Mold temperature detection method and system

By training the mold area division and temperature prediction model, combined with the injection molding process data, the abnormal problem of temperature detection during the mold injection material is solved, real-time monitoring and abnormal detection of mold temperature is realized, and material performance and molding quality are improved. .

CN119017668AActive Publication Date: 2024-11-26林细弟

Patent Information

Application Number
CN202411124631.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-16
Publication Date
2024-11-26
Estimated Expiration
2044-08-16

AI Technical Summary

Technical Problem

The prior art cannot detect abnormal changes in mold temperature in a timely manner during the mold injection process, resulting in a decline in material performance.

Method used

By dividing the mold into multiple areas, using infrared images to obtain the average temperature data, and combining the feed speed, injection pressure and nozzle temperature data of the injection molding machine, training the temperature prediction model, predicting the mold temperature, calculating the temperature detection error score, and judging Whether the mold temperature is qualified.

Benefits of technology

It improves the accuracy of mold temperature prediction, promptly detects temperature abnormalities, reduces finished product quality problems caused by temperature, and improves material performance and molding quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119017668A_ABST
    Figure CN119017668A_ABST
Patent Text Reader

Abstract

The invention provides a mold temperature detection method and system, and relates to the technical field of mold temperature detection. The method comprises the following steps: dividing a mold into a plurality of areas, and obtaining an infrared image of each area at a plurality of moments in a current detection period; according to the infrared image, determining average temperature data of the mold at multiple moments in the current detection period; acquiring feeding speed data, injection pressure data and nozzle temperature data of the injection molding machine at a plurality of moments of the current detection period; the feeding speed data, the injection pressure data and the nozzle temperature data are processed through a trained temperature prediction model, and predicted temperature data at multiple moments in the current detection period are obtained; determining a temperature detection error score according to the predicted temperature data and the average temperature data; and according to the temperature detection error score and the temperature detection error score threshold, determining whether the mold temperature detection is qualified. According to the invention, the abnormal change of the mold temperature can be detected in time in the process of injecting the material into the mold.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mold temperature detection, and in particular to a mold temperature detection method and system. Background Art

[0002] In the related art, CN115673295A provides a mold baking preheating device and method, which includes an air delivery module, a natural gas delivery module, a mixer, a surface burner, a temperature sensor and a controller. The air delivery module and the natural gas delivery module are both connected to the mixer, and the mixer is used to mix a certain proportion of air and natural gas and then deliver them to the surface burner. The air delivery module, the natural gas delivery module and the temperature sensor are all connected to the controller for communication, and the temperature sensor is used to monitor the surface temperature of the mold and transmit the temperature information to the controller, and the controller is used to control the working state of the air delivery module and the natural gas delivery module according to the temperature information. The device can automatically monitor the surface temperature of the mold, and the temperature monitoring result is highly accurate and the feedback is timely; the temperature information is fed back to the controller, and the controller automatically controls the flame temperature according to the temperature information, thereby improving the quality of mold baking preheating, and the mold heating quality is high and the service life is long.

[0003] CN116275005A relates to a mold temperature detection method and system, belonging to the field of mold temperature control technology, the method includes obtaining an infrared image of the mold after the initial heating; dividing the mold in the infrared image into different areas, and determining the difference between the temperature in each area and the standard temperature; determining the heating temperature rise rate and temperature decay rate corresponding to each area of ​​the mold; determining the heating time required for each area according to the difference, heating temperature rise rate and temperature decay rate, so as to heat the mold according to the heating time. According to this scheme, the problem that uniform heating cannot be guaranteed during the existing mold preheating process is solved.

[0004] Therefore, in the related art, although the mold can be preheated, the temperature of the mold is not detected during the process of injecting materials into the mold, and the abnormal changes in the mold temperature cannot be detected in time during the process of injecting materials into the mold, resulting in a decrease in material performance.

[0005] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0006] The present invention provides a mold temperature detection method and system, which can solve the technical problem that the abnormal change of the mold temperature cannot be detected in time during the process of injecting materials into the mold.

[0007] According to a first aspect of the present invention, there is provided a mold temperature detection method, comprising: dividing the mold into a plurality of areas, obtaining infrared images of each area at a plurality of moments in a current detection cycle; determining average temperature data of the mold at a plurality of moments in the current detection cycle based on the infrared images; obtaining feed speed data, injection pressure data and nozzle temperature data of an injection molding machine at a plurality of moments in the current detection cycle; processing the feed speed data, the injection pressure data and the nozzle temperature data through a trained temperature prediction model to obtain predicted temperature data at a plurality of moments in the current detection cycle; determining a temperature detection error score based on the predicted temperature data and the average temperature data; and determining whether the mold temperature detection is qualified based on the temperature detection error score and a temperature detection error score threshold.

[0008] According to the present invention, based on the infrared image, the average temperature data of the mold at multiple moments in the current detection cycle is determined, including: obtaining the average regional temperature data of each area at multiple moments in the current detection cycle based on the pixel values ​​of the pixel points of the infrared image; averaging the average regional temperature data to determine the average temperature data of the mold at multiple moments in the current detection cycle.

[0009] According to the present invention, the training step of the temperature prediction model includes: obtaining the historical average temperature data of the mold at multiple moments in multiple historical detection cycles, wherein the previous historical detection cycle of the current detection cycle is the first historical detection cycle; obtaining the historical feed speed data, historical injection pressure data and historical nozzle temperature data of the injection molding machine at multiple moments in multiple historical detection cycles; processing the historical feed speed data, the historical injection pressure data and the historical nozzle temperature data through the temperature prediction model to obtain the historical predicted temperature data at multiple moments in multiple historical detection cycles; determining the loss function of the temperature prediction model based on the historical average temperature data, the historical feed speed data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data; and training the temperature prediction model based on the training loss function to obtain the trained temperature prediction model.

[0010] According to the present invention, the loss function of the temperature prediction model is determined based on the historical average temperature data, the historical feed rate data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data, including: according to the formula Determine the loss function Loss of the temperature prediction model, where M s,k is the historical feed rate data at the kth moment of the sth historical detection cycle, M s,1 is the historical feed rate data at the first moment of the sth historical detection cycle, P s,kis the historical injection pressure data at the kth moment of the sth historical detection cycle, P s,1 is the historical injection pressure data at the first moment of the sth historical detection cycle, T s,k is the historical nozzle temperature data at the kth moment of the sth historical detection cycle, T s,1 is the historical nozzle temperature data at the first moment of the sth historical detection cycle, is the historical average temperature data at the kth moment of the sth historical detection cycle, The historical predicted temperature data at the kth moment of the sth historical detection cycle output by the temperature prediction model, m is the number of historical detection cycles, n is the number of moments in the detection cycle, k≤n, s≤m, and k, n, s and m are all positive integers; according to the training loss function, the temperature prediction model is trained to obtain the trained temperature prediction model.

[0011] According to the present invention, a temperature detection error score is determined based on the predicted temperature data and the average temperature data, including: fitting the predicted temperature data and the moments in the current detection cycle to determine the predicted temperature function in the current detection cycle; determining the predicted temperature derivative function based on the predicted temperature function; determining the predicted temperature change rate at multiple moments in the current detection cycle based on the predicted temperature derivative function; fitting the average temperature data and the moments in the current detection cycle to determine the average temperature function in the current detection cycle; determining the average temperature derivative function based on the average temperature function; determining the average temperature change rate at multiple moments in the current detection cycle based on the average temperature derivative function; obtaining a predicted temperature state vector based on the predicted temperature data and the predicted temperature change rate of the mold at multiple moments in the current detection cycle; obtaining an average temperature state vector based on the average temperature data and the average temperature change rate of the mold at multiple moments in the current detection cycle; and determining a temperature detection error score based on the predicted temperature state vector and the average temperature state vector.

[0012] According to the present invention, determining the temperature detection error score according to the predicted temperature state vector and the average temperature state vector includes: according to the formula Determine the temperature detection error score D, where is the average temperature data at the kth moment of the current detection cycle, is the average temperature change rate at the kth moment in the current detection cycle, is the predicted temperature data at the kth moment in the current detection cycle, is the predicted temperature change rate at the kth moment in the current detection cycle, is the average temperature state vector at the kth moment of the current detection cycle, is the predicted temperature state vector at the kth moment in the current detection cycle, n is the number of moments in the detection cycle, k≤n, and both k and n are positive integers, and min is the minimum value function.

[0013] According to the present invention, determining whether the mold temperature detection is qualified according to the temperature detection error score and the temperature detection error score threshold includes: according to the formula Get the first condition C1 and the second condition C2, where D is the temperature detection error score, D p is the temperature detection error score threshold; when the first condition C1 is met, the mold temperature detection is determined to be unqualified; when the second condition C2 is met, the mold temperature detection is determined to be qualified.

[0014] According to a second aspect of the present invention, there is provided a mold temperature detection system, comprising: an infrared image module, for dividing a mold into a plurality of regions, and obtaining infrared images of each region at a plurality of moments in a current detection cycle; an average temperature data module, for determining average temperature data of the mold at a plurality of moments in the current detection cycle based on the infrared images; an injection molding machine data module, for obtaining feed speed data, injection pressure data and nozzle temperature data of the injection molding machine at a plurality of moments in the current detection cycle; a predicted temperature data module, for processing the feed speed data, the injection pressure data and the nozzle temperature data through a trained temperature prediction model, and obtaining predicted temperature data at a plurality of moments in the current detection cycle; a temperature detection error scoring module, for determining a temperature detection error score based on the predicted temperature data and the average temperature data; and a qualified judgment module, for determining whether the mold temperature detection is qualified based on the temperature detection error score and a temperature detection error score threshold.

[0015] Technical effect: According to the present invention, the mold temperature when injecting material into the mold is predicted by combining the trained temperature prediction model with the feed rate data, injection pressure data and nozzle temperature data of the injection molding process, which can improve the accuracy of mold temperature prediction. The predicted temperature data output by the temperature prediction model is compared with the average temperature data actually measured by the mold, so that abnormal changes in mold temperature can be discovered in time, thereby reducing the quality problems of finished products caused by temperature and improving material properties and molding quality. When training the temperature prediction model, the influence of the feed rate data, injection pressure data and nozzle temperature data on the mold temperature can be used to determine the influence of the above data on the error of the predicted temperature data, so as to train the loss function based on the influence and the relative error of the historical predicted temperature data, and the weights are set based on the characteristics that the longer the time interval, the shorter the cycle interval, and the lower the prediction accuracy, so as to improve the objectivity and accuracy of the loss function, thereby reducing the training loss function during the training process, which can improve the training intensity and training efficiency, and improve the accuracy of the temperature prediction model. When determining the temperature detection error score, the temperature detection error score can be determined by predicting the temperature state vector and the average temperature state vector, which can realize real-time monitoring and abnormal detection of the mold temperature during the material injection process, so that the material can be processed under suitable temperature conditions, improving material performance and molding quality. When determining whether the mold temperature detection is qualified, the temperature detection error score and the temperature detection error score threshold can be used to determine whether the mold temperature detection is qualified. If the mold temperature detection is qualified, it can be judged that the temperature of the mold during the material injection process is kept within a reasonable range, thereby improving the consistency of material performance.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and do not limit the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other embodiments can be obtained based on these drawings without creative work.

[0018] Figure 1 A schematic diagram exemplarily shows a flow chart of a mold temperature detection method according to an embodiment of the present invention;

[0019] Figure 2 A block diagram of a mold temperature detection system according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0022] Figure 1 A flow chart of a mold temperature detection method according to an embodiment of the present invention is exemplarily shown, and the method comprises: step S101, dividing the mold into multiple areas, and obtaining infrared images of each area at multiple moments in the current detection cycle; step S102, determining the average temperature data of the mold at multiple moments in the current detection cycle based on the infrared images; step S103, obtaining the feed speed data, injection pressure data and nozzle temperature data of the injection molding machine at multiple moments in the current detection cycle; step S104, processing the feed speed data, injection pressure data and nozzle temperature data through a trained temperature prediction model to obtain predicted temperature data at multiple moments in the current detection cycle; step S105, determining a temperature detection error score based on the predicted temperature data and the average temperature data; step S106, determining whether the mold temperature detection is qualified based on the temperature detection error score and the temperature detection error score threshold.

[0023] According to the mold temperature detection method of an embodiment of the present invention, the mold temperature when injecting material into the mold is predicted through a trained temperature prediction model combined with the feed speed data, injection pressure data and nozzle temperature data of the injection molding process, thereby improving the accuracy of mold temperature prediction. The predicted temperature data output by the temperature prediction model is compared with the average temperature data actually measured for the mold, so that abnormal changes in the mold temperature can be discovered in time, thereby reducing quality problems of finished products caused by temperature and improving material properties and molding quality.

[0024] According to one embodiment of the present invention, in step S101, each detection cycle can be set to 1 minute, 2 minutes, etc., and the interval between adjacent moments can be set to 1 second, 2 seconds, etc., which is not limited by the present invention. The mold is divided into multiple different areas. The area can be divided according to the shape, size or function of the mold to perform more accurate temperature monitoring. At multiple moments in the current detection cycle, an infrared imaging device is used to shoot each area and record the temperature distribution information of the area.

[0025] According to an embodiment of the present invention, in step S102, average temperature data of the mold at multiple moments in the current detection cycle is determined based on the infrared image.

[0026] According to one embodiment of the present invention, step S102 includes: obtaining average regional temperature data of each area at multiple moments in the current detection cycle based on the pixel values ​​of the pixel points of the infrared image; averaging the average regional temperature data to determine the average temperature data of the mold at multiple moments in the current detection cycle.

[0027] According to one embodiment of the present invention, each infrared image is analyzed to extract the pixel value of each pixel in the image. Each pixel value represents the temperature value of the position. The temperature values ​​are added and divided by the number of pixels in the area to calculate the average regional temperature data of the area. The average regional temperature data of each area at multiple moments in the current detection cycle are added and divided by the number of areas to obtain the average temperature data of the mold at multiple moments in the current detection cycle, and the average temperature data is used as the actual measured temperature of the mold.

[0028] According to one embodiment of the present invention, in step S103, the feed rate refers to the rate at which the injection molding machine injects material into the mold. The injection pressure refers to the pressure applied when the material is injected into the mold during the injection molding process, and the nozzle temperature is the temperature of the material before it is injected into the mold. The data is automatically collected by a sensor connected to the injection molding machine to improve the accuracy and real-time nature of the data.

[0029] According to one embodiment of the present invention, in step S104, the feed speed data, injection pressure data and nozzle temperature data of the injection molding machine have a greater impact on the mold temperature. For example, when the feed speed is fast, the material flows rapidly in the mold, generating higher shear heat, causing the mold temperature to rise. Higher injection pressure makes the material more closely contact with the mold surface, accelerating heat transfer, causing the mold temperature to rise. Higher nozzle temperature causes the material to overheat before entering the mold, causing the mold temperature to rise too quickly. The temperature prediction model is a deep learning neural network model, which can be processed based on the feed speed data, injection pressure data and nozzle temperature data to obtain predicted temperature data at multiple moments in the current detection cycle.

[0030] According to one embodiment of the present invention, the training step of the temperature prediction model includes: obtaining the historical average temperature data of the mold at multiple moments in multiple historical detection cycles, wherein the previous historical detection cycle of the current detection cycle is the first historical detection cycle; obtaining the historical feed speed data, historical injection pressure data and historical nozzle temperature data of the injection molding machine at multiple moments in multiple historical detection cycles; processing the historical feed speed data, the historical injection pressure data and the historical nozzle temperature data through the temperature prediction model to obtain the historical predicted temperature data at multiple moments in multiple historical detection cycles; determining the loss function of the temperature prediction model based on the historical average temperature data, the historical feed speed data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data; and training the temperature prediction model based on the training loss function to obtain the trained temperature prediction model.

[0031] According to one embodiment of the present invention, the loss function of the temperature prediction model is used to measure the difference between the model prediction result and the actual situation, and is a key indicator for training the model. During the training process, the loss function is back-propagated and the model parameters are adjusted to minimize the loss function, thereby improving the accuracy of the temperature prediction model and obtaining a trained temperature prediction model.

[0032] According to one embodiment of the present invention, the loss function of the temperature prediction model is determined according to the historical average temperature data, the historical feed rate data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data, including: determining the loss function Loss of the temperature prediction model according to formula (1),

[0033]

[0034] Among them, M s,k is the historical feed rate data at the kth moment of the sth historical detection cycle, M s,1 is the historical feed rate data at the first moment of the sth historical detection cycle, P s,k is the historical injection pressure data at the kth moment of the sth historical detection cycle, P s,1 is the historical injection pressure data at the first moment of the sth historical detection cycle, T s,k is the historical nozzle temperature data at the kth moment of the sth historical detection cycle, T s,1 is the historical nozzle temperature data at the first moment of the sth historical detection cycle, is the historical average temperature data at the kth moment of the sth historical detection cycle, The historical predicted temperature data at the kth moment of the sth historical detection cycle output by the temperature prediction model, m is the number of historical detection cycles, n is the number of moments in the detection cycle, k≤n, s≤m, and k, n, s and m are all positive integers; according to the training loss function, the temperature prediction model is trained to obtain the trained temperature prediction model.

[0035] According to an embodiment of the present invention, the error in the prediction at the first moment of any detection cycle is small, the prediction result is more accurate, and the size of the error is positively correlated with the moment in the detection cycle. In formula (1), It is the ratio of the historical feed speed data at the kth moment in the sth historical detection cycle to the historical feed speed data at the 1st moment. The closer the ratio is to 1, the smaller the impact of the change in historical feed speed data on the mold temperature. Thus, the impact of the change in historical feed speed data on the error of the predicted temperature data is smaller. It is the ratio of the historical injection pressure data at the kth moment in the sth historical detection cycle to the historical injection pressure data at the 1st moment. The closer the ratio is to 1, the smaller the impact of the change in historical injection pressure data on the mold temperature. Thus, the impact of the change in historical injection pressure data on the error of the predicted temperature data is smaller. It is the ratio of the historical nozzle temperature data at the kth moment of the sth historical detection cycle to the historical nozzle temperature data at the first moment. The closer this ratio is to 1, the smaller the impact of the change in historical nozzle temperature data on the mold temperature. Thus, the smaller the impact of the change in historical nozzle temperature data on the error of the predicted temperature data. The product of the above three items indicates that the feed rate data, injection pressure data and nozzle temperature data are positively correlated with the mold temperature. For example, a faster feed rate causes the material to flow rapidly in the mold, generating higher shear heat, thereby increasing the mold temperature. A higher injection pressure allows the material to be in closer contact with the mold surface, thereby accelerating heat transfer, causing the mold temperature to rise. A higher nozzle temperature causes the material to overheat before entering the mold, causing the mold temperature to rise too quickly. Therefore, the larger the feed rate data, injection pressure data and nozzle temperature data at the kth moment of the sth historical detection cycle relative to the data at the first moment, that is, and The larger the value of , the higher the mold temperature, and the greater the impact on the error of the predicted temperature data. It is the relative error between the historical average temperature data at the kth moment of the sth historical detection period and the historical predicted temperature data at the kth moment of the sth historical detection period output by the temperature prediction model. It represents the weight at the kth moment, which is used to reasonably weight the relative errors at different moments in the loss function. For the sth historical detection period, the accuracy of the historical predicted temperature data at the kth moment output by the temperature prediction model is usually higher than the accuracy of the historical predicted temperature data at the k+1th moment. That is, the longer the time interval between a certain moment in a historical detection period and the first moment, the less accurate the prediction result, and therefore the lower the weight. Conversely, the more accurate the prediction result, the higher the weight. It represents the weighted summation of the relative errors at each moment in the sth historical detection cycle. is the weight of the sth historical detection cycle, and the previous historical detection cycle of the current detection cycle is the first historical detection cycle. The longer the mold is used, the more it will affect the temperature measurement, such as blockage of the heat exchange tube and poor cooling water channel. That is, the longer the time interval between a historical detection cycle and the current detection cycle, the more accurate the prediction result, and therefore, the higher its weight. Conversely, the less accurate the prediction result, the lower its weight. Through the weight of the historical detection cycle, the weighted sum of the relative errors at each moment in the above sth historical detection cycle is weighted and summed to obtain the loss function of the temperature prediction model. During the training process, the loss function is back-propagated and the model parameters are adjusted to reduce the value of the loss function of the temperature prediction model, thereby improving the accuracy of the temperature prediction model and obtaining the temperature prediction model.

[0036] In this way, the influence of feed rate data, injection pressure data and nozzle temperature data on the mold temperature can be used to determine the influence of the above data on the error of predicted temperature data. Based on this influence and the relative error of historical predicted temperature data, and based on the characteristics that the longer the moment interval, the shorter the cycle interval, the lower the prediction accuracy, weights are set to train the loss function to improve the objectivity and accuracy of the loss function, thereby reducing the training loss function during the training process, improving the training intensity and efficiency, and improving the accuracy of the temperature prediction model.

[0037] According to an embodiment of the present invention, in step S105, a temperature detection error score is determined based on the predicted temperature data and the average temperature data.

[0038] According to one embodiment of the present invention, step S105 includes: fitting the predicted temperature data and the moments in the current detection cycle to determine the predicted temperature function in the current detection cycle; determining the predicted temperature derivative function based on the predicted temperature function; determining the predicted temperature change rate at multiple moments in the current detection cycle based on the predicted temperature derivative function; fitting the average temperature data and the moments in the current detection cycle to determine the average temperature function in the current detection cycle; determining the average temperature derivative function based on the average temperature function; determining the average temperature change rate at multiple moments in the current detection cycle based on the average temperature derivative function; obtaining a predicted temperature state vector based on the predicted temperature data and the predicted temperature change rate of the mold at multiple moments in the current detection cycle; obtaining an average temperature state vector based on the average temperature data and the average temperature change rate of the mold at multiple moments in the current detection cycle; determining a temperature detection error score based on the predicted temperature state vector and the average temperature state vector.

[0039] According to one embodiment of the present invention, the predicted temperature data and multiple moments in the current detection cycle are fitted to obtain a predicted temperature function for describing the law of the predicted temperature data in the current detection cycle changing over time, the predicted temperature function is derived to determine the predicted temperature derivative function, multiple moments in the current detection cycle are substituted into the predicted temperature derivative function, and the predicted temperature change rate at multiple moments in the current detection cycle is determined. The average temperature data and multiple moments in the current detection cycle are fitted to obtain an average temperature function for describing the law of the average temperature data in the current detection cycle changing over time, the average temperature function is derived to determine the average temperature derivative function, multiple moments in the current detection cycle are substituted into the average temperature derivative function, and the average temperature change rate at multiple moments in the current detection cycle is determined. The two parameters of the predicted temperature data and the predicted temperature change rate are combined into a vector to obtain a predicted temperature state vector. The two parameters of the average temperature data and the average temperature change rate are combined into a vector to obtain an average temperature state vector.

[0040] According to an embodiment of the present invention, determining a temperature detection error score according to the predicted temperature state vector and the average temperature state vector includes: determining a temperature detection error score D according to formula (2),

[0041]

[0042] in, is the average temperature data at the kth moment of the current detection cycle, is the average temperature change rate at the kth moment in the current detection cycle, is the predicted temperature data at the kth moment in the current detection cycle, is the predicted temperature change rate at the kth moment in the current detection cycle, is the average temperature state vector at the kth moment of the current detection cycle, is the predicted temperature state vector at the kth moment in the current detection cycle, n is the number of moments in the detection cycle, k≤n, and both k and n are positive integers, and min is the minimum value function.

[0043] According to one embodiment of the present invention, in formula (2), It is the cosine similarity between the average temperature state vector at the kth moment of the current detection cycle and the predicted temperature state vector at the kth moment of the current detection cycle. The closer the cosine similarity is to 0, the greater the difference between the average temperature and the average temperature change rate actually measured by the mold at the kth moment, and the predicted temperature and the predicted temperature change rate, that is, it indicates that the mold temperature is abnormal and has a large deviation from the predicted reasonable value. For example, after the mold has been used for a long time, the heat exchange efficiency of the heat exchange tube decreases due to the accumulation of impurities, causing the mold temperature to be abnormal, or the external environment temperature is too high, causing the mold temperature to be abnormal. Take the minimum value of the cosine similarity between the average temperature state vector and the predicted temperature state vector at multiple moments of the current detection cycle as the temperature detection error score, that is, the maximum difference between the actual temperature measured by the mold and the predicted temperature, as the temperature detection error score of the mold. The larger the temperature detection error score, the greater the degree of abnormality of the mold temperature when injecting materials.

[0044] In this way, the temperature detection error score can be determined by predicting the temperature state vector and the average temperature state vector, and the real-time monitoring and abnormality detection of the mold temperature during the material injection process can be achieved, so that the material can be processed under suitable temperature conditions and the material performance and molding quality can be improved.

[0045] According to an embodiment of the present invention, in step S106, whether the mold temperature detection is qualified is determined by comparing the temperature detection error score with the temperature detection error score threshold.

[0046] According to an embodiment of the present invention, determining whether the mold temperature detection is qualified according to the temperature detection error score and the temperature detection error score threshold includes: obtaining a first condition C1 and a second condition C2 according to formula (3),

[0047]

[0048] Where D is the temperature detection error score, D p is the temperature detection error score threshold; when the first condition C1 is met, the mold temperature detection is determined to be unqualified; when the second condition C2 is met, the mold temperature detection is determined to be qualified.

[0049] According to one embodiment of the present invention, in formula (3), the first condition C1 indicates that the temperature detection error score is greater than or equal to the temperature detection error score threshold, the mold temperature is detected to be abnormal, and the mold temperature detection is unqualified. The second condition C2 indicates that the temperature detection error score is less than the temperature detection error score threshold, the mold temperature is not detected to be abnormal, and the mold temperature detection is qualified.

[0050] In this way, the temperature detection error score and the temperature detection error score threshold can be used to determine whether the mold temperature detection is qualified. If the mold temperature detection is qualified, it can be judged that the temperature of the mold during the material injection process is kept within a reasonable range, thereby improving the consistency of material performance.

[0051] According to the mold temperature detection method of the embodiment of the present invention, the mold temperature when the mold is injected with material is predicted by combining the trained temperature prediction model with the feed rate data, injection pressure data and nozzle temperature data of the injection molding process, which can improve the accuracy of mold temperature prediction. The predicted temperature data output by the temperature prediction model is compared with the average temperature data actually measured by the mold, so that abnormal changes in mold temperature can be discovered in time, thereby reducing the quality problems of finished products caused by temperature and improving material properties and molding quality. When training the temperature prediction model, the influence of the feed rate data, injection pressure data and nozzle temperature data on the mold temperature can be used to determine the influence of the above data on the error of the predicted temperature data, so that based on the influence and the relative error of the historical predicted temperature data, and based on the characteristics that the longer the time interval, the shorter the cycle interval, and the lower the prediction accuracy, the weight is set to train the loss function to improve the objectivity and accuracy of the loss function, so that the training loss function is reduced during the training process, which can improve the training intensity and training efficiency, and improve the accuracy of the temperature prediction model. When determining the temperature detection error score, the temperature detection error score can be determined by predicting the temperature state vector and the average temperature state vector, which can realize real-time monitoring and abnormal detection of the mold temperature during the material injection process, so that the material can be processed under suitable temperature conditions, improving material performance and molding quality. When determining whether the mold temperature detection is qualified, the temperature detection error score and the temperature detection error score threshold can be used to determine whether the mold temperature detection is qualified. If the mold temperature detection is qualified, it can be judged that the temperature of the mold during the material injection process is kept within a reasonable range, thereby improving the consistency of material performance.

[0052] Figure 2A block diagram of a mold temperature detection system according to an embodiment of the present invention is exemplarily shown, and the system includes: an infrared image module, which is used to divide the mold into multiple areas and obtain infrared images of each area at multiple moments in the current detection cycle; an average temperature data module, which is used to determine the average temperature data of the mold at multiple moments in the current detection cycle based on the infrared image; an injection molding machine data module, which is used to obtain the feed speed data, injection pressure data and nozzle temperature data of the injection molding machine at multiple moments in the current detection cycle; a predicted temperature data module, which is used to process the feed speed data, the injection pressure data and the nozzle temperature data through a trained temperature prediction model to obtain the predicted temperature data at multiple moments in the current detection cycle; a temperature detection error scoring module, which is used to determine the temperature detection error score based on the predicted temperature data and the average temperature data; and a qualified judgment module, which is used to determine whether the mold temperature detection is qualified based on the temperature detection error score and the temperature detection error score threshold.

[0053] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. A mold temperature detection method, characterized in that: include: Divide the mold into multiple areas and obtain infrared images of each area at multiple moments in the current detection cycle; Determine average temperature data of the mold at multiple moments in a current detection cycle according to the infrared image; At multiple moments in the current detection cycle, the feed speed data, injection pressure data and nozzle temperature data of the injection molding machine are obtained; the feed speed data, the injection pressure data and the nozzle temperature data are processed by a trained temperature prediction model to obtain predicted temperature data at multiple moments in the current detection cycle; and a temperature detection error score is determined based on the predicted temperature data and the average temperature data; Whether the mold temperature detection is qualified is determined according to the temperature detection error score and the temperature detection error score threshold.

2. The mold temperature detection method according to claim 1, characterized in that: According to the infrared image, the average temperature data of the mold at multiple moments in the current detection cycle is determined, including: according to the pixel values ​​of the pixels of the infrared image, obtaining the average regional temperature data of each area at multiple moments in the current detection cycle; averaging the average regional temperature data to determine the average temperature data of the mold at multiple moments in the current detection cycle.

3. The mold temperature detection method according to claim 1, characterized in that: The training step of the temperature prediction model includes: obtaining the historical average temperature data of the mold at multiple moments in multiple historical detection cycles, wherein the previous historical detection cycle of the current detection cycle is the first historical detection cycle; obtaining the historical feed speed data, historical injection pressure data and historical nozzle temperature data of the injection molding machine at multiple moments in multiple historical detection cycles; processing the historical feed speed data, the historical injection pressure data and the historical nozzle temperature data through the temperature prediction model to obtain the historical predicted temperature data at multiple moments in multiple historical detection cycles; determining the loss function of the temperature prediction model based on the historical average temperature data, the historical feed speed data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data; and training the temperature prediction model based on the training loss function to obtain the trained temperature prediction model.

4. The mold temperature detection method according to claim 3, characterized in that: Determining the loss function of the temperature prediction model according to the historical average temperature data, the historical feed rate data, the historical injection pressure data, the historical nozzle temperature data and the historical predicted temperature data includes: according to the formula Determine the loss function Loss of the temperature prediction model, where M s,k is the historical feed rate data at the kth moment in the sth historical detection cycle, M s,1 is the historical feed rate data at the first moment of the sth historical detection cycle, P s,k is the historical injection pressure data at the kth moment of the sth historical detection cycle, P s,1 is the historical injection pressure data at the first moment of the sth historical detection cycle, T s,k is the historical nozzle temperature data at the kth moment of the sth historical detection cycle, T s,1 is the historical nozzle temperature data at the first moment of the sth historical detection cycle, is the historical average temperature data at the kth moment of the sth historical detection cycle, The historical predicted temperature data at the kth moment of the sth historical detection cycle output by the temperature prediction model, m is the number of historical detection cycles, n is the number of moments in the detection cycle, k≤n, s≤m, and k, n, s and m are all positive integers; according to the training loss function, the temperature prediction model is trained to obtain the trained temperature prediction model.

5. The mold temperature detection method according to claim 1, characterized in that: Determine a temperature detection error score based on the predicted temperature data and the average temperature data, including: fitting the predicted temperature data and the moments in the current detection cycle to determine the predicted temperature function in the current detection cycle; determining the predicted temperature derivative function based on the predicted temperature function; determining the predicted temperature change rate at multiple moments in the current detection cycle based on the predicted temperature derivative function; fitting the average temperature data and the moments in the current detection cycle to determine the average temperature function in the current detection cycle; determining the average temperature derivative function based on the average temperature function; determining the average temperature change rate at multiple moments in the current detection cycle based on the average temperature derivative function; obtaining a predicted temperature state vector based on the predicted temperature data and the predicted temperature change rate of the mold at multiple moments in the current detection cycle; obtaining an average temperature state vector based on the average temperature data and the average temperature change rate of the mold at multiple moments in the current detection cycle; and determining a temperature detection error score based on the predicted temperature state vector and the average temperature state vector.

6. The mold temperature detection method according to claim 5, characterized in that: Determining a temperature detection error score according to the predicted temperature state vector and the average temperature state vector includes: determining a temperature detection error score according to the formula Determine the temperature detection error score D, where is the average temperature data at the kth moment of the current detection cycle, is the average temperature change rate at the kth moment in the current detection cycle, is the predicted temperature data at the kth moment in the current detection cycle, is the predicted temperature change rate at the kth moment in the current detection cycle, is the average temperature state vector at the kth moment of the current detection cycle, is the predicted temperature state vector at the kth moment in the current detection cycle, n is the number of moments in the detection cycle, k≤n, and both k and n are positive integers, and min is the minimum value function.

7. The mold temperature detection method according to claim 1, characterized in that: According to the temperature detection error score and the temperature detection error score threshold, determining whether the mold temperature detection is qualified includes: according to the formula Get the first condition C1 and the second condition C2, where D is the temperature detection error score, D p is the temperature detection error score threshold; when the first condition C1 is met, the mold temperature detection is determined to be unqualified; when the second condition C2 is met, the mold temperature detection is determined to be qualified.

8. A mold temperature detection system, characterized in that: include: Infrared imaging module, used to divide the mold into multiple areas and obtain infrared images of each area at multiple moments in the current detection cycle; An average temperature data module, used to determine average temperature data of the mold at multiple moments in the current detection cycle according to the infrared image; The injection molding machine data module is used to obtain the injection molding machine's feed speed data, injection pressure data, and nozzle temperature data at multiple moments in the current detection cycle; A predicted temperature data module, used to process the feed rate data, the injection pressure data and the nozzle temperature data through a trained temperature prediction model to obtain predicted temperature data at multiple moments in the current detection cycle; a temperature detection error scoring module, configured to determine a temperature detection error score based on the predicted temperature data and the average temperature data; The qualified judgment module is used to determine whether the mold temperature detection is qualified according to the temperature detection error score and the temperature detection error score threshold.

Citation Information

Patent Citations

  • Die baking preheating device and method

    CN115673295A

  • Mold temperature detection method and system

    CN116275005A

  • Injection molding machine

    CN108698295A

  • Weight prediction method and device for injection molding object and key parameter adjustment method

    CN114043694A

  • Mold maintenance prediction model training method, mold maintenance prediction method and mold maintenance prediction system

    CN115169617A

Cited By

  • Mould abnormity management system and method based on machine vision

    CN119494645A

  • Preheating system and preheating method for nuclear fuel assembly

    CN119983549A

  • Preheating System and Preheating Method for Nuclear Fuel Assembly

    CN119983549B

  • Accurate temperature regulation and control method for flexible mold

    CN120645401A