Sintering furnace operation control method based on tensor decomposition

Abstract

The invention discloses a sintering furnace operation control method based on tensor decomposition, relates to the field of information fusion technology, and solves the technical problem that the prior art does not consider the influence of other factors on the operation of the sintering furnace, does not fuse and analyze data, and thus the final analysis result has a certain gap with the optimal state; the invention collects real-time status data of the sintering furnace, pre-processes the collected real-time status data, ensures the integrity of subsequent analysis data, and is conducive to improving the efficiency of data collection; uses tensor decomposition to decompose the real-time status data of the sintering furnace to obtain decomposed data, extracts characteristic parameters from the decomposed data, and integrates the characteristic parameters into an evaluation sequence; uses a parameter evaluation model to analyze the evaluation sequence to obtain an evaluation coefficient, adjusts the operation state of the sintering furnace according to the evaluation data, and continuously detects the evaluation data of the sintering furnace to keep the sintering furnace in the optimal operation state, which is conducive to maintaining the quality of the products produced and improving the production efficiency.

Description

Technical Field

[0001] The invention belongs to the field of information fusion, and relates to a sintering furnace information fusion technology, and specifically to a sintering furnace operation control method based on tensor decomposition. Background Art

[0002] Pressure sintering furnace is a material preparation equipment that combines vacuum / atmosphere, hot pressing and high-temperature sintering. It is suitable for sintering new materials such as silicon nitride, cemented carbide and functional ceramics in a high-pressure nitrogen atmosphere, which is beneficial to increase the sintering density of the material and improve the mechanical properties of the material. Pressure sintering furnaces are widely used in materials science, electricity, optics, mechanical engineering and other fields. Precise control of temperature, atmosphere, time and other parameters in the sintering furnace can obtain a stable sintering process, thereby improving product quality and consistency. Adjusting the parameters in the sintering furnace can adapt to the sintering requirements of different materials, which is beneficial to adjust the corresponding parameters according to different materials. The use of automated control can improve production efficiency and reduce the possibility of manual intervention and misoperation.

[0003] The prior art (the invention patent application with the publication number CN117268119A) discloses an atmosphere control system for a vacuum sintering furnace, a data acquisition unit, a data analysis unit, an abnormal analysis unit, a normal monitoring unit, an abnormal adjustment unit and an information output unit. The present invention relates to the technical field of sintering furnaces, and solves the technical problem that atmosphere control based on the experience of operators will cause large errors in the working process, which will further directly affect the final performance of the product. The present invention monitors the atmosphere inside the sintering furnace, monitors its oxygen content and pressure, and reasonably corrects past data based on the real-time status data generated inside the sintering furnace, further adjusts the atmosphere system inside the sintering furnace based on the acquired real-time status data, and performs secondary mediation analysis for abnormal situations, so as to avoid large errors in the working process caused by adjustment based on the experience of operators, thereby avoiding affecting the subsequent overall quality; however, the prior art adjusts the atmosphere in the sintering furnace by monitoring its oxygen content and pressure, without considering the influence of other factors on the operation of the sintering furnace, and without fusing and analyzing the data, so that the final analysis result has a certain gap with the optimal state.

[0004] The present invention provides a sintering furnace operation control method based on tensor decomposition to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a sintering furnace operation control method based on tensor decomposition, which is used to solve the technical problem that the prior art does not consider the impact of other factors on the operation of the sintering furnace and does not fuse the data for analysis, resulting in a certain gap between the final analysis result and the optimal state.

[0006] To achieve the above object, a first aspect of the present invention provides a sintering furnace operation control method based on tensor decomposition, comprising:

[0007] Step S1: collecting real-time status data of the sintering furnace through a number of data sensors; preprocessing the collected real-time status data, converting the processed real-time status data into a tensor format, and obtaining tensor data; wherein the real-time status data includes real-time temperature, real-time pressure and real-time atmosphere; the data sensors are arranged according to the structure of the sintering furnace;

[0008] Step S2: decomposing the tensor data using a tensor decomposition algorithm to obtain decomposition data, and extracting the characteristics of the sintering furnace based on the decomposition data to obtain characteristic parameters;

[0009] Step S3: Integrate the characteristic parameters into an evaluation sequence, and call a parameter evaluation model based on the evaluation sequence to evaluate the operating state of the sintering furnace to obtain an evaluation coefficient; wherein the parameter evaluation model is constructed based on an artificial intelligence model;

[0010] Step S4: adjusting the operating state of the sintering furnace based on the evaluation coefficient, and continuously monitoring the operating state of the sintering furnace.

[0011] Preferably, the data sensor is arranged according to the structure of the sintering furnace, including:

[0012] Determine the key parameters that need to be measured according to the requirements of the sintering process, as well as the key points at which the corresponding key parameters are measured in the sintering furnace; obtain the data sensor's collection range and the internal space of the sintering furnace, and determine the number of data sensors to be set; spread the data sensors evenly around the key points; among them, the key parameters refer to the real-time status data, and the key points refer to the points where the key parameters fluctuate the most.

[0013] The present invention arranges data sensors according to the structure of the sintering furnace, and determines the position of the sensor according to the fluctuation of key parameters, so as to focus on places where the key parameters fluctuate greatly; and according to the collection range of the data sensor, it is evenly arranged from the key point to the surrounding areas, so that the collection range can be wider, which is beneficial to the integrity of the collected data.

[0014] Preferably, the preprocessing of the collected real-time status data includes:

[0015] Retrieving a number of collected real-time status data, determining whether the number of real-time status data contains sub-data of all data types, if yes, retaining the group of real-time status data; if no, discarding the group of real-time status data;

[0016] Determine whether there are abnormal difference data in several groups of real-time status data; if yes, remove the corresponding real-time status data; if no, retain the corresponding real-time status data.

[0017] It should be noted that abnormal difference data refers to the situation in which the absolute value of the difference between the largest data and the second largest data in a group of data is greater than a set threshold, or the absolute value of the difference between the smallest data and the second smallest data is greater than a set threshold. In this case, the largest data or the smallest data in the group is called abnormal difference data.

[0018] The present invention preprocesses the collected real-time status data and removes data with missing values ​​and abnormalities, which can ensure the integrity of subsequent analysis data and is conducive to improving the efficiency of data collection.

[0019] Preferably, converting the processed real-time status data into a tensor format includes:

[0020] The tensor dimension is determined according to the dimension of the real-time status data, and an empty tensor object is created by using a programming tool; the real-time status data is filled into the tensor object according to the data type to obtain tensor data; wherein the programming tool includes NumPy or TensorFlow in Python.

[0021] The present invention utilizes a programming tool to convert the preprocessed real-time status data into a tensor format, which can facilitate subsequent tensor decomposition of the tensor data and is beneficial to the fusion of the sintering furnace data.

[0022] Preferably, the decomposing the tensor data using a tensor decomposition algorithm comprises:

[0023] Retrieve the tensor data and mark it as x, and decompose the tensor into three-order tensors according to the quantity type of real-time status data Among them, the size of tensor x is n 1 ×n 2 ×n 3 ; The size is r 1 × 2 × 3 ; The size of the matrix U is n 1 × 1 ; The size of the matrix V is n 2 × 2 ; The size of the matrix W is n 3 × 3 ;

[0024] Simplify the tensor expression to get the simplified tensor The least squares method is used alternately for each iteration and each matrix factor.

[0025] Preferably, the alternately using the least square method for each iteration and each matrix factor comprises:

[0026] S1: Compute the mode-n horizontal expansion X(n) of the tensor x, n=1,…N;

[0027] S2: Calculate X(n) = U∑V T , n = 1, ... N, determine its effective rank R n , let U (n) ←U(:,1:R n );

[0028] S3: Compute

[0029] S4: When the core tensor When U (1) , U (2) , …U (n) ; Otherwise, loop S2-S3.

[0030] Preferably, the extracting the characteristics of the sintering furnace based on the decomposition data comprises:

[0031] Retrieve the decomposed data, extract the change trend, periodicity and stability of the real-time status data from the decomposed data; integrate the change trend, periodicity and stability into characteristic parameters.

[0032] It should be noted that the judgment of the change trend is determined based on the degree of curve change of the real-time status data within the set time period; the judgment of periodicity is based on whether there is periodic change in the real-time status data within the set time period; the judgment of stability is based on the fluctuation range and frequency of the real-time status data, and the stability is determined by comparing with the stability table; the stability table is comprehensively obtained based on the historical data of the sintering furnace.

[0033] The present invention extracts characteristic parameters of real-time status data changes from decomposed data, which can provide a basis for subsequent evaluation of the operating status of the sintering furnace.

[0034] Preferably, the step of evaluating the operating state of the sintering furnace by calling the parameter evaluation model based on the evaluation sequence includes:

[0035] Extract the evaluation sequence and call the parameter evaluation model; input the evaluation sequence into the parameter evaluation model to obtain the evaluation label, and match the corresponding evaluation coefficient according to the evaluation label; wherein the evaluation label is set to a positive integer.

[0036] The present invention utilizes a parameter evaluation model to evaluate the operating state of the sintering furnace and obtains an evaluation coefficient, which provides a basis for subsequent adjustment of the operating parameters of the sintering furnace. When the operating state of the sintering furnace does not reach an ideal state, the operating parameters of the sintering furnace can be adjusted in time to avoid malfunctions of the sintering furnace due to poor operating conditions.

[0037] Preferably, the parameter evaluation model is constructed based on an artificial intelligence model, including:

[0038] Acquire standard training data, wherein the standard training data includes standard input data consistent with the evaluation sequence content attributes, and standard output data consistent with the evaluation label content attributes;

[0039] The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as a parameter evaluation model; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

[0040] The present invention trains the artificial intelligence model according to standard training data, and uses the trained artificial intelligence model to analyze the characteristic parameters of the sintering furnace, which can make the analysis result more accurate and is conducive to evaluating the operating status of the sintering furnace.

[0041] Preferably, the adjusting the operating state of the sintering furnace based on the evaluation coefficient includes:

[0042] Retrieve the evaluation coefficient and determine whether the evaluation coefficient is greater than the coefficient threshold; if yes, continue to detect the operating status of the sintering furnace; if no, gradually adjust the real-time status data of the sintering furnace;

[0043] When adjusting the real-time status data of the sintering furnace, the evaluation coefficient of the sintering furnace is continuously detected. When the evaluation coefficient is less than the coefficient threshold, the real-time status data of the sintering furnace is continuously adjusted until the evaluation coefficient is greater than the coefficient threshold.

[0044] The present invention adjusts the operating state of the sintering furnace and continuously detects the evaluation data of the sintering furnace, so that the sintering furnace can be in the optimal operating state, which is beneficial to maintaining the quality of produced goods and production efficiency.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. The present invention collects real-time status data of a sintering furnace, and pre-processes the collected real-time status data, and eliminates data with missing values ​​and abnormalities, which can ensure the integrity of subsequent analysis data and is beneficial to improving the efficiency of data collection; the processed real-time status data is converted into a tensor format; it can facilitate subsequent tensor decomposition of the tensor data, and is beneficial to the fusion of the sintering furnace data; the information of the sintering furnace is fused by using tensor decomposition, which is beneficial to more accurate analysis of the operation data of the sintering furnace.

[0047] 2. The present invention extracts characteristic parameters from the decomposed data and integrates the characteristic parameters into an evaluation sequence; uses a parameter evaluation model to analyze the evaluation sequence to obtain an evaluation coefficient; uses a trained artificial intelligence model to analyze the characteristic parameters of the sintering furnace, which can make the analysis result more accurate and is conducive to evaluating the operating status of the sintering furnace; adjusts the operating status of the sintering furnace and continuously detects the evaluation data of the sintering furnace, which can make the sintering furnace in the optimal operating state, which is conducive to maintaining the quality of the products produced and improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 It is a schematic diagram of the specific step framework of the present invention;

[0050] Figure 2 It is a schematic diagram of the specific process steps of the present invention;

[0051] Figure 3 It is a schematic diagram of the steps of decomposing the sintering furnace data tensor according to the present invention;

[0052] Figure 4 It is a schematic diagram of the steps of evaluating and adjusting the operating status of a sintering furnace according to the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] See also Figure 1-Figure 2The first aspect of the present invention provides a sintering furnace operation control method based on tensor decomposition, comprising:

[0055] Step S1: collecting real-time status data of the sintering furnace through a number of data sensors; preprocessing the collected real-time status data, converting the processed real-time status data into a tensor format, and obtaining tensor data; wherein the real-time status data includes real-time temperature, real-time pressure and real-time atmosphere; the data sensors are arranged according to the structure of the sintering furnace;

[0056] Step S2: decomposing the tensor data using a tensor decomposition algorithm to obtain decomposition data, and extracting the characteristics of the sintering furnace based on the decomposition data to obtain characteristic parameters;

[0057] Step S3: Integrate the characteristic parameters into an evaluation sequence, and call a parameter evaluation model based on the evaluation sequence to evaluate the operating state of the sintering furnace to obtain an evaluation coefficient; wherein the parameter evaluation model is constructed based on an artificial intelligence model;

[0058] Step S4: adjusting the operating state of the sintering furnace based on the evaluation coefficient, and continuously monitoring the operating state of the sintering furnace.

[0059] See also Figure 3 , collect the real-time status data of the sintering furnace through several data sensors, judge whether several real-time status data contain sub-data of all data types, if yes, keep the group of real-time status data; if no, remove the group of real-time status data; judge whether there are abnormal difference data in several groups of real-time status data; if yes, remove the corresponding real-time status data; if no, keep the corresponding real-time status data; determine the tensor dimension according to the dimension of the real-time status data, and create an empty tensor object by using the programming tool; fill the real-time status data into the tensor object according to the data type to obtain the tensor data; retrieve the tensor data and mark it as x, and decompose the tensor into a third-order tensor according to the quantity type of the real-time status data. Among them, the size of tensor x is n 1 ×n 2 ×n 3 ; The size is r 1 × 2 × 3 ; The size of the matrix U is n 1 × 1 ; The size of the matrix V is n 2 × 2 ; The size of the matrix W is n 3 × 3 ; Simplify the tensor expression to get the simplified tensor The least squares method is applied alternately for each iteration and each matrix factor as follows:

[0060] S1: Compute the mode-n horizontal expansion X(n) of the tensor x, n=1,…N;

[0061] S2: Calculate X(n) = U∑V T , n = 1, ... N, determine its effective rank R n , let U (n) ←U(:,1:R n );

[0062] S3: Compute

[0063] S4: When the core tensor When U (1) , U (2) , …U (n) ; Otherwise, loop S2-S3.

[0064] For example, there are three sets of data as follows:

[0065] Data set 1: temperature = 1200.5°C, pressure = 1.2 atm, atmosphere = N2; Data set 2: temperature = 1199.8°C, pressure = 1.21 atm, atmosphere = H2; Data set 3: temperature = 1300.0°C, pressure = 1.1 atm;

[0066] Data group 1 and data group 2 contain complete real-time temperature, real-time pressure and real-time atmosphere data, so they are retained; data group 3 lacks real-time atmosphere data, so it should be eliminated;

[0067] Assuming that the number of data sets we retain is 2, the dimension of the tensor is (2, 3), where 2 is the number of data sets and 3 is the number of dimensions of each data set (temperature, pressure, atmosphere); output is a tensor object with a shape of (2, 3).

[0068] See also Figure 4 , retrieve the decomposed data, extract the change trend, periodicity and stability of the real-time status data from the decomposed data; integrate the change trend, periodicity and stability into characteristic parameters; integrate the characteristic parameters into an evaluation sequence, and call the parameter evaluation model; input the evaluation sequence into the parameter evaluation model to obtain the evaluation label, and match the corresponding evaluation coefficient according to the evaluation label; wherein the evaluation label is set to a positive integer; judge whether the evaluation coefficient is greater than the coefficient threshold; if yes, continue to detect the operating status of the sintering furnace; if not, gradually adjust the real-time status data of the sintering furnace; when adjusting the real-time status data of the sintering furnace, continue to detect the evaluation coefficient of the sintering furnace, and when the evaluation coefficient is less than the coefficient threshold, continue to adjust the real-time status data of the sintering furnace until the evaluation coefficient is greater than the coefficient threshold.

[0069] It should be noted that the steps for training the parameter evaluation model are as follows:

[0070] Acquire standard training data, wherein the standard training data includes standard input data consistent with the evaluation sequence content attributes, and standard output data consistent with the evaluation label content attributes;

[0071] The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as a parameter evaluation model; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

[0072] For example: extract the changing trend, periodicity and stability of real-time status data:

[0073] Trend of change: The trend of data over time can be evaluated by calculating the slope, growth rate or moving average of the data.

[0074] Periodicity: Periodic patterns in data can be identified through Fourier transforms, autocorrelation functions, or periodicity detection algorithms.

[0075] Stability: The stability of the data can be assessed by calculating its variance, standard deviation, or coefficient of variation.

[0076] Based on the integrated feature parameters, an evaluation sequence is created and a pre-trained parameter evaluation model is called. After the evaluation sequence is input into the parameter evaluation model, the model outputs an evaluation label and the corresponding evaluation coefficient.

[0077] The evaluation coefficient is compared with a preset coefficient threshold. If the evaluation coefficient is greater than the coefficient threshold, it means that the real-time status data of the sintering furnace is within an acceptable range; otherwise, adjustment is required;

[0078] After adjusting the real-time status data of the sintering furnace, it is necessary to continuously monitor and evaluate it. This can be achieved by regularly collecting and analyzing the real-time status data. If it is found that the evaluation coefficient is lower than the coefficient threshold again, it means that the adjustment effect is not ideal or there are problems, and further adjustments and optimizations are needed.

[0079] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0080] The working principle of the present invention is as follows: the present invention collects the real-time status data of the sintering furnace through a number of data sensors, determines whether the real-time status data contain sub-data of all data types, if yes, retains the group of real-time status data; if no, removes the group of real-time status data; determines whether there are abnormal difference data in the groups of real-time status data; if yes, removes the corresponding real-time status data; if no, retains the corresponding real-time status data; determines the tensor dimension according to the dimension of the real-time status data, and creates an empty tensor object by using a programming tool; fills the real-time status data into the tensor object according to the data type to obtain tensor data; decomposes the tensor data to obtain decomposed data;

[0081] The present invention extracts the change trend, periodicity and stability of real-time status data from the decomposed data; integrates the change trend, periodicity and stability into characteristic parameters; integrates the characteristic parameters into an evaluation sequence, and calls a parameter evaluation model; inputs the evaluation sequence into the parameter evaluation model to obtain an evaluation label, and matches the corresponding evaluation coefficient according to the evaluation label; wherein the evaluation label is set to a positive integer; judges whether the evaluation coefficient is greater than a coefficient threshold; if yes, continuously detects the operating status of the sintering furnace; if no, gradually adjusts the real-time status data of the sintering furnace; when adjusting the real-time status data of the sintering furnace, continuously detects the evaluation coefficient of the sintering furnace, and when the evaluation coefficient is less than the coefficient threshold, continuously adjusts the real-time status data of the sintering furnace until the evaluation coefficient is greater than the coefficient threshold.

[0082] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A sintering furnace operation control method based on tensor decomposition, characterized in that: include: Step S1: collecting real-time status data of the sintering furnace through a number of data sensors; Preprocess the collected real-time status data; The processed real-time state data is converted into a tensor format to obtain tensor data; wherein the real-time state data includes real-time temperature, real-time pressure and real-time atmosphere; the data sensor is set according to the structure of the sintering furnace; Step S2: decomposing the tensor data using a tensor decomposition algorithm to obtain decomposition data, and extracting the characteristics of the sintering furnace based on the decomposition data to obtain characteristic parameters; Step S3: Integrate the characteristic parameters into an evaluation sequence, and call a parameter evaluation model based on the evaluation sequence to evaluate the operating state of the sintering furnace to obtain an evaluation coefficient; wherein the parameter evaluation model is constructed based on an artificial intelligence model; Step S4: adjusting the operating state of the sintering furnace based on the evaluation coefficient, and continuously monitoring the operating state of the sintering furnace.

2. The sintering furnace operation control method based on tensor decomposition according to claim 1 is characterized in that: The data sensor is arranged according to the structure of the sintering furnace, including: Determine the key parameters that need to be measured according to the requirements of the sintering process, as well as the key points at which the corresponding key parameters are measured in the sintering furnace; obtain the data sensor's collection range and the internal space of the sintering furnace, and determine the number of data sensors to be set; spread the data sensors evenly around the key points; among them, the key parameters refer to the real-time status data, and the key points refer to the points where the key parameters fluctuate the most.

3. The sintering furnace operation control method based on tensor decomposition according to claim 1, characterized in that: The preprocessing of the collected real-time status data includes: Retrieving a number of collected real-time status data, determining whether the number of real-time status data contains sub-data of all data types, if yes, retaining the group of real-time status data; if no, discarding the group of real-time status data; Determine whether there are abnormal difference data in several groups of real-time status data; if yes, remove the corresponding real-time status data; if no, retain the corresponding real-time status data.

4. The sintering furnace operation control method based on tensor decomposition according to claim 3 is characterized in that: The step of converting the processed real-time status data into a tensor format includes: The tensor dimension is determined according to the dimension of the real-time status data, and an empty tensor object is created by using a programming tool; the real-time status data is filled into the tensor object according to the data type to obtain tensor data; wherein the programming tool includes NumPy or TensorFlow in Python.

5. The sintering furnace operation control method based on tensor decomposition according to claim 4 is characterized in that: Decomposing the tensor data using a tensor decomposition algorithm includes: Retrieve the tensor data and mark it as x, and decompose the tensor into three-order tensors according to the quantity type of real-time status data Among them, the size of tensor x is n1×n2×n3; The size of is r1×r2×r3; the size of matrix U is n1×r1; the size of matrix V is n2×r2; the size of matrix W is n3×r3; Simplify the tensor expression to get the simplified tensor The least squares method is used alternately for each iteration and each matrix factor.

6. The sintering furnace operation control method based on tensor decomposition according to claim 5 is characterized in that: The least square method is used alternately for each iteration and each matrix factor, including: S1: Compute the mode-n horizontal expansion X(n) of the tensor x, n=1,…N; S2: Calculate X(n) = U∑V T , n = 1, ... N, determine its effective rank R n , let U (n) ←U(:,1:R n ); S3: Compute S4: When the core tensor When U (1) , U (2) , …U (n) ; Otherwise, loop S2-S3.

7. The sintering furnace operation control method based on tensor decomposition according to claim 1 is characterized in that: The extracting of the features of the sintering furnace based on the decomposition data includes: Retrieve the decomposed data, extract the change trend, periodicity and stability of the real-time status data from the decomposed data; integrate the change trend, periodicity and stability into characteristic parameters.

8. The sintering furnace operation control method based on tensor decomposition according to claim 7 is characterized in that: The step of evaluating the operating state of the sintering furnace by calling the parameter evaluation model based on the evaluation sequence includes: Extract the evaluation sequence and call the parameter evaluation model; input the evaluation sequence into the parameter evaluation model to obtain the evaluation label, and match the corresponding evaluation coefficient according to the evaluation label; wherein the evaluation label is set to a positive integer.

9. The sintering furnace operation control method based on tensor decomposition according to claim 8, characterized in that: The parameter evaluation model is constructed based on an artificial intelligence model and includes: Acquire standard training data, wherein the standard training data includes standard input data consistent with the evaluation sequence content attributes, and standard output data consistent with the evaluation label content attributes; The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as a parameter evaluation model; wherein the artificial intelligence model includes a convolutional neural network model or a BP neural network model.

10. The sintering furnace operation control method based on tensor decomposition according to claim 1, characterized in that: The step of adjusting the operating state of the sintering furnace based on the evaluation coefficient includes: Retrieve the evaluation coefficient and determine whether the evaluation coefficient is greater than the coefficient threshold; if yes, continue to detect the operating status of the sintering furnace; if no, gradually adjust the real-time status data of the sintering furnace; When adjusting the real-time status data of the sintering furnace, the evaluation coefficient of the sintering furnace is continuously detected. When the evaluation coefficient is less than the coefficient threshold, the real-time status data of the sintering furnace is continuously adjusted until the evaluation coefficient is greater than the coefficient threshold.

Images