Method, device and electronic equipment for identifying samples on tray

Abstract

The present invention provides a method, device and electronic device for identifying samples on a tray, and relates to the field of Raman spectroscopy analysis. The method comprises: removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal; when it is determined that the Raman intensity corresponding to the Raman data after removal is less than or equal to a preset noise signal, processing the Raman data after removal to obtain all peak data, sorting the Raman intensities corresponding to all peak data to obtain a target arrangement array; respectively calculating the 25th percentile and the 95th percentile of the target arrangement array, determining the noise interval difference, and determining the intensity threshold according to the noise interval difference and the 95th percentile; when it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, determining that there is a sample on the current tray, otherwise, determining that there is no sample on the current tray. The present invention effectively eliminates the interference of these special signals and reduces the false positive rate by identifying the silicon wafer peak and the characteristic peak of the tray.

Description

Technical Field

[0001] The invention relates to the field of Raman spectroscopy analysis, and in particular to a method, a device and an electronic device for identifying a sample on a tray. Background Art

[0002] In the field of material research and development, Raman spectroscopy is an important characterization method and is widely used to record and analyze the peak position and intensity information of material properties. However, in actual operation, the interpretation of Raman spectra faces a series of challenges, especially when judging whether the sample to be measured actually exists on the carrier to be tested. First, even if there is no real sample on the carrier to be tested (such as a silicon wafer), the Raman spectrometer will still characterize it and generate noise and background characteristic signals. The existence of these signals requires relevant personnel to identify the real sample characteristics from the Raman spectrum to ensure the accuracy of subsequent data analysis.

[0003] The traditional identification process mainly relies on manual work. Researchers will manually set the noise threshold according to the characteristics of the Raman spectrum and check whether there is a signal above the threshold. If so, it is considered that the sample peak exists. However, this method has obvious defects: it is highly dependent on manual analysis and cannot be seamlessly integrated with modern automated equipment, thus limiting the efficiency and accuracy of detection. In addition, although the manual method can ensure that the characteristic information of the tray material is not scanned by the silicon wafer through operating skills, in automated equipment, due to the accuracy limitations of the robotic arm, there is a risk of scanning the tray material and generating a false positive judgment. This is because the Raman characteristic signal of the tray material may be similar to the sample signal, resulting in algorithm misjudgment.

[0004] In summary, the existing technology has two major defects in the interpretation of Raman spectra: one is that it relies on manual analysis and cannot be combined with automated equipment; the other is that there is a risk of false positive judgment due to scanning of tray materials. These defects limit the automation and precision application of Raman spectroscopy in the field of material research and development. At present, there is no technical solution that can solve the above technical problems, and there is no method, device and electronic equipment for identifying samples on trays. Summary of the invention

[0005] The present invention provides a method, device and electronic device for identifying samples on a tray, which can effectively avoid false positive identification of tray samples, use the quantile method to identify Raman data to determine whether there is a sample on the tray, and improve the accuracy of identification.

[0006] In a first aspect, the present invention provides a method for identifying a sample on a tray, comprising:

[0007] Eliminating characteristic peaks of the silicon wafer from target Raman data to obtain Raman data after elimination, wherein the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray;

[0008] When it is determined that the Raman intensity corresponding to the eliminated Raman data is less than or equal to the preset noise signal, the eliminated Raman data is processed by a peak recognition algorithm to obtain all peak data, and the Raman intensities corresponding to all peak data are sorted in descending order to obtain a target arrangement array;

[0009] Calculate the 25th quantile and the 95th quantile of the target array respectively, determine the noise interval difference according to the 25th quantile and the 95th quantile, and determine the intensity threshold according to the noise interval difference and the 95th quantile;

[0010] When it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, it is determined that there is a sample on the tray; otherwise, it is determined that there is no sample on the tray.

[0011] According to the method for identifying samples on a tray provided by the present invention, before using a peak identification algorithm to process the eliminated Raman data to obtain all peak data, the method further includes:

[0012] identifying the noise signal according to the Raman spectrum corresponding to the Raman data after elimination;

[0013] The preset noise signal is determined according to the product of the noise signal and a preset coefficient.

[0014] According to the method for identifying samples on a tray provided by the present invention, the method of processing the eliminated Raman data using a peak identification algorithm to obtain all peak data includes:

[0015] For any frequency point in the Raman data after elimination, when the Raman intensity at the frequency point is greater than the Raman intensity at an adjacent frequency point, determining the frequency point as peak data;

[0016] The adjacent frequency points include a frequency point before the frequency point and a frequency point after the frequency point;

[0017] Traverse all frequency points and get all peak data.

[0018] According to the method for identifying samples on a tray provided by the present invention, the step of respectively calculating the 25th percentile and the 95th percentile of the target arrangement array comprises:

[0019]

[0020] in, is the 25th percentile, is the integer part of the 25th percentile position, is the decimal part of the 25th percentile position, Arrange the array for the target, is the 95th percentile, is the integer part of the 95th percentile position, is the decimal part of the 95th percentile position;

[0021] Among them, the 25th percentile position POS 25 0.25n, 95th percentile position POS 95 is 0.95n, where n is the total number of the target arrangement arrays.

[0022] According to the method for identifying samples on a tray provided by the present invention, determining the noise interval difference according to the 25th percentile and the 95th percentile includes:

[0023]

[0024] in, is the 95th percentile, is the 25th percentile, The noise interval difference is the noise interval difference.

[0025] According to the method for identifying samples on a tray provided by the present invention, determining the intensity threshold according to the noise interval difference and the 95th percentile includes:

[0026]

[0027] in, is the preset empirical coefficient, The value of is 6. is the intensity threshold. is the intensity threshold.

[0028] According to the method for identifying samples on a tray provided by the present invention, after obtaining the Raman data after elimination, the method further includes:

[0029] When it is determined that the Raman intensity corresponding to the Raman data after elimination is greater than a preset noise signal, it is determined that there is a sample on the tray.

[0030] According to the method for identifying samples on a tray provided by the present invention, before removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal, the method further includes:

[0031] Inputting the original Raman data into a preset fluorescence elimination model to obtain target fluorescence background data output by the preset fluorescence elimination model;

[0032] Eliminating the target fluorescence background data from the original Raman data to obtain the target Raman data;

[0033] The preset fluorescence elimination model is determined after training all sample Raman data obtained under a preset experimental environment and each sample fluorescence background data after fluorescence marking in each sample Raman data.

[0034] In a second aspect, a device for identifying samples on a tray is provided, comprising:

[0035] A rejection unit, the rejection unit is used to reject characteristic peaks of silicon wafers from target Raman data to obtain Raman data after rejection, the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray;

[0036] A processing unit, wherein the processing unit is used to process the Raman data after elimination by using a peak recognition algorithm to obtain all peak data when it is determined that the Raman intensity corresponding to the Raman data after elimination is less than or equal to a preset noise signal, and sort the Raman intensities corresponding to all peak data in order from large to small to obtain a target arrangement array;

[0037] A calculation unit, the calculation unit is used to calculate the 25th quantile and the 95th quantile of the target arrangement array respectively, determine the noise interval difference according to the 25th quantile and the 95th quantile, and determine the intensity threshold according to the noise interval difference and the 95th quantile;

[0038] A determination unit is used to determine that there is a sample on the tray if it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, otherwise, determine that there is no sample on the tray.

[0039] In a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a method for identifying samples on the tray when executing the program.

[0040] In automated equipment, due to the accuracy of the robot arm, it is possible to scan the characteristic signal of the tray material, resulting in a false positive judgment. The present invention effectively eliminates the interference of these special signals by identifying the silicon wafer peak and the tray characteristic peak, thereby reducing the false positive rate. Since the algorithm can identify and avoid the interference of the tray signal, in the design and implementation of automated equipment, it is not necessary to strictly require that the sample be embedded in the tray at a certain specific angle, which increases the fault tolerance of the automated equipment and makes the equipment more flexible and stable in use.

[0041] The present invention adopts the quantile method to identify the noise in the Raman spectrum, determines the distribution range of the noise by statistically analyzing the distribution of peaks and calculating the values ​​of the 25th and 95th percentiles, thereby more accurately identifying the noise signal, and can dynamically calculate the appropriate threshold value according to the noise information analyzed from the Raman spectrum signal. Compared with the method of setting a fixed threshold value, it is more flexible and robust. By optimizing the noise recognition algorithm, the present invention can more accurately identify the characteristic peaks of the sample, reduce the situation of misjudgment and missed judgment, and improve the accuracy and reliability of sample recognition; the Raman spectrum data is automatically read and analyzed by an intelligent algorithm without manual intervention, which significantly improves the efficiency of sample recognition, accurately identifies the noise and signal in the Raman spectrum, avoids the situation of misjudging the noise signal as a sample peak, and improves the accuracy of recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 This is one of the flow diagrams of the method for identifying samples on a tray provided by the present invention;

[0044] Figure 2 This is the second flow chart of the method for identifying samples on a tray provided by the present invention;

[0045] Figure 3 It is a schematic structural diagram of a device for identifying samples on a tray provided by the present invention;

[0046] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Figure 1 This is one of the flow diagrams of the method for identifying a sample on a tray provided by the present invention, and the method for identifying a sample on a tray comprises:

[0049] Step 101, removing characteristic peaks of the silicon wafer from target Raman data to obtain Raman data after removal, wherein the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray;

[0050] Step 102: When it is determined that the Raman intensity corresponding to the eliminated Raman data is less than or equal to the preset noise signal, the eliminated Raman data is processed using a peak recognition algorithm to obtain all peak data, and the Raman intensities corresponding to all peak data are sorted in descending order to obtain a target arrangement array;

[0051] Step 103, respectively calculating the 25th quantile and the 95th quantile of the target arrangement array, determining the noise interval difference according to the 25th quantile and the 95th quantile, and determining the intensity threshold according to the noise interval difference and the 95th quantile;

[0052] Step 104: When it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, it is determined that there is a sample on the current tray; otherwise, it is determined that there is no sample on the current tray.

[0053] In step 101, this step is to remove the characteristic peak generated by the silicon wafer itself in the Raman spectrum to prevent misidentification in the subsequent signal intensity judgment. The characteristic peak of the silicon wafer usually appears at about 520cm -1 The present invention uses an algorithm or preprocessing method to identify and remove this characteristic peak. Assuming that the Raman spectrum data set contains the characteristic peak of silicon wafers and other possible noise and sample peaks, the algorithm is used to identify the peak at 520cm -1 The characteristic peaks near the sample are removed from the data set, and the data set after removal will only contain noise, other possible sample peaks and possible tray characteristic peaks.

[0054] Optionally, for samples carried on silicon wafers, the peak of the silicon wafer needs to be excluded first to ensure that there is no misidentification when judging the data intensity. In Raman spectroscopy analysis, the characteristic peak of the silicon wafer mainly appears at about 520cm -1 The peak corresponds to the lattice vibration mode of the silicon wafer, that is, the characteristic Raman scattering caused by the optical phonon mode of the silicon crystal. The position and shape of this peak can provide information about the crystal quality, stress state and doping status of the silicon material. For high-quality single-crystal silicon, the peak is generally sharp and symmetrical, so such peaks should be eliminated in the algorithm logic, and then the signal comparison and recognition should be performed. Because it is an automated device, in some special cases, it will scan outside the silicon wafer during scanning. At this time, the algorithm needs to be able to judge the tray data scanned outside the silicon wafer and cannot be mistakenly identified as a sample.

[0055] Optionally, before removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal, the method further includes:

[0056] Inputting the original Raman data into a preset fluorescence elimination model to obtain target fluorescence background data output by the preset fluorescence elimination model;

[0057] Eliminating the target fluorescence background data from the original Raman data to obtain the target Raman data;

[0058] The preset fluorescence elimination model is determined after training all sample Raman data obtained under a preset experimental environment and each sample fluorescence background data after fluorescence marking in each sample Raman data.

[0059] Optionally, the raw Raman data is data directly acquired by a Raman spectrometer, which is input into a pre-trained fluorescence elimination model, and the preset fluorescence elimination model is used to identify the target fluorescence background data, and then remove the fluorescence background signal from the raw Raman data, which may be caused by various factors in the experimental environment, such as light source, detector, sample itself, etc. After the fluorescence elimination model processes the raw Raman data, it will output a data set representing the fluorescence background, which represents those signals that do not belong to the sample itself but are caused by the preset experimental environment, namely the target fluorescence background data. Further, the obtained fluorescence background data is removed from the raw Raman data to obtain purer target Raman data, so that the data after elimination more accurately reflects the Raman spectral characteristics of the sample itself.

[0060] Optionally, the present invention can significantly reduce the noise and interference in the data by removing the fluorescence background data, thereby improving the accuracy of the Raman spectrum data and helping to more accurately identify the chemical composition and structural information in the sample; the use of a pre-trained fluorescence elimination model can adapt to different experimental environments and sample types, thereby enhancing the robustness of the algorithm and more effectively handling various complex situations, such as different types of trays, different sample placement methods, etc. Since this embodiment can more accurately identify sample signals and exclude interference signals, it can significantly improve fault tolerance when used in automated equipment. Even if the robot arm accuracy is not high enough or the sample placement is not accurate enough, this embodiment can correctly identify the sample and avoid false positive judgments.

[0061] Optionally, after obtaining the Raman data after elimination, the method further includes:

[0062] When it is determined that the Raman intensity corresponding to the Raman data after elimination is greater than a preset noise signal, it is determined that there is a sample on the current tray.

[0063] Optionally, after obtaining the Raman data after removing the characteristic peaks of the silicon wafer, the present invention adds a judgment step: when it is determined that the Raman intensity corresponding to the Raman data after the removal is greater than the preset noise signal, it is determined that there is a sample on the current tray, and whether there is a sample is determined by judging the Raman intensity. The present invention further processes the Raman data after removing the characteristic peaks of the silicon wafer, which may include identifying and removing the characteristic peaks of the tray, and judging whether the Raman intensity corresponding to the Raman data after the removal has an intensity value greater than the preset noise signal after the processed Raman data. If the Raman intensity corresponding to the Raman data after the removal is greater than the preset noise signal, it is determined that there is a sample on the current tray.

[0064] The present invention can determine that there is a sample on the current tray when it is determined that the Raman intensity is greater than a preset noise signal, without using a subsequent quantile algorithm to judge subsequent tray samples, thereby shortening the calculation steps and the recognition procedure, improving the recognition efficiency, and reducing the labor cost and time cost.

[0065] In step 102, first check whether the Raman intensity of the Raman data after removing the characteristic peak of the silicon wafer is less than or equal to the preset noise signal. If so, further use the peak recognition algorithm to identify and extract all possible peak data, and then sort these peak data according to Raman intensity. Through sorting and screening, we can more easily identify the high-intensity peaks that may represent the sample, and provide a basis for the noise interval difference calculation and intensity threshold determination in the subsequent steps.

[0066] Optionally, before using a peak recognition algorithm to process the eliminated Raman data to obtain all peak data, the method further includes:

[0067] identifying the noise signal according to the Raman spectrum corresponding to the Raman data after elimination;

[0068] The preset noise signal is determined according to the product of the noise signal and a preset coefficient.

[0069] In such an embodiment, the present invention first inputs the original Raman data into a preset fluorescence elimination model to obtain target fluorescence background data, and removes the background data from the original Raman data to obtain target Raman data; then removes the characteristic peak of the silicon wafer from the target Raman data. The characteristic peak of the silicon wafer is a known and fixed signal in Raman spectrum analysis. Removing it can avoid misjudgment in subsequent analysis; after removing the characteristic peak of the silicon wafer, the noise signal is identified according to the Raman spectrum corresponding to the removed Raman data. This step is to find out the noise component therein by analyzing the Raman spectrum to provide a basis for subsequent judgment. The preset noise signal is determined according to the product of the identified noise signal and a preset coefficient. The preset coefficient is an empirical value, and its size can be adjusted according to actual conditions. After determining the preset noise signal, it can be further determined that "when it is determined that the Raman intensity corresponding to the removed Raman data is less than or equal to the preset noise signal, the removed Raman data is processed using a peak recognition algorithm to obtain all peak data, and the Raman intensities corresponding to all peak data are sorted in order from large to small to obtain a target arrangement array."

[0070] Optionally, the use of a peak recognition algorithm to process the eliminated Raman data to obtain all peak data includes:

[0071] For any frequency point in the Raman data after elimination, when the Raman intensity at the frequency point is greater than the Raman intensity at an adjacent frequency point, determining the frequency point as peak data;

[0072] The adjacent frequency points include a frequency point before the frequency point and a frequency point after the frequency point;

[0073] Traverse all frequency points and get all peak data.

[0074] Optionally, the present invention first defines a recognition standard for peak data. For any frequency point in the Raman data after elimination, if the Raman intensity at this frequency point is greater than the Raman intensity at its adjacent frequency points (including the frequency points before the frequency point and the frequency points after the frequency point), then the frequency point is determined to be peak data. This standard is based on the basic characteristics of the peak in the Raman spectrum, that is, the Raman intensity at the peak top is usually higher than the Raman intensity at its adjacent positions. Next, all peak data that meet the above conditions are found by traversing all frequency points. In this way, all peak data in the Raman data after eliminating the characteristic peaks of silicon wafers can be obtained, providing a basis for subsequent recognition algorithms.

[0075] Optionally, first use a peak identification algorithm to obtain all peaks in the Raman spectrum, that is, all data points that satisfy the following formula:

[0076] f(x)>f(x+1) and f(x)>f(x-1);

[0077] where x is any frequency point (micrometers) and f(x) is the Raman intensity at that frequency.

[0078] In step 103, the concept of quantile in statistics is used to calculate the noise interval difference, and an intensity threshold is determined accordingly. This intensity threshold will be used in subsequent steps to determine whether there is a sample peak. By calculating the intensity threshold, a standard is set to determine which peaks may represent sample peaks. This intensity threshold is dynamically calculated based on the statistical characteristics of Raman data, so it has certain versatility and robustness.

[0079] Optionally, respectively calculating the 25th percentile and the 95th percentile of the target arrangement array includes:

[0080]

[0081] in, is the 25th percentile, is the integer part of the 25th percentile position, is the decimal part of the 25th percentile position, Arrange the array for the target, is the 95th percentile, is the integer part of the 95th percentile position, is the decimal part of the 95th percentile position;

[0082] Among them, the 25th percentile position POS 25 0.25n, 95th percentile position POS 95 is 0.95n, where n is the total number of the target arrangement arrays.

[0083] Optionally, determining the noise interval difference according to the 25th percentile and the 95th percentile includes:

[0084]

[0085] in, is the 95th percentile, is the 25th percentile, is the noise interval difference.

[0086] Optionally, determining the intensity threshold according to the noise interval difference and the 95th percentile includes:

[0087]

[0088] in, is the preset empirical coefficient, The value of is 6. is the intensity threshold.

[0089] Optionally, in the present invention, some of the peak data in the target array are noise and some are sample peaks, and the noise needs to be removed by the following algorithm. By referring to the concept of box plot, the distribution of these peaks is statistically analyzed to obtain the 25th and 95th percentile values ​​to determine the distribution of noise. The quantile algorithm is as follows:

[0090] Sort the target array F = f(p1), f(p2) ... f(pn)) from large to small, where p1, p2 ... pn are the peak positions obtained by the above peak recognition algorithm, and n is the total number of peaks. The quantile algorithm is as follows, taking the 25th quantile as an example: 25th quantile position POS 25 = 0.25 × n = i + f, F[i 25 ] represents the i+1th number in the array F. For example: suppose F = 3, 5, 6, 7, 10, then POS 25 = 0.25 × 5 = 1.25, so i = 1, f = 0.25. Then according to the calculation formula for the 25th percentile, q25 = F[1] + 0.25 × F[2]-F[1] = 5 + 0.25×(6-5) = 5.25. The calculation method for the 95th percentile is similar.

[0091] Optionally, the noise identification algorithm is: R = ANY (F (x) > T), R is the final judgment result, and the peak identification algorithm is traversed to obtain the target peaks in all Raman spectra. If the signal of any target peak is greater than T, it is considered that there is a sample.

[0092] In step 104, step 104 is the final judgment step, which judges whether there is a sample on the current tray according to the intensity threshold calculated in the previous step, traverses these peak data, and checks whether its Raman intensity is greater than the intensity threshold T. If there is any peak whose Raman intensity is greater than T, it is considered that there is a sample on the current tray; otherwise, it is considered that there is no sample on the current tray.

[0093] In automated equipment, due to the accuracy of the robot arm, it is possible to scan the characteristic signal of the tray material, resulting in a false positive judgment. The present invention effectively eliminates the interference of these special signals by identifying the silicon wafer peak and the tray characteristic peak, thereby reducing the false positive rate. Since the algorithm can identify and avoid the interference of the tray signal, in the design and implementation of automated equipment, it is not necessary to strictly require that the sample be embedded in the tray at a certain specific angle, which increases the fault tolerance of the automated equipment and makes the equipment more flexible and stable in use.

[0094] The present invention adopts the quantile method to identify the noise in the Raman spectrum, determines the distribution range of the noise by statistically analyzing the distribution of peaks and calculating the values ​​of the 25th and 95th percentiles, thereby more accurately identifying the noise signal, and can dynamically calculate the appropriate threshold value according to the noise information analyzed from the Raman spectrum signal. Compared with the method of setting a fixed threshold value, it is more flexible and robust. By optimizing the noise recognition algorithm, the present invention can more accurately identify the characteristic peaks of the sample, reduce the situation of misjudgment and missed judgment, and improve the accuracy and reliability of sample recognition; the Raman spectrum data is automatically read and analyzed by an intelligent algorithm without manual intervention, which significantly improves the efficiency of sample recognition, accurately identifies the noise and signal in the Raman spectrum, avoids the situation of misjudging the noise signal as a sample peak, and improves the accuracy of recognition.

[0095] Figure 2 This is the second flow chart of the method for identifying samples on a tray provided by the present invention. As an optional embodiment of the present invention, the current routine laboratory practice relies on manual judgment of the noise size, and confirms whether there is a sample by comparing the Raman data intensity with the noise signal. When it is less than or equal to "n times the noise signal", it is assumed that there is no sample, and when it is greater than, it is determined that there is a sample. This n times requires specific data to support it, and analysis and parameter adjustment are performed.

[0096] After analyzing and adjusting the Raman characteristic data of some test sample features, non-sample (silicon wafer) features, and tray materials, it was finally found that all sample identification scenarios can be covered when n is 6. Since the tray data is identified by the algorithm, some false positives are found, and the peak shape is also close to the sample. At this time, special identification is performed based on the tray sample characteristics to ensure that no false positive scenarios occur.

[0097] In the present invention, if Figure 2 As shown, it includes: step 1, the program directly reads the Raman data through the machine API; step 2, the Raman data is defluoresced; step 3, the characteristic peaks of the silicon wafer itself in the Raman data are eliminated to ensure that no misjudgment occurs; step 4, based on the Raman data, it is detected whether it is a PTFE tray; step 5, the intelligent algorithm analyzes the Raman data to determine whether there is a sample characteristic peak in the data; step 6, the intelligent algorithm completes the judgment and returns whether it is a sample.

[0098] The noise identification algorithm of the present invention enables the sample identification operation to be automated. By identifying the signal of the Raman spectrum and the noise information, the algorithm can directly determine whether there is a sample based on the Raman spectrum information, avoiding the misconception that there is a sample due to the noise signal. Since the method is to parse the noise from the Raman spectrum signal and calculate the appropriate threshold, the method has a certain versatility and can be used for different Raman devices and different systems. It is more robust than the method of setting a fixed threshold. At the same time, the algorithm can identify and avoid the situation where the Raman spectrum presents the characteristic signal of the tray material due to the Raman device scanning the tray, causing the false positive. Therefore, in the design and implementation of the automated equipment, there is no need to deliberately require the sample to be embedded in the tray at a certain specific angle, which increases the fault tolerance of the automated equipment and the robustness of the peak identification algorithm. The present invention identifies the noise in the Raman spectrum by calculating the quantile method to determine whether the scanned object corresponding to the Raman spectrum contains a sample; by identifying the silicon wafer peak and the tray characteristic peak to exclude the special signal in the Raman spectrum, the false positive situation in the sample identification is avoided.

[0099] Figure 3 : is a structural schematic diagram of a sample identification device on a tray provided by the present invention, the sample identification device on the tray comprises a rejection unit 1, the rejection unit is used to reject characteristic peaks of silicon wafers from target Raman data to obtain Raman data after rejection, the target Raman data is determined after reading the current tray, the silicon wafer is clamped at a preset position of the current tray, the working principle of the rejection unit 1 can refer to the aforementioned step 101, and will not be repeated here.

[0100] The device for identifying samples on the tray also includes a processing unit 2, which is used to process the Raman data after elimination using a peak recognition algorithm to obtain all peak data when it is determined that the Raman intensity corresponding to the Raman data after elimination is less than or equal to a preset noise signal, and sort the Raman intensities corresponding to all peak data in order from large to small to obtain a target arrangement array. The working principle of the processing unit 2 can refer to the aforementioned step 102 and will not be repeated here.

[0101] The device for identifying samples on the tray also includes a calculation unit 3, which is used to calculate the 25th percentile and the 95th percentile of the target arrangement array, respectively, determine the noise interval difference according to the 25th percentile and the 95th percentile, and determine the intensity threshold according to the noise interval difference and the 95th percentile. The working principle of the calculation unit 3 can be referred to the aforementioned step 103 and will not be repeated here.

[0102] The device for identifying samples on the tray further includes a determination unit 4, which is used to determine that there is a sample on the current tray when it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold; otherwise, it is determined that there is no sample on the current tray. The working principle of the determination unit 4 can be referred to the aforementioned step 104 and will not be described in detail here.

[0103] In automated equipment, due to the accuracy of the robot arm, it is possible to scan the characteristic signal of the tray material, resulting in a false positive judgment. The present invention effectively eliminates the interference of these special signals by identifying the silicon wafer peak and the tray characteristic peak, thereby reducing the false positive rate. Since the algorithm can identify and avoid the interference of the tray signal, in the design and implementation of automated equipment, it is not necessary to strictly require that the sample be embedded in the tray at a certain specific angle, which increases the fault tolerance of the automated equipment and makes the equipment more flexible and stable in use.

[0104] The present invention adopts the quantile method to identify the noise in the Raman spectrum, determines the distribution range of the noise by statistically analyzing the distribution of peaks and calculating the values ​​of the 25th and 95th percentiles, thereby more accurately identifying the noise signal, and can dynamically calculate the appropriate threshold value according to the noise information analyzed from the Raman spectrum signal. Compared with the method of setting a fixed threshold value, it is more flexible and robust. By optimizing the noise recognition algorithm, the present invention can more accurately identify the characteristic peaks of the sample, reduce the situation of misjudgment and missed judgment, and improve the accuracy and reliability of sample recognition; the Raman spectrum data is automatically read and analyzed by an intelligent algorithm without manual intervention, which significantly improves the efficiency of sample recognition, accurately identifies the noise and signal in the Raman spectrum, avoids the situation of misjudging the noise signal as a sample peak, and improves the accuracy of recognition.

[0105] Figure 4 Schematic diagram of the structure of the electronic device provided by the present invention. Figure 4As shown, the electronic device may include: a processor 110 , a communications interface 120 , a memory 130 and a communications bus 140 , wherein the processor 110 , the communications interface 120 , and the memory 130 communicate with each other via the communications bus 140 . The processor 110 can call the logic instructions in the memory 130 to execute the method for identifying the sample on the tray, which includes: removing the characteristic peak of the silicon wafer from the target Raman data to obtain the removed Raman data, wherein the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray; when it is determined that the Raman intensity corresponding to the removed Raman data is less than or equal to the preset noise signal, using the peak recognition algorithm to process the removed Raman data to obtain all peak data, and sorting the Raman intensities corresponding to all peak data in order from large to small to obtain a target arrangement array; respectively calculating the 25th quantile and the 95th quantile of the target arrangement array, determining the noise interval difference according to the 25th quantile and the 95th quantile, and determining the intensity threshold according to the noise interval difference and the 95th quantile; when it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, determining that the sample exists on the current tray, otherwise, determining that the sample does not exist on the current tray.

[0106] In addition, the logic instructions in the above-mentioned memory 130 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0107] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute a method for identifying a sample on a tray provided by the above methods, the method comprising: removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal, the target Raman data is determined after reading the current tray, the silicon wafer is clamped at a preset position of the current tray; after determining that the Raman intensity corresponding to the Raman data after removal is less than or equal to a preset noise In the case of a sound signal, a peak recognition algorithm is used to process the eliminated Raman data to obtain all peak data, and the Raman intensities corresponding to all peak data are sorted in order from large to small to obtain a target arrangement array; the 25th quantile and the 95th quantile of the target arrangement array are respectively calculated, and a noise interval difference is determined according to the 25th quantile and the 95th quantile, and an intensity threshold is determined according to the noise interval difference and the 95th quantile; when it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, it is determined that there is a sample on the current tray, otherwise, it is determined that there is no sample on the current tray.

[0108] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented by a processor to perform the identification method of samples on a tray provided by the above methods when executed by the processor, the method comprising: removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal, wherein the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray; when it is determined that the Raman intensity corresponding to the Raman data after removal is less than or equal to a preset noise signal, processing the Raman data after removal by using a peak recognition algorithm to obtain all peak data, sorting the Raman intensities corresponding to all peak data in order from large to small, and obtaining a target arrangement array; respectively calculating the 25th quantile and the 95th quantile of the target arrangement array, determining the noise interval difference according to the 25th quantile and the 95th quantile, and determining the intensity threshold according to the noise interval difference and the 95th quantile; when it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, determining that there is a sample on the current tray, otherwise, determining that there is no sample on the current tray.

[0109] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0110] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying samples on a tray, characterized in that: include: Eliminating characteristic peaks of the silicon wafer from target Raman data to obtain Raman data after elimination, wherein the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray; When it is determined that the Raman intensity corresponding to the eliminated Raman data is less than or equal to the preset noise signal, the eliminated Raman data is processed by a peak recognition algorithm to obtain all peak data, and the Raman intensities corresponding to all peak data are sorted in descending order to obtain a target arrangement array; Calculate the 25th quantile and the 95th quantile of the target array respectively, determine the noise interval difference according to the 25th quantile and the 95th quantile, and determine the intensity threshold according to the noise interval difference and the 95th quantile; When it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold, it is determined that there is a sample on the current tray; otherwise, it is determined that there is no sample on the current tray.

2. The method for identifying samples on a tray according to claim 1, characterized in that: Before using a peak recognition algorithm to process the eliminated Raman data to obtain all peak data, the method further includes: identifying the noise signal according to the Raman spectrum corresponding to the Raman data after elimination; The preset noise signal is determined according to the product of the noise signal and a preset coefficient.

3. The method for identifying samples on a tray according to claim 1, characterized in that: The peak recognition algorithm is used to process the Raman data after elimination to obtain all peak data, including: For any frequency point in the Raman data after elimination, when the Raman intensity at the frequency point is greater than the Raman intensity at an adjacent frequency point, determining the frequency point as peak data; The adjacent frequency points include a frequency point before the frequency point and a frequency point after the frequency point; Traverse all frequency points and get all peak data.

4. The method for identifying samples on a tray according to claim 1, characterized in that: The step of respectively calculating the 25th percentile and the 95th percentile of the target arrangement array comprises: ； in, is the 25th percentile, is the integer part of the 25th percentile position, is the decimal part of the 25th percentile position, Arrange the array for the target, is the 95th percentile, is the integer part of the 95th percentile position, is the decimal part of the 95th percentile position; Among them, the 25th percentile position POS 25 0.25n, 95th percentile position POS 95 is 0.95n, where n is the total number of the target arrangement arrays.

5. The method for identifying samples on a tray according to claim 1, characterized in that: Determining the noise interval difference according to the 25th percentile and the 95th percentile includes: ； in, is the 95th percentile, is the 25th percentile, is the noise interval difference.

6. The method for identifying samples on a tray according to claim 5, characterized in that: Determining the intensity threshold according to the noise interval difference and the 95th percentile includes: ； in, is the preset empirical coefficient, The value of is 6. is the intensity threshold. is the intensity threshold.

7. The method for identifying samples on a tray according to claim 1, characterized in that: After obtaining the post-elimination Raman data, the method further includes: When it is determined that the Raman intensity corresponding to the Raman data after elimination is greater than a preset noise signal, it is determined that there is a sample on the current tray.

8. The method for identifying samples on a tray according to claim 1, characterized in that: Before removing characteristic peaks of silicon wafers from target Raman data to obtain Raman data after removal, the method further includes: Inputting the original Raman data into a preset fluorescence elimination model to obtain target fluorescence background data output by the preset fluorescence elimination model; Eliminating the target fluorescence background data from the original Raman data to obtain the target Raman data; The preset fluorescence elimination model is determined after training all sample Raman data obtained under a preset experimental environment and each sample fluorescence background data after fluorescence marking in each sample Raman data.

9. A device for identifying samples on a tray, characterized in that: include: A rejection unit, the rejection unit is used to reject characteristic peaks of silicon wafers from target Raman data to obtain Raman data after rejection, the target Raman data is determined after reading the current tray, and the silicon wafer is clamped at a preset position of the current tray; A processing unit, wherein the processing unit is used to process the Raman data after elimination by using a peak recognition algorithm to obtain all peak data when it is determined that the Raman intensity corresponding to the Raman data after elimination is less than or equal to a preset noise signal, and sort the Raman intensities corresponding to all peak data in order from large to small to obtain a target arrangement array; A calculation unit, the calculation unit is used to calculate the 25th quantile and the 95th quantile of the target arrangement array respectively, determine the noise interval difference according to the 25th quantile and the 95th quantile, and determine the intensity threshold according to the noise interval difference and the 95th quantile; A determination unit is used to determine that there is a sample on the current tray if it is determined that the Raman intensity corresponding to any peak data is greater than the intensity threshold; otherwise, determine that there is no sample on the current tray.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for identifying samples on a tray according to any one of claims 1 to 8 is implemented.

Images