Sand moisture and clay blue absorption collaborative control based casting density optimization system and method

Abstract

The application discloses a foundry compactness optimization system and method based on sand moisture and clay blue absorption amount synergistic regulation and control, relates to the technical field of foundry control, and synchronously collects sand high-frequency eddy current signals and multi-frequency spectrum near-infrared signals in a sand mixing process through a self-cleaning wear-resistant alloy probe, inputs a pre-trained support vector machine regression signal fusion model, and obtains real-time moisture content and clay blue absorption amount; in combination with a preset compactness target threshold value, a compactness-sand parameter correlation database is queried to determine a parameter target interval, a deviation value is calculated, and a regulation and control instruction is generated to control a water adding device and a clay conveying device to adjust the dosages; the compactness of the foundry product after regulation and control is detected, and if the compactness does not reach the standard, the model output weight is iteratively optimized for correction until the compactness reaches the standard; the system comprises a detection module, a signal processing module, an execution regulation and control module, a compactness detection and correction module and a central control module, realizes closed-loop control, and improves the compactness stability of the foundry product.

Description

Technical Field

[0001] This invention relates to the field of casting control technology, and more specifically, to a casting density optimization system and method based on the synergistic regulation of molding sand moisture and clay bluing amount. Background Technology

[0002] In the field of precision casting manufacturing, molding sand, as the core carrier of the mold, directly affects the compactness, permeability, and collapseability of the mold due to the dynamic coupling characteristics of its moisture content and clay blue absorption, thus determining the density and mechanical property stability of the casting. Currently, the industry mostly uses offline intermittent analysis for molding sand parameter detection, such as thermogravimetric analysis for moisture content and methylene blue colorimetry for clay blue absorption. These methods require manual sampling and laboratory processing, resulting in significant detection delays and an inability to capture in real time the fluctuations in raw material batches during sand mixing, such as parameter mutations caused by differences in clay mineral composition and changes in the surface moisture content of the original sand, leading to lag in control response. At the same time, traditional detection probes are affected by high-speed scouring and adhering contamination of molding sand particles, resulting in severe wear of the probe end and signal distortion. The coil impedance value during high-frequency eddy current detection is also affected. Susceptible to temperature drift interference, multi-spectral near-infrared detection suffers from light scattering errors due to sand accumulation on the probe surface, leading to a continuous decline in parameter detection accuracy. In terms of control logic, existing technologies mostly adopt independent control of a single parameter, ignoring the constraint relationship between clay adsorption and water absorption capacity. When clay adsorption is insufficient, the water retention capacity of molding sand decreases, making the mold prone to cracking; when adsorption is too high, the bonding force of molding sand is too strong, making the casting prone to porosity defects. In addition, the correlation between density and molding sand parameters lacks material-specific quantitative models, relying on manual experience to set parameter ranges. The adaptability of molding sand parameters for different materials, such as cast iron, cast steel, and aluminum alloys, is poor, making it difficult to meet the stringent requirements of high-end equipment for casting density. There is an urgent need to construct a closed-loop technology system of real-time detection, collaborative control, and dynamic correction.

[0003] Therefore, existing technologies suffer from detection lag distortion, lack of coupling in regulation, and lack of material-specific quantitative correlation in density. Summary of the Invention

[0004] In order to overcome the problems of detection lag distortion, lack of coupling in regulation, and lack of material-specific quantitative correlation in density of existing technologies, this invention discloses a casting density optimization system and method based on the synergistic regulation of molding sand moisture and clay blue absorption, which can effectively solve the above-mentioned technical problems.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A method for optimizing casting density based on the synergistic regulation of molding sand moisture and clay bluing capacity, the method comprising:

[0007] Real-time detection signals of molding sand during the sand mixing process are obtained. The real-time detection signals include high-frequency eddy current signals and multi-spectral near-infrared signals collected by a self-cleaning wear-resistant alloy probe. The high-frequency eddy current signals include coil impedance and inductive reactance values. The multi-spectral near-infrared signals include absorbance values ​​at water-sensitive wavelengths and absorbance values ​​at clay-sensitive reference wavelengths.

[0008] The real-time detection signal is input into a pre-trained signal fusion model to obtain the real-time moisture content of the molding sand and the real-time clay blue absorption. The signal fusion model is a support vector machine regression model trained with multiple sets of standard molding sand sample data. The standard molding sand sample data includes high-frequency eddy current signals and multi-spectral near-infrared signals corresponding to different moisture contents and different clay blue absorption.

[0009] Based on the preset target threshold for casting density, the density-molding sand parameter correlation database is queried to determine the target range for molding sand moisture content and the target range for clay bluing amount.

[0010] Calculate the first deviation value between the real-time moisture content and the target range of the moisture content, and the second deviation value between the real-time clay blue absorption amount and the target range of the clay blue absorption amount;

[0011] Based on the first deviation value, a moisture control command is generated to control the water addition device of the sand mixer to adjust the instantaneous water addition; based on the second deviation value, a clay control command is generated to control the clay conveying device to adjust the conveying amount per unit time.

[0012] Obtain the density test result of the molded casting after regulation. If the density test result does not reach the density target threshold, then based on the difference between the density test result and the density target threshold, correct the output weight of the signal fusion model. Iteratively execute the above steps of obtaining the real-time detection signal of the molding sand during the sand mixing process, inputting the real-time detection signal into the pre-trained signal fusion model to obtain real-time parameters, calculating the deviation value and generating the moisture regulation command, and correcting the output weight of the signal fusion model until the density of the casting reaches the density target threshold.

[0013] Preferably, the real-time detection signal of molding sand during the sand mixing process includes:

[0014] The high-frequency eddy current module of the self-cleaning wear-resistant alloy probe generates an alternating electromagnetic field of a fixed frequency. When molding sand flows through the probe's detection end, the impedance and reactance values ​​of the coil under the action of the electromagnetic field are collected.

[0015] The multi-spectral near-infrared module of the self-cleaning wear-resistant alloy probe is synchronously controlled to emit near-infrared light of water-sensitive wavelength and clay-sensitive reference wavelength, receive the light signal after diffuse reflection of molding sand, and convert it into the corresponding absorbance value.

[0016] At preset time intervals, the compressed air pulse jet device of the self-cleaning wear-resistant alloy probe mounting base is activated to remove the molding sand adhering to the surface of the probe end, ensuring the stability of the detection signal.

[0017] Preferably, the training process of the pre-trained signal fusion model includes:

[0018] Prepare multiple sets of standard molding sand samples. The moisture content and clay blue absorption of each set of samples are within a predetermined range, and the number of samples in each set is not less than the predetermined number.

[0019] High-frequency eddy current signals and multi-spectral near-infrared signals were collected from each group of standard molding sand samples to construct a sample dataset;

[0020] The sample dataset is divided into a training set and a validation set according to the proportion. The training set is used to train a support vector machine regression model with high-frequency eddy current signal and multi-spectral near-infrared signal as input features and the actual moisture content and actual clay blue absorption of standard molding sand samples as output labels.

[0021] The model's prediction accuracy is verified using the validation set. If the prediction error for moisture content exceeds ± a preset threshold or the prediction error for clay blue absorption exceeds a preset threshold, the number of samples is increased and the model kernel function parameters are adjusted until the prediction accuracy meets the requirements, thus completing model training.

[0022] Preferably, the step of querying the density-molding sand parameter correlation database to determine the target range of molding sand moisture content and the target range of clay bluing amount based on the preset casting density target threshold includes:

[0023] A density-molding sand parameter correlation database is constructed, which stores historical data of casting density corresponding to different casting materials under different molding sand moisture contents and different clay bluing amounts.

[0024] Based on the aforementioned associated database, a linear regression algorithm was used to fit the functional relationship between moisture content, clay blue absorption, and casting density.

[0025] Substituting the preset target threshold for casting density into the functional relationship, the corresponding target ranges for molding sand moisture content and clay blue absorption are obtained by reverse calculation.

[0026] If the material is a new casting material, several small-batch trial productions are carried out first to collect density data corresponding to different molding sand parameters, which are then added to the associated database. After correcting the functional relationship, the target range is determined.

[0027] Preferably, the step of generating a moisture control command based on the first deviation value to control the water supply device of the sand mixer to adjust the instantaneous water supply; and generating a clay control command based on the second deviation value to control the clay conveying device to adjust the conveying volume per unit time includes:

[0028] If the first deviation value is negative, calculate the required water replenishment amount = absolute value of the first deviation value × instantaneous flow rate of molding sand. Based on the maximum water supply capacity of the water supply device, determine the adjustment range of the instantaneous water replenishment amount and generate a water control command.

[0029] If the first deviation value is positive, a moisture control command is generated to pause water addition or reduce the amount of water added, until the real-time moisture content falls back to the target range.

[0030] If the second deviation value is negative, calculate the amount of clay to be added = absolute value of the second deviation value × instantaneous flow rate of molding sand ÷ actual amount of clay absorbed. Based on the conveying efficiency of the clay conveying device, determine the adjustment range of the conveying amount per unit time and generate a clay control command.

[0031] If the second deviation value is positive, a clay control command is generated to pause clay transport or reduce the transport volume until the real-time clay absorption volume falls back to the target range.

[0032] Preferably, the density test results of the regulated molding sand casting include:

[0033] Several samples were randomly selected from the castings produced after adjustment, and the volume percentage of internal defects in each sample was detected by an ultrasonic flaw detector.

[0034] Calculate the actual density of the casting = 1 - defect volume ratio, and take the average of the actual density values ​​of multiple samples as the density test result;

[0035] If the deviation between the actual value and the average value of a single sample compaction exceeds a preset threshold, the sample is re-extracted for testing to ensure the reliability of the test results.

[0036] Preferably, the casting density optimization system based on the synergistic regulation of molding sand moisture and clay bluing absorption includes:

[0037] The self-cleaning molding sand testing module includes a wear-resistant alloy probe, a high-frequency eddy current unit, a multi-spectral near-infrared unit, and a pulse cleaning unit. The wear-resistant alloy probe is fixed to the side wall of the sand mixer, with the probe end facing the molding sand flow path. The high-frequency eddy current unit is used to generate an alternating electromagnetic field and collect coil impedance and inductive reactance values. The multi-spectral near-infrared unit is used to emit dual-wavelength near-infrared light and collect absorbance values. The pulse cleaning unit is used to periodically clean the molding sand from the probe surface.

[0038] The signal processing and model calculation module includes a microprocessor and a pre-stored signal fusion model. The microprocessor is used to receive the real-time signal from the self-cleaning molding sand detection module, call the signal fusion model to calculate the real-time moisture content and the real-time clay blue absorption, and compare the deviation value with the target interval.

[0039] The control module includes a water supply device for the sand mixer, a clay conveying device, and a control unit.

[0040] The control unit receives the control instructions from the signal processing and model calculation module, and drives the water adding device to adjust the water adding amount and drives the clay conveying device to adjust the conveying amount.

[0041] The density detection and model correction module includes an ultrasonic flaw detector and a data storage unit.

[0042] The ultrasonic flaw detector is used to detect the density of the casting, and the data storage unit records the density deviation data, which is used to correct the output weights of the signal fusion model.

[0043] The central control module is communicatively connected to the self-cleaning molding sand detection module, signal processing and model calculation module, execution control module, and density detection and model correction module, respectively, to coordinate the working sequence of each module and realize closed-loop control of data acquisition, calculation, control, and detection.

[0044] Preferably, the detection end of the wear-resistant alloy probe adopts a multi-layer composite structure, consisting of a wear-resistant insulating ceramic layer, a high-frequency eddy current coil, and a heat insulation layer from the outside to the inside. A multi-spectral near-infrared light channel is set in the central axis of the probe, and an LED light source and photodetector are built in.

[0045] The central control module adopts a communication method combining industrial Ethernet and IO-Link bus, and communicates with the signal processing and model calculation module. The central control module receives the working status feedback signals of each module in real time. If a module fails, it will immediately issue an alarm signal and suspend the sand mixing operation.

[0046] The density-molding sand parameter association database is stored in the data storage unit. The data storage unit supports real-time data updates and queries, and has a data backup function to prevent the loss of historical data. The signal processing and model calculation module supports online correction of model parameters, and can complete model optimization without stopping system operation.

[0047] Preferably, an electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the optimization method described above.

[0048] Preferably, a computer-readable storage medium stores a computer program that, when executed by a processor, implements the optimization method described above.

[0049] Compared with existing technologies, the beneficial effects of this invention are as follows: To solve the problems of detection lag distortion, decoupling of control, and lack of material-specific quantitative correlation of density in existing technologies, this technology forms a closed-loop solution through multi-dimensional targeted design: Regarding detection lag distortion, a self-cleaning wear-resistant alloy probe fixed to the side wall of the sand mixer is used. Its multi-layer composite structure (outer layer of wear-resistant insulating ceramic to resist molding sand erosion, inner layer of heat insulation to reduce temperature interference) can simultaneously acquire high-frequency eddy current signals and multi-spectral near-infrared signals. Combined with compressed air pulse jet cleaning to periodically remove sand accumulation on the probe surface, this eliminates the lag of traditional offline detection and avoids signal distortion caused by wear and adhesion, achieving real-time and accurate detection of molding sand parameters. Regarding decoupling of control, a pre-trained support vector machine regression signal fusion model is used to convert the detection signal into real-time moisture content and clay adsorption, combined with the density target threshold to determine the target parameters. The system generates control commands based on quantitative logic within a specified range, achieving coupled and coordinated control between the two. Simultaneously, it iteratively corrects the model output weights based on casting density detection results, continuously optimizing parameter matching. For the lack of material-specific quantitative correlation in density measurement, a correlation database storing historical data on density and molding sand parameters for different casting materials is constructed. Linear regression is used to fit the material-specific quantitative function relationship. For new materials, data is supplemented through small-batch trial production to correct the function, and then the precise parameter range is inferred. This design provides reliable data support for control due to real-time and accurate detection. Coupled control avoids mold cracking or porosity defects caused by single parameter deviations. Material-specific quantitative correlation ensures parameter compatibility for castings of different materials, ultimately improving the casting density compliance rate, reducing casting scrap rate to lower production costs, and supporting online model correction and real-time data updates to ensure production continuity and simultaneously improve production efficiency and casting quality stability. Attached Figure Description

[0050] To more clearly illustrate the embodiments of the present invention or the technical solutions in 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 merely exemplary. For those skilled in the art, other embodiments can be derived from the provided drawings without creative effort.

[0051] Figure 1 This is a diagram illustrating the steps of the method of the present invention;

[0052] Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0053] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0054] To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions;

[0055] It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.

[0056] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0057] Example 1

[0058] Please see Figure 1 A method for optimizing casting density based on the synergistic regulation of molding sand moisture and clay bluing absorption, the method comprising:

[0059] Real-time detection signals of molding sand during the sand mixing process are obtained. The real-time detection signals include high-frequency eddy current signals and multi-spectral near-infrared signals collected by a self-cleaning wear-resistant alloy probe. The high-frequency eddy current signals include coil impedance and inductive reactance values. The multi-spectral near-infrared signals include absorbance values ​​at water-sensitive wavelengths and absorbance values ​​at clay-sensitive reference wavelengths.

[0060] The real-time detection signal is input into a pre-trained signal fusion model to obtain the real-time moisture content of the molding sand and the real-time clay blue absorption. The signal fusion model is a support vector machine regression model trained with multiple sets of standard molding sand sample data. The standard molding sand sample data includes high-frequency eddy current signals and multi-spectral near-infrared signals corresponding to different moisture contents and different clay blue absorption.

[0061] Based on the preset target threshold for casting density, the density-molding sand parameter correlation database is queried to determine the target range for molding sand moisture content and the target range for clay bluing amount.

[0062] Calculate the first deviation value between the real-time moisture content and the target range of the moisture content, and the second deviation value between the real-time clay blue absorption amount and the target range of the clay blue absorption amount;

[0063] Based on the first deviation value, a moisture control command is generated to control the water addition device of the sand mixer to adjust the instantaneous water addition; based on the second deviation value, a clay control command is generated to control the clay conveying device to adjust the conveying amount per unit time.

[0064] Obtain the density test result of the molded casting after regulation. If the density test result does not reach the density target threshold, then based on the difference between the density test result and the density target threshold, correct the output weight of the signal fusion model. Iteratively execute the above steps of obtaining the real-time detection signal of the molding sand during the sand mixing process, inputting the real-time detection signal into the pre-trained signal fusion model to obtain real-time parameters, calculating the deviation value and generating the moisture regulation command, and correcting the output weight of the signal fusion model until the density of the casting reaches the density target threshold.

[0065] The real-time detection signal of molding sand during the sand mixing process includes:

[0066] The high-frequency eddy current module of the self-cleaning wear-resistant alloy probe generates an alternating electromagnetic field of a fixed frequency. When molding sand flows through the probe's detection end, the impedance and reactance values ​​of the coil under the action of the electromagnetic field are collected.

[0067] The multi-spectral near-infrared module of the self-cleaning wear-resistant alloy probe is synchronously controlled to emit near-infrared light of water-sensitive wavelength and clay-sensitive reference wavelength, receive the light signal after diffuse reflection of molding sand, and convert it into the corresponding absorbance value.

[0068] At preset time intervals, the compressed air pulse jet device of the self-cleaning wear-resistant alloy probe mounting base is activated to remove the molding sand adhering to the surface of the probe end, ensuring the stability of the detection signal.

[0069] The training process of the pre-trained signal fusion model includes:

[0070] Prepare multiple sets of standard molding sand samples. The moisture content and clay blue absorption of each set of samples are within a predetermined range, and the number of samples in each set is not less than the predetermined number.

[0071] High-frequency eddy current signals (impedance value, reactance value) and multi-spectral near-infrared signals (absorbance value at water-sensitive wavelength and absorbance value at clay-sensitive reference wavelength) were collected for each group of standard molding sand samples to construct a sample dataset;

[0072] The sample dataset is divided into a training set and a validation set according to the proportion. The training set is used to train a support vector machine regression model with high-frequency eddy current signal and multi-spectral near-infrared signal as input features and the actual moisture content and actual clay blue absorption of standard molding sand samples as output labels.

[0073] The model's prediction accuracy is verified using the validation set. If the prediction error for moisture content exceeds ± a preset threshold or the prediction error for clay blue absorption exceeds a preset threshold, the number of samples is increased and the model kernel function parameters are adjusted until the prediction accuracy meets the requirements, thus completing model training.

[0074] The step of querying the density-molding sand parameter correlation database based on the preset casting density target threshold to determine the target range of molding sand moisture content and clay bluing absorption includes:

[0075] A density-molding sand parameter correlation database is constructed, which stores historical data of casting density for different casting materials (cast iron, cast steel, aluminum alloy) under different molding sand moisture contents and different clay bluing amounts.

[0076] Based on the aforementioned associated database, a linear regression algorithm was used to fit the functional relationship between moisture content, clay blue absorption, and casting density.

[0077] Substituting the preset target threshold for casting density into the functional relationship, the corresponding target ranges for molding sand moisture content and clay blue absorption are obtained by reverse calculation.

[0078] If the material is a new casting material, several small-batch trial productions are carried out first to collect density data corresponding to different molding sand parameters, which are then added to the associated database. After correcting the functional relationship, the target range is determined.

[0079] The step of generating a moisture control command based on the first deviation value to control the water supply device of the sand mixer to adjust the instantaneous water supply; and generating a clay control command based on the second deviation value to control the clay conveying device to adjust the conveying volume per unit time, includes:

[0080] If the first deviation value is negative (the real-time moisture content is lower than the lower limit of the target range), calculate the amount of water to be added = the absolute value of the first deviation value × the instantaneous flow rate of the molding sand. Based on the maximum water supply capacity of the water supply device, determine the adjustment range of the instantaneous water addition and generate a moisture control command.

[0081] If the first deviation value is positive (the real-time moisture content is higher than the upper limit of the target range), a moisture control command is generated to pause water addition or reduce the amount of water added, until the real-time moisture content falls back to the target range.

[0082] If the second deviation value is negative (the real-time clay absorption is lower than the lower limit of the target range), calculate the amount of clay to be supplemented = the absolute value of the second deviation value × the instantaneous flow rate of molding sand ÷ the actual absorption of clay. Based on the conveying efficiency of the clay conveying device, determine the adjustment range of the conveying amount per unit time and generate a clay control command.

[0083] If the second deviation value is positive (the real-time clay absorption is higher than the upper limit of the target range), a clay control command is generated to suspend clay delivery or reduce the delivery amount until the real-time clay absorption falls back to the target range.

[0084] The density test results of the regulated molding sand castings include:

[0085] Several samples were randomly selected from the castings produced after adjustment, and the volume percentage of internal defects (porosity, shrinkage) in each sample was detected by ultrasonic flaw detector.

[0086] Calculate the actual density of the casting = 1 - defect volume ratio, and take the average of the actual density values ​​of multiple samples as the density test result;

[0087] If the deviation between the actual value and the average value of a single sample compaction exceeds a preset threshold, the sample is re-extracted for testing to ensure the reliability of the test results.

[0088] This embodiment uses the actual production scenario of an automotive parts manufacturing company producing engine block castings as a background to elaborate on the complete implementation process of the optimization method.

[0089] Equipment debugging: Select a self-cleaning wear-resistant alloy probe suitable for aluminum alloy molding sand detection, and debug its high-frequency eddy current module and multi-spectral near-infrared module. The high-frequency eddy current module must ensure that it can stably generate an alternating electromagnetic field suitable for detecting molding sand. The multi-spectral near-infrared module must accurately calibrate the emission intensity of the water-sensitive wavelength and the clay-sensitive reference wavelength to ensure the accuracy of signal acquisition. At the same time, check the operating status of the water supply device and clay conveying device of the sand mixer to ensure that the water flow rate and clay conveying volume are smoothly adjusted without jamming or leakage.

[0090] Basic data collection: Collect historical data on the production of similar aluminum alloy engine block castings by the company, including the moisture content of molding sand, the amount of clay adsorption, and the corresponding casting density test results of different production batches. It is also necessary to record factors that may affect the performance of molding sand, such as sand mixing time, ambient temperature and humidity during the production process, to provide data for building a density-molding sand parameter correlation database.

[0091] Key parameter calibration: Through laboratory tests, the basic performance parameters of aluminum alloy molding sand are determined. For example, the basic blue absorption range of clay in molding sand is measured, and the variation law of the fluidity and compactness of molding sand under different moisture contents is clarified to ensure that the target range of molding sand parameters set later meets the actual production needs.

[0092] Activate the high-frequency eddy current module of the self-cleaning wear-resistant alloy probe to generate a stable alternating electromagnetic field. When the molding sand in the sand mixer flows through the probe's detection end, the moisture and clay components in the molding sand will change the distribution of the electromagnetic field, thereby affecting the impedance and reactance of the coil. The probe collects the impedance and reactance values ​​of the coil in real time, with the collection frequency set to 10 times per second to ensure timely capture of dynamic changes in molding sand parameters. During the collection process, care should be taken to avoid interference from metallic foreign objects near the probe. If abnormal signal fluctuations are detected, the environment around the probe should be checked immediately and interference eliminated.

[0093] Synchronous with the high-frequency eddy current signal acquisition, the multi-spectral near-infrared module of the probe is activated. This module emits near-infrared light at water-sensitive wavelengths and clay-sensitive reference wavelengths. After the light shines on the molding sand surface, it undergoes diffuse reflection. The probe receives the diffusely reflected light signal and converts it into corresponding absorbance values ​​based on the intensity changes of the light signal. The absorbance value at the water-sensitive wavelength reflects the moisture content in the molding sand, while the absorbance value at the clay-sensitive reference wavelength reflects the clay content in the molding sand. To ensure signal accuracy, the near-infrared light emission and reception windows need to be cleaned regularly to prevent dust or molding sand particles from affecting the transmission of the light signal.

[0094] According to the preset time interval (set to 1 minute in this embodiment), the compressed air pulse jetting device of the self-cleaning wear-resistant alloy probe mounting base is started. The compressed air is ejected in the form of high-pressure pulses to remove the molding sand particles attached to the probe detection end surface, so as to avoid the accumulation of molding sand affecting the accuracy of subsequent signal acquisition. During the jetting process, the pressure of compressed air and the jetting time need to be controlled to ensure the cleaning effect while preventing the high-pressure airflow from causing excessive interference to the flow state of the molding sand.

[0095] The collected real-time detection signals are input into a pre-trained signal fusion model to obtain the key parameters of the molding sand. The specific steps are as follows:

[0096] Since the high-frequency eddy current signals and multi-spectral near-infrared signals may be subject to noise interference, such as equipment vibration and changes in ambient light, the signals need to be preprocessed. Filtering algorithms, such as Kalman filtering, are used to remove high-frequency noise from the signals. At the same time, the signals are normalized to convert the signal values ​​of different dimensions to the same numerical range, which is convenient for subsequent model calculations.

[0097] The pre-processed real-time detection signal is input into a pre-trained support vector machine regression model. This model uses high-frequency eddy current signals (coil impedance value, inductive reactance value) and multi-spectral near-infrared signals (absorbance value of water-sensitive wavelength and absorbance value of clay-sensitive reference wavelength) as input features, and the actual moisture content of molding sand and the actual clay blue absorption amount as output labels. Through training with a large number of standard samples in the early stage, it has achieved high prediction accuracy. After the model runs, it quickly outputs the real-time moisture content of molding sand and the real-time clay blue absorption amount, and displays the results on the production monitoring terminal in real time, so that operators can keep track of the status of molding sand in real time.

[0098] To ensure the reliability of the model prediction results, a portion of molding sand samples are selected every hour, and traditional laboratory testing methods, such as drying method to determine moisture content and methylene blue method to determine clay blue absorption, are used to verify the model prediction results. If the deviation between the prediction results and the laboratory test results exceeds the preset threshold (in this embodiment, the deviation of moisture content is set to no more than ±0.1%, and the deviation of clay blue absorption is set to no more than ±0.2mL / 100g), the model operation status should be checked in time, and the model parameters should be recalibrated if necessary.

[0099] Based on the preset target threshold for casting density, and combined with the density-molding sand parameter correlation database, the target range for key molding sand parameters is determined. The specific process is as follows:

[0100] Open the density-molding sand parameter association database. This database stores historical data on the density of castings of different materials under different molding sand moisture contents and different clay bluing amounts. Filter historical data that are consistent with the type of aluminum alloy engine block castings currently being produced and have similar production processes from the database, and use them as a reference for determining the target range.

[0101] Based on the selected historical data, a linear regression algorithm is used to fit the functional relationship between molding sand moisture content, clay bluing amount and casting density. Through this functional relationship, the preset casting density target threshold (not less than 99%) is substituted into the equation to back-calculate the corresponding target ranges for molding sand moisture content and clay bluing amount. For example, the calculation shows that the target range for molding sand moisture content of the aluminum alloy engine cylinder block casting in this embodiment is 3.5%-4.0%, and the target range for clay bluing amount is 28-32mL / 100g.

[0102] If the casting material to be produced this time is a newly introduced aluminum alloy, and there is a lack of sufficient historical data in the database to support the determination of the target range, then a small-batch trial production should be carried out first. During the trial production, different combinations of molding sand moisture content and clay bluing amount should be set, and the casting density data corresponding to each combination should be collected. These data should be added to the density-molding sand parameter correlation database, and after refitting the functional relationship, the accurate target range of molding sand parameters should be determined.

[0103] Based on the deviation between real-time parameters and the target range, corresponding control commands are generated to control the relevant equipment to adjust the molding sand parameters. The specific operation is as follows:

[0104] Compare the real-time moisture content of the molding sand with the target moisture content range to calculate the first deviation value; compare the real-time clay blue absorption with the target clay blue absorption range to calculate the second deviation value. If the real-time moisture content is lower than the lower limit of the target range, the first deviation value is negative; if it is higher than the upper limit of the target range, the first deviation value is positive; if it is within the range, the deviation value is zero. The deviation value of clay blue absorption is calculated in the same way.

[0105] When the first deviation value is negative (the real-time moisture content is lower than the lower limit of the target range), the amount of water to be added is calculated based on the instantaneous flow rate of the molding sand (the amount of water to be added = the absolute value of the first deviation value × the instantaneous flow rate of the molding sand). Combined with the maximum water supply capacity of the water supply device of the sand mixer, the adjustment range of the instantaneous water addition is determined, and a moisture control command is generated. For example, if the amount of water to be added is 5L / min, and the maximum water supply capacity of the water supply device meets the requirements, then the water supply device is instructed to increase the instantaneous water addition by 5L / min.

[0106] When the first deviation value is positive (the real-time moisture content is higher than the upper limit of the target range), a control command is generated to pause water addition or reduce the amount of water added. If the moisture content exceeds the upper limit by a small amount, the amount of water added can be reduced appropriately; if it exceeds it by a large amount, water addition is paused until the real-time moisture content falls back to the target range.

[0107] When the second deviation value is negative (the real-time clay absorption is lower than the lower limit of the target range), the amount of clay to be supplemented is calculated based on the instantaneous flow rate of molding sand and the actual absorption of clay (the amount of clay to be supplemented = the absolute value of the second deviation value × the instantaneous flow rate of molding sand ÷ the actual absorption of clay). Based on the conveying efficiency of the clay conveying device, the adjustment range of the conveying volume per unit time is determined, and a clay control command is generated. For example, if the amount of clay to be supplemented is 3 kg / min and the conveying efficiency of the clay conveying device is met, then the command is given to increase the conveying volume per unit time by 3 kg / min.

[0108] When the second deviation value is positive (the real-time clay absorption is higher than the upper limit of the target range), a control command is generated to suspend clay delivery or reduce the delivery amount until the real-time clay absorption falls back to the target range.

[0109] After the control operation is completed, the density of the formed casting is tested. Based on the test results, it is determined whether the signal fusion model needs to be corrected to form a closed-loop optimization. The specific steps are as follows:

[0110] From the aluminum alloy engine block castings produced after adjustment, several samples were randomly selected (10 samples were selected in this embodiment). Each sample was tested using an ultrasonic flaw detector. By detecting the volume ratio of internal defects such as porosity and shrinkage, the actual density value of the casting was calculated (actual density value of casting = 1 - volume ratio of defects). The average value of the actual density values ​​of the 10 samples was taken as the density test result after this adjustment.

[0111] If the density test result reaches the preset density target threshold (not less than 99%), it indicates that the current adjustment is effective. Production will continue according to the current parameters, and the production data will be added to the density-molding sand parameter association database to provide more data for subsequent production.

[0112] If the density test result does not reach the target threshold, the difference between the density test result and the target threshold is calculated. Based on this difference, the reasons for the prediction deviation of the model are analyzed, and the output weights of the signal fusion model are corrected. For example, if the actual density value is 0.5% lower than the target threshold, and the analysis finds that the moisture content predicted by the model is too high, resulting in insufficient moisture in the actual molding sand, the output weights related to the moisture content prediction in the model can be appropriately adjusted to reduce the prediction deviation. After the model is corrected, the above signal acquisition, parameter calculation, control and detection steps are repeated until the density of the casting reaches the target threshold.

[0113] During the testing process, if the deviation between the actual density value of a single sample and the average value exceeds a preset threshold (set to ±0.3% in this embodiment), it is considered that the sample may have a testing error or special defect, and the sample needs to be re-extracted for testing to ensure the reliability of the density test results and avoid the influence of individual abnormal samples on the judgment of the control effect.

[0114] Example 2

[0115] Please see Figure 2 A casting density optimization system based on the synergistic regulation of molding sand moisture and clay bluing capacity, the system comprising:

[0116] The self-cleaning molding sand testing module includes a wear-resistant alloy probe, a high-frequency eddy current unit, a multi-spectral near-infrared unit, and a pulse cleaning unit. The wear-resistant alloy probe is fixed to the side wall of the sand mixer, with the probe end facing the molding sand flow path. The high-frequency eddy current unit is used to generate an alternating electromagnetic field and collect coil impedance and inductive reactance values. The multi-spectral near-infrared unit is used to emit dual-wavelength near-infrared light and collect absorbance values. The pulse cleaning unit is used to periodically clean the molding sand from the probe surface.

[0117] The signal processing and model calculation module includes a microprocessor and a pre-stored signal fusion model. The microprocessor is used to receive the real-time signal from the self-cleaning molding sand detection module, call the signal fusion model to calculate the real-time moisture content and the real-time clay blue absorption, and compare the deviation value with the target interval.

[0118] The execution control module includes a sand mixer water supply device, a clay conveying device, and a control unit. The control unit receives control commands from the signal processing and model calculation module, driving the water supply device to adjust the water supply amount and the clay conveying device to adjust the conveying amount. The density detection and model correction module includes an ultrasonic flaw detector and a data storage unit. The ultrasonic flaw detector is used to detect the density of the casting, and the data storage unit records density deviation data, which is used to correct the output weights of the signal fusion model.

[0119] The central control module is communicatively connected to the self-cleaning molding sand detection module, signal processing and model calculation module, execution control module, and density detection and model correction module, respectively, to coordinate the working sequence of each module and realize closed-loop control of data acquisition, calculation, control, and detection.

[0120] The detection end of the wear-resistant alloy probe adopts a multi-layer composite structure, consisting of a wear-resistant insulating ceramic layer (thickness 0.5-1mm), a high-frequency eddy current coil (wound with copper enameled wire, 50-100 turns), and a heat insulation layer (ceramic fiber material, thickness 0.3-0.5mm) from the outside to the inside. A multi-spectral near-infrared light channel is set in the central axis of the probe, and an LED light source and photodetector are built in.

[0121] The central control module adopts a communication method combining industrial Ethernet and IO-Link bus with the signal processing and model calculation module. The central control module receives the working status feedback signals of each module in real time. If a module malfunctions, such as probe signal interruption or actuator jamming, an alarm signal is immediately issued and the sand mixing operation is suspended.

[0122] The density-molding sand parameter association database is stored in the data storage unit. The data storage unit supports real-time data updates and queries, and has a data backup function to prevent the loss of historical data. The signal processing and model calculation module supports online correction of model parameters, and can complete model optimization without stopping system operation.

[0123] An electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the optimization method described above.

[0124] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the optimization method described above.

[0125] This embodiment, based on the above-mentioned aluminum alloy engine cylinder block casting production scenario, elaborates in detail the composition of the optimization system, the functions of each module, and the workflow, ensuring that the system can efficiently support the implementation of the optimization method.

[0126] The casting density optimization system based on the coordinated regulation of molding sand moisture and clay blue absorption consists of five core modules: a self-cleaning molding sand detection module, a signal processing and model calculation module, an execution and control module, a density detection and model correction module, and a central control module. The modules interact with each other and transmit commands through an industrial communication network, forming a complete closed-loop control system to ensure efficient and accurate regulation of molding sand parameters and optimization of casting density.

[0127] The self-cleaning molding sand detection module mainly includes a wear-resistant alloy probe, a high-frequency eddy current unit, a multi-spectral near-infrared unit, and a pulse cleaning unit. The wear-resistant alloy probe is made of high-strength wear-resistant material, which can adapt to the harsh working environment inside the sand mixer and extend its service life. The probe is fixed to the side wall of the sand mixer, and the detection end is directly facing the sand flow path to ensure full contact and detection of the molding sand.

[0128] The high-frequency eddy current unit generates a stable alternating electromagnetic field. When molding sand flows through the probe, the electromagnetic field changes due to the influence of the molding sand composition (moisture, clay). The unit collects the impedance and reactance values ​​of the coil in real time and transmits the collected signals to the signal processing and model calculation module. To ensure the stability of the electromagnetic field, a voltage and frequency stabilization circuit is set inside the unit to effectively avoid voltage fluctuations from interfering with the electromagnetic field.

[0129] The multi-spectral near-infrared unit includes a near-infrared light source, a light signal receiver, and a signal conversion circuit. The near-infrared light source can accurately emit light at water-sensitive wavelengths and clay-sensitive reference wavelengths. The light signal receiver receives the light signal after diffuse reflection from the molding sand. The signal conversion circuit converts the light signal into the corresponding absorbance value and then transmits the absorbance value signal to the subsequent modules. The unit is also equipped with a temperature compensation circuit to reduce the impact of ambient temperature changes on the detection accuracy of near-infrared light signals.

[0130] The pulse cleaning unit consists of a compressed air storage tank, a pulse solenoid valve, and a jetting pipeline. According to the time interval set by the central control module, the pulse solenoid valve opens, and compressed air is sprayed into the detection end of the wear-resistant alloy probe in the form of high-pressure pulses through the jetting pipeline to remove the molding sand particles attached to the surface. The jetting pressure and jetting time can be flexibly adjusted by the central control module to ensure the cleaning effect while avoiding excessive impact on the flow of molding sand.

[0131] Hardware composition of the signal processing and model calculation module: This module is based on a high-performance microprocessor, and is equipped with hardware devices such as data storage chips and signal interface circuits. The microprocessor has powerful data processing capabilities and can quickly process real-time signals from the self-cleaning molding sand detection module; the data storage chip is used to store pre-trained signal fusion models, historical detection data and calculation results; the signal interface circuit realizes signal interaction with other modules and supports multiple communication protocols.

[0132] Software Functions: Performs preprocessing operations such as filtering and normalization on the received high-frequency eddy current signals and multi-spectral near-infrared signals to remove noise interference, unify the signal format, and provide high-quality data input for subsequent model calculations.

[0133] The pre-stored support vector machine regression model is loaded, and the pre-processed real-time signal is input into the model. The real-time moisture content of the molding sand and the real-time clay adsorption are calculated by the model. During the calculation process, the running status of the model is monitored in real time. If any abnormality occurs, such as calculation timeout or results exceeding the reasonable range, an alarm signal is immediately issued and fed back to the central control module.

[0134] The real-time moisture content and real-time clay blue absorption are compared with the target range obtained from the density-molding sand parameter association database. The first deviation value and the second deviation value are calculated, and the deviation value data is transmitted to the central control module to provide a basis for generating control instructions.

[0135] It supports online correction of model parameters. When the density detection result does not reach the target threshold, it receives the deviation data transmitted between the density detection and model correction modules, and adjusts the output weight of the signal fusion model according to the preset correction algorithm. Model optimization can be completed without stopping the system, ensuring the continuity of production.

[0136] The control module includes a sand mixer water supply device, a clay conveying device, and a control unit. The sand mixer water supply device consists of a water tank, an electromagnetic flow valve, and water pipes. The electromagnetic flow valve can precisely control the water flow rate. The clay conveying device uses a screw conveyor, and the conveying volume per unit time is adjusted by adjusting the motor speed. The control unit consists of a microcontroller, a drive circuit, and a status detection sensor. It is responsible for receiving control commands and driving the actuator to move, while also detecting the operating status of the actuator.

[0137] Workflow: The control unit receives moisture control instructions and clay control instructions from the central control module, parses the instructions, and extracts key parameters such as the adjustment range of water addition or clay delivery.

[0138] Based on the analyzed parameters, the control unit drives the electromagnetic flow valve through the drive circuit to adjust the opening, thereby adjusting the instantaneous water addition of the sand mixer's water addition device; and drives the screw conveyor motor to change its speed, thereby adjusting the conveying capacity of the clay conveying device per unit time.

[0139] The status detection sensor monitors the actual water volume added by the water adding device, the actual conveying volume of the clay conveying device, and the operating status of the actuator in real time, such as whether the motor is running normally and whether the pipeline is blocked. The detection results are fed back to the central control module to form a closed-loop control and ensure that the control commands are executed accurately.

[0140] The density detection and model correction module includes:

[0141] The density testing unit uses an ultrasonic flaw detector as its core testing equipment, along with auxiliary equipment such as probes and coupling agent application devices. The ultrasonic flaw detector emits ultrasonic waves that penetrate the casting sample and detects the location, size, and shape of internal defects based on the reflected ultrasonic signals. It then calculates the volume ratio of the defects to obtain the actual density value of the casting. During the testing process, the software can automatically record the test data and generate a test report, facilitating subsequent data analysis and traceability.

[0142] The data storage unit uses a high-capacity hard drive as the storage medium to build a density-molding sand parameter correlation database. It stores molding sand parameters (moisture content, clay bluing amount) and corresponding casting density data for different casting materials and different production batches. It supports real-time data updates and query functions. Operators can query historical data at any time through terminal devices to provide reference for production decisions. It also has a data backup function, which regularly backs up the database data to external storage devices to prevent the loss of historical data due to hardware failure.

[0143] When the density detection result does not reach the target threshold, the correlation between the density deviation and the predicted deviation of the molding sand parameters is analyzed. Model correction parameters are generated according to the preset correction logic and transmitted to the signal processing and model calculation module to guide it to complete the adjustment of the output weight of the signal fusion model. During the correction process, the model parameters and density detection results before and after the correction are recorded to form a model correction log, which provides a basis for subsequent model optimization.

[0144] The central control module is responsible for coordinating the working sequence of each module, realizing closed-loop control of data acquisition, calculation, regulation, and detection. It uses an industrial-grade touch screen as the human-machine interface, allowing operators to view the working status of each module, real-time parameters of molding sand, and casting density test results in real time.

[0145] The same or similar labels correspond to the same or similar parts;

[0146] The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.

[0147] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all implementation methods here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A method for optimizing casting density based on the synergistic control of molding sand moisture and clay bluing absorption, characterized in that, The method includes: Real-time detection signals of molding sand during the sand mixing process are acquired. These real-time detection signals include high-frequency eddy current signals and multi-spectral near-infrared signals collected by a self-cleaning wear-resistant alloy probe. The high-frequency eddy current signals include coil impedance and inductive reactance values, and the multi-spectral near-infrared signals include absorbance values ​​at water-sensitive wavelengths and absorbance values ​​at clay-sensitive reference wavelengths. At preset time intervals, the compressed air pulse jet cleaning device of the self-cleaning wear-resistant alloy probe mounting base is activated to remove molding sand adhering to the probe end surface, ensuring the stability of the detection signals. The real-time detection signal is input into a pre-trained signal fusion model to obtain the real-time moisture content of the molding sand and the real-time clay blue absorption. The signal fusion model is a support vector machine regression model trained with multiple sets of standard molding sand sample data. The standard molding sand sample data includes high-frequency eddy current signals and multi-spectral near-infrared signals corresponding to different moisture contents and different clay blue absorption. Based on the preset target threshold for casting density, the density-molding sand parameter correlation database is queried to determine the target range for molding sand moisture content and the target range for clay bluing amount. Calculate the first deviation value between the real-time moisture content and the target range of the moisture content, and the second deviation value between the real-time clay blue absorption amount and the target range of the clay blue absorption amount; Based on the first deviation value, a moisture control command is generated to control the water addition device of the sand mixer to adjust the instantaneous water addition; based on the second deviation value, a clay control command is generated to control the clay conveying device to adjust the conveying amount per unit time. Obtain the density test result of the molding sand casting after regulation. If the density test result does not reach the density target threshold, then based on the difference between the density test result and the density target threshold, correct the output weight of the signal fusion model. Iteratively execute the above steps of obtaining the real-time detection signal of molding sand during the sand mixing process, inputting the real-time detection signal into the pre-trained signal fusion model to obtain real-time parameters, calculating the deviation value and generating a moisture regulation command, and correcting the output weight of the signal fusion model until the density of the casting reaches the density target threshold. The step of querying the density-molding sand parameter correlation database based on the preset casting density target threshold to determine the target range of molding sand moisture content and clay bluing absorption includes: A density-molding sand parameter correlation database is constructed, which stores historical data of casting density corresponding to different casting materials under different molding sand moisture contents and different clay bluing amounts. Based on the aforementioned associated database, a linear regression algorithm was used to fit the functional relationship between moisture content, clay blue absorption, and casting density. Substituting the preset target threshold for casting density into the functional relationship, the corresponding target ranges for molding sand moisture content and clay blue absorption are obtained by reverse calculation. If the material is a new casting material, several small-batch trial productions are carried out first to collect density data corresponding to different molding sand parameters, which are then added to the associated database. After correcting the functional relationship, the target range is determined.

2. The method according to claim 1, characterized in that, The real-time detection signal of molding sand during the sand mixing process includes: The high-frequency eddy current module of the self-cleaning wear-resistant alloy probe generates an alternating electromagnetic field of a fixed frequency. When molding sand flows through the probe's detection end, the impedance and reactance values ​​of the coil under the action of the electromagnetic field are collected. The multi-spectral near-infrared module of the self-cleaning wear-resistant alloy probe is synchronously controlled to emit near-infrared light of water-sensitive wavelength and clay-sensitive reference wavelength, receive the light signal after diffuse reflection of molding sand, and convert it into the corresponding absorbance value. At preset time intervals, the compressed air pulse jet device of the self-cleaning wear-resistant alloy probe mounting base is activated to remove the molding sand adhering to the surface of the probe end, ensuring the stability of the detection signal.

3. The method according to claim 1, characterized in that, The training process of the pre-trained signal fusion model includes: Prepare multiple sets of standard molding sand samples. The moisture content and clay blue absorption of each set of samples are within a predetermined range, and the number of samples in each set is not less than the predetermined number. High-frequency eddy current signals and multi-spectral near-infrared signals were collected from each group of standard molding sand samples to construct a sample dataset; The sample dataset is divided into a training set and a validation set according to the proportion. The training set is used to train a support vector machine regression model with high-frequency eddy current signal and multi-spectral near-infrared signal as input features and the actual moisture content and actual clay blue absorption of standard molding sand samples as output labels. The model's prediction accuracy is verified using the validation set. If the prediction error for moisture content exceeds ± a preset threshold or the prediction error for clay blue absorption exceeds a preset threshold, the number of samples is increased and the model kernel function parameters are adjusted until the prediction accuracy meets the requirements, thus completing model training.

4. The method according to claim 1, characterized in that, The moisture control command is generated based on the first deviation value to control the water addition device of the sand mixer to adjust the instantaneous water addition. Based on the second deviation value, a clay control command is generated to control the clay conveying device to adjust the conveying rate per unit time, including: If the first deviation value is negative, calculate the required water replenishment amount = absolute value of the first deviation value × instantaneous flow rate of molding sand. Based on the maximum water supply capacity of the water supply device, determine the adjustment range of the instantaneous water replenishment amount and generate a water control command. If the first deviation value is positive, a moisture control command is generated to pause water addition or reduce the amount of water added, until the real-time moisture content falls back to the target range. If the second deviation value is negative, calculate the amount of clay to be added = absolute value of the second deviation value × instantaneous flow rate of molding sand ÷ actual amount of clay absorbed. Based on the conveying efficiency of the clay conveying device, determine the adjustment range of the conveying amount per unit time and generate a clay control command. If the second deviation value is positive, a clay control command is generated to pause clay transport or reduce the transport volume until the real-time clay absorption volume falls back to the target range.

5. The method according to claim 1, characterized in that, The density test results of the regulated molding sand castings include: Several samples were randomly selected from the castings produced after adjustment, and the volume percentage of internal defects in each sample was detected by an ultrasonic flaw detector. Calculate the actual density of the casting = 1 - defect volume ratio, and take the average of the actual density values ​​of multiple samples as the density test result; If the deviation between the actual value and the average value of a single sample compaction exceeds a preset threshold, the sample is re-extracted for testing to ensure the reliability of the test results.

6. A casting density optimization system based on the synergistic regulation of molding sand moisture and clay bluing capacity, characterized in that, The system includes: The self-cleaning molding sand testing module includes a wear-resistant alloy probe, a high-frequency eddy current unit, a multi-spectral near-infrared unit, and a pulse cleaning unit. The wear-resistant alloy probe is fixed to the side wall of the sand mixer, with the probe end facing the molding sand flow path. The high-frequency eddy current unit is used to generate an alternating electromagnetic field and collect coil impedance and inductive reactance values. The multi-spectral near-infrared unit is used to emit dual-wavelength near-infrared light and collect absorbance values. The pulse cleaning unit is used to periodically clean the molding sand from the probe surface. The signal processing and model calculation module includes a microprocessor and a pre-stored signal fusion model. The microprocessor is used to receive the real-time signal from the self-cleaning molding sand detection module, call the signal fusion model to calculate the real-time moisture content and the real-time clay blue absorption, and compare the deviation value with the target interval. The control module includes a water supply device for the sand mixer, a clay conveying device, and a control unit. The control unit receives the control instructions from the signal processing and model calculation module, and drives the water adding device to adjust the water adding amount and drives the clay conveying device to adjust the conveying amount. The density detection and model correction module includes an ultrasonic flaw detector and a data storage unit. The ultrasonic flaw detector is used to detect the density of the casting, and the data storage unit records the density deviation data, which is used to correct the output weights of the signal fusion model. The central control module is communicatively connected to the self-cleaning molding sand detection module, signal processing and model calculation module, execution and control module, and density detection and model correction module, respectively, to coordinate the working sequence of each module and realize closed-loop control of data acquisition, calculation, control, and detection; The determination of the target interval includes: A density-molding sand parameter correlation database is constructed, which stores historical data of casting density corresponding to different casting materials under different molding sand moisture contents and different clay bluing amounts. Based on the aforementioned associated database, a linear regression algorithm was used to fit the functional relationship between moisture content, clay blue absorption, and casting density. Substituting the preset target threshold for casting density into the functional relationship, the corresponding target ranges for molding sand moisture content and clay blue absorption are obtained by reverse calculation. If the material is a new casting material, several small-batch trial productions are carried out first to collect density data corresponding to different molding sand parameters, which are then added to the associated database. After correcting the functional relationship, the target range is determined.

7. The system according to claim 6, characterized in that, The detection end of the wear-resistant alloy probe adopts a multi-layer composite structure, consisting of a wear-resistant insulating ceramic layer, a high-frequency eddy current coil, and a heat insulation layer from the outside to the inside. A multi-spectral near-infrared light channel is set in the central axis of the probe, and an LED light source and photodetector are built in. The central control module adopts a communication method combining industrial Ethernet and IO-Link bus, and communicates with the signal processing and model calculation module. The central control module receives the working status feedback signals of each module in real time. If a module fails, it will immediately issue an alarm signal and suspend the sand mixing operation. The density-molding sand parameter association database is stored in the data storage unit. The data storage unit supports real-time data updates and queries, and has a data backup function to prevent the loss of historical data. The signal processing and model calculation module supports online correction of model parameters, and can complete model optimization without stopping system operation.

8. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the optimization method according to any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the optimization method according to any one of claims 1-5.

Images