Integrated optical device segmentation design method based on heuristic iteration and simulation feedback
By employing heuristic iteration and simulation feedback methods, combined with physical insights and computer simulation, the problems of high computational cost and low efficiency in the segmented design of integrated optical devices were solved, achieving efficient and systematic optimal design.
Patent Information
- Application Number
- CN202511108574.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-08
- Publication Date
- 2025-12-16
AI Technical Summary
Existing segmented design methods for integrated optical devices are computationally expensive, inefficient, lack systematicity, and make it difficult to find optimal solutions that are both high-performance and compact.
A heuristic iterative and simulation feedback approach is adopted. Key regions are identified through physical analysis, an initial width set is manually selected, and iterative optimization is performed using efficient simulation tools to form a closed-loop iterative process of analysis-adjustment-resimulation until a balance between performance and complexity is achieved.
Significantly improves design efficiency, ensures high performance, provides a clear and operable design process, enables human-machine collaboration, and approaches the optimal solution.
Smart Images
Figure CN121145601A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the field of integrated optoelectronics, and specifically relates to a segmented design method for integrated optical devices based on heuristic iteration and simulation feedback. BACKGROUND
[0002] In integrated optics, many high-performance devices (such as mode converters, couplers, spot size converters, etc.) rely on smooth, gradually changing geometric profiles to achieve efficient light field evolution. However, when designing photomasks, in order to facilitate manufacturing, it is often necessary to "segment" or "step" such continuous curve profiles, i.e., to approximate them with a series of rectangular segments with constant width (or pitch).
[0003] How to determine the number of these segments, their respective widths and lengths, is a core challenge in device design. Currently, there are the following main methods and their drawbacks:
[0004] (1) Complete algorithmic optimization: using global optimization algorithms (such as genetic algorithms, particle swarm algorithms, etc.) to perform a brute-force search of all segment widths and lengths as free variables. The disadvantages of this method are that the parameter space is huge, the computational cost is extremely high, the convergence speed is slow, and it is easy to fall into a local optimal solution, and it may not be able to find a physically reasonable optimal structure.
[0005] (2) Pure manual design: completely relying on the designer's experience to "manually select", lacking systematic guidance and feedback, the design process is arbitrary, inefficient, and it is difficult to guarantee the performance and compactness of the final design.
[0006] (3) Simple rule-based segmentation: for example, simply performing equal-length or equal-width step segmentation. This "one-size-fits-all" approach does not take into account the differences in optical behavior in different regions of the device, especially in regions where the mode changes dramatically, and rough segmentation can introduce significant loss.
[0007] As stated in the design principles, "in the mode hybridization region, the segmentation should be more refined", which is a correct physical direction. However, existing technologies do not provide a systematic design process that can translate this principle into an operational, efficient, and optimal solution. Therefore, there is an urgent need for a design method that can integrate the designer's physical insight with the computer's precise simulation capabilities. SUMMARY
[0008] The present application aims to overcome the problems of high computational cost, low efficiency and lack of systematicness in the existing segmented optical device design method. The present application provides a segmented design method for integrated optical devices based on heuristic iteration and simulation feedback, which can efficiently find a nearly optimal segmented device design with high performance and compact size by combining heuristic manual selection and rapid simulation iteration.
[0009] To achieve the above-mentioned application purposes, the technical solutions adopted by the present application are as follows: a segmented design method for integrated optical devices based on heuristic iteration and simulation feedback, comprising the following steps:
[0010] S1, physical analysis and heuristic initial width selection; S2, model construction and simulation feedback; S3, iterative manual adjustment and refinement; S4, convergence and final design determination.
[0011] Further, as a preferred technical solution of the present application, S1 comprises: (1) identifying key regions: performing preliminary mode analysis on the ideal continuous contour device, such as calculating the variation curve of its effective refractive index (neff) along the light propagation direction. The region with the most dramatic mode evolution, i.e. the "mode hybridization region" with the fastest variation of effective refractive index or the closest effective refractive index curves of different modes, is identified as the key region.
[0012] (2) Heuristic width set selection: the designer manually selects a set of discrete width values with "good" numerical values (easy to layout, such as 0.50, 0.55, 0.60 μm). This selection is not arbitrary, but follows a core principle: in the identified key regions, the distribution of width values is denser; in the regions with smooth mode variation, the distribution of width values can be sparser. This step uses human physical intuition to pre-set a promising and limited "width material library".
[0013] Further, as a preferred technical solution of the present application, S2 comprises: (1) constructing a segmented model: using the width value set selected in S1 to construct a segmented device model. The length of each segment can be initially set to be equal, or optimized by a simple subroutine.
[0014] (2) Performance simulation: using an efficient simulation tool (such as Eignemode Expansion, EME) to simulate the optical performance of the current segmented model, and obtaining its transmission efficiency, insertion loss, mode crosstalk and other key performance indicators.
[0015] Further, as a preferred technical solution of the present invention, S3 includes: (1) Analysis feedback: Check the simulation results. If the overall performance does not meet the standard, or the simulation shows that the loss is mainly concentrated at a few specific width jumps, the designer will make manual adjustments. (2) Adjustment strategy: The adjustment can be: Adding segments: In areas with high loss, insert a new intermediate width value from the preset width set to make the transition smoother. Adjusting width selection: Change or fine-tune some values in the width set. Adjusting segment length: Lengthen or shorten the length of certain key segments. (3) Loop: Return to S2 and perform simulation again with the adjusted new model to form a closed loop of "analysis-adjustment-resimulation".
[0016] As a further preferred embodiment of the present invention, S4 repeats the iterative cycle of S3. When the designer observes that further increasing the number of segments or making fine adjustments results in negligible improvement in device performance (i.e., diminishing returns, performance tending to saturate), the iteration process terminates. At this point, the design is considered to have reached a near-optimal balance between performance and complexity (or compactness). The final structure (width and length of each segment) is determined.
[0017] The segmented design method for integrated optical devices based on heuristic iteration and simulation feedback described in this invention has the following technical advantages compared with existing technologies:
[0018] (1) Significantly improve design efficiency: Compared with fully automatic brute-force optimization algorithms, this method greatly reduces the parameter space that needs to be explored through human heuristic selection, thereby significantly reducing computation time and cost.
[0019] (2) Guarantee high performance: Through iterative refinement based on physical insights, this method can systematically approximate the optimal solution, avoiding the arbitrariness of purely manual design and the coarseness of simple rule division, and the final designed device has excellent performance.
[0020] (3) Clear and practical process: This invention provides a clear and operable engineering design process, which standardizes the design process of complex devices and makes them easy for engineers to master and apply.
[0021] (4) Intelligent integration of human-machine advantages: This method perfectly combines human physical intuition (identifying key areas and heuristic selection) with the precise computing power of computers (rapid simulation feedback), and is a model of human-machine collaborative design. Attached Figure Description
[0022] Figure 1 This is a flowchart of the method of the present invention in an embodiment of the present invention;
[0023] Figure 2The curves showing the effective refractive index of TE0 and TE1 modes as a function of waveguide width in the embodiments of the present invention are shown. Detailed Implementation
[0024] The present invention will be further explained in detail below with reference to the accompanying drawings, so that those skilled in the art can better understand and implement the present invention. However, the following examples are only used to explain the present invention and are not intended to limit the present invention.
[0025] This invention discloses a segmented design method for integrated optical devices based on heuristic iteration and simulation feedback. The core idea is to avoid relying solely on optimization algorithms for all geometric parameters. Instead, designers heuristically select key parameters based on physical principles, and then use simulation tools for rapid performance verification and iterative optimization, thereby searching within a significantly reduced and more efficient parameter space.
[0026] The specific implementation steps of this method are as follows:
[0027] Step 1: Physical Analysis and Heuristic Initial Width Selection
[0028] (1) Identify key regions: Perform preliminary mode analysis on ideal continuous profile devices, such as calculating the change curve of their effective refractive index (neff) along the light propagation direction. The region with the most dramatic mode evolution, that is, the “mode hybridization region” where the effective refractive index changes the fastest or where the effective refractive index curves of different modes are closest, is identified as a key region.
[0029] (2) Heuristic width set selection: The designer manually selects a set of discrete width values with “good” values (easy to plot, such as 0.50, 0.55, 0.60 μm). This selection is not arbitrary, but follows a core principle: the width values are more densely distributed in identified critical areas; and the width values can be more sparsely distributed in areas where the pattern changes gradually. This step uses human physical intuition to pre-define a promising, finite “width material library”.
[0030] Step 2: Model Building and Simulation Feedback
[0031] (1) Construct a segmented model: Use the set of width values selected in step one to construct a segmented device model. The length of each segment can be initially set to be equal, or it can be initially optimized by a simple subroutine.
[0032] (2) Performance simulation: Use efficient simulation tools (such as Eignemode Expansion, EME) to perform optical performance simulation on the current segmented model and obtain its key performance indicators such as transmission efficiency, insertion loss, and mode crosstalk.
[0033] Step 3: Iterative manual adjustment and refinement
[0034] (1) Analysis and feedback: Check the simulation results. If the overall performance does not meet the standards, or if the simulation shows that the loss is mainly concentrated at a few specific width transition points, the designer will make manual adjustments.
[0035] (2) Adjustment strategy: Adjustment can be:
[0036] Add segments: In areas with high loss, insert a new intermediate width value from the preset width set to make the transition smoother.
[0037] Adjust width selection: Change or fine-tune certain values in the width set.
[0038] Adjust segment length: lengthen or shorten the length of certain key segments.
[0039] (3) Loop: Return to step two and perform simulation again with the adjusted new model to form a closed loop of "analysis-adjustment-resimulation".
[0040] Step 4: Convergence and Final Design Determination
[0041] Repeat the iterative cycle of step three. The iteration process terminates when the designer observes that further increasing the number of segments or making fine-tuning yields negligible improvements in device performance (i.e., diminishing returns, performance saturation). At this point, the design is considered to have reached a near-optimal balance between performance and complexity (or compactness). The final structure (width and length of each segment) is determined.
[0042] In practical implementation, we will take the design of a segmented mode converter as an example to illustrate the implementation of this method. The design flowchart is as follows: Figure 1 As shown.
[0043] (1) Initial Analysis: The design goal is to convert the TE0 mode to the TE1 mode. First, mode analysis is performed on an ideal continuous profile mode converter, plotting the effective refractive index of the TE0 and TE1 modes as a function of waveguide width. For example... Figure 2 As shown: Figure 2 This is a physical analysis plot used for heuristic selection, showing the effective refractive index of two coupled modes in a device as a function of propagation distance, and marking the "mode hybridization region" that requires fine subdivision. It was found that when the waveguide width is between 1.4 μm and 1.5 μm, the two curves are very close, which is the key region of "mode hybridization".
[0044] (2) Heuristic selection of width set: Based on the above analysis, the designer manually selects a width set, for example: {0.5, 0.7, 0.8, 0.85, 0.9, 0.95, 1.0, 1.05, 1.1, 1.2, 1.4, 1.6} μm. It can be seen that in the critical region of 1.4-1.5 μm, the width increment (e.g., 0.05 μm or less) is much smaller than in other regions.
[0045] (3) First iteration: An initial model was constructed using this width set, with the length of each segment tentatively set to 2 μm. Simulation using the EME method yielded a transmission efficiency of 85%, but with significant losses. The simulation field distribution showed that energy leakage mainly occurred around the 1.4 μm and 1.5 μm widths.
[0046] (4) Second iteration and adjustment: Based on feedback, the designers decided to further densify the segments in the critical area. The width set was modified to {...1.4,1.42,1.44,...,1.48,1.5,...}. Meanwhile, simulation results showed that the transition loss of 0.5-0.7μm was very small, and the length of this segment could be appropriately shortened, with the saved length allocated to the critical segment.
[0047] (5) Convergence: After several rounds of "adjustment-simulation" cycles, when the total number of segments reaches 15, the transmission efficiency reaches 98.5%. At this point, if more segments are added (for example, to 20), the efficiency only increases to 98.6%. The designer judged that the performance had saturated, so he stopped iterating and adopted the design with 15 segments as the final solution.
[0048] Through this process, a high-performance, structurally sound segmented device can be designed efficiently.
[0049] The specific implementation schemes described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific implementation schemes of the present invention and are not intended to limit the scope of the present invention. Any equivalent changes and modifications made by those skilled in the art without departing from the concept and principles of the present invention should fall within the scope of protection of the present invention.
Claims
1. A segmented design method for integrated optical devices based on heuristic iteration and simulation feedback, characterized in that, Includes the following steps: S1. Physical analysis and heuristic initial width selection; S2. Model building and simulation feedback; S3, Iterative manual adjustment and refinement; S4. Convergence and final design determination.
2. The segmented design method for integrated optical devices based on heuristic iteration and simulation feedback according to claim 1, characterized in that, S1 specifically includes the following steps: S1.1 Identifying Key Regions: Perform preliminary pattern analysis on an ideal continuous profile device and calculate its effective refractive index n. eff The region of most dramatic mode evolution along the direction of light propagation, i.e. the region of mode hybridization where the effective refractive index changes the fastest or where the effective refractive index curves of different modes are closest, is identified as the key region. S1.2 Heuristic width set selection: Select a set of discrete width values that are easy to draw in the layout.
3. The segmented design method for integrated optical devices based on heuristic iteration and simulation feedback according to claim 2, characterized in that, S2 specifically includes the following steps: S2.1 Constructing a segmented model: Use the set of width values selected in S1.2 to construct a segmented device model; S2.2 Performance Simulation: Use simulation tools to perform optical performance simulation on the current segmented model to obtain its key performance indicators such as transmission efficiency, insertion loss, and mode crosstalk.
4. The segmented design method for integrated optical devices based on heuristic iteration and simulation feedback according to claim 3, characterized in that, S3 specifically includes the following steps: S3.1 Analysis and Feedback: Check the simulation results. If the overall performance does not meet the standard, or if the simulation shows that the loss is mainly concentrated at a few specific width jumps, manual adjustment will be performed. S3.2 The adjustment strategy includes: adding segments: in areas with high loss, inserting a new intermediate width value from the preset width set to make the transition smoother; adjusting width selection: changing or fine-tuning certain values in the width set; adjusting segment length: lengthening or shortening the length of certain key segments. S3.3, Loop: Return to S2 and perform simulation again using the adjusted new model, forming a closed loop of analysis-adjustment-resimulation.
5. The segmented design method for integrated optical devices based on heuristic iteration and simulation feedback according to claim 4, characterized in that, The iteration cycle of S4 repeats the iteration of S3. When it is observed that further increasing the number of segments or making fine adjustments has a negligible effect on the device performance (i.e., diminishing returns) and the performance tends to saturate, the iteration process terminates. At this point, the design is considered to have reached a near-optimal balance between performance and complexity or compactness. The final structure, i.e., the width and length of each segment, is determined.