Curve synchronous adjustment method

Abstract

The invention relates to the technical field of detection imaging analysis, and provides a curve synchronous adjustment method which comprises the following steps: establishing a proportion reference, modifying and detecting a target curve, extracting modification information, determining a to-be-modified correlation curve, determining adjustment parameters and adjusting the curve in batches. According to the curve adjustment method provided by the invention, automatic synchronous adjustment of the multiple curves is realized by executing the engines in batches, the adjustment time is greatly shortened, the curve adjustment efficiency is improved, and the adjustment accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of detection imaging analysis, in particular to a curve synchronous adjustment method. BACKGROUND

[0002] In detection analysis, such as scientific research experiment data analysis, industrial detection, etc., when there are multiple associated analysis projects, they often need to be generated into multiple curves for comparative analysis. These curves are usually layered and aligned. As shown in the figure, the data result graph has been layered and matched and aligned. The structure of the elements of each curve is different, and the depth and point number are also different. Figure 1

[0003] When the analyst learns that the previous reference curve parameter input is incorrect, or the parameter needs to be adjusted due to the need for analysis, the parameter of a certain curve may be modified, such as modifying the depth data of a certain curve. In order to ensure the alignment state between curves, the corresponding adjustment needs to be made to other curves.

[0004] In the technical solution provided by the prior art, when one of the curves is modified and the other curves need to be adjusted synchronously to adapt to the new analysis requirements, a manual adjustment mode is adopted. The operator needs to first determine the modification range of the target curve as the target curve, and then manually input the adjustment value to complete the modification of a single curve by adjusting the parameters (such as depth, rate value, etc.) of each layer of the other curves one by one. If the number of curves is large or the number of layers is complex, the above steps need to be repeated dozens or even hundreds of times.

[0005] This method has the following significant disadvantages:

[0006] Operation is complicated: each layer of each associated curve needs to be manually operated separately, and the steps are highly repetitive. Especially when the number of curves exceeds 5 or the number of layers exceeds 3, the operation process is lengthy.

[0007] Time-consuming: The process of manually calculating the proportion and inputting the adjustment value is time-consuming. For a data set containing 10 curves and 4 layers, a single adjustment takes more than 10 minutes, which seriously affects the data processing efficiency.

[0008] High error risk: When manually calculating the proportion and inputting the adjustment value, calculation errors or input deviations may occur due to fatigue or negligence, which may lead to a decrease in curve alignment accuracy and the need for repeated verification and correction. SUMMARY

[0009] Based on the above problems of the prior art, the present application provides a curve synchronous adjustment method to solve the problem of low efficiency and high error risk caused by manual modification of the curve adjustment provided by the prior art.

[0010] ​The curve synchronous adjustment method provided by the embodiment of the application comprises the following steps:

[0011] S100, a proportion reference is established, a set of curves that have been aligned by layers is taken as a reference set, initial proportion parameters of different layers between curves in the reference set are identified and extracted, and an initial proportion matrix is formed;

[0012] S200, target curve modification detection, whether a curve is modified is detected in real time, when it is detected that the parameters of a certain curve are modified, the curve on which the modification operation is performed is taken as a target curve, and information of the target curve is recorded;

[0013] S300, modification information extraction, curve data after modification of the target curve is analyzed, and at least layer index information involved in the modification, modification value size information and modification range information are extracted;

[0014] S400, determination of a curve to be modified, the curve to be modified is determined through dimensional matching according to the information of the target curve and the proportion reference;

[0015] S500, determination of an adjustment parameter, an adjustment proportion value of each curve to be modified and the modification information of the target curve are obtained according to the initial proportion matrix, and an adjustment parameter of each curve to be modified is generated;

[0016] S600, batch adjustment of a curve, a batch execution engine is started, and all curves to be modified are synchronously adjusted according to the generated adjustment parameters.

[0017] The curve synchronous adjustment method provided by the embodiment of the application provides a basis for obtaining adjustment parameters of a set of curves through parameters between the curves by establishing a proportion reference. When it is detected that a target curve is modified, the adjustment value of a parameter of another curve to be modified in the same standard set as the target curve can be calculated through the modification information and the proportion reference. The automatic synchronous adjustment of multiple curves is realized through a batch execution engine, manual operation is replaced, the adjustment time is greatly shortened, and the efficiency is improved, especially in a large number of curve scenarios. The problem that calculation errors or input deviations are prone to occur due to fatigue and negligence when a proportion is calculated and an adjustment value is input is solved, and the adjustment accuracy is improved.

[0018] Preferably, after the step S100 of forming the initial proportion matrix, the method further comprises the following steps:

[0019] S110, dynamic proportion model construction, a time attenuation factor and a layer weight coefficient are introduced, and a dynamic proportion model is constructed;

[0020] S120, proportion library management, a proportion library system is established to manage the dynamic proportion model.

[0021] Preferably, the step S400 is specifically: based on the dynamic scaling model characteristics in the scaling library system, the target curve and other associated curves are subjected to curve type consistency verification, hierarchical structure similarity comparison, and historical adjustment record correlation analysis.

[0022] Preferably, the adjustment scaling value is obtained by dynamic calculation in the step S500, and the dynamic calculation is specifically: according to the extracted modification information and the matching result, a dynamic scaling model is called, and the adjustment scaling value of each hierarchical layer of the associated curve to be modified is calculated in real time in combination with a time attenuation factor and a hierarchical weight coefficient.

[0023] Preferably, after the adjustment parameter is determined in the step S500, an anomaly detection step is further included, which specifically includes:

[0024] Dual threshold detection: a hard threshold and a soft threshold are set for anomaly detection; the hard threshold is a limit value range based on the physical characteristics of the curve, when the modification value of the adjustment parameter or the calculated adjustment parameter adjustment value exceeds the hard threshold, a warning is triggered and the adjustment is refused to be performed; the soft threshold is calculated by the 3σ principle to obtain a normal fluctuation range, when the adjustment value deviation exceeds 2σ, a backup scaling value is used for adjustment.

[0025] Preferably, after the anomaly detection step, the following steps are further included:

[0026] Breakpoint recovery: key scaling data and adjustment records are notarized by using a blockchain technology, when the system is interrupted abnormally, the notarized data is used to backtrack to the nearest valid state;

[0027] Abnormal positioning and prompting: when an anomaly is detected, the abnormal source is located through log analysis, and a prompt is given.

[0028] Preferably, after the step S600, a step S700 of scaling model iterative upgrading is further included, which specifically includes:

[0029] S710, data collection and analysis: modification record data, verification result data, and abnormal handling situation data in the batch adjustment process are collected, a machine learning model is used to analyze the applicability and accuracy of the scaling model, and the deviation law of the scaling model in different scenarios is identified;

[0030] S720, model optimization and upgrading: the scaling model is optimized based on the analysis result, including at least one of adjusting a time attenuation factor coefficient, updating a hierarchical weight coefficient, and adding a scaling calculation rule for a special scenario.

[0031] The embodiment of the application further provides an electronic device including a memory and a processor, the memory stores a computer executable program, and the processor calls the computer executable program in the memory to realize the curve synchronous adjustment method.

[0032] The embodiment of the present application further provides a storage medium, which is a computer readable storage medium, and a computer executable program is stored on the computer readable storage medium, and the computer executable program is executed by a processor to realize the curve synchronous adjustment method.

[0033] The embodiment of the present application further provides a computer program product for realizing curve synchronous adjustment, including codes for realizing the curve synchronous adjustment method. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0035] Figure 1 For the related curve schematic diagram generated in the prior art detection and analysis;

[0036] Figure 2 For the curve synchronous adjustment method flowchart in the embodiment of the present application;

[0037] Figure 3 For the target curve detection flowchart provided by the embodiment of the present application;

[0038] Figure 4 For the state backtracking flowchart in the embodiment of the present application;

[0039] Figure 5 For the adjustment parameter generation schematic diagram in the embodiment of the present application. DETAILED DESCRIPTION

[0040] In the following, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present application. Therefore, the drawings and the description are considered to be exemplary in nature rather than limiting.

[0041] It should be noted that the terms "first", "second", "symmetric", "array" and the like are only used for distinguishing description and position description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "symmetric" and the like can explicitly or implicitly include one or more features; similarly, for some features that are not limited in number by the words "two", "three" and the like, it should be noted that the features also belong to explicitly or implicitly including one or more features.

[0042] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "fixing" and the like should be understood in a broad sense; for example, it can be fixedly connected, or detachably connected, or integrally formed; it can be mechanically connected, it can be directly connected, it can be welded, it can be indirectly connected through an intermediate medium, it can be the communication or interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the description and drawings in combination with specific circumstances.

[0043] The technical solutions of the present application will be described in detail below with reference to the drawings.

[0044] As shown in the drawings, the present application provides a curve synchronous adjustment method, which specifically includes the following steps: Figure 2

[0045] S100, proportion reference establishment, taking a group of layered and aligned curves as a reference group, identifying and extracting the initial proportion parameters of different layers between the curves in the reference group to form an initial proportion matrix.

[0046] The proportion reference establishment provided by the present application specifically includes the following steps:

[0047] A group of curves with verified depth data is selected as the reference group by the parameter extraction module, and the initial proportion parameters of different layers between the curves are identified and extracted, including layering speed proportion, depth proportion, etc., to form an initial proportion matrix.

[0048] In the embodiments provided by the present application, for the curve group that has been layered and aligned (i.e. verified depth data), any one of the curves can be selected as the reference curve, and the proportion values of the other curves relative to the reference curve are calculated. Among them, the proportion of the i-th layer of the j-th curve to the reference curve is defined as:

[0049]

[0050] Where: R i,j : the proportion of the i-th layer of the j-th curve to the reference curve, V i,j ​: the i-th stratigraphic parameter value (e.g. depth) of the j-th curve, V i,base : the i-th stratigraphic parameter value of the reference curve.

[0051] Suppose there are 4 curves (A, B, C, D) that have been stratigraphically aligned, each curve contains 3 stratigraphies, and the depth data is as follows:

[0052] Curve Surface layer (pm) Intermediate layer (pm) Bottom layer (pm) A 100 200 300 B 120 240 360 C 80 160 240 D 110 220 330

[0053] Step S101: Set the reference curve: select curve A as the reference curve (base=A).

[0054] Step S102: Calculate the stratigraphic proportion of each curve to the reference curve, and form the proportion matrix:

[0055] Proportion matrix of curve B to A:

[0056]

[0057] Proportion matrix of curve C to A:

[0058]

[0059] Proportion matrix of curve D to A:

[0060]

[0061] Step S103: Organize into a proportion matrix:

[0062]

[0063] (row: stratigraphy 1-3, column: curve B-D).

[0064] Then, curve B can be selected as the reference curve, and the proportion matrix of other curves in the reference group to curve B is calculated respectively. The specific calculation steps can be referred to the above example of curve A, and will not be described here.

[0065] S200, target curve modification detection, real-time detection whether there is a curve modification operation, when detecting that the parameter of a certain curve is modified, the curve of the modification operation is taken as the target curve, and the information of the target curve is recorded.

[0066] The target curve modification detection provided in the application can refer to Figure 3When the user modifies the parameters of a curve in the operation interface, such as modifying the depth of the middle layer of curve C001 from 200 um to 250 um, the data interface layer monitors the modification operation of all curves in real time, and when it detects that the parameters (such as depth data) of curve C001 have changed, the curve on which the modification operation occurs is taken as the target curve. The target curve modification information is recorded, including the curve identifier, the layer index, the values before and after modification, the modification time, and other key information.

[0067] S300, modification information extraction, the curve data after modification of the target curve is analyzed, and at least the layer index information involved in the modification, the modification value size information, and the modification range information are extracted.

[0068] S400, determination of the associated curve to be modified, the associated curve to be modified is determined through dimensional matching according to the information of the target curve and the proportion reference.

[0069] S500, determination of the adjustment parameter, the adjustment proportion value of each associated curve to be modified and the modification information of the target curve are obtained according to the initial proportion matrix, and the adjustment parameter of each associated curve to be modified is generated.

[0070] When the i-th layer parameter value of the reference curve is modified from to , the synchronous adjustment formula of the associated curve is:

[0071]

[0072] Wherein: The i-th layer adjusted value of the j-th curve.

[0073] Still taking the above-mentioned reference curve group with four curves A, B, C, and D as an example, assuming that the user modifies the depth of the middle layer of reference curve A (from 200 um to 220 um), then:

[0074] The adjusted depth of the middle layer of curve B:

[0075]

[0076] The adjusted depth of the middle layer of curve C:

[0077]

[0078] The adjusted depth of the middle layer of curve D:

[0079]

[0080] The final result after adjustment:

[0081] Curve Surface layer (pm) Intermediate layer (pm) Bottom layer (pm) A 100 220 300 B 120 264 360 C 80 176 240 D 110 242 330

[0082] S600, batch adjustment curve, start batch execution engine, synchronize all associated curves to be modified according to the generated adjustment parameters.

[0083] In the preferred embodiment provided by the application, after establishing the proportion reference, step S100 further establishes a dynamic proportion reference, specifically including the following steps:

[0084] S110, dynamic proportion model construction, introduce time decay factor and hierarchical weight coefficient to construct dynamic proportion model.

[0085] Introduce time decay factor and hierarchical weight coefficient to construct dynamic proportion model. Time decay factor is calculated according to the formula R(t) = R0 x e^(-kt), where R0 is the initial proportion, k is the decay coefficient, which can be set according to the device characteristics, t is the data collection interval time, which is used to correct the proportion deviation caused by device drift; hierarchical weight coefficient W i Based on historical data fitting, used to distinguish the importance of different layers, the dynamic proportion calculation formula is V ji ×V templati ×(R ij ×W i ×R(t)), where V ji is the correction value of the jth curve of the ith layer, V templati is the modified parameter value of the ith layer of the target curve, R ij is the initial proportion. By introducing time decay factor and hierarchical weight coefficient to construct dynamic proportion model, real-time correction of curve proportion is realized, the problem of insufficient accuracy under device drift and complex hierarchical scene is solved, and long-term synchronization alignment accuracy of the curve is ensured.

[0086] For time decay factor, it is set according to the following factors:

[0087] 1. Measurement stability

[0088] Characteristic description: the ability of the device to maintain measurement accuracy over a long period of time. Devices with poor stability will produce more obvious drift over time (such as sensor zero point shift caused by temperature change).

[0089] Typical devices: precision mass spectrometer, laser range finder.

[0090] 2. Response delay characteristics

[0091] Characteristic description: the response speed of the device to parameter changes. Devices with high delay will cause the data timestamp to be out of sync with the actual physical time when continuously sampling.

[0092] Typical devices: electrochemical sensor, slow response thermometer.

[0093] 3. Environmental sensitivity

[0094] Characteristic description: The degree to which the performance of a device is affected by external environment (temperature, humidity, electromagnetic interference, etc.).

[0095] Typical devices: Optical microscopes, semiconductor detectors.

[0096] 4. Sampling frequency

[0097] Characteristic description: The number of times a device collects data per unit of time. A device with low sampling frequency may experience large drifts between data collection intervals.

[0098] Typical devices: Geological exploration equipment, weather station sensors.

[0099] Method for setting the attenuation coefficient (k value)

[0100] The attenuation coefficient k determines the rate at which the proportion decays over time. The larger the value, the more rapid the proportion correction. The k value should be set in combination with the characteristics of the device and the application scenario. The specific method is as follows:

[0101] 1. Empirical value based on device stability test

[0102] Test method:

[0103] Under constant conditions, use the device to continuously collect curve data of the same object (such as 24 hours of uninterrupted sampling);

[0104] Calculate the proportion between curves every fixed time interval (such as 1 hour) and compare it with the initial proportion. Record the curve of the proportion deviation over time;

[0105] Fit the deviation curve to an exponential decay model: Deviation (t) = Initial deviation × e -kt , solve for k value.

[0106] Example: A mass spectrometer stability test shows that its proportion deviation reaches 50% of the initial value after 2 hours. Substitute into the formula:

[0107] 2. Theoretical calculation based on device specifications

[0108] Formula method:

[0109] Where, the hourly drift rate of the device can be obtained from the device specification sheet (such as "0.1% drift per hour").

[0110] Example: A laser range finder specification sheet indicates "0.05% drift per hour", then:

[0111] 3. Adaptive dynamic adjustment

[0112] Method: Compare the difference between the current calculation and the historical ratio in real time during system operation, and dynamically adjust the k value through feedback mechanism.

[0113]

[0114] Where:

[0115] ΔR: Absolute difference between current ratio and historical ratio;

[0116] R: Historical ratio value;

[0117] α: Learning rate (usually 0.1ˉ0.3).

[0118] Advantages: Automatically adapt to device aging, environmental changes, etc., without manual intervention.

[0119] Recommended range of k value for different device characteristics

[0120]

[0121]

[0122] The following is an exemplary explanation of the setting of the attenuation coefficient K by setting the attenuation coefficient of the SIMS device:

[0123] Secondary ion mass spectrometer (SIMS) is used for material depth analysis. When the device works continuously, the depth measurement drifts due to ion source fluctuations, and the attenuation coefficient needs to be set to correct the ratio.

[0124] Setting steps

[0125] Device characteristics analysis:

[0126] Stability test shows: After 8 hours of continuous work, the depth measurement deviation reaches 3%;

[0127] Sampling frequency: 2 minutes / point;

[0128] Environmental sensitivity: moderate (requires constant temperature environment).

[0129] k value calculation:

[0130] Assuming the initial ratio is 1 and the deviation is 3% after 8 hours, the ratio becomes

[0131] Actual application:

[0132] Set k = 0.004 in the system;

[0133] When the user adjusts the curve, the system calculates R(t) = R0×e -0.004×tDynamic correction ratio (t is the data acquisition interval, unit: hour).

[0134] By introducing time decay factor and hierarchical weight coefficient to construct dynamic ratio model, real-time correction of curve ratio is realized, and the problems of equipment drift and insufficient accuracy in complex hierarchical scenes are solved, ensuring long-term synchronization and alignment accuracy of curves.

[0135] S120, ratio library management, to manage dynamic ratio model by establishing ratio library system.

[0136] Establishing ratio library system supports classified storage (such as by curve type, application scenario, etc.), query, editing and version control of ratio model, realizing reuse and sharing of ratio model.

[0137] Taking the ratio database system established by a material analysis for example, which is used to store dynamic ratio model of SIMS curve (secondary ion mass spectrum curve), the core structure includes three levels:

[0138] First-level classification: by curve type (such as depth curve, concentration curve);

[0139] Second-level classification: by application scenario (such as semiconductor chip analysis, thin film material analysis);

[0140] Third-level classification: by sample type (such as silicon wafer, compound semiconductor).

[0141] Classification storage example

[0142] 1. Storage path design

[0143] Ratio library system:

[0144]

[0145] 2. Model storage content (take "silicon wafer depth curve model V1.0" as an example)

[0146] Each model contains:

[0147] Basic information: model name, creation time, creator, number of applicable curves (such as 5 curves), number of layers (such as 4 layers);

[0148] Core data: initial ratio matrix (4 rows × 4 columns, corresponding to hierarchical ratio of 4 associated curves and reference curve), decay coefficient k value (set to 0.004 for SIMS device characteristics);

[0149] Associated files: original curve data snapshot, adjustment record log, verification report (such as alignment deviation of adjusted curve ≤1%).

[0150] Query function example

[0151] User needs to call the historical proportion model for "depth curve of silicon wafer (semiconductor chip analysis scenario)", the query process is as follows:

[0152] Multi-level screening: Select "depth curve→semiconductor chip analysis→silicon wafer" in the system interface in turn;

[0153] Condition filtering: Support filtering according to "creation time (such as the last 3 months)" and "applicable layer number (such as 4 layers)", and display the models (such as V1.0 and V2.0) that meet the conditions;

[0154] Preview comparison: Click the model to view details, such as the proportion matrix, k value and historical adjustment effect (with a visual curve comparison chart) of V1.0, to assist in determining whether it is applicable.

[0155] Editing function example

[0156] If the user finds that the decay coefficient k value (0.004) of "silicon wafer depth curve model V1.0" is too large on the new device, resulting in a large deviation after adjustment, the editing steps are as follows:

[0157] Model copying: Call out V1.0 from the library and select "copy as new version" (to avoid directly modifying the original model);

[0158] Parameter modification: Adjust the k value to 0.003 (based on the stability test results of the new device), and record the modification reason ("adapt to the drift characteristics of the new SIMS device (model XXX)");

[0159] Save and take effect: Save to the original path as "V2.0", overwrite the original V2.0 or add a new version, and generate a modification log (including parameter comparison before and after modification).

[0160] In the preferred embodiment provided by the embodiment of the application, step S400 determines the associated curve (referred to as the associated curve) to be modified through multi-dimensional matching. The multi-dimensional matching is based on the characteristics of the proportion model in the proportion library, and performs, on the target curve and other associated curves, consistency verification of curve types, similarity comparison of hierarchical structures, and association analysis of historical adjustment records, to ensure the accuracy of matching.

[0161] The core of multi-dimensional matching is a three-level verification mechanism of "feature extraction→hierarchical verification→association analysis", which ensures the matching accuracy of the target curve and the associated curve. The following takes the SIMS curve synchronous adjustment as an example to introduce the process of three-level verification. The specific process is as follows:

[0162] First-level verification: consistency verification of curve types (basic screening):

[0163] 1. Feature extraction

[0164] Retrieve the ratio model features to be matched from the ratio library, and extract the core attributes of the target curve and the associated curve to be modified:

[0165] Mandatory features:

[0166] Curve type (such as "depth curve", "concentration curve", "sputtering rate curve");

[0167] Physical unit (such as "A", "μm", "nm" for depth curve, "atoms / cm 3 " for concentration curve);

[0168] Sample type label (such as "silicon-based wafer", "compound semiconductor", "metal thin film").

[0169] 2. Consistency judgment

[0170] If the curve type, unit, and sample label of the target curve and the associated curve are completely consistent, proceed to the next level of verification;

[0171] If there is any feature mismatch (such as the target curve is "depth curve (μm)", and the associated curve is "concentration curve (atoms / cm 3 )"), directly exclude the associated curve and mark it as "mismatch".

[0172] Examples are as follows:

[0173] Target curve features: type = depth curve, unit = μm, sample = silicon-based wafer;

[0174] Associated curve A features: type = depth curve, unit = μm, sample = silicon-based wafer → pass the first level of verification;

[0175] Associated curve B features: type = concentration curve, unit = atoms / cm 3 , sample = silicon-based wafer → not pass, excluded.

[0176] Second-level verification: hierarchical structure similarity comparison (structure matching):

[0177] For the associated curve that passes the first level of verification, verify the consistency of its hierarchical structure with the target curve:

[0178] 1. Hierarchical feature extraction

[0179] Number of layers (such as the target curve contains 3 layers, and the associated curve also needs to contain 3 layers);

[0180] Layer boundary value (such as the boundary between the surface layer and the middle layer: the target curve is 100 μm, and the associated curve needs to be within the range of [90 μm, 110 μm], allowing a ±10% error);

[0181] Layered trend (e.g. the direction of each layer parameter change with depth: target curve layering: surface layer→middle layer "increasing", correlation curve consistent).

[0182] 2. Similarity calculation

[0183] Using "layered matching degree" to quantify structural similarity:

[0184] If the matching degree is ≥80% (i.e. only 1 layer is allowed to be unmatched), enter the next level of verification;

[0185] If the matching degree is <80%, mark it as "low matching" and manually confirm whether to continue.

[0186] Examples are as follows:

[0187] Target curve layering: 3 layers, boundary = [100 μm, 200 μm], trend = increasing;

[0188] Correlation curve A: 3 layers, boundary = [95 μm, 205 μm], trend = increasing→layered matching degree = 100%→pass;

[0189] Correlation curve C: 3 layers, boundary = [150 μm, 250 μm], trend = decreasing→layered matching degree = 0%→low matching, manual intervention.

[0190] Three-level verification: historical adjustment record correlation analysis (experience verification):

[0191] Determine whether there are collaborative adjustment records between the two curves through historical data to enhance the reliability of matching:

[0192] 1. Historical data extraction

[0193] Retrieve from system log:

[0194] In the last 3 adjustment records, whether the target curve and the correlation curve belong to the same curve group;

[0195] Proportion deviation of historical adjustment (e.g. whether the layered proportion deviation of the two after the past synchronous adjustment is ≤2%);

[0196] Operator label (e.g. whether it has been labeled as "correlation curve" by the same analyst in history).

[0197] 2. Correlation score

[0198] Score according to the following rules (full score 100 points):

[0199] Same group adjustment record: +20 points each time (maximum 60 points);

[0200] Historical proportion deviation ≤2%: +20 points;

[0201] Manual marking association: +20 points.

[0202] Score ≥ 60 points: determined as "high association", matching passed;

[0203] Score < 60 points: determined as "low association", prompting the user to confirm whether to force matching.

[0204] Example:

[0205] Association curve A: all the last three times are adjusted with the target curve in the same group (60 points), historical deviation 1.5% (20 points) → total score 80 points → high association, passed;

[0206] Association curve D: no same group record (0 points), historical deviation 5% (0 points) → total score 0 points → low association, manual confirmation required.

[0207] Matching result output and processing:

[0208] Automatic matching result:

[0209] Association curves that pass the three-level verification are automatically included in the synchronous adjustment range and marked as "automatic matching passed";

[0210] Curves that do not pass any level of verification are marked as "not matched" or "low matching" and are displayed separately in a list.

[0211] Manual intervention mechanism:

[0212] For "low matching" curves, a visual comparison interface is provided (showing unmatched feature items, such as layer boundary difference diagram);

[0213] Provide "force matching" option and feedback to user, receive user confirmation information for forced matching, and record intervention reasons (for subsequent proportion model iteration).

[0214] Step S500 determines that the adjustment parameters need to be calculated first: according to the extracted modification information and matching results, call the dynamic proportion model, combine the time decay factor and layer weight coefficient, and calculate the proportion value of each layer of the associated curve in real time to ensure that the calibrated proportion can accurately reflect the current curve state.

[0215] Dynamic proportion calculation specific process and formula (taking SIMS depth curve synchronous adjustment as an example)

[0216] The core of dynamic proportion calculation is to generate the adjustment proportion of each layer of the associated curve based on modification information, matching results and dynamic model parameters in real time to ensure that the calibrated proportion is consistent with the current curve state. The specific process and formula are as follows:

[0217] I. Parameter preparation before calculation

[0218] 1. Basic parameter extraction

[0219] The following key parameters are obtained from the system:

[0220] Target curve information:

[0221] Target curve i-th layer value before modification: V base old (i);

[0222] Target curve i-th layer value after modification: V base new (i);

[0223] Modification time: t modify (Current timestamp);

[0224] Associated curve information:

[0225] Initial ratio of associated curve j to reference curve (i.e. selecting target curve as reference curve) (extracted from ratio library): R _init(i,j) (i-th layer, j-th associated curve);

[0226] Last adjustment time of associated curve j: t last(j) ;

[0227] Dynamic model parameters:

[0228] Time decay factor coefficient: k (preset according to device characteristics, such as 0.004 / h);

[0229] Layer weight coefficient: W(i) (weight of i-th layer, based on historical data fitting, range 0ˉ1);

[0230] II. Dynamic ratio calculation process

[0231] Step 1: Calculate time decay factor (correct device drift)

[0232] According to the time difference between the reference curve and the associated curve, calculate the time decay factor R t (i,j):

[0233]

[0234] Meaning: If the associated curve has not been adjusted for a long time (large time difference), R t value decreases, the correction amplitude of the ratio increases, offsetting the influence of device drift;

[0235] Unit: t modify -t last(j) in "hours" (matching the unit of k).

[0236] Example:

[0237] Associated curve j last adjustment time: t last(j) = 2024-07-20 10:00;

[0238] Current modification time: t modify = 2024-07-20 14:00 (time difference = 4h);

[0239] k = 0.004 / h, then R t (i,j) = e^(-0.004x4) ≈ e^(-0.016) ≈ 0.984;

[0240] Step 2: Determine the layer weight coefficient (highlight key layers);

[0241] Layer weight W(i) is preset according to layer importance (can be edited in the proportion library):

[0242] Key layers (such as middle doped layers of semiconductor chips): W(i) = 0.9ˉ1.0;

[0243] Secondary layers (such as surface oxide layers): W(i) = 0.3ˉ0.5;

[0244] Example:

[0245] Layer 1 (surface layer): W(1) = 0.4;

[0246] Layer 2 (middle layer): W(2) = 1.0 (key layer);

[0247] Layer 3 (bottom layer): W(3) = 0.6;

[0248] Step 3: Calculate the dynamic adjustment proportion (comprehensive correction);

[0249] Associated curve j dynamic adjustment proportion R dynamic (i,j) of the i-th layer:

[0250] R dynamic (i,j) = R init (i,j) x R t (i,j) x W(i)

[0251] Meaning: Initial proportion R init After time decay (R t ) and layer weight (W) correction, the dynamic proportion that fits the current state is obtained.

[0252] Example:

[0253] Initial proportion R init(2,j) = 1.2 (intermediate layer, initial ratio of associated curve j to reference curve)

[0254] Time decay factor R t (2,j) = 0.984 (step 1 result)

[0255] Layered weight W(2) = 1.0 (step 2 result)

[0256] Dynamic ratio: R dynamic (2,j) = 1.2 x 0.984 x 1.0 ≈ 1.181

[0257] Step 4: Calculate the adjusted value of the associated curve:

[0258] Adjusted value V of the i-th layer of the associated curve j j_new (i):

[0259] V j_new (i) = V base_new (i) x R dynamic (i,j)

[0260] Example:

[0261] Modified value V of the intermediate layer in the reference curve base_new (2) = 250 μm

[0262] Dynamic ratio R dynamic (2,j) = 1.181 (step 3 result)

[0263] Adjusted value of the intermediate layer in the associated curve j: 250 x 1.181 ≈ 295.25 μm

[0264] Three, complete calculation process example (multi-layer scenario)

[0265] Assuming the reference curve (A) modifies the intermediate layer (layer 2), the calculation process of the associated curve (B) is as follows:

[0266]

[0267] Four, key logic explanation

[0268] Time dimension correction: through R t Offset the drift error of long-term operation of the device, the longer the correction range is larger;

[0269] Spatial dimension optimization: through W(i) to ensure the priority of the ratio accuracy of key layers, avoid the influence of secondary layer error on the overall result;

[0270] Real-time guarantee: all parameters (such as t modify , V base_new) are all real-time extraction, and the calculation process is completed within 100 ms, meeting the real-time demand of synchronous adjustment.

[0271] Visual matching interface: The matching results and dynamic scaling calculation process of the target curve and other curves are displayed in an interactive table and graphical manner, with matching degree and scaling confidence marked, supporting manual confirmation or adjustment of matching relationship and scaling parameters by the operator.

[0272]

[0273]

[0274] The curve synchronous adjustment method provided by the embodiments of the present application further includes an abnormality detection step after determining the adjustment parameter in step S500, specifically including:

[0275] Dual threshold detection: Set a hard threshold and a soft threshold for abnormality detection; the hard threshold is a limit value range based on the physical characteristics of the curve, when the modified value of the adjustment parameter or the calculated adjustment parameter adjustment value exceeds the hard threshold, a warning is triggered and the adjustment is refused to be executed; the soft threshold calculates the normal fluctuation range through the 3σ principle, when the adjustment value deviation exceeds 2σ, a backup scaling value is used for adjustment, and the backup scaling is the mean value of the last 3 effective scalings.

[0276] Dual threshold detection formula:

[0277] Hard threshold formula (taking depth value as an example): Hard threshold range = [0, maximum range of equipment] (out of range is determined as abnormal).

[0278] Soft threshold core formula:

[0279] Soft threshold range = [μ-2σ, μ+2σ]

[0280] (n = 30, i.e. taking the last 30 effective data as samples).

[0281] Backup scaling calculation:

[0282] (R1, R2, R3 are the scaling values of the last 3 effective adjustments).

[0283] Breakpoint recovery: The blockchain technology is used to store evidence of key scaling data and adjustment records, when the system is interrupted abnormally, the last effective state can be traced back based on the stored evidence data, avoiding data loss and error accumulation;

[0284] Abnormality type and determination basis

[0285]

[0286]

[0287] State backtracking can refer to the attached Figure 4 .

[0288] Abnormal positioning and prompting: when an abnormality is detected, the source of the abnormality is located through log analysis, such as input errors, scale model failures, etc., and the operator is prompted in an intuitive manner to assist in problem solving.

[0289] The curve synchronization method provided by the application adopts double threshold detection and breakpoint recovery function of blockchain storage, effectively identifies and handles abnormal situations, improves the robustness and reliability of the system, and reduces the curve adjustment errors caused by abnormalities.

[0290] The curve synchronization adjustment method provided by the embodiment of the application is illustrated as shown in Figure 5 The step S500 is exemplarily illustrated as follows:

[0291] Adjustment parameter generation:

[0292] The total depth of the target curve (reference curve H) is modified from 4.78722 μm to 6 μm;

[0293] The first layer rate of H becomes 147.89, and the second layer becomes 71.44

[0294] There are four associated curves (Al, C, O, Si), and the scale values after dynamic calibration are as follows:

[0295] The adjustment parameters to be generated include: the depth adjustment value of each curve and each layer, and the difference (Δ) before and after adjustment.

[0296] The scale matrix after dynamic calibration (has combined time attenuation and layer weight)

[0297]

[0298]

[0299] Step S600: batch adjustment of curves, specifically: the algorithm execution is realized in parallel in a multi-threaded manner. Examples:

[0300] Suppose the curves to be adjusted are A, B, C, D, and E (all pass the parameter verification), and the server is a 4-core CPU (thread pool initialization 6 threads):

[0301] Task decomposition: divided into 3 task groups (group 1: A, B; group 2: C; group 3: D, E);

[0302] Thread allocation: thread 1 handles group 1, thread 2 handles group 2, and thread 3 handles group 3;

[0303] Parallel execution:

[0304] Thread 1: adjust A first (all sub-layers 1-3 succeed), then adjust B (sub-layer 2 fails, triggering rollback, B is marked as "failed");

[0305] Thread 2: adjust C (all sub-layers succeed);

[0306] Thread 3: adjust D (succeed), E (succeed);

[0307] Status monitoring: UI displays "Total progress: 4 / 5 (Curve B failed)" in real time;

[0308] Result summary:

[0309] Successful curves: A, C, D, E;

[0310] Failed curve: B (reason: sub-layer 2 adjustment value exceeds soft threshold);

[0311] Conflict detection: sub-layer proportion deviation of A and C is 1.2% (normal), no conflict.

[0312] The curve synchronous adjustment method provided by the application combines a multi-dimensional matching mechanism and a batch execution engine to achieve accurate matching and efficient batch adjustment of modified curves and associated curves, greatly improving operation efficiency and reducing manual intervention costs.

[0313] The curve synchronous adjustment method provided by the embodiment of the application further includes S700 proportion model iterative upgrade after batch synchronous adjustment, specifically including:

[0314] S710, data collection and analysis, collecting modification record data, verification result data, and abnormal processing situation data in the batch adjustment process, analyzing the applicability and accuracy of the proportion model through a machine learning model, and identifying the deviation law of the proportion model in different scenarios;

[0315] S720, model optimization and upgrade: based on the analysis results, at least one of the following is performed on the proportion model: adjusting the time decay factor coefficient, updating the sub-layer weight coefficient, and adding a proportion calculation rule for a special scenario. For scenarios where low confidence proportions frequently occur, a new proportion model is recommended to be created to realize dynamic iterative upgrade of the proportion model and continuously improve the performance of curve synchronous adjustment.

[0316] The curve synchronous adjustment method of the further preferred embodiment of the application enables the system to adapt to different curve characteristic scenarios such as continuous sub-layers and discrete sub-layers, and the performance is continuously optimized as the usage time increases. Through automatic verification, interactive optimization, and machine learning driven model iterative upgrade, accurate control of adjustment results and continuous optimization of proportion models are realized, enhancing the adaptability of the system to different scenarios.

[0317] The above description is merely that of a specific implementation of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various changes or replacements within the technical scope disclosed by the present application, and these should be encompassed within the scope of protection of the present application. Therefore, the scope of protection of the present application should be subject to the scope of protection of the claims.

Claims

1. A method of curve synchronization adjustment, characterized by, The method comprises the following steps of: S100, establishing a proportion reference, taking a set of layered and aligned curves as a reference group, identifying and extracting initial proportion parameters of different layers between curves in the reference group to form an initial proportion matrix; S200, detecting target curve modification, detecting in real time whether a curve is modified, taking the curve subjected to the modification operation as a target curve when it is detected that the parameters of the curve are modified, and recording information of the target curve; S300, extracting modification information, analyzing curve data after modification of the target curve, and extracting at least layer index information, modification value size information and modification range information involved in the modification; S400, determining a curve to be modified, determining a curve to be modified according to information of the target curve and the proportion reference, and determining the curve to be modified through dimensional matching; S500, determining an adjustment parameter, obtaining an adjustment proportion value of each curve to be modified and modification information of the target curve according to the initial proportion matrix, and generating an adjustment parameter of each curve to be modified; S600, adjusting the curves in batches, starting a batch execution engine, and synchronously adjusting all curves to be modified according to the generated adjustment parameters.

2. The curve synchronization adjustment method of claim 1, wherein, After the step S100 of forming the initial proportion matrix, the method further comprises the following steps of: S110, constructing a dynamic proportion model, introducing a time attenuation factor and a layer weight coefficient, and constructing a dynamic proportion model; S120, managing a proportion library, and managing the dynamic proportion model by establishing a proportion library system.

3. The curve synchronization adjustment method of claim 2, wherein, The step S400 specifically comprises: based on characteristics of the dynamic proportion model in the proportion library system, performing curve type consistency verification, layer structure similarity comparison and historical adjustment record correlation analysis on the target curve and other associated curves.

4. The curve synchronization adjustment method of claim 3, wherein, In the step S500, the adjustment proportion value is obtained through dynamic calculation. The dynamic calculation specifically comprises: calling the dynamic proportion model according to the extracted modification information and the matching result, combining the time attenuation factor and the layer weight coefficient, and calculating the adjustment proportion value of each layer of the curve to be modified in real time.

5. The method of curve synchronization adjustment as claimed in claim 2, wherein, After the step S500 of determining the adjustment parameter, the method further comprises an abnormality detection step, specifically comprising: Double threshold detection: setting a hard threshold and a soft threshold for abnormality detection; the hard threshold is a limit value range based on physical characteristics of the curve, when the modification value of the adjustment parameter or the calculated adjustment parameter adjustment value exceeds the hard threshold, a warning is triggered and the adjustment is refused to be executed; the soft threshold is a normal fluctuation range calculated by the 3σ principle, when the adjustment value deviation exceeds 2σ, a backup proportion value is used for adjustment.

6. The curve synchronization adjustment method of claim 5, wherein, After the abnormality detection step, the method further comprises the following steps: Breakpoint recovery: using a blockchain technology to store evidence of key proportion data and adjustment records, and when the system is abnormally interrupted, the system is backtracked to the nearest valid state based on the stored evidence data; Abnormality positioning and prompting: when the abnormality is detected, the abnormality source is positioned through log analysis, and a prompt is given.

7. The curve synchronization adjustment method of claim 6, wherein, After the step S600, the method further comprises a step S700 of proportion model iterative upgrading, specifically comprising: S710, data collection and analysis, collect modification record data, verification result data, abnormal processing situation data in batch adjustment process, analyze the applicability and accuracy of the proportion model through machine learning model, and identify the deviation law of the proportion model in different scenarios; S720, model optimization and upgrading: based on the analysis result, at least one of the following is included: adjusting the time decay factor coefficient, updating the hierarchical weight coefficient, and adding a special scene proportion calculation rule.

8. An electronic device, comprising: The memory stores a computer executable program, and the processor calls the computer executable program in the memory to realize the curve synchronous adjustment method of any one of claims 1 to 7.

9. A storage medium, characterized by The storage medium is a computer readable storage medium, and the computer readable storage medium stores a computer executable program, and the computer executable program is executed by the processor to realize the curve synchronous adjustment method of any one of claims 1 to 7.

10. A computer program product for enabling curve synchronization adjustment, characterized in that, The code for realizing the curve synchronous adjustment method of any one of claims 1 to 7.

Images