Industrial Silicon Smelting Process Prediction Method and System Based on Data Analysis

Abstract

The present invention discloses a prediction method and system for the industrial silicon smelting process based on data analysis, which relates to the technical field of industrial silicon smelting management and control; analyzes the physical structure of the smelting furnace, locates the thermocouple nodes as temperature mapping points in the virtual three-dimensional model, and establishes a spatial topological relationship; collects historical smelting data, aligns the temperature monitoring parameters and process operation parameters according to time nodes, and constructs a smelting parameter group with spatio-temporal correlation; generates a three-dimensional temperature representation based on the temperature mapping point data, classifies the temperature representation through temperature change curve fitting and heat conduction correlation analysis; matches the current process parameters with the reference historical process operation parameter sequence library in real time, and outputs the three-dimensional temperature field prediction result corresponding to the optimal historical parameter group; the present invention effectively solves the problems of low efficiency and large error in traditional manual analysis, and provides a reliable technical support for the intelligent control of industrial silicon smelting.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial silicon smelting control, and in particular to a method and system for predicting the industrial silicon smelting process based on data analysis. Background Art

[0002] During the industrial silicon smelting process, the real-time and accurate prediction of the three-dimensional temperature field in the furnace is the core requirement for optimizing process control, reducing energy consumption, and ensuring production safety. Currently, the industry generally uses thermocouple arrays to monitor the temperature at fixed points in the furnace, and relies on operators to manually analyze the temperature distribution trend by combining historical data and empirical rules. However, the manual analysis mode has significant defects: on the one hand, limited by the sparsity of thermocouple layout (usually only covering 15%-20% of the key areas of the furnace body), the temperature in the unmonitored area needs to be estimated by linear interpolation or empirical formulas, resulting in low spatial resolution (the local area prediction error exceeds ±80°C), and it is difficult to capture the transient temperature anomalies in key areas such as around the electrodes and at the junction of furnace charges (such as sudden overheating or cooling) in a timely manner; on the other hand, manual analysis requires integrating multiple sets of discrete data and relying on the subjective experience of the operator. When facing dynamic working conditions such as raw material composition fluctuations and current phase adjustments, the analysis takes up to 15-30 minutes, seriously lagging behind the actual temperature change rate in the furnace (the local temperature change can reach 5-10°C / s), and easily leading to the risk of process out of control. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for predicting the industrial silicon smelting process that can predict the change of the temperature field in the smelting furnace.

[0004] The present invention discloses a method for predicting the industrial silicon smelting process based on data analysis, including:

[0005] Analyze the structure of the smelting furnace, construct a virtual three-dimensional smelting furnace, and set temperature mapping points in the virtual three-dimensional smelting furnace based on the position nodes to which each thermocouple belongs;

[0006] Obtain a number of historical smelting process data, determine the historical temperature monitoring parameters and historical process operation parameters at different time nodes, and construct a historical temperature monitoring parameter sequence and a historical process operation parameter sequence. The combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data is recorded as a historical smelting parameter group;

[0007] Based on the historical temperature monitoring parameters in the historical smelting parameter group, construct the temperature manifestation state of the virtual three-dimensional smelting furnace;

[0008] Classify the temperature manifestation states of the virtual three-dimensional smelting furnace to obtain several virtual three-dimensional smelting furnace manifestation groups. Compare the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace manifestation group to determine the parameter mapping intervals of the corresponding process factor parameters, and configure the parameter mapping intervals under the corresponding process factor parameters in the historical process operation parameter sequences. Construct a reference historical process operation parameter sequence library from several historical process operation parameter sequences;

[0009] Use the reference historical process operation parameter sequence library to compare and analyze the current process operation parameters of the smelting furnace, determine the most suitable historical process operation parameter sequence, and recognize the manifestation of the virtual three-dimensional smelting furnace corresponding to the historical process operation parameter sequence as the prediction reference manifestation.

[0010] In some embodiments disclosed by the present invention, the method for constructing the temperature manifestation state of the virtual three-dimensional smelting furnace includes:

[0011] Based on the historical temperature monitoring parameters in the historical smelting parameter group, determine the mapping point temperature change curve of each temperature mapping point in the virtual three-dimensional smelting furnace;

[0012] Based on the relative position characteristics between the temperature mapping points and the matching characteristics of the mapping point temperature change curves, connect the temperature mapping points to each other to obtain the in-furnace temperature manifestation line.

[0013] In some embodiments disclosed by the present invention, the method for classifying the temperature manifestation states of the virtual three-dimensional smelting furnace includes:

[0014] Compare the temperature manifestation states of the virtual three-dimensional smelting furnace with each other to determine the temperature equal manifestation degree. If the temperature equal manifestation degree is greater than or equal to the preset value, classify the corresponding temperature manifestation states of the virtual three-dimensional smelting furnace into one category to obtain several virtual three-dimensional smelting furnace manifestation groups.

[0015] In some embodiments disclosed by the present invention, the method for comparing the temperature manifestation states of the virtual three-dimensional smelting furnace with each other includes:

[0016] Align the temperature manifestation states between the virtual three-dimensional smelting furnaces in terms of time, compare the relative in-furnace temperature manifestation lines, and construct a first temperature equal manifestation operator based on the comparison results;

[0017] Compare each pair of relative mapping point temperature change curves, and correct the first temperature equal manifestation operator based on the comparison results to obtain the temperature equal manifestation degree;

[0018] Among them, the expression for calculating the temperature equal manifestation degree is:

[0019] ;

[0020] Among them, D is the temperature equivalent performance degree, is the temperature equivalent parameter of the in-furnace temperature performance line at the t-th time node, is the equivalent influence adjustment coefficient of the in-furnace temperature performance line, is the equivalent influence adjustment constant of the in-furnace temperature performance line, T is the total number of all time nodes participating in the comparison, is the temperature equivalent parameter of the temperature change curve of the h-th mapping point at the t-th time node, H is the total number of temperature change curves of the mapping points participating in the comparison, is the equivalent influence adjustment coefficient of the temperature change curve of the mapping point, is the equivalent influence adjustment constant of the temperature change curve of the mapping point.

[0021] In some embodiments disclosed by the present invention, the method for determining the temperature equivalent parameter of the in-furnace temperature performance line includes:

[0022] Determine the relative distance between mapping points between each pair of opposite temperature mapping points. If it is less than or equal to the preset value, it is determined that the corresponding temperature mapping points are equivalent, and record one mapping point equivalence. Based on the number of recorded mapping point equivalences, determine the temperature equivalent parameter of the in-furnace temperature performance line;

[0023] Among them, the expression for calculating the temperature equivalent parameter of each in-furnace temperature performance line is:

[0024] ;

[0025] Among them, is the temperature mapping point equivalence judgment function. If the x1-th temperature mapping point is recorded as a mapping point equivalence, then output 1, and y1 is the total number of temperature mapping points participating in the comparison;

[0026] The method for determining the temperature equivalent parameter of the temperature change curve of the mapping point includes:

[0027] Determine the time nodes for comparison on the temperature change curve, denoted as comparison time nodes. Centered on the comparison time nodes, intercept comparison curve segments on the temperature change curve according to the preset time length;

[0028] Compare the relative comparison curve segments. The comparison method includes uniformly setting a number of curve comparison points on the comparison curve segments, determining the vertical axis difference amount between the relative curve comparison points. If the vertical axis difference amount of the curve is less than or equal to the preset value, it is determined that the corresponding curve comparison points are equivalent curve comparison points, determine the equivalent comparison point ratio of the equivalent curve comparison points to all curve comparison points, and based on the comparison point ratio interval to which the equivalent comparison point ratio belongs, determine the corresponding temperature equivalent parameter .

[0029] In some embodiments disclosed by the present invention, the method for determining the parameter mapping intervals of different process factor parameters includes:

[0030] Construct a parameter mapping axis for each type of process factor parameter, and map the corresponding process factor parameters onto the parameter mapping axis in the form of factor parameter mapping points;

[0031] Determine the distribution of factor parameter mapping points on the parameter mapping axis, determine the distribution intervals of factor parameter mapping points with a distribution density greater than or equal to a preset value, and determine the parameter mapping intervals of the corresponding process factor parameters based on the lengths of the distribution intervals of factor parameter mapping points.

[0032] In some embodiments disclosed by the present invention, the method for comparing and analyzing the current process operation parameters of a smelting furnace by using a reference historical process operation parameter sequence library:

[0033] Compare the current process operation parameters in the order of time series with different reference historical process operation parameter sequences in the reference historical process operation parameter sequence library. The comparison method includes comparing the process factor parameters in the current process operation parameters with the parameter mapping intervals in the reference historical process operation parameter sequences. If the process factor parameters belong to the corresponding parameter mapping intervals, the corresponding process factor parameters are determined as matching process factor parameters;

[0034] Record in real time the number of matches of the matching process factor parameters of each parameter type. If the number of matches within a preset time period is greater than or equal to a preset value, determine that the parameter type is a matching parameter type. If each parameter type is a matching parameter type, determine that the corresponding reference historical process operation parameter sequence is the most suitable historical process operation parameter sequence.

[0035] In some embodiments disclosed by the present invention, an industrial silicon smelting process prediction system based on data analysis is further disclosed, which is characterized by including:

[0036] A first module for analyzing the structure of the smelting furnace, constructing a virtual three-dimensional smelting furnace, and setting temperature mapping points in the virtual three-dimensional smelting furnace based on the position nodes to which each thermocouple belongs;

[0037] A second module for obtaining a number of historical smelting process data, determining the historical temperature monitoring parameters and historical process operation parameters at different time nodes, constructing a historical temperature monitoring parameter sequence and a historical process operation parameter sequence, and recording the combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data as a historical smelting parameter group;

[0038] A third module, configured to construct a temperature representation state of a virtual three-dimensional smelting furnace based on historical temperature monitoring parameters in a historical smelting parameter group;

[0039] A fourth module, configured to classify the temperature representation state of the virtual three-dimensional smelting furnace to obtain several virtual three-dimensional smelting furnace representation groups, compare the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace representation group, determine a parameter mapping interval of the corresponding process factor parameter, and configure the parameter mapping interval under the corresponding process factor parameter in the historical process operation parameter sequence, and construct several historical process operation parameter sequences into a reference historical process operation parameter sequence library;

[0040] A fifth module, configured to perform a comparison and analysis on the current process operation parameters of the smelting furnace by using the reference historical process operation parameter sequence library, determine the most suitable historical process operation parameter sequence, and recognize the representation of the virtual three-dimensional smelting furnace corresponding to the historical process operation parameter sequence as a prediction reference representation.

[0041] The present invention discloses an industrial silicon smelting process prediction method and system based on data analysis, relating to the technical field of industrial silicon smelting management and control; analyzing the physical structure of a smelting furnace, positioning thermocouple nodes as temperature mapping points in a virtual three-dimensional model, and establishing a spatial topological relationship; collecting historical smelting data, aligning temperature monitoring parameters and process operation parameters according to time nodes, and constructing a smelting parameter group with spatio-temporal correlation; generating a three-dimensional temperature representation state based on temperature mapping point data, classifying the temperature representation state through temperature change curve fitting and heat conduction correlation analysis; matching current process parameters with a reference historical process operation parameter sequence library in real time, and outputting a three-dimensional temperature field prediction result corresponding to an optimal historical parameter group; the present invention effectively solves the problems of low efficiency and large error in traditional manual analysis, and provides reliable technical support for the intelligent control of industrial silicon smelting.

[0042] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Description of the Drawings

[0043] Figure 1 It is a method and method steps for predicting an industrial silicon smelting process based on data analysis disclosed in an embodiment of the present invention. Detailed Embodiments

[0044] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and should not be construed as limiting the protection scope of the present invention. Those skilled in the art can make some non-essential improvements and adjustments based on the content of the present invention below. In the present invention, unless otherwise clearly specified and limited, the technical terms used in the present invention should have the ordinary meaning understood by those skilled in the art of the present invention.

[0046] Embodiment:

[0047] The present invention discloses a prediction method for the industrial silicon smelting process based on data analysis. Refer to Figure 1 , including:

[0048] Step S100, analyze the structure of the smelting furnace, construct a virtual three-dimensional smelting furnace, and set temperature mapping points in the virtual three-dimensional smelting furnace based on the position nodes to which each thermocouple belongs.

[0049] The core of this step is to establish a three-dimensional model consistent with the actual smelting furnace structure through digital means and accurately map the positions of temperature monitoring points. First, based on the design drawings or actual measurement data of the smelting furnace (such as furnace body dimensions, electrode layout, refractory material distribution, etc.), key geometric parameters (such as hearth diameter, electrode insertion depth) are extracted, and a parametric three-dimensional model is constructed using computer-aided design (CAD) software (such as SolidWorks) to ensure that the model can dynamically simulate changes in the furnace body structure (such as electrode lifting). Subsequently, according to the actual installation positions of the thermocouples (such as layered arrangement on the furnace wall, annular distribution around the electrodes), the corresponding temperature mapping points are marked in the virtual model, and the spatial coordinates of each mapping point need to be strictly aligned with the actual positions, usually achieved through coordinate system conversion (such as from a local coordinate system to a global coordinate system). In the model verification stage, by comparing the actual temperature measurement data with the model simulation results (such as the temperature distribution under steady state), the mesh division or material property parameters are adjusted to ensure that the model error is controlled within a reasonable range (such as ±5°C). The technical significance of this step is to provide a reference framework for the spatial positioning of subsequent temperature data. For example, the influence of the molten pool depth on the thermal field distribution is reflected through temperature mapping points at different heights, laying a physical consistency foundation for multi-source data fusion.

[0050] Step S200, obtain a number of historical smelting process data, determine the historical temperature monitoring parameters and historical process operation parameters at different time nodes, and construct a historical temperature monitoring parameter sequence and a historical process operation parameter sequence. The combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data is denoted as a historical smelting parameter group.

[0051] The goal of this step is to extract structured data from historical smelting records and establish a time-series correlation between temperature and process parameters. The specific process includes: obtaining temperature monitoring data (such as thermocouple second-level sampling values) and process operation parameters (such as current, voltage, and feeding amount) from industrial control systems (such as DCS and SCADA), and at the same time integrating manual records (such as fault logs). Arrange the data of the same smelting batch in chronological order to form a temperature monitoring parameter sequence (such as T1(t), T2(t),..., Tn(t)) and a process operation parameter sequence (such as current I(t), voltage V(t)), and bind the two into a "historical smelting parameter group" to ensure that the temperature and process parameters at each time point correspond one by one. The technical significance of this step is to construct a highly credible structured data set. For example, through time alignment, the dynamic correlation between current fluctuations and temperature changes in a specific area can be analyzed, providing an input basis for subsequent pattern mining.

[0052] Step S300: Based on the historical temperature monitoring parameters in the historical smelting parameter group, construct the temperature representation state of the virtual three-dimensional smelting furnace.

[0053] This step aims to transform discrete temperature monitoring data into a characteristic representation reflecting the overall state of the temperature field in the furnace. The specific methods include: drawing a curve of temperature change over time for each temperature mapping point (such as the typical trends of temperature rise in the initial stage of smelting, stability in the middle stage, and cooling in the later stage), and calculating the morphological similarity of adjacent node curves through dynamic time warping (DTW) or Pearson correlation coefficient. If the curve similarity of adjacent nodes exceeds a threshold (such as the correlation coefficient > 0.8) and the spatial distance meets the requirements (such as less than 0.5 meters), then connect them as a "temperature representation line", representing the co-variation area of the local temperature field. For example, the dense connection of multiple representation lines in the molten pool area may reflect the uniformity of molten silicon. Finally, by integrating all representation lines, construct the dynamic change pattern of the three-dimensional temperature field. The technical significance of this step is to upgrade discrete point data to a continuous field representation. For example, through the representation line, hot spots or abnormal temperature gradients in the furnace can be identified, breaking through the limitations of traditional single-point analysis.

[0054] Step S400: Classify the temperature representation states of the virtual three-dimensional smelting furnace to obtain several virtual three-dimensional smelting furnace representation groups. Compare the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace representation group to determine the parameter mapping intervals of the corresponding process factor parameters, and configure the parameter mapping intervals under the corresponding process factor parameters in the historical process operation parameter sequences. Construct a reference historical process operation parameter sequence library from several historical process operation parameter sequences.

[0055] Classification of temperature representation states:

[0056] First, standardize the three-dimensional temperature performance states of different smelting batches to ensure that the time span is aligned with the smelting stage (for example, uniformly intercept the data of the first 2 hours of the smelting period). By calculating the similarity scores between the temperature performance states (combining indicators such as the consistency of spatial distribution and the matching degree of curve shapes), group the performance states with high similarity into the same category. For example, if the temperature gradient change trends in the molten pool area of two temperature fields are the same, and the overlap degree of the hot spot positions exceeds the set threshold, they are determined to be of the same category. The classification process uses an automated clustering algorithm (such as density-based clustering) to finally form several "virtual three-dimensional smelting furnace performance groups", and each group represents a typical temperature field evolution pattern.

[0057] Determination of the parameter mapping interval:

[0058] Conduct statistical analysis on the historical process parameters (such as current, voltage, feeding frequency) corresponding to each performance group, and extract their distribution laws. For example, if the current values of a certain performance group are mainly concentrated in the range of 2000 - 2200A, and the temperature field stability is relatively high within this range, then this range is defined as the "mapping interval" of the current parameter. When defining the interval boundaries, process constraint conditions (such as equipment safety thresholds) and data distribution density need to be considered to avoid the interval being too wide or too narrow due to extreme values. Finally, the mapping intervals of all process parameters are bound to the corresponding historical sequences to form a reference historical process operation parameter sequence library.

[0059] Step S500, use the reference historical process operation parameter sequence library to compare and analyze the current process operation parameters of the smelting furnace, determine the most suitable historical process operation parameter sequence, and recognize the performance of the virtual three-dimensional smelting furnace corresponding to the historical process operation parameter sequence as the prediction reference performance.

[0060] This step realizes the prediction of the smelting process through the matching of real-time data with the historical database. The specific logic is as follows: Real-time process parameters (such as current and voltage) are collected from the DCS system, smoothed according to a fixed time window (such as 10 seconds) to eliminate instantaneous noise. Subsequently, the current parameters are compared item by item with the mapping intervals of historical parameters. For example, it is judged whether the current current is within the range of 2000 - 2200A of a certain historical group, and the continuous matching times of each parameter type are counted. If a certain parameter falls within the same interval in N consecutive time windows (such as N = 5), it is marked as "stable matching"; only when all process parameters meet the stable matching conditions, the corresponding historical sequence is determined as "the most suitable". If multiple sequences meet the conditions simultaneously, the sequence with the highest D value is selected as the prediction reference. Finally, the temperature performance states (such as the distribution of the performance line and the position of the hot spot) corresponding to the most suitable sequence are used as the prediction reference for the current smelting process, such as warning of the risk of abnormal temperature rise in the electrode area. The technical significance of this step lies in realizing the rapid diagnosis of the process state through real-time data drive. For example, the matching results can directly guide the operator to adjust the current or feeding strategy, avoid temperature runaway, and improve the stability and energy efficiency of the smelting process.

[0061] In some embodiments disclosed by the present invention, the method for constructing the temperature performance state of the virtual three-dimensional smelting furnace includes:

[0062] Step S301, based on the historical temperature monitoring parameters in the historical smelting parameter group, determine the mapping point temperature change curve of each temperature mapping point in the virtual three-dimensional smelting furnace.

[0063] Based on the monitoring data of each temperature mapping point in the historical smelting parameter group, extract its temperature values in the entire smelting cycle in chronological order to form an independent time series curve. For example, the curve of a certain furnace wall mapping point can reflect the rapid temperature rise in the initial stage of smelting (0 - 30 minutes), the steady state maintenance in the middle stage (30 - 120 minutes), and the temperature drop in the later stage (120 - 180 minutes). After the curve is drawn, key feature points (such as the peak temperature and the mutation point of the heating rate) need to be marked for subsequent analysis of the regularity and abnormality of the local temperature behavior.

[0064] Step S302, based on the relative position characteristics between the temperature mapping points and the matching characteristics of the mapping point temperature change curves, connect the temperature mapping points to each other to obtain the temperature performance line in the furnace.

[0065] Based on the generation of single-point curves, combining the spatial position relationship of mapping points and the similarity of curve shapes, discrete nodes are connected into continuous temperature representation lines. First, calculate the physical distance between adjacent mapping points (for example, a distance less than 0.5 meters is defined as "spatially adjacent"), and screen out potential connected node pairs; secondly, compare the curve shapes of adjacent nodes (such as the consistency of fluctuation trends and the alignment degree of key feature points). If the similarity exceeds a threshold (such as a trend coincidence degree > 80%), it is determined as a "connectable node pair". Finally, a three-dimensional temperature representation line network is formed by cross-connecting the node pairs that meet the conditions. For example, multiple adjacent nodes in the molten pool area form a dense cluster of representation lines due to the synchronous temperature rise trend, which can intuitively reflect the spatial heat conduction path; while the radial distribution of the representation lines around the electrode may indicate problems of uneven current distribution.

[0066] In some embodiments disclosed by the present invention, the method for classifying the temperature representation states of a virtual three-dimensional smelting furnace includes:

[0067] Step S401: Compare the temperature representation states of the virtual three-dimensional smelting furnace with each other to determine the temperature equivalent representation degree. If the temperature equivalent representation degree is greater than or equal to a preset value, classify the corresponding temperature representation states of the virtual three-dimensional smelting furnace into one category to obtain several virtual three-dimensional smelting furnace representation groups.

[0068] In some embodiments disclosed by the present invention, the method for comparing the temperature representation states of the virtual three-dimensional smelting furnace with each other includes:

[0069] Step S402: Align the temperature representation states between the virtual three-dimensional smelting furnaces in terms of time, compare the relative in-furnace temperature representation lines, and based on the comparison results, construct a first temperature equivalent representation operator.

[0070] Step S403: Compare each pair of relative mapping point temperature change curves, and based on the comparison results, correct the first temperature equivalent representation operator to obtain the temperature equivalent representation degree.

[0071] Among them, the expression for calculating the temperature equivalent representation degree is:

[0072] .

[0073] Among them, D is the temperature equivalent representation degree, is the temperature equivalent parameter of the in-furnace temperature representation line at the t-th time node, is the in-furnace temperature representation line equivalent influence adjustment coefficient, is the in-furnace temperature representation line equivalent influence adjustment constant, T is the number of all time nodes participating in the comparison, is the temperature equivalent parameter of the temperature change curve of the h-th mapping point at the t-th time node, where H is the total number of temperature change curves of mapping points participating in the comparison. is the equivalent influence adjustment coefficient of the mapping point temperature change curve. is the equivalent influence adjustment constant of the mapping point temperature change curve.

[0074] In some embodiments disclosed by the present invention, the method for determining the temperature equivalent parameter of the in-furnace temperature performance line includes:

[0075] Step S4031: Determine the relative distance between mapping points of each pair of opposite temperature mapping points. If it is less than or equal to the preset value, it is considered that the corresponding temperature mapping points are equivalent, and record one instance of mapping point equivalence. Based on the number of recorded mapping point equivalences, determine the temperature equivalent parameter of the in-furnace temperature performance line.

[0076] Among them, the expression for calculating the temperature equivalent parameter of each in-furnace temperature performance line is:

[0077] .

[0078] Among them, is the mapping point equivalence judgment function. If the x1-th temperature mapping point is recorded as a mapping point equivalence, then output 1, where y1 is the total number of temperature mapping points participating in the comparison;

[0079] The method for determining the temperature equivalent parameter of the mapping point temperature change curve includes:

[0080] Determine the time nodes for comparison on the temperature change curve, denoted as comparison time nodes. Centered on the comparison time nodes, intercept comparison curve segments on the temperature change curve according to the preset time length.

[0081] Compare the opposite comparison curve segments. The comparison method includes uniformly setting several curve comparison points on the comparison curve segments, determining the vertical axis difference of the curves between the opposite curve comparison points. If the vertical axis difference of the curves is less than or equal to the preset value, it is considered that the corresponding curve comparison points are equivalent curve comparison points, determine the proportion of equivalent curve comparison points among all curve comparison points, and based on the comparison point proportion interval to which the proportion of equivalent comparison points belongs, determine the corresponding temperature equivalent parameter. .

[0082] In some embodiments disclosed by the present invention, the method for determining the parameter mapping interval of different process factor parameters includes:

[0083] Step S404: Construct a parameter mapping axis for each type of process factor parameter, and map the corresponding process factor parameters onto the parameter mapping axis in the form of factor parameter mapping points.

[0084] Step S405: Determine the distribution of factor parameter mapping points on the parameter mapping axis, determine the distribution interval of factor parameter mapping points with a distribution density greater than or equal to a preset value, and determine the parameter mapping interval of the corresponding process factor parameter based on the length of the distribution interval of factor parameter mapping points.

[0085] In some embodiments disclosed by the present invention, a method for comparing and analyzing the current process operation parameters of a smelting furnace by using a reference historical process operation parameter sequence library:

[0086] Step S501: Compare the current process operation parameters with different reference historical process operation parameter sequences in the reference historical process operation parameter sequence library in a time-sequence corresponding manner. The comparison method includes comparing the process factor parameters in the current process operation parameters with the parameter mapping intervals in the reference historical process operation parameter sequences. If the process factor parameter belongs to the corresponding parameter mapping interval, the corresponding process factor parameter is determined to be a matching process factor parameter.

[0087] Step S502: Record the matching times of the matching process factor parameters of each parameter type in real time. If the number of matching times within a preset time period is greater than or equal to a preset value, determine that the parameter type is a matching parameter type. If each parameter type is a matching parameter type, determine that the corresponding reference historical process operation parameter sequence is the most suitable historical process operation parameter sequence.

[0088] In some embodiments disclosed by the present invention, an industrial silicon smelting process prediction system based on data analysis is also disclosed, which is characterized by including:

[0089] The first module is used to analyze the structure of the smelting furnace, construct a virtual three-dimensional smelting furnace, and set temperature mapping points in the virtual three-dimensional smelting furnace based on the position nodes to which each thermocouple belongs.

[0090] The second module is used to obtain a number of historical smelting process data, determine the historical temperature monitoring parameters and historical process operation parameters at different time nodes, construct a historical temperature monitoring parameter sequence and a historical process operation parameter sequence, and record the combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data as a historical smelting parameter group.

[0091] The third module is used to construct the temperature performance state of the virtual three-dimensional smelting furnace based on the historical temperature monitoring parameters in the historical smelting parameter group.

[0092] The fourth module is used to classify the temperature manifestation states of the virtual three-dimensional smelting furnace, obtain several virtual three-dimensional smelting furnace manifestation groups, compare the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace manifestation group, determine the parameter mapping intervals of the corresponding process factor parameters, and configure the parameter mapping intervals under the corresponding process factor parameters in the historical process operation parameter sequences, and construct several historical process operation parameter sequences into a reference historical process operation parameter sequence library;

[0093] The fifth module is used to compare and analyze the current process operation parameters of the smelting furnace by using the reference historical process operation parameter sequence library, determine the most suitable historical process operation parameter sequence, and recognize the manifestation of the virtual three-dimensional smelting furnace corresponding to the historical process operation parameter sequence as the prediction reference manifestation.

[0094] The present invention discloses a prediction method and system for the industrial silicon smelting process based on data analysis, which relates to the technical field of industrial silicon smelting control; analyzes the physical structure of the smelting furnace, locates the thermocouple nodes in the virtual three-dimensional model as temperature mapping points, and establishes a spatial topological relationship; collects historical smelting data, aligns the temperature monitoring parameters and process operation parameters according to time nodes, and constructs a spatio-temporally correlated smelting parameter group; generates a three-dimensional temperature manifestation state based on the temperature mapping point data, classifies the temperature manifestation state through temperature change curve fitting and heat conduction correlation analysis; matches the current process parameters with the reference historical process operation parameter sequence library in real time, and outputs the prediction result of the three-dimensional temperature field corresponding to the optimal historical parameter group; the present invention effectively solves the problems of low efficiency and large error in traditional manual analysis, and provides a reliable technical support for the intelligent control of industrial silicon smelting.

[0095] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present invention.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting industrial silicon smelting process based on data analysis, characterized in that: include: Analyze the structure of the smelting furnace, build a virtual three-dimensional smelting furnace, and set the temperature mapping point in the virtual three-dimensional smelting furnace based on the location node of each thermocouple; Acquire a number of historical smelting process data, determine the historical temperature monitoring parameters and historical process operation parameters at different time nodes, and construct a historical temperature monitoring parameter sequence and a historical process operation parameter sequence, and record the combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data as a historical smelting parameter group; Based on the historical temperature monitoring parameters in the historical smelting parameter group, the temperature performance of the virtual three-dimensional smelting furnace is constructed; The temperature performance states of the virtual three-dimensional smelting furnace are classified to obtain a number of virtual three-dimensional smelting furnace performance groups, and the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace performance group are compared to determine the parameter mapping interval of the corresponding process factor parameter, and the parameter mapping interval is configured under the corresponding process factor parameter in the historical process operation parameter sequence, and the several historical process operation parameter sequences are constructed as a reference historical process operation parameter sequence library; The current process operating parameters of the smelting furnace are compared and analyzed using the reference historical process operating parameter sequence library to determine the most suitable historical process operating parameter sequence, and the performance of the virtual three-dimensional smelting furnace corresponding to the historical process operating parameter sequence is identified as the predicted reference performance.

2. The method for predicting industrial silicon smelting process based on data analysis according to claim 1, characterized in that: The method for constructing the temperature representation state of a virtual three-dimensional smelting furnace includes: Determine a mapping point temperature change curve of each temperature mapping point in the virtual three-dimensional smelting furnace based on the historical temperature monitoring parameters in the historical smelting parameter group; Based on the relative position characteristics between the temperature mapping points and the matching characteristics of the temperature change curves of the mapping points, the temperature mapping points are connected to each other to obtain the temperature performance line in the furnace.

3. The method for predicting industrial silicon smelting process based on data analysis according to claim 2, characterized in that: Methods for classifying the temperature performance of a virtual three-dimensional smelting furnace include: The temperature performance states of the virtual three-dimensional smelting furnaces are compared with each other to determine the degree of temperature equivalence. If the degree of temperature equivalence is greater than or equal to a preset value, the temperature performance states of the corresponding virtual three-dimensional smelting furnaces are classified into one category to obtain several virtual three-dimensional smelting furnace performance groups.

4. The method for predicting industrial silicon smelting process based on data analysis according to claim 3, characterized in that: The method for comparing the temperature performance of the virtual three-dimensional smelting furnace with each other includes: The temperature expression states between the virtual three-dimensional smelting furnaces are aligned in time, and the relative temperature expression lines in the furnaces are compared, and based on the comparison results, a first temperature equivalent expression operator is constructed; Comparing the temperature change curves of each relative mapping point, and based on the comparison result, correcting the first temperature equivalent expression operator to obtain the temperature equivalent expression degree; Among them, the expression for calculating the degree of temperature equivalent performance is: ； Where D is the temperature equivalent performance level, is the temperature equivalent parameter of the furnace temperature expression line at the t-th time node, The adjustment coefficient of the furnace temperature performance line is equivalent to the influence of the adjustment coefficient. is the adjustment constant for the equivalent impact of the furnace temperature performance line, T is the number of all time nodes involved in the comparison, is the temperature equivalent parameter of the temperature change curve of the hth mapping point at the tth time node, H is the total number of temperature change curves of the mapping points involved in the comparison, The temperature change curve of the mapping point is equivalent to the impact adjustment coefficient, Adjust the constants for the equivalent impact of the temperature change curve at the mapping point.

5. The method for predicting industrial silicon smelting process based on data analysis according to claim 4, characterized in that: Methods for determining temperature equivalent parameters of the temperature performance line in the furnace include: Determine the relative distance between each relative temperature mapping point. If the relative distance is less than or equal to a preset value, the corresponding temperature mapping points are deemed equal, and the mapping points are recorded equal once. Based on the recorded number of times the mapping points are equal, determine the temperature equalization parameter of the temperature expression line in the furnace. Among them, the expression for calculating the temperature equivalent parameter of each furnace temperature expression line is: ； in, is the temperature mapping point equality judgment function. If the x1th temperature mapping point is recorded as the mapping point equality, then Output 1, y1 is the total number of temperature mapping points involved in the comparison; The method for determining the temperature equivalent parameter of the temperature change curve of the mapping point includes: Determine a time node for comparison on the temperature change curve, record it as the comparison time node, and intercept a comparison curve segment on the temperature change curve according to a preset time length with the comparison time node as the center; Comparing the relative comparison curve segments, the comparison method includes evenly setting a number of curve comparison points on the comparison curve segment, determining the curve vertical axis difference between the relative curve comparison points, if the curve vertical axis difference is less than or equal to a preset value, then identifying the corresponding curve comparison point as an equivalent curve comparison point, determining the equivalent comparison point ratio of the equivalent curve comparison point to all the curve comparison points, and determining the corresponding temperature equivalent parameter based on the comparison point ratio interval to which the equivalent comparison point ratio belongs .

6. The method for predicting industrial silicon smelting process based on data analysis according to claim 1, characterized in that: Methods for determining parameter mapping intervals for different process factor parameters include: Construct a parameter mapping axis for each type of process factor parameter, and map the corresponding process factor parameter on the parameter mapping axis in the form of factor parameter mapping points; The factor parameter mapping point distribution of the parameter mapping axis is determined, a factor parameter mapping point distribution interval having a distribution density greater than or equal to a preset value is determined, and based on the length of the factor parameter mapping point distribution interval, a parameter mapping interval of the corresponding process factor parameter is determined.

7. The method for predicting industrial silicon smelting process based on data analysis according to claim 1, characterized in that: Method for comparing and analyzing the current process operation parameters of the smelting furnace by using the reference historical process operation parameter sequence library: Comparing the current process operation parameters with different reference historical process operation parameter sequences in the reference historical process operation parameter sequence library in a time-series corresponding manner, wherein the comparison method includes comparing the process factor parameters in the current process operation parameters with the parameter mapping intervals in the reference historical process operation parameter sequences, and if the process factor parameters belong to the corresponding parameter mapping intervals, then the corresponding process factor parameters are identified as matching process factor parameters; The matching number of the matching process factor parameters of each parameter type is recorded in real time. If the matching number within the preset time period is greater than or equal to the preset value, the parameter type is determined to be a matching parameter type. If each parameter type is a matching parameter type, the corresponding reference historical process operation parameter sequence is determined to be the most suitable historical process operation parameter sequence.

8. The industrial silicon smelting process prediction system based on data analysis is characterized by: include: The first module is used to analyze the structure of the smelting furnace, construct a virtual three-dimensional smelting furnace, and set the temperature mapping point in the virtual three-dimensional smelting furnace based on the position node of each thermocouple; The second module is used to obtain a number of historical smelting process data, determine the historical temperature monitoring parameters and historical process operation parameters at different time nodes, and construct a historical temperature monitoring parameter sequence and a historical process operation parameter sequence, and record the combination of the historical temperature monitoring parameter sequence and the historical process operation parameter sequence belonging to the same historical smelting process data as a historical smelting parameter group; The third module is used to construct the temperature performance state of the virtual three-dimensional smelting furnace based on the historical temperature monitoring parameters in the historical smelting parameter group; The fourth module is used to classify the temperature performance states of the virtual three-dimensional smelting furnace to obtain a number of virtual three-dimensional smelting furnace performance groups, compare the historical process operation parameter sequences corresponding to each virtual three-dimensional smelting furnace performance group, determine the parameter mapping interval of the corresponding process factor parameter, and configure the parameter mapping interval under the corresponding process factor parameter in the historical process operation parameter sequence, and construct a number of historical process operation parameter sequences as a reference historical process operation parameter sequence library; The fifth module is used to compare and analyze the current process operation parameters of the smelting furnace using the reference historical process operation parameter sequence library, determine the most suitable historical process operation parameter sequence, and identify the performance of the virtual three-dimensional smelting furnace corresponding to the historical process operation parameter sequence as the predicted reference performance.

Images