Pyrometallurgy copper temperature abnormity identification system and method based on machine learning and platinum-rhodium thermocouple

Abstract

The invention belongs to the technical field of copper pyrometallurgy, and particularly relates to a copper pyrometallurgy temperature anomaly recognition system and method based on machine learning and platinum-rhodium thermocouples, and the system comprises a platinum-rhodium thermocouple measurement module which is used for installing the platinum-rhodium thermocouples in a smelting furnace and measuring temperature data in the smelting process in real time; the data acquisition and transmission module is used for acquiring temperature data and transmitting the temperature data to the central processing system through a wireless communication technology; the machine learning temperature anomaly identification module is used for analyzing and processing the temperature data by utilizing a machine learning algorithm, establishing a temperature anomaly identification model and identifying abnormal temperature; and the early warning and control module is used for triggering an early warning mechanism and automatically adjusting the heating power and air volume parameters of the smelting furnace when the abnormal temperature is identified. By combining the high-precision temperature measurement technology of the platinum-rhodium thermocouple and the machine learning algorithm, real-time monitoring and accurate early warning of temperature abnormity in the copper pyrometallurgy process are achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of copper pyrometallurgy, and particularly relates to a copper pyrometallurgy temperature anomaly identification system and method based on machine learning and platinum-rhodium thermocouples. BACKGROUND

[0002] Copper pyrometallurgy is a high-temperature process that converts copper ore into copper, usually in a high-temperature furnace. Temperature control is a key factor in the smelting process, and too high or too low temperatures can affect the yield and quality of copper. Traditional temperature measurement methods mainly rely on manual detection or simple sensor monitoring, which is difficult to achieve real-time and accurate temperature anomaly identification for complex smelting processes.

[0003] Therefore, the prior art has defects and deficiencies, and needs to be further improved and developed.

[0004] In the prior art, temperature monitoring of copper pyrometallurgy mainly relies on manual detection or simple sensor monitoring. These methods have problems such as low measurement accuracy, inability to monitor in real time, poor anomaly identification ability, and lagging control means, leading to unstable smelting process, high energy consumption, and many safety hazards. Traditional sensors are easily disturbed in high-temperature and strongly corrosive smelting environments, manual detection has a long cycle and cannot monitor temperature changes in real time, and it is difficult to accurately identify temperature anomalies. Even if an anomaly is detected, it is not possible to quickly adjust the heating power, air volume, and other parameters of the smelting furnace. Manual detection and traditional sensor monitoring have significant shortcomings in temperature monitoring of copper pyrometallurgy. Manual detection has a long sampling period and cannot monitor temperature changes in real time, and has large errors, making it difficult to accurately reflect the true temperature (as shown in Figure 1 and Figure 2 ). Although traditional sensors can achieve continuous monitoring, they are affected by high temperatures and strong corrosive environments, have low measurement accuracy, and are subject to noise interference (as shown in Figure 2 ). In addition, manual detection and sensor monitoring perform poorly in anomaly identification. Manual detection is difficult to detect temperature anomalies in a timely manner due to low sampling frequency and large errors; sensor monitoring can identify anomalies, but is affected by noise and has unstable identification results (as shown in Figure 3 ). These problems lead to unstable smelting processes, high energy consumption, and many safety hazards, and there is an urgent need for more advanced monitoring technology to address the limitations of existing methods.

[0005] Therefore, the present application aims to solve the above-mentioned problems of temperature monitoring and control in the existing technology of copper pyrometallurgy. By using platinum rhodium thermocouples, high-precision and high-stability temperature measurement data are provided; through the data acquisition and wireless transmission module, real-time and continuous temperature data acquisition and transmission are realized; by using machine learning algorithms (such as long short-term memory network, LSTM), a temperature anomaly identification model is established to accurately identify temperature anomalies; through the early warning and control module, when temperature anomalies are identified, the heating power, air volume and other parameters of the smelting furnace are automatically adjusted to quickly correct abnormal conditions. The present application can significantly improve the temperature control precision and abnormal identification capability of the copper pyrometallurgy process, optimize the smelting process, reduce energy consumption and production cost, and improve production safety and stability. SUMMARY

[0006] To solve the problems existing in the prior art, the present application provides a temperature anomaly identification system and method for copper pyrometallurgy based on machine learning and platinum rhodium thermocouples, aiming to solve the problems of low temperature monitoring precision, inability to accurately identify abnormalities in real time, etc. in the prior art, improve the stability and safety of the smelting process, reduce energy consumption and production cost, and realize real-time monitoring and accurate early warning of temperature anomalies in the copper pyrometallurgy process by combining high-precision temperature measurement technology of platinum rhodium thermocouples and machine learning algorithms.

[0007] To achieve the above-mentioned purposes, the present application provides the following solutions:

[0008] The temperature anomaly identification system for copper pyrometallurgy based on machine learning and platinum rhodium thermocouples comprises: a platinum rhodium thermocouple measurement module, a data acquisition and transmission module, a machine learning temperature anomaly identification module, and a warning and control module.

[0009] The platinum rhodium thermocouple measurement module is used to install platinum rhodium thermocouples in the smelting furnace to measure temperature data in real time during the smelting process.

[0010] The data acquisition and transmission module is used to acquire temperature data and transmit the temperature data to the central processing system through wireless communication technology.

[0011] The machine learning temperature anomaly identification module is used to analyze and process temperature data by using machine learning algorithms, establish a temperature anomaly identification model, and identify abnormal temperatures.

[0012] The warning and control module is used to trigger the warning mechanism and automatically adjust the heating power and air volume parameters of the smelting furnace when abnormal temperatures are identified.

[0013] Preferably, the platinum rhodium thermocouple measurement module comprises a coating unit and an arrangement unit.

[0014] The coating unit is used to add a self-repairing oxidation-resistant coating to the measurement end of the platinum rhodium thermocouple.

[0015] The arrangement unit is used for selecting S-type or B-type platinum-rhodium thermocouples according to the high-temperature environment characteristics of copper pyrometallurgy, and arranging a plurality of platinum-rhodium thermocouples in the middle, top and bottom of the hearth wall of the smelting furnace to form a temperature monitoring network.

[0016] Preferably, the machine learning temperature anomaly identification module comprises: establishing a temperature anomaly identification model by using a long short-term memory network, and the output of the lth hidden layer in the temperature anomaly identification model at a time step t is Wherein, the neurons H l ∈[50,100] of the lth hidden layer, and the stack structure of the hidden layer is represented as:

[0017]

[0018] Wherein, X t is an input tensor at a time step t;

[0019] The output layer adopts a single neuron+Sigmoid activation function to output a temperature anomaly probability p t ∈(0,1):

[0020]

[0021] Wherein, L is the total number of hidden layers, h t ( L) is the output of the Lth hidden layer at a time step t, is a Sigmoid function, and x is a variable of the Sigmoid function; and are the weights and bias of the output layer, and T is a transpose;

[0022] The weighted loss function is:

[0023]

[0024] Wherein, is a loss function, M is the number of samples, y m represents the true label of the mth sample, p m represents the predicted probability of the mth sample, and alpha∈(0,1) is an abnormal sample weight.

[0025] Preferably, the early warning and control module comprises an early warning mechanism unit and a control strategy unit.

[0026] The early warning mechanism unit is used for triggering an early warning mechanism to notify smelting operators to take measures by means of sound-light alarm and short message notification when an abnormal temperature is identified.

[0027] A control strategy unit is configured to automatically adjust the heating power and air volume parameters of the smelting furnace to correct the temperature anomaly when the abnormal temperature is identified.

[0028] Preferably, the automatic adjustment in the control strategy unit is based on real-time measured temperature data and historical experience data, and a fuzzy control algorithm is used to predict the future temperature change trend, and then adjust the heating power and air volume parameters of the smelting furnace to correct the temperature anomaly.

[0029] The application also provides a method for identifying temperature anomalies in copper fire smelting based on machine learning and platinum-rhodium thermocouples, which is implemented by using the system described above, and the method comprises the following steps:

[0030] A platinum-rhodium thermocouple is installed in the smelting furnace to measure the temperature data in real time during the smelting process.

[0031] The temperature data is collected and transmitted to the central processing system through wireless communication technology.

[0032] The temperature data is analyzed and processed by using a machine learning algorithm to establish a temperature anomaly identification model to identify abnormal temperatures.

[0033] When the abnormal temperature is identified, a warning mechanism is triggered and the heating power and air volume parameters of the smelting furnace are automatically adjusted.

[0034] Preferably, the platinum-rhodium thermocouple is installed in the smelting furnace to measure the temperature data in real time during the smelting process, which comprises the following steps:

[0035] A self-repairing oxidation-resistant coating is added to the measurement end of the platinum-rhodium thermocouple.

[0036] According to the high-temperature environmental characteristics of copper fire smelting, S-type or B-type platinum-rhodium thermocouples are selected, and multiple platinum-rhodium thermocouples are arranged in the middle, top and bottom of the hearth wall of the smelting furnace to form a temperature monitoring network.

[0037] Preferably, the temperature data is analyzed and processed by using a machine learning algorithm to establish a temperature anomaly identification model to identify abnormal temperatures, which comprises the following steps: a long short-term memory network is used to establish a temperature anomaly identification model, and the output of the lth hidden layer in the temperature anomaly identification model at time step t is wherein the neurons H l of the lth hidden layer are

[0038]

[0039] wherein, X t is the input tensor at time step t;

[0040] The output layer uses a single neuron + Sigmoid activation function to output the temperature anomaly probability pt ∈(0,1):

[0041]

[0042] Where L is the total number of hidden layers, h t ( L) This is the output of the Lth hidden layer at time step t. Here, x is the Sigmoid function, and x is the variable for the Sigmoid function. and Here, T represents the weights and biases of the output layer, and T is the transpose.

[0043] The weighted loss function is:

[0044]

[0045] in, Let M be the loss function, M be the number of samples, and y be the loss function. m p represents the true label of the m-th sample. m Let α represent the predicted probability of the m-th sample, and α∈(0,1) be the weight of the outlier sample.

[0046] Preferably, upon detecting an abnormal temperature, an early warning mechanism is triggered, and the heating power and air volume parameters of the smelting furnace are automatically adjusted, including:

[0047] When an abnormal temperature is detected, an early warning mechanism is triggered to notify smelting operators to take measures via audible and visual alarms and SMS notifications.

[0048] When an abnormal temperature is detected, the heating power and air volume parameters of the smelting furnace are automatically adjusted to correct the temperature anomaly.

[0049] Preferably, the automatic adjustment is based on real-time measured temperature data and historical experience data, using a fuzzy control algorithm to predict future temperature change trends, and then adjusts the heating power and air volume parameters of the smelting furnace to correct temperature anomalies.

[0050] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0051] This invention enables real-time monitoring and accurate identification of temperature anomalies, allowing for timely corrective measures to prevent smelting accidents caused by temperature fluctuations and improve the stability of the smelting process. Stable temperature control improves copper yield and quality, reduces copper loss and impurity content due to temperature anomalies, thereby enhancing the economic benefits of copper smelting. Optimized temperature control strategies improve energy efficiency, reduce energy consumption and production costs during smelting, bringing significant economic benefits to enterprises. Timely detection and handling of temperature anomalies reduces safety risks to high-temperature furnaces and equipment, ensures operator safety, and enhances the safety of pyrometallurgical copper production.

[0052] The system has wide application prospects, is suitable for copper smelting plants of various scales and types, can bring a technological innovation to the copper pyrometallurgy industry, promote the intelligent and accurate development of smelting processes, and has important economic and social values. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiments are briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0054] Figure 1 A comparison diagram of temperature changes between artificial detection and sensor monitoring in the background art of the present application;

[0055] Figure 2 A comparison diagram of measurement accuracy between artificial detection and sensor monitoring in the background art of the present application;

[0056] Figure 3 A comparison diagram of abnormality recognition ability between artificial detection and sensor monitoring in the background art of the present application;

[0057] Figure 4 A whole structure diagram of the pyrometallurgical copper temperature abnormality recognition system based on machine learning and platinum-rhodium thermocouples in the embodiment of the present application;

[0058] Figure 5 A copper smelting furnace temperature monitoring and abnormality recognition diagram in the embodiment of the present application. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0060] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0061] Embodiment one

[0062] As Figure 4As shown, the present application provides a temperature anomaly identification system for copper fire smelting based on machine learning and platinum-rhodium thermocouple, which includes a platinum-rhodium thermocouple measurement module, a data acquisition and transmission module, a machine learning temperature anomaly identification module, and a warning and control module.

[0063] The platinum-rhodium thermocouple measurement module is used to install platinum-rhodium thermocouples in the smelting furnace to measure temperature data in real time during the smelting process.

[0064] The data acquisition and transmission module is used to collect temperature data and transmit the temperature data to the central processing system through wireless communication technology.

[0065] The machine learning temperature anomaly identification module is used to analyze and process temperature data using machine learning algorithms, establish a temperature anomaly identification model, and identify abnormal temperatures.

[0066] The warning and control module is used to trigger the warning mechanism and automatically adjust the heating power and air volume parameters of the smelting furnace when an abnormal temperature is identified.

[0067] Further, the specific implementation process of the present application is as follows:

[0068] The system of the present application includes a platinum-rhodium thermocouple measurement module, a data acquisition and transmission module, a machine learning temperature anomaly identification module, and a warning and control module, which work together to complete the temperature anomaly identification task. The platinum-rhodium thermocouple measurement module is responsible for installing platinum-rhodium thermocouples at key positions in the smelting furnace to measure temperature data in real time during the smelting process, providing a high-quality data basis for subsequent temperature anomaly identification. Platinum-rhodium thermocouples have excellent high-temperature stability and anti-interference ability, and can work stably in high-temperature and strong-corrosion smelting environments for a long time, providing accurate and reliable temperature measurement data. The measurement principle of platinum-rhodium thermocouples is to use the temperature-dependent characteristics of the thermoelectric potential difference between platinum-rhodium alloy and platinum alloy to determine the temperature value by measuring the thermoelectric potential difference, which has high measurement accuracy and a wide measurement range. In different areas of the smelting furnace, such as the middle, top, and bottom of the furnace wall, multiple platinum-rhodium thermocouples are arranged to form a temperature monitoring network to comprehensively grasp the temperature distribution during the smelting process.

[0069] (1) The platinum-rhodium thermocouple measurement module includes a coating unit and a layout unit.

[0070] The coating unit is used to add a self-repairing oxidation-resistant coating to the measurement end of the platinum-rhodium thermocouple.

[0071] The layout unit is used to select S-type or B-type platinum-rhodium thermocouples according to the high-temperature environmental characteristics of copper fire smelting, and arrange multiple platinum-rhodium thermocouples in the middle, top, and bottom of the furnace wall of the smelting furnace to form a temperature monitoring network.

[0072] Aiming at the strong corrosive environment of high-temperature metal vapor and sulfide coexisting in copper pyrometallurgy, the traditional platinum-rhodium thermocouple is prone to "hot corrosion-oxidation synergistic failure", which leads to reduced service life. The embodiment adds a self-repairing oxidation-resistant coating to the measuring end of the platinum-rhodium thermocouple:

[0073] (1) Electrochemically depositing a 5-μm-thick aluminum-cerium (Al-Ce) active coating on the surface of the platinum-rhodium alloy;

[0074] (2) When the coating is eroded by sulfide and microcracks are generated, Ce 3+ ions diffuse at high temperature to form CeO2, which seals the cracks in real time;

[0075] (3) The coating process is fully compatible with the existing S-type / B-type filament preparation process.

[0076] The above newly added coating can prolong the service life of the platinum-rhodium thermocouple, reduce the annual drift of the thermoelectric potential, and also reduce the annual maintenance cost of a single furnace (loss of shutdown + labor).

[0077] According to the characteristics of the high-temperature environment of copper pyrometallurgy, S-type platinum-rhodium thermocouples (platinum-rhodium 10-platinum) are selected. Their measurement range is controlled at 0-1700℃, and the accuracy level is I level, which can well meet the temperature measurement requirements in the process of copper pyrometallurgy. In some areas with ultra-high humidity, B-type platinum-rhodium thermocouples (platinum-rhodium 30-platinum-rhodium 6) can be selected, which have a wider measurement range and higher stability and corrosion resistance at high temperatures. The error limit continues to be ±0.25%, which needs to be noted that the above error limit refers to the deviation of the actual measured temperature of the temperature sensing element under normal working conditions.

[0078] (II) Data acquisition and transmission module

[0079] The data acquisition and transmission module is used to collect the temperature data measured by the platinum-rhodium thermocouple and transmit the data to the central processing system through wireless communication technology, ensuring the timeliness and accuracy of the data. The data acquisition device has high sampling rate and low delay characteristics, and can collect temperature data in real time and continuously, meeting the temperature monitoring requirements of copper pyrometallurgy which changes rapidly.

[0080] 1. Specification parameters of data acquisition device

[0081] A high-precision, high-sampling-rate data acquisition device is selected, and its sampling frequency can be set according to the dynamic characteristics of the smelting process, generally recommended to be set at 10-100 Hz. The resolution of the data acquisition device should not be less than 16 bits to ensure that the weak millivolt-level signal output by the platinum-rhodium thermocouple can be accurately collected, thereby improving the accuracy of temperature measurement.

[0082] 2. Data preprocessing function and algorithm

[0083] The data collector has built-in data preprocessing functions such as filtering and denoising. In terms of filtering, a digital low-pass filter can be used with a cutoff frequency set to 5-10 Hz to effectively filter out high-frequency noise interference. The denoising algorithm can use the Kalman filter algorithm to further reduce the influence of random noise on temperature measurement values through dynamic estimation and correction of collected data, improving data stability and reliability.

[0084] 3. Selection of wireless communication technology and equipment

[0085] Considering the environmental characteristics and data transmission requirements of the smelting site, LoRa wireless communication technology is selected for data transmission. LoRa has advantages such as long-distance transmission, low power consumption, and strong anti-interference ability, and can adapt to the complex electromagnetic environment of the smelting workshop. The working frequency of the wireless communication module can be selected at 433 MHz or 868 MHz frequency band, and the communication distance can generally reach 1-3 kilometers, meeting the data transmission distance requirements between the smelting furnace and the central processing system. The specific adjustment can be made according to the actual needs.

[0086] (Three) Machine learning temperature anomaly identification module

[0087] The machine learning temperature anomaly identification module is responsible for using machine learning algorithms to analyze and process the collected temperature data, and establishes a temperature anomaly identification model. The selection of machine learning algorithms should consider their processing ability for time series data and anomaly detection performance, and algorithms such as long short-term memory network (LSTM) can be used to capture long-term dependencies and short-term fluctuation characteristics in temperature data.

[0088] First of all, the input layer dimension, according to the number of platinum-rhodium thermocouples installed and the sampling frequency to determine the number of neurons in the input layer. For example, if there are 10 platinum-rhodium thermocouples and the sampling frequency is 10 Hz, the input layer dimension is (10, 10), that is, 10 thermocouples input 10 sampling temperature values within the past 1 second.

[0089] Second, the number of hidden layers and neurons, after several experiments, set the hidden layer to 2-3 layers, and the number of neurons in each layer to 50-100, which can achieve good temperature anomaly identification effect. The activation function of the hidden layer neurons uses the sigmoid function, which can limit the output of the neurons to the range of (0, 1), which helps to alleviate the gradient disappearance problem and improve the training efficiency of the model.

[0090] Finally, the output layer dimension and activation function, the output layer is used to judge whether the temperature is abnormal, so the output layer has only one neuron, and the activation function selects the Sigmoid function, which can convert the model output into a value representing the temperature anomaly probability. If the value is greater than the set threshold (such as 0.8), it is determined that the temperature is abnormal.

[0091] Specifically, the input layer structure design is space-time coupling modeling:

[0092] Since the coupling relationship between the spatial distribution of multiple thermocouples and the time sequence is not considered in the traditional method, the spatial dimension (multiple platinum rhodium thermocouples) and the time dimension (sliding window) are jointly modeled.

[0093] Suppose that N thermocouples are installed, the sampling frequency is f Hz, and the sliding window length is T' seconds, then the input tensor of time step t is:

[0094]

[0095] In this embodiment: N = 10, f = 10 Hz, T' = 1 s, then the input tensor of time step t is:

[0096]

[0097] LSTM hidden layer structure design - deep time sequence modeling:

[0098] The traditional shallow LSTM is difficult to capture the complex mode of long-period thermal inertia and short-period disturbance in the smelting process, and the present application adopts 2-3 layers of stacked LSTM, with 50-100 neurons in each layer.

[0099] Suppose that the output of the lth hidden layer at time step t is Where the neurons H l ∈ [50, 100] of the lth hidden layer are stacked to form a structure:

[0100]

[0101] Where, X t is the input tensor of time step t.

[0102] Output layer activation function - abnormal probability output:

[0103] The traditional binary classification model outputs a hard label, which cannot quantify the abnormal confidence, and the present application uses a single neuron + Sigmoid activation function for the output layer to output the temperature abnormal probability p t ∈ (0, 1) :

[0104]

[0105] Where, L is the total number of hidden layers, h t ( L) is the output of the Lth hidden layer at time step t, is a Sigmoid function, and x is a variable of the Sigmoid function; and For the weight and bias of the output layer, T is the transpose;

[0106] Loss function and optimizer - for unbalanced abnormal samples:

[0107] Normal samples are much more than abnormal samples, which leads to model bias towards majority class. Therefore, the invention adopts a weighted binary cross-entropy loss function, and introduces an Adam optimizer.

[0108] Let the true label be y m ∈{0,1}, the prediction probability is p m , the weighted loss function is:

[0109]

[0110] Where, is the loss function, M is the number of samples, alpha ∈ (0,1) is the weight of abnormal samples (alpha = 0.8 in this embodiment), y m and p m respectively represent the true label and prediction probability of the mth sample.

[0111] Adam optimizer parameters: learning rate η = 0.001, β1 = 0.95, β2 = 0.99.

[0112] In addition, in the model training process, first, the historical temperature data needs to be preprocessed, including data cleaning, normalization and other steps, to eliminate noise and outliers in the data and improve the effect of model training. The preprocessed data is divided into training set and test set, and the training set is used to train the machine learning model, and the model parameters are adjusted through the optimization algorithm (such as gradient descent method) to make it accurately fit the temperature change rule in the normal smelting process.

[0113] The optimization algorithm in the invention is improved:

[0114] 1. Adaptive learning rate adjustment: Based on the original basis, a method of adaptively adjusting the learning rate based on the characteristics of the data set is introduced, and the learning rate will be dynamically adjusted according to the performance of the model on the validation set to speed up the convergence and avoid overfitting.

[0115] 2. Improved momentum parameters: The momentum parameters β1 and β2 of the Adam optimizer are optimized, and through experiments, for the temperature anomaly identification task, setting β1 = 0.95, β2 = 0.99 can more effectively balance the short-term and long-term gradient information of the model, thereby improving the generalization ability of the model.

[0116] 3. Weighted loss function: To address the problem of temperature abnormal samples in the copper pyrometallurgical process, a weighted binary cross-entropy loss function is introduced. By assigning a higher weight (e.g., α = 0.8) to abnormal samples, the model can pay more attention to abnormal samples, thereby improving the accuracy of anomaly detection.

[0117] 4. Early stopping mechanism: To prevent overfitting, an early stopping mechanism is introduced. If the model's performance on the validation set does not significantly improve over several consecutive epochs, training is terminated early, which helps save computational resources and avoid model overfitting.

[0118] After training is complete, the model is validated using the test set to evaluate its anomaly detection performance, such as accuracy (e.g., over 95%), recall, F1 score, and other indicators. The false positive rate and false negative rate are both controlled within 5% to ensure the reliability and effectiveness of the model in actual application. In the actual smelting process, real-time collected temperature data is input into the trained machine learning model, which quickly and accurately judges whether the current temperature is abnormal based on the input data and outputs the abnormal identification result. When the model detects that the temperature data deviates significantly from the normal temperature variation pattern, such as a sudden increase or decrease in temperature exceeding the set threshold, or abnormal frequency of temperature fluctuations, it identifies the temperature as abnormal and triggers the early warning mechanism.

[0119] (IV) Early warning and control module, including early warning mechanism unit and control strategy unit;

[0120] Early warning mechanism unit, used to trigger the early warning mechanism to notify smelting operators to take measures through sound and light alarm, SMS notification when abnormal temperature is identified;

[0121] Control strategy unit, used to automatically adjust the heating power and air volume parameters of the smelting furnace to correct temperature abnormalities when abnormal temperature is identified.

[0122] The early warning and control module is responsible for triggering the early warning mechanism and automatically adjusting the heating power, air volume and other parameters of the smelting furnace when temperature abnormalities are identified, so as to correct the temperature abnormalities and ensure the stable operation of the smelting process. The early warning mechanism timely notifies the operator through various ways such as audible and visual alarms, short message notifications, etc., to ensure that the operator can receive the early warning information in time and take appropriate measures. The audible and visual alarm device can be installed at the smelting site and the control room. When a temperature abnormality occurs, it sends out obvious sound and light signals to remind the operator to take immediate action. Among them, the audible alarm uses a 90-100 decibel high-pitched siren, and the light alarm uses a red stroboscopic light with a flashing frequency of 2-5 times per second. When a temperature abnormality occurs, the audible and visual alarm device starts immediately and continues to work until the operator manually resets it or the temperature returns to normal. The short message notification function can send abnormal information to the operator's mobile phone through the mobile communication network. The short message content includes key information such as the exact time of the abnormality, the specific location (such as the center of the hearth, the left side of the hearth bottom, etc.), the degree of abnormality (such as the percentage of temperature exceeding the normal range), etc. The delay time of short message sending should be controlled within 10-20 seconds to ensure that the operator can receive the early warning information in time.

[0123] The early warning and control module automatically adjusts the heating power, air volume and other parameters of the smelting furnace according to the abnormal identification results to correct the temperature abnormalities. First of all, the heating power adjustment. When the hearth temperature is detected to be too high, the heating power is automatically reduced according to the degree of temperature exceeding the standard. For example, if the temperature exceeds the normal range by 5%-10%, the heating power is reduced by 10%-15%; if the temperature exceeds the normal range by more than 10%, the heating power is reduced by 20%-30%. The adjustment of heating power can be realized by controlling the input voltage, current or fuel flow of the heating element (such as resistance wire, gas nozzle, etc.). Then, the air volume adjustment. When the hearth bottom temperature is too low, the air volume can be increased by adjusting the speed of the air blower to increase the oxygen supply of the hearth bottom and promote fuel combustion, thereby increasing the hearth bottom temperature. The range of air volume adjustment is generally ±20% of the rated air volume, and the adjustment speed should be dynamically adjusted according to the temperature change trend to avoid excessive air volume adjustment leading to intensified temperature fluctuations in the furnace.

[0124] The automatic adjustment in the control strategy unit is based on real-time measured temperature data and historical experience data, using a fuzzy control algorithm to predict future temperature trends and then adjust the heating power, air volume parameters of the smelting furnace to correct temperature abnormalities. For example, if the current hearth temperature shows a rapid upward trend and has approached the upper limit temperature, the system will predict in advance and appropriately reduce the heating power to achieve more precise and stable temperature control, improving the response speed and control accuracy of the system.

[0125] As Figure 5The copper smelting furnace temperature monitoring and abnormality identification diagram shows that through high-precision measurement by platinum-rhodium thermocouples and data preprocessing, the system can obtain real-time and accurate temperature data of the copper smelting furnace, ensuring high precision of temperature monitoring. The LSTM model can effectively capture long-term dependencies and short-term fluctuation characteristics in temperature data, with an accuracy rate of temperature anomaly identification of over 95%. Through continuous collection and analysis of temperature data, the adaptive ability of the model is continuously enhanced, further improving the accuracy of anomaly identification. The system has fast response capability, and the entire process from temperature anomaly occurrence to triggering of early warning and automatic adjustment of parameters is completed within 1 minute, ensuring stable operation of the copper smelting furnace. Through the optimized temperature control strategy, the application of the system significantly improves energy utilization efficiency, reduces energy consumption and production cost of the copper smelting furnace. Implementation cases show that the application of the system reduces energy consumption of the copper smelting furnace by 15% and production cost by 10%. Timely discovery and handling of temperature anomalies reduce the risk of high temperature and equipment damage of the copper smelting furnace, ensuring the safety of operating personnel and enhancing the safety of copper smelting furnace production.

[0126] The embodiment realizes real-time monitoring and accurate early warning of temperature anomalies of the copper smelting furnace by combining high-precision temperature measurement technology of platinum-rhodium thermocouples and LSTM machine learning algorithms. The system performs well in improving temperature monitoring accuracy, anomaly identification accuracy, system response speed, energy consumption reduction, and production cost optimization, and has significant economic and safety benefits. The system is suitable for copper smelting furnaces of various scales and types, bringing a technological innovation to the copper smelting industry and promoting the intelligent and precise development of smelting processes.

[0127] Through the above specific embodiments, the problems of low precision, poor real-time performance, weak anomaly identification ability, and lagging control means in temperature monitoring during copper pyrometallurgical smelting can be effectively solved. The improved technology can improve the precision and stability of temperature monitoring, realize real-time monitoring of the smelting process, quickly identify temperature anomalies and timely adjust parameters such as heating power and air volume, thereby optimizing the smelting process, reducing energy consumption, reducing safety hazards, and improving production efficiency. It should be noted that the specific embodiments may vary depending on actual application conditions, and adjustments and optimizations should be made in actual operation according to factors such as the scale of the smelting furnace, process requirements, environmental conditions, etc., to ensure the best performance and reliability of the monitoring system.

[0128] In summary, the system of the present application can timely correct abnormal conditions by real-time monitoring and accurate identification of temperature anomalies, avoid smelting accidents caused by temperature fluctuations, and improve the stability of the smelting process. Stable temperature control is conducive to improving the yield and quality of copper, reducing copper loss and impurity content caused by temperature anomalies, and thus improving the economic benefits of copper smelting. Optimized temperature control strategy can improve energy utilization efficiency, reduce energy consumption and production cost in the smelting process, and bring significant economic benefits to enterprises. Timely detection and handling of temperature anomalies can reduce the safety risk of high-temperature furnace body and equipment, protect the safety of operators, and enhance the safety of copper production by fire smelting.

[0129] The system of the present application has wide application prospects and is suitable for copper smelting plants of various scales and types, can bring a technological innovation to the fire smelting copper industry, promote the intelligent and precise development of smelting process, and has important economic and social value.

[0130] Embodiment two

[0131] The present application also provides a fire smelting copper temperature anomaly identification method based on machine learning and platinum-rhodium thermocouples, which is implemented by using the system of the preceding embodiment. The method comprises:

[0132] Platinum-rhodium thermocouples are installed in the smelting furnace to measure temperature data in real time during the smelting process;

[0133] Temperature data are collected and transmitted to the central processing system through wireless communication technology;

[0134] Machine learning algorithms are used to analyze and process the temperature data, establish a temperature anomaly identification model, and identify abnormal temperatures;

[0135] When an abnormal temperature is identified, a warning mechanism is triggered and the heating power and air volume parameters of the smelting furnace are automatically adjusted.

[0136] Further, platinum-rhodium thermocouples are installed in the smelting furnace to measure temperature data in real time during the smelting process, including:

[0137] A self-repairing oxidation-resistant coating is added to the measurement end of the platinum-rhodium thermocouple;

[0138] According to the high-temperature environmental characteristics of fire smelting copper, S-type or B-type platinum-rhodium thermocouples are selected, and multiple platinum-rhodium thermocouples are arranged in the middle, top and bottom of the hearth wall of the smelting furnace to form a temperature monitoring network.

[0139] Further, machine learning algorithms are used to analyze and process the temperature data, establish a temperature anomaly identification model, and identify abnormal temperatures, including: a long short-term memory network is used to establish a temperature anomaly identification model, and the output of the lth hidden layer in the temperature anomaly identification model at time step t is wherein the neurons H of the lth hidden layer l ∈ [50, 100], the stacked structure of the hidden layers is represented as:

[0140]

[0141] wherein X t is the input tensor of time step t;

[0142] The output layer adopts a single neuron + Sigmoid activation function, and outputs the temperature anomaly probability p t ∈ (0, 1):

[0143]

[0144] wherein L is the total number of hidden layers, h t (L) is the output of the Lth hidden layer at time step t, is a Sigmoid function, and x is the variable of the Sigmoid function; and are the weights and bias of the output layer, and T is the transpose;

[0145] The weighted loss function is:

[0146]

[0147] wherein, is the loss function, M is the number of samples, y m represents the true label of the mth sample, p m represents the predicted probability of the mth sample, and a ∈ (0, 1) is the weight of the abnormal sample.

[0148] Further, when an abnormal temperature is identified, a warning mechanism is triggered and the heating power and air volume parameters of the smelting furnace are automatically adjusted, including:

[0149] When an abnormal temperature is identified, the warning mechanism is triggered to notify the smelting operator to take measures through sound-light alarm and short message notification;

[0150] When an abnormal temperature is identified, the heating power and air volume parameters of the smelting furnace are automatically adjusted to correct the temperature anomaly.

[0151] Further, the automatic adjustment is based on real-time measured temperature data and historical experience data, and a fuzzy control algorithm is adopted to predict the future temperature change trend, and then the heating power and air volume parameters of the smelting furnace are adjusted to correct the temperature anomaly.

[0152] Embodiment Three

[0153] This embodiment illustrates the system described in the foregoing embodiments through a specific case:

[0154] Before implementing the system, all platinum-rhodium thermocouple measuring ends are pre-coated with self-repairing oxidation-resistant coatings (Al-Ce active coatings) to ensure long-term stable operation in high-temperature sulfidation corrosion environment. When implementing the system, first select the appropriate platinum-rhodium thermocouple model to ensure that its measurement range and accuracy meet the requirements of the high-temperature environment of copper pyrometallurgical smelting. According to the structure and process characteristics of the smelting furnace, determine the specific installation position of the platinum-rhodium thermocouple, such as the middle, top and bottom of the furnace wall, to comprehensively monitor temperature changes.

[0155] Using special high-temperature installation tools and materials, the platinum-rhodium thermocouple is firmly installed at key positions of the smelting furnace measurement points to ensure its stability in high-temperature, strong corrosion environment. The installed platinum-rhodium thermocouple is tested preliminarily to check the accuracy and stability of its measurement data, and to eliminate problems such as poor contact that may occur during installation.

[0156] Configure the parameters of the data collector, set appropriate sampling rate and data transmission mode to collect temperature data measured by the platinum-rhodium thermocouple in real time and accurately. Choose appropriate wireless communication technology such as Wi-Fi or LoRa to build a data transmission network to ensure stable connection between the data collector and the central processing system.

[0157] In the central processing system, install and configure the software environment and tool library required for the machine learning temperature anomaly recognition module to provide support for model training and recognition. Collect historical temperature data during the process of copper pyrometallurgical smelting, including data during normal smelting and when temperature anomalies occur, as the basic data set for machine learning model training. Preprocess the historical temperature data, including data cleaning, normalization and other steps, to eliminate noise and outliers in the data and improve the effectiveness of model training.

[0158] Select an appropriate machine learning algorithm, such as long short-term memory network (LSTM), to build a temperature anomaly recognition model and use the preprocessed data to train the model. During model training, adjust model parameters such as learning rate, number of layers, number of neurons, etc. through cross-validation and other methods to obtain the best model performance. Use the test set to verify the trained machine learning model and evaluate its recognition accuracy, recall rate and other indicators under different temperature anomaly conditions. Based on the model verification results, optimize and adjust the model to improve its application effect in actual smelting process.

[0159] Configure the parameters of the warning and control module, set the temperature anomaly warning threshold and control strategy, such as the adjustment range of heating power, the adjustment speed of air volume, etc. Connect the warning and control module with the heating system and ventilation system of the smelting furnace to ensure that it can automatically adjust the relevant parameters when temperature anomalies are identified.

[0160] Before the system is put into use, overall testing and debugging are carried out to check the cooperation between modules and the stability of the system. In the actual pyrometallurgical copper production process, the system collects temperature data in real time and identifies abnormalities through machine learning models. When the system identifies temperature abnormalities, it immediately triggers a warning mechanism and notifies the operator through sound and light alarms, SMS notifications, and other means. The operator checks the operation status and process parameters of the smelting furnace in a timely manner according to the warning information and takes appropriate measures to handle it.

[0161] The system automatically adjusts the heating power and air volume of the smelting furnace and other parameters according to the preset control strategy to correct temperature abnormalities and restore normal smelting processes. During system operation, temperature data and abnormal identification results are continuously collected for online model updating and optimization to improve the system's self-adaptability.

[0162] Periodically calibrate and maintain platinum rhodium thermocouples to ensure long-term stable and accurate temperature measurement data. Regularly check and maintain data collectors and wireless communication equipment to ensure the stability and reliability of data transmission. According to changes in smelting processes and production needs, retrain and adjust machine learning models to adapt to new production conditions. Store and analyze system operation data and abnormal handling records to provide production optimization and equipment maintenance decision support for enterprises.

[0163] During system implementation, train operators to familiarize them with system usage methods and abnormal handling processes to improve system application effectiveness. The system's functions and modules can be expanded according to the scale and complexity of the smelting furnace, such as adding more temperature monitoring points, introducing other types of sensors, and integrating with the enterprise's production management system to achieve coordinated optimization of temperature anomaly identification and production scheduling, quality control, etc. During system implementation, focus on data security and privacy protection and take appropriate encryption and access control measures.

[0164] Optimize the system's interface and operation process according to actual application conditions to improve user experience. The system's temperature anomaly identification function can be extended to other metal smelting fields such as iron, aluminum, zinc, etc., with broad application prospects. During system implementation, collect user feedback and improvement suggestions to continuously optimize system performance and functions. Establish system monitoring and alarm logs to record the time, location, cause, and treatment results of each temperature anomaly occurrence to provide a basis for subsequent analysis and improvement.

[0165] Case 1

[0166] In a large copper smelting plant, the system successfully identified multiple abnormal temperature rises in the furnace, triggered an early warning and automatically adjusted the heating power, avoiding uneven melting of the furnace charge and damage to the furnace body, and improving the yield and quality of copper.

[0167] Case 2

[0168] In another copper smelting plant, the system detected an abnormal decrease in furnace bottom temperature, which could affect the reduction reaction of copper. The system immediately issued a warning and automatically increased the heating power of the furnace bottom, quickly restoring the furnace bottom temperature to the normal range, ensuring the stability of the smelting process and the quality of the copper.

[0169] Case 3

[0170] After using the system in a copper smelting plant, through continuous monitoring and analysis of temperature data, it was found that the furnace wall temperature had periodic fluctuations. After investigation, it was found that the aging of the furnace wall insulation material caused an increase in heat loss. After replacing the insulation material in time, the system monitored that the furnace wall temperature tended to be stable, and the smelting energy consumption was significantly reduced.

[0171] Case 4

[0172] In a small copper smelting plant test, the system successfully identified an abnormal rise in furnace temperature at the beginning of smelting, which may be due to insufficient combustion caused by wet furnace charge. After the system triggered a warning, the operator adjusted the furnace charge ratio and ventilation parameters in time, and the furnace temperature returned to normal, improving the smelting efficiency and the purity of copper.

[0173] Case 5

[0174] After using the system for a period of time in a copper smelting plant, a large amount of temperature data and abnormal identification results were accumulated. Through in-depth analysis of these data, the smelting process parameters such as furnace charge particle size and furnace atmosphere were optimized, further improving the yield and quality of copper, reducing production costs, and enhancing the market competitiveness of the enterprise. DETAILED DESCRIPTION

[0176] 1. System composition and working principle: (1) Platinum-rhodium thermocouple measurement module: Install platinum-rhodium thermocouples at key positions of the copper smelting furnace (such as the hearth, the bottom of the furnace, and the furnace wall) to measure temperature data in real time. (2) Data acquisition and transmission module: Collect temperature data measured by platinum-rhodium thermocouples and transmit the data to the central processing system through wireless communication technology (such as Wi-Fi or LoRa). (3) Machine learning temperature anomaly identification module: Use machine learning algorithms such as long short-term memory network (LSTM) to analyze and process the collected temperature data, and establish a temperature anomaly identification model. (4) Early warning and control module: When temperature anomalies are identified, trigger the early warning mechanism and timely notify the operator through sound and light alarms, SMS notifications, and other means, while automatically adjusting the heating power, air volume, and other parameters of the copper smelting furnace to correct temperature anomalies.

[0177] 2. Data acquisition and preprocessing: (1) Use a high sampling rate and low delay data collector to collect temperature data measured by platinum-rhodium thermocouples in real time; (2) filter and denoise the collected raw data for data preprocessing, thereby improving the accuracy and stability of the data.

[0178] 3. Machine learning model training: Collect historical copper smelting furnace temperature data, including normal smelting and temperature anomaly data. Clean and normalize the historical temperature data to eliminate noise and outliers in the data. Then, use the preprocessed data to train the LSTM model, adjust the model parameters through optimization algorithms (such as gradient descent), and make it accurately fit the temperature variation law in the normal smelting process. Finally, use the test set to verify the trained model and evaluate its anomaly detection performance, such as accuracy, recall rate, F1 score, and other indicators.

[0179] 4. Real-time monitoring and anomaly identification: Input the real-time collected temperature data into the trained LSTM model. The model quickly and accurately judges whether the current temperature is abnormal based on the input data and outputs the anomaly identification result.

[0180] 5. Early warning and control: When temperature anomalies are detected, the system timely notifies the operator through sound and light alarms, SMS notifications, and other means. According to the anomaly identification result, automatically adjust the heating power, air volume, and other parameters of the copper smelting furnace to correct temperature anomalies. For example, when detecting that the hearth temperature is too high, the system automatically reduces the heating power or increases the air volume; when detecting that the bottom of the furnace temperature is too low, the system automatically increases the heating power or reduces the air volume.

[0181] The above described embodiments are only to illustrate the preferred modes of the present application, and are not intended to limit the scope of the present application. Any modification and improvement made by those skilled in the art to the technical solutions of the present application without departing from the design spirit of the present application shall fall within the protection scope of the present application as defined by the claims.

Claims

1. A pyrometallurgical copper temperature anomaly recognition system based on machine learning and platinum-rhodium thermocouple, characterized in that, The system comprises a platinum-rhodium thermocouple measurement module, a data acquisition and transmission module, a machine learning temperature anomaly identification module, and a warning and control module. The platinum-rhodium thermocouple measurement module is used to install platinum-rhodium thermocouples in the smelting furnace to measure temperature data in real time during the smelting process. The data acquisition and transmission module is used to collect temperature data and transmit the temperature data to the central processing system through wireless communication technology. The machine learning temperature anomaly identification module is used to analyze and process the temperature data using machine learning algorithms, establish a temperature anomaly identification model, and identify abnormal temperatures. The warning and control module is used to trigger the warning mechanism and automatically adjust the heating power and air volume parameters of the smelting furnace when an abnormal temperature is identified.

2. The system of claim 1, wherein, The platinum-rhodium thermocouple measurement module comprises a coating unit and an arrangement unit. The coating unit is used to add a self-repairing oxidation-resistant coating to the measurement end of the platinum-rhodium thermocouple. The arrangement unit is used to select S-type or B-type platinum-rhodium thermocouples according to the high-temperature environment characteristics of copper fire smelting, and arrange multiple platinum-rhodium thermocouples in the middle, top, and bottom of the hearth wall of the smelting furnace to form a temperature monitoring network.

3. The system of claim 1, wherein, The machine learning temperature anomaly identification module comprises: a long short-term memory network is used to establish a temperature anomaly identification model, and the output of the lth hidden layer in the temperature anomaly identification model at the time step t is wherein the neuron H of the lth hidden layer l ∈ [50, 100], and the stacking structure of the hidden layer is represented as: where X t is the input tensor for time step t; The output layer adopts single neuron+Sigmoid activation function, and outputs temperature anomaly probability p t ∈(0,1): where L is the total number of hidden layers, h t (L) is the output of the Lth hidden layer at time step t, is a Sigmoid function, x is a variable of the Sigmoid function; and are weights and biases of the output layer, and T is a transpose. The weighted loss function is: wherein, is a loss function, M is the number of samples, y m represents the true label of the mth sample, p m represents the prediction probability of the mth sample, and a e (0, 1) is an abnormal sample weight.

4. The system of claim 1, wherein, The warning and control module comprises a warning mechanism unit and a control strategy unit. The warning mechanism unit is used to trigger the warning mechanism to notify smelting operators to take measures through sound and light alarms and SMS notifications when an abnormal temperature is identified. The control strategy unit is used to automatically adjust the heating power and air volume parameters of the smelting furnace to correct temperature abnormalities when an abnormal temperature is identified.

5. The system of claim 4, wherein, The automatic adjustment in the control strategy unit is based on real-time measured temperature data and historical experience data, uses a fuzzy control algorithm to predict future temperature trends, and then adjusts the heating power and air volume parameters of the smelting furnace to correct temperature abnormalities.

6. A method for identifying temperature anomalies in copper pyro-metallurgical smelting based on machine learning and platinum rhodium thermocouples, the method being implemented using the system of any one of claims 1 to 5, characterized in that, The method comprises: Installing platinum-rhodium thermocouples in the smelting furnace to measure temperature data in real time during the smelting process. Collecting temperature data and transmitting the temperature data to the central processing system through wireless communication technology. Using machine learning algorithms to analyze and process the temperature data, establishing a temperature anomaly identification model, and identifying abnormal temperatures. Triggering the warning mechanism and automatically adjusting the heating power and air volume parameters of the smelting furnace when an abnormal temperature is identified.

7. The method of claim 6, wherein, Installing platinum-rhodium thermocouples in the smelting furnace to measure temperature data in real time during the smelting process, including: Adding a self-repairing oxidation-resistant coating to the measurement end of the platinum-rhodium thermocouple. Selecting S-type or B-type platinum-rhodium thermocouples according to the high-temperature environment characteristics of copper fire smelting, and arranging multiple platinum-rhodium thermocouples in the middle, top, and bottom of the hearth wall of the smelting furnace to form a temperature monitoring network.

8. The method of claim 6, wherein, The temperature data is analyzed and processed by using a machine learning algorithm, a temperature anomaly recognition model is established, and abnormal temperature is recognized, including: a long short-term memory network is used to establish a temperature anomaly recognition model, and the output of the lth hidden layer in the temperature anomaly recognition model at the time step t is where the neurons H l ∈[50,100], and the stacking structure of the hidden layer is represented as: where X t is the input tensor for time step t; The output layer adopts single neuron+Sigmoid activation function, and outputs temperature anomaly probability p t ∈(0,1): where L is the total number of hidden layers, h t (L) is the output of the Lth hidden layer at time step t, is a Sigmoid function, x is a variable of the Sigmoid function; and are weights and biases of the output layer, and T is a transpose. The weighted loss function is: wherein, is a loss function, M is the number of samples, y m represents the true label of the mth sample, p m represents the predicted probability of the mth sample, and a e (0, 1) is an abnormal sample weight.

9. The method of claim 6, wherein, Triggering the warning mechanism and automatically adjusting the heating power and air volume parameters of the smelting furnace when an abnormal temperature is identified, including: Triggering the warning mechanism to notify smelting operators to take measures through sound and light alarms and SMS notifications when an abnormal temperature is identified. Automatically adjusting the heating power and air volume parameters of the smelting furnace to correct temperature abnormalities when an abnormal temperature is identified.

10. The method of claim 9, wherein, Automatic adjustment is based on real-time measurement of temperature data and historical data, using fuzzy control algorithm, predict future temperature trends, and then adjust the heating power of the smelting furnace, air volume parameters, correct temperature anomalies.

Images