A method and system for controlling a glass article manufacturing process

Abstract

The application discloses a kind of control method and system in glass product manufacturing process, specifically relates to glass product manufacturing technical field;Through real-time monitoring the voltage fluctuation and signal noise of temperature sensor, analyze the voltage fluctuation and signal noise data in several time periods, calculate the temperature sensor data accuracy index in combination with weight assignment method, and through oxygen sensor real-time monitoring furnace oxygen concentration fluctuation, evaluate the influence of atmosphere change, monitor data accuracy and oxygen concentration fluctuation are comprehensively evaluated using fuzzy logic algorithm, dynamically adjust heating power, to optimize glass melting process, prevent temperature out of control or incomplete melting, reduce the risk of equipment damage, improve production efficiency, reduce maintenance and downtime, effectively improve the accuracy of temperature monitoring, enhance the stability of system, ensure the efficient, safe operation of glass production process.

Description

Technical Field

[0001] This invention relates to the field of glass product manufacturing technology, and specifically to a control method and system for the glass product manufacturing process. Background Technology

[0002] Control in the glass manufacturing process refers to the real-time monitoring and adjustment of key parameters such as temperature, pressure, time, and chemical composition through technical means at every stage of glass production to ensure stable product quality, improved process efficiency, and optimized resource utilization. This control process spans every stage from raw material mixing, melting, forming, annealing to subsequent processing, aiming to avoid defects caused by process deviations, such as bubbles, streaks, uneven thickness, or insufficient strength. Specifically, during the glass melting stage, the control system monitors the temperature distribution of the furnace and the uniformity of the raw materials to ensure the purity and fluidity of the molten glass; during the forming stage, precise adjustment of the mold temperature and the flow rate of the molten glass yields products with the required dimensions and shapes; during annealing, the cooling rate is strictly controlled to avoid brittleness caused by residual stress. This end-to-end control not only improves the quality and reliability of glass products but also effectively reduces energy consumption and production costs.

[0003] The existing technology has the following shortcomings:

[0004] Control systems rely on sensors to acquire real-time data and make decisions. If a feedback loop error occurs during data transmission, the system may fall into an incorrect adjustment cycle. For example, a temperature sensor might transmit incorrect data for some reason (such as electrical interference, malfunction, or signal delay). The control system adjusts based on this data, but the new adjustments, in turn, affect the temperature sensor's data, creating a vicious cycle. Simultaneously, feedback loop errors can cause temperature sensors to continuously send incorrect temperature data to the control system, which then automatically adjusts the heating power based on this erroneous data. Since the adjusted temperature data is then transmitted to the control system again, forming a vicious cycle, it can lead to uncontrolled temperature fluctuations, either too high or too low. If the furnace temperature is too high, the equipment and furnace materials may be damaged by overheating, potentially leading to furnace wall melting or equipment burnout, resulting in significant repair costs and downtime. If the temperature is too low, the glass may not melt completely, the reaction rate inside the furnace may slow down, impurities may remain in the glass, and in severe cases, the final glass product may be impossible to extract. Summary of the Invention

[0005] The purpose of this invention is to provide a control method and system for the manufacturing process of glass products, so as to overcome the shortcomings of the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a control method in the manufacturing process of glass products, comprising the following steps:

[0007] S1: Install a voltage monitor on the power supply line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and install a noise monitor in the communication line between the temperature sensor and the control system to monitor the noise data in the signal transmission in real time, and monitor the heating power output of the heater in real time through a power meter.

[0008] S2: Obtain the power supply voltage fluctuation data and signal transmission noise data of the temperature sensor in several time periods, analyze them, determine the weight of the accuracy of the temperature sensor monitoring data in each time period, and calculate the monitoring data accuracy index by weighted average summation of the weight of the accuracy of the temperature sensor monitoring data in each time period.

[0009] S3: The oxygen concentration in the furnace is monitored in real time by an oxygen sensor, and the degree of abnormal change in the oxygen content of the furnace atmosphere is judged based on the fluctuation of the oxygen concentration in the furnace gas within a fixed time period.

[0010] S4: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted using fuzzy logic to optimize the melting process.

[0011] Preferably, in step S2, after analyzing the fluctuation data of the temperature sensor power supply voltage, a voltage fluctuation frequency anomaly index is generated. The method for obtaining the voltage fluctuation frequency anomaly index is as follows:

[0012] Collect voltage fluctuation data within the Q time period Where v(s) is the value of the voltage signal at time s, and N is the number of sampling points. The signal in the time domain is converted into a frequency domain signal by Fast Fourier Transform (FFT). The FFT converts the voltage signal V(s) into the frequency domain signal V(f), and its calculation formula is as follows: ; It is frequency The corresponding complex Fourier coefficients, where j is the imaginary unit. ; Here, T is the corresponding frequency, T is the signal sampling period, and k is the frequency index; to analyze the frequency components of the voltage fluctuation signal, its amplitude spectrum is calculated. That is, the energy distribution at different frequencies, expressed as: In the formula, Re( ) and Im They are respectively The real and imaginary parts of the spectrum; calculate the total energy of the entire spectrum. As a reference value, the expression is: ; Identify amplitude value If the amplitude is greater than the normal fluctuation range, Mark these frequencies as anomalous frequencies and calculate their energy contribution. The expression is: The voltage fluctuation frequency anomaly index is calculated using the following expression: In the formula, AK is the voltage fluctuation frequency anomaly index.

[0013] Preferably, in step S2, the signal collected by the temperature sensor is discretized into a discrete numerical sequence, and the signal is set. Given a signal sequence of length Z, divide it into discrete intervals to obtain discrete values. Define the intervals. To represent different discrete values, calculate the probability of each discrete value in the signal X, forming the probability distribution P(X) of the signal, and statistically analyze each discrete value in the signal. Frequency of occurrence Calculate the probability of each discrete value. The expression is: The information entropy is calculated based on the probability distribution of the signal. The calculation expression is as follows: ;in, The entropy of signal X is calculated by comparing the entropy of the original signal with that of the denoised signal. The signal noise interference index is then calculated, and the entropy H(X) of the entire temperature sensor signal sequence is calculated. Noise suppression is then applied to the signal to obtain the denoised signal sequence. Then calculate its entropy. Next, the signal-to-noise interference index is calculated, expressed as: In the formula, This represents the signal-to-noise interference index.

[0014] Preferably, in S2, the voltage fluctuation frequency anomaly index and the signal noise interference index are converted into a comprehensive feature vector. The comprehensive feature vector is used as the input to the machine learning model. The machine learning model uses the weighted labels of the accuracy of temperature sensor monitoring data in each time period predicted by each set of comprehensive feature vectors as the prediction objective. The training objective is to minimize the sum of prediction errors of the weighted labels of the accuracy of temperature sensor monitoring data in all time periods. The machine learning model is trained until the sum of prediction errors converges and the model training stops. The weighted labels of the accuracy of temperature sensor monitoring data in each time period are determined according to the model output. The machine learning model is a multinomial regression model, and the accuracy index of monitoring data is obtained by weighted averaging and summing the weighted labels of the accuracy of temperature sensor monitoring data in each time period.

[0015] Preferably, in S3, after analyzing the fluctuation of oxygen concentration in the furnace gas within a fixed time period, an oxygen concentration fluctuation deviation index is generated to determine the degree of abnormal change in oxygen content in the furnace atmosphere. The method for obtaining the oxygen concentration fluctuation deviation index is as follows:

[0016] Collect oxygen concentration data within a fixed time period, and set the oxygen concentration data as... Where t represents the timestamp, Given the oxygen concentration data at time t; select the order of the autoregressive moving average (ARMA) model. The ARMA model includes an autoregressive (AR) term and a moving average (MA) term, and the model is expressed as: ;in, The coefficient of the AR term, The coefficient of the MA term. For noise terms, In time The white noise residuals at time step 1 are used to determine the orders p and q using the autocorrelation function and partial autocorrelation function; based on the chosen orders p and q, the model parameters are estimated. and , Or q; use the fitted ARMA model to predict future oxygen concentrations, and calculate the difference between actual observations and model predictions, i.e.: ;in, For residuals, To calculate the oxygen concentration fluctuation deviation index based on the oxygen concentration value predicted by the ARMA model, the expression is as follows: In the formula, Y is the total number of data points, and KM is the oxygen concentration fluctuation deviation index.

[0017] Preferably, in step S4, based on the calculated accuracy index of the monitoring data and the fluctuation of oxygen concentration in the furnace gas, fuzzy logic is applied to dynamically adjust the heating power and optimize the melting process, specifically:

[0018] The accuracy index RY of monitoring data and the oxygen concentration fluctuation deviation index KM are used as input terms of fuzzy logic, and the heating power of the melting furnace is used as the output term.

[0019] The accuracy index of monitoring data, the oxygen concentration fluctuation deviation index, and the heating power of the melting furnace were all fuzzed.

[0020] Construct fuzzy rules based on RY and KM, which connect the relationship between input and output variables through fuzzy logic;

[0021] The measured RY and KM values ​​are converted into fuzzy sets;

[0022] Based on the defined fuzzy rules, the fuzzy value of the heating power is obtained through a reasoning process;

[0023] During the reasoning process, a fuzzy heating power value is obtained, and at the same time, defuzzification is performed to obtain a specific heating power value.

[0024] The defuzzified heating power value is fed back to the automated control system, which then adjusts the heating power in the furnace according to the value, thereby optimizing the glass melting process.

[0025] The present invention also provides a control system for the manufacturing process of glass products, including a sensor monitoring module, a data analysis module, an oxygen concentration monitoring module, and a fuzzy logic control module;

[0026] Sensor monitoring module: A voltage monitor is installed on the power line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and a noise monitor is installed in the communication line between the temperature sensor and the control system to monitor the noise data in signal transmission in real time, and the heating power output of the heater is monitored in real time through a power meter.

[0027] Data Analysis Module: Acquires power supply voltage fluctuation data and noise data in signal transmission of temperature sensors over several time periods. After analysis, determines the weighting of the accuracy of temperature sensor monitoring data in each time period. The accuracy index of monitoring data is obtained by weighted averaging and summing the weighting of the accuracy of temperature sensor monitoring data in each time period.

[0028] Oxygen concentration monitoring module: It monitors the oxygen concentration of the gas in the furnace in real time through an oxygen sensor, and judges the degree of abnormal change in the oxygen content of the atmosphere in the furnace based on the fluctuation of the oxygen concentration of the gas in the furnace within a fixed time period.

[0029] Fuzzy logic control module: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted after applying fuzzy logic to optimize the melting process.

[0030] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0031] 1. This invention provides a strong guarantee for the accuracy of temperature data by installing a voltage monitor, noise monitor, and power meter on the power line of the temperature sensor to detect power fluctuations, signal noise, and heater power in real time. By analyzing voltage fluctuations and signal noise, and combining them with a machine learning model to generate a monitoring data accuracy index, and then using fuzzy logic analysis based on the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted, effectively avoiding the risk of temperature runaway or incomplete melting.

[0032] 2. This invention can accurately assess the accuracy of temperature sensor monitoring data and changes in the furnace atmosphere in real time, and precisely control the melting process based on these indicators. In particular, while solving the problem of vicious feedback loops, it can optimize the glass melting process, improve production efficiency, reduce the risk of equipment damage, and effectively reduce production downtime and maintenance costs caused by temperature fluctuations. It provides the glass melting industry with an efficient, safe, and intelligent production control method, offering significant economic benefits and safety assurance. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0034] Figure 1 This is a flowchart of the method of the present invention.

[0035] Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Example 1, please refer to Figure 1 and Figure 2 As shown in this embodiment, a control method for the manufacturing process of glass products includes the following steps:

[0038] S1: Install a voltage monitor on the power supply line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and install a noise monitor in the communication line between the temperature sensor and the control system to monitor the noise data in the signal transmission in real time, and monitor the heating power output of the heater in real time through a power meter.

[0039] S2: Obtain the power supply voltage fluctuation data and signal transmission noise data of the temperature sensor in several time periods, analyze them, determine the weight of the accuracy of the temperature sensor monitoring data in each time period, and calculate the monitoring data accuracy index by weighted average summation of the weight of the accuracy of the temperature sensor monitoring data in each time period.

[0040] S3: The oxygen concentration in the furnace is monitored in real time by an oxygen sensor, and the degree of abnormal change in the oxygen content of the furnace atmosphere is judged based on the fluctuation of the oxygen concentration in the furnace gas within a fixed time period.

[0041] S4: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted using fuzzy logic to optimize the melting process.

[0042] In S1, a voltage monitor is installed on the power supply line of the temperature sensor to detect real-time fluctuations in the power supply voltage, and a noise monitor is installed on the communication line between the temperature sensor and the control system to monitor noise data during signal transmission in real time. A power meter is also used to monitor the heating power output of the heater in real time. Specifically:

[0043] A voltage monitor is used to monitor power supply voltage fluctuations of a temperature sensor in real time, and can detect errors caused by electrical interference, voltage instability, and other problems. The operating accuracy of a temperature sensor is greatly affected by the power supply voltage; voltage fluctuations may cause the sensor to output inaccurate temperature data.

[0044] Select a suitable voltage monitor based on the operating voltage range of the temperature sensor. The voltage monitor should have sufficient accuracy and response speed to monitor power supply voltage fluctuations in real time. Install the voltage monitor at the power input terminal of the temperature sensor. Ensure that the voltage monitor can directly measure the voltage received by the sensor, unaffected by interference from external power sources or equipment. By monitoring voltage fluctuation data, it is possible to determine if there are any voltage anomalies. If the power supply voltage fluctuation exceeds a set threshold, an alarm system can be triggered, and compensatory measures can be taken, such as adjusting the power supply or switching to a backup power supply. Real-time monitoring of voltage fluctuations can identify the effects of electrical interference or power instability, avoiding sensor data errors caused by voltage anomalies. If the temperature sensor transmits data incorrectly due to voltage fluctuations, it can lead to erroneous adjustments in the automated control system, resulting in temperature runaway. A voltage monitor helps to detect voltage anomalies early and prevent temperature regulation errors.

[0045] Noise monitors are used to monitor noise in real-time during signal transmission between temperature sensors and control systems, identifying and eliminating communication errors caused by electromagnetic interference (EMI), signal attenuation, or crosstalk. The noise monitor should be able to capture noise in the signal line, including both high-frequency and low-frequency interference. Common noise sources include power fluctuations, motor operation, and external electromagnetic fields. Connect the noise monitor to the data communication line between the temperature sensor and the control system. Ensure the monitor can detect noise data in the signal in real time and analyze signal quality. The noise monitor can monitor noise changes in the communication signal in real time and generate noise level data. If the noise exceeds a set threshold, it can trigger an alarm or automatic calibration system to correct communication errors. The noise monitor can promptly identify and quantify signal interference between the sensor and the control system, avoiding erroneous data transmission caused by noise interference. Noise in signal transmission may cause the temperature sensor to transmit incorrect temperature data, thus affecting the regulation of the automated control system. The noise monitor can improve the reliability and accuracy of data transmission.

[0046] Monitoring the heater's power output with a power meter allows for real-time tracking of its operating status, ensuring that its output power matches the temperature setpoint and preventing temperature runaway caused by abnormal heater power. The power meter should have the function of measuring heater power output in real time, detecting instantaneous power changes and long-term power trends. A multi-functional power meter capable of monitoring current, voltage, and power factor should be selected. The power meter should be installed in the heater circuit, ensuring accurate measurement of the heater's current and voltage, and calculation of the power output. The power meter should be directly connected to the heater's input. The power meter can transmit real-time power data to the central control system, analyzing changes in power output to determine if the heater is operating normally. If the power output is abnormal, the control system can immediately adjust the heater power or activate a backup heater. The power meter provides data on the actual power output of the heater, ensuring that the heater power matches the temperature control requirements. If the power output is too high or too low, the system can automatically adjust to prevent excessively high or low temperatures. Real-time monitoring of power output allows for timely detection of heater malfunctions or performance degradation, preventing overheating due to excessive power or incomplete glass melting due to insufficient power.

[0047] By installing voltage and noise monitors, the temperature sensor can operate under stable power and signal conditions, thus improving the accuracy of temperature monitoring data. Inaccurate temperature sensor data can lead to continuous erroneous adjustments in the control system, creating a vicious feedback loop. Real-time monitoring of voltage fluctuations, signal noise, and heater power output allows for early warnings and corrective actions before problems occur, preventing this erroneous cycle. Through multiple monitoring and real-time data analysis, the system can adaptively adjust its control strategy to ensure that the temperature and heating power during glass melting remain at optimal levels, thereby improving production efficiency and reducing equipment damage.

[0048] In this application, by installing a voltage monitor on the temperature sensor power line, a noise monitor on the signal transmission line, and a power meter to monitor the heater's power output, temperature control problems caused by electrical interference, signal transmission errors, or abnormal heater power can be effectively reduced. This comprehensive monitoring and feedback mechanism not only enables real-time monitoring and optimization of system parameters but also allows for timely corrective measures in case of anomalies, ensuring temperature control accuracy and equipment stability during the glass melting process.

[0049] S2: Acquire the power supply voltage fluctuation data and noise data in signal transmission of the temperature sensor within several time periods. After analysis, determine the weight of the accuracy of the temperature sensor monitoring data within each time period. Then, calculate the weighted average summation of the weighted values ​​of the accuracy of the temperature sensor monitoring data within each time period to obtain the monitoring data accuracy index.

[0050] Based on the sampling frequency of the temperature sensor and the response time of the control system, the initial time period length is set. For example, if the temperature sensor samples once per second and the control system adjusts every 5 seconds, the time period can be set to 5 seconds or 10 seconds.

[0051] Real-time collection of power supply voltage fluctuation data from a temperature sensor over several time periods; analysis of the power supply voltage fluctuation data to generate a voltage fluctuation frequency anomaly index; the method for obtaining the voltage fluctuation frequency anomaly index is as follows:

[0052] Collect voltage fluctuation data within the Q time period Where v(s) is the value of the voltage signal at time s, and N is the number of sampling points. The signal in the time domain is converted into a frequency domain signal by Fast Fourier Transform (FFT). The FFT converts the voltage signal V(s) into the frequency domain signal V(f), and its calculation formula is as follows: ; It is frequency The corresponding complex Fourier coefficients, where j is the imaginary unit. ; Here, T is the corresponding frequency, T is the signal sampling period, and k is the frequency index; to analyze the frequency components of the voltage fluctuation signal, its amplitude spectrum is calculated. That is, the energy distribution at different frequencies, expressed as: In the formula, Re( ) and Im They are respectively The real and imaginary parts of the spectrum; calculate the total energy of the entire spectrum. As a reference value, the expression is: ; Identify amplitude value If the amplitude is greater than the normal fluctuation range, Mark these frequencies as anomalous frequencies and calculate their energy contribution. The expression is: The voltage fluctuation frequency anomaly index is calculated using the following expression: In the formula, AK is the voltage fluctuation frequency anomaly index.

[0053] A higher voltage fluctuation frequency anomaly index typically indicates significant abnormal fluctuations in the power supply line or signal transmission channel where the temperature sensor is located. These abnormal fluctuations may be caused by factors such as electrical interference, equipment failure, and signal delay, thus affecting the accuracy of the data collected by the temperature sensor. When the voltage fluctuation frequency anomaly index is high, it means that the power supply voltage is unstable, and the sensor's operating environment is adversely affected, leading to errors or deviations in the temperature sensor's monitoring data. In this case, the temperature sensor may continuously collect erroneous data, affecting the decision-making and adjustment of the entire automated control system, and even causing temperature runaway during the production process, increasing the risk of equipment failure and downtime.

[0054] On the other hand, a lower voltage fluctuation frequency anomaly index indicates less voltage fluctuation in the power supply line where the temperature sensor is located, less interference during signal transmission, and a more stable power supply. In this case, the temperature data collected by the sensor is more likely to be accurate and reliable because the impact of electrical interference, signal delay, and other fault factors is smaller. Therefore, a lower anomaly index is usually associated with higher data accuracy, indicating that the sensor can provide stable and reliable data, which helps the automated control system make precise temperature regulation and production decisions, ensuring the smooth operation of the production process.

[0055] Noise data from temperature sensor signal transmission is collected in real time over several time periods. After analyzing the noise data, a signal noise interference index is generated. The method for obtaining the signal noise interference index is as follows:

[0056] The signal collected by the temperature sensor is discretized into a discrete numerical sequence, and the signal is set. Given a signal sequence of length Z, divide it into discrete intervals to obtain discrete values. Define the intervals. To represent different discrete values, calculate the probability of each discrete value in the signal X, forming the probability distribution P(X) of the signal, and statistically analyze each discrete value in the signal. Frequency of occurrence Calculate the probability of each discrete value. The expression is: Information entropy is calculated based on the probability distribution of the signal. Information entropy measures the uncertainty or disorder of a signal. Higher entropy indicates more uncertainty in the signal, potentially representing stronger noise. The calculation expression is: ;in, It is the entropy of signal X. If the probability distribution of the signal is relatively uniform, that is, the probability of each discrete value appearing is similar, then the entropy is high, indicating that the signal has high complexity or noise.

[0057] By comparing the entropy of the original signal and the entropy of the denoised signal, the signal noise interference index is calculated, and the entropy H(X) of the entire temperature sensor signal sequence is calculated. Noise suppression or filtering processing (such as low-pass filtering, wavelet denoising, etc.) is then applied to the signal to obtain the denoised signal sequence. Then calculate its entropy. Next, the signal-to-noise interference index is calculated, expressed as: In the formula, This represents the signal-to-noise interference index.

[0058] A higher signal-to-noise interference (SNOI) index indicates lower accuracy of the temperature sensor's monitoring data. A large SNOI signifies strong noise components and a high entropy value in the signal detected by the temperature sensor, indicating greater uncertainty and disorder. This typically suggests significant external interference or inherent sensor noise, leading to unstable and unreliable measurement data. Noise interference affects sensor accuracy and response speed, potentially causing erroneous data to be input into the control system, resulting in incorrect decisions and adjustments, ultimately impacting the stability and efficiency of the entire glass production process. In such cases, the accuracy of the temperature sensor's monitoring data is low, and decisions made by the control system based on this data may lead to equipment damage, increased energy consumption, or decreased product quality.

[0059] A lower signal-to-noise ratio (SNR) indicates higher accuracy of the temperature sensor's monitoring data. A low SNR signifies lower noise content and entropy in the signal, meaning the signal is relatively stable, clear, and less susceptible to external interference or sensor malfunction. In this case, the temperature sensor can provide more accurate temperature data, allowing the control system to make more effective adjustments based on this accurate data, thus ensuring the precision and consistency of temperature control in the glass production process. This not only improves production efficiency but also reduces energy waste and equipment wear, ensuring product quality meets standards. Therefore, a lower SNR indicates higher accuracy of the temperature sensor's monitoring data, contributing to improved overall production performance.

[0060] The voltage fluctuation frequency anomaly index and signal noise interference index are converted into a comprehensive feature vector. This comprehensive feature vector is used as the input to a machine learning model. The machine learning model uses the weighted labels of the accuracy of temperature sensor monitoring data in each time period predicted by each set of comprehensive feature vectors as the prediction objective. The training objective is to minimize the sum of prediction errors of the weighted labels of the accuracy of temperature sensor monitoring data in all time periods. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The weighted labels of the accuracy of temperature sensor monitoring data in each time period are determined based on the model output. The machine learning model is a multinomial regression model, and the accuracy index of the monitoring data is obtained by weighted averaging and summing the weighted labels of the accuracy of temperature sensor monitoring data in each time period.

[0061] The method for obtaining the weights for the accuracy of temperature sensor monitoring data in each time period is as follows: Obtain the corresponding function expression from the training data of the comprehensive feature vector of the trained machine learning model. In the formula, This is the output function of the model, where AK is the voltage fluctuation frequency anomaly index and GH is the signal noise interference index. Assign weights to the accuracy of temperature sensor monitoring data for each time period.

[0062] S3: The oxygen concentration in the furnace is monitored in real time by an oxygen sensor, and the degree of abnormal change in the oxygen content of the furnace atmosphere is judged based on the fluctuation of the oxygen concentration in the furnace gas over a fixed period of time.

[0063] Oxygen sensors are used to measure the oxygen concentration in the furnace gas in real time and transmit the data to the control system. This data reflects the oxygen content in the furnace, thus indirectly inferring the atmosphere within the furnace. Changes in oxygen concentration directly affect the stability and quality of the glass melting process. Oxygen concentration data is typically a continuous real-time data stream. The sensor records and transmits oxygen concentration values ​​(such as percentage concentration or PPM value) at regular intervals. A fixed set of time periods is selected for data acquisition. For example, the data acquisition cycle can be divided into monitoring every few minutes or every hour. The choice of time period depends on the furnace's operating cycle, reaction time, and production process requirements. For example, time periods of 10 minutes, 30 minutes, or 1 hour can be selected to ensure coverage of fluctuations in the oxygen concentration within the furnace and to obtain sufficient time-series data for analysis.

[0064] An oxygen concentration fluctuation deviation index is generated by analyzing the oxygen concentration fluctuations in the furnace gas over a fixed time period to determine the degree of abnormal changes in the oxygen content of the furnace atmosphere. The method for obtaining the oxygen concentration fluctuation deviation index is as follows:

[0065] Collect oxygen concentration data over fixed time periods. This data should be recorded at fixed intervals (e.g., every minute, every hour). Set the oxygen concentration data as... Where t represents the timestamp, This represents the oxygen concentration data at time t.

[0066] Choose the order of the autoregressive moving average (ARMA) model. The ARMA model contains an autoregressive (AR) term and a moving average (MA) term, and the model is represented as follows: ;in, The coefficient of the AR term (autoregressive coefficient). The coefficient (moving average coefficient) of the MA term. For noise terms, In time The white noise residuals at time step 1 are used to determine the appropriate p and q orders using the autocorrelation function (ACF) and partial autocorrelation function (PACF).

[0067] Depending on the chosen orders p and q, the parameters of the model are estimated using either the least squares method or the maximum likelihood estimation method. and , Or q; use the fitted ARMA model to predict future oxygen concentrations, and calculate the difference between actual observations and model predictions, i.e.: ;in, For residuals, To calculate the oxygen concentration fluctuation deviation index based on the oxygen concentration value predicted by the ARMA model, the expression is as follows: In the formula, Y is the total number of data points, and KM is the oxygen concentration fluctuation deviation index.

[0068] A larger oxygen concentration fluctuation deviation index indicates a more severe abnormal change in the oxygen content of the furnace atmosphere. A larger index signifies more drastic fluctuations in the oxygen concentration within the furnace. This usually indicates a significant change in oxygen concentration over a given period, potentially caused by instability due to factors within the furnace (such as incomplete combustion, uneven oxygen supply, or changes in the external environment). Frequent fluctuations in oxygen concentration lead to unstable temperatures within the furnace, affecting the glass melting process and potentially resulting in substandard glass quality, equipment damage, or furnace malfunction. In this case, a large oxygen concentration fluctuation deviation index can be considered a signal of poor atmosphere control, indicating potential uneven oxygen supply or other abnormalities within the furnace, which can severely impact production efficiency and quality.

[0069] A smaller oxygen concentration fluctuation deviation index indicates a less severe abnormal change in the oxygen content of the furnace atmosphere. Conversely, a smaller index suggests a more stable oxygen concentration with smaller fluctuations. A stable oxygen concentration indicates precise atmosphere control within the furnace, with oxygen supply and exhaust maintained within a reasonable range, which is crucial for the glass melting process. In this case, the glass melting temperature is more stable, the melting reaction efficiency is higher, impurity removal is better, and the final product quality is more reliable. Therefore, a smaller oxygen concentration fluctuation deviation index generally means less fluctuation in the oxygen concentration within the furnace atmosphere, indicating a well-functioning oxygen control system, a stable production process, and a lower risk of abnormal changes.

[0070] S4: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted using fuzzy logic to optimize the melting process.

[0071] The accuracy index RY of monitoring data and the oxygen concentration fluctuation deviation index KM are used as input terms of fuzzy logic, and the heating power of the melting furnace is used as the output term.

[0072] The accuracy index (RY) for monitoring data typically ranges from 0 to 1 (a higher RY indicates more accurate data). Fuzzy definitions: Low Accuracy: RY is less than a certain threshold (e.g., 0.3). Medium Accuracy: RY is between 0.3 and 0.7. High Accuracy: RY is greater than 0.7.

[0073] The oxygen concentration fluctuation deviation index (KM) represents the degree of fluctuation in oxygen concentration, ranging from 0 to a certain maximum value. Generally, the larger the value, the more severe the fluctuation. Fuzzy definitions: Low Variation: KM is less than a certain threshold (e.g., 0.3). Medium Variation: KM is between 0.3 and 0.7. High Variation: KM is greater than 0.7.

[0074] The output variable is the furnace heating power, which is dynamically adjusted based on the inputs RY and KM. For simplicity, it is assumed that the heating power ranges from 0 to 100, representing the percentage of the furnace's heating power. Low Power: 0 - 30%; Medium Power: 30% - 70%; High Power: 70% - 100%.

[0075] Construct a fuzzy rule system based on RY and KM. Fuzzy rules connect the relationship between input and output variables through fuzzy logic. Here are some possible rules:

[0076] Rule 1: If RY is low accuracy and KM is high fluctuation, then the heating power is low. Rule 2: If RY is medium accuracy and KM is medium fluctuation, then the heating power is medium. Rule 3: If RY is high accuracy and KM is low fluctuation, then the heating power is high. Rule 4: If RY is low accuracy and KM is low fluctuation, then the heating power is medium. Rule 5: If RY is high accuracy and KM is high fluctuation, then the heating power is medium.

[0077] The measured RY and KM values ​​are converted into fuzzy sets. For example, for RY=0.6 and KM=0.4, their fuzzy values ​​are determined according to the defined membership functions.

[0078] Based on the defined fuzzy rules, a fuzzy value for the heating power is derived through a reasoning process. The reasoning process combines the rules and inputs through fuzzy operations (such as minimum value and weighted average) to generate the fuzzy output result.

[0079] The reasoning process yields a fuzzy heating power value, for example, a fuzzy value within the "medium power" range. To transform this fuzzy result into a specific heating power value, defuzzification is required. A commonly used defuzzification method is the weighted average method.

[0080] After defuzzification, a specific heating power value is obtained, which represents the adjustment value of the heating power in the melting furnace.

[0081] The defuzzified heating power value is fed back to the automated control system, which then adjusts the heating power in the melting furnace based on this value, thereby optimizing the glass melting process. Adjustments to the heating power affect the furnace temperature control, thus improving glass melting efficiency and the quality of the final product.

[0082] In a fuzzy logic system, by analyzing the accuracy (RY) of the temperature sensor's monitoring data and the deviation of oxygen concentration fluctuations (KM), a specific heating power value is finally obtained through fuzzy inference and defuzzification processes. For example, the defuzzified heating power might be 68% (i.e., the heating power of the furnace should be adjusted to 68%).

[0083] The defuzzified heating power value is fed back to the automated control system. This feedback is a real-time control signal representing the power output required by the heating system. The automated control system uses this value as a control variable and compares it with the current furnace heating power to determine whether the power output of the heating equipment needs to be adjusted.

[0084] The automated control system adjusts the furnace temperature by controlling the power output of the heating equipment based on the current heating power and the target heating power (the defuzzified value). Adjustments may be made for: electric heaters in the furnace: increasing or decreasing the power input to raise or lower the heating power; and gas flow in the gas furnace: regulating the gas supply to achieve the required heating power. The control system employs precise feedback mechanisms (such as PID controllers) to adjust the heating power, ensuring the furnace temperature remains stable within the optimal range and preventing glass quality problems caused by excessive temperature fluctuations.

[0085] By precisely adjusting the heating power, the temperature inside the melting furnace can be gradually increased or decreased as needed, thereby optimizing the glass melting process: maintaining the furnace temperature within the optimal range for glass melting (typically 1300°C to 1500°C) ensures uniform glass melting without bubbles or impurities. If the glass melting rate is slow or the oxygen concentration fluctuates significantly, increasing the heating power can accelerate the melting reaction and improve production efficiency. Dynamically adjusting the heating power based on the accuracy of monitoring data and fluctuations in oxygen concentration helps reduce energy waste and avoids excessive energy consumption due to excessively high or low temperatures. By optimizing the heating power and controlling the oxygen concentration, the uniformity of glass composition can be ensured, preventing a decline in the quality of the finished glass product due to temperature instability or atmospheric fluctuations.

[0086] The control process is real-time, therefore the automation system needs to continuously monitor data from temperature sensors, oxygen sensors, and the actual output of heating power. The system periodically acquires this data and re-runs the fuzzy logic control to dynamically adjust the heating power. This closed-loop feedback control helps to cope with any changes in production, such as variations in raw materials, fluctuations in equipment performance, or changes in environmental conditions.

[0087] In this embodiment, a voltage monitor is installed on the power line of the temperature sensor, a noise monitor is installed in the communication line between the sensor and the control system, and the heating power of the heater is monitored in real time using a power meter. The voltage fluctuations and signal noise data of the temperature sensor power supply are acquired and analyzed. Then, the weighted values ​​of the accuracy of the temperature sensor monitoring data in each time period are calculated, and a weighted average is used to obtain the monitoring data accuracy index. Simultaneously, an oxygen sensor is used to monitor changes in oxygen concentration inside the furnace to identify abnormal changes in the oxygen content of the furnace atmosphere. Combining the monitoring data accuracy index and oxygen concentration fluctuations, fuzzy logic is used to adjust the heating power, thereby optimizing the glass melting process, improving production efficiency, reducing energy waste, and ensuring the stability of glass quality.

[0088] Example 2: A control system for the glass manufacturing process described in this example includes a sensor monitoring module, a data analysis module, an oxygen concentration monitoring module, and a fuzzy logic control module.

[0089] Sensor monitoring module: A voltage monitor is installed on the power line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and a noise monitor is installed in the communication line between the temperature sensor and the control system to monitor the noise data in signal transmission in real time, and the heating power output of the heater is monitored in real time through a power meter.

[0090] Data Analysis Module: Acquires power supply voltage fluctuation data and noise data in signal transmission of temperature sensors over several time periods. After analysis, determines the weighting of the accuracy of temperature sensor monitoring data in each time period. The accuracy index of monitoring data is obtained by weighted averaging and summing the weighting of the accuracy of temperature sensor monitoring data in each time period.

[0091] Oxygen concentration monitoring module: It monitors the oxygen concentration of the gas in the furnace in real time through an oxygen sensor, and judges the degree of abnormal change in the oxygen content of the atmosphere in the furnace based on the fluctuation of the oxygen concentration of the gas in the furnace within a fixed time period.

[0092] Fuzzy logic control module: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted after applying fuzzy logic to optimize the melting process.

[0093] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A control method in the manufacturing process of glass products, characterized in that: Includes the following steps; S1: Install a voltage monitor on the power supply line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and install a noise monitor in the communication line between the temperature sensor and the control system to monitor the noise data in the signal transmission in real time, and monitor the heating power output of the heater in real time through a power meter. S2: Obtain the power supply voltage fluctuation data and signal transmission noise data of the temperature sensor in several time periods, analyze them, determine the weight of the accuracy of the temperature sensor monitoring data in each time period, and calculate the monitoring data accuracy index by weighted average summation of the weight of the accuracy of the temperature sensor monitoring data in each time period. In S2, after analyzing the fluctuation data of the temperature sensor power supply voltage, a voltage fluctuation frequency anomaly index is generated. The method for obtaining the voltage fluctuation frequency anomaly index is as follows: Collect voltage fluctuation data within the Q time period Where v(s) is the value of the voltage signal at time s, and N is the number of sampling points. The signal in the time domain is converted into a frequency domain signal by Fast Fourier Transform (FFT). The FFT converts the voltage signal V(s) into the frequency domain signal V(f), and its calculation formula is as follows: ; It is frequency The corresponding complex Fourier coefficients, where j is the imaginary unit. ; Here, T is the corresponding frequency, T is the signal sampling period, and k is the frequency index; to analyze the frequency components of the voltage fluctuation signal, its amplitude spectrum is calculated. That is, the energy distribution at different frequencies, expressed as: In the formula, Re( ) and Im They are respectively The real and imaginary parts of the spectrum; calculate the total energy of the entire spectrum. As a reference value, the expression is: ; Identify amplitude value If the amplitude is greater than the normal fluctuation range, Mark these frequencies as anomalous frequencies and calculate their energy contribution. The expression is: The voltage fluctuation frequency anomaly index is calculated using the following expression: In the formula, AK is the voltage fluctuation frequency anomaly index; S3: The oxygen concentration in the furnace is monitored in real time by an oxygen sensor, and the degree of abnormal change in the oxygen content of the furnace atmosphere is judged based on the fluctuation of the oxygen concentration in the furnace gas within a fixed time period. S4: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted using fuzzy logic to optimize the melting process.

2. The control method in the glass product manufacturing process according to claim 1, characterized in that: In S2, the signal collected by the temperature sensor is discretized into a discrete numerical sequence, and the signal is set. Given a signal sequence of length Z, divide it into discrete intervals to obtain discrete values. Define the intervals. To represent different discrete values, calculate the probability of each discrete value in the signal X, forming the probability distribution P(X) of the signal, and statistically analyze each discrete value in the signal. Frequency of occurrence Calculate the probability of each discrete value. The expression is: The information entropy is calculated based on the probability distribution of the signal. The calculation expression is as follows: ;in, The entropy of signal X is calculated by comparing the entropy of the original signal with that of the denoised signal. The signal noise interference index is then calculated, and the entropy H(X) of the entire temperature sensor signal sequence is calculated. Noise suppression is then applied to the signal to obtain the denoised signal sequence. Then calculate its entropy. Next, the signal-to-noise interference index is calculated, expressed as: In the formula, This represents the signal-to-noise interference index.

3. The control method in the glass product manufacturing process according to claim 2, characterized in that: In S2, the voltage fluctuation frequency anomaly index and signal noise interference index are converted into a comprehensive feature vector. This comprehensive feature vector is used as the input to the machine learning model. The machine learning model uses the weighted labels of the accuracy of temperature sensor monitoring data in each time period predicted by each set of comprehensive feature vectors as the prediction objective. The training objective is to minimize the sum of prediction errors of the weighted labels of the accuracy of temperature sensor monitoring data in all time periods. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The weighted labels of the accuracy of temperature sensor monitoring data in each time period are determined based on the model output. The machine learning model is a multinomial regression model, and the accuracy index of the monitoring data is obtained by weighted averaging and summing the weighted labels of the accuracy of temperature sensor monitoring data in each time period.

4. The control method in the glass product manufacturing process according to claim 3, characterized in that: In S3, after analyzing the fluctuation of oxygen concentration in the furnace gas over a fixed time period, an oxygen concentration fluctuation deviation index is generated to determine the degree of abnormal change in the oxygen content of the furnace atmosphere. The method for obtaining the oxygen concentration fluctuation deviation index is as follows: Collect oxygen concentration data within a fixed time period, and set the oxygen concentration data as... Where t represents the timestamp, Given the oxygen concentration data at time t; select the order of the autoregressive moving average (ARMA) model. The ARMA model includes an autoregressive (AR) term and a moving average (MA) term, and the model is expressed as: ;in, The coefficient of the AR term, The coefficient of the MA term. For noise terms, In time The white noise residuals at time step 1 are used to determine the orders p and q using the autocorrelation function and partial autocorrelation function; based on the chosen orders p and q, the model parameters are estimated. and , Or q; use the fitted ARMA model to predict future oxygen concentrations, and calculate the difference between actual observations and model predictions, i.e.: ;in, For residuals, To calculate the oxygen concentration fluctuation deviation index based on the oxygen concentration value predicted by the ARMA model, the expression is as follows: In the formula, Y is the total number of data points, and KM is the oxygen concentration fluctuation deviation index.

5. The control method in the glass product manufacturing process according to claim 4, characterized in that: In S4, based on the calculated accuracy index of the monitoring data and the fluctuation of oxygen concentration in the furnace gas, fuzzy logic is used to dynamically adjust the heating power and optimize the melting process. Specifically: The accuracy index RY of monitoring data and the oxygen concentration fluctuation deviation index KM are used as input terms of fuzzy logic, and the heating power of the melting furnace is used as the output term. The accuracy index of monitoring data, the oxygen concentration fluctuation deviation index, and the heating power of the melting furnace were all fuzzed. Construct fuzzy rules based on RY and KM, which connect the relationship between input and output variables through fuzzy logic; The measured RY and KM values ​​are converted into fuzzy sets; Based on the defined fuzzy rules, the fuzzy value of the heating power is obtained through a reasoning process; During the reasoning process, a fuzzy heating power value is obtained, and at the same time, defuzzification is performed to obtain a specific heating power value. The defuzzified heating power value is fed back to the automated control system, which then adjusts the heating power in the furnace according to the value, thereby optimizing the glass melting process.

6. A control system for a glass product manufacturing process, used to implement the control method for a glass product manufacturing process according to any one of claims 1-5, characterized in that: It includes a sensor monitoring module, a data analysis module, an oxygen concentration monitoring module, and a fuzzy logic control module; Sensor monitoring module: A voltage monitor is installed on the power line of the temperature sensor to detect the fluctuation data of the power supply voltage in real time, and a noise monitor is installed in the communication line between the temperature sensor and the control system to monitor the noise data in signal transmission in real time, and the heating power output of the heater is monitored in real time through a power meter. Data Analysis Module: Acquires power supply voltage fluctuation data and noise data in signal transmission of temperature sensors over several time periods. After analysis, determines the weighting of the accuracy of temperature sensor monitoring data in each time period. The accuracy index of monitoring data is obtained by weighted averaging and summing the weighting of the accuracy of temperature sensor monitoring data in each time period. Oxygen concentration monitoring module: It monitors the oxygen concentration of the gas in the furnace in real time through an oxygen sensor, and judges the degree of abnormal change in the oxygen content of the atmosphere in the furnace based on the fluctuation of the oxygen concentration of the gas in the furnace within a fixed time period. Fuzzy logic control module: Based on the accuracy index of the calculated monitoring data and the fluctuation of oxygen concentration in the furnace gas, the heating power is dynamically adjusted after applying fuzzy logic to optimize the melting process.

Images