A TOSA optical path active alignment method and system

Abstract

This invention discloses a TOSA (Optical Oriented Array Alignment) optical path active alignment method and system, relating to the field of optoelectronic device alignment technology. The invention drives the TOSA device to output a test optical signal and collects the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned. Based on the acquisition dimension parameters and optical path communication transmission status parameters, an adaptive search strategy is generated. The effectiveness of the generated adaptive search strategy is monitored, and a decision is made on whether to adopt an optimization strategy. Based on sample alignment data, a joint evaluation is performed and the search path is dynamically corrected to determine the target alignment position of the device to be aligned. This helps to solve the problem that existing methods struggle to simultaneously consider historical sample utilization efficiency, search path optimization, and the rapid and accurate determination of the optimal alignment position of the current device in scenarios where there are many acquisition dimensions and dynamically changing communication transmission statuses in the TOSA optical path active alignment process.

Description

Technical Field

[0001] This invention relates to the field of optoelectronic device alignment technology, and in particular to a TOSA optical path active alignment method and system. Background Technology

[0002] In optical communication transmitter modules, the TOSA (Transmitter Optical Sub-Assembly) is a crucial optoelectronic component for converting electrical signals into optical signals and outputting them. It typically consists of a laser chip, lens, isolator, sleeve, pigtail, and related support structures. The TOSA optical path refers to the transmission path of the light beam emitted from the laser or light-emitting diode, which sequentially undergoes optical transmission processes such as collimation, focusing, isolation, or coupling before finally entering the optical fiber or target waveguide. The quality of this optical path construction directly affects the transmitted optical power, coupling efficiency, return loss, and transmission stability, and is one of the key factors determining the performance and packaging quality of the optical module.

[0003] Because the optical components in the TOSA optical path are small in size and have limited assembly spacing, and because beam transmission is highly sensitive to spatial position deviations, any positional offset, angular deviation, or focal length mismatch between the laser output end, lens center, isolator optical axis, and fiber coupling end can easily lead to increased coupling loss, decreased output power, or even component mismatch. Therefore, to ensure the transmission accuracy and coupling effect of the TOSA optical path, precise alignment of related optical components is usually required during the packaging manufacturing process. Among these methods, active alignment, by monitoring the coupled light power or related optical feedback signals in real time during the emission state and dynamically adjusting the relative positions of the components accordingly, can more intuitively reflect the actual optical path state and has thus become an important alignment method in existing TOSA packaging processes.

[0004] The existing active alignment process is as follows: In existing technologies, active alignment of TOSA or similar optical emitting components typically does not rely solely on point-by-point light finding of a single device, but includes module installation, test optical signal output, coupling state detection, position parameter acquisition, coordinated position adjustment, predictive assisted search, and fixed retesting. Specifically, the light source sub-module, optical sub-module, and coupling end are first installed on the corresponding platform, driving the light source to output a test optical signal and acquiring power feedback and position parameters under the current coupling state. Then, by adjusting the position and distance between the light source sub-module and the optical sub-module, the overall optical path is actively optimized, or a predictive model trained with sample alignment data is used to output the position parameters to be moved and guide the relative position adjustment of the current device. After reaching the preset coupling index, fixing and retesting are performed to complete the active alignment. This type of existing technology shows that the active alignment process has gradually evolved from a simple axis-by-axis search relying solely on real-time power feedback to a comprehensive alignment method combining module-level coordinated adjustment and historical sample data-assisted search.

[0005] In existing technologies, the components work together to form a complete test optical path: In the signal output component, the driving power supply provides a stable preset driving current for the laser emitting unit, and the signal control unit generates and outputs an appropriate modulation signal and enable control signal, which together drive the laser emitting unit to emit a test optical signal that meets the test requirements; the laser emitting unit is installed on the light source mount of the light source sub-module. After the test optical signal is exported from the light source sub-module, it is transmitted to the optical sub-module, where the lens collimates the beam, the isolator suppresses reflected light interference, and then it is transmitted to the coupling end through the optical path fixed by the optical bench. Finally, it is received by the silicon photonics chip coupler end, forming a detectable complete test optical path, which provides a stable and standard optical signal basis for subsequent alignment parameter acquisition, search strategy generation, and target position determination.

[0006] Based on the aforementioned active alignment concept, existing technologies are not limited to independently adjusting the position of a single light-emitting device or a single-stage coupling. Instead, they have gradually developed active alignment schemes oriented towards module-level optical path relationships. For example, the optical emission sub-module structure and its active alignment method disclosed in Chinese invention patent application CN117518371A include: a light source sub-module, which includes a light source component assembled on a light source holder; an optical sub-module, which includes an optical component assembled on an optical bench; and a silicon chip, which includes a coupler. Light emitted by the light source component is received along the optical path by the coupler via the optical component. The light source holder and the optical bench conduct heat for heat dissipation. The light source sub-module and the optical sub-module are actively aligned simultaneously, and their positions and distances are flexibly adjusted during the active alignment process.

[0007] For example, Chinese invention patent application CN105319655B discloses an automatic coupling method and system for optical integrated chips and optical fiber assemblies, comprising: constructing a neural network including an input layer, a hidden layer, and an output layer; pre-acquiring multiple sets of sample parameter values ​​for aligning the optical fiber assembly with multiple sample optical integrated chips; training the neural network with the multiple sets of sample parameter values; acquiring multiple sets of actual parameter values ​​for the optical fiber assembly and the optical integrated chip to be aligned, with each set of actual parameter values ​​being input to the neural network as an input value to the neural network input layer, and obtaining an output value including a set of output parameters from the neural network output layer; and adjusting the position of the optical fiber assembly relative to the optical integrated chip according to the position parameters to be moved.

[0008] During the active alignment process of the TOSA optical path, when outputting test optical signals and adjusting and aligning the position based on coupling state detection results, position parameters, and prediction results, the need to acquire sample alignment data from multiple angles among devices in the same batch, as well as fluctuations in optical communication transmission indicators, may lead to deviations between the actual optimal alignment position of the current device and the historical search center, predicted output position, or instantaneous power peak position. This results in a large search range and insufficient alignment efficiency. The specific reasons are as follows: On the one hand, to obtain sample alignment data that can be used for predictive search, multiple rounds of sampling are usually required for the same batch of devices at different angles, relative positions, and coupling states. This results in a large number of sample collection dimensions and a cumbersome data acquisition process. Furthermore, the historical search center or predicted output position formed based on such samples is more of a statistical reference result and is difficult to accurately characterize the individual deviations and instantaneous optimal alignment positions of the current devices under actual assembly and adjustment conditions. On the other hand, during active alignment, optical communication transmission indicators may fluctuate with changes in transmission state, coupling state, and detection conditions. This means that the instantaneously detected power peak position, predicted output position, and the actual optimal position that meets communication transmission requirements are not completely consistent. This can easily lead to a lengthy search path, an unreasonable search order, and a deviation in target position determination, thus affecting the efficiency of active alignment.

[0009] Existing methods typically rely on real-time power feedback for stepwise searching or on prediction results trained on sample data to guide the search. The former is prone to lengthy alignment processes due to dispersed search starting points and large scanning ranges, while the latter is prone to deviations in target position guidance due to insufficient representation of common patterns among devices in the same batch and the individual state of the current device by the prediction results. Thus, it is difficult to simultaneously achieve both search efficiency and accuracy in target position determination.

[0010] In summary, during the active alignment process of the TOSA optical path, there may be scenarios where there are many dimensions of sample collection for the same batch of devices and the communication transmission status changes dynamically. Existing methods are difficult to balance the efficiency of historical sample utilization, search path optimization, and the rapid and accurate determination of the optimal alignment position of the current device. Summary of the Invention

[0011] This invention provides a TOSA optical path active alignment method and system. The technical solution provided by this application is as follows: Firstly, a TOSA optical path active alignment method is provided. The specific implementation of this method is as follows: S1, the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned are acquired; S2, an adaptive search strategy is generated based on the acquisition dimension parameters and optical path communication transmission status parameters, the effectiveness of the generated adaptive search strategy is monitored, and it is decided whether to adopt the search strategy optimization strategy; S3, joint evaluation is performed based on sample alignment data and the search path is dynamically corrected, and the target alignment position of the device to be aligned is determined.

[0012] Secondly, a TOSA optical path active alignment system is provided. This system applies a TOSA optical path active alignment method, including: a TOSA device alignment parameter acquisition module, used to acquire the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned; a search strategy generation module, used to generate an adaptive search strategy based on the acquisition dimension parameters and optical path communication transmission status parameters, monitor the effectiveness of the generated adaptive search strategy, and decide whether to adopt the search strategy optimization strategy; and a device alignment position acquisition module, used to perform joint evaluation based on sample alignment data and dynamically correct the search path, and determine the target alignment position of the device to be aligned.

[0013] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. By collecting the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned, a comprehensive and accurate characterization of the individual characteristics and real-time operating conditions of the device can be achieved, providing reliable data support for subsequent search strategy generation and path correction. Based on the acquisition dimension parameters and optical path communication transmission status parameters, an adaptive search strategy is generated. Monitoring the effectiveness of the generated adaptive search strategy and deciding whether to adopt a search strategy optimization strategy helps improve the adaptability and operational efficiency of the search strategy to the current operating conditions, avoiding ineffective searches caused by strategy rigidity, thereby shortening the active alignment time and ensuring rapid locking of the target alignment position. Joint evaluation based on sample alignment data and dynamic correction of the search path, as well as determining the target alignment position of the device to be aligned, helps improve the accuracy of target alignment position determination, thus improving the consistency of alignment among devices in the same batch.

[0014] 2. Based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, the acquisition dimension complexity level and communication transmission status level are determined respectively. Based on the qualified level set, the corresponding initial adaptive search strategy is matched. Compared with the shortcomings of low initial search efficiency in existing technologies, this helps to improve the matching accuracy of the initial search strategy, realize the accurate adaptation of the search strategy with the current acquisition complexity and transmission status, and better ensure the search efficiency in the initial stage of active alignment, reduce invalid search links, and lay a solid foundation for subsequent path optimization and location locking.

[0015] 3. Based on the sample alignment data, the power evaluation results corresponding to each candidate position are determined, and the search path that does not meet the preset change trend is dynamically corrected. When the change trend data meets the qualified change trend conditions, the target alignment position of the current device to be aligned is determined based on the change trend data. When it does not meet the conditions, the search path is dynamically corrected. Compared with the shortcomings of the target position convergence lag in the existing technology, it helps to realize the adaptive dynamic optimization of the search path, accurately guides the search direction to converge towards the optimal target area, and is more conducive to ensuring the accuracy and stability of the target alignment position, while taking into account the efficiency and accuracy of active alignment.

[0016] 4. When a candidate position in a certain search direction shows a continuous increase in coupled optical power, but the corresponding extinction ratio and signal-to-noise ratio do not improve synchronously, or even show increased fluctuations, if the obtained trend consistency parameter is not greater than the corresponding qualified trend threshold, a TOSA optical path search fluctuation alarm is sent to the preset control center. Otherwise, it continues to judge whether the path convergence parameter, which represents the degree of convergence of the current search path towards the target area, is within the corresponding convergence interval. Compared with the shortcomings of existing technologies that rely on coupled optical power indicators, this helps to realize timely identification and early warning of abnormal working conditions during the alignment process, thereby facilitating the controllability and robustness of the active alignment process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a TOSA optical path active alignment method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of two-dimensional light propagation provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of three-dimensional light propagation provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the adaptive search strategy matching provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the search strategy optimization strategy provided in the embodiments of the present invention; Figure 6 This is a schematic diagram of the structure of a TOSA optical path active alignment system provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0020] Before providing a detailed explanation of the embodiments of this application, the application scenarios of these embodiments will be described first.

[0021] The above-disclosed embodiments are merely some examples of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

[0022] In the first embodiment, this invention provides a TOSA optical path active alignment method. For example... Figure 1 The flowchart shown is a TOSA optical path active alignment method. The processing flow of this method may include the following steps: S1, TOSA device alignment parameter acquisition: During the active alignment of the TOSA optical path, the TOSA device is driven to output a test optical signal, and the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned are acquired. The acquisition dimension parameters represent a set of data used to characterize the number of dimensions and the complexity of dimension combinations during the acquisition of samples from the same batch of devices. The optical path communication transmission status parameters represent a set of data used to characterize the quality of optical signal transmission. By acquiring the alignment parameters of the TOSA device, the comprehensiveness and accuracy of the alignment parameters can be improved, and the parameter acquisition deviation caused by individual differences of devices in the same batch and fluctuations in optical transmission status can be reduced. This helps to provide reliable data support for the subsequent search strategy generation and avoids deviation in search direction and misjudgment of target position due to missing or biased parameters.

[0023] S2, Search Strategy Generation: An adaptive search strategy is generated based on the acquired dimension parameters and optical path communication transmission status parameters. The effectiveness of the generated adaptive search strategy is monitored, and a decision is made on whether to adopt the search strategy optimization strategy. Through search strategy generation, the adaptability and rationality of the search process are ensured, dynamically adapting to the individual state of the device to be aligned and real-time optical transmission changes. This provides efficient path guidance for the rapid acquisition of the device alignment position, reduces invalid search steps, and improves the overall efficiency of the alignment process.

[0024] S3, Device Alignment Position Acquisition: Based on sample alignment data, joint evaluation is performed and the search path is dynamically corrected to gradually converge the search direction towards the target area that meets the preset communication transmission requirements, and the target alignment position of the device to be aligned is determined. By acquiring the device alignment position, it is possible to quickly and accurately lock the optimal target alignment position of the device to be aligned, ensuring that the alignment result meets the preset communication transmission index and coupling efficiency requirements. Among them, the sample alignment data refers to the data set collected, recorded and used for search strategy generation, search effect evaluation and search path correction during the active alignment process of the TOSA optical path to characterize the optical path coupling state, communication transmission state and search path change characteristics under different candidate positions. This includes, but is not limited to, adjustment coordinates, adjustment angles, axial spacing, etc. in various directions.

[0025] The embodiments of this invention provide a database that includes a series of parameters, including pre-stored acquisition dimension threshold ranges, qualified trend thresholds, etc. The data in this database provides a basis for parameter determination, search strategy generation, search path correction, and abnormal alarm triggering for a TOSA optical path active alignment method.

[0026] It should be noted that the process of constructing parameters such as the pre-stored threshold range for collection dimensions or the qualified trend threshold is as follows: Collect sample data generated during the historical active alignment process of the TOSA optical path. The sample data includes at least the acquisition dimension parameters, trend consistency parameters, and the corresponding final alignment results. The alignment results are used to characterize whether the corresponding sample data meets the preset alignment requirements. Then, the sample data is divided into qualified sample sets and unqualified sample sets according to the final alignment results. Statistical analysis is performed on various acquisition dimension parameters in the qualified sample set to obtain the mean and standard deviation of the corresponding acquisition dimension parameters. The pre-stored acquisition dimension threshold range corresponding to each acquisition dimension parameter is determined by adding or subtracting a preset multiple of the standard deviation from the mean. For example, the number of acquisition angles in the qualified sample set is statistically analyzed, and its mean is 8 and standard deviation is 2. When the preset multiple is 1.5, the corresponding threshold range of the number of acquisition angles can be determined to be 5 to 11. Similarly, the distribution statistics of the trend consistency parameters corresponding to the qualified and unqualified sample sets can be performed to obtain the cross-distribution boundary of the trend consistency parameters in the two types of sample sets, the point of maximum classification accuracy, or the preset quantile. Any one of these results is determined as the qualified trend threshold. Finally, the constructed pre-stored acquisition dimension threshold range and qualified trend threshold are written into a preset database for subsequent determination of acquisition dimension complexity level, search path change trend, and alarm triggering conditions during the TOSA optical path active alignment process.

[0027] The TOSA device alignment parameter acquisition, search strategy generation, and device alignment position acquisition provided in this invention application are closely related, specifically in the following aspects: First, the synergistic effect of data-driven and strategy generation: TOSA device alignment parameter acquisition provides a high-dimensional, individualized raw data foundation for search strategy generation, enabling adaptive search strategies to be dynamically constructed based on the real-time state of the device and environmental fluctuations. This effectively avoids the universal bias caused by relying solely on historical statistical data, realizing a shift from passive response to proactive prediction. Second, the synergistic effect of strategy guidance and path optimization: Search strategy generation provides clear directional guidance and an efficient search paradigm for device alignment position acquisition. By monitoring the effectiveness of the strategy in real time and dynamically optimizing it, the approach time to the target area is greatly shortened, solving the problems of excessively large search range and lengthy paths in existing technologies, and significantly improving positioning efficiency. Third, the synergistic effect of position feedback and system closed loop: The device alignment position acquisition result is not only the final alignment target, but also a direct verification and feedback of the accuracy of parameter acquisition and the rationality of strategy generation.

[0028] For example, before actively aligning the same batch of TOSA devices, in order to establish a sample library that can be used for predictive search, it is usually necessary to select several sample devices and collect corresponding sample data under conditions such as lens fine-tuning around different angles, light source sub-modules and optical sub-modules being in different relative positions, and coupling light power being in different states. Assuming that during the initial modeling of a certain batch of devices, it is necessary to record the coupling power and communication indicators under multiple rotation angles, multiple axial positions, and multiple lateral offset positions, the sample acquisition process often requires repeating multiple rounds of light finding, measurement, recording, and correction operations. Although this can obtain the historical search center or predicted output position, such results are more based on the reference position formed by the overall statistical characteristics of the same batch of devices. For the device to be aligned, if there are slight differences between the laser center emission position or the initial installation state of the coupling end and the sample device, the historical search center may only be able to roughly indicate the search direction, but cannot directly correspond to the current device's instantaneous optimal alignment position. In the end, it is still necessary to continue searching within a large range. However, through the interaction between TOSA device alignment parameter acquisition, search strategy generation and device alignment position, it is helpful to combine the real-time parameters of the current device to dynamically correct the reference deviation of the historical samples, combine the statistical commonalities of the sample data with the individual characteristics of the current device, thereby quickly narrowing the search range, optimizing the search path, and achieving precise locking of the optimal alignment position, while taking into account both alignment efficiency and alignment accuracy.

[0029] like Figure 2The diagram shown is a two-dimensional light propagation schematic provided by an embodiment of this invention: the horizontal axis represents the position Z [mm] in the direction of light propagation, and the vertical axis represents the position Y [mm] of the light in the vertical direction. Different colored lines in the diagram represent light propagation paths at different incident positions or angles, gray outlines represent optical elements, and the vertical line on the right can be understood as the target receiving surface, image plane, or reference plane. As can be seen from the diagram, after entering from the left, the light passes through two sets of optical elements and undergoes refraction during propagation, causing the direction of light propagation at different initial positions to change in subsequent space.

[0030] like Figure 3 The figure shows a three-dimensional light propagation diagram provided in an embodiment of the present invention: As can be seen in the figure, multiple beams of light of different colors enter from the left, pass through two sets of lens-like optical elements in sequence, and then propagate to the target surface on the right. The large disk on the right can be understood as the receiving surface, the detection surface, the reference target surface, or the coupling target position. The figure shows the propagation state of multiple beams of light after passing through the two sets of optical elements in sequence in space, as well as the spatial distribution of the beams in the direction of the target surface, which is used to illustrate the overall structural relationship of the emitted light path and the beam transmission path.

[0031] Furthermore, the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned are collected. The specific process is as follows: During the test optical path output process, the coupling status of the test optical path is monitored to obtain the acquisition dimension parameters, and the transmission quality corresponding to the test optical signal is monitored to obtain the optical path communication transmission status parameters. The acquisition dimension parameters include one or more of the number of acquisition angles and the number of position sampling points. The number of acquisition angles refers to the number of different angle states actually acquired by the TOSA device to be aligned based on the counter monitoring during the acquisition process of the same batch of device samples in order to obtain alignment data under different posture conditions. Different angle states may include one or more of the angle values ​​corresponding to pitch angle, yaw angle, or roll angle. The number of position sampling points refers to the number of position points actually collected within a preset search range for different relative adjustment coordinates of the TOSA device to be aligned, based on counter monitoring, in order to obtain alignment data under different spatial position conditions during the sample acquisition process of the same batch of devices. Different relative adjustment coordinates may include one or more of the X-direction position, Y-direction position, and Z-direction position. Optical path communication transmission status parameters include one or more of the received optical power and optical path signal-to-noise ratio. Received optical power refers to the actual optical power received at the coupling end under the current optical path transmission conditions based on the monitoring of the optical power meter. Optical path signal-to-noise ratio refers to the ratio between the optical signal power and noise power monitored by the optical power meter and the spectrum analyzer.

[0032] It should be added that, by controlling the signal output component, a preset drive current, modulation signal, and enable control signal are applied to the laser emitting unit in the TOSA device, so that the TOSA device outputs the corresponding test optical signal, which is then formed into the test optical path to be tested via the light source sub-module, optical sub-module, and coupling terminal. The signal output component includes a drive power supply and a signal control unit. The signal control unit is a control module used to generate, adjust, and output preset modulation signals and enable control signals, and can work with the drive power supply to accurately control the electrical signal parameters applied to the laser emitting unit. It includes a signal generator, modulator, control chip, and signal conditioning circuit, etc. The laser emitting unit includes light-emitting devices such as lasers. The light source sub-module includes a light source base and a laser emitting unit. The optical sub-module includes lenses, isolators, and optical benches, etc. The coupling terminal includes silicon photonics chip coupler terminals, etc.

[0033] like Figure 4 The diagram illustrates an adaptive search strategy matching method provided in an embodiment of this invention. The adaptive search strategy is generated based on acquisition dimension parameters and optical path communication transmission status parameters. Specifically, the acquisition dimension complexity level and communication transmission status level are determined based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, respectively. An initial adaptive search strategy is then matched based on a set of acceptable levels. The specific process is as follows: When all acquisition dimension parameters are within the corresponding pre-stored acquisition dimension threshold range, the corresponding acquisition dimension complexity level is classified as a first-level complexity level; when some acquisition dimension parameters are not within the corresponding pre-stored acquisition dimension threshold range, the corresponding acquisition dimension complexity level is classified as a second-level complexity level, etc. Level 1: When all optical path communication transmission status parameters are within the corresponding pre-stored optical path communication threshold range (including, but not limited to, the received optical power threshold range corresponding to received optical power and the optical path signal-to-noise ratio threshold range corresponding to optical path signal-to-noise ratio), the corresponding communication transmission status level is classified as Level 1 optical path communication level. When some optical path communication transmission status parameters are not within the corresponding pre-stored optical path communication threshold range, the corresponding communication transmission status level is classified as Level 2 optical path communication level. The set of levels consisting of Level 1 complexity level and Level 1 optical path communication level is recorded as the qualified level set, and the set of levels consisting of other acquisition dimension complexity level and communication transmission status level is recorded as the unqualified level set.

[0034] Based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, the acquisition dimension complexity level and communication transmission status level are determined respectively, and a corresponding initial adaptive search strategy is matched. That is, when all acquisition dimension parameters are within the corresponding pre-set pre-stored acquisition dimension threshold range, the corresponding acquisition dimension complexity level is classified as level one complexity level. When there are acquisition dimension parameters outside the corresponding pre-stored acquisition dimension threshold range, the corresponding acquisition dimension complexity level is classified as level two complexity level. This helps to improve the accuracy of acquisition dimension complexity determination, quickly distinguish the complexity of sample acquisition scenarios, and thus provide a quantitative basis for the accurate matching of the initial adaptive search strategy. It avoids search strategy mismatch caused by acquisition dimension complexity being too high or too low, reduces invalid search time, and improves the targeting of the initial search.

[0035] When all optical path communication transmission status parameters are within the corresponding pre-set, pre-stored optical path communication threshold range, the corresponding communication transmission status level is classified as Level 1 optical path communication level. When there are optical path communication transmission status parameters that are not within the corresponding pre-stored optical path communication threshold range, the corresponding communication transmission status level is classified as Level 2 optical path communication level. This is beneficial for dynamically adjusting the focus of the search strategy according to the transmission status level, ensuring that the search process is both efficient and accurate, and avoiding alignment deviations caused by ignoring transmission status fluctuations.

[0036] The effectiveness of the generated adaptive search strategy is monitored, and a search strategy optimization strategy is decided upon. This means that based on the type of the detected grade set, a search strategy optimization strategy is selectively adopted. The specific process is as follows: When an unqualified grade set is detected, the search strategy optimization strategy is initiated in the next preset search time period; when a qualified grade set is detected, the initially set adaptive search strategy is used to search, and the sample alignment data is jointly evaluated and the search path is dynamically corrected. The adaptive search strategy includes at least one or more of the following: search direction, search window range, and search order. The search direction refers to the relative alignment between the driving light source submodule and the optical submodule during the TOSA optical path active alignment process. During position adjustment, the selected displacement adjustment direction includes a set of adjustment directions used to guide optical path coupling optimization, such as the lateral offset direction, the axial approach or distance direction, and the lens rotation angle direction. The search window range refers to the spatial or angular range used to limit the relative position adjustment interval of the device. The search order refers to the execution order of each search direction and adjustment dimension dynamically determined according to preset priority or real-time operating conditions. The search strategy optimization strategy is used to specifically modify the search window range and search order in the initially set adaptive search strategy based on the anomaly sources corresponding to the set of unqualified levels, in order to reduce the proportion of invalid search areas, reduce the number of repeated samplings, and improve the search efficiency of the target alignment position.

[0037] like Figure 5 The diagram shows a search strategy optimization strategy provided in an embodiment of this invention: when the set of unqualified levels is found to consist only of level 2 complexity levels, a complexity-constrained search strategy optimization is performed; when the set of unqualified levels is found to consist only of level 2 optical communication levels, a communication-priority search strategy optimization is performed; and when the set of unqualified levels is found to include both level 2 complexity levels and level 2 optical communication levels, a joint correction search strategy optimization is performed.

[0038] Specifically, the search strategy optimization process is as follows: Further judgment is made based on the set of unqualified levels; when the set of unqualified levels is found to consist only of second-level complexity levels, it is determined that the current TOSA device to be aligned mainly has an excessive acquisition dimension complexity problem, and a complexity-constrained search strategy optimization is executed; The complexity-constrained search strategy optimization is as follows: the sample alignment data corresponding to the second-level complexity level is input into the search window optimization model, the optimized search window size is output and a search is performed, and a joint evaluation is conducted based on the sample alignment data, and the search path is dynamically corrected.

[0039] When the set of unqualified levels is detected to consist only of level 2 optical path communication levels, it is determined that the current TOSA device to be aligned mainly has an abnormal optical path communication transmission status problem, and a communication-priority search strategy optimization is executed. Specifically, the sample alignment data corresponding to the level 2 optical path communication level is input into the search order optimization model, the search order is output and the search is performed, and a joint evaluation is performed based on the sample alignment data and the search path is dynamically corrected.

[0040] When the set of non-compliant levels is detected to include both Level 2 complexity and Level 2 optical communication, it is determined that the current TOSA device to be aligned has both problems of exceeding the limit of acquisition dimensional complexity and abnormal communication transmission status. A joint corrective search strategy optimization is then performed. The joint corrective search strategy optimization means that the search is performed using the obtained joint search window size and joint search order, and the search path is dynamically corrected based on the sample alignment data. The joint search window size and joint search order are obtained by inputting the sample alignment data corresponding to both Level 2 complexity and Level 2 optical communication into the search window and outputting it to the order optimization model.

[0041] This invention provides a method for active alignment of optical paths using TOSA, which pre-constructs a class of models trained using existing machine learning algorithms. These models and their corresponding training data include, but are not limited to: Search window optimization model and corresponding training data: sample alignment data corresponding to level 2 complexity and preset search window size; search order optimization model and corresponding training data: sample alignment data corresponding to level 2 optical communication and preset search order; search window and order optimization model and corresponding training data: sample alignment data corresponding to both level 2 complexity and level 2 optical communication, preset joint search window size and joint search order; search performance evaluation model and corresponding training data: sample alignment data and preset trend data; search path correction model and corresponding training data: trend data and preset search order and search window size.

[0042] The aforementioned model possesses the ability to learn, fit, and output the mapping relationship between input data and target parameters. Based on sample alignment data with different anomaly levels or search states, it can adaptively output search window size, search order, etc., that are compatible with the current TOSA active alignment process. In specific applications, the sample alignment data corresponding to the second-level complexity level needs to be input into the search window optimization model to output the search window size; the sample alignment data corresponding to the second-level optical path communication level needs to be input into the search order optimization model to output the search order; the sample alignment data corresponding to both the second-level complexity level and the second-level optical path communication level needs to be input into the search window and order optimization model to output the joint search window size and joint search order; the sample alignment data needs to be input into the search effect evaluation model to output trend data; and the trend data needs to be input into the search path correction model to output the search order and search window size.

[0043] Taking the construction process of the search window optimization model as an example, it is constructed using the random forest algorithm. The specific process is as follows: First, historical sample alignment data corresponding to the second level of complexity is obtained, and each group of sample alignment data is labeled with a pre-set search window size to form a model training sample set. Then, outlier removal, missing value completion, and feature normalization are performed on the training sample set to extract input features that characterize the dimensional complexity of sample collection and search difficulty, and the pre-set search window size is used as the output label. Subsequently, the mapping relationship between input features and output labels is trained using the random forest algorithm. By constructing multiple decision trees and integrating the output results of each decision tree, the search window optimization model is obtained. Finally, the accuracy of the constructed search window optimization model is verified using validation samples. When the output result meets the preset error requirements, the construction of the search window optimization model is completed.

[0044] When a set of non-compliant grades is detected, a search strategy optimization strategy is initiated. This involves outputting the corresponding search window size or search order based on the composition of the non-compliant grade set. Compared to existing methods that blindly expand the search range, have rigid and fixed search parameters, lack targeted optimization, and are prone to getting trapped in local optima, this approach helps improve the adaptability and targeting of the search strategy, achieving precise optimization of the search path. This significantly shortens alignment time and reduces invalid search intervals. When a set of compliant grades is detected, an initially set adaptive search strategy is used to search, jointly evaluating sample alignment data and dynamically correcting the search path. Compared to existing technologies with redundant search paths and target position determination biases, this approach is more conducive to improving the convergence speed and rationality of the search path, taking into account the reference value of historical samples and the real-time status of the current device. This enables rapid and accurate locking of the optimal alignment position, ensuring the consistency and stability of alignment for devices in the same batch.

[0045] In this embodiment, the acquisition dimension complexity level and communication transmission status level are determined based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, respectively. The initial adaptive search strategy is matched with the qualified level set, which helps to improve the adaptability of the initial search strategy to the current working condition of the device to be aligned, avoids resource waste and path redundancy caused by indiscriminate search, and thus ensures the initial search efficiency of active alignment, laying the foundation for rapid target location locking in the future. Moreover, based on the type of the monitored level set, the search strategy optimization strategy is selectively adopted, which helps to dynamically adapt to the real-time changes in the working condition, correct search deviations in a timely manner, and ensure that the search strategy is always in a state of efficient operation.

[0046] It is worth mentioning that there is a correlation between generating adaptive search strategies and monitoring the effectiveness of the generated adaptive search strategies, as follows: Generating adaptive search strategies is the prerequisite and foundation for monitoring their effectiveness. Only by first generating an initial adaptive search strategy based on the matching of the level set can the judgment criteria and evaluation dimensions for effectiveness monitoring be clarified, ensuring that the monitoring results are targeted. On the other hand, monitoring the effectiveness of adaptive search strategies is the core basis for optimizing search strategies and improving strategy adaptability. The type and degree of effectiveness deviations detected can guide the direction of strategy optimization in reverse, avoiding blind optimization. At the same time, the monitoring results can also verify the rationality of the initial search strategy matching.

[0047] Furthermore, a joint evaluation and dynamic correction of the search path are performed based on the sample alignment data. Specifically, the power evaluation results corresponding to each candidate position are determined based on the sample alignment data, and search paths that do not meet the preset trend are dynamically corrected. The specific process is as follows: First, the sample alignment data is input into a pre-built search performance evaluation model, which outputs the trend data corresponding to the current search path. The trend data includes a power trend parameter characterizing the change in coupled optical power along the current search path and a communication trend parameter characterizing the change in optical path communication transmission status along the current search path. The power trend parameter is represented by the coupled optical power fluctuation trend, which is a statistical measure of the difference in coupled optical power between adjacent candidate positions within a preset local window. The communication trend parameter is represented by the change in optical path signal-to-noise ratio (SNR), which is a statistical measure of the difference in optical path SNR between adjacent candidate positions within a preset local window. It should be noted that candidate positions represent the set of positions to be evaluated formed by the current TOSA device under different relative adjustment coordinates, different angle states, or different distance states under the guidance of the adaptive search strategy.

[0048] Then, it is determined whether the trend data corresponding to the output search path meets the qualified trend conditions, that is, whether the trend data is all within the corresponding preset range. If it meets the conditions, the current search direction is determined to be valid, and the target alignment position of the device to be aligned is determined based on the trend data. If it does not meet the conditions, it is determined whether the current search direction is invalid, and dynamic correction of the search path is performed. Dynamic correction of the search path is used to reduce the proportion of invalid search areas, suppress the risk of misjudgment caused by the inconsistency between power trend and communication trend, and improve the positioning accuracy of the target alignment position.

[0049] Determining the target alignment position of the device to be aligned based on trend data means inputting the corresponding trend data into a preset neural network model and outputting the corresponding target alignment position to guide the relative position adjustment of the current TOSA device, that is, adjusting the TOSA device to be aligned to the output target alignment position. For example, a BP (BackPropagation) neural network model is used. After being trained and optimized with historical sample alignment data of the same batch of devices, this model can accurately explore the nonlinear mapping relationship between trend data and target alignment position, adapting to the needs of dynamic changes in working conditions during the active alignment of the TOSA optical path. This model uses error backpropagation as its core principle. It calculates the network output result through forward propagation and updates the weights and bias parameters of each layer of neurons in reverse layer by layer based on the error between the output result and the actual label, continuously iterating to reduce the prediction error, thereby establishing a stable nonlinear mapping relationship between trend data and target alignment position. This can effectively adapt to the complex working conditions of dynamic fluctuations in transmission status and individual differences in devices during the active alignment of the TOSA optical path, and achieve accurate prediction output of the target alignment position.

[0050] Specifically, the dynamic correction of the search path is as follows: In the next preset search time period, the change trend data is input into the preset search path correction model, the search order and search window size are output, the search is performed again with the output search order and search window size, and the change trend data corresponding to the search path is checked to see if it meets the qualified change trend conditions. If it does, the target alignment position of the device to be aligned is determined based on the change trend data. If it still does not meet the conditions, a search path alarm is pushed to the preset control center.

[0051] like Figure 6 The diagram shows a schematic of a TOSA optical path active alignment system provided in an embodiment of this invention. The system includes: a TOSA device alignment parameter acquisition module, used to drive the TOSA device to output test optical signals during active alignment of the TOSA optical path, and to acquire the acquisition dimension parameters and optical path communication transmission status parameters of the current TOSA device to be aligned. The acquisition dimension parameters represent a set of data used to characterize the number of dimensions and the complexity of dimension combinations during the acquisition of samples from the same batch of devices. The optical path communication transmission status parameters represent a set of data used to characterize the quality of optical signal transmission. A search strategy generation module is used to generate an adaptive search strategy based on the acquisition dimension parameters and optical path communication transmission status parameters, monitor the effectiveness of the generated adaptive search strategy, and decide whether to adopt a search strategy optimization strategy. A device alignment position acquisition module is used to perform joint evaluation based on sample alignment data and dynamically correct the search path, so that the search direction gradually converges towards the target area that meets the preset communication transmission requirements, and determines the target alignment position of the current device to be aligned. The candidate positions represent the set of evaluation positions formed by the current TOSA device to be aligned under different relative adjustment coordinates, different angle states, or different distance states under the guidance of the adaptive search strategy.

[0052] In this embodiment, the search path is dynamically corrected by performing joint evaluation based on sample alignment data. The obtained trend data is used to further determine whether the qualified trend conditions are met. If they are met, the target alignment position of the device to be aligned is determined based on the trend data. Otherwise, the search path is dynamically corrected. This improves the convergence speed of the search path and reduces invalid search steps. It also helps to avoid errors in target position determination caused by historical sample deviations and fluctuations in optical transmission status. This ensures that the target alignment position matches the individual characteristics of the device to be aligned, balancing the efficiency and accuracy of active alignment.

[0053] By adopting dynamic search path correction, that is, re-searching with the output search order and search window size, and checking whether the change trend data corresponding to the search path meets the qualified change trend conditions, compared with the shortcomings of the target position convergence lag that usually exists in existing methods, it is more conducive to achieving adaptive adjustment of the search path, accurately matching the dynamic changes of the current acquisition dimension complexity and optical transmission state, thereby ensuring that the search process always approaches the optimal target area.

[0054] In the second embodiment, based on the method provided in the first embodiment, although the candidate positions in a certain search direction show a continuous increase in coupled optical power, the corresponding extinction ratio and signal-to-noise ratio do not improve synchronously, and even show increased fluctuations. In this case, if convergence is continued solely based on the power trend, it is easy to misjudge the locally more powerful positions as the target alignment positions. Therefore, a joint evaluation is performed based on sample alignment data, and the search path is dynamically corrected. The specific process is as follows: Obtain the trend consistency parameter, which characterizes the degree of matching between the power trend parameter and the communication trend parameter; the trend consistency parameter is obtained by calculating the correlation coefficient between the power trend parameter and the communication trend parameter using a correlation analysis algorithm, such as the Pearson correlation coefficient analysis algorithm, and is represented by the Pearson correlation coefficient; when the trend consistency parameter is not greater than the corresponding qualified trend threshold, it indicates that the degree of matching between the trend of coupled optical power change and the trend of optical path communication transmission status change on the current search path is low, indicating that there is a power imbalance in the current search direction. If the trend and communication trend are inconsistent, the current search direction is deemed invalid, and a TOSA optical path search fluctuation alarm is sent to the preset control center. When the trend consistency parameter is greater than the corresponding qualified trend threshold, it is further determined whether the path convergence parameter, which represents the convergence degree of the current search path towards the target area, is within the corresponding pre-set convergence interval. If so, the current search direction is deemed valid, and the target alignment position of the device to be aligned is determined based on the change trend data. Otherwise, the current search direction is deemed invalid, and a TOSA optical path active alignment alarm is sent to the preset control center. The path convergence parameter is represented by the total decrease in distance between multiple consecutive candidate positions on the current search path and the center of the target area. The total decrease refers to the cumulative sum of all distance reductions monitored by the grating ruler displacement sensor in the distance changes of multiple consecutive candidate positions on the current search path relative to the center of the target area, which is used to characterize the overall convergence degree of the current search path towards the center of the target area.

[0055] In this embodiment, when a candidate position in a certain search direction shows a continuous increase in coupled optical power, but the corresponding extinction ratio and signal-to-noise ratio do not improve synchronously, or even show increased fluctuations, a joint evaluation is performed based on sample alignment data, and the search path is dynamically corrected. When the trend consistency parameter is greater than the corresponding qualified trend threshold, it is further determined whether the path convergence parameter, which characterizes the degree of convergence of the current search path towards the target region, is within the corresponding convergence interval. This helps to achieve accurate identification and dynamic calibration of the search path, reduce alignment deviations caused by misleading single coupled optical power indicators, and thus take into account both coupling efficiency and optical communication transmission quality requirements. This ensures that the finally determined target alignment position meets both the coupling power index and the preset optical communication transmission standard, thereby improving the reliability and stability of active alignment.

[0056] The above-disclosed embodiments are merely some examples of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for active alignment of a TOSA optical path, characterized in that, The method includes: S1, Collect the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned; S2, based on the acquisition dimension parameters and optical path communication transmission status parameters, generates an adaptive search strategy, monitors the effectiveness of the generated adaptive search strategy, and decides whether to adopt the search strategy optimization strategy. S3 performs joint evaluation based on sample alignment data and dynamically corrects the search path, and determines the target alignment position of the device to be aligned.

2. The TOSA optical path active alignment method as described in claim 1, characterized in that, The specific process for acquiring the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned is as follows: During the test optical path output process, the coupling state of the test optical path is monitored, the acquisition dimension parameters are obtained, the transmission quality corresponding to the test optical signal is monitored, and the optical path communication transmission status parameters are obtained. The acquisition dimension parameters include one or more of the number of acquisition angles and the number of location sampling points; The optical path communication transmission status parameters include one or more of the received optical power and the optical path signal-to-noise ratio.

3. The TOSA optical path active alignment method as described in claim 1, characterized in that, The adaptive search strategy generated based on the acquisition dimension parameters and optical path communication transmission status parameters specifically means that the acquisition dimension complexity level and communication transmission status level are determined based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, respectively, and the corresponding initial adaptive search strategy is matched based on the qualified level set. The acquisition dimension complexity level and communication transmission status level are determined based on the acquired acquisition dimension parameters and optical path communication transmission status parameters, respectively. The specific process is as follows: When all the parameters of the collection dimension are within the corresponding pre-stored collection dimension threshold range, the corresponding collection dimension complexity level is classified as a level one complexity level. When a data collection dimension parameter is outside the corresponding pre-stored data collection dimension threshold range, the corresponding data collection dimension complexity level is classified as a second-level complexity level. When all optical path communication transmission status parameters are within the corresponding pre-stored optical path communication threshold range, the corresponding communication transmission status level is classified as a first-level optical path communication level. When the optical path communication transmission status parameters are not within the corresponding pre-stored optical path communication threshold range, the corresponding communication transmission status level will be classified as a level two optical path communication level. The set of levels consisting of the first-level complexity level and the first-level optical path communication level is denoted as the qualified level set, and the set of levels consisting of the remaining acquisition dimension complexity level and communication transmission status level is denoted as the unqualified level set.

4. The TOSA optical path active alignment method as described in claim 3, characterized in that, The effectiveness of the monitored adaptive search strategy is assessed, and a decision is made on whether to adopt the search strategy optimization strategy. This means that based on the type of the monitored level set, a search strategy optimization strategy is selectively adopted. The specific process is as follows: When a set of non-compliant levels is detected, the search strategy optimization strategy is activated; When a set of qualified levels is detected, a search is performed using the initially set adaptive search strategy, and the sample alignment data is jointly evaluated and the search path is dynamically corrected.

5. The TOSA optical path active alignment method as described in claim 4, characterized in that, The search strategy optimization strategy follows the following process: Further judgment will be made based on the set of non-compliant grades; When it is detected that the set of unqualified levels consists only of second-level complexity levels, a complexity-constrained search strategy is executed for optimization. When the set of unqualified grades is detected to consist only of the second-level optical path communication grade, a communication priority search strategy optimization is performed. Specifically, the sample alignment data corresponding to the second-level optical path communication grade is input into the search order optimization model, the search order is output and the search is performed, and a joint evaluation is performed based on the sample alignment data and the search path is dynamically corrected. When the set of non-compliant levels is detected to include both level 2 complexity level and level 2 optical path communication level, a joint correction search strategy optimization is performed.

6. The TOSA optical path active alignment method as described in claim 5, characterized in that, The optimization of the complexity-constrained search strategy is as follows: Input the sample alignment data corresponding to the second level of complexity into the search window optimization model, output the optimized search window size and perform the search, perform joint evaluation based on the sample alignment data and dynamically correct the search path; The joint corrective search strategy optimization means that the search is performed using the obtained joint search window size and joint search order, and the search path is dynamically corrected based on the sample alignment data for joint evaluation.

7. The TOSA optical path active alignment method as described in claim 1, characterized in that, The joint evaluation and dynamic correction of the search path based on sample alignment data specifically means that the power evaluation result corresponding to each candidate position is determined based on the sample alignment data, and the search path that does not meet the preset trend is dynamically corrected. The specific process is as follows: The sample alignment data is input into a pre-built search performance evaluation model, and the corresponding trend data is output. The trend data includes power trend parameters and communication trend parameters; Determine whether the change trend data corresponding to the output search path meets the qualified change trend conditions. If it does, determine the target alignment position of the current device to be aligned based on the change trend data. If it does not meet the conditions, perform dynamic correction of the search path. The step of determining the target alignment position of the device to be aligned based on the trend data means that the corresponding trend data is input into a preset neural network model, and the corresponding target alignment position is output to guide the relative position adjustment of the current TOSA device.

8. The TOSA optical path active alignment method as described in claim 7, characterized in that, The specific representation of the dynamic correction of the search path is as follows: The trend data is input into the preset search path correction model, and the search order and search window size are output. The search is performed again with the output search order and search window size, and the trend data corresponding to the search path is checked to see if it meets the qualified trend conditions. If it does, the target alignment position of the device to be aligned is determined based on the trend data. If it still does not meet the conditions, a search path alarm is pushed to the preset control center.

9. The TOSA optical path active alignment method as described in claim 1, characterized in that, The specific process of performing joint evaluation and dynamically correcting the search path based on sample alignment data is as follows: Obtain trend consistency parameters that characterize the degree of matching between power trend parameters and communication trend parameters; When the trend consistency parameter is not greater than the corresponding qualified trend threshold, the current search direction is determined to be invalid, and a TOSA optical path search fluctuation alarm is sent to the preset control center. When the trend consistency parameter is greater than the corresponding qualified trend threshold, it is further determined whether the path convergence parameter, which represents the degree of convergence of the current search path toward the target area, is within the corresponding convergence interval. If so, the current search direction is determined to be valid, and the target alignment position of the device to be aligned is determined based on the change trend data. Otherwise, the current search direction is determined to be invalid, and a TOSA optical path active alignment alarm is sent to the preset control center.

10. A TOSA optical path active alignment system, the system being used to implement the TOSA optical path active alignment method as described in any one of claims 1-9, characterized in that, include: The TOSA device alignment parameter acquisition module is used to acquire the acquisition dimension parameters and optical path communication transmission status parameters of the TOSA device to be aligned. The search strategy generation module is used to generate an adaptive search strategy based on the acquisition dimension parameters and optical path communication transmission status parameters, monitor the effectiveness of the generated adaptive search strategy, and decide whether to adopt the search strategy optimization strategy. The device alignment position acquisition module is used to perform joint evaluation based on sample alignment data, dynamically correct the search path, and determine the target alignment position of the device to be aligned.

Images