avoiding reconstruction errors introduced by the algorithm, ensuring extremely high spectral fidelity. Furthermore, this invention achieves multi-wavelength resolution using single-point detection, greatly reducing the integration complexity and manufacturing cost of photoelectric detection arrays, providing a high-performance, low-power solution for portable and highly integrated spectral analysis. Therefore, this invention effectively overcomes the various shortcomings of existing technologies and has high industrial application value.

[0057] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A hybrid spectrometer, characterized in that, At least including: An adjustable weighting unit is configured to receive multiple sets of discrete spectral components in the receiving time sequence, apply light intensity weights to each set of discrete spectral components and superimpose them to form a composite optical signal; wherein, the multiple sets of discrete spectral components correspond to the same incident broadband light, and the discrete spectral components of adjacent sets have a preset uniform step displacement feature in the spectral domain. The processing unit is configured to acquire the electrical signal corresponding to the composite optical signal, and calculate the light intensity of the discrete spectral components according to the preset uniform step displacement characteristics and the applied weights, thereby reconstructing the spectrum of the incident broadband light.

2. The hybrid spectrometer according to claim 1, characterized in that, It also includes a discrete sampling unit configured to filter the incident broadband light to generate the multiple sets of discrete spectral components; the frequency range of the broadband light is divided into several detection ranges on an average basis; the discrete sampling unit is configured to acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, thereby forming an initial set of discrete spectral components.

3. The hybrid spectrometer according to claim 2, characterized in that, The discrete sampling unit is configured to synchronously acquire one discrete spectral component from each detection range at an initial filtering position corresponding to the start frequency of each detection range, thereby forming a set of initial discrete spectral components.

4. The hybrid spectrometer according to claim 2, characterized in that, The discrete sampling unit is further configured to perform frequency step scanning starting from the starting frequency position corresponding to each detection range with a preset frequency step size, so that at each frequency position, a set of several discrete spectral components are synchronously acquired from the several detection ranges.

5. The hybrid spectrometer according to claim 4, characterized in that, The bandwidth of a single detection range is scanned at a preset frequency step size.

6. The hybrid spectrometer according to claim 3, characterized in that, The adjustable weighting unit is configured to receive several discrete spectral components synchronously output by the discrete sampling unit for each frequency position, and to perform several transmittance modulations on the several discrete spectral components to perform intensity weighting and superposition on the several discrete spectral components to form a composite optical signal, wherein the number of discrete spectral components output by the discrete sampling unit each time is the same as the number of transmittance modulations corresponding to the performance of transmittance modulation on that set of discrete spectral components.

7. The hybrid spectrometer according to claim 6, characterized in that, Each time the transmittance modulation is applied to the plurality of discrete spectral components, a set of transmittance combinations that are linearly independent of the other transmittance modulations are used.

8. The hybrid spectrometer according to claim 6, characterized in that, The processing unit is configured to, for each frequency position and for each transmittance modulation, use the light intensity corresponding to the several discrete spectral components as unknowns, multiply each unknown by the corresponding transmittance in the transmittance modulation, and add them together to make it equal to the value of the composite optical signal converted into an electrical signal, thereby establishing a linear equation; obtain several linear equations established by several transmittance modulations to form a system of linear equations, and obtain the light intensity of the several discrete spectral components at each frequency position according to the system of linear equations.

9. The hybrid spectrometer according to claim 1, characterized in that, It also includes a photoelectric detection unit, configured to receive the composite optical signal output by the adjustable weighting unit, and convert the composite optical signal into an electrical signal and output it to the processing unit.

10. The hybrid spectrometer according to claim 9, characterized in that, The photoelectric detection unit employs a single broadband photodetector or an array of photodetectors.

11. The hybrid spectrometer according to claim 3, characterized in that, The discrete sampling unit is equipped with a tuning mechanism to perform the preset frequency step scanning. The tuning mechanism causes the transmission spectrum of the discrete sampling unit to be translated on the frequency axis, thereby changing the filter center frequency according to the preset uniform frequency step.

12. The hybrid spectrometer according to claim 4, characterized in that, The preset frequency step size is less than or equal to half the full width at half maximum (FWHM) of the filtered resonant peak of the discrete sampling unit.

13. The hybrid spectrometer according to claim 6, characterized in that, The processing unit is adapted to be configured in the form of The matrix equation is as follows: I is an n×1 measurement vector composed of electrical signal values ​​corresponding to n transmittance modulations; X is an n×1 unknown vector composed of light intensities of the n discrete spectral components; A is an n×n transmission matrix, where the k-th row element corresponds to the transmittance of the adjustable weighting unit for the n discrete spectral components during the k-th transmittance modulation.

14. The hybrid spectrometer according to claim 6, characterized in that, The modulation scheme of the adjustable weighting unit is configured to ensure that the transmission matrix A is a full-rank matrix or a non-singular matrix; the processing unit uses equations The light intensity of the several discrete spectral components is obtained.

15. The hybrid spectrometer according to claim 1, characterized in that, The processing unit reconstructs the spectrum by sorting and splicing the light intensity data calculated from all frequency positions according to frequency magnitude to form a continuous spectral distribution.

16. A spectral reconstruction method for a hybrid spectrometer, characterized in that, Includes the following steps: Multiple sets of discrete spectral components corresponding to the same incident broadband light in time sequence are obtained, wherein the discrete spectral components of adjacent sets have a preset uniform step displacement feature in the spectral domain. Each set of discrete spectral components is weighted by light intensity and then superimposed to form a composite optical signal. Obtain the electrical signal corresponding to the composite optical signal; Based on the preset uniform step displacement characteristics and applied weights, the light intensity of the discrete spectral components is calculated by processing the electrical signals corresponding to the multiple sets of discrete spectral components, thereby reconstructing the spectrum of the incident broadband light.

17. The spectral reconstruction method for a hybrid spectrometer according to claim 16, characterized in that, The acquisition of multiple sets of discrete spectral components includes: controlling the discrete sampling unit of the hybrid spectrometer to be in the initial filtering position, dividing the frequency range of the incident broadband light into several detection ranges on average, and simultaneously acquiring one discrete spectral component from each detection range to obtain a set of initial discrete spectral components composed of several discrete spectral components.

18. The spectral reconstruction method for a hybrid spectrometer according to claim 16, characterized in that, The process of forming a composite optical signal and the corresponding electrical signal includes: keeping the filtering center frequency of the discrete sampling unit unchanged, controlling the adjustable weighting unit of the hybrid spectrometer to perform several transmittance modulations on a current set of several discrete spectral components, wherein the transmittance combinations used in the several modulations are linearly independent of each other; during each modulation, detecting the composite optical signal formed by the weighted superposition of the several discrete spectral components, and converting the composite optical signal into an electrical signal.

19. The spectral reconstruction method for a hybrid spectrometer according to claim 16, characterized in that, The acquisition and processing of multiple sets of discrete spectral components to reconstruct the spectrum includes: constructing a system of linear equations based on the electrical signal values ​​corresponding to the several modulations and the known transmittance combination, and solving for the light intensity corresponding to each of the discrete spectral components in the current set; controlling the discrete sampling unit to change the filter center frequency with a preset frequency step size, entering the next frequency position, and repeatedly performing the process of forming a composite optical signal, acquiring its corresponding electrical signal, and calculating the light intensity of the discrete spectral components for each of the several discrete spectral components until the accumulated scanning range covers the bandwidth of a single detection range; and stitching together the light intensity data corresponding to the multiple sets of discrete spectral components calculated at all frequency positions in order of frequency magnitude to form a continuous spectral distribution.

20. The spectral reconstruction method for a hybrid spectrometer according to claim 16, characterized in that, The process of constructing and solving the system of linear equations includes: constructing the system in the form of... The matrix equation; where I is an n×1 measurement vector, whose elements Let X represent the electrical signal value measured during the k-th modulation; X is an n×1 unknown vector, whose elements... Represents the intensity of the j-th discrete spectral component; A is an n×n transfer matrix, and its element in the k-th row and j-th column is... This represents the transmittance of the adjustable weighting unit for the j-th discrete spectral component at the k-th modulation; it is calculated... Solving for the light intensity vector X of the n discrete spectral components at the current step position yields the result.

21. The spectral reconstruction method for a hybrid spectrometer according to claim 16, characterized in that, The transmission matrix A is obtained through system pre-calibration, including: pre-measuring the spectral response curves of the adjustable weighting unit under different driving conditions, and extracting the transmittance values ​​at the n discrete spectral components corresponding to the current step position.

Images