Temperature real-time control method applied to injection molding machine

Abstract

The invention relates to the technical field of injection molding machine temperature control, and particularly discloses a real-time temperature control method applied to an injection molding machine, which uses a PID (Proportion Integration Differentiation) controller to carry out temperature control in an injection molding process. Firstly, a target temperature value is set, an automatic tuning algorithm is used for determining a preliminary PI D parameter of a real-time temperature PID controller, then an artificial intelligence technology based on deep learning is used for carrying out time sequence analysis on real-time temperature data in the injection molding process so as to reveal the effectiveness of the current PID parameter on temperature control, and the accuracy of temperature control is improved. Therefore, the self-adaptive optimization of the initial PID parameters is carried out based on the time sequence change mode of the temperature, so that the accuracy and the response speed of temperature control are improved. According to the invention, the temperature change can be quickly responded, and more efficient and more stable temperature management is realized, so that overshoot and oscillation phenomena are reduced, and more stable and accurate temperature control in the injection molding process is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of temperature control of injection molding machines, and more specifically, to a real-time temperature control method applied to injection molding machines. Background Art

[0002] In the injection molding industry, temperature control is one of the key factors to ensure product quality and production efficiency. The injection molding machine heats the plastic material to a molten state and injects it into the mold cavity to form the desired shape. In this process, too high a temperature may cause plastic degradation and mold damage, while too low a temperature will affect the filling effect of the plastic and the physical properties of the product. Therefore, precise temperature control is essential to maintain the fluidity of the material, the filling speed, and the dimensional stability and surface quality of the final product. However, due to changes in external environmental conditions, the thermal inertia of the machine itself, and fluctuations in process parameters, maintaining a constant and ideal temperature environment becomes complex and challenging.

[0003] Traditional injection molding machine temperature control systems usually use PID (proportional-integral-differential) controllers to adjust the output power of the heater to maintain the set target temperature. Although PID controllers are widely used in the industry because of their simplicity and ease of use, their temperature control accuracy is highly dependent on the parameter settings of the PID (proportional-integral-differential) controller. Inappropriate PID parameters may lead to overshoot or undershoot, causing excessive temperature fluctuations and affecting product quality. In addition, due to the complexity of ambient temperature fluctuations, plastic melting, flow and heat transfer during the injection molding process, fixed PID parameters may no longer be applicable, resulting in limited temperature control accuracy and response speed, and even overshoot, oscillation and other problems.

[0004] Therefore, an optimized real-time temperature control method for injection molding machines is needed. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a real-time temperature control method for an injection molding machine, which uses a PID controller to perform temperature control during the injection molding process. First, a target temperature value is set, and the preliminary PID parameters of the real-time temperature PID controller are determined using an automatic tuning algorithm. Then, the real-time temperature data in the injection molding process is analyzed in a time series manner using artificial intelligence technology based on deep learning to reveal the effectiveness of the current PID parameters for temperature control, thereby adaptively optimizing the preliminary PID parameters based on the time series change pattern of the temperature, so as to improve the accuracy and response speed of the temperature control. In this way, it is possible to respond quickly to temperature changes and achieve more efficient and stable temperature management, thereby reducing overshoot and oscillation phenomena and ensuring that the temperature control of the injection molding process is more stable and precise.

[0006] According to one aspect of the present application, a real-time temperature control method for an injection molding machine is provided, comprising:

[0007] Set the target temperature value;

[0008] Determine preliminary PID parameters of real-time temperature PID controller using auto-tuning algorithm;

[0009] A time queue for receiving temperature values ​​collected by a temperature sensor;

[0010] Based on the time queue of the temperature values, performing parameter optimization on preliminary PID parameters of the real-time temperature PID controller to obtain optimized PID parameters;

[0011] extracting a current temperature value from the time queue of temperature values;

[0012] Calculating the difference between the current temperature value and the target temperature value, and inputting the difference into the real-time temperature PID controller to obtain a temperature control instruction;

[0013] The method of optimizing the preliminary PID parameters of the real-time temperature PID controller based on the time queue of the temperature value to obtain the optimized PID parameters includes:

[0014] Performing time series feature extraction based on modal decomposition on the time queue of the temperature values ​​to obtain a set of time series encoding feature vectors of temperature intrinsic modal signal components;

[0015] Performing graph walk context-aware aggregation on the set of temporal encoding feature vectors of the temperature intrinsic modal signal components to obtain a significant aggregation feature vector of the temperature intrinsic modal set;

[0016] Based on the significant aggregation feature vector of the temperature intrinsic mode set, the preliminary PID parameters are optimized to obtain the optimized PID parameters.

[0017] Preferably, performing time series feature extraction based on modal decomposition on the time queue of the temperature values ​​to obtain a set of time series encoding feature vectors of temperature intrinsic modal signal components includes:

[0018] performing variational modal decomposition on the time sequence of the temperature values ​​to obtain a set of temperature intrinsic modal signal components;

[0019] Each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components is time-series encoded respectively to obtain a set of time-series encoded feature vectors of the temperature intrinsic modal signal components.

[0020] Preferably, performing time-series coding on each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components to obtain a set of time-series coding feature vectors of the temperature intrinsic modal signal components comprises:

[0021] Each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components is respectively input into a temperature modal component time series feature extractor based on an LSTM model to obtain a set of time series encoding feature vectors of the temperature intrinsic modal signal components.

[0022] Preferably, performing graph walk context-aware aggregation on the set of temporal encoding feature vectors of the temperature intrinsic modal signal components to obtain a significant aggregation feature vector of the temperature intrinsic modal set includes:

[0023] Extracting the topological structure features of the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a temperature intrinsic modal signal component graph walk topological feature matrix;

[0024] Based on the temperature intrinsic modal signal component graph walk topology feature matrix, context-aware enhancement is performed on the set of time-series encoding feature vectors of the temperature intrinsic modal signal component to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component;

[0025] Based on the feature correlation between the set of context-aware enhanced feature vectors of the temperature eigenmodal signal components and the set of time-series encoding feature vectors of the temperature eigenmodal signal components, significance aggregation is performed on the set of time-series encoding feature vectors of the temperature eigenmodal signal components to obtain a significant aggregation feature vector of the temperature eigenmodal set.

[0026] Preferably, extracting the topological structure features of the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a temperature intrinsic modal signal component graph walk topological feature matrix includes:

[0027] Inputting each of the time-series encoding feature vectors of the temperature intrinsic modal signal component in the set of the time-series encoding feature vectors of the temperature intrinsic modal signal component into a hyperbolic space mapper to obtain a set of time-series encoding feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping;

[0028] Calculating the Poincare distance between any two time-series coded eigenvectors of the temperature intrinsic modal signal components after hyperbolic space mapping in the set of time-series coded eigenvectors of the temperature intrinsic modal signal components after hyperbolic space mapping to obtain a temperature intrinsic modal signal component graph walk topology matrix;

[0029] The temperature intrinsic modal signal component graph walk topology matrix is ​​subjected to dilated convolution coding to obtain the temperature intrinsic modal signal component graph walk topology feature matrix.

[0030] Preferably, based on the temperature intrinsic modal signal component graph walk topology feature matrix, context-aware enhancement is performed on the set of time-series encoding feature vectors of the temperature intrinsic modal signal component to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component, including:

[0031] The graph walk topological feature matrix of the temperature intrinsic modal signal component and the set of temporal coding feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping are input into a global context walk encoder based on a graph convolutional neural network model to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component.

[0032] Preferably, based on the feature correlation between the set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component and the set of time-series coded feature vectors of the temperature intrinsic modal signal component, the set of time-series coded feature vectors of the temperature intrinsic modal signal component is significantly aggregated to obtain the temperature intrinsic modal set significantly aggregated feature vector, comprising:

[0033] Inputting each corresponding group of the temperature intrinsic modal signal component context-aware enhanced feature vectors and the temperature intrinsic modal signal component time-series encoded feature vectors in the set of the temperature intrinsic modal signal component context-aware enhanced feature vectors and the set of the temperature intrinsic modal signal component time-series encoded feature vectors into an autocorrelation gating unit to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors;

[0034] Inputting the set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors into a normalization unit based on a Softmax function to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence weight factors;

[0035] Based on the set of temperature intrinsic modal signal component autocorrelation gated significant confidence weight factors, a position-weighted sum of a set of time-series encoding feature vectors of the temperature intrinsic modal signal component is calculated to obtain the temperature intrinsic modal set significant aggregated feature vector.

[0036] Preferably, optimizing the preliminary PID parameters based on the significant aggregation eigenvector of the temperature intrinsic mode set to obtain the optimized PID parameters comprises:

[0037] Inputting the significant aggregation feature vector of the temperature intrinsic mode set into a decoder-based PID parameter dynamic optimizer to obtain a parameter optimization factor;

[0038] The parameter optimization factor is multiplied by the preliminary PID parameters to obtain the optimized PID parameters.

[0039] This application has at least the following technical effects:

[0040] Compared with the prior art, the real-time temperature control method for an injection molding machine provided by the present application uses a PID controller to perform temperature control during the injection molding process. First, the target temperature value is set, and the preliminary PID parameters of the real-time temperature PID controller are determined using an automatic tuning algorithm. Then, the real-time temperature data during the injection molding process is analyzed in time series using artificial intelligence technology based on deep learning to reveal the effectiveness of the current PID parameters for temperature control, thereby performing adaptive optimization of the preliminary PID parameters based on the time series change pattern of the temperature, so as to improve the accuracy and response speed of the temperature control. The present application can respond quickly to temperature changes, achieve more efficient and stable temperature management, thereby reducing overshoot and oscillation phenomena, and ensuring that the temperature control of the injection molding process is more stable and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0042] Figure 1 The figure is a flow chart of a real-time temperature control method applied to an injection molding machine according to an embodiment of the present application.

[0043] Figure 2 Schematic diagram of data flow of a real-time temperature control method applied to an injection molding machine according to an embodiment of the present application.

[0044] Figure 3 This is a flowchart of sub-step S4 of the real-time temperature control method applied to an injection molding machine according to an embodiment of the present application.

[0045] Figure 4 This is a flowchart of sub-step S41 of the real-time temperature control method applied to an injection molding machine according to an embodiment of the present application.

[0046] Figure 5 This is a flowchart of sub-step S42 of the real-time temperature control method applied to an injection molding machine according to an embodiment of the present application.

[0047] Figure 6 This is a flowchart of sub-step S421 of the real-time temperature control method applied to an injection molding machine according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0049] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0050] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0051] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0052] It is worth noting that in this application, all actions to obtain data are carried out in compliance with the relevant data protection laws and policies of the country where the data is located, and with the authorization given by the owner of the corresponding device.

[0053] In view of the technical problems described in the above background technology, the present application proposes an optimized real-time temperature control method for injection molding machines, which uses a PID controller to perform temperature control during the injection molding process. First, the target temperature value is set, and the preliminary PID parameters of the real-time temperature PID controller are determined using an automatic tuning algorithm. Then, the real-time temperature data in the injection molding process is analyzed in time series using artificial intelligence technology based on deep learning to reveal the effectiveness of the current PID parameters for temperature control, thereby adaptively optimizing the preliminary PID parameters based on the time series change pattern of the temperature, so as to improve the accuracy and response speed of the temperature control. In this way, it is possible to respond quickly to temperature changes and achieve more efficient and stable temperature management, thereby reducing overshoot and oscillation phenomena and ensuring that the temperature control of the injection molding process is more stable and accurate.

[0054] Figure 1 The figure is a flow chart of a real-time temperature control method applied to an injection molding machine according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the real-time temperature control method for an injection molding machine according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the real-time temperature control method applied to the injection molding machine includes the following steps: S1, setting a target temperature value; S2, determining preliminary PID parameters of a real-time temperature PID controller by using an automatic tuning algorithm; S3, receiving a time queue of temperature values ​​collected by a temperature sensor; S4, based on the time queue of the temperature values, performing parameter optimization on the preliminary PID parameters of the real-time temperature PID controller to obtain optimized PID parameters; S5, extracting a current temperature value from the time queue of the temperature values; S6, calculating a difference between the current temperature value and the target temperature value, and inputting the difference into the real-time temperature PID controller to obtain a temperature control instruction.

[0055] In the above-mentioned real-time temperature control method applied to the injection molding machine, the step S1 sets the target temperature value. It should be understood that different plastic materials have their specific melting temperature ranges, within which the material can achieve optimal fluidity, thereby ensuring a good filling effect. By setting the target temperature value, a benchmark can be established for subsequent temperature control to ensure that the material maintains its ideal physical and chemical properties during processing. The process of setting the target temperature value requires comprehensive consideration of multiple factors such as the thermal properties of the plastic material, the effectiveness of the mold design and cooling system, the injection molding process parameters, the production environment conditions, the product quality requirements, and the economic benefits.

[0056] In the process of setting the target temperature value, first of all, it is crucial to understand the thermal properties of the plastic material used. Each plastic has its own specific melting point, glass transition temperature (Tg), heat deformation temperature and other characteristics, which determine the behavior of the material when heated and its ability to withstand heat without degradation or adverse changes. Therefore, the target temperature set must ensure the fluidity of the material without causing damage to the material.

[0057] Secondly, the design of the mold and the cooling system also have an important influence on the temperature setting. The internal flow channel layout of the mold, the gate location, and the location and size of the cooling channel will determine the speed and uniformity of heat transfer. If the cooling system is not efficient enough, the target temperature may need to be lowered to prevent overheating of the product; while an efficient cooling system allows a higher melt temperature to ensure a good filling effect. At the same time, the parameters of the injection molding process itself, such as injection speed, holding time, and back pressure, will also change the flow behavior of the plastic in the mold. For example, a faster injection speed usually requires a higher melt temperature to ensure the filling effect, while a longer holding time and high back pressure may cause the target temperature to be slightly lowered to avoid defects caused by excessive pressure.

[0058] The ambient temperature and humidity of the production workshop also affect the selection of the target temperature. A high temperature environment may cause the plastic to soften prematurely, affecting the processing accuracy; a humid environment may cause the plastic to absorb moisture, resulting in bubbles or cracks during the heating process. Therefore, when setting the target temperature, the influence of external environmental conditions should be considered and appropriate adjustments should be made.

[0059] In addition, different applications have different requirements for the quality of the finished product. For precision parts or optical components, surface finish and flatness are critical, which may mean tighter temperature control to reduce any factors that may cause surface defects. For structural parts, strength and toughness may be more important considerations, and a slightly looser temperature range may be allowed.

[0060] Finally, cost is also an important consideration. Although higher operating temperatures can help improve material fluidity, they also increase energy consumption and the risk of equipment wear, and may also lead to increased material waste. Therefore, it is also necessary to balance the relationship between performance and cost when setting the target temperature.

[0061] In the above-mentioned real-time temperature control method for injection molding machine, the step S2 uses an automatic tuning algorithm to determine the preliminary PID parameters of the real-time temperature PID controller. It should be understood that considering that manual adjustment of PID parameters is a time-consuming and professional knowledge-required process, the present application first uses an automatic tuning algorithm to determine the preliminary PID parameters of the real-time temperature PID controller, thereby simplifying the manual operation process and improving work efficiency. In a specific example of the present application, the Ziegler-Nichols law is used for automatic tuning. The Ziegler-Nichols method is a manual / semi-automatic tuning method based on the critical proportionality method, which gradually increases the proportional gain Kp of the controller until the system output begins to oscillate continuously (critical oscillation). The proportional gain Ku and the oscillation period Tu at this time are recorded, and then the optimal parameters of the PID controller are calculated according to a specific empirical formula. In this way, the preliminary PID parameters can be quickly determined, providing a good starting point for subsequent temperature control.

[0062] After selecting the appropriate automatic tuning method, first of all, the selected algorithm needs to be integrated into the existing control system architecture, which may involve software programming and hardware interface design. Then, the automatic tuning process is started in a safe and controllable environment. During this period, the algorithm continuously monitors the input and output data of the system and iteratively updates the PID parameters according to the preset objective function (such as minimizing the sum of squared errors). In order to ensure the effectiveness of the tuning results, a series of verification experiments are usually required. These experiments can help confirm whether the new PID parameters have indeed improved the performance of the system, such as faster response speed, smaller steady-state error, and better anti-disturbance ability. If it is found that some aspects fail to meet expectations, the automatic tuning strategy can be further adjusted or other more suitable methods can be reselected.

[0063] After completing the automatic tuning, a set of preliminary PID parameters is obtained. However, due to many unforeseen factors in the injection molding process, such as differences between raw material batches, mold wear, ambient temperature fluctuations, etc., these may affect the final control effect. Therefore, even with preliminary parameters, it is still necessary to regularly check and evaluate the performance of the system and make necessary adjustments based on actual conditions.

[0064] In the above-mentioned real-time temperature control method applied to the injection molding machine, the step S3 receives a time queue of temperature values ​​collected by the temperature sensor. Specifically, the present application continuously monitors the temperature changes inside the injection molding machine to analyze the adaptability of the current PID parameters to temperature control, so as to adjust the parameters of the PID controller in time to adapt to the temperature fluctuations during the injection molding process.

[0065] Specifically, during the injection molding process, changes in temperature directly affect product quality. Therefore, the system needs to have the ability to monitor and record temperature changes in real time. As a key component for sensing the temperature of the environment or material, the temperature sensor's output data must be connected to the data acquisition unit (DAQ) through a physical interface (such as an analog signal line or a digital communication bus). DAQ is responsible for converting the raw electrical signal from the sensor into digital information that the computer can understand. Depending on the specific application requirements, different transmission methods can be selected. For example, analog signals are suitable for simple point-to-point connections, but are susceptible to electromagnetic interference; while digital communication protocols such as Modbus, CAN bus or industrial Ethernet provide more stable data transmission and support multi-point network architecture.

[0066] After collecting the data from the temperature sensor, the next step is to organize the data into an ordered time series. This step is usually completed in an embedded controller or PC application. Each new measurement result is accompanied by a timestamp, indicating the exact moment when the sample was obtained. These time-tagged data are then stored in a ring buffer or other form of temporary storage structure to form a so-called "time queue". The ring buffer is a data structure that is particularly suitable for real-time applications, allowing old data to be efficiently overwritten without having to move the positions of other elements. When new data arrives, it replaces the earliest data that enters the buffer, so that the latest N samples are always available. This design helps to reduce memory usage while ensuring that all operations are completed in a fixed time, which is very important for maintaining the real-time response characteristics of the system.

[0067] The data taken from the buffer is then passed to the upper-level software for further processing. At this point, a series of operations such as filtering, smoothing, and anomaly detection may be involved to remove noise or identify potential problems. In addition, statistics such as average, maximum / minimum, and standard deviation can be calculated to better understand and evaluate the current temperature status.

[0068] Accordingly, choosing the right temperature sensor is the basis for building an effective temperature monitoring system. Different types of temperature sensors have their own advantages and disadvantages. When selecting a temperature sensor, multiple factors such as accuracy, response speed, working range, cost and installation convenience should be considered comprehensively. Thermocouples are one of the most commonly used industrial-grade temperature sensors. They are based on the Seebeck effect, that is, the voltage generated at the contact point of two different metals changes with temperature. The advantages of thermocouples are low cost, high temperature resistance and fast response speed, but the output signal is weak, usually requiring additional amplification circuits, and drift may occur after long-term use, affecting the measurement accuracy. On the other hand, resistance temperature detectors (RTDs) are based on the principle that the resistance of metals changes with temperature, providing higher accuracy and stability, especially for accurate measurements in the low to medium temperature range. However, RTDs are expensive and have relatively slow response speeds.

[0069] In addition to the two common types of temperature sensors mentioned above, there are also thermistors and infrared temperature sensors to choose from. Thermistors are characterized by high sensitivity and fast response, but their performance degrades in high temperature environments. Infrared temperature sensors can measure the surface temperature of objects non-contactly, and are very suitable for measuring the temperature of objects that are difficult to access or in motion. However, infrared temperature sensors are usually more expensive and are greatly affected by environmental factors such as dust and steam. Therefore, when selecting a temperature sensor, it is necessary to make the best choice based on the specific process requirements and working environment. For example, in injection molding, considering the different positions inside and outside the mold and the temperature range requirements, it may be necessary to use a combination of multiple types of sensors to ensure that the temperature throughout the production process can be accurately monitored and controlled.

[0070] In the above-mentioned real-time temperature control method applied to an injection molding machine, the step S4 optimizes the preliminary PID parameters of the real-time temperature PID controller based on the time queue of the temperature value to obtain the optimized PID parameters. Figure 3 FIG. 4 is a flow chart of sub-step S4 of the real-time temperature control method for an injection molding machine according to an embodiment of the present application. Figure 3 As shown, the step S4 includes the steps of: S41, performing time series feature extraction based on modal decomposition on the time queue of the temperature values ​​to obtain a set of time series encoding feature vectors of temperature intrinsic modal signal components; S42, performing graph walking context-aware aggregation on the set of time series encoding feature vectors of the temperature intrinsic modal signal components to obtain a significantly aggregated feature vector of the temperature intrinsic modal set; S43, optimizing the preliminary PID parameters based on the significantly aggregated feature vector of the temperature intrinsic modal set to obtain the optimized PID parameters.

[0071] Figure 4 FIG. 4 is a flowchart of sub-step S41 of the real-time temperature control method for an injection molding machine according to an embodiment of the present application. Figure 4 As shown, the step S41 includes the steps of: S411, performing variational modal decomposition on the time queue of the temperature values ​​to obtain a set of temperature intrinsic modal signal components; S412, performing time series encoding on each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components to obtain a set of time series encoding feature vectors of the temperature intrinsic modal signal components.

[0072] More specifically, in step S411, variational mode decomposition is performed on the time queue of the temperature value to obtain a set of temperature intrinsic mode signal components. It should be understood that the present application takes into account that the temperature change in the injection molding process may contain characteristics on multiple time scales, such as fast-response heating / cooling cycles, slower thermal inertia changes, and the influence of ambient temperature fluctuations. Therefore, in order to capture the temporal change pattern of temperature data more comprehensively and meticulously, the present application uses variational mode decomposition (VMD) technology to process the time queue of the temperature value to decompose the original temperature signal into a series of intrinsic mode functions (IMFs) to form a set of temperature intrinsic mode signal components. Among them, each temperature intrinsic mode signal component corresponds to a different time scale and frequency range, representing the characteristics of the temperature change of the injection molding machine on a specific time scale. In this way, by analyzing each temperature intrinsic mode signal component separately, different aspects of temperature change can be better understood, providing more detailed and accurate data support for subsequent temperature control.

[0073] More specifically, in a specific example of the present application, the step S412 includes: inputting each temperature intrinsic modal signal component in the set of the temperature intrinsic modal signal components into a temperature modal component time series feature extractor based on the LSTM model to obtain a set of time series encoding feature vectors of the temperature intrinsic modal signal components. It should be known to those skilled in the art that the LSTM model is a special recursive neural network (RNN) that is specifically designed to process and predict long-term dependency problems in time series data. In the present application, the LSTM model can effectively capture the temporal dynamic characteristics of the temperature intrinsic modal signal components through its unique structure, including a forget gate, an input gate, and an output gate. The working principle of the LSTM model is that it can selectively remember or forget past information, thereby avoiding the common gradient vanishing problem in traditional RNNs, so that the model can learn dependencies over a long time span. When the temperature intrinsic modal signal component is input into the LSTM model, the model automatically adjusts the internal state according to previous learning experience and generates the most appropriate response to the current input. This mechanism ensures that even in the presence of complex nonlinearities and long delays, the LSTM model can accurately extract the key features of temperature changes, and then generate a corresponding time-series encoding feature vector for each temperature intrinsic modal signal component, forming a set of time-series encoding feature vectors of temperature intrinsic modal signal components for subsequent analysis and prediction tasks.

[0074] Specifically, the step S42 performs graph walking context-aware aggregation on the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a significantly aggregated feature vector of the temperature intrinsic modal set. It should be understood that the set of time-series encoding feature vectors of the temperature intrinsic modal signal components respectively represents the time-series change characteristics of temperature on different time scales. In order to fully understand the temporal dynamic change characteristics of temperature, it is necessary to further aggregate the set of time-series encoding feature vectors of the temperature intrinsic modal signal components. In particular, considering that different modal components may have different importance, the present application proposes a graph walking context-aware aggregation method, which uses a graph walking algorithm to mine the correlation and context information between each temperature intrinsic modal signal component, so as to evaluate the relative importance of each modal component, and perform weighted aggregation accordingly, thereby effectively highlighting the contribution of key modal components, while suppressing the influence of noise and unimportant modal components, so as to improve the accuracy of feature expression. Among them, Figure 5 FIG. 4 is a flowchart of sub-step S42 of the real-time temperature control method for an injection molding machine according to an embodiment of the present application. Figure 5 As shown, the step S42 includes the steps of: S421, extracting the topological structure characteristics of the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a graph walk topology feature matrix of the temperature intrinsic modal signal components; S422, based on the graph walk topology feature matrix of the temperature intrinsic modal signal components, performing context-aware enhancement on the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal components; S423, based on the feature correlation between the set of context-aware enhanced feature vectors of the temperature intrinsic modal signal components and the set of time-series encoding feature vectors of the temperature intrinsic modal signal components, performing significance aggregation on the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a significant aggregation feature vector of the temperature intrinsic modal set.

[0075] Figure 6 FIG. 4 is a flowchart of sub-step S421 of the real-time temperature control method for an injection molding machine according to an embodiment of the present application. Figure 6As shown, the step S421 includes the steps of: S4211, inputting each temperature intrinsic modal signal component time-coding feature vector in the set of temperature intrinsic modal signal component time-coding feature vectors into the hyperbolic space mapper respectively to obtain a set of temperature intrinsic modal signal component time-coding feature vectors after hyperbolic space mapping; S4212, calculating the Poincare distance between any two temperature intrinsic modal signal component time-coding feature vectors after hyperbolic space mapping in the set of temperature intrinsic modal signal component time-coding feature vectors after hyperbolic space mapping to obtain a temperature intrinsic modal signal component graph walk topology matrix; S4213, performing hole convolution coding on the temperature intrinsic modal signal component graph walk topology matrix to obtain the temperature intrinsic modal signal component graph walk topology feature matrix.

[0076] More specifically, the step S4211 is expressed as follows:

[0077] V={v 1 ,v 2 ,...,v i , ..., v n}

[0078] h i =W 1 v i W 2

[0079] Wherein, V represents the set of time-series encoding feature vectors of the temperature intrinsic modal signal component, v 1 、v 2 、v i and v n represents the first, second, i-th and n-th time-series coded feature vectors of the temperature intrinsic modal signal component in the set of time-series coded feature vectors of the temperature intrinsic modal signal component respectively, n is the number of feature vectors in the set of time-series coded feature vectors of the temperature intrinsic modal signal component, W 1 and W 2 Represent the first linear mapping matrix and the second linear mapping matrix respectively, h i Represents the i-th time-series encoding feature vector of the temperature intrinsic modal signal component after hyperbolic space mapping in the set of time-series encoding feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping.

[0080] Specifically, in order to better represent the correlation between each temperature intrinsic modal signal component, the present application first maps the set of time-series coded feature vectors of the temperature intrinsic modal signal component to a hyperbolic space. Compared with Euclidean space, the exponentially growing volume of the hyperbolic space allows a more natural embedding of tree or tree-like structures, making it more suitable for representing hierarchical data sets. Therefore, through hyperbolic space mapping, the hyperbolic geometric characteristics can be fully utilized to maintain the hierarchical structure and long-tail distribution in high-dimensional data, so as to enhance the relative distance representation capability between the features of the temperature intrinsic modal signal components.

[0081] More specifically, the step S4212 is expressed by the formula:

[0082]

[0083] Among them, h j represents the jth time-series coded feature vector of the temperature intrinsic modal signal component after hyperbolic space mapping in the set of time-series coded feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping, ||·|| represents the norm of the calculation vector, arccosh(·) represents the inverse cosine function, d P (·,·) represents the Poincare distance metric function, D ij Represents the element value at the (i, j)th position in the walk topology matrix of the temperature intrinsic modal signal component graph.

[0084] That is, by calculating the Poincare distance between the time-series encoding eigenvectors of each temperature intrinsic modal signal component after hyperbolic space mapping, the relative proximity between each temperature intrinsic modal signal component is represented, and a temperature intrinsic modal signal component graph walk topology matrix is ​​constructed to reveal the global topological structure of the set of time-series encoding eigenvectors of the temperature intrinsic modal signal component.

[0085] More specifically, the step S4213 is expressed by the formula:

[0086] D t =DilatedConv(D)

[0087] Where D represents the temperature intrinsic modal signal component graph walk topology matrix, DilatedConv(·) represents the dilated convolution operation, and D t Represents the topological characteristic matrix of the temperature intrinsic mode signal component graph.

[0088] That is, by applying the dilated convolution coding technology to the walk topology matrix of the temperature intrinsic modal signal component graph, the sparse connection characteristics of the dilated convolution coding technology can be utilized to further capture the dependencies between the temperature intrinsic modal signal components within a long distance, thereby enhancing the feature expression capability.

[0089] More specifically, in a specific example of the present application, the step S422 includes: inputting the set of the graph walk topology feature matrix of the temperature intrinsic modal signal component and the time series encoding feature vector of the temperature intrinsic modal signal component after the hyperbolic space mapping into the global context walk encoder based on the graph convolutional neural network model to obtain the set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component, which is expressed by the formula:

[0090]

[0091] Among them, GCN(·,·) represents graph convolutional neural network, s i Indicates the v i The corresponding context-aware enhanced feature vector of the temperature intrinsic modal signal component.

[0092] That is, the time-series encoding feature vectors of each temperature intrinsic modal signal component are used as nodes in the graph structure, and the graph walk topological feature matrix of the temperature intrinsic modal signal component is used as the edge of the graph walk. The walk is performed on the entire graph structure through the graph convolution operation, and information transfer and accumulation are performed between each temperature intrinsic modal signal component. This allows each node to access the information of its multi-hop neighbors to utilize contextual information to enhance its feature representation, thereby extracting a set of context-aware enhanced feature vectors of temperature intrinsic modal signal components with rich contextual information.

[0093] More specifically, in a specific example of the present application, the step S423 includes: inputting each corresponding group of the temperature intrinsic modal signal component context-aware enhanced feature vector and the temperature intrinsic modal signal component time-series encoded feature vector in the set of the temperature intrinsic modal signal component context-aware enhanced feature vector and the set of the temperature intrinsic modal signal component time-series encoded feature vector into an autocorrelation gating unit to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors, which is expressed by the formula:

[0094]

[0095] Among them, c i Indicates the v i The corresponding temperature intrinsic mode signal component autocorrelation gated significant confidence factor, g(·,·) represents the autocorrelation gated unit, represents positional subtraction, and softmax[·] represents a classifier based on the Softmax function.

[0096] That is, the temperature intrinsic modal signal component features after context-aware enhancement and the original temperature intrinsic modal signal component features are sent to the autocorrelation gating unit to measure the consistency and correlation between the two. It should be understood that if the original temperature intrinsic modal signal component features and the feature representation after context-aware enhancement have a high degree of consistency and correlation, it can be considered that the temperature intrinsic modal signal component plays a leading role in the characterization of the global temperature time series features. In this way, the relative importance of each temperature intrinsic modal signal component can be effectively revealed, and a set of temperature intrinsic modal signal component autocorrelation gating significant confidence factors can be generated to guide the subsequent feature aggregation process.

[0097] In a specific example of the present application, the step S423 further includes: inputting the set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors into a normalization unit based on a Softmax function to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence weight factors, which is expressed as:

[0098]

[0099] Among them, exp(·) represents the exponential function with the natural constant as the base, w i Indicates the v i The corresponding temperature intrinsic mode signal component autocorrelation gated significant confidence weight factor.

[0100] In a specific example of the present application, the step S423 further includes: based on the set of autocorrelation gated significant confidence weight factors of the temperature intrinsic modal signal component, calculating the position-weighted sum of the set of time-series coded feature vectors of the temperature intrinsic modal signal component to obtain the significant aggregation feature vector of the temperature intrinsic modal set, which is expressed by the formula:

[0101]

[0102] Among them, ⊙ represents the point product by position, and F represents the significant aggregate eigenvector of the temperature intrinsic mode set.

[0103] That is, a normalization unit based on a Softmax function is used to process the set of autocorrelation gated significant confidence factors of the temperature intrinsic modal signal component to obtain normalized weight factors, ensuring that the sum of the weight factors is 1, and each weight factor is applied to the set of temporal encoding feature vectors of the temperature intrinsic modal signal component, and through a weighted aggregation operation, the key features are emphasized and the noise information is suppressed to obtain a significant aggregation feature vector of the temperature intrinsic modal set. In this way, not only the temperature change information on each time scale is comprehensively considered, but also the feature quality is further improved by learning the interaction and influence between different modal components.

[0104] Specifically, in a specific example of the present application, the step S43 includes: inputting the significant aggregation feature vector of the temperature intrinsic mode set into the decoder-based PID parameter dynamic optimizer to obtain a parameter optimization factor; multiplying the parameter optimization factor by the preliminary PID parameter to obtain the optimized PID parameter. Specifically, the decoder adopts a multi-layer perceptron structure, and learns the complex patterns and inherent laws of temperature changes, such as periodic characteristics, long-term trends, abnormal points, etc., by performing multi-layer feature extraction on the significant aggregation feature vector of the temperature intrinsic mode set, and generates corresponding parameter optimization factors based on this, which are used to reflect the need for adjustment of PID controller parameters under the current temperature change conditions. Then, by multiplying the parameter optimization factor with the preliminary PID parameter, fine-tuning is performed on the basis of maintaining the original control structure, so that the PID controller can respond more accurately to temperature fluctuations during the injection molding process while maintaining the stability of the PID controller.

[0105] Considering that each temperature intrinsic modal signal component time series encoding feature vector in the set of temperature intrinsic modal signal component time series encoding feature vectors respectively represents the frequency domain correlation implicit coding features of the temperature time series distribution in different frequency bands, when performing feature sequence aggregation analysis, the heterogeneity of temperature frequency domain correlation features in different frequency bands will lead to the lack of decision-making of the aggregation feature instances of the significant aggregation feature vectors of the temperature intrinsic modal set, thereby affecting the accuracy of the parameter optimization factor obtained by the decoder-based PID parameter dynamic optimizer input thereto.

[0106] Based on this, before the temperature intrinsic mode set significant aggregation feature vector is input into the decoder-based PID parameter dynamic optimizer, the temperature intrinsic mode set significant aggregation feature vector is first optimized, including the steps of:

[0107] Based on the L1 norm, L2 norm and its own characteristic scale of the significant aggregation feature vector of the temperature intrinsic mode set, a first temperature intrinsic mode set significant aggregation normalization factor and a second temperature intrinsic mode set significant aggregation normalization factor are constructed, which are expressed as:

[0108] δ=||V|| 1 / L

[0109]

[0110] Among them, ||V|| 1 and ||V|| 2 denote the L1 norm and L2 norm of the significant aggregation feature vector of the temperature intrinsic mode set, respectively, denote the own characteristic scale of the significant aggregation feature vector of the temperature intrinsic mode set, δ denotes the significant aggregation normalization factor of the first temperature intrinsic mode set, and ε denotes the significant aggregation normalization factor of the second temperature intrinsic mode set;

[0111] Taking the first temperature eigenmode set significant aggregation normalization factor as a scaling factor and the second temperature eigenmode set significant aggregation normalization factor as a bias factor, the temperature eigenmode set significant aggregation feature vector is feature transformed to obtain the first temperature eigenmode set significant aggregation deviation measurement vector, which is expressed as:

[0112]

[0113] Among them, ⊙ represents the point multiplication by position, sigmoid represents the activation function, Indicates subtraction by position, V 1 represents the significant aggregate deviation measure vector of the first temperature intrinsic mode set;

[0114] Taking the first temperature intrinsic mode set significant aggregation normalization factor as the bias factor and the second temperature intrinsic mode set significant aggregation normalization factor as the scaling factor, the temperature intrinsic mode set significant aggregation feature vector is feature transformed to obtain the second temperature intrinsic mode set significant aggregation deviation measurement vector, which is expressed as:

[0115]

[0116] Among them, V 2 represents the significant aggregate deviation measure vector of the second temperature intrinsic mode set;

[0117] After the second temperature eigenmode set significant aggregation deviation measurement vector is measured in an informationized manner, the first temperature eigenmode set significant aggregation deviation measurement vector is used to perform compensation correction explicitness to obtain the first temperature eigenmode set significant aggregation compensation correction vector, which is expressed as:

[0118] V 3 =|log 2 (V 2 )|☉V 1 ☉δ

[0119] Among them, V 3 represents the significant aggregation compensation correction vector of the first temperature intrinsic mode set;

[0120] After the first temperature intrinsic mode set significant aggregation deviation measurement vector is measured in an informationized manner, the second temperature intrinsic mode set significant aggregation deviation measurement vector is used to perform compensation correction explicitness to obtain the second temperature intrinsic mode set significant aggregation compensation correction vector, which is expressed as:

[0121] V 4 =|log 2 (V 1 )|☉V 2 ☉ε

[0122] Among them, V 4 represents the significant aggregation compensation correction vector of the second temperature intrinsic mode set;

[0123] Based on the first temperature eigenmode set significant aggregation compensation correction vector and the second temperature eigenmode set significant aggregation compensation correction vector, the temperature eigenmode set significant aggregation feature vector is compensated and corrected to obtain an optimized temperature eigenmode set significant aggregation feature vector, which is expressed as:

[0124]

[0125] Among them, α and β represent weighted hyperparameters, represents addition by position, and V′ represents the significantly aggregated eigenvector of the optimized temperature intrinsic mode set.

[0126] Accordingly, in this preferred embodiment, a composite sparse restriction control mechanism is constructed to act on the significant aggregation feature vector of the temperature intrinsic mode set to form an interactive fairness evaluation mechanism, and the interactive fairness evaluation mechanism is used to correct the deviation of the cluster attribute adaptive association evolution mechanism. On this basis, the integrated fusion feature information feedback module is used as a balancing factor for multi-level fairness regulation to construct a robust fairness representation architecture, so as to support the significant aggregation feature vector of the temperature intrinsic mode set to establish a joint decision-making effect under the feature distribution framework through this architecture, thereby improving the significance of its feature value as an example for the decoding regression decision, and improving the accuracy of the parameter optimization factor obtained by the significant aggregation feature vector of the temperature intrinsic mode set input into the decoder-based PID parameter dynamic optimizer.

[0127] In the above-mentioned real-time temperature control method applied to the injection molding machine, the step S5 extracts the current temperature value from the time queue of the temperature value. That is, by extracting the current temperature value from the time queue of the temperature value, it is ensured that the PID controller makes an immediate response based on the latest temperature information.

[0128] During the injection molding process, temperature sensors continuously collect temperature data inside and outside the mold and transmit this data to the data acquisition unit (DAQ) through a physical interface (such as an analog signal line or a digital communication bus). In order to ensure the timeliness of the data, a high-frequency sampling strategy must be adopted so that the data points in the time queue are dense and close to the actual situation. In addition, choosing a suitable transmission protocol (such as Modbus, CAN bus or industrial Ethernet) can provide more stable data transmission and support multi-point network architecture, reducing the risk of data loss or delay. Introducing a synchronization mechanism is also an important means to ensure that the data acquisition time and the control system reading time are consistent. For example, use hardware timers or software lock-step technology to coordinate the time when the DAQ collects data with the time when the control system reads data to ensure that the two are strictly synchronized. In this way, the latest temperature information can be passed to the PID controller in the shortest time, avoiding control lag caused by time difference.

[0129] Setting a high priority for data acquisition and processing tasks in the operating system ensures that these critical operations can be performed at the first time without being affected by other non-critical tasks. This can minimize delays in the data processing process and enable the PID controller to always adjust according to the latest temperature information. It is crucial to develop an efficient algorithm for quickly accessing the latest data in the ring buffer. For example, a pointer indexing method can be used to directly point to the latest data location instead of traversing the entire queue. Such an approach can significantly increase the speed of data extraction and ensure that the PID controller receives the latest temperature information. In addition to efficient data extraction, the extracted data needs to be verified and exceptions handled. Confirming that the data is not lost or damaged can be done by methods such as checksums and cyclic redundancy checks (CRC) to ensure the integrity of data transmission. Common methods for removing possible measurement noise or interference signals include low-pass filtering and sliding average filtering. Identify data points that are significantly deviated from the normal range and take appropriate measures, such as triggering an alarm or automatically switching to a backup sensor, to ensure that the data received by the PID controller is accurate and reliable. To further improve the response speed of the PID controller, a high-frequency update strategy can be considered. This means that the PID controller checks temperature changes more frequently and adjusts the output accordingly to respond to temperature fluctuations more quickly. In addition, combined with adaptive control or intelligent optimization algorithms (such as genetic algorithm, particle swarm optimization), PID parameters can be dynamically adjusted according to real-time feedback to obtain better control effects.

[0130] In the above-mentioned real-time temperature control method applied to the injection molding machine, the step S6 calculates the difference between the current temperature value and the target temperature value, and inputs the difference into the real-time temperature PID controller to obtain a temperature control instruction. Specifically, the real-time temperature PID controller can dynamically adjust the output temperature control instruction according to the error signal and the optimized PID parameters to drive the heater or other temperature regulating equipment, thereby achieving precise control of the temperature of the injection molding process, ensuring that the temperature of the entire injection molding process is controlled within an ideal range, so as to meet production needs.

[0131] In summary, a real-time temperature control method for an injection molding machine based on an embodiment of the present application is explained, which uses a PID controller to perform temperature control during the injection molding process. First, a target temperature value is set, and the preliminary PID parameters of the real-time temperature PID controller are determined using an automatic tuning algorithm. Then, the real-time temperature data during the injection molding process is analyzed in a time series manner using artificial intelligence technology based on deep learning to reveal the effectiveness of the current PID parameters for temperature control, thereby adaptively optimizing the preliminary PID parameters based on the time series change pattern of the temperature, thereby improving the accuracy and response speed of the temperature control. In this way, it is possible to respond quickly to temperature changes, achieve more efficient and stable temperature management, thereby reducing overshoot and oscillation phenomena, and ensuring that the temperature control of the injection molding process is more stable and accurate.

[0132] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0133] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0134] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0135] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0136] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A real-time temperature control method for an injection molding machine, characterized in that: include: Set the target temperature value; Determine preliminary PID parameters of real-time temperature PID controller using auto-tuning algorithm; A time queue for receiving temperature values ​​collected by a temperature sensor; Based on the time queue of the temperature values, performing parameter optimization on preliminary PID parameters of the real-time temperature PID controller to obtain optimized PID parameters; extracting a current temperature value from the time queue of temperature values; Calculating the difference between the current temperature value and the target temperature value, and inputting the difference into the real-time temperature PID controller to obtain a temperature control instruction; The method of optimizing the preliminary PID parameters of the real-time temperature PID controller based on the time queue of the temperature value to obtain the optimized PID parameters includes: Performing time series feature extraction based on modal decomposition on the time queue of the temperature values ​​to obtain a set of time series encoding feature vectors of temperature intrinsic modal signal components; Performing graph walk context-aware aggregation on the set of temporal encoding feature vectors of the temperature intrinsic modal signal components to obtain a significant aggregation feature vector of the temperature intrinsic modal set; Based on the significant aggregation feature vector of the temperature intrinsic mode set, the preliminary PID parameters are optimized to obtain the optimized PID parameters.

2. The real-time temperature control method for an injection molding machine according to claim 1, characterized in that: Performing time series feature extraction based on modal decomposition on the time queue of the temperature value to obtain a set of time series encoding feature vectors of temperature intrinsic modal signal components, including: performing variational modal decomposition on the time sequence of the temperature values ​​to obtain a set of temperature intrinsic modal signal components; Each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components is time-series encoded respectively to obtain a set of time-series encoded feature vectors of the temperature intrinsic modal signal components.

3. The real-time temperature control method for an injection molding machine according to claim 2, characterized in that: The steps of respectively performing time-series coding on each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components to obtain a set of time-series coding feature vectors of the temperature intrinsic modal signal components include: Each temperature intrinsic modal signal component in the set of temperature intrinsic modal signal components is respectively input into a temperature modal component time series feature extractor based on an LSTM model to obtain a set of time series encoding feature vectors of the temperature intrinsic modal signal components.

4. The real-time temperature control method for an injection molding machine according to claim 3, characterized in that: Performing graph walk context-aware aggregation on the set of temporal encoding feature vectors of the temperature intrinsic modal signal components to obtain a significant aggregation feature vector of the temperature intrinsic modal set, including: Extracting the topological structure features of the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a temperature intrinsic modal signal component graph walk topological feature matrix; Based on the temperature intrinsic modal signal component graph walk topology feature matrix, context-aware enhancement is performed on the set of time-series encoding feature vectors of the temperature intrinsic modal signal component to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component; Based on the feature correlation between the set of context-aware enhanced feature vectors of the temperature eigenmodal signal components and the set of time-series encoding feature vectors of the temperature eigenmodal signal components, significance aggregation is performed on the set of time-series encoding feature vectors of the temperature eigenmodal signal components to obtain a significant aggregation feature vector of the temperature eigenmodal set.

5. The real-time temperature control method for an injection molding machine according to claim 4, characterized in that: Extracting the topological structure features of the set of time-series encoding feature vectors of the temperature intrinsic modal signal components to obtain a temperature intrinsic modal signal component graph walk topological feature matrix includes: Inputting each of the time-series encoding feature vectors of the temperature intrinsic modal signal component in the set of the time-series encoding feature vectors of the temperature intrinsic modal signal component into a hyperbolic space mapper to obtain a set of time-series encoding feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping; Calculating the Poincare distance between any two time-series coded eigenvectors of the temperature intrinsic modal signal components after hyperbolic space mapping in the set of time-series coded eigenvectors of the temperature intrinsic modal signal components after hyperbolic space mapping to obtain a temperature intrinsic modal signal component graph walk topology matrix; The temperature intrinsic modal signal component graph walk topology matrix is ​​subjected to dilated convolution coding to obtain the temperature intrinsic modal signal component graph walk topology feature matrix.

6. The real-time temperature control method for an injection molding machine according to claim 5, characterized in that: Based on the temperature intrinsic modal signal component graph walk topology feature matrix, context-aware enhancement is performed on the set of time-series encoding feature vectors of the temperature intrinsic modal signal component to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component, including: The graph walk topological feature matrix of the temperature intrinsic modal signal component and the set of temporal coding feature vectors of the temperature intrinsic modal signal component after hyperbolic space mapping are input into a global context walk encoder based on a graph convolutional neural network model to obtain a set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component.

7. The real-time temperature control method for an injection molding machine according to claim 6, characterized in that: Based on the feature correlation between the set of context-aware enhanced feature vectors of the temperature intrinsic modal signal component and the set of time-series coded feature vectors of the temperature intrinsic modal signal component, performing significance aggregation on the set of time-series coded feature vectors of the temperature intrinsic modal signal component to obtain the significant aggregation feature vector of the temperature intrinsic modal set, including: Inputting each corresponding group of the temperature intrinsic modal signal component context-aware enhanced feature vectors and the temperature intrinsic modal signal component time-series encoded feature vectors in the set of the temperature intrinsic modal signal component context-aware enhanced feature vectors and the set of the temperature intrinsic modal signal component time-series encoded feature vectors into an autocorrelation gating unit to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors; Inputting the set of temperature intrinsic modal signal component autocorrelation gated significant confidence factors into a normalization unit based on a Softmax function to obtain a set of temperature intrinsic modal signal component autocorrelation gated significant confidence weight factors; Based on the set of temperature intrinsic modal signal component autocorrelation gated significant confidence weight factors, a position-weighted sum of a set of time-series encoding feature vectors of the temperature intrinsic modal signal component is calculated to obtain the temperature intrinsic modal set significant aggregated feature vector.

8. The real-time temperature control method for an injection molding machine according to claim 7, characterized in that: Based on the significant aggregation feature vector of the temperature intrinsic mode set, the preliminary PID parameters are optimized to obtain the optimized PID parameters, including: Inputting the significant aggregation feature vector of the temperature intrinsic mode set into a decoder-based PID parameter dynamic optimizer to obtain a parameter optimization factor; The parameter optimization factor is multiplied by the preliminary PID parameters to obtain the optimized PID parameters.

Images