Intelligent early warning method and system for crystal rod quality

Abstract

The present invention relates to the field of data processing technology, and in particular to an intelligent early warning method and system for crystal rod quality. The method includes the steps of: obtaining a preset surrounding data segment of each temperature data point in a temperature data sequence of a crystal rod; calculating the necessity of filtering the temperature data point; correcting the necessity of filtering by the deviation between the temperature data sequence and the corresponding flow data sequence to obtain the possibility of filtering the temperature data point; obtaining the temperature data point to be filtered by comparing the filtering possibility of the temperature data point with a preset threshold; filtering and denoising the temperature data point to be filtered in the temperature data sequence, and processing the target temperature data sequence obtained after filtering and denoising by an ARIMA model to realize the early warning of the crystal rod quality, which effectively improves the accuracy of the early warning of the crystal rod quality.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent early warning method and system for crystal rod quality. Background Art

[0002] Crystal rods are widely used in semiconductor, photovoltaic and other high-tech industries due to their high hardness, high density and good strength. They are the basic materials for manufacturing high-tech products such as semiconductor devices and solar cells. The quality of crystal rods directly affects the product performance and reliability of related industries. If there are defects or quality problems in the crystal rods, it may cause the performance of the product to decline or fail. Therefore, it is necessary to monitor and warn the quality of the crystal rods.

[0003] In the existing schemes, the early warning of material quality is mostly predicted by obtaining the data deviation between the actual data and the predicted data in the material production process. For example, the patent application document with publication number CN117132019A discloses a coke quality prediction method, which provides historical data of raw coal ad value and coke ad value; preprocesses the historical data; sets parameters; continuously optimizes and iterates in combination with the plant growth network model; reduces the influence of interference factors through data collection and cleaning methods; and optimizes in combination with the coke loss function. The model updates the parameters in each iteration and adjusts the parameters according to the difference between the experimental prediction results and the experimental true values.

[0004] Although the above existing solutions can remove noise to a certain extent and realize early warning of material quality, they do not record how to retain the real original data while realizing filtering and denoising, so as to facilitate subsequent accurate early warning.

[0005] Based on this, how to accurately realize the early warning of crystal rod quality is a problem that needs to be solved urgently by technical personnel in this field. Summary of the invention

[0006] In order to solve the technical problem of how to accurately realize early warning of crystal rod quality, the present invention provides an intelligent early warning method and system for crystal rod quality.

[0007] In a first aspect, the present invention provides an intelligent early warning method for crystal ingot quality, which adopts the following technical solution:

[0008] An intelligent early warning method for crystal ingot quality comprises the following steps:

[0009] Acquire a preset surrounding data segment of each temperature data point in the temperature data sequence of the crystal rod;

[0010] Calculate the filtering necessity for the temperature data point: ;

[0011] In the formula, Indicates Necessity of filtering temperature data points, Indicates The temperature value of the temperature data point, Indicates The mean temperature in the data segment around the temperature data point, Indicates The number of modes in the data segment around the temperature data point, Indicates The mean of the temperature differences of all adjacent temperature data points in the data segment around the temperature data point, represents an exponential function with e as the base; the necessity of filtering is corrected by the deviation between the temperature data sequence and the corresponding flow data sequence to obtain the filtering possibility of the temperature data point; the temperature data point to be filtered is obtained by comparing the filtering possibility of the temperature data point with a preset threshold; the temperature data point to be filtered in the temperature data sequence is filtered and denoised, and the target temperature data sequence obtained after filtering and denoising is processed by the ARIMA model to realize early warning of crystal rod quality.

[0012] The present invention monitors and warns the quality of the crystal ingot by analyzing the temperature data on the surface of the crystal ingot. In this process, the present invention takes into account that noise data may interfere with the accuracy of the warning. Therefore, the temperature data point is filtered by obtaining the data performance of the temperature data point in the surrounding data segment to obtain the necessity of filtering the temperature data point, thereby accurately realizing the quality warning of the crystal ingot; on this basis, the present invention also takes into account that the noise data may be mixed with the real abnormal data. In order to further distinguish them, the present invention also corrects the necessity of filtering by obtaining flow data that is correlated with the temperature data, thereby effectively improving the accuracy of the crystal ingot quality warning.

[0013] According to an intelligent early warning method for crystal ingot quality provided by the present invention, the step of obtaining a preset surrounding data segment of each temperature data point in a temperature data sequence of the crystal ingot also includes: collecting temperature data of the surface of the crystal ingot by a temperature sensor, and preprocessing the temperature data to obtain the temperature data sequence of the crystal ingot.

[0014] The present invention takes into account that the overall quality of the originally collected temperature data is poor, which is not conducive to subsequent data processing, and therefore preprocessing is performed to improve the overall quality of the data.

[0015] According to an intelligent early warning method for crystal rod quality provided by the present invention, the necessity of filtering is corrected by the deviation between the temperature data sequence and the corresponding flow data sequence to obtain the filtering possibility of the temperature data point, including: obtaining the surrounding data segments of the flow data point in the flow data sequence, the flow data point and the temperature data point corresponding one by one; determining the difference between the temperature data point and the corresponding flow data point, the difference is used to characterize the trend difference between the temperature data point and the flow data point in their respective surrounding data segments; obtaining the temperature data point to be filtered by comparing the filtering possibility of the temperature data point with a preset threshold, the filtering possibility is positively correlated with the filtering necessity and the difference.

[0016] The present invention also takes into account that it is impossible to accurately distinguish abnormal temperature data from noise data only through the data characteristics of the temperature data itself. Therefore, the necessity of filtering the temperature data is corrected by further acquiring flow data that is correlated with the temperature data. While accurately denoising, the data characteristics of the normal temperature data can be retained, further improving the accuracy of the crystal rod quality warning.

[0017] According to an intelligent early warning method for ingot quality provided by the present invention, the difference between the temperature data point and the corresponding flow data point satisfies the relationship:

[0018] ;

[0019] In the formula, Indicates The temperature data point is The difference between the traffic data points, Indicates The data segment around the temperature data point is greater than The number of data points for each temperature data point, Indicates The data segment around the flow data point is greater than The number of data points for each flow data point, represents a hyperparameter.

[0020] According to an intelligent early warning method for crystal ingot quality provided by the present invention, the filtering possibility of the temperature data point satisfies the relationship:

[0021] ;

[0022] In the formula, Indicates The filtering possibility of temperature data points, Indicates Necessity of filtering temperature data points, Indicates The temperature data point is The difference between the traffic data points, Indicates The average sampling time interval between adjacent extreme points in the data segment around the temperature data point, Indicates The average sampling time interval between adjacent extreme value points in the data segment around the flow data point, represents the hyperparameter, represents the linear normalization function, Represents the absolute value symbol.

[0023] The present invention provides a method for calculating the filtering possibility of an accurate temperature data point. By analyzing the deviation between the surrounding data segments of the temperature data point and the surrounding data segments of the corresponding flow data point, the filtering possibility of the temperature data point can be accurately obtained to prepare for subsequent quality warnings.

[0024] According to an intelligent early warning method for crystal rod quality provided by the present invention, the temperature data point to be filtered is obtained by comparing the filtering possibility of the temperature data point with a preset threshold, including: a preset threshold; if the filtering possibility of the temperature data point is greater than the preset threshold, the temperature data point is a noise data point to be filtered; if the filtering possibility of the temperature data point is not greater than the preset threshold, the temperature data point is a normal data point.

[0025] According to an intelligent early warning method for ingot quality provided by the present invention, filtering and denoising the temperature data points to be filtered in the temperature data sequence includes: constructing a reference data segment of a preset length of the temperature data point to be filtered with the temperature data point to be filtered as the center; obtaining L adjacent data segments of the reference data segment, where L represents the number of preset adjacent data segments; and calculating the filtering weight of each adjacent data segment: ;

[0026] In the formula, Indicates The filtering weights of adjacent data segments, Indicates the reference data segment and The sampling interval of the center temperature data points of adjacent data segments is Represents the mean and the The difference between the means of adjacent data segments, Indicates the number of adjacent data segments. represents an exponential function with e as the base; weighted averaging is performed on each adjacent data segment according to the filtering weights of each adjacent data segment to achieve filtering processing of the temperature data point to be filtered.

[0027] According to an intelligent early warning method for crystal ingot quality provided by the present invention, the target temperature data sequence obtained after filtering and denoising by ARIMA model is used to realize early warning of crystal ingot quality, including: obtaining the absolute value of the difference between the predicted value and the actual value of each temperature data point by ARIMA model, if the absolute value of the difference is greater than a preset abnormal threshold, the temperature data point is an abnormal data point; if the absolute value of the difference is not greater than the preset abnormal threshold, the temperature data point is a normal data point; if the proportion of the number of abnormal data points in the historical data segment of the temperature data point is greater than the preset early warning threshold, a quality abnormality early warning is issued externally.

[0028] The present invention takes into account that when the quality of the crystal rod is abnormal, there may be safety hazards and may cause losses to the equipment that subsequently uses the crystal rod. Therefore, when the quality of the crystal rod is found to be abnormal, an early warning prompt is issued in time to reduce the risk.

[0029] In a second aspect, the present invention provides an intelligent early warning system for crystal ingot quality, which adopts the following technical solution:

[0030] An intelligent early warning system for crystal ingot quality includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned intelligent early warning method for crystal ingot quality is implemented.

[0031] By adopting the above technical solution, the above-mentioned intelligent early warning method for crystal rod quality is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0032] The present invention has the following technical effects:

[0033] Based on the above technical solution, when realizing the quality warning of the crystal rod, the present invention obtains the abnormal temperature data on the surface of the crystal rod by analyzing the temperature data characteristics on the surface of the crystal rod, thereby obtaining the quality monitoring result of the crystal rod. In this process, the present invention takes into account that the noise data will interfere with the accuracy of the warning, so the temperature data point is filtered by obtaining the data performance of the temperature data point in the surrounding data segment to obtain the necessity of filtering the temperature data point, thereby accurately realizing the quality warning of the crystal rod. On this basis, the present invention also takes into account that it is impossible to accurately distinguish abnormal temperature data from noise data only by the data characteristics of the temperature data itself, so the necessity of filtering the temperature data is corrected by further obtaining the flow data that is correlated with the temperature data, and the data characteristics of the normal temperature data can be retained while accurately removing noise, further improving the accuracy of the quality warning of the crystal rod. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0035] Figure 1 A schematic flow chart of an intelligent early warning method for crystal ingot quality provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0037] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0038] It should be noted that changes in the surface temperature of the crystal rod may cause the crystal rod to become brittle, oxidized or undergo structural changes. Therefore, temperature is a more intuitive parameter when monitoring the quality of the crystal rod. However, when collecting temperature data for monitoring, due to the influence of factors such as the collection environment, the collected temperature data may be interfered by noise data, thereby affecting the accuracy of the crystal rod quality monitoring.

[0039] Based on this, an embodiment of the present invention discloses an intelligent early warning method for crystal rod quality. The temperature data points that need to be filtered and denoised are obtained by analyzing the overall changes in temperature data. The real original data is retained while accurately filtering and denoising. Accurate crystal rod quality abnormality monitoring results are obtained through the denoised temperature data and early warning is issued.

[0040] See also Figure 1 As shown, Figure 1 A schematic flow chart of an intelligent early warning method for crystal ingot quality provided by an embodiment of the present invention, the method specifically comprises the following steps S1 to S4.

[0041] S1: Obtaining a preset surrounding data segment of each temperature data point in the temperature data sequence of the crystal rod.

[0042] The number of temperature data points included in the surrounding data segments of the temperature data points can be preset to 21, which can be specifically set according to actual needs.

[0043] It should be noted that in order to accurately obtain the temperature data point that needs to be filtered, the data performance of the temperature data point can be analyzed in the surrounding data segments of the temperature data point, so as to obtain the necessity of filtering and realize accurate filtering.

[0044] For example, in an embodiment of the present invention, obtaining a preset surrounding data segment of each temperature data point in a temperature data sequence of a crystal rod also includes: collecting temperature data of the surface of the crystal rod by a temperature sensor, and obtaining the temperature data sequence of the crystal rod after preprocessing the temperature data.

[0045] Specifically, a corresponding acquisition frequency is set for the temperature sensor, and the temperature sensor regularly acquires temperature data on the crystal surface, and the temperature value acquired at each acquisition time is taken as a temperature data point.

[0046] The acquisition frequency may be once every 3 seconds, which may be set according to actual needs. The preprocessing method may be to convert the temperature data into a format, etc., which may be set according to actual needs.

[0047] For example, when obtaining the surrounding data segments of the temperature data point, there are two possible implementation methods:

[0048] In a possible implementation, a number of temperature data points may be equally acquired on both sides of the current temperature data point with the current temperature data point as the center, and the current temperature data point and the number of temperature data points together constitute a surrounding data segment of the current temperature data point.

[0049] In another possible implementation, the current temperature data point may be used as an end point, and several temperature data points may be obtained from the historical data of the current temperature data point. The current temperature data point and the several temperature data points together constitute a surrounding data segment of the current temperature data point.

[0050] After obtaining the surrounding data segments of the temperature data point based on any of the above possible implementations, the necessity of filtering the temperature data point can be obtained according to the noise performance of the temperature data point in the surrounding data segments, that is, the following steps are performed.

[0051] S2: Calculate the filtering necessity of the temperature data point, correct the filtering necessity by the deviation between the temperature data sequence and the corresponding flow data sequence, and obtain the filtering possibility of the temperature data point.

[0052] It should be noted that compared with the normally changing temperature data, the data value of the noise will be more prominent, and the data change between the noise data and the surrounding normal temperature data will be more drastic, while the normally changing temperature data is relatively gentle. Therefore, the more prominent the data value of the temperature data point is and the more drastic the data change between it and the surrounding data segments, the greater the possibility that the temperature data point is noise data, and the greater the necessity of corresponding filtering.

[0053] For example, in an embodiment of the present invention, the necessity of filtering the temperature data point is determined by the noise performance of the temperature data point in the surrounding data segment, which can be specifically referred to in the following relationship:

[0054] ;

[0055] In the formula, Indicates Necessity of filtering temperature data points, Indicates The temperature value of the temperature data point, Indicates The mean temperature in the data segment around the temperature data point, Indicates The number of modes in the data segment around the temperature data point, Indicates The mean of the temperature differences of all adjacent temperature data points in the data segment around the temperature data point, Represents an exponential function with base e.

[0056] In the above formula, Indicates The relative size of the data value of the temperature data point in the surrounding data segment. The larger the value, the The greater the difference between a temperature data point and the average level of temperature data points in the surrounding data segments, the greater the possibility that the temperature data point belongs to noise.

[0057] Indicates The number of modes in the data segments surrounding a temperature data point. The mode can represent the data value with the highest frequency in the surrounding data segments. The smaller the total number of modes, the more drastic the overall data changes in the surrounding data segments, and the greater the necessity of filtering the corresponding temperature data points.

[0058] Indicates The smaller the ratio is, the more stable the temperature change in the data segment is, and the mode is dominant in the surrounding data segments. The larger the value is, the more consistent the data feature changes in the surrounding data segments are with the characteristics of the noise data changes, and the greater the possibility that the temperature data point is noise data.

[0059] After obtaining the necessity of filtering for each temperature data point in the crystal rod temperature data sequence based on the above formula, the necessity of filtering is corrected by the deviation between the temperature data sequence and the corresponding flow data sequence, and the filtering possibility of the temperature data point can be obtained. The filtering possibility of the temperature data point is compared with the threshold to obtain the temperature data point that needs to be filtered, that is, execute the following steps.

[0060] S3: Obtain the temperature data point to be filtered by comparing the filtering possibility of the temperature data point with a preset threshold.

[0061] It should be noted that based on the above steps, the filtering possibility of the temperature data point can be obtained, that is, the possibility that the temperature data point is noise data. By comparing the filtering possibility of the temperature data point with the preset threshold, the temperature data point with a higher degree of noise can be obtained. Such data points are the data points that need to be filtered. Filtering the temperature data points with a higher degree of noise can effectively reduce the impact of noise data on the quality warning of the crystal rod.

[0062] By way of example, in an embodiment of the present invention, a temperature data point to be filtered is obtained by comparing the filtering possibility of the temperature data point with a preset threshold, including: a preset threshold; if the filtering possibility of the temperature data point is greater than the preset threshold, the temperature data point is a noise data point to be filtered; if the filtering possibility of the temperature data point is not greater than the preset threshold, the temperature data point is a normal data point.

[0063] The preset threshold may be set to 0.55, and may be adjusted according to actual needs, and the embodiment of the present invention does not impose too many limitations on this.

[0064] After obtaining the noise data to be filtered based on the above steps, a filtering algorithm can be used to filter and denoise the temperature data sequence, that is, perform the following steps.

[0065] S4: Filter and denoise the temperature data points to be filtered in the temperature data sequence, and process the target temperature data sequence obtained after filtering and denoising through the ARIMA model to achieve early warning of crystal ingot quality.

[0066] It should be noted that the embodiment of the present invention can obtain temperature data points that are subject to greater noise interference through the above steps, and then combine conventional filtering algorithms to process high noise points, so that the filtering algorithm can allocate resources more effectively and significantly reduce errors caused by noise, thereby effectively improving the overall accuracy of the temperature data.

[0067] By way of example, in an embodiment of the present invention, the temperature data points to be filtered in the temperature data sequence are filtered and denoised, including: constructing a reference data segment of a preset length of the temperature data point to be filtered with the temperature data point to be filtered as the center; obtaining L adjacent data segments of the reference data segment, where L represents the number of preset adjacent data segments; calculating the filtering weights of each adjacent data segment, and performing weighted averaging on each adjacent data segment according to the filtering weights of each adjacent data segment to achieve filtering processing of the temperature data point to be filtered.

[0068] For example, in the embodiment of the present invention, the filtering weights of adjacent data segments are determined, and the specific details can be referred to the following relationship:

[0069] ;

[0070] In the formula, Indicates The filtering weights of adjacent data segments, Indicates the reference data segment and The sampling interval of the center temperature data points of adjacent data segments is Represents the mean and the The difference between the means of adjacent data segments, Indicates the number of adjacent data segments. Represents an exponential function with base e.

[0071] In the above formula, Indicates The weight factor when filtering adjacent data segments, Represents the sum of weight factors of all adjacent data segments.

[0072] It should be noted that by processing the actual values ​​of the temperature data points in the denoised temperature data sequence with the Autoregressive Integrated Moving Average Model (ARIMA model), the predicted values ​​of each temperature data point can be obtained. Combined with the deviation between the actual value and the predicted value, the abnormal data in the temperature data sequence can be accurately obtained.

[0073] By way of example, in an embodiment of the present invention, the target temperature data sequence obtained after filtering and denoising is processed by an ARIMA model to achieve early warning of crystal rod quality, including: obtaining the absolute value of the difference between the predicted value and the actual value of each temperature data point in the target temperature data sequence through the ARIMA model, if the absolute value of the difference is greater than a preset abnormal threshold, the temperature data point is an abnormal data point; if the absolute value of the difference is not greater than the preset abnormal threshold, the temperature data point is a normal data point; if the proportion of the number of abnormal data points in the historical data segment of the temperature data point is greater than the preset warning threshold, a quality abnormality warning is issued.

[0074] The preset abnormal threshold can be set to 5°C, which can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this. The preset warning threshold can be set to 10%, which can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0075] In this way, the abnormal state of the crystal rod can be accurately identified and an early warning can be issued by obtaining the abnormal temperature of the crystal rod surface.

[0076] The warning methods include sound warning and signal light warning, etc., which can be set according to actual needs.

[0077] It can be seen that in the embodiment of the present invention, when realizing accurate early warning of crystal rod quality, the preset surrounding data segment of each temperature data point in the temperature data sequence of the crystal rod can be obtained; the filtering necessity and filtering possibility of the temperature data point are calculated, and the temperature data point to be filtered is obtained by comparing the filtering possibility of the temperature data point with the preset threshold value; the temperature data point to be filtered in the temperature data sequence is filtered and denoised, and the target temperature data sequence obtained after filtering and denoising is processed by the ARIMA model to realize early warning of crystal rod quality.

[0078] In this way, the embodiment of the present invention monitors and warns the quality of the crystal rod by analyzing the temperature data on the surface of the crystal rod. In this process, the embodiment of the present invention takes into account that noise data will interfere with the accuracy of the warning. Therefore, the temperature data points are filtered by obtaining the data performance of the temperature data points in the surrounding data segments to obtain the necessity of filtering the temperature data points, thereby accurately realizing the quality warning of the crystal rod.

[0079] Based on the above Figure 1 The embodiment shown is further explained through the following steps to facilitate understanding of how to correct the necessity of filtering through the deviation between the temperature data sequence and the corresponding flow data sequence in the above step S3 to obtain the filtering possibility of the temperature data point.

[0080] It should be noted that the necessity of filtering the temperature data points obtained based on the above steps can only reflect whether the temperature data points are noise data to a certain extent, but cannot distinguish between noise data and abnormal temperature data on the surface of the crystal rod.

[0081] Based on this, the embodiment of the present invention can further distinguish between noise data and abnormal temperature data.

[0082] It should be further explained that during the chemical vapor deposition process of the crystal rod, the increase in gas flow rate may increase the reaction rate, thereby generating more heat on the surface of the crystal rod. Therefore, at the same sampling time, there is a correlation between the temperature of the crystal rod surface and the flow data flowing into the crystal rod, and the noise data points are not actually generated by the crystal rod itself, so the noise data cannot satisfy the correlation between temperature and gas flow. Based on this, in order to further distinguish between noise data and abnormal temperature data, the correlation between the temperature data and flow data of the crystal rod can be obtained, and the necessity of filtering the temperature data points can be corrected through the correlation, so as to obtain noise data more accurately.

[0083] By way of example, in an embodiment of the present invention, the necessity of filtering is corrected by the deviation between the temperature data sequence and the corresponding flow data sequence to obtain the filtering possibility of the temperature data point, including: obtaining the surrounding data segments of the flow data point in the flow data sequence, the flow data points and the temperature data points corresponding one by one; determining the difference between the temperature data point and the corresponding flow data point, the difference is used to characterize the trend difference between the temperature data point and the flow data point in their respective surrounding data segments; determining the comparison result of the filtering possibility of the temperature data point with a preset threshold, the filtering possibility is positively correlated with the difference.

[0084] For example, in an embodiment of the present invention, obtaining the surrounding data segments of the flow data points in the flow data sequence of the crystal rod also includes: collecting flow data through a flow sensor, and preprocessing the flow data to obtain the flow data sequence of the crystal rod.

[0085] The acquisition frequency may be once every 3 seconds, which may be set according to actual needs. The preprocessing method may be to convert the temperature data into a format, etc., which may be set according to actual needs.

[0086] After the flow data sequence of the crystal rod is obtained by the above-mentioned processing method, the surrounding data segments of each flow data point can be obtained.

[0087] It should be noted that when the gas flow changes, the temperature does not change rapidly, and there is a time difference between the flow data and the corresponding temperature data. Therefore, when determining the surrounding data segments of the flow data points, the embodiment of the present invention iteratively adjusts the flow data points included in the flow data segments, obtains the Pearson correlation coefficient between the temperature data segments and the flow data segments, and obtains the flow data segments corresponding to the maximum Pearson correlation coefficient, which is the surrounding data segments of the flow data points, to eliminate the time error.

[0088] Specifically, when obtaining the surrounding data segments of the flow data points, the initial surrounding data segments of the flow data points corresponding to the temperature data points can be obtained in the flow data sequence based on the method of obtaining the surrounding data segments of the temperature data points, and the Pearson correlation coefficient between the initial surrounding data segments and the surrounding data segments of the current temperature data points can be obtained. The initial surrounding data segments are then moved forward as a whole by the length of a flow data point to obtain the Pearson correlation coefficient between the new initial surrounding data segments and the surrounding data segments of the current temperature data points, and so on, until the Pearson correlation coefficients between Q initial surrounding data segments and the surrounding data segments of the current temperature data points are obtained, where Q is the preset number of initial surrounding data segments; the initial surrounding data segment corresponding to the maximum Pearson correlation coefficient is used as the surrounding data segment of the flow data point corresponding to the current temperature data point.

[0089] The value of Q may be 100, and may be set according to actual needs. The embodiment of the present invention does not impose too many restrictions on this.

[0090] After obtaining the surrounding data segments of the flow data points corresponding to the temperature data points based on the above steps, the difference between the temperature data points and the flow data points can be obtained based on the deviation between the surrounding data segments of the temperature data points and the flow data points.

[0091] For example, in an embodiment of the present invention, the difference between a temperature data point and a corresponding flow data point is determined by referring to the following relationship:

[0092] ;

[0093] In the formula, Indicates The temperature data point is The difference between the traffic data points, Indicates The temperature value of the data segment around the temperature data point is greater than The number of data points for each temperature data point, Indicates The flow value in the data segment around the flow data point is greater than The number of data points for each flow data point, represents a hyperparameter.

[0094] Among them, the hyperparameter can be 0.01, which can be set according to actual needs.

[0095] In the above formula, Indicates The temperature value of the data segment around the temperature data point is greater than The number of data points for the temperature data point is the same as that for the The flow value in the data segment around the flow data point is greater than The difference in the number of data points of the flow data points, that is, The temperature data points and The difference in the relative size of the traffic data points in their respective data segments. The larger the value, the The greater the gap between the surrounding data segment of a temperature data point and the flow data.

[0096] After obtaining the difference between the temperature data point and the corresponding flow data point based on the above formula, the filtering possibility of the temperature data point can be obtained based on the difference between the two.

[0097] For example, in the embodiment of the present invention, the filtering possibility of the temperature data point is determined, and the specific relationship can be referred to as follows:

[0098] ;

[0099] In the formula, Indicates The filtering possibility of temperature data points, Indicates Necessity of filtering temperature data points, Indicates The temperature data point is The difference between the traffic data points, Indicates The average sampling time interval between adjacent extreme points in the data segment around the temperature data point, Indicates The average sampling time interval between adjacent extreme value points in the data segment around the flow data point, represents the hyperparameter, represents the linear normalization function, Represents the absolute value symbol.

[0100] In the above formula, It represents the absolute difference of the average sampling time interval between the adjacent extreme value points of the surrounding data segments of the temperature data point and the flow data point. The larger the value, the more inconsistent the overall trend of the surrounding data segments of the temperature data point and the flow data point, and the worse the correlation between the two. The greater the possibility that a temperature data point is noise data, the more it needs to be filtered.

[0101] After the filtering possibility of each temperature data point is obtained based on the above steps, the noise data can be obtained by comparing the filtering possibility of the temperature data point with a preset threshold.

[0102] It can be seen that in the embodiment of the present invention, the embodiment of the present invention also takes into account that it is impossible to accurately distinguish abnormal temperature data from noise data only through the data characteristics of the temperature data itself. Therefore, the necessity of filtering the temperature data is corrected by further acquiring flow data that is correlated with the temperature data. While accurately denoising, the data characteristics of the normal temperature data can be retained, further improving the accuracy of the crystal rod quality warning.

[0103] An embodiment of the present invention further discloses an intelligent early warning system for crystal rod quality, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent early warning method for crystal rod quality according to the present invention is implemented.

[0104] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0105] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM, a dynamic random access memory DRAM, a static random access memory SRAM, an enhanced dynamic random access memory EDRAM, a high bandwidth memory HBM, a hybrid memory cube HMC, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0106] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0107] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent early warning method for crystal ingot quality, characterized in that: include: Acquire a preset surrounding data segment of each temperature data point in the temperature data sequence of the crystal rod; Calculate the filtering necessity for the temperature data point: ； In the formula, Indicates Necessity of filtering temperature data points, Indicates The temperature value of the temperature data point, Indicates The mean temperature in the data segment around the temperature data point, Indicates The number of modes in the data segment around the temperature data point, Indicates The mean of the temperature differences of all adjacent temperature data points in the data segment around the temperature data point, represents an exponential function with base e; Correcting the necessity of filtering by the deviation between the temperature data sequence and the corresponding flow data sequence to obtain the filtering possibility of the temperature data point includes: Acquire surrounding data segments of flow data points in a flow data sequence, wherein the flow data points correspond to temperature data points one by one; determine the difference between the temperature data points and the corresponding flow data points, wherein the difference is used to characterize the trend difference between the temperature data points and the flow data points in their respective surrounding data segments; obtain the temperature data points to be filtered by comparing the filtering possibility of the temperature data points with a preset threshold, wherein the filtering possibility is positively correlated with both the filtering necessity and the difference; The temperature data point to be filtered is obtained by comparing the filtering possibility of the temperature data point with a preset threshold; the temperature data point to be filtered in the temperature data sequence is filtered and denoised, and the target temperature data sequence obtained after filtering and denoising is processed by the ARIMA model to realize early warning of crystal rod quality.

2. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The step of obtaining the preset surrounding data segments of each temperature data point in the temperature data sequence of the crystal rod further includes: The temperature data of the surface of the crystal rod is collected by a temperature sensor, and the temperature data sequence of the crystal rod is obtained after preprocessing the temperature data.

3. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The difference between the temperature data point and the corresponding flow data point satisfies the relationship: ； In the formula, Indicates The temperature data point is The difference between the traffic data points, Indicates The data segment around the temperature data point is greater than The number of data points for each temperature data point, Indicates The data segment around the flow data point is greater than The number of data points for each flow data point, represents a hyperparameter.

4. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The filtering possibility of the temperature data point satisfies the relationship: ； In the formula, Indicates The filtering possibility of temperature data points, Indicates Necessity of filtering temperature data points, Indicates The temperature data point is The difference between the traffic data points, Indicates The average sampling time interval between adjacent extreme points in the data segment around the temperature data point, Indicates The average sampling time interval between adjacent extreme value points in the data segment around the flow data point, represents the hyperparameter, represents the linear normalization function, Represents the absolute value symbol.

5. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The step of obtaining the temperature data point to be filtered by comparing the filtering possibility of the temperature data point with a preset threshold value includes: Preset threshold value; If the filtering possibility of the temperature data point is greater than a preset threshold, the temperature data point is a noise data point to be filtered; If the filtering possibility of the temperature data point is not greater than a preset threshold, the temperature data point is a normal data point.

6. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The filtering and denoising of the temperature data points to be filtered in the temperature data sequence comprises: Taking the temperature data point to be filtered as the center, construct a reference data segment of a preset length of the temperature data point to be filtered; obtain L adjacent data segments of the reference data segment, where L represents the number of preset adjacent data segments; calculate the filtering weight of each adjacent data segment: ; In the formula, Indicates The filtering weights of adjacent data segments, Indicates the reference data segment and The sampling interval of the center temperature data points of adjacent data segments is Represents the mean and the The difference between the means of adjacent data segments, Indicates the number of adjacent data segments. represents an exponential function with e as the base; weighted averaging is performed on each adjacent data segment according to the filtering weights of each adjacent data segment to achieve filtering processing of the temperature data point to be filtered.

7. The intelligent early warning method for crystal ingot quality according to claim 1, characterized in that: The target temperature data sequence obtained after filtering and denoising by the ARIMA model is processed to achieve early warning of the quality of the crystal ingot, including: The absolute value of the difference between the predicted value and the actual value of each temperature data point in the target temperature data sequence is obtained through the ARIMA model. If the absolute value of the difference is greater than the preset abnormal threshold, the temperature data point is an abnormal data point; if the absolute value of the difference is not greater than the preset abnormal threshold, the temperature data point is a normal data point; If the proportion of the number of abnormal data points in the historical data segment of the temperature data point is greater than the preset warning threshold, a quality abnormality warning is issued.

8. An intelligent early warning system for crystal ingot quality, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent early warning method for crystal rod quality according to any one of claims 1 to 7 is implemented.

Images