Online monitoring method and system for process parameters of waste sand regeneration production line

Abstract

The invention discloses an online monitoring method and system for process parameters of a waste sand regeneration production line, and relates to the technical field of casting. According to the online monitoring method and system for the process parameters of the waste sand regeneration production line, multi-parameter real-time sensing of collaborative calibration, trend prediction based on a parameter coupling model and dynamic self-adaptive threshold adjustment are combined; high-precision and stable closed-loop control of the waste sand regeneration process is realized, so that the quality consistency and the production efficiency of regenerated sand are improved, and the accuracy and the reliability of online monitoring data are remarkably improved by introducing a collaborative calibration mechanism among parameters. According to the method, the parameter coupling model is constructed, real-time monitoring data is converted into prediction of a future process trend, and a self-adaptive threshold value is dynamically generated based on the prediction, so that process control is converted into active prevention from traditional passive response, quality risks such as waste sand overburning or insufficient regeneration are effectively avoided, and the stability of the production process is greatly enhanced.

Description

Technical Field

[0001] This invention relates to the field of casting technology, and more specifically to an online monitoring method and system for process parameters of a waste sand recycling production line. Background Technology

[0002] In casting production, molding sand is the primary molding material for making casting molds and cores. After casting is completed, a large amount of molding sand becomes waste sand. To conserve resources, reduce costs, and protect the environment, recycling waste sand to restore its usability is an indispensable key step in modern foundry enterprises. Waste sand recycling production lines typically include processes such as crushing, magnetic separation, roasting, cooling, and micron powder removal. The stability of these process parameters directly determines the quality of the recycled sand, which in turn affects the quality of subsequent castings.

[0003] Existing waste sand recycling production lines typically monitor process parameters by installing independent sensors at key workstations to monitor single parameters such as temperature and humidity online. Control systems mostly operate based on preset fixed thresholds; when a monitored value exceeds the set range, it triggers an adjustment mechanism or alarm. Furthermore, for complex parameters such as particle size distribution and binder residue, manual sampling at regular intervals is often required, with samples sent to a laboratory for offline analysis, resulting in significant time lag in the results.

[0004] Existing technologies have several shortcomings. First, the monitoring of each parameter is conducted in isolation, failing to consider the physicochemical coupling relationships between parameters such as temperature, humidity, and binder residue. Furthermore, measurements from a single sensor are easily affected by interference from other operating conditions, leading to deviations. Second, control methods based on fixed thresholds are passive responses, only correcting errors after they occur. They cannot anticipate changes in process trends, easily causing overshooting or fluctuations in process parameters, affecting the stability of recycled sand quality. Finally, parameters relying on offline monitoring cannot provide effective guidance for real-time control, resulting in significant delays in production adjustments and difficulty in responding to unexpected situations such as raw material fluctuations. Summary of the Invention

[0005] The purpose of this invention is to provide an online monitoring method and system for process parameters of a waste sand recycling production line. By combining multi-parameter real-time sensing with collaborative calibration, trend prediction based on parameter coupling model, and dynamic adaptive threshold adjustment, it can achieve high-precision and stable closed-loop control of the waste sand recycling process, thus solving the problems existing in the background technology.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The first aspect of the present invention provides an online monitoring method for process parameters of a waste sand recycling production line, including: S1, acquiring temperature parameters, humidity parameters, particle size parameters and binder residue parameters of multiple key nodes on the waste sand recycling production line, and integrating them to obtain an original parameter set.

[0007] S2. Perform collaborative calibration on the original parameter set to eliminate measurement interference between parameters and generate a calibrated parameter set.

[0008] S3. Perform correlation analysis on the input parameter coupling model of the calibrated parameter set to generate parameter trend prediction results.

[0009] S4. Based on the predicted trend of the parameters, dynamically adjust the control threshold of the process parameters to generate an adaptive threshold.

[0010] S5. Based on the adaptive threshold, generate and send control commands to the actuator to adjust the waste sand recycling process.

[0011] A second aspect of the present invention provides a system for implementing the online monitoring method for process parameters of the waste sand recycling production line described in the present invention, comprising: an integrated sensor module for acquiring the original parameter set.

[0012] The central processing unit, connected to the integrated sensor module, is used to generate the calibrated parameter set, the parameter trend prediction results, the adaptive threshold, and the control commands.

[0013] An actuator, connected to the central processing unit, is used to receive and execute the control commands to adjust the waste sand recycling process.

[0014] An alarm device, connected to the central processing unit, is used to trigger an alarm upon receiving a trigger signal based on a verification conclusion.

[0015] The beneficial effects of the present invention are as follows: (1) The present invention achieves high-precision and stable closed-loop control of the waste sand regeneration process by combining multi-parameter real-time sensing with collaborative calibration, trend prediction based on parameter coupling model and dynamic adaptive threshold adjustment, thereby improving the quality consistency and production efficiency of regenerated sand.

[0016] (2) By introducing a collaborative calibration mechanism between parameters, especially by using temperature parameters to compensate and correct humidity parameters, this invention significantly improves the accuracy and reliability of online monitoring data. This method overcomes the shortcomings of traditional independent sensors that are susceptible to cross-interference under complex conditions such as high temperature and high humidity, providing a high-quality data foundation for subsequent precise analysis and control, and ensuring that the decision-making source of the entire monitoring system is authentic and reliable.

[0017] (3) This invention transforms real-time monitoring data into predictions of future process trends by constructing a parameter coupling model, and dynamically generates adaptive thresholds based on this. This core innovation transforms process control from traditional passive response to proactive prevention, enabling pre-adjustment before substantial deviations in process parameters occur, effectively avoiding quality risks such as over-burning of waste sand or insufficient regeneration, thereby significantly enhancing the stability of the production process.

[0018] (4) This invention also integrates intelligent alarm verification and model self-optimization mechanisms. The multi-parameter consistency verification before alarms effectively filters out false alarms caused by single-point sensor failures, reducing unnecessary production interruptions. The model's periodic self-updating capability ensures that the system can adapt to the evolution of production conditions over a long period of time. This self-diagnosis and self-evolution feature greatly improves the robustness and long-term effectiveness of the entire online monitoring method, realizing intelligent and highly reliable operation and maintenance of the production process. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention.

[0021] Figure 2 This is a schematic diagram of the system structure connection of the present invention.

[0022] Figure 3 This is a schematic diagram of the parameter coupling model of the present invention. Detailed Implementation

[0023] 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, and 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.

[0024] Reference Figure 1 As shown, the first aspect of the present invention provides an online monitoring method for process parameters of a waste sand recycling production line, comprising: S1, acquiring temperature parameters, humidity parameters, particle size parameters and binder residue parameters of multiple key nodes on the waste sand recycling production line, and integrating them to obtain an original parameter set.

[0025] In a specific embodiment of the present invention, the step of integrating to obtain the original parameter set includes: reading real-time monitoring data from integrated sensor modules deployed at key nodes of the waste sand recycling production line, including but not limited to the crushing section, heating section and cooling section.

[0026] Specifically, the integrated sensor module is a physically integrated unit containing multiple sensors to ensure synchronous multi-dimensional parameter measurements of the same batch of waste sand samples, thereby eliminating data misalignment caused by spatial and event differences.

[0027] During the data acquisition phase, the sensors within the integrated sensor module work in parallel, and temperature and humidity parameters are acquired in real time through high-precision thermocouples and capacitive humidity sensors.

[0028] The particle size parameters are obtained using laser scattering technology. This technology involves emitting a collimated laser beam into a flowing waste sand sample and measuring the distribution of scattered light energy by the particles. Then, the complete particle size distribution characteristics are derived from the scattering spectrum, and finally, representative particle size parameters are generated.

[0029] For measuring the adhesive residue parameter, near-infrared spectroscopy is used. This technique uses near-infrared light of a specific wavelength to irradiate the waste sand. The organic molecules in the adhesive will absorb the light of a specific wavelength. The residual amount of adhesive is quantitatively calculated by analyzing the absorption peak intensity of the reflection spectrum.

[0030] Data collected by all sensors at the same time, including temperature, humidity, particle size parameters, and adhesive residue parameters, is integrated into a single data vector. The data vector obtained from the i-th key node at time t can be represented as... ,in, For temperature parameters, For humidity parameters, For particle size parameters, To determine the adhesive residue parameter, the data vectors generated at all key nodes at each time point are collected to form the original parameter set of the time series. Where N is the total number of key nodes, the original parameter set provides a high-fidelity data foundation for subsequent calibration and correlation analysis.

[0031] S2. Perform collaborative calibration on the original parameter set to eliminate measurement interference between parameters and generate a calibrated parameter set.

[0032] In a specific embodiment of the present invention, the step of generating the calibrated parameter set includes: receiving the original parameter set, and independently performing error analysis on the temperature parameter, humidity parameter, particle size parameter, and binder residue parameter in the original parameter set based on a preset error model, wherein the error model can be specifically expressed as: ,in It is one of the following parameters: temperature, humidity, particle size, and adhesive residue. For input The corresponding temporary value after error analysis, , The inputs are respectively The corresponding sensor gain correction coefficient and zero-point offset correction coefficient.

[0033] Specifically, the gain correction coefficients and zero-point offset correction coefficients of the various sensors are dynamically updated, not static. Their determination can be broken down into three steps: First, based on the sensor's factory benchmark, the manufacturer calibrates it using professional equipment, providing initial values ​​'a' and 'b' adapted to the sensor itself as baseline values. Next, in conjunction with on-site calibration under the waste sand recycling production line conditions, using standard equipment of sufficient accuracy, such as a standard thermometer, and the production line sensors, the same process node parameter, such as temperature, is measured simultaneously. The actual values ​​are compared with the sensor readings, and 'a' and 'b' are adjusted to adapt to the high-temperature, dusty, and other on-site environments. Finally, optimization is performed using historical production line data. Based on accumulated sensor readings, recycled sand quality, and other data, 'a' and 'b' are continuously fine-tuned to ensure accurate correction of the original parameters.

[0034] The gain correction coefficient corrects the amplification ratio of the sensor signal. For example, if the sensor amplifies the actual signal by 100 times, the reading will be too high if the amplification is too great, and a will be less than 1 if the amplification is too small, and a will be greater than 1 if the amplification is too small. The typical range is 0.95-1.05. The zero-point offset coefficient corrects the basic deviation when the sensor has no signal. For example, when the actual value of the sensor is 0, the reading is b, which may be positive or negative. Therefore, the unit of b is consistent with the measurement parameters, such as temperature is ℃, humidity is %RH, and particle size is μm.

[0035] Example 1: The determination of the gain correction coefficient and zero-point offset correction coefficient of the thermocouple temperature sensor in the heating section is as follows: Temperature monitoring is required in the waste sand recycling heating section, with a common range of 50-300℃. Assuming a certain type of thermocouple sensor is used, the on-site calibration steps are as follows: Using a standard thermometer with an accuracy of ±0.1℃ (i.e., industry-standard calibration equipment), simultaneously collect two sets of data at the same measuring point in the heating section: Actual temperature 100℃, sensor initial reading 101℃; Actual temperature 200℃, sensor initial reading 201℃.

[0036] By substituting into the formula and solving the system of equations, we get: gain correction coefficient a = 1.0, indicating that the sensor amplification ratio has no deviation; zero-point offset correction coefficient b = -1, indicating that the sensor reading is always 1℃ higher than the true value.

[0037] Verification: If the subsequent sensor reading is 150℃, after correction it will be 149℃, which is consistent with the actual temperature.

[0038] Example 2: The determination of the gain correction coefficient and zero-point offset correction coefficient of the capacitive humidity sensor in the cooling section is as follows: The humidity in the cooling section is easily affected by temperature. Assuming a capacitive humidity sensor is used with a measurement range of 0-100%RH, the coefficients a and b are determined by calibrating at different temperature points, and a temperature-humidity sensor coefficient correlation table is established as shown in Table 1.

[0039] Table 1. Correlation Table of Temperature-Humidity Sensor Coefficients

[0040] It should be noted that temperature compensation is performed when the gain correction coefficient and zero offset correction coefficient of the humidity sensor are determined. Essentially, this is to make the coefficients adapt to the actual operating temperature of the humidity sensor, so as to avoid using the gain correction coefficient and zero offset correction coefficient of 25℃ to correct the original humidity of 35℃, and reduce the deviation from the true value.

[0041] It should be added that historical production line data is used for optimization. Based on accumulated sensor readings, recycled sand quality, and other data, a and b are continuously fine-tuned to ensure accurate correction of the original parameters. The historical production line data includes at least: 1. Raw sensor data: Real-time raw readings of the corresponding sensors for each batch of production, with timestamps and production batch numbers. 2. Temporary error analysis values: Temporary values ​​calculated based on the current a and b after error analysis. 3. Recycled sand quality verification data: Offline test results for each batch of recycled sand, such as the moisture content of recycled sand related to humidity parameters, and the strength of recycled sand related to binder residue parameters, used to reverse-verify the accuracy of the temporary values ​​after error analysis, and thus determine whether a and b need adjustment.

[0042] Fine-tuning of a and b is triggered when any of the following conditions are met: 1. In production batches exceeding the first preset quantity, the average deviation between the temporary value of a sensor after error analysis and the measured value of the standard equipment on site increases by more than 10% compared to the previous calibration. For example, the deviation between the temporary value of the humidity sensor after error analysis and the measured value of the standard humidity generator increases from 0.5%RH to 0.55%RH. 2. In production batches exceeding the second preset quantity, the temporary value of a parameter, such as the amount of adhesive residue, is normal after error analysis, but the quality of the recycled sand, such as the fluctuation of sand particle strength, exceeds the process requirement range, such as the strength deviation > 2MPa. This indicates that the temporary values ​​corrected by a and b do not fully match the actual process and need to be optimized in reverse by combining quality data.

[0043] When fine-tuning a and b, the goal is to minimize the deviation between the temporary value after error analysis and the actual value, and a linear regression fitting method is used.

[0044] Using the temperature parameters from the original parameter set, temperature compensation calculations are performed on the humidity parameters at the same acquisition time to obtain the corrected humidity parameters. The correction process is expressed as follows: ,in The corrected humidity parameters are as follows. These are temporary values ​​for the humidity parameter after error analysis. These are temporary values ​​of temperature parameters measured at the same time and location, after error analysis. This is a temperature compensation function that reflects the relationship between humidity measurement error and temperature. Specifically, it is obtained in advance through fitting experimental data to ensure the accuracy of humidity readings at different operating temperatures.

[0045] The calibrated humidity parameter is combined with other parameters in the original parameter set to generate the calibrated parameter set.

[0046] The technical effect of this method is a significant improvement in the accuracy and reliability of online monitoring data. By combining single-item correction with a pre-set error model and cross-correction using the physical correlation between parameters, this scheme not only corrects individual sensor errors but, more importantly, eliminates cross-interference in multi-parameter measurement environments, particularly addressing the industry challenge of the impact of high temperature on humidity measurement. This collaborative correction mechanism produces a superposition effect beyond simple error correction, enabling the data output of the entire multi-parameter sensor network to form an internally self-consistent and highly reliable whole. The final calibrated parameter set more realistically reflects the actual physicochemical state of waste sand during the regeneration process, thus providing high-quality data input for subsequent process trend prediction and adaptive control, and is the cornerstone of ensuring the effectiveness of the entire online monitoring method.

[0047] S3. Perform correlation analysis on the input parameter coupling model of the calibrated parameter set to generate parameter trend prediction results.

[0048] Reference Figure 3 As shown, in a specific embodiment of the present invention, the step of generating parameter trend prediction results includes: inputting the calibrated parameter set into a model that quantifies the dynamic coupling relationship between each parameter, i.e., a parameter coupling model; using the model that quantifies the dynamic coupling relationship between each parameter, deducing the change trajectory of each parameter in the calibrated parameter set within a future preset time, i.e., a prediction curve, to generate the parameter trend prediction results. For example, the future preset time for the crushing section and cooling section is 3-5 minutes, and the future preset time for the heating section is 5-8 minutes.

[0049] The parameter coupling model is specifically trained based on historical process data.

[0050] The parameter coupling model is a fusion model of physical mechanism and data-driven approach. The physical mechanism layer can construct particle size distribution models of waste sand in the crushing section and heat and mass transfer models in the heating section based on the laws of mass and energy conservation, such as Fourier's law and Fick's law, as well as pyrolysis kinetic models of binders like the Arrhenius equation, thereby establishing preliminary quantitative relationships between various parameters. The data-driven layer can employ deep learning models, such as Long Short-Term Memory (LSTM) networks or Transformer models, or tree-based models like XG Boost or Light GBM, to learn from historical process data and correct biases in the physical mechanism model, capturing nonlinear, high-dimensional, and complex correlations. The fusion mechanism can employ multi-task learning, model stacking, or hybrid expert systems based on gated units to combine the advantages of mechanistic knowledge and data-driven approaches.

[0051] Specifically, the parameter coupling model is a mathematical model that can describe and quantify the interaction between various parameters. The historical process data includes at least: historical monitoring data of sensors at key nodes, such as time series data of temperature, humidity, particle size, and binder residue; process setting parameters for the corresponding time period, such as heating section power and cooling section air volume; and quality test results of the final recycled sand, such as moisture content, compressive strength, and particle size qualification rate. The data must cover typical operating conditions of the waste sand recycling production line, such as different initial moisture contents of waste sand and different production loads, so that the model can capture the complex dynamic coupling relationship in the real production environment.

[0052] Specifically, within the model, the input calibrated parameter set is analyzed through its solidified structure and parameters. The core of the model is to identify and calculate the correlation between different parameter change rates. For example, it can quantify the rate at which the humidity parameter will decrease when the temperature parameter increases at a certain rate, and the trend of the decay curve of the adhesive residue parameter and the distribution change of the particle size parameter.

[0053] The above process can be represented as: ,in These are the predicted parameter values ​​output by the model. This represents the parameter coupling model itself. It is the set of calibrated parameters input at the current moment. This represents the coupling relationship knowledge that the model learns from historical process data.

[0054] Based on this profound understanding of dynamic coupling relationships, the model extrapolates forward, generating parameter trend predictions for a predetermined time period. These predictions, presented as a series of curves, clearly demonstrate the possible trajectory of each parameter's change over a future period, such as the next five minutes, under the current operating conditions. These parameter trend predictions provide crucial decision-making support for subsequent adaptive control and risk warning.

[0055] The technical effect of this method is a fundamental shift from post-event monitoring to pre-event prediction. By introducing a parameter coupling model trained on historical data, this approach no longer views each process parameter in isolation, but rather as an interconnected and organic whole. This analysis of dynamic coupling relationships enables the system to anticipate potential future chain reactions triggered by current operations. This transcends the limitations of traditional monitoring methods that can only reflect the current situation. The resulting parameter trend predictions transform process control from passive correction to proactive and predictive optimization, laying the foundation for higher levels of production automation and intelligence.

[0056] S4. Based on the predicted trend of the parameters, dynamically adjust the control threshold of the process parameters to generate an adaptive threshold.

[0057] In a specific embodiment of the present invention, step S4, the step of generating an adaptive threshold includes: comparing the predicted parameter trend with a preset process safety and quality boundary; when the predicted parameter trend exceeds the process safety and quality boundary, it is determined that there is a potential process anomaly risk; identifying the risk type; and quantifying the process anomaly risk index according to the magnitude and speed of the deviation.

[0058] The risk types include over-burning risk, insufficient roasting risk, excessive cooling (i.e., excessive humidity risk), excessive drying (i.e., insufficient humidity risk), excessively coarse particle size risk, and excessively fine particle size risk.

[0059] Specifically, process anomaly risk indicators It can be represented as ,in It is the maximum deviation between the predicted parameter value and the process safety and quality boundary. It is the predicted maximum rate of change. and It is a normalization function, and w1 and w2 are the weighting coefficients for deviation magnitude and speed, respectively.

[0060] Based on the type and indicators of the process anomaly risk, the dynamic process parameter control boundary is calculated, as shown in the following formula: ,in, This is the dynamically adjusted new upper limit threshold. The default standard upper limit threshold, As a quantitative indicator of process anomaly risk, It is an adjustable sensitivity coefficient, with dimensions consistent with the parameter being adjusted.

[0061] For example, if an overheating risk is identified, the upper temperature threshold for the current control cycle is lowered, thereby tightening the control range in advance and increasing the safety margin.

[0062] The dynamic process parameter control boundary is set as the adaptive threshold.

[0063] The technical effect of this method is to upgrade the process control paradigm from passive response to proactive prevention. By combining parameter prediction capabilities with a threshold adjustment mechanism, the system no longer waits for deviations to actually occur before correcting them, but instead mitigates risks by adjusting control boundaries at the nascent stage of deviations. This dynamic, feedforward control approach cannot be achieved by either prediction technology or control technology alone; the combination of the two produces a synergistic effect, namely, achieving predictive and stable control of the process. The resulting adaptive threshold allows the production line to anticipate problems and respond proactively, thereby greatly improving the stability of recycled sand quality and the safety of the production process, effectively avoiding defective products or downtime events caused by uncontrolled process parameters.

[0064] S5. Based on the adaptive threshold, generate and send control commands to the actuator to adjust the waste sand recycling process.

[0065] Specifically, this step is a key step in achieving closed-loop control, transforming the generated adaptive threshold into actual physical control of the production line. This process takes as input real-time sensor monitoring data from the sensor network and the adaptive threshold generated in the previous step.

[0066] In a specific embodiment of the present invention, the step of generating and sending control commands includes: continuously comparing real-time acquired sensor monitoring data with a corresponding adaptive threshold, wherein the adaptive threshold includes an upper limit value and a lower limit value.

[0067] When the sensor-monitored data exceeds the adaptive threshold, an adjustment command is calculated and generated, the specific process of which is as follows: ,in This is the difference between the current parameter monitoring value and the corresponding adaptive threshold, i.e., the error. and These are preset proportional and integral control coefficients used to ensure the speed and stability of the adjustment action. for The difference between the parameter monitoring value at time and the corresponding adaptive threshold, u(t) is the adjustment amount of the actuator. The calculated u(t) is encapsulated into a standardized adjustment instruction that can be recognized by the actuator.

[0068] The adjustment command is sent to the actuator, which includes a heater or a fan.

[0069] Specifically, the actuator is a physical device on the production line, such as a heater that controls heating power or a fan that regulates airflow. Upon receiving an adjustment command, the actuator will immediately execute it; for example, the heater will reduce its power, or the fan will increase its speed, thereby precisely fine-tuning the waste sand recycling process and pulling the deviated process parameters back to the optimal range defined by the adaptive threshold. For example, if the actuator is a heater, i.e., a device that controls temperature, such as… This indicates a 4% reduction in heater power. If the actuator is a fan, i.e., a device that controls humidity and temperature, such as... This means the fan airflow increases by 7%.

[0070] The technical effect of this method is to achieve precise, stable, and efficient closed-loop control of the production process. By combining real-time data with dynamic adaptive thresholds, the control system is no longer a simple mechanical response to a fixed target, but becomes an intelligent adjustment mechanism with predictive and adaptive capabilities. This prediction-based fine-tuning control can suppress process fluctuations with smaller adjustment amplitudes and faster response speeds, effectively avoiding parameter overshoot and oscillation phenomena common in traditional control methods. It seamlessly transforms the intelligent results of front-end monitoring, analysis, and prediction into precise physical actions of back-end actuators, truly opening up the entire link from data to control, ensuring a high degree of consistency in the quality of recycled sand products, and improving overall energy efficiency by reducing unnecessary energy fluctuations.

[0071] In a specific embodiment of the present invention, after generating the adaptive threshold, the method further includes: when the process abnormality risk index exceeds the risk threshold, starting an alarm delay program and generating a delay trigger signal.

[0072] It should be noted that when the calculated process anomaly risk index exceeds the preset risk threshold, an alarm is not triggered immediately. Instead, an alarm delay program is started, an internal delay trigger signal is generated, and a short timing window is opened, for example, a few seconds. The timing window provides the necessary time for subsequent verification steps.

[0073] During the alarm delay period, the parameter coupling model is used to verify the consistency of real-time sensor monitoring data and generate a verification conclusion, which indicates whether the real-time sensor monitoring data is consistent or inconsistent.

[0074] It should be noted that the established parameter coupling model is retrieved at this point, but its purpose is not prediction, but verification. The real-time parameter set obtained from the sensor network at the current moment, which includes the latest readings of all key parameters, is input into the model that quantifies the dynamic coupling relationship between the parameters. This parameter coupling model performs reverse verification based on the physicochemical correlation laws between parameters that it has learned internally. That is, based on the real-time value of one parameter, such as temperature, it calculates the theoretical values ​​of other related parameters, such as humidity and adhesive residue. Then, the calculated theoretical values ​​are compared with the real-time values ​​measured by the sensors. If the average similarity of all parameters is greater than or equal to a preset similarity threshold, it indicates that the data from each sensor are mutually supportive, and the process anomaly is systematic and real. If significant differences occur, such as extremely high temperature readings but no corresponding changes in other parameters, it indicates that the temperature sensor itself is likely malfunctioning. Based on the comparison results, a verification conclusion is generated, i.e., consistency or inconsistency. That is, when the average similarity is greater than or equal to the preset similarity threshold, the output of real-time sensor monitoring data is consistent; otherwise, the output is inconsistent.

[0075] The similarity can be measured by calculating the Euclidean distance, cosine similarity, or correlation coefficient between the measured values ​​of the sensor and the theoretical values ​​derived from the parameter coupling model. Then, the average similarity is obtained by weighting the similarity of all parameters.

[0076] Based on the verification results, control the triggering state of the alarm device.

[0077] Specifically, when the verification conclusion is that the real-time sensor monitoring data is consistent, it indicates that the process is indeed heading towards a dangerous state. The delay program is immediately stopped, and alarm devices such as audible and visual alarms are activated. When the verification conclusion is inconsistent, the main alarm is suppressed, and a maintenance alarm is generated for a specific sensor to notify the operator to check the sensor.

[0078] The technical effect of this method is a significant improvement in the robustness and reliability of the entire monitoring system. By combining predictive risk assessment with real-time multi-parameter consistency verification, this synergistic mechanism achieves a synergistic effect greater than the sum of its parts, meaning the parameter-coupled model is endowed with both predictive and diagnostic functions. This not only avoids unnecessary production interruptions and increases the effective uptime of the production line, but also enhances operators' trust in the automation system. When an alarm sounds, its reliability is extremely high, ensuring a rapid and accurate response to real process anomalies.

[0079] Specifically, this step introduces adaptive learning and continuous optimization capabilities into the entire online monitoring method, ensuring that the parameter-coupled model can evolve over time and maintain its long-term accuracy in prediction and diagnosis. This process is executed periodically or event-triggered, rather than in real time.

[0080] In a specific embodiment of the present invention, the method further includes: collecting real-time monitoring data and corresponding historical process data during the production process to form an updated dataset.

[0081] The parameter coupling model is retrained using the updated dataset to generate an updated parameter coupling model.

[0082] A smooth transition mechanism is used to replace the original parameter coupling model with the updated parameter coupling model for subsequent correlation analysis.

[0083] Specifically, the updated parameter coupling model is loaded into a spare memory area, and during the verification period, the old and new parameter coupling models run in parallel, but the new parameter coupling model is in shadow mode, and its output does not participate in actual control. By comparing the prediction results of the old and new parameter coupling models with the actual production data, and confirming that the performance of the new parameter coupling model is better than or equal to that of the old model, a seamless switch operation is performed. The main program's call pointer to the model is then set to the updated parameter coupling model. From this point on, all relevant functions of the entire online monitoring system, such as trend prediction and consistency verification, will be executed based on this updated parameter coupling model.

[0084] The confirmation that the performance of the new parameter coupling model is better than or equal to that of the old model is achieved by comparing the mean absolute error and root mean square error between the predicted values ​​and actual monitored values ​​of the new and old parameter coupling models for particle size parameters, adhesive residue parameters, temperature parameters, and humidity parameters.

[0085] The technical effect of this method is to endow the entire online monitoring system with the ability to self-evolve and adapt over the long term. By periodically training and iterating the core parameter coupling model with the latest production data, this solution ensures that the system's intelligent core does not become outdated or sluggish due to changes in production conditions. This data-driven self-optimization cycle is unparalleled by static models. It combines the advantages of real-time monitoring with the depth of offline learning, producing a synergistic effect—the continuous improvement of system performance. This makes the entire online monitoring method not only advanced in the initial deployment phase but also able to maintain its high accuracy and reliability over the long term, maximizing its adaptability to changes throughout the entire lifecycle of the production line and achieving true long-term intelligent operation and maintenance.

[0086] In a specific embodiment of the present invention, the method further includes: storing the calibrated parameter set, the parameter trend prediction results and the adaptive threshold, and constructing a historical database.

[0087] Specifically, the key data streams generated in previous steps are captured and archived in real time, including a precisely calibrated set of parameters representing the most accurate physicochemical state of the waste sand; parameter trend predictions generated by the parameter coupling model; and adaptive thresholds dynamically calculated to address risks. These data are systematically stored in a specially constructed relational database with timestamps and production batch numbers, forming a rich and queryable historical database where each record is fully linked to the actual state at a given moment, future predictions, and control decisions.

[0088] Based on the data correlation in the historical database, a quality report is generated. The quality report includes, but is not limited to, the accuracy of the parameter coupling model, the stability score of the production process, and the correlation between process parameter fluctuations and product quality.

[0089] It should be noted that the above process is automatically executed by the report generation module. It can query the historical database and perform a series of analytical operations according to a preset cycle. When analyzing the accuracy of the parameter coupling model, the mean absolute error (MAE) is used as the core indicator. The MAE of each parameter is calculated. If the MAE of each parameter conforms to the preset acceptable MAE range for the corresponding parameter, the accuracy of the parameter coupling model is deemed acceptable. When performing stability scoring of the production process: the number of times the adaptive threshold is triggered within a set time period is counted. The maximum deviation of the parameter from the threshold each time it is triggered. According to the scoring formula , , These are the weighting coefficients. Correlation coefficient analysis is used to examine the relationship between process parameter fluctuations and product quality: process parameter fluctuation data for a specific batch is extracted from the historical database and compared with the final quality inspection results of that batch. The correlation coefficient is calculated using a formula, with a range of (-1, 1). The closer the absolute value is to 1, the stronger the correlation.

[0090] The quality report is then output to a display device.

[0091] The technical benefits of this method include the visualization, traceability, and analyzability of process information, thus providing solid data support for management and optimization. By constructing a historical database that links real-time data, predictive information, and control decisions, this solution provides an evidentiary basis for post-event quality traceability and fault diagnosis. The generated quality report transforms complex dynamic data into an intuitive process report, which not only serves as proof of product quality but also as a crucial basis for continuous process improvement and control strategy optimization, enabling operators and managers to make data-driven decisions.

[0092] Reference Figure 2 As shown, a second aspect of the present invention provides a system for implementing an online monitoring method for process parameters of a waste sand recycling production line according to the present invention, comprising: an integrated sensor module for acquiring the original parameter set.

[0093] The central processing unit, connected to the integrated sensor module, is used to generate the calibrated parameter set, the parameter trend prediction results, the adaptive threshold, and the control commands.

[0094] An actuator, connected to the central processing unit, is used to receive and execute the control commands to adjust the waste sand recycling process.

[0095] An alarm device, connected to the central processing unit, is used to trigger an alarm upon receiving a trigger signal based on a verification conclusion.

[0096] This invention proposes a collaborative intelligent online monitoring method. Its technical principle lies in constructing a complete technical chain from multi-dimensional data perception to closed-loop control. First, an integrated multi-parameter sensor network acquires real-time snapshots of the waste sand's state at key process nodes, forming the raw data foundation. Next, the physical correlation between parameters is used for collaborative data calibration, generating a high-fidelity calibrated parameter set, providing accurate input for subsequent analysis. The core lies in introducing a parameter coupling model trained on historical data. This model deeply understands the dynamic interactions between various process parameters and transforms the real-time calibrated parameter set into accurate predictions of future process trends. Based on these predictions, the system no longer passively waits for deviations to occur but actively and dynamically adjusts the boundaries of process control, i.e., generating adaptive thresholds. Finally, based on this dynamically changing adaptive threshold, the system generates and issues precise control commands, using actuators to fine-tune the production process in real time, thus forming an intelligent closed-loop control system capable of anticipating risks, proactively avoiding them, and continuously self-optimizing.

[0097] This invention, through the aforementioned technical principles, achieves significant technical effects. First, by employing multi-parameter collaborative monitoring and calibration, the accuracy and reliability of online data are greatly improved, enabling a more realistic reflection of the real-time state of the recycled sand and laying a solid foundation for enhancing the consistency and pass rate of the final product quality. Second, the introduction of a parameter coupling model to achieve trend prediction transforms process control from a traditional reactive mode to an advanced preventative mode, effectively avoiding process anomalies such as overheating and insufficient regeneration caused by parameter fluctuations, significantly enhancing the stability and safety of the production process. Finally, the combination of prediction-based adaptive threshold adjustment and closed-loop control enables refined and intelligent control of the process, optimizing energy consumption and reducing unnecessary equipment intervention and downtime, comprehensively improving the overall operating efficiency and economic benefits of the waste sand recycling production line.

[0098] It should be added that the formulas mentioned above, through the principle of dimensional consistency and mathematical standardization methods such as normalization, dimensionless parameter conversion, or unit system unification, can translate physical quantities with different properties into unitless standard values ​​or parameters that can be superimposed in the same dimension. This eliminates the interference of different dimensions on the computational logic, allowing the formulas to retain the original data distribution characteristics while possessing mathematical rationality and adaptability to objective laws. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the invention.

[0099] It should also be noted that the various thresholds in this invention, such as the risk threshold and similarity threshold for process safety and quality boundaries, process anomaly risk indicators, etc., are set according to unified and reasonable rules: they are all based on historical production / failure data of waste sand recycling production lines and the results of a large number of process experiments and tests, combined with the experience of experts in casting, automation, intelligent control and other fields, while comprehensively considering the actual operating environment of the equipment, such as the heating furnace's tolerance temperature, the air volume range of the cooling fan, the design life of key components such as the life of wear-resistant parts of the crusher, and the process quality requirements such as the strength and particle size standard calibration of recycled sand. This setting method has both scientific basis and practicality, and the technology is relatively mature in the existing technology, so it will not be elaborated here.

[0100] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0101] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0102] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations 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. Therefore, the scope of protection of this application should be determined by the scope of protection of the claims.

[0103] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications or additions should fall within the protection scope of the present invention.

Claims

1. A method for online monitoring of process parameters in a waste sand recycling production line, characterized in that, include: S1. Obtain temperature parameters, humidity parameters, particle size parameters, and binder residue parameters of multiple key nodes in the waste sand recycling production line, and integrate them to obtain the original parameter set; S2. Perform collaborative calibration on the original parameter set to eliminate measurement interference between parameters and generate a calibrated parameter set; The step of generating the calibrated parameter set includes: The system receives the original parameter set and independently performs error analysis on the temperature, humidity, particle size, and adhesive residue parameters in the original parameter set based on a preset error model. The error model can be specifically expressed as follows: ,in It is one of the following parameters: temperature, humidity, particle size, and adhesive residue. For input The corresponding temporary value after error analysis, , The inputs are respectively The corresponding sensor's gain correction coefficient and zero-point offset correction coefficient; Using the temperature parameters from the original parameter set, temperature compensation calculations are performed on the humidity parameters at the same acquisition time to obtain the corrected humidity parameters. The correction process is expressed as follows: ,in The corrected humidity parameters are as follows. These are temporary values ​​for the humidity parameter after error analysis. These are temporary values ​​of temperature parameters measured at the same time and location, after error analysis. The temperature compensation function reflects the relationship between humidity measurement error and temperature. Specifically, it is obtained in advance through fitting experimental data to ensure the accuracy of humidity readings at different operating temperatures. The corrected humidity parameter is combined with other parameters in the original parameter set to generate the calibrated parameter set; S3. Input the calibrated parameter set into the parameter coupling model and perform correlation analysis to generate parameter trend prediction results; S4. Based on the parameter trend prediction results, dynamically adjust the control threshold of the process parameters to generate an adaptive threshold. S5. Based on the adaptive threshold, generate and send control commands to the actuator to adjust the waste sand recycling process.

2. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, The steps for integrating to obtain the original parameter set include: Real-time monitoring data is read from integrated sensor modules deployed at key nodes, including but not limited to the crushing, heating, and cooling sections of waste sand recycling production lines; Particle size parameters and binder residue parameters were obtained through laser scattering and near-infrared spectroscopy measurements, respectively. These parameters, along with temperature and humidity parameters, were then integrated into a raw parameter set, which is represented as follows: Where N is the total number of critical nodes, These are respectively represented as the data vectors obtained from the first key node, the second key node, and the Nth key node at time t.

3. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, The steps for generating parameter trend prediction results include: The calibrated parameter set is input into a model that quantifies the dynamic coupling relationship between parameters, i.e., a parameter coupling model. The model that quantifies the dynamic coupling relationship between parameters is used to deduce the change trajectory of each parameter in the calibrated parameter set within a future preset time, i.e., the prediction curve, and generate the parameter trend prediction result. The parameter coupling model is specifically trained based on historical process data; The parameter coupling model is a fusion model of physical mechanism and data-driven approach.

4. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, In step S4, the step of generating the adaptive threshold includes: The predicted parameter trend results are compared with the preset process safety and quality boundaries. If the predicted trend result of any parameter exceeds the process safety and quality boundaries, it is determined that there is a potential process anomaly risk. The risk type is identified, and the process anomaly risk index is obtained by quantifying the deviation magnitude and speed. Based on the type and indicators of the process anomaly risk, the dynamic process parameter control boundary is calculated, as shown in the following formula: ,in, This is the dynamically adjusted new upper limit threshold. The default standard upper limit threshold, As a quantitative indicator of process anomaly risk, It is an adjustable sensitivity coefficient, with dimensions consistent with the parameter being adjusted; The dynamic process parameter control boundary is set as the adaptive threshold.

5. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, The step of generating and sending control commands includes: The real-time sensor monitoring data is continuously compared with the corresponding adaptive threshold, which includes an upper limit and a lower limit. When the sensor-monitored data exceeds the adaptive threshold, an adjustment command is calculated and generated, the specific process of which is as follows: ,in This is the difference between the current monitored parameter value and the corresponding adaptive threshold, i.e., the error; and These are preset proportional and integral control coefficients used to ensure the speed and stability of the adjustment action. for The difference between the parameter monitoring value at time and the corresponding adaptive threshold, u(t) is the adjustment amount of the actuator. The calculated u(t) is encapsulated into a standardized adjustment instruction that can be recognized by the actuator. The adjustment command is sent to the actuator, which includes a heater or a fan.

6. The online monitoring method for process parameters of a waste sand recycling production line according to claim 4, characterized in that, After generating the adaptive threshold, the following is also included: When the process abnormality risk index exceeds the risk threshold, an alarm delay program is initiated and a delay trigger signal is generated; During the alarm delay period, the parameter coupling model is used to verify the consistency of real-time sensor monitoring data and generate a verification conclusion, which is whether the real-time sensor monitoring data is consistent or inconsistent. Based on the verification results, control the triggering state of the alarm device.

7. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, Also includes: Collect real-time monitoring data and corresponding historical process data during the production process to form an updated dataset; The parameter coupling model is retrained using the updated dataset to generate an updated parameter coupling model; A smooth transition mechanism is used to replace the original parameter coupling model with the updated parameter coupling model for subsequent correlation analysis.

8. The online monitoring method for process parameters of a waste sand recycling production line according to claim 1, characterized in that, Also includes: Store the calibrated parameter set, the parameter trend prediction results, and the adaptive threshold to construct a historical database; Based on the data correlation in the historical database, a quality report is generated. The quality report includes, but is not limited to, the accuracy of the parameter coupling model, the stability score of the production process, and the correlation between process parameter fluctuations and product quality. The quality report is then output to a display device.

9. A system for online monitoring of process parameters in a waste sand recycling production line according to any one of claims 1-8, characterized in that, include: An integrated sensor module is used to acquire the original parameter set; The central processing unit, connected to the integrated sensor module, is used to generate the calibrated parameter set, the parameter trend prediction result, the adaptive threshold, and the control command. An actuator, connected to the central processing unit, is used to receive and execute the control commands to adjust the waste sand recycling process; An alarm device, connected to the central processing unit, is used to trigger an alarm upon receiving a trigger signal based on a verification conclusion.

Images