A method, system, storage medium, and electronic device for detecting the composition of melt in a furnace.

Abstract

This invention belongs to the field of laser spectral analysis technology and proposes a method, system, storage medium, and electronic device for detecting the composition of melt in a furnace. The method includes: acquiring a raw multi-furnace, multi-period, multi-point spectral dataset; preprocessing the raw multi-furnace, multi-period, multi-point spectral dataset to obtain a preprocessed multi-furnace, multi-period, multi-point spectral dataset; constructing multiple quantitative detection models based on the preprocessed multi-furnace, multi-period, multi-point spectral dataset; selecting a suitable quantitative detection model from the multiple quantitative detection models; and using the suitable quantitative detection model to detect the composition of melt in the furnace and obtain the detection results. This invention overcomes the shortcomings of single-point sampling representativeness by the sampling rod, reduces the influence of physical properties and texture of melts in different furnace periods on the detection results, and improves the detection accuracy and precision of the detection results.

Description

Technical Field

[0001] This invention belongs to the field of laser spectral analysis technology, and particularly relates to a method, system, storage medium and electronic device for detecting the composition of melt in a furnace. Background Technology

[0002] During the smelting process, it is necessary to monitor the composition of the materials in the reactor in real time and adjust the core smelting parameters such as material ratio and feed amount in a timely manner.

[0003] Currently, fluorescence analysis is mainly used to conduct periodic high-frequency tests on materials. However, fluorescence analysis requires manual sampling, sample preparation, and testing, and the test results are subject to a certain lag.

[0004] Laser-induced breakdown spectroscopy (LIBS) is an emerging atomic emission spectroscopic analysis technique with advantages such as no sample preparation required and online in-situ detection, enabling real-time online material detection and assisting in real-time process control. However, it is affected by factors such as flue gas, large liquid level fluctuations, and surface scum. While sampling with a sampling rod can reduce the impact of on-site conditions, the sampling rod passes through a surface scum layer, causing scum to adhere to the outermost layer of the rod. Furthermore, the surface morphology and texture of the material on the sampling rod vary significantly across different furnace phases in a single smelting process in the reactor. Direct detection and modeling can affect the quantitative analysis accuracy of laser-induced breakdown spectroscopy. Summary of the Invention

[0005] This invention proposes a method, system, storage medium, and electronic device for detecting the composition of melt in a furnace.

[0006] The present invention provides a method for detecting the composition of melt in a furnace, the method comprising:

[0007] Obtain the original multi-furnace, multi-furnace, multi-period, multi-point spectral dataset, and preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset.

[0008] All furnace periods of each furnace batch are divided into multiple segments. The preprocessed multi-furnace multi-furnace period multi-point spectral dataset is divided according to the segment time to obtain a multi-furnace multi-segment spectral dataset.

[0009] The multi-furnace, multi-segment spectral datasets are combined according to the segment sequence number to obtain multiple segmented combined spectral datasets;

[0010] Based on the segmented combined spectral dataset, a quantitative detection model is constructed to obtain multiple quantitative detection models;

[0011] Select a suitable quantitative detection model from among the multiple quantitative detection models, and use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results.

[0012] Furthermore,

[0013] The original multi-furnace, multi-furnace period, multi-site spectral dataset is specifically represented as follows:

[0014] {(S1,S2,……,S i )1,(S1,S2,……,S i )2,……,(S1,S2,……,S i ) j},

[0015] Where i represents the furnace period; j represents the furnace number; S i This is the original multi-site spectral dataset for a single furnace period; (S1,S2,……,S i ) j This is the original single-furnace, multi-furnace, multi-site spectral dataset.

[0016] Furthermore,

[0017] The original single-furnace multi-site spectral dataset is specifically represented as follows:

[0018] {(S 11 ,S 12 ,……,S 1n ),(S 21 ,S 22 ,……,S 2n ),……,(S m1 ,S m2 ,……,S mn )},

[0019] Where m is the location; n is the height of the sampling rod; S mn This is the original single-point spectral data; (S m1 ,S m2 ,……,S mn () is the original single-point spectral dataset.

[0020] Furthermore,

[0021] The original multi-furnace, multi-period, multi-site spectral dataset is preprocessed to obtain the preprocessed multi-furnace, multi-period, multi-site spectral dataset, including:

[0022] The original multi-furnace, multi-period, multi-point spectral dataset is subjected to outlier removal and scum spectrum removal processing to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset {(S e1 ,Se2 ,……,S ei )1,(S e1 ,S e2 ,……,S ei )2,……,(S e1 ,S e2 ,……,S ei ) j}, where i is the furnace period; j is the furnace number; S ei This is a preprocessed multi-site spectral dataset for a single furnace period; (S) e1 ,S e2 ,……,S ei ) j This is a preprocessed multi-furnace, multi-period, multi-site spectral dataset for a single furnace.

[0023] Furthermore,

[0024] Based on the form and texture of the material inside the furnace, all furnace periods of each furnace batch are divided into K segments, where K ≤ I, and I is the number of furnace periods in a single furnace batch.

[0025] Furthermore,

[0026] The multiple segmented combined spectral datasets are specifically represented as follows:

[0027] (data1,data2,……,data k ),

[0028] Where k represents the segment; data k This is a segmented combined spectral dataset.

[0029] Furthermore,

[0030] The various quantitative detection models are specifically represented as follows:

[0031] {y=f1(x),y=f2(x),……,y=f l (x)},

[0032] Where x represents the spectral features extracted from the spectral data; y represents the result of the melt composition detection in the furnace; l represents the serial number of the quantitative detection model; y = f l (x) represents a quantitative detection model.

[0033] The present invention provides an in-furnace melt composition detection system for implementing the aforementioned in-furnace melt composition detection method, the system comprising:

[0034] The preprocessing module is used to acquire the original multi-furnace, multi-period, multi-point spectral dataset, and to preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset.

[0035] The partitioning module is used to divide all furnace periods of each furnace batch into multiple segments, and to partition the preprocessed multi-furnace multi-furnace period multi-point spectral dataset according to the segment time to obtain a multi-furnace multi-segment spectral dataset.

[0036] The combination module is used to combine the multi-furnace multi-segment spectral datasets according to the segment sequence number to obtain multiple segmented combined spectral datasets.

[0037] The construction module is used to construct a quantitative detection model based on the segmented combined spectral dataset, thereby obtaining multiple quantitative detection models;

[0038] The detection module is used to select a suitable quantitative detection model from multiple quantitative detection models, and to use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results.

[0039] The present invention provides a computer-readable storage medium storing a program or instructions that, when executed on a computer, cause the computer to perform the aforementioned furnace melt composition detection method.

[0040] An electronic device according to the present invention includes a processor coupled to a memory; the processor is used to read and execute a computer program stored in the memory to implement the aforementioned method for detecting the composition of melt in a furnace.

[0041] Compared with the prior art, the beneficial effects of this invention are:

[0042] This invention overcomes the shortcomings of single-point sampling by using a sampling rod to take samples at multiple points. By aggregating data from multiple furnace cycles and the same segment, a quantitative detection model is established. The most suitable quantitative detection model is selected based on actual working conditions to detect the melt composition. This reduces the influence of physical properties and texture of the melt in different furnace cycles on the detection results, thereby improving the detection accuracy and precision. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the fully automated sampling method of the present invention;

[0045] Figure 2 This is a flowchart of the furnace melt composition detection method of the present invention;

[0046] Figure 3 This is a schematic diagram comparing the Sn element detection results of tin-rich slag before the implementation of the furnace melt composition detection method of the present invention;

[0047] Figure 4 This is a schematic diagram comparing the Sn element detection results of tin-rich slag after the implementation of the furnace melt composition detection method of the present invention;

[0048] Figure 5 This is a schematic diagram of the furnace melt composition detection system of the present invention;

[0049] Figure 6 This is a schematic diagram of the structure of the electronic device of the present invention.

[0050] Explanation of reference numerals in the attached figures:

[0051] 1-Sampling rod, 2-Sampling rod system, 3-Optical detection system, 201-Preprocessing module, 202-Division module, 203-Combination module, 204-Construction module, 205-Detection module, 301-Processor, 302-Memory. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Furthermore, in order to gain a deeper understanding of this invention, some of the terms used herein are explained as follows:

[0054] A furnace run refers to the entire process of smelting in a reactor, from the start of smelting to the discharge.

[0055] A furnace period refers to different stages or time periods within the same furnace batch. Based on the different degrees of smelting and reduction, a furnace batch is divided into multiple furnace periods.

[0056] Before implementing the in-furnace melt composition detection method provided in this embodiment of the invention, it is necessary to collect spectral data inside the reactor (hereinafter referred to as "furnace") using a fully automatic sampling rod (hereinafter referred to as "sampling rod"). The following is the process of collecting spectral data for multiple furnace cycles:

[0057] Before collecting spectral data, select a specific furnace period of a specific furnace batch and set the sampling points and the trajectory of the sampling rod.

[0058] first step:

[0059] like Figure 1 As shown, a sampling rod is inserted into a point inside the furnace to collect material from the furnace. The sampling rod is then extracted to a specified height, and the spectral data at that specified height is collected using the LIBS device (LIBS device refers to laser-induced breakdown spectroscopy device) on the sampling rod. A single spectral data point is obtained, and the single spectral data point obtained at one point is defined as a single-point spectral data point.

[0060] By extracting the sampling rod to different heights and collecting spectral data at different heights, multiple spectral data points can be obtained at a single location. These multiple spectral data points obtained at a single location are defined as a single-location spectral dataset.

[0061] By performing a single-point acquisition of the spectral data generated by the reactor during a specific furnace cycle, a single-point spectral dataset for a single furnace cycle and a specific furnace period is obtained.

[0062] Step Two:

[0063] Following the sampling rod's trajectory, move the sampling rod to the next point and continue collecting spectral data at that point using the same method, obtaining another set of single-point spectral datasets. Continue this process until spectral data has been collected from all points using the sampling rod's trajectory, resulting in multiple sets of single-point spectral datasets. Define the multiple spectral data points obtained from all points as a multi-point spectral dataset, and each multi-point spectral dataset comprises multiple sets of single-point spectral datasets.

[0064] In summary, following the methods described in steps one and two, multiple multi-point acquisitions are performed on the spectral data generated by the reactor during a specific furnace cycle, resulting in multiple sets of multi-point spectral datasets for a single furnace cycle and a single furnace period. For example, if a specific furnace cycle is from 12:00 to 13:00, the spectral data generated by the reactor during that specific furnace cycle and a single furnace period is acquired through five multi-point acquisitions, resulting in five sets of multi-point spectral datasets for a single furnace cycle and a single furnace period.

[0065] Step 3:

[0066] For all furnace periods of a single furnace, spectral data are collected according to the methods described in steps one and two, ultimately resulting in a multi-point spectral dataset for a single furnace across multiple furnace periods.

[0067] Step 4:

[0068] For all furnace cycles and all furnace periods, spectral data were collected according to the methods described in steps one, two, and three, ultimately resulting in a multi-furnace, multi-furnace, multi-point spectral dataset.

[0069] The single-furnace single-period single-point spectral dataset, the single-furnace single-period multi-point spectral dataset, the single-furnace multi-period multi-point spectral dataset, and the multi-furnace multi-period multi-point spectral dataset all have corresponding acquisition times.

[0070] In addition, when selecting a sampling rod, a sampling rod with a circular, rectangular or square cross-section can be selected. Preferably, the sampling rod has a square cross-section because selecting a square cross-section can ensure the flatness and usability of each surface of the sampling rod.

[0071] Figure 2 This is a flowchart of a method for detecting the composition of melt in a furnace according to an embodiment of the present invention. In one embodiment, it specifically includes the following steps:

[0072] S101: Obtain the original multi-furnace, multi-period, multi-point spectral dataset, preprocess the original multi-furnace, multi-period, multi-point spectral dataset, and obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset.

[0073] S101-1: Obtain the original multi-furnace, multi-period, multi-point spectral dataset {(S1,S2,……,S...} i )1,(S1,S2,……,S i )2,……,(S1,S2,……,S i ) j}, where i is the furnace period; j is the furnace number; S i This is the original multi-site spectral dataset for a single furnace period; (S1,S2,……,S i ) j This is the original single-furnace, multi-furnace, multi-site spectral dataset. The original multi-furnace, multi-site spectral dataset is the raw, unprocessed dataset directly collected by the sampling rod.

[0074] Specifically, S i ={(S 11 ,S 12 ,……,S 1n ),(S 21 ,S 22 ,……,S 2n ),……,(S n1 ,Sn2 ,……,S nb )}, where m is the location; n is the height of the sampling rod; S mn This refers to the raw, single-point spectral data, i.e., the raw spectral data collected at a specific altitude from a single point; (S m1 ,S m2 ,……,S mn () is the original single-point spectral dataset.

[0075] S101-2: Preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset.

[0076] Pretreatment methods include scum spectrum removal and abnormal point removal.

[0077] Outlier spectrum removal refers to removing abnormal spectral data from all single-site spectral datasets. The processing method for a specific set of single-site spectral datasets involves: extracting features from the dataset, and cleaning the data using outlier removal methods such as correlation and 3sigma to remove outlier spectra and obtain a valid single-site spectral dataset.

[0078] Outlier removal refers to removing outliers from all multi-site spectral datasets. The processing method for a specific multi-site spectral dataset involves: feature extraction, outlier removal methods such as correlation and 3sigma, and data cleaning based on smelting mechanisms to remove outliers and obtain a valid multi-site spectral dataset. The smelting mechanism refers to the fact that during the smelting process, certain process parameters may deviate from the normal range due to equipment failure, raw material changes, or improper operation, leading to the generation of abnormal data. By comparing the data range and trends under normal smelting mechanisms, these abnormal data can be more easily identified.

[0079] By performing scum spectrum removal processing on all single-point spectral datasets in the original multi-furnace multi-period multi-point spectral dataset, and by performing outlier removal processing on all multi-point spectral datasets in the original multi-furnace multi-period multi-point spectral dataset, an effective multi-furnace multi-period multi-point spectral dataset is obtained. This effective multi-furnace multi-period multi-point spectral dataset is also called the preprocessed multi-furnace multi-period multi-point spectral dataset. The preprocessed multi-furnace multi-period multi-point spectral dataset is defined as {(S... e1 ,S e2 ,……,S ei )1,(S e1 ,S e2 ,……,S ei )2,……,(S e1 ,S e2,……,S ei ) j}, where i is the furnace period; j is the furnace number; S ei This is a preprocessed multi-site spectral dataset for a single furnace period; (S) e1 ,S e2 ,……,S ei ) j This is a preprocessed multi-furnace, multi-period, multi-site spectral dataset for a single furnace.

[0080] S102: Divide all furnace periods of each furnace batch into multiple segments, and divide the preprocessed multi-furnace multi-furnace period multi-point spectral dataset according to the segment time to obtain a multi-furnace multi-segment spectral dataset.

[0081] Based on the form and texture of the materials inside the furnace, all furnace periods of a single furnace cycle are divided into multiple continuous and non-overlapping segments. The number of segments is defined as K, K≤I, where I is the number of all furnace periods in a single furnace cycle. All furnace periods of each furnace cycle are divided according to the method described in step S102-1, and all furnace periods of each furnace cycle are divided into K segments.

[0082] Each segment has its corresponding time and sequence number. The time corresponding to each segment is defined as the segment time, and the sequence number corresponding to each segment is defined as the segment sequence number.

[0083] The acquisition time of the preprocessed multi-furnace, multi-period, multi-location spectral dataset is obtained. The segmented time is aligned with the acquisition time in units. The preprocessed multi-furnace, multi-period, multi-location spectral dataset is divided according to the segmented time to obtain the spectral dataset of all segments of all furnaces. The spectral dataset of all segments of all furnaces is referred to as the "multi-furnace, multi-segment spectral dataset".

[0084] S103: Combine the multi-furnace, multi-segment spectral datasets according to the segment number to obtain multiple segmented combined spectral datasets.

[0085] Multiple segmented combined spectral datasets are defined as (data1, ata2, ..., ata). k ), where k is the segment; data k This is a segmented combined spectral dataset.

[0086] For example, the spectral datasets of the first segment of the first furnace, the first segment of the second furnace, and the first segment of the third furnace are combined to obtain three combined spectral datasets (data1, ata2, ata3).

[0087] S104: Quantitative detection models are constructed based on segmented combined spectral datasets, resulting in multiple quantitative detection models.

[0088] Multiple quantitative detection models are defined as {y = f1(x), = f2(x), ..., = f...} l (x)}, where x is the spectral feature extracted from the spectral data; y is the result of the in-furnace melt composition detection; l is the serial number of the quantitative detection model; y = f l (x) represents a quantitative detection model. The results of the melt composition detection in the furnace are specifically expressed as element concentrations.

[0089] It is worth noting that the number of quantitative detection models is equal to the number of segments.

[0090] S105: Select a suitable quantitative detection model from multiple quantitative detection models, use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results.

[0091] A multi-point spectral dataset of a single furnace run and a single furnace period to be detected is obtained. Spectral features are extracted from the dataset. Based on the furnace period setting and process time (process time refers to the time difference between furnace periods), the specific position of the multi-point spectral dataset of a single furnace run and a single furnace period to be detected within the furnace period setting is determined. Combining the previous detection results, a quantitative detection model applicable to the multi-point spectral dataset of a single furnace run and a single furnace period to be detected is selected. The spectral features are input into the applicable quantitative detection model. The applicable quantitative detection model is used to detect the melt composition in the furnace and obtain the detection results, i.e., the element concentration.

[0092] The system acquires a multi-point spectral dataset of a single furnace run and a single furnace period to be tested. Spectral features are extracted from the multi-point spectral dataset of a single furnace run and a single furnace period. The system can also manually select the applicable quantitative detection model, input the spectral features into the applicable quantitative detection model, and use the applicable quantitative detection model to detect the melt composition in the furnace and obtain the detection results, i.e., the element concentration.

[0093] Verification Example

[0094] Verification was conducted at a tin smelter.

[0095] like Figure 3 As shown, the horizontal axis represents the furnace batch, with each batch comprising multiple furnace periods; the vertical axis represents the concentration of melt composition within the furnace; the blue curve represents the reference value for Sn element concentration; and the orange curve represents the actual detected value for Sn element concentration. Using the previous model to detect Sn element using this method yielded poor results, prone to anomalies, and with a significant deviation between the actual detected value and the reference value.

[0096] like Figure 4As shown, the horizontal axis represents the furnace batch, with each batch comprising multiple furnace periods; the vertical axis represents the concentration of melt components within the furnace; the blue curve represents the reference value for Sn element concentration; and the orange curve represents the actual detected value for Sn element concentration. The quantitative detection model implemented using this method rarely produces abnormal results for Sn element, and the actual detected values ​​are largely consistent with the reference values. The detection effect is significantly improved and better reflects the smelting mechanism, namely: during the reduction of tin-rich slag, the Sn element content gradually decreases, while the contents of elements such as Fe, Si, and Ca gradually increase.

[0097] Therefore, the furnace melt composition detection method of the present invention improves the accuracy of furnace melt composition detection.

[0098] Embodiments of the present invention also provide an in-furnace melt composition detection system, such as... Figure 5 As shown, it includes:

[0099] The preprocessing module 201 is used to obtain the original multi-furnace, multi-period, multi-point spectral dataset, and to preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset.

[0100] The segmentation module 202 is used to divide all furnace periods of each furnace batch into multiple segments, and to divide the preprocessed multi-furnace multi-furnace period multi-point spectral dataset according to the segment time to obtain a multi-furnace multi-segment spectral dataset.

[0101] The combination module 203 is used to combine multi-furnace multi-segment spectral datasets according to the segment sequence number to obtain multiple segmented combined spectral datasets.

[0102] Module 204 is used to build quantitative detection models based on segmented combined spectral datasets, resulting in multiple quantitative detection models.

[0103] The detection module 205 is used to select a suitable quantitative detection model from multiple quantitative detection models, and to use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results.

[0104] It should be noted that the preprocessing module 201, the division module 202, the combination module 203, the construction module 204 and the detection module 205 mentioned above correspond to steps S101 to S105 in the embodiment of the furnace melt composition detection method. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the above embodiments.

[0105] Embodiments of the present invention also provide a computer-readable storage medium storing a program or instructions that, when executed on a computer, cause the computer to perform the furnace melt composition detection method as described in the above method embodiments.

[0106] like Figure 6 As shown, embodiments of the present invention also provide an electronic device, including: a processor 301, the processor 301 being coupled to a memory 302, the processor 301 being used to read and execute a computer program stored in the memory 302 to implement the furnace melt composition detection method as described in the above method embodiments.

[0107] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the composition of melt in a furnace, characterized in that, include: Obtain the original multi-furnace, multi-furnace, multi-period, multi-point spectral dataset, and preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset. All furnace periods of each furnace batch are divided into multiple segments. The preprocessed multi-furnace multi-furnace period multi-point spectral dataset is divided according to the segment time to obtain a multi-furnace multi-segment spectral dataset. The multi-furnace, multi-segment spectral datasets are combined according to the segment sequence number to obtain multiple segmented combined spectral datasets; Based on the segmented combined spectral dataset, a quantitative detection model is constructed to obtain multiple quantitative detection models; Select a suitable quantitative detection model from among the multiple quantitative detection models, and use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results; The original multi-furnace, multi-period, multi-site spectral dataset is preprocessed to obtain the preprocessed multi-furnace, multi-period, multi-site spectral dataset, including: The original multi-furnace, multi-period, multi-site spectral dataset is subjected to outlier removal and outlier spectral data removal processing to obtain the preprocessed multi-furnace, multi-period, multi-site spectral dataset. ,in, For the furnace period; For furnace batches; This is a preprocessed multi-point spectral dataset for a single furnace period; This is a preprocessed multi-furnace, multi-period, multi-site spectral dataset for a single furnace.

2. The method according to claim 1, characterized in that, The original multi-furnace, multi-furnace period, multi-site spectral dataset is specifically represented as follows: ， in, For the furnace period; For furnace batches; This is the original multi-point spectral dataset for a single furnace period; This is the original single-furnace, multi-furnace, multi-site spectral dataset.

3. The method according to claim 2, characterized in that, The original single-furnace multi-site spectral dataset is specifically represented as follows: ， in, For the location; The height of the sampling rod; This is the original single-point spectral data; This is the original single-point spectral dataset.

4. The method according to claim 1, characterized in that, Based on the form and texture of the material inside the furnace, all furnace periods of each furnace batch are divided into K segments, where K ≤ I, and I is the number of furnace periods in a single furnace batch.

5. The method according to claim 1 or 4, characterized in that, The multiple segmented combined spectral datasets are specifically represented as follows: ， in, For segmentation; This is a spectral dataset for a certain segment.

6. The method according to claim 1 or 4, characterized in that, The various quantitative detection models are specifically represented as follows: ， in, These are spectral features extracted from spectral data. The results of the analysis of the melt composition inside the furnace; This is the serial number of the quantitative detection model; For a certain quantitative detection model.

7. A furnace melt composition detection system, characterized in that, A method for performing the in-furnace melt composition detection method according to any one of claims 1-6, comprising: The preprocessing module is used to acquire the original multi-furnace, multi-period, multi-point spectral dataset, and to preprocess the original multi-furnace, multi-period, multi-point spectral dataset to obtain the preprocessed multi-furnace, multi-period, multi-point spectral dataset. The partitioning module is used to divide all furnace periods of each furnace batch into multiple segments, and to partition the preprocessed multi-furnace multi-furnace period multi-point spectral dataset according to the segment time to obtain a multi-furnace multi-segment spectral dataset. The combination module is used to combine the multi-furnace multi-segment spectral datasets according to the segment sequence number to obtain multiple segmented combined spectral datasets. The construction module is used to construct a quantitative detection model based on the segmented combined spectral dataset, thereby obtaining multiple quantitative detection models; The detection module is used to select a suitable quantitative detection model from multiple quantitative detection models, and to use the suitable quantitative detection model to detect the composition of the melt in the furnace and obtain the detection results.

8. A computer-readable storage medium, characterized in that, The system stores a program or instructions that, when executed on a computer, cause the computer to perform the furnace melt composition detection method as described in any one of claims 1-6.

9. An electronic device, characterized in that, Includes a processor, which is coupled to a memory; The processor is used to read and execute the computer program stored in the memory to implement the furnace melt composition detection method as described in any one of claims 1-6.

Images