A sintering working condition evaluation method based on multi-source heterogeneous information fusion

Abstract

The application provides a sintering condition evaluation method based on multi-source heterogeneous information fusion, comprising the following steps: S1: collecting data information of a sintering process, and extracting exhaust gas temperature and machine tail section information from the data information; S2: extracting the best observation section in a sintering cycle from the machine tail section information to obtain a machine tail section key frame, and extracting combustion zone feature information from the machine tail section key frame; S3: calculating and analyzing the sintering process data features according to the exhaust gas temperature and the combustion zone feature information; S4: establishing a sintering condition evaluation model according to the sintering process data features, and calculating a sintering condition comprehensive index according to the sintering condition evaluation model, and evaluating the sintering condition through the sintering condition comprehensive index, and the application realizes detection of the sintering section key frame. The detection accuracy of the method for the sintering cycle is 100%, and the detection of the key frame can meet the needs of sintering production personnel and condition evaluation.

Description

Technical Field

[0001] This invention relates to the field of intelligent identification technology for sintering processes, and in particular to a method for evaluating sintering conditions by fusing multi-source heterogeneous information. Background Technology

[0002] In steel production, sintering is a crucial step in pyrometallurgy. The quality of sinter directly determines the cost and product quality of subsequent processes, including blast furnace smelting. The sintering process is complex, and raw material and state parameters such as the sintering raw material ratio and ignition temperature have a significant impact on sintering conditions, chemical composition, and physical properties of the sinter.

[0003] In practice, due to the harsh production environment, high cost of online detection, and low accuracy of online detection during the sintering process, current production conditions can only be obtained through offline detection and manual observation. This makes it impossible to monitor production, adjust processes, and improve quality in a timely and effective manner. Therefore, real-time monitoring of the sintering production process and efficient control of sintering operations are crucial to ensuring the quality of sintered products. How to achieve real-time assessment of sintering conditions in harsh production environments is a pressing problem and challenge that needs to be addressed. Summary of the Invention

[0004] The purpose of this invention is to provide a sintering condition evaluation method based on multi-source heterogeneous information fusion, so as to achieve accurate evaluation of sintering conditions.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a sintering condition evaluation method based on multi-source heterogeneous information fusion, comprising: S1: collecting data information of the sintering process, and extracting exhaust gas temperature and tail section information from the data information; S2: extracting the best observation section within the sintering cycle from the tail section information to obtain a tail section keyframe, and extracting combustion zone feature information from the tail section keyframe; S3: calculating and analyzing sintering process data features based on the exhaust gas temperature and the combustion zone feature information; S4: establishing a sintering condition evaluation model based on the sintering process data features, calculating a comprehensive sintering condition index based on the sintering condition evaluation model, and evaluating the sintering condition through the comprehensive sintering condition index.

[0006] Further, S2 specifically includes: S21: using a convolutional neural network method to process the tail section information, thereby detecting the sintering cycle; S22: using the SIFT algorithm and K-Means algorithm to calculate and analyze the sintering cycle, and obtain the tail section keyframe.

[0007] Further, S22 specifically includes: S221: using the SIFT algorithm to extract feature points from each cross-sectional image within the sintering cycle, and constructing a feature vector from the feature points of each cross-sectional image; S222: calculating the similarity between each cross-sectional image based on the constructed feature vector; S223: clustering the cross-sectional images using the K-Means algorithm based on the similarity between each cross-sectional image; S224: selecting the cross-sectional image with the most feature points after clustering as the sintering keyframe sequence; S225: averaging all the cross-sectional images in the keyframe sequence to obtain the tail section keyframe.

[0008] Further, S3 specifically includes: S31: Calculating the sintering transverse heterogeneity index STHI by processing the combustion zone image in the combustion zone feature information using the Lorentz curve; S32: Calculating the final sintering endpoint index BTP, specifically including: S321: Establishing an exhaust gas temperature curve using the exhaust gas temperature, and obtaining the preliminary sintering endpoint through the exhaust gas temperature curve; S322: Correcting the preliminary sintering endpoint by introducing the exhaust gas temperature from the main flue to obtain the corrected sintering endpoint BTP. T S323: Utilize the features of the combustion zone image to determine the corrected sintering endpoint BTP. T The final sintering endpoint index BTP is obtained by making corrections; S33: The thickness H of the combustion zone is calculated based on the combustion zone characteristic information.

[0009] Further, S4 specifically includes: establishing the sintering condition evaluation model using the sintering transverse heterogeneity index STHI, the final sintering endpoint index BTP, and the combustion zone thickness H, and calculating the sintering condition comprehensive index using the sintering condition evaluation model.

[0010] Furthermore, the formula for calculating the Sintering Condition Comprehensive Index (SCI) is as follows: The Sintering Condition Comprehensive Index (SCI) ranges from [0, 1].

[0011] Furthermore, the sintering conditions are divided into multiple sintering condition levels, and the value of the Sintering Condition Comprehensive Index (SCI) is compared with the sintering condition level to complete the evaluation of the sintering conditions.

[0012] Furthermore, the number of sintering condition levels is divided into four; the value ranges of the four sintering condition levels are respectively: [0, 0.2), [0.2, 0.5), [0.5, 0.8) and [0.8, 1].

[0013] Furthermore, S323 specifically includes: calculating the thermal inertia I of the combustion zone based on the generated combustion zone image.h Entropy of combustion zone height distribution E h and the average heat storage value M in the combustion zone a ;Utilizing the combustion-induced thermal inertia I h The entropy of the combustion zone height distribution E h and the average heat storage value M in the combustion zone a The corrected sintering endpoint BTP T The final sintering endpoint index BTP is obtained by making corrections.

[0014] Furthermore, the corrected formula for the final sintering endpoint index BTP is: BTP = BTP T +β1I h +β2E h +β3M a β1, β2, and β3 are coefficients.

[0015] Analysis reveals that this invention discloses a sintering condition evaluation method based on multi-source heterogeneous information fusion. It also discloses an adaptive sintering cycle keyframe detection method based on the SIFT and K-Means algorithms, enabling the detection of keyframes on the sintering cross-section. This method achieves 100% accuracy in detecting the sintering cycle, and its keyframe detection meets the needs of sintering production personnel and condition evaluation. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. Wherein:

[0017] Figure 1 A flowchart of an embodiment of the present invention. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. Various examples are provided by way of explanation and not by way of limitation. Indeed, those skilled in the art will recognize that modifications and variations can be made to the invention without departing from its scope or spirit. For example, a feature shown or described as part of one embodiment may be used in another embodiment to produce yet another embodiment. Therefore, it is desirable that the present invention encompass such modifications and variations falling within the scope of the appended claims and their equivalents.

[0019] like Figure 1 As shown in the embodiment of the present invention, a sintering condition evaluation method based on multi-source heterogeneous information fusion is provided, which includes: S1: collecting data information of the sintering process, and extracting exhaust gas temperature and tail section information from the data information;

[0020] S2: Extract the best observation section within the sintering cycle from the tail section information to obtain the tail section key frame, and extract the combustion zone feature information from the tail section key frame;

[0021] S2 specifically includes: S21: using a convolutional neural network method to process the tail section information, thereby detecting the sintering cycle;

[0022] S22: The sintering cycle is calculated and analyzed using the SIFT algorithm and the K-Means algorithm to obtain the key frame of the tail section.

[0023] S22 specifically includes: S221: using the SIFT algorithm to extract feature points of each cross-sectional image during the sintering cycle, and constructing a feature vector from the feature points of each cross-sectional image;

[0024] S222: Calculate the similarity between each of the cross-sectional images based on the constructed feature vectors;

[0025] S223: Based on the similarity between each cross-sectional image, the K-Means algorithm is used to cluster the cross-sectional images. This invention first uses the SIFT algorithm to extract feature points from each cross-sectional image during the sintering cycle and constructs feature vectors. Then, the similarity between each feature vector is calculated, and the K-Means algorithm is used to cluster the cross-sectional images within the sintering cycle. Considering that a complete sintering cycle can be divided into pre-, mid-, and post-cycles, three data clusters are selected for clustering. Secondly, the class with the most average feature points is selected as the sintering keyframe sequence. Finally, all images in the keyframe sequence are averaged to obtain the sintering cycle keyframes.

[0026] S224: Select the cross-sectional image with the most feature points after clustering as the sintering keyframe sequence;

[0027] S225: Average all the cross-sectional images in the keyframe sequence to obtain the tail section keyframe.

[0028] S3: Based on the exhaust gas temperature and the combustion zone characteristic information, the sintering process data characteristics are calculated and analyzed;

[0029] S3 specifically includes: S31: Using the Lorentz curve to process the combustion zone image in the combustion zone feature information to calculate the sintering transverse heterogeneity index STHI;

[0030] S32: Calculate the final sintering endpoint index (BTP), specifically including:

[0031] S321: Establish an exhaust gas temperature curve using the exhaust gas temperature, and obtain the preliminary sintering endpoint through the exhaust gas temperature curve;

[0032] S322: The temperature of the exhaust gas introduced into the main flue is used to correct the preliminary sintering endpoint to obtain the corrected sintering endpoint BTP. T ;

[0033] S323: Utilize the features of the combustion zone image to determine the corrected sintering endpoint BTP. T The final sintering endpoint index BTP was obtained by making corrections.

[0034] S33: The thickness H of the combustion zone is calculated based on the combustion zone characteristic information.

[0035] S4: Establish a sintering condition evaluation model based on the sintering process data characteristics, calculate the sintering condition comprehensive index based on the sintering condition evaluation model, and evaluate the sintering condition through the sintering condition comprehensive index.

[0036] S4 specifically includes: establishing the sintering condition evaluation model using the sintering transverse heterogeneity index STHI, the final sintering endpoint index BTP, and the combustion zone thickness H, and calculating the sintering condition comprehensive index using the sintering condition evaluation model.

[0037] The formula for calculating the Sintering Condition Comprehensive Index (SCI) is as follows:

[0038]

[0039] The Sintering Condition Comprehensive Index (SCI) ranges from [0, 1].

[0040] The sintering conditions are divided into multiple sintering condition levels. The sintering condition comprehensive index (SCI) value is compared with the sintering condition level to complete the evaluation of the sintering conditions.

[0041] The number of sintering condition levels is divided into four; the value ranges of the four sintering condition levels are [0, 0.2), [0.2, 0.5), [0.5, 0.8) and [0.8, 1], respectively, which correspond to Level 4 (poor), Level 3 (medium), Level 2 (good) and Level 1 (excellent).

[0042] S323 specifically includes: calculating the thermal inertia I of the combustion zone based on the generated combustion zone image. h Entropy of combustion zone height distribution E h and the average heat storage value M in the combustion zone a ;Utilizing the combustion-induced thermal inertia I h The entropy of the combustion zone height distribution E h and the average heat storage value M in the combustion zone a The corrected sintering endpoint BTP TThe final sintering endpoint index BTP is obtained by making corrections.

[0043] The corrected formula for the final sintering endpoint index BTP is: BTP = BTP T +β1I h +β2E h +β3M a .

[0044] Compared with existing technologies, this invention addresses the shortcomings of current manual capture of key frames from the tail section of the sintering machine, such as reliance on experience, lag, and randomness. It proposes a multi-source heterogeneous information fusion method for sintering condition assessment, which has achieved excellent results in practical applications. This method first utilizes deep learning to detect sintering cycle videos from the sintering machine tail monitoring video. Then, it employs an adaptive sintering cycle key frame detection method based on the SIFT and K-Means algorithms to detect key frames of the sintering cross-section. This method achieves 100% accuracy in detecting the sintering cycle and meets the needs of sintering production personnel and condition assessment.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating sintering conditions through multi-source heterogeneous information fusion, characterized in that, include: S1: Collect data information from the sintering process, and extract exhaust gas temperature and tail section information from the data information; S2: Extract the best observation section within the sintering cycle from the tail section information to obtain the tail section key frame, and extract the combustion zone feature information from the tail section key frame; S3: Based on the exhaust gas temperature and the combustion zone characteristic information, the sintering process data characteristics are calculated and analyzed; S4: Establish a sintering condition evaluation model based on the sintering process data characteristics, calculate a comprehensive sintering condition index based on the sintering condition evaluation model, and evaluate the sintering condition using the comprehensive sintering condition index. S2 specifically includes: S21: The sintering cycle is detected by processing the tail section information using a convolutional neural network method. S22: The sintering cycle is calculated and analyzed using the SIFT and K-Means algorithms to obtain the keyframe of the tail section. S22 specifically includes: S221: Use the SIFT algorithm to extract feature points from each cross-sectional image during the sintering cycle, and construct a feature vector from the feature points of each cross-sectional image; S222: Calculate the similarity between each of the cross-sectional images based on the constructed feature vectors; S223: Based on the similarity between each cross-sectional image, the cross-sectional images are clustered using the K-Means algorithm; S224: Select the cross-sectional image with the most feature points after clustering as the sintering keyframe sequence; S225: Average all the cross-sectional images in the keyframe sequence to obtain the tail section keyframe. S3 specifically includes: S31: The sintering transverse heterogeneity index STHI is calculated by processing the combustion zone image in the combustion zone feature information using the Lorentz curve; S32: Calculate the final sintering endpoint index (BTP), specifically including: S321: Establish an exhaust gas temperature curve using the exhaust gas temperature, and obtain the preliminary sintering endpoint through the exhaust gas temperature curve; S322: The temperature of the exhaust gas introduced into the main flue is used to correct the preliminary sintering endpoint to obtain the corrected sintering endpoint BTP. T ; S323: Utilize the features of the combustion zone image to determine the corrected sintering endpoint BTP. T The final sintering endpoint index BTP was obtained by making corrections. S33: The thickness H of the combustion zone is calculated based on the combustion zone characteristic information, wherein S4 specifically includes: The sintering condition evaluation model is established using the sintering transverse heterogeneity index STHI, the final sintering endpoint index BTP, and the combustion zone thickness H. The comprehensive sintering condition index (SCI) is then calculated using this model. The formula for calculating the comprehensive sintering condition index (SCI) is as follows: The Sintering Condition Comprehensive Index (SCI) ranges from [0, 1].

2. The sintering condition evaluation method based on multi-source heterogeneous information fusion according to claim 1, characterized in that, The sintering conditions are divided into multiple sintering condition levels. The sintering condition comprehensive index (SCI) value is compared with the sintering condition level to complete the evaluation of the sintering conditions.

3. The sintering condition evaluation method based on multi-source heterogeneous information fusion according to claim 2, characterized in that, The number of sintering condition levels is divided into four; The value ranges of the four sintering condition levels are [0, 0.2), [0.2, 0.5), [0.5, 0.8), and [0.8, 1].

4. The sintering condition evaluation method based on multi-source heterogeneous information fusion according to claim 1, characterized in that, S323 specifically includes: The thermal inertia I of the combustion zone is calculated based on the generated combustion zone image. h Entropy of combustion zone height distribution E h and the average heat storage value M in the combustion zone a ; Utilizing the combustion thermal inertia I h The entropy of the combustion zone height distribution E h and the average heat storage value M in the combustion zone a The corrected sintering endpoint BTP T The final sintering endpoint index BTP is obtained by making corrections.

5. The sintering condition evaluation method based on multi-source heterogeneous information fusion according to claim 4, characterized in that, The corrected formula for the final sintering endpoint index BTP is: , , , is a coefficient.

Images