Visual melt temperature measurement method

Abstract

The invention relates to an apparent melt temperature measurement method, which comprises the following steps of: selecting natural gas, oxygen and nitrogen as influence factor parameters to obtain melt radiation measurement temperature and standard measurement temperature sample data at a plurality of moments in a circulation period in a copper smelting furnace body; performing sample data fitting training on the radiation measurement temperature by taking the standard measurement temperature as a reference, and constructing a temperature adjustment model of the melt in the atmosphere of natural gas, oxygen and nitrogen; and the melt radiation measurement temperature at each moment in the cycle period is obtained in real time, adjustment is performed through the temperature adjustment model, the measurement output temperature of the melt is obtained, and the effect of continuously obtaining the accurate measurement temperature is achieved.

Description

Technical Field

[0001] This invention belongs to the field of non-ferrous metal smelting, and particularly relates to a method for measuring the temperature of molten metal by observing its appearance. Background Technology

[0002] In the field of non-ferrous metal smelting, the measurement and control of the temperature of the molten metal in the furnace has a crucial impact on optimizing smelting processes and saving energy. Traditional furnace molten metal temperature measurement is divided into contact temperature measurement technology and non-contact temperature measurement technology. Thermocouple contact temperature measurement has high accuracy, but it can only be used for single measurements; otherwise, the high temperature of the molten metal will quickly damage the thermocouple sensor, and manual operation of inserting and removing the thermocouple is required, which has a high risk factor. Temperature measurement using thermal radiation methods such as infrared does not directly contact the molten metal and allows for continuous measurement, but it is easily affected by various factors and is difficult to adapt to the harsh working environment of industrial production.

[0003] For example, in the copper smelting industry, the temperature of molten copper in the anode furnace is usually measured by taking a sample from the copper outlet after the molten copper or slag has been poured out. This method provides outdated temperature data for the molten liquid inside the furnace, which cannot meet the demands of today's high-quality production. The temperature inside the anode furnace can reach 1200℃-1250℃, and existing temperature measuring equipment cannot perform continuous temperature measurements at this temperature. Therefore, currently, observation holes are opened on the anode furnace body during copper smelting. When the molten copper level is below the observation hole, an infrared thermometer is used to measure the temperature inside the furnace from the outside through the observation hole. However, because the area above the molten copper surface is filled with flowing air, natural gas, etc., the accuracy of the temperature measurement cannot be guaranteed. During the feeding, oxidation, reduction, and casting stages, various intermediate interfering media such as gases, dust, and fumes in the optical path attenuate the radiation energy of the target object through absorption, scattering, and reflection. The degree of influence depends on the content of the interfering media and the spectral response characteristics of the infrared radiation thermometer. Furthermore, during the casting stage, the furnace tilt angle and melt level may also affect the temperature measurement results. In existing technologies, there is no quantitative analysis and evaluation of the various factors causing interference; the impact of interference factors in the optical path on measurement accuracy is dynamic and highly uncertain. Summary of the Invention

[0004] This invention provides a method for measuring the temperature of flammable melts, which achieves continuous and accurate temperature measurement by establishing, training, and using a large temperature measurement model.

[0005] A method for measuring melt temperature using flame-induced radiation is proposed. Natural gas, oxygen, and nitrogen are selected as influencing factors. Sample data of melt radiation measurement temperature and standard measurement temperature are obtained at multiple moments within one cycle in a copper smelting furnace. The sample data are used to fit and train the radiation measurement temperature based on the standard measurement temperature to construct a temperature adjustment model for the melt under natural gas, oxygen, and nitrogen atmospheres. The melt radiation measurement temperature is acquired in real time at each moment within the cycle, and adjusted through the temperature adjustment model to obtain the measured output temperature of the melt.

[0006] Preferably, a dual-color thermometry method is used to obtain the radiation measurement temperature. Two characteristic wavelengths λ1 and λ2 are selected, and their emissivity R1 and R2 are measured respectively. The ratio of the emissivity or the logarithm of the ratio is used as the adjustment parameter.

[0007] Preferably, a filtering algorithm is used to reduce temperature fluctuations in radiation measurements.

[0008] Preferably, the copper smelting furnace body is an anode furnace, and the melt is the melt inside the anode furnace.

[0009] Preferably, the endpoint of the oxidation-reduction process in the anode furnace is used as the anchor point for training the sample data fitting within the cycle.

[0010] Preferably, the influencing factors also include sulfur dioxide, dust, furnace tilt angle, liquid level and / or the distance from the melt to the radiation temperature probe.

[0011] Preferably, the redox endpoint of the anode furnace is determined by correlating the sulfur dioxide concentration.

[0012] Preferably, a filtering algorithm is used to reduce temperature fluctuations in the radiation measurement temperature within the cycle. The filtering algorithm divides the rate of change of the radiation measurement temperature into two or more stages, and establishes a filtering fitting model and temperature compensation for each stage.

[0013] Preferably, a temperature adjustment model with multiple cycles is obtained, and supervised learning, semi-supervised learning, deep learning, neural network algorithms or artificial intelligence are used to search for the optimal values ​​of each parameter of the temperature adjustment model to improve the accuracy of the measured output temperature.

[0014] Preferably, the melt temperature measured by a thermocouple is used as the standard measurement temperature.

[0015] This invention achieves continuous measurement of melt temperature by establishing and using a large-scale temperature measurement model. It replaces the method of estimating melt temperature by manually observing the color of the melt inside the furnace, thus realizing the effect of visually monitoring the melt. Detailed Implementation

[0016] Example: In the anode furnace section of copper smelting, a temperature measuring device is installed and fixed in contact with or without contact with the anode furnace body. The radiation temperature probe maintains a distance from the molten material inside the furnace to collect information about the melt. During the smelting process, feeding, oxidation, reduction, and casting form a cycle, with each stage proceeding sequentially and repeatedly. The liquid level, natural gas, nitrogen, oxygen, sulfur dioxide, and dust content of the molten material inside the anode furnace are constantly changing. Especially when the temperature measuring device is inserted at an angle downwards or installed on the furnace body, the distance between the molten material and the temperature probe changes constantly with the different inclination angles of the anode furnace body, and the collected information about the molten material inside the furnace also changes.

[0017] Model building and training. Parameters such as natural gas, nitrogen, oxygen, sulfur dioxide, flue gas, furnace inclination angle, liquid level, and distance from the melt to the temperature probe are selected as influence coefficients [θ1, θ2, θ3, θ4, ..., θ]. n ], various influencing factors θ n Set weight K i Other influencing factors in the industrial production process are represented by unknown quantities X. j This indicates that factors such as smoke and dust that are difficult to measure accurately, the distance between the melt and the temperature probe, or other unknown factors are weighted by K. j .

[0018] The radiation thermometer is a dual-color thermometer, selecting two characteristic wavelengths λ1 and λ2 to measure their emissivity R1 and R2 respectively. The ratio or logarithm of the emissivity is used as an adjustment parameter. When two wavelengths λ1 and λ2 are selected, their corresponding radiance can be denoted as L1 and L2. Considering the emissivity of the target object... The difference is such that the two-color ratio is defined as R as:

[0019]

[0020] Taking the logarithm of both sides of the comparison value, we have

[0021] +

[0022] Define the emissivity ratio compensation term:

[0023]

[0024] Summarized as follows:

[0025]

[0026] Substituting the Planck function into the right side of the above formula, the target temperature T can be numerically solved.

[0027] Parameter description: To avoid absorption peaks in the two selected infrared bands; Emissivity; R: Luminance ratio; Emissivity to logarithmic difference; T: temperature after instrument dichromatography.

[0028] In industrial applications, the near-infrared region can be selected, such as... wait.

[0029] Simplification of Planck's function (Wien approximation):

[0030]

[0031] Substituting the compensation ratio R, we obtain the temperature inversion formula.

[0032]

[0033] Temperature measurement of the radiation thermometer A filtering algorithm is employed to reduce temperature value fluctuations and improve temperature accuracy. The algorithm establishes different filtering fitting models at different stages of the heat source temperature change rate, and different temperature iteration compensation models at different stages of the heat source temperature change, thereby improving the precise positioning of the heat source temperature.

[0034] One approach is to use a moving average to smooth out fluctuations, specifically:

[0035]

[0036] Wherein, N: the length of the sliding window, preferably 3 to 5;

[0037] The original temperature value at time t;

[0038]

[0039]

[0040]

[0041] ,

[0042] P: Error covariance, initialized to a constant.

[0043] Q: Process noise,

[0044] R: Measurement noise.

[0045] Before and after the redox endpoint, the filtering algorithm can divide the rate of temperature change in radiation measurement into two stages, with a separate filtering fitting model and temperature compensation for each stage. Alternatively, it can be divided into three, four, or more stages based on feeding, oxidation, reduction, and casting, with a separate filtering fitting model and temperature compensation for each stage. Furthermore, based on the temperature change in radiation measurement, the rate of temperature change can be divided into multiple stages, with a separate filtering fitting model and temperature compensation for each stage.

[0046] Using feeding, oxidation, reduction, and casting as stages within a cycle, the radiation temperature measurement device collects temperatures at different times within the cycle. Temperature measurement with thermocouple Temperature is measured using thermocouples as the standard and by utilizing a radiation thermometer. Thermocouple for measuring temperature The model is trained, and the trained model is used to compensate for or adjust the output temperature.

[0047] Supervised learning, semi-supervised learning, deep learning, neural network algorithms, or artificial intelligence are used to search for the optimal values ​​of each parameter until the required accuracy range is reached.

[0048]

[0049] ,

[0050] ,

[0051] ,

[0052]

[0053] Bias term (constant)

[0054] ,

[0055] ,

[0056] The SO2 compensation coefficients trained by the model

[0057] ,

[0058] ;

[0059] For coefficient θ i θ j and weight K i K j It satisfies normalization.

[0060] One approach uses the endpoint of the redox reaction in the anode furnace as the anchor point for fitting and training the sample data within the cycle. Furthermore, it correlates the redox endpoint determination of the anode furnace with sulfur dioxide concentration. The invention patent CN111291501B, "Intelligent Endpoint Determination System for Redox Reduction in Copper Smelting Anode Furnaces," discloses an implementation method that correlates the redox endpoint determination of the anode furnace with sulfur dioxide concentration.

[0061] By using data-driven learning, the model parameters are optimized to compensate for the temperature difference. The objective function is to minimize the error relative to the true value, which is the minimum mean square error.

[0062]

[0063]

[0064] Simultaneously add parameter normalization constraints:

[0065]

[0066] Alternatively, soft constraints can be implemented using L1 / L2 regularization terms:

[0067]

[0068] Collect sample data

[0069] Time t <![CDATA[Thermocouple temperature T s > <![CDATA[Thermometer temperature T n > <![CDATA[SO2 concentration]]> Gas flow rate <![CDATA[t1]]> 1200℃ 1180℃ 2.3% 1002.3 Nm³ / h … … … … …

[0070] calculate

[0071] Use linear regression to fit the following form:

[0072]

[0073] Training using the standard gradient descent algorithm And after each update Forced normalization.

[0074] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for measuring the temperature of a flaming melt, characterized in that: Natural gas, oxygen, and nitrogen were selected as influencing parameters. Sample data of the radiative temperature and standard temperature of the melt were obtained at multiple moments in one cycle of the copper smelting furnace. The radiative temperature was fitted and trained based on the standard temperature to construct a temperature adjustment model of the melt under the atmosphere of natural gas, oxygen, and nitrogen. The melt radiation measurement temperature is acquired in real time at each moment within the cycle, and adjusted through a temperature adjustment model to obtain the measured output temperature of the melt.

2. The method for measuring the temperature of a flaming melt according to claim 1, characterized in that: The radiation measurement temperature is obtained by using a two-color thermometry method. Two characteristic wavelengths λ1 and λ2 are selected, and their emissivity R1 and R2 are measured respectively. The ratio of the emissivity or the logarithm of the ratio is used as the adjustment parameter.

3. The method for measuring the temperature of a flaming melt according to claim 1, characterized in that: A filtering algorithm is used to reduce temperature fluctuations in radiation measurements.

4. The method for measuring the temperature of a flaming melt according to claim 1, characterized in that: The copper smelting furnace body is an anode furnace, and the melt is the melt inside the anode furnace.

5. The method for measuring the temperature of a flaming melt according to claim 4, characterized in that: The endpoint of the oxidation-reduction process in the anode furnace is used as the anchor point for training the sample data fitting within the cycle.

6. The method for measuring the temperature of a flaming melt according to claim 4, characterized in that: Other influencing parameters include sulfur dioxide, dust, furnace tilt angle, liquid level and / or the distance from the melt to the radiation temperature probe.

7. The method for measuring the temperature of a flaming melt according to claim 6, characterized in that: The sulfur dioxide concentration is used to determine the redox endpoint of the anode furnace.

8. The method for measuring the temperature of a flaming melt according to claim 4, characterized in that: Within the cycle, a filtering algorithm is used to reduce temperature fluctuations in the radiation measurement temperature. The filtering algorithm divides the rate of change of the radiation measurement temperature into two or more stages, and establishes a filtering fitting model and temperature compensation for each stage.

9. The method for measuring the temperature of a flaming melt according to claim 1, characterized in that: A temperature adjustment model with multiple cycles is obtained. Supervised learning, semi-supervised learning, deep learning, neural network algorithms or artificial intelligence are used to search for the optimal values ​​of each parameter of the temperature adjustment model, thereby improving the accuracy of the measured output temperature.

10. The method for measuring the temperature of a flame-induced melt according to claim 1, characterized in that: The melt temperature measured by a thermocouple is used as the standard measurement temperature.

Images