Mineral casting production data analysis method and system

Abstract

The invention provides a mineral casting production data analysis method, and relates to the technical field of production monitoring, and the method comprises the steps: outputting the prediction probability of the existence of micro holes and cracks based on the correlation mapping of mineral particle distribution and curing defects according to the time-frequency characteristics in an input vibration frequency spectrum and the energy entropy of an acoustic emission signal, and then carrying out the prediction of the existence of the micro holes and cracks; and inputting the time-frequency characteristics and the energy entropy into a target depth model, and regulating and controlling an execution terminal based on an output result of the target depth model. According to the mineral casting production data analysis method provided by the invention, the traditional independently monitored vibration spectrum and the acoustic emission signal are fused through the attention mechanism, the problem that defect prediction cannot be carried out by a single data source is solved, and meanwhile, process parameters are dynamically adjusted in real-time defect probability prediction, and defect prevention type production is realized.

Description

Technical Field

[0001] The present application relates to the field of production monitoring, and in particular to a method and system for analyzing mineral casting production data. Background Art

[0002] Mineral castings, also known as artificial granite, are used as basic structural components for high-end equipment. They are generally made of an epoxy resin matrix and mineral fillers (quartz sand / basalt).

[0003] The curing process of mineral castings involves the coupling of multiple physical fields, including non-Newtonian fluid flow, particle settling, and exothermic reactions. Traditional empirical process control methods struggle to ensure consistency. Microvoids and microcracks typically develop during the curing phase, but X-ray inspection requires 24 hours, resulting in high scrap rates. Therefore, timely control of the mineral casting production process based on available monitoring data is a pressing issue. Summary of the Invention

[0004] The embodiments of the present application provide a method and system for analyzing mineral casting production data to improve the above-mentioned problems.

[0005] To achieve the above objectives, this application adopts the following technical solutions:

[0006] In a first aspect, the present application proposes a mineral casting production data analysis method applicable to a mineral casting production system, wherein the mineral casting production system includes a monitoring terminal, an execution terminal, and a control terminal, wherein the monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal, and the execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and control the casting speed. The method is applicable to the control terminal and includes the following steps:

[0007] The control terminal obtains the vibration spectrum of the mineral casting to be tested during the pouring and molding stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be tested after solidification based on the ultrasonic detection terminal;

[0008] The control terminal obtains a target deep learning model, which is based on the correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal;

[0009] The time-frequency features and energy entropy are input into the target deep model, and the execution terminal is regulated based on the output results of the target deep model.

[0010] In conjunction with the first aspect, in some embodiments, a target deep learning model is obtained. The target deep learning model is based on a correlation mapping between mineral particle distribution and solidification defects, and is used to output a predicted probability of the presence of microvoids and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal, including:

[0011] The control terminal decomposes the acoustic emission signal into multiple sub-bands using wavelet packets and extracts the energy proportion of each sub-band as a feature vector;

[0012] The control terminal obtains the Mel-frequency cepstral coefficients of the vibration spectrum, fuses the Mel-frequency cepstral coefficients with multiple feature vectors, and inputs the fused target vector into the target model;

[0013] The control terminal obtains training data, trains the target model based on the training data, and stops training and obtains the trained target model when the training process meets the preset conditions.

[0014] In combination with the first aspect, in some embodiments, obtaining Mel-frequency cepstral coefficients of a vibration spectrum and fusing the Mel-frequency cepstral coefficients with a plurality of eigenvectors includes:

[0015] The control terminal performs time axis matching on the Mel-frequency cepstral coefficients of the vibration spectrum and the multiple eigenvectors;

[0016] The control terminal performs dimension-raising and splicing on the acoustic emission feature vector and the Mel-frequency cepstral coefficient to obtain multiple target vectors, and normalizes the target vectors.

[0017] In conjunction with the first aspect, in some embodiments, the control terminal obtains training data, trains a target model based on the training data, and when the training process meets a preset condition, stops training and obtains the trained target model, including:

[0018] The control terminal obtains production line sensor data in real time through the industrial Internet of Things interface. The production line sensor data includes the vibration spectrum of the mineral casting to be tested during the casting and molding stage, and the acoustic emission signal of the mineral casting to be tested after solidification.

[0019] The control terminal synchronously retrieves the physical verification results from the quality inspection database through the industrial Internet of Things interface. The verification results include defect location and size data from X-ray inspection and internal crack depth reports recorded by ultrasonic inspection.

[0020] The control terminal inputs the production line sensor data and physical verification results into the target depth model for iterative calculation and obtains the error value. When the error value is less than the preset value, the calculation is stopped and the trained target depth model is output.

[0021] In conjunction with the first aspect, in some embodiments, the control terminal inputs the production line sensor data and the physical verification results into the target depth model for iterative calculation, and obtains an error value. When the error value is less than a preset value, the calculation is stopped and the trained target depth model is output, including:

[0022] The control terminal inputs the production line sensor data and physical verification results into the target depth model and obtains the initial prediction results of the target depth model;

[0023] The control terminal compares the initial prediction results with the physical verification results, determines the error value based on the comparison results, and stops the calculation when the error value is less than the preset value and outputs the trained target depth model.

[0024] In conjunction with the first aspect, in some embodiments, the time-frequency features and energy entropy are input into a target depth model, and based on the output result of the target depth model, the execution terminal is regulated, including:

[0025] The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the variance of the solidification energy distribution is less than or equal to a preset threshold.

[0026] In conjunction with the first aspect, in some embodiments, the monitoring terminal further includes a temperature monitoring terminal, and the control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the curing energy distribution variance is less than or equal to a preset threshold, including:

[0027] The control terminal uses a fuzzy PID algorithm, in which the input variables are the predicted probability of the existence of micropores and cracks, and the standard deviation of the temperature gradient during the casting and molding stage of the mineral casting to be tested. The output variables are the vibration table frequency compensation Δf and the casting valve opening correction coefficient k. Among them, when the predicted probability of the existence of micropores and cracks is greater than 0.3 and the standard deviation of the temperature gradient is greater than 8℃ / cm, Δf is increased by 15-20Hz and k is reduced to 0.7-0.8.

[0028] In conjunction with the first aspect, in some embodiments, the method further includes:

[0029] When the predicted probability of the existence of micro-voids and cracks exceeds the target threshold, the sensor data stream of the associated process is traced back;

[0030] Based on graph convolutional networks, key abnormal factors in mixing uniformity, pouring angle, and ambient humidity are identified, and a root cause analysis report is generated.

[0031] In a second aspect, the embodiments of the present application further provide a mineral casting production data analysis system, which includes a monitoring terminal, an execution terminal, and a control terminal. The monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal. The execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and to control the casting speed. The system is configured as follows:

[0032] The control terminal obtains the vibration spectrum of the mineral casting to be tested during the pouring and molding stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be tested after solidification based on the ultrasonic detection terminal;

[0033] The control terminal obtains a target deep learning model, which is based on the correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal;

[0034] The time-frequency features and energy entropy are input into the target deep model, and the execution terminal is regulated based on the output results of the target deep model.

[0035] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0036] Obtain a target deep learning model. The target deep learning model is based on the correlation mapping between mineral particle distribution and solidification defects. It is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal, including:

[0037] The control terminal decomposes the acoustic emission signal into multiple sub-bands using wavelet packets and extracts the energy proportion of each sub-band as a feature vector;

[0038] The control terminal obtains the Mel-frequency cepstral coefficients of the vibration spectrum, fuses the Mel-frequency cepstral coefficients with multiple feature vectors, and inputs the fused target vector into the target model;

[0039] The control terminal obtains training data, trains the target model based on the training data, and stops training and obtains the trained target model when the training process meets the preset conditions.

[0040] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0041] Obtain the Mel-frequency cepstral coefficients of the vibration spectrum and fuse them with multiple eigenvectors, including:

[0042] The control terminal performs time axis matching on the Mel-frequency cepstral coefficients of the vibration spectrum and the multiple eigenvectors;

[0043] The control terminal performs dimension-raising and splicing on the acoustic emission feature vector and the Mel-frequency cepstral coefficient to obtain multiple target vectors, and normalizes the target vectors.

[0044] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0045] The control terminal obtains training data and trains the target model based on the training data. When the training process meets the preset conditions, the training is stopped and the trained target model is obtained, including:

[0046] The control terminal obtains production line sensor data in real time through the industrial Internet of Things interface. The production line sensor data includes the vibration spectrum of the mineral casting to be tested during the casting and molding stage, and the acoustic emission signal of the mineral casting to be tested after solidification.

[0047] The control terminal synchronously retrieves the physical verification results from the quality inspection database through the industrial Internet of Things interface. The verification results include defect location and size data from X-ray inspection and internal crack depth reports recorded by ultrasonic inspection.

[0048] The control terminal inputs the production line sensor data and physical verification results into the target depth model for iterative calculation and obtains the error value. When the error value is less than the preset value, the calculation is stopped and the trained target depth model is output.

[0049] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0050] The control terminal inputs the production line sensor data and physical verification results into the target depth model for iterative calculation and obtains the error value. When the error value is less than the preset value, the calculation stops and the trained target depth model is output, including:

[0051] The control terminal inputs the production line sensor data and physical verification results into the target depth model and obtains the initial prediction results of the target depth model;

[0052] The control terminal compares the initial prediction results with the physical verification results, determines the error value based on the comparison results, and stops the calculation when the error value is less than the preset value and outputs the trained target depth model.

[0053] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0054] The time-frequency features and energy entropy are input into the target deep model. Based on the output of the target deep model, the execution terminal is regulated, including:

[0055] The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the variance of the solidification energy distribution is less than or equal to a preset threshold.

[0056] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0057] The monitoring terminal also includes a temperature monitoring terminal. The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the curing energy distribution variance is less than or equal to the preset threshold, including:

[0058] The control terminal uses a fuzzy PID algorithm, in which the input variables are the predicted probability of the existence of micropores and cracks, and the standard deviation of the temperature gradient during the casting and molding stage of the mineral casting to be tested. The output variables are the vibration table frequency compensation Δf and the casting valve opening correction coefficient k. Among them, when the predicted probability of the existence of micropores and cracks is greater than 0.3 and the standard deviation of the temperature gradient is greater than 8℃ / cm, Δf is increased by 15-20Hz and k is reduced to 0.7-0.8.

[0059] In conjunction with the second aspect, in some embodiments, the system is configured to:

[0060] When the predicted probability of the existence of micro-voids and cracks exceeds the target threshold, the sensor data stream of the associated process is traced back;

[0061] Based on graph convolutional networks, key abnormal factors in mixing uniformity, pouring angle, and ambient humidity are identified, and a root cause analysis report is generated.

[0062] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0063] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method proposed in the first aspect of the embodiment of the present invention.

[0064] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in the first aspect of the embodiment of the present invention.

[0065] In summary, the above method and system have the following technical effects:

[0066] This application proposes a method for analyzing mineral casting production data. This method uses a correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal. The time-frequency characteristics and energy entropy are then input into a target depth model, and the execution terminal is regulated based on the output of the target depth model. This application proposes a method for analyzing mineral casting production data. This method fuses the traditional independently monitored vibration spectrum and acoustic emission signals through an attention mechanism, solving the problem of a single data source being unable to predict defects. It also dynamically adjusts process parameters based on real-time defect probability prediction to achieve defect-preventive production. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a flow chart of a mineral casting production data analysis method proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0069] This application proposes a mineral casting production data analysis method, which is applicable to a mineral casting production system. The mineral casting production system includes a monitoring terminal, an execution terminal, and a control terminal. The monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal. The execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and to control the casting speed. The method is applicable to the control terminal. Figure 1 , including the following steps:

[0070] S101: The control terminal obtains the vibration spectrum of the mineral casting to be detected during the casting and molding stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be detected after solidification based on the ultrasonic detection terminal.

[0071] During the pouring process of mineral castings, the flow mixing of the resin matrix and the mineral filler generates mechanical vibrations. After eliminating the externally applied vibrations, its spectral characteristics can directly reflect the uniformity of particle distribution and flow shear force. Generally speaking, high-frequency harmonics (>1kHz) characterize the interface friction between the resin and the filler. In some cases, it may also be caused by bubble generation. The spectrum can be obtained by fixing the base to the outer wall of the casting mold or other positions, which is not limited in this application.

[0072] When mineral castings solidify, epoxy resin cross-links and releases heat, and microcracks / pores are generated inside, which release elastic waves. For example, when cracks expand, burst acoustic emission signals (duration <100μs) are generated. When the interface peels, it manifests as continuous acoustic emission (energy concentrated at 30-80kHz). When the pores collapse, they are characterized by low-frequency modal waves (<20kHz).

[0073] S102: The control terminal obtains a target deep learning model, which is based on the correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the existence of micropores and cracks based on the time-frequency characteristics in the input vibration spectrum and the energy entropy of the acoustic emission signal.

[0074] As you can understand, this model is based on a correlation mapping between mineral particle distribution and solidification defects. It is designed to analyze and process input data. Specifically, it predicts the probability of the presence of microvoids and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal. In this way, the model can provide an assessment of the internal structural integrity of the material, which is important for ensuring material quality and preventing potential structural failures.

[0075] Specifically, the control terminal decomposes the acoustic emission signal into multiple sub-bands through wavelet packets, and extracts the energy proportion of each sub-band as a feature vector.

[0076] For example, to cover the 0-100 kHz damage characteristic frequency band, 16 sub-bands are used as an example. In other embodiments, other numbers can be used, which is not limited in this application. It is understandable that a sudden increase in energy in the high-frequency band corresponds to microcracks, while a change in the low-frequency band reflects pore collapse.

[0077] For example, the AE signal is completely decomposed into four layers using a db10 wavelet basis that matches the fracture frequency response of basalt fiber, generating 16 sub-bands (6.25 kHz bandwidth) covering the critical damage response frequency band of 0-100 kHz.

[0078] Then, the control terminal obtains the Mel-cepstral coefficients of the vibration spectrum, fuses the Mel-cepstral coefficients with multiple eigenvectors, and inputs the fused target vector into the target model. Exemplarily, in this embodiment, a splicing fusion is adopted, that is, after the Mel-cepstral coefficients of the vibration spectrum are matched with multiple eigenvectors on the time axis, the acoustic emission eigenvectors are spliced ​​with the Mel-cepstral coefficients in a dimension-upgraded manner to obtain multiple target vectors, and the target vectors are normalized. After normalization, the dimensional effect can be eliminated and the model training effect can be improved. The fused eigenvector is input into the target deep learning model (such as CNN, LSTM or Transformer) to predict the probability of micropores and cracks.

[0079] The control terminal obtains training data, trains the target model based on the training data, and stops training and obtains the trained target model when the training process meets the preset conditions.

[0080] For example, training data can be physical verification results from a quality inspection database, retrieved synchronously via an industrial IoT interface. These verification results include defect location and size data from X-ray inspections and internal crack depth reports from ultrasonic testing. Production line sensor data and physical verification results are then fed into the target depth model for iterative calculations. Error values ​​are then obtained. When the error value falls below a preset value, the calculations are terminated and the trained target depth model is output.

[0081] Specifically, the control terminal can input the production line sensor data and physical verification results into the target depth model and obtain the initial prediction results of the target depth model; the control terminal compares the initial prediction results with the physical verification results, determines the error value based on the comparison results, and stops the calculation when the error value is less than the preset value and outputs the trained target depth model.

[0082] S103: Input the time-frequency features and energy entropy into the target depth model, and regulate the execution terminal based on the output results of the target depth model.

[0083] As can be understood, in this embodiment, the control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time, so that the variance of the solidification energy distribution is less than or equal to a preset threshold. Exemplarily, the monitoring terminal in this embodiment also includes a temperature monitoring terminal, and the control terminal uses a fuzzy PID algorithm, in which the input variables are the predicted probability of the presence of micropores and cracks, and the standard deviation of the temperature gradient during the casting and molding stage of the mineral casting to be tested. The energy distribution variance is a characteristic of the uniformity of solidification within the casting. Therefore, the variance of the solidification energy distribution can be made less than or equal to a preset threshold to ensure that the final predicted product has fewer defects.

[0084] In other embodiments, when the predicted probability of the existence of microvoids and cracks exceeds the target threshold, the control terminal traces back the sensor data stream of the associated process to more accurately determine the location of the problem. At the same time, it can also identify key abnormal factors in mixing uniformity, pouring inclination, and ambient humidity based on the graph convolutional network to generate a root cause analysis report.

[0085] This application proposes a method for analyzing mineral casting production data. This method uses a correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal. The time-frequency characteristics and energy entropy are then input into a target depth model, and the execution terminal is regulated based on the output of the target depth model. This application proposes a method for analyzing mineral casting production data. This method fuses the traditional independently monitored vibration spectrum and acoustic emission signals through an attention mechanism, solving the problem of a single data source being unable to predict defects. It also dynamically adjusts process parameters based on real-time defect probability prediction to achieve defect-preventive production.

[0086] Based on the same inventive concept, the present application also proposes a mineral casting production data analysis system. The mineral casting production data analysis system includes a monitoring terminal, an execution terminal, and a control terminal. The monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal. The execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and to control the casting speed. The system is configured as follows:

[0087] The control terminal obtains the vibration spectrum of the mineral casting to be tested during the pouring and molding stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be tested after solidification based on the ultrasonic detection terminal;

[0088] The control terminal obtains a target deep learning model, which is based on the correlation mapping between mineral particle distribution and solidification defects, and is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal;

[0089] The time-frequency features and energy entropy are input into the target deep model, and the execution terminal is regulated based on the output results of the target deep model.

[0090] In some embodiments, the system is configured to:

[0091] Obtain a target deep learning model. The target deep learning model is based on the correlation mapping between mineral particle distribution and solidification defects. It is used to output the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal, including:

[0092] The control terminal decomposes the acoustic emission signal into multiple sub-bands using wavelet packets and extracts the energy proportion of each sub-band as a feature vector;

[0093] The control terminal obtains the Mel-frequency cepstral coefficients of the vibration spectrum, fuses the Mel-frequency cepstral coefficients with multiple feature vectors, and inputs the fused target vector into the target model;

[0094] The control terminal obtains training data, trains the target model based on the training data, and stops training and obtains the trained target model when the training process meets the preset conditions.

[0095] In some embodiments, the system is configured to:

[0096] Obtain the Mel-frequency cepstral coefficients of the vibration spectrum and fuse them with multiple eigenvectors, including:

[0097] The control terminal performs time axis matching on the Mel-frequency cepstral coefficients of the vibration spectrum and the multiple eigenvectors;

[0098] The control terminal performs dimension-raising and splicing on the acoustic emission feature vector and the Mel-frequency cepstral coefficient to obtain multiple target vectors, and normalizes the target vectors.

[0099] In some embodiments, the system is configured to:

[0100] The control terminal obtains training data and trains the target model based on the training data. When the training process meets the preset conditions, the training is stopped and the trained target model is obtained, including:

[0101] The control terminal obtains production line sensor data in real time through the industrial Internet of Things interface. The production line sensor data includes the vibration spectrum of the mineral casting to be tested during the casting and molding stage, and the acoustic emission signal of the mineral casting to be tested after solidification.

[0102] The control terminal synchronously retrieves the physical verification results from the quality inspection database through the industrial Internet of Things interface. The verification results include defect location and size data from X-ray inspection and internal crack depth reports recorded by ultrasonic inspection.

[0103] The control terminal inputs the production line sensor data and physical verification results into the target depth model for iterative calculation and obtains the error value. When the error value is less than the preset value, the calculation is stopped and the trained target depth model is output.

[0104] In some embodiments, the system is configured to:

[0105] The control terminal inputs the production line sensor data and physical verification results into the target depth model for iterative calculation and obtains the error value. When the error value is less than the preset value, the calculation stops and the trained target depth model is output, including:

[0106] The control terminal inputs the production line sensor data and physical verification results into the target depth model and obtains the initial prediction results of the target depth model;

[0107] The control terminal compares the initial prediction results with the physical verification results, determines the error value based on the comparison results, and stops the calculation when the error value is less than the preset value and outputs the trained target depth model.

[0108] In some embodiments, the system is configured to:

[0109] The time-frequency features and energy entropy are input into the target deep model. Based on the output of the target deep model, the execution terminal is regulated, including:

[0110] The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the variance of the solidification energy distribution is less than or equal to a preset threshold.

[0111] In some embodiments, the system is configured to:

[0112] The monitoring terminal also includes a temperature monitoring terminal. The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the curing energy distribution variance is less than or equal to the preset threshold, including:

[0113] The control terminal uses a fuzzy PID algorithm, in which the input variables are the predicted probability of the existence of micropores and cracks, and the standard deviation of the temperature gradient during the casting and molding stage of the mineral casting to be tested. The output variables are the vibration table frequency compensation Δf and the casting valve opening correction coefficient k. Among them, when the predicted probability of the existence of micropores and cracks is greater than 0.3 and the standard deviation of the temperature gradient is greater than 8℃ / cm, Δf is increased by 15-20Hz and k is reduced to 0.7-0.8.

[0114] In some embodiments, the system is configured to:

[0115] When the predicted probability of the existence of micro-voids and cracks exceeds the target threshold, the sensor data stream of the associated process is traced back;

[0116] Based on graph convolutional networks, key abnormal factors in mixing uniformity, pouring angle, and ambient humidity are identified, and a root cause analysis report is generated.

[0117] This application proposes a mineral casting production data analysis system that uses a correlation mapping between mineral particle distribution and solidification defects. It outputs the predicted probability of the presence of micropores and cracks based on the time-frequency characteristics of the input vibration spectrum and the energy entropy of the acoustic emission signal. The time-frequency characteristics and energy entropy are then input into a target depth model, and the execution terminal is regulated based on the output of the target depth model. This application proposes a mineral casting production data analysis system that fuses the traditional independently monitored vibration spectrum and acoustic emission signals through an attention mechanism, solving the problem of a single data source being unable to predict defects. At the same time, it dynamically adjusts process parameters based on real-time defect probability prediction to achieve defect-preventive production.

[0118] Based on the same inventive concept, an embodiment of the present application further provides an electronic device, the electronic device comprising:

[0119] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the mineral casting production data analysis method according to an embodiment of the present application.

[0120] In addition, to achieve the above-mentioned purpose, an embodiment of the present application further proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the mineral casting production data analysis method of the embodiment of the present application.

[0121] The following is a detailed introduction to the various components of electronic equipment:

[0122] The term "processor" is the control center of an electronic device and may be a single processor or a collective term for multiple processing elements. For example, the processor may be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0123] Optionally, the processor can perform various functions of the electronic device by running or executing a software program stored in the memory, and calling data stored in the memory.

[0124] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0125] Alternatively, the memory may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processor through an interface circuit of the electronic device, and this is not specifically limited in the embodiments of the present invention.

[0126] A transceiver is used to communicate with network devices or terminal devices.

[0127] Optionally, the transceiver may include a receiver and a transmitter, wherein the receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0128] Optionally, the transceiver may be integrated with the processor, or may exist independently and be coupled to the processor via an interface circuit of the router, which is not specifically limited in the embodiment of the present invention.

[0129] In addition, the technical effects of the electronic device can refer to the technical effects of the data transmission method in the above method embodiment, and will not be repeated here.

[0130] It should be understood that the processor in the embodiments of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0131] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0132] The above embodiments can be implemented in whole or in part through software, hardware (such as circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function according to the embodiments of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired method (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available media can be magnetic media (such as floppy disks, hard disks, tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0133] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0134] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0135] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0136] Those skilled in the art will appreciate that the units and algorithm steps of each example 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 performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

Claims

1. A method for analyzing mineral casting production data, characterized in that: The invention is applicable to a mineral casting production system, which includes a monitoring terminal, an execution terminal, and a control terminal, wherein the monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal, and the execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and control the casting speed. The method is applicable to the control terminal and includes the following steps: The control terminal obtains the vibration spectrum of the mineral casting to be detected during the pouring and forming stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be detected after solidification based on the ultrasonic detection terminal; The control terminal obtains a target deep learning model, where the target deep learning model is based on a correlation mapping between mineral particle distribution and solidification defects, and is used to output a predicted probability of the presence of micropores and cracks based on the input time-frequency characteristics in the vibration spectrum and the energy entropy of the acoustic emission signal; The time-frequency features and the energy entropy are input into the target depth model, and the execution terminal is regulated based on an output result of the target depth model.

2. A mineral casting production data analysis method according to claim 1, characterized in that: Obtaining a target deep learning model, wherein the target deep learning model is based on a correlation mapping between mineral particle distribution and solidification defects, and is used to output a predicted probability of the presence of micropores and cracks based on the time-frequency characteristics in the input vibration spectrum and the energy entropy of the acoustic emission signal, including: The control terminal decomposes the acoustic emission signal into multiple sub-bands using wavelet packets, and extracts the energy proportion of each sub-band as a feature vector; The control terminal obtains the Mel-cepstral coefficients of the vibration spectrum, fuses the Mel-cepstral coefficients with the plurality of feature vectors, and inputs the fused target vector into the target model; The control terminal obtains training data, trains the target model based on the training data, and stops training and obtains the trained target model when the training process meets preset conditions.

3. A mineral casting production data analysis method according to claim 2, characterized in that: Obtaining Mel-frequency cepstral coefficients of the vibration spectrum and fusing the Mel-frequency cepstral coefficients with the plurality of feature vectors, including: The control terminal performs time axis matching on the Mel-frequency cepstral coefficients of the vibration spectrum and the plurality of feature vectors; The control terminal performs dimension-raising splicing on the acoustic emission feature vector and the Mel-frequency cepstral coefficient to obtain a plurality of target vectors, and normalizes the target vectors.

4. A mineral casting production data analysis method according to claim 2, characterized in that: The control terminal acquires training data, trains the target model based on the training data, and stops training and acquires the trained target model when the training process meets a preset condition, including: The control terminal acquires production line sensor data in real time through the industrial Internet of Things interface, wherein the production line sensor data includes the vibration spectrum of the produced mineral casting to be tested during the casting and molding stage, and the acoustic emission signal of the produced mineral casting to be tested after solidification; The control terminal synchronously retrieves the physical verification results from the quality inspection database through the industrial Internet of Things interface, wherein the verification results include defect location and size data detected by X-ray flaw detection and internal crack depth report recorded by ultrasonic flaw detection; The control terminal inputs the production line sensor data and the physical verification results into the target depth model for iterative calculation and obtains an error value. When the error value is less than a preset value, the calculation is stopped and the trained target depth model is output.

5. A mineral casting production data analysis method according to claim 4, characterized in that: The control terminal inputs the production line sensor data and the physical verification result into the target depth model for iterative calculation and obtains an error value. When the error value is less than a preset value, the calculation is stopped and the trained target depth model is output, including: The control terminal inputs the production line sensor data and the physical verification result into the target depth model, and obtains an initial prediction result of the target depth model; The control terminal compares the initial prediction result with the physical verification result, determines an error value based on the comparison result, and stops calculating and outputs the trained target depth model when the error value is less than a preset value.

6. A mineral casting production data analysis method according to claim 1, characterized in that: Inputting the time-frequency feature and the energy entropy into the target depth model, and regulating the execution terminal based on an output result of the target depth model, including: The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the solidification energy distribution variance is less than or equal to a preset threshold.

7. A mineral casting production data analysis method according to claim 6, characterized in that: The monitoring terminal further includes a temperature monitoring terminal. The control terminal sends instructions to the execution terminal to adjust the pouring flow rate and vibration frequency of the execution terminal in real time so that the curing energy distribution variance is less than or equal to a preset threshold, including: The control terminal adopts a fuzzy PID algorithm, wherein the input variables are the predicted probability of the existence of the micropores and cracks and the standard deviation of the temperature gradient during the casting and molding stage of the mineral casting to be tested, and the output variables are the vibration table frequency compensation Δf and the casting valve opening correction coefficient k. Among them, when the predicted probability of the existence of the micropores and cracks is greater than 0.3 and the standard deviation of the temperature gradient is greater than 8°C / cm, Δf is increased by 15-20Hz and k is reduced to 0.7-0.

8.

8. A mineral casting production data analysis method according to claim 1, characterized in that: The method further comprises: When the predicted probability of the existence of the micro-voids and cracks exceeds a target threshold, the sensor data stream of the associated process is backtracked; Based on graph convolutional networks, key abnormal factors in mixing uniformity, pouring angle, and ambient humidity are identified, and a root cause analysis report is generated.

9. A mineral casting production data analysis system, characterized in that: The mineral casting production data analysis system includes a monitoring terminal, an execution terminal, and a control terminal. The monitoring terminal includes a vibration detection terminal and an ultrasonic detection terminal. The execution terminal is used to apply vibration to the mineral casting to be detected during the casting and molding stage and to control the casting speed. The system is configured as follows: The control terminal obtains the vibration spectrum of the mineral casting to be detected during the pouring and forming stage based on the vibration detection terminal, and obtains the acoustic emission signal of the mineral casting to be detected after solidification based on the ultrasonic detection terminal; The control terminal obtains a target deep learning model, where the target deep learning model is based on a correlation mapping between mineral particle distribution and solidification defects, and is used to output a predicted probability of the presence of micropores and cracks based on the input time-frequency characteristics in the vibration spectrum and the energy entropy of the acoustic emission signal; The time-frequency features and the energy entropy are input into the target depth model, and the execution terminal is regulated based on an output result of the target depth model.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively coupled to at least one of the processors; The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors so that the at least one processor can execute any one of the methods proposed in items 1-8 of the present invention.

Images