A method and system for intelligently detecting energy consumption during crystal growth

Abstract

The present invention relates to the field of data processing technology, and in particular to an intelligent detection method and system for energy consumption in a crystal growth process. The method comprises the steps of collecting heating power data and heat flow data, dividing the heating power data into multiple groups of heating power data, and constructing multiple box plots according to the multiple groups of heating power data, analyzing the heating power data in each box plot, and obtaining the noise level of the heating power data in each box plot; obtaining the expansion factor of each box plot according to the noise level of the heating power data in each box plot, correcting the maximum expansion length of the box plot according to the expansion factor of each box plot, obtaining the expansion length of each box plot, obtaining the updated box plot of each box plot according to the expansion length of each box plot, and performing abnormal judgment on each heating power data according to the updated box plot of each box plot. The present invention improves the accuracy of abnormal detection and avoids the influence of noise data.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent detection method and system for energy consumption in a crystal growth process. Background Art

[0002] Crystal growth plays a key role in the fields of semiconductors, optoelectronic materials and drug manufacturing. However, this process is usually accompanied by high consumption and waste of resources. In the context of major challenges such as the global energy crisis, climate change and environmental pollution, improving industrial energy efficiency and sustainable development have become important goals for all companies. The intelligent energy consumption detection method implemented in this solution can monitor the heating power of the heating system during the crystal growth process in real time, identify and optimize the links with high energy consumption, thereby effectively reducing energy consumption and carbon emissions. This is of great significance for mitigating climate change and protecting the environment.

[0003] The patent application document with the publication number CN115375348A currently discloses a data outlier detection method based on box plot indicators, including the following steps S1: establishing a box plot indicator calculation and data outlier detection system; S2: collecting orders generated by users in real time through a data collection module, and sending the order data to a downstream box plot indicator calculation module, etc.

[0004] The box plot algorithm in the above method detects data anomalies based on the size of the data within the box to which it belongs compared to the upper and lower limits of the anomaly. However, when collecting the heating power data of the heating system during the crystal growth process, the heating power data may contain noise data due to electromagnetic interference from other equipment around the sensor. This noise data will have a relatively close value to the abnormal heating power data generated by the heating system itself. Therefore, when using the box plot algorithm to detect abnormal data in the heating power data, the noise data may be detected as abnormal data, resulting in errors in the abnormal detection results. Summary of the invention

[0005] In order to solve the technical problem that when a box plot algorithm detects abnormal data in heating power data, noise data may be detected as abnormal data, resulting in errors in abnormal detection results, the present invention provides an intelligent detection method and system for energy consumption in a crystal growth process.

[0006] In a first aspect, the present invention provides a method for intelligently detecting energy consumption during a crystal growth process, which adopts the following technical solution:

[0007] A method for intelligently detecting energy consumption during a crystal growth process, comprising the steps of:

[0008] Collect heating power data and heat flow data;

[0009] Divide the heating power data into multiple groups of heating power data; construct multiple box plots based on the multiple groups of heating power data; obtain the noise level of the heating power data in each box plot; obtain the expansion factor of each box plot:

[0010] , Represents the expansion factor of the xth box plot; Represents the noise level of the heating power data in the xth box plot; Represents the noise level of the heating power data in the previous boxplot of the xth boxplot; Represents the number of heating power data in the xth box plot; Represents the value of the bth heating power data in the xth box plot; Represents the value of the b-th heating power data in the previous box plot of the x-th box plot; Represents the preset second hyperparameter;

[0011] The extended length of each box plot is obtained, where the extended length is positively correlated with the expansion factor of each box plot; and an updated box plot of each box plot is obtained according to the extended length of each box plot.

[0012] The innovation of the present invention lies in that the expansion factor of each box plot is obtained by analyzing the noise level of the heating power data in each box plot, and then the expansion length of each box plot is adaptively adjusted according to the expansion factor of each box plot. In this way, it is possible to more accurately judge whether the heating power data point is abnormal. This method reduces the interference of noise data on abnormal judgment, thereby improving the accuracy of detection. Furthermore, the expansion factor of each box plot is obtained by evaluating the performance of the noise level of the heating power data in each box plot and the previous box plot of each box plot, so that the result of the expansion factor is more accurate.

[0013] Preferably, dividing the heating power data into multiple groups of heating power data includes:

[0014] The number of data N is preset, and according to the number of data N, the heating power data in the heating power data sequence is divided into a plurality of groups of heating power data in sequence.

[0015] Preferably, constructing a plurality of box plots according to a plurality of sets of heating power data comprises:

[0016] A box plot algorithm is used to construct a box plot for each set of heating power data, and the number of abnormal data in each box plot is obtained.

[0017] The more abnormal data there are, the greater the noise level of the data in the box plot is, which facilitates the subsequent acquisition of the noise level of the heating power data in each box plot.

[0018] Preferably, obtaining the noise level of the heating power data in each box plot includes:

[0019] Get the reference boxplot for each boxplot;

[0020] ;

[0021] In the formula, Represents the noise level of the heating power data in the xth box plot; Represents the number of abnormal data in the xth box plot; Represents the preset hyperparameters; The number of reference box plots representing the xth box plot; represents the Pearson correlation coefficient between the heating power data in the rth reference boxplot of the xth boxplot and the heat flux data corresponding to its time series; Represents the Pearson correlation coefficient between the heating power data in the x-th box plot and the heat flow data corresponding to its time series; norm() represents the normalization function.

[0022] It is convenient to adaptively obtain the expansion factor of each box plot according to the greater the noise level of the heating power data in each box plot.

[0023] Preferably, obtaining the extended length of each box plot includes:

[0024] ;

[0025] In the formula, Represents the extended length of the xth box plot; Represents the expansion factor of the xth box plot; Represents the maximum extension length of the box plot; Represents the floor symbol.

[0026] This facilitates the subsequent expansion of the data in each box plot and eliminates the interference of noise data on the anomaly detection results as much as possible.

[0027] Preferably, obtaining an updated box plot of each box plot according to the extended length of each box plot comprises:

[0028] Get before the sampling time of the xth box plot The heating power data is recorded as the extended data of the x-th box plot. According to the heating power data in the x-th box plot and the extended data of the x-th box plot, an updated box plot of the x-th box plot is constructed, where: Represents the extended length of the x-th box plot.

[0029] Expand the box plots with more noise data to eliminate the interference of noise data on anomaly detection results as much as possible.

[0030] Preferably, obtaining a reference box plot for each box plot comprises:

[0031] The number of box plots M is preset, and the M box plots before the time series and the M box plots after the time series of each box plot are recorded as reference box plots of each box plot.

[0032] In a second aspect, the present invention provides an intelligent detection system for energy consumption during crystal growth, which adopts the following technical solution:

[0033] A crystal growth process energy consumption intelligent detection system comprises: 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 crystal growth process energy consumption intelligent detection method is implemented.

[0034] By adopting the above technical solution, the above-mentioned intelligent detection method of energy consumption in the crystal growth process 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.

[0035] The present invention has the following technical effects: the purpose of the present invention is to obtain the expansion factor of each box plot by analyzing the noise level of the heating power data in each box plot, and then adapt the expansion length of each box plot according to the expansion factor of each box plot, so that the box plots containing more noise data are expanded, and the interference of the noise data on the abnormal detection results is eliminated as much as possible, so that it is possible to more accurately judge whether the heating power data point is abnormal, thereby improving the accuracy of detection. Furthermore, the expansion factor of each box plot is obtained by evaluating the performance of the noise level of the heating power data in each box plot and the previous box plot of each box plot, so that the result of the expansion factor is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] 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.

[0037] Figure 1 It is a method flow chart of an intelligent detection method for energy consumption in a crystal growth process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] 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.

[0039] 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.

[0040] The embodiment of the present invention discloses a method for intelligently detecting energy consumption during crystal growth. Figure 1 , comprising steps S1 to S3:

[0041] S1: Collect heating power data and heat flow data.

[0042] In an embodiment of the present invention, the preset sampling time is 5 seconds / time, and the collection time period is the entire growth process of the crystal. The power data collector and the heat flow meter are used to collect the heating power data and the heat flow data of the heating system during the crystal growth process to obtain the heating power data sequence and the heat flow data sequence. It should be noted that the heating power data and the heat flow data have time correspondence.

[0043] S2: Divide the heating power data into multiple groups of heating power data, construct multiple box plots according to the multiple groups of heating power data, analyze the heating power data in each box plot, and obtain the noise level of the heating power data in each box plot.

[0044] It should be noted that the box plot algorithm is based on the size of the data in the box plot to which it belongs compared to the upper and lower limits of the abnormality. However, when collecting the heating power data of the heating system during the crystal growth process, the electromagnetic interference of other equipment around the sensor may cause the heating power data to contain noise data. This noise data is closer to the value of the abnormal heating power data generated by the heating system itself. Therefore, when using the box plot algorithm to detect abnormal data in the heating power data, the noise data may be detected as abnormal data, resulting in errors in the abnormal detection results. Therefore, the present invention first divides the data in the heating power data sequence into multiple groups of heating power data and constructs multiple box plots, analyzes the data in each box plot, expands the box plots containing more noise data, and then smoothes out the interference of the noise data on the abnormal detection results as much as possible.

[0045] It should be noted that the more abnormal data there are in any box plot, the greater the noise level of the heating power data in the box plot. It is known that in a heating system, the heating power (input heat) is usually balanced with the heat flow (output heat) to maintain the required temperature. Therefore, if the heat flow increases, it is usually necessary to increase the heating power to keep the temperature stable. This balance relationship directly leads to a certain correlation between the heating power data and the heat flow data. Therefore, the weaker the connection between the heating power data in the box plot and the heat flow data corresponding to its time series, the greater the noise level of the heating power data in the box plot.

[0046] In the embodiment of the present invention, the number of data N is preset, and according to the number of data N, the heating power data in the heating power data sequence is divided into multiple groups of heating power data in turn, and a box plot algorithm is used to construct a box plot for each group of heating power data, and the number of abnormal data in each box plot is obtained. It should be noted that if the data in the last heating power data sequence does not meet N, it is padded with 0, and constructing a box plot is a well-known technology, and in the embodiment of the present invention, it will not be described in detail.

[0047] The number of box plots M is preset, and the M box plots before the time series and the M box plots after the time series of each box plot are recorded as reference box plots of each box plot. In an embodiment of the present invention, the number of box plots M is preset to 5. In other embodiments, the implementer may preset the value of the number of box plots M according to the specific implementation situation.

[0048] Get the degree of noise in the heating power data in each boxplot:

[0049] ;

[0050] In the formula, Represents the noise level of the heating power data in the xth box plot; Represents the number of abnormal data in the xth box plot; represents a preset hyperparameter. In the embodiment of the present invention, the preset , in order to prevent the number of abnormal data in the xth box plot from being 0; The number of reference box plots representing the xth box plot; represents the Pearson correlation coefficient between the heating power data in the rth reference boxplot of the xth boxplot and the heat flux data corresponding to its time series; represents the Pearson correlation coefficient between the heating power data in the xth box plot and the heat flow data corresponding to its time series; norm() represents the normalization function;

[0051] The larger it is, the greater the noise level of the heating power data in the xth box plot; The larger the value is, the weaker the connection between the heating power data in the x-th box plot and the heat flow data corresponding to the time series is. Then, the greater the possibility of noise data in the heating power data in the box plot, the more potential noise data points there are. It can also be further explained that the more abnormal data points in the x-th box plot, the greater the credibility, and the greater the noise level in the x-th box plot.

[0052] S3: Obtain an expansion factor of each box plot according to the noise level of the heating power data in each box plot.

[0053] It should be noted that the above steps obtain the noise level of the heating power data in the box plot, but this indicator is obtained based on the changing characteristics of the heating power data in the box plot, and does not take into account the noise performance of the thermal power data that needs to be included when the box plot is subsequently expanded; if the noise performance of the thermal power data that needs to be included is greater, more thermal power data needs to be included in the box plot to smooth the impact of noise data on anomaly detection, but it is not accurate enough to determine the expansion factor of the box plot only by the noise level of the heating power data in the box plot and the noise level of the heating power data in the previous box plot of the box plot. It is also necessary to consider the sum of the differences between the box plot and all the one-to-one corresponding heating power data in the previous box plot. The larger the value, the greater the trend difference between the previous box plot of the box plot and the heating power data in the box plot, and the worse the noise smoothing effect and utilization value of the heating power data in the previous box plot of the box plot, then the box plot should continue to include more heating power data to smooth the noise impact.

[0054] In an embodiment of the present invention, the expansion factor of each box plot is obtained:

[0055] ;

[0056] In the formula, Represents the expansion factor of the xth box plot; Represents the noise level of the heating power data in the xth box plot; Represents the noise level of the heating power data in the previous boxplot of the xth boxplot; It represents the number of heating power data in the x-th box plot. It should be noted that the number of heating power data in the previous box plot of the x-th box plot is consistent with the number of heating power data in the x-th box plot; Represents the value of the bth heating power data in the xth box plot; Represents the value of the b-th heating power data in the previous box plot of the x-th box plot; Represents the preset second hyperparameter to prevent The value of is 0. In the embodiment of the present invention, the preset In other embodiments, the implementer may preset The value of

[0057] The larger the value is, the greater the noise level of the heating power data in the x-th box plot is. Therefore, when performing anomaly detection on the heating power data in the x-th box plot, the x-th box plot should contain more heating power data to minimize the interference of noise data on the anomaly detection results. Therefore, the larger the expansion factor of the x-th box plot is;

[0058] The larger the value is, the stronger the noise level of the heating power data in the previous box plot of the x-th box plot is. Therefore, when performing anomaly detection on the heating power data in the x-th box plot, more heating power data should be included in the x-th box plot to minimize the interference of noise data on the anomaly detection results. Therefore, the larger the expansion factor of the x-th box plot is;

[0059] It represents the sum of the differences of all one-to-one corresponding heating power data in the box plot before the x-th box plot. The larger the value is, the greater the trend difference between the heating power data in the box plot before the x-th box plot and the x-th box plot is, which means that the noise smoothing effect and utilization value of the heating power data in the box plot before the x-th box plot are worse for the x-th box plot. Then the x-th box plot should continue to include more heating power data to smooth the noise impact, and thus the expansion factor of the x-th box plot will be larger.

[0060] S4: According to the expansion factor of each box plot, the maximum expansion length of the box plot is corrected to obtain the expansion length of each box plot, and according to the expansion length of each box plot, an updated box plot of each box plot is obtained, and an abnormality judgment is made on each heating power data according to the updated box plot of each box plot.

[0061] It should be noted that the expansion factor of each box plot is obtained, so the expansion length of each box plot can be adaptively obtained according to the expansion factor of each box plot. The larger the expansion factor of any box plot, the greater the possibility that the heating power data in the box plot contains noise data and the more potential noise data points. Therefore, when performing anomaly detection on the heating power data points in the box plot, it is necessary to expand the box plot to include more heating power data to minimize the interference of noise data on the anomaly detection results, so the expansion length of the box plot will be larger.

[0062] In the embodiment of the present invention, the maximum extension length of the box plot is preset to DA. In the embodiment of the present invention, DA=200 is preset. In other embodiments, the implementer may preset the value of DA according to the specific implementation situation.

[0063] Get the extended length of each boxplot:

[0064] ;

[0065] In the formula, Represents the extended length of the xth box plot; Represents the expansion factor of the xth box plot; Represents the maximum extension length of the box plot; Represents the floor symbol; The larger the value of , the larger the extended length of the x-th box plot.

[0066] Get before the sampling time of the xth box plot Heating power data are recorded as the extended data of the x-th box plot. According to the heating power data in the x-th box plot and the extended data of the x-th box plot, an updated box plot of the x-th box plot is constructed. Similarly, an updated box plot of each box plot is obtained.

[0067] According to the performance of each heating power data in the updated box plot of the box plot to which it belongs, an abnormality judgment is made for each heating power data. It should be noted that the abnormality judgment of the data in the box plot is a well-known technology and will not be described in detail in the embodiments of the present invention.

[0068] An embodiment of the present invention further discloses a system for intelligently detecting energy consumption during a crystal growth process, comprising 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 method for intelligently detecting energy consumption during a crystal growth process according to the present invention is implemented.

[0069] 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.

[0070] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction 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, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, 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 the device or accessible or connectable to the device.

[0071] 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.

[0072] 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. A method for intelligent detection of energy consumption in a crystal growth process, characterized in that: Includes steps: Collect heating power data and heat flow data; Divide the heating power data into multiple groups of heating power data; construct a box plot for each group of heating power data to obtain the number of abnormal data in each box plot; obtain the noise level of the heating power data in each box plot , Represents the noise level of the heating power data in the xth box plot; as well as Represent the number of abnormal data in the x-th box plot and the number of reference box plots respectively; Represents the preset hyperparameters; represents the Pearson correlation coefficient between the heating power data in the rth reference boxplot of the xth boxplot and the heat flux data corresponding to its time series; represents the Pearson correlation coefficient between the heating power data in the xth box plot and the heat flow data corresponding to its time series; norm() represents the normalization function; Get the expansion factor for each boxplot , Represents the expansion factor of the xth box plot; Represents the noise level of the heating power data in the previous boxplot of the xth boxplot; Represents the number of heating power data in the xth box plot; Represents the value of the bth heating power data in the xth box plot; Represents the value of the b-th heating power data in the previous box plot of the x-th box plot; Represents the preset second hyperparameter; Get the updated box plot for each box plot based on the expansion factor of each box plot: the maximum expansion length of the x-th box plot is equal to After the product of is rounded down, it is recorded as ; For the heating power data in the x-th box plot and the data before the sampling time of the x-th box plot heating power data, construct an updated box-whisker plot of the x-th box-whisker plot; and perform abnormal judgment on each heating power data according to the updated box-whisker plot of each box-whisker plot.

2. The method for intelligently detecting energy consumption during crystal growth process according to claim 1, characterized in that: The step of dividing the heating power data into a plurality of groups of heating power data comprises: The number of data N is preset, and according to the number of data N, the heating power data in the heating power data sequence is divided into a plurality of groups of heating power data in sequence.

3. The method for intelligently detecting energy consumption during crystal growth process according to claim 1, characterized in that: The acquisition of the reference box plot includes: The number of box plots M is preset, and the M box plots before the time series and the M box plots after the time series of each box plot are recorded as reference box plots of each box plot.

4. An intelligent detection system for energy consumption during crystal growth, 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, a method for intelligently detecting energy consumption in a crystal growth process according to any one of claims 1 to 3 is implemented.

Images