An intelligent temperature control system for a stamping die of aluminum alloy sheet

Abstract

The present invention discloses an intelligent temperature control system for a stamping die of an aluminum alloy sheet, which relates to the field of aluminum alloy sheet processing and includes a temperature acquisition, data transmission, analysis and processing, control decision-making, heating and cooling execution module; for temperature acquisition, a variety of sensors and adaptive layout are used to ensure accurate data; for data transmission, multiple technologies and blockchain authentication are combined to ensure stability; for analysis and processing, a quantum neural network and causal analysis are introduced to mine features; for control decision-making, multiple algorithms are combined and multiple parameters are considered to achieve intelligent decision-making; for heating and cooling, a new device and phase change materials are used, and combined with an advanced control strategy to accurately control the temperature. The present invention accurately acquires the temperature, multiple technologies ensure data transmission, quantum computing improves the analysis and decision-making ability, and the new device and control strategy achieve accurate temperature control; it can also monitor the die state, diagnose faults, improve the stamping quality and efficiency, extend the die life, and reduce costs.

Description

Technical Field

[0001] The present invention relates to the technical field of aluminum alloy sheet processing, and particularly relates to an intelligent temperature control system for a stamping die of an aluminum alloy sheet. Background Art

[0002] Due to excellent properties such as low density, high strength, and corrosion resistance, aluminum alloy sheets have been widely used in many fields such as aerospace, automobile manufacturing, and electronic equipment. As an important process for aluminum alloy sheet processing, stamping can efficiently process aluminum alloy sheets into various complex-shaped parts. However, the die temperature control during the stamping process of aluminum alloy sheets has always been a key factor restricting the stamping quality and production efficiency.

[0003] In the traditional aluminum alloy sheet stamping process, the die temperature is mainly controlled by the experience of operators and simple temperature adjustment devices. This method has many drawbacks. First, the experience of operators has limitations, and it is difficult to accurately grasp the optimal temperature of the die in different stamping stages and different working conditions. Moreover, the manual adjustment has a slow response speed and cannot respond in time to the rapid change of the die temperature, easily resulting in large fluctuations in the die temperature. When the die temperature is too high, the aluminum alloy sheet may be overly softened, resulting in a decrease in the dimensional accuracy of the formed part, a deterioration in the surface quality, and even defects such as cracks; while when the die temperature is too low, the plasticity of the aluminum alloy sheet decreases, the stamping difficulty increases, the die wear is easily aggravated, and the service life of the die is reduced.

[0004] Secondly, the traditional temperature adjustment devices have a single function and limited adjustment accuracy. For example, some heating devices can only provide a fixed heating power and cannot be dynamically adjusted according to the actual temperature requirements of the die; the coolant flow rate and temperature of the cooling device are also difficult to accurately control, and cannot achieve precise adjustment of the die temperature. In addition, the traditional temperature acquisition method mainly uses a few thermocouple sensors, which can only obtain the temperature information of the local part of the die and cannot comprehensively and accurately reflect the overall temperature distribution of the die, resulting in a lack of sufficient basis for temperature control decisions.

[0005] With the development of industrial automation and intelligence, although some advanced temperature control technologies have been gradually applied to the stamping die of aluminum alloy sheets, there are still some problems. The existing intelligent temperature control systems need to improve in terms of data processing and analysis capabilities, cannot fully explore the potential information behind the temperature data, and are difficult to achieve precise prediction and intelligent control of the die temperature. At the same time, these systems lack sufficient adaptability and flexibility when dealing with complex and changeable stamping working conditions and die states, and cannot adjust the control strategy in time to meet the production requirements.

[0006] In addition, during the long-term use of the mold, faults such as wear and cracks will occur. These faults will not only affect the temperature distribution of the mold, but also reduce the quality and efficiency of stamping forming. However, the existing temperature control systems often lack the functions of real-time monitoring and fault diagnosis of the mold state, and cannot detect potential problems of the mold in time for early warning and repair, increasing the uncertainty and cost in the production process.

[0007] Therefore, it is of great practical significance to develop an intelligent temperature control system for aluminum alloy sheet stamping molds. This system needs to be able to collect mold temperature information in real time and accurately, process and analyze data efficiently, achieve precise control of the mold temperature, and have the functions of mold state monitoring and fault diagnosis and early warning, so as to improve the quality and production efficiency of aluminum alloy sheet stamping forming and reduce production costs. Summary of the Invention

[0008] The intelligent temperature control system for aluminum alloy sheet stamping molds proposed by the present invention aims to solve the problems mentioned in the above existing technologies.

[0009] To achieve the above object, the present invention adopts the following technical solutions: An intelligent temperature control system for aluminum alloy sheet stamping molds, comprising:

[0010] Temperature acquisition module: Thermocouple sensors and fiber Bragg grating sensors are evenly distributed at key parts of the mold, and an adaptive layout algorithm is used to dynamically adjust the distribution positions of the sensors; the uniformity of the mold temperature distribution is evaluated through the temperature uniformity evaluation formula where is the temperature collected by the th sensor, is the average temperature, n is the number of sensors, is the weight determined according to the sensor position and importance;

[0011] Data transmission module: Using 5G communication technology and industrial Ethernet, adopting a data transmission authentication mechanism based on blockchain, and evaluating the data transmission stability through the data transmission stability evaluation formula where is the amount of correctly transmitted data, is the total amount of transmitted data, is the time of the

[0012] jth transmission interruption, and T is the total transmission time; Data analysis and processing module: Using a noise reduction algorithm to denoise the collected data, and introducing a quantum neural network for feature extraction; evaluating the effectiveness of the features through the data feature effectiveness evaluation formula where is the weight of the is the causal association strength between the j-th feature and the die temperature control target, and m is the number of features;

[0013] Control decision-making module: An intelligent decision-making model based on fuzzy logic control and particle swarm optimization algorithm is adopted. According to the real-time state of the die and the dynamic changes of the stamping process, the working parameters of the heating and cooling execution module are adjusted; A multi-parameter coupling control decision-making model is established, and the decision accuracy evaluation formula is used to evaluate the decision accuracy, is the number of correct decisions, is the total number of decisions, is the deviation between the die temperature and the target temperature after the k-th decision, and L is the total number of decisions;

[0014] Heating and cooling execution module: Equipped with an electromagnetic induction heating device and a microchannel liquid cooling device, and intelligent phase change materials are introduced. When the die temperature is too high, the phase change material absorbs heat and undergoes a phase change. When the temperature drops, the phase change material releases heat to maintain the stability of the die temperature.

[0015] Furthermore, it also includes:

[0016] Die state detection module: Monitor the pressure change during the stamping process through a pressure sensor, monitor the vibration of the die using a vibration sensor, adopt an anomaly detection algorithm based on wavelet transform, and introduce an anomaly perception technology of quantum entanglement state; Use a multi-sensor data fusion algorithm to fuse and process the data of pressure, vibration, and quantum entanglement state perception; Adopt the anomaly detection accuracy evaluation formula to evaluate the anomaly detection effect, is the number of anomalies correctly detected, is the actual number of anomalies, is the delay time of the n-th anomaly detection, is the total monitoring time.

[0017] Furthermore, it also includes:

[0018] Fault diagnosis and warning module: Use fault tree analysis method and Bayesian network to analyze the operation data of each module, and introduce a deep learning autoencoder for fault feature extraction; Adopt a fault reasoning mechanism based on knowledge graph to structurally represent and associate fault knowledge to achieve fault diagnosis and reasoning; When a potential fault is detected, display the fault location and maintenance guidance to the operator through virtual reality and augmented reality technologies; Adopt the fault prediction accuracy evaluation formula to evaluate the fault prediction effect, is the number of faults correctly predicted, is the actual number of faults that occurred, is the deviation between the j-th fault prediction time and the actual fault occurrence time, is the total fault monitoring time.

[0019] Furthermore, the temperature acquisition module uses self-calibration technology to regularly calibrate the sensor. It applies a calibration algorithm based on Kalman filtering and introduces quantum calibration technology; adopts a calibration method based on quantum bit error correction code to correct quantum noise and errors in the sensor measurement process; uses an adaptive calibration strategy to dynamically adjust the calibration period and calibration parameters according to the usage time and working environment factors of the sensor; evaluates the calibration effect through the calibration error evaluation formula where is the true temperature value, is the temperature value measured by the sensor, is the number of calibrations, is the weight determined according to the calibration time and environmental conditions.

[0020] Furthermore, the data transmission module introduces data encryption technology to encrypt the transmitted data. It adopts a quantum encryption algorithm and combines it with homomorphic encryption technology; uses a hybrid encryption mechanism based on blockchain and quantum key distribution, combining the distributed ledger of blockchain and the uncrackability of quantum keys. Evaluates the security of data encryption through the data encryption security evaluation formula where is the amount of securely transmitted data, is the total amount of data, is the probability that the -th encrypted data is attempted to be cracked,

[0021] Furthermore, the data analysis and processing module introduces transfer learning technology. It uses the temperature data and processing experience of the mold to introduce cross-domain transfer learning; transfers the temperature control data and processing methods in the aerospace and automotive manufacturing fields to the temperature control system of the aluminum alloy sheet stamping die, broadening the data source and processing ideas; adopts a transfer learning method based on meta-learning to learn the characteristics and processing patterns of data in different fields; evaluates the effect of transfer learning through the transfer learning effect evaluation formula where is the improvement value of the -th data analysis index after transfer learning, is the value of the -th data analysis index before transfer learning, n is the number of data analysis indicators, is the weight determined according to the importance of the indicators.

[0022] Furthermore, the control decision-making module adopts a multi-objective optimization algorithm to ensure the stability of the die temperature, taking into account both the quality of stamping forming and production efficiency; the non-dominated sorting genetic algorithm is used for optimization, and the quantum multi-objective optimization algorithm is introduced; considering the service life and energy consumption of the die, a multi-objective collaborative optimization model is established, and a multi-objective optimization effect evaluation formula is adopted. Evaluate the effect of multi-objective optimization. $w_j$ is the weight of the $j$-th objective. $e_j$ is the optimization effect index of the $j$-th objective. $c_{j,k}$ is the collaborative correlation strength between the $j$-th objective and other objectives, and $m$ is the number of objectives.

[0023] Furthermore, the heating and cooling execution module adopts an intelligent flow regulating valve and a variable frequency heating power supply to achieve the control of the coolant flow rate and heating power; the traditional PID control algorithm is adopted, the model predictive control algorithm is introduced, and combined with the adaptive control strategy, according to the real-time state of the die and environmental changes, the parameters of the control algorithm are automatically adjusted; through the control accuracy evaluation formula Evaluate the control accuracy. $x_0$ is the set control parameter value. $x_i$ is the actual control parameter value. $n$ is the number of control times. $w_i$ is the weight determined according to the control time and importance.

[0024] Furthermore, the die state detection module uses machine learning algorithms to classify and predict pressure and vibration data, using a method that combines support vector machines and neural networks, and introducing convolutional neural networks and recurrent neural networks of deep learning; adopting an ensemble learning-based method to fuse the results of multiple machine learning algorithms, and evaluating the classification accuracy through the formula Evaluate the classification accuracy. TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, FN is the number of false negative examples. $\Delta_{conf}^i$ is the confidence deviation of the $i$-th classification. $N$ is the total number of classifications.

[0025] Furthermore, the fault diagnosis and warning module uses a method that combines an expert system and machine learning for fault diagnosis. The expert system makes a preliminary diagnosis based on pre-set rules and experience, and the machine learning algorithm learns and analyzes the fault data; introducing a knowledge graph deep learning fault diagnosis model, adopting a case-based reasoning fault diagnosis method, and using historical fault cases for analogical reasoning to locate and solve newly emerging faults; adopting a fault diagnosis accuracy evaluation formula Evaluate the effect of fault diagnosis. $TP$ is the number of faults correctly diagnosed. $N$ is the total number of diagnoses. is the deviation between the j-th fault diagnosis time and the actual fault occurrence time, and is the total fault diagnosis time.

[0026] Compared with the existing technologies, the beneficial effects of the present invention are as follows:

[0027] In terms of temperature acquisition, the system adopts a variety of advanced sensors, including quantum dot temperature sensors, which can capture the temperature changes of the mold comprehensively and with high precision. The adaptive layout algorithm can dynamically adjust the sensor positions according to the characteristics of the mold. The improved evaluation formula can more accurately evaluate the temperature uniformity, providing an accurate data basis for subsequent temperature control.

[0028] In terms of data transmission, by combining 5G, industrial Ethernet and terahertz communication technologies, it ensures fast and stable data transmission. The blockchain authentication mechanism guarantees the integrity and traceability of data. The optimized evaluation formula comprehensively considers the situation of transmission interruption, making the data transmission more reliable.

[0029] The data analysis and processing module introduces quantum neural networks and causal analysis methods, which can mine deep-level features and find key causal associations. Cross-domain transfer learning and meta-learning methods broaden the thinking of data processing. The improved evaluation formula scientifically evaluates the effect of transfer learning, improving the efficiency and accuracy of data analysis.

[0030] The control decision-making module combines fuzzy logic, particle swarm optimization and reinforcement learning algorithms, considering multi-parameter coupling, and can quickly adjust the control parameters according to the real-time state of the mold and process changes. The quantum multi-objective optimization algorithm can quickly find the global optimal solution, taking into account multiple objectives. The improved evaluation formula comprehensively evaluates the optimization effect, realizing precise temperature control decision-making.

[0031] The heating and cooling execution module adopts new heating and cooling devices, such as bionic microchannel liquid cooling devices and intelligent phase change materials, which can efficiently regulate the temperature of the mold. The model predictive control algorithm and adaptive control strategy improve the control precision and response speed. The improved evaluation formula reasonably evaluates the control effect.

[0032] In addition, the mold condition monitoring module introduces quantum entanglement state sensing technology and multi-sensor data fusion algorithms, which can detect minor damages of the mold in advance. The improved evaluation formula comprehensively considers the detection delay, improving the accuracy of anomaly detection. The fault diagnosis and early warning module combines deep learning and knowledge graphs, provides maintenance guidance through VR and AR technologies. The improved evaluation formula accurately evaluates the fault prediction effect, and can detect and solve potential faults in a timely manner.

[0033] Generally speaking, this system improves the quality and production efficiency of aluminum alloy sheet stamping, extends the service life of the mold, reduces the production cost, and has good application prospects and economic value. Description of the Drawings

[0034] Figure 1 The schematic block diagram of an intelligent temperature control system for a stamping die of an aluminum alloy sheet proposed by the present invention; Detailed Embodiments

[0035] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present invention.

[0037] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality" means two or more unless otherwise specifically defined. In addition, the terms "mounted", "connected" and "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the internal communication of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the accompanying drawings.

[0038] Refer to Figure 1 : An intelligent temperature control system for a stamping die of an aluminum alloy sheet, comprising:

[0039] Temperature acquisition module: The temperature acquisition module plays a crucial role in the temperature monitoring and control of the mold. At key parts of the mold, in addition to evenly distributing high-precision thermocouple sensors and fiber Bragg grating sensors, quantum dot temperature sensors are innovatively introduced. The thermocouple sensor is based on the Seebeck effect. When the two ends of two conductors made of different materials are connected to form a loop and there is a temperature difference at both ends, a thermoelectromotive force will be generated in the loop, and the temperature can be obtained by measuring this electromotive force. The fiber Bragg grating sensor utilizes the reflection characteristics of the grating in the optical fiber for light with a specific wavelength. When the temperature changes, the period and effective refractive index of the grating change, resulting in the drift of the center wavelength of the reflected light. The temperature can be determined by detecting the amount of wavelength drift. The working principle of the quantum dot temperature sensor is based on the fluorescence characteristics of quantum dots. Quantum dots are a kind of nanoscale semiconductor material that emits fluorescence when excited by light, and the fluorescence emission wavelength and intensity change with temperature. By precisely measuring this fluorescence characteristic, ultra-high-precision temperature perception can be achieved. Its unique advantage is that it can capture temperature changes in tiny areas of the mold, which is difficult for traditional sensors to achieve. Moreover, its response speed is several times faster than that of traditional sensors, meaning that temperature information can be obtained more timely. In terms of sensor layout, an adaptive layout algorithm is adopted. This algorithm will first accurately model the shape and size of the mold, and use computer-aided design (CAD) and computer-aided engineering (CAE) technologies to simulate the temperature field distribution inside the mold under different stamping processes. Then, according to the simulation results, combined with stamping process requirements such as stamping speed, pressure magnitude, material properties, etc., the distribution positions of the sensors are dynamically adjusted. For example, in the stress concentration area or the area with drastic temperature changes of the mold, the number of sensors will be appropriately increased to ensure accurate temperature data acquisition and achieve the best temperature acquisition effect. In order to more accurately evaluate the uniformity of the mold temperature distribution, an improved temperature uniformity evaluation formula (where is the temperature collected by the th sensor, is the average temperature, n is the number of sensors, is the weight determined according to the sensor position and importance) is used to more accurately evaluate the uniformity of the mold temperature distribution.

[0040] Data Transmission Module: The data transmission module is an important part to ensure the accurate and stable transmission of key data such as mold temperature. Based on the application of 5G communication technology and industrial Ethernet, terahertz communication technology is introduced as a backup communication link. 5G communication technology is based on millimeter-wave frequency bands and technologies such as massive MIMO (Multiple-Input Multiple-Output), capable of providing peak data rates of up to dozens of Gbps, ultra-low latency in the millisecond range, and connection numbers in the millions per square kilometer. Industrial Ethernet adopts the IEEE802.3 standard, featuring openness, low cost, high reliability, etc., and is widely used in industrial environments to achieve high-speed data transmission between devices. Terahertz communication technology operates in the frequency band of 0.1 - 10 THz, with extremely high transmission rates and strong anti-interference capabilities. Its extremely high bandwidth enables it to transmit a large amount of data in a short time. For example, when the mold temperature data suddenly increases, it can also transmit quickly. Moreover, terahertz waves have strong directivity and are not easily interfered by signals in other frequency bands. When 5G and industrial Ethernet fail, such as being affected by electromagnetic interference, network congestion, or physical damage, the terahertz communication system can quickly switch. It uses an intelligent monitoring module to continuously monitor the status of the primary communication link. Once an anomaly is detected, the terahertz communication link will be activated within an extremely short time (nanosecond level) to ensure uninterrupted data transmission. To ensure data integrity and source traceability, a blockchain-based data transmission authentication mechanism is adopted. Blockchain is a decentralized distributed ledger technology that divides data into multiple data blocks, and each data block contains all data transaction records within a certain period of time. During data transmission, a unique digital signature is added to each data transmission packet. This digital signature is generated through an asymmetric encryption algorithm. The sender uses its own private key to sign the data, and the receiver uses the sender's public key for verification. At the same time, each block in the blockchain contains the hash value of the previous block, forming a chained structure. Once the data is tampered with during transmission, the receiver can detect the anomaly by verifying the digital signature and hash value, thus ensuring data integrity. Also, through the distributed ledger of the blockchain, the source and transmission path of the data can be traced to ensure the source traceability of the data. When evaluating the data transmission stability, an optimized data transmission stability evaluation formula is adopted (where is the amount of data correctly transmitted, is the total amount of transmitted data, is the time of the j-th transmission interruption, and T is the total transmission time), considering the transmission interruption situation comprehensively to evaluate the data transmission stability more comprehensively.

[0041] Data analysis and processing module: The data analysis and processing module undertakes the key data processing and feature extraction tasks in the entire mold temperature management system. First, the collected data is denoised using a deep learning-based denoising algorithm. Deep learning algorithms rely on multi-layer neural network structures, such as common convolutional neural networks (CNNs) or recurrent neural networks (RNNs) and their variants (such as LSTM, GRU, etc.). When processing temperature data, for CNN, the temperature data can be structured in the time and space dimensions to make it similar to image data. The local and global features in the data are automatically extracted through components such as convolution layers and pooling layers. The convolution kernel is used to slide convolution on the data to capture the patterns and regularities in the data and remove noise interference. For RNN and its variants, since temperature data has time series characteristics, they can effectively process this sequence data, capture the long-term dependencies in the data through memory units, identify the differences between noise signals and true temperature signals, and then filter and remove noise. At the same time, quantum neural networks are introduced for feature extraction. Quantum neural networks use the superposition and entanglement characteristics of quantum bits. Unlike traditional bits, which have only two states, 0 and 1, quantum bits (qubits) can be in a superposition state of 0 and 1 at the same time, which enables quantum neural networks to have the ability to parallelize calculations when processing data and can process multiple data states at the same time. The entanglement property allows a special connection between quantum bits. Even if they are separated in space, the state change of one quantum bit will instantly affect other entangled quantum bits. When processing complex temperature data, quantum neural networks can more efficiently explore the high-dimensional space of data and dig out deep-level features hidden in the data, such as the subtle change pattern of mold temperature under complex working conditions. In terms of feature screening, a feature screening method based on causal analysis is used. This method combines domain knowledge and data mining technology. Domain knowledge includes professional knowledge such as mold material characteristics and stamping process principles. For example, the difference in thermal conductivity of different mold materials will affect the temperature distribution, and the parameters such as stamping speed and pressure in the stamping process are closely related to temperature changes. Data mining technology uses algorithms to discover potential patterns and laws from large amounts of data. In causal analysis, methods such as Granger causality test are used to determine the causal relationship between different features and between features and mold temperature. For example, by analyzing and judging whether the feature of the mold surface coating thickness is the Granger cause that affects the mold temperature change, the key features that have a direct causal relationship with the mold temperature control can be found. When evaluating the effectiveness of the features, the improved data feature effectiveness evaluation formula is used (in is the weight of the jth feature, is the contribution rate of the jth feature, is the causal relationship strength between the jth feature and the mold temperature control target, and m is the number of features), which can more accurately evaluate the effectiveness of the features.

[0042] Control decision-making module: The control decision-making module is the core component for precise control of the mold temperature. It comprehensively applies a variety of advanced algorithms and models to achieve efficient and intelligent control decisions. This module adopts an intelligent decision-making model based on fuzzy logic control and particle swarm optimization algorithm, and on this basis, a reinforcement learning algorithm is introduced. Fuzzy logic control is a control method that mimics the human way of thinking. It transforms precise input data into fuzzy sets through membership functions. For example, it converts the precise value of the mold temperature into fuzzy concepts such as "low temperature", "medium temperature", "high temperature", etc. Then, according to pre-set fuzzy rules, such as "if the mold temperature is low and the temperature change rate is small, then appropriately increase the heating power", fuzzy reasoning is carried out. Finally, through defuzzification operation, the fuzzy output is converted into a precise control quantity, such as the specific value of heating power adjustment. The particle swarm optimization algorithm (PSO) is an optimization algorithm based on swarm intelligence. In the scenario of mold temperature control, each particle represents a set of possible control parameters (such as heating power, cooling flow rate, etc.). The particles fly in the solution space and find the optimal solution by continuously updating their own positions. The position update of the particles is based on their own historical optimal positions and the global optimal position of the group. Through iterative search, it gradually converges to the optimal combination of control parameters to achieve optimal control of the mold temperature. The reinforcement learning algorithm interacts with the environment and continuously learns the optimal control strategy. In mold temperature control, the environment is the mold and its stamping process system. The reinforcement learning algorithm takes the real-time state of the mold (such as the current temperature, temperature change trend, stamping process stage, etc.) as input and outputs corresponding control actions (such as adjusting the working parameters of heating or cooling equipment). After each execution of the control action, the system will give a reward or punishment signal according to the change of the mold temperature. The algorithm continuously tries different control actions, adjusts the strategy according to the reward feedback, and gradually learns the optimal control strategy in different states, so that it can quickly adjust the working parameters of the heating and cooling execution modules according to the real-time state of the mold and the dynamic changes of the stamping process. At the same time, considering factors such as the thermal properties of the mold material (such as thermal conductivity, specific heat capacity, etc.), the material and thickness of the aluminum alloy sheet, a control decision-making model with multi-parameter coupling is established. The thermal conductivity of the mold material determines the heat conduction speed inside the mold, and the specific heat capacity affects the ease of temperature change of the mold; different materials of aluminum alloy sheets have different heat dissipation and deformation characteristics during the stamping process, and the thickness will affect the heat exchange efficiency between the sheet and the mold. Through comprehensive analysis and modeling of these parameters, using methods such as finite element analysis to simulate the temperature change of the mold under different parameter combinations, a control decision-making model that can accurately reflect the multi-parameter coupling relationship is constructed, providing a more comprehensive and accurate basis for control decisions. In terms of evaluating the decision-making accuracy, an improved decision-making accuracy evaluation formula is adopted (where is the number of correct decisions, is the total number of decisions, is the deviation between the mold temperature and the target temperature after the k-th decision, and L is the total number of decisions). Considering the temperature deviation after the decision comprehensively, the decision accuracy can be evaluated more scientifically.

[0043] Heating and Cooling Execution Module: As the direct executor of mold temperature regulation, the heating and cooling execution module is equipped with an electromagnetic induction heating device and a microchannel liquid cooling device, ensuring precise control of the mold temperature through a variety of advanced technologies. The electromagnetic induction heating device adopts a new type of high-frequency induction coil material and topological structure. In terms of materials, a special alloy material with high magnetic permeability, low hysteresis loss, and low eddy current loss is selected. This material can efficiently generate induced current under a high-frequency alternating magnetic field, converting electrical energy into heat energy and significantly improving the heating efficiency. For example, some ferrite-metal composite materials have a much higher magnetic permeability than traditional copper coil materials. Under the same electromagnetic excitation, they can generate a stronger induced magnetic field, thereby increasing the heating power. In terms of topological structure, an optimized multi-turn spiral or array design is adopted. The multi-turn spiral structure can form a relatively uniform magnetic field distribution on the mold surface, making each part of the mold heat more evenly. The array structure can flexibly adjust the position and parameters of each induction unit according to the shape and heating requirements of the mold, achieving precise heating of local areas of the mold. Through precise electromagnetic field simulation and experimental optimization, the optimal number of coil turns, spacing, and arrangement method are determined, further improving the heating uniformity and efficiency. The microchannel liquid cooling device adopts a bionic microchannel design. This design mimics the blood vessel network in organisms to construct a complex and delicate microchannel structure. These microchannels have a very small inner diameter, usually between dozens and hundreds of micrometers, greatly increasing the contact area between the coolant and the inner wall of the mold. When the coolant flows through the microchannels, it can more effectively remove the heat inside the mold. At the same time, the branch structure of the microchannels is similar to the branches of blood vessels, enabling the coolant to be evenly distributed in various regions inside the mold, avoiding local overheating or uneven cooling problems that may occur in traditional cooling methods. Through reasonable fluid mechanics design, the shape, size, and layout of the microchannels are optimized to ensure the minimum flow resistance of the coolant in the microchannels and achieve the best cooling effect. In addition, this module introduces intelligent phase change materials. These phase change materials have a specific phase change temperature range. When the mold temperature is too high, the phase change material absorbs heat and changes from a solid state to a liquid state or other phase states, undergoing a phase change process. During this process, the phase change material absorbs a large amount of latent heat, playing a role in buffering and regulating the temperature to prevent the mold temperature from rising sharply. For example, some organic polymer phase change materials can absorb a large amount of heat while maintaining a relatively stable temperature when the temperature reaches their phase change point. When the temperature decreases, the phase change material changes from a liquid state to a solid state, releasing the stored heat to maintain the stability of the mold temperature. By precisely controlling the addition amount and distribution position of the phase change material, it can play the best role in mold temperature regulation, further improving the stability and accuracy of mold temperature control.

[0044] In the present invention, the following modules are further included:

[0045] Mold Status Detection Module, which plays a crucial role in monitoring the mold status. It ensures accurate control of the mold status through various advanced technical means. First, the module monitors the pressure changes during the stamping process through pressure sensors. The pressure sensors adopt high-precision piezoresistive or piezoelectric principles. The piezoresistive pressure sensor is based on the piezoresistive effect. When subjected to pressure, the internal resistance value will change, and the pressure magnitude can be accurately obtained by measuring the change in the resistance value. The piezoelectric pressure sensor utilizes the property of piezoelectric materials to generate charges when subjected to pressure, and determines the pressure by measuring the magnitude of the charges. These pressure sensors are precisely installed at the key stress-bearing parts of the mold, capable of capturing the dynamic changes in pressure during the stamping process in real-time and accurately, providing basic data for analyzing the stress state of the mold. At the same time, vibration sensors are used to monitor the vibration of the mold. Vibration sensors mostly adopt accelerometers or displacement sensors. The accelerometer can measure the acceleration changes of the mold in various directions, and the characteristics such as the vibration frequency and amplitude of the mold can be understood by analyzing the acceleration signals. The displacement sensor can accurately measure the displacement of the mold components, thereby judging whether there is looseness or abnormal displacement in the mold. These vibration sensors are distributed at different positions of the mold to comprehensively monitor the vibration status of the mold. In addition to adopting the anomaly detection algorithm based on wavelet transform, the anomaly perception technology based on quantum entanglement state is also introduced. Quantum entanglement state is a quantum mechanics phenomenon. For a quantum system in an entangled state, even if they are spatially separated from each other, the state change of one quantum will instantaneously affect the state of the other quantum. In mold monitoring, the quantum entanglement state is extremely sensitive to tiny physical changes. For example, tiny damages or defects inside the mold will cause extremely tiny changes in physical properties such as local stress and strain, and these changes will affect the quantum system in the entangled state through the surrounding physical fields. Quantum sensors can capture these extremely weak change signals, thereby pre-sensing the tiny damages and defects inside the mold, providing an important basis for the preventive maintenance of the mold. In order to improve the accuracy and reliability of anomaly detection, a multi-sensor data fusion algorithm is adopted. This algorithm first pre-processes the data sensed by pressure, vibration, and quantum entanglement state, including data cleaning, filtering, etc., to remove noise and interference signals. Then, data fusion techniques, such as the data fusion method based on Bayesian network, are used to fuse the data from different sensors. Bayesian network can comprehensively judge the status of the mold according to the probability relationship between the data from different sensors. For example, when the pressure sensor detects an abnormal increase in pressure, the vibration sensor detects an abnormal vibration frequency, and the quantum entanglement state senses a tiny change inside the mold, through the inference calculation of the Bayesian network, it can more accurately judge whether there is an anomaly in the mold, as well as the type and location of the anomaly. In terms of evaluating the effect of anomaly detection, an improved anomaly detection accuracy evaluation formula (where is the number of anomalies correctly detected, is the actual number of exceptions, is the delay time of the th anomaly detection,

[0046] is the total monitoring time), comprehensively considering the detection delay situation, and more reasonably evaluating the anomaly detection effect.Fault Diagnosis and Early Warning Module: This module plays a crucial role in ensuring the stable operation of the system. It accurately diagnoses and warns of system faults through a variety of advanced technical means. The module uses the fault tree analysis method and Bayesian network to analyze the operation data of each module of the system. The fault tree analysis method is a graphical deduction method. Starting from a major fault (top event) that the system does not want to occur, it gradually analyzes various direct and indirect causes (intermediate events and bottom events) that lead to the top event. These events are connected by logical gates (such as AND gates, OR gates, etc.) to construct an inverted tree-shaped logical causal relationship diagram. For example, taking the out-of-control mold temperature as the top event, the possible causes of this event include intermediate events such as faults in the heating and cooling execution module and errors in the data analysis and processing module. Further analysis can also find bottom events such as sensor faults and control algorithm errors. Through this method, the causal relationship of system faults can be clearly sorted out, providing clues for fault troubleshooting. The Bayesian network is a graphical network model based on probabilistic reasoning. It uses nodes to represent variables (such as the operation parameters of each module) and directed edges to represent the probabilistic dependence relationships between variables. By learning and analyzing historical operation data, the conditional probability distribution between each node is determined. During fault diagnosis, according to the currently observed operation data, the probability of each fault occurring is calculated using Bayes' formula to determine the most likely type of fault in the system. On this basis, an autoencoder of deep learning is introduced for fault feature extraction. The autoencoder is a special neural network structure composed of an encoder and a decoder. The encoder compresses the input high-dimensional operation data into a low-dimensional feature representation, and the decoder attempts to restore these feature representations to the original data. During the training process, the autoencoder learns the inherent features of the data by minimizing the reconstruction error. When the system fails, the features of the operation data will change, and the autoencoder can capture these changes and discover potential fault patterns. For example, by training on temperature, pressure and other data in normal operation and fault states, the autoencoder can identify abnormal changes in data features during faults, thus extracting fault features. A fault reasoning mechanism based on a knowledge graph is adopted to structurally represent and associate fault knowledge. The knowledge graph is a semantic network that stores knowledge in the form of entity-relationship-entity triples. In the intelligent mold temperature control system, various fault types, fault causes, fault manifestations, repair methods, etc. are used as entities, and the causal relationships, association relationships, etc. between them are used as relationships to construct a fault knowledge graph. When an abnormality is detected in the system, by reasoning and querying in the knowledge graph, the cause and solution of the fault can be quickly determined, realizing more accurate fault diagnosis and reasoning. When potential faults are detected, the fault location and repair guidance are displayed to the operator through virtual reality (VR) and augmented reality (AR) technologies.VR technology creates a completely virtual environment. Operators can wear VR devices to view the structure and fault locations of the mold in all directions in the virtual environment, and at the same time obtain detailed repair steps and operation guides. AR technology superimposes virtual information on the real scene. Operators can, through devices such as AR glasses, see the markings of fault locations and repair prompt information on the actual mold, and conduct fault troubleshooting and repair more intuitively. In evaluating the fault prediction effect, an improved fault prediction accuracy evaluation formula is adopted. (where is the number of faults correctly predicted, is the number of actually occurring faults, is the deviation between the j - th fault prediction time and the actual fault occurrence time, is the total fault monitoring time), comprehensively considering the prediction time deviation, to more accurately evaluate the fault prediction effect.

[0047] In the present invention, the temperature acquisition module uses self - calibration technology to regularly calibrate the sensor. In addition to applying the calibration algorithm based on Kalman filtering, quantum calibration technology is introduced. Quantum calibration technology utilizes the stability and repeatability of quantum states to provide a high - precision calibration reference for the sensor. By adopting the calibration method based on quantum bit error - correcting codes, it can effectively correct the quantum noise and errors in the sensor measurement process. An adaptive calibration strategy is used to dynamically adjust the calibration period and calibration parameters according to factors such as the usage time and working environment of the sensor. An improved calibration error evaluation formula is adopted (where is the true temperature value, is the temperature value measured by the sensor, is the number of calibrations, is the weight determined according to the calibration time and environmental conditions), to more reasonably evaluate the calibration effect.

[0048] In the present invention, the data transmission module introduces data encryption technology to encrypt the transmitted data. In addition to adopting the algorithm based on quantum encryption, homomorphic encryption technology is combined. Homomorphic encryption allows direct calculation on encrypted data without decryption, ensuring the security of data during the processing. A hybrid encryption mechanism based on blockchain and quantum key distribution is adopted, combining the distributed ledger of blockchain and the uncrackability of quantum keys to further improve the security of data encryption. An improved data encryption security evaluation formula is adopted (where is the amount of securely transmitted data, is the total amount of data, is the probability that the - th encrypted data is attempted to be cracked, is the total number of encrypted data transmissions), comprehensively considers the risk of encrypted data being cracked, and more comprehensively evaluates the security of data encryption.

[0049] In the present invention, the data analysis and processing module introduces transfer learning technology. In addition to using the temperature data and processing experience of other similar molds, cross-domain transfer learning is introduced. The temperature control data and processing methods in the fields of aerospace, automobile manufacturing, etc. are transferred to the temperature control system of aluminum alloy sheet stamping forming molds to broaden the data source and processing ideas. The meta-learning-based transfer learning method is used to quickly learn the characteristics and processing modes of data in different fields, thereby improving the efficiency and effect of transfer learning. The improved transfer learning effect evaluation formula is used (in After transfer learning The improvement value of data analysis indicators, It is the first The value of the data analysis indicator, n is the number of data analysis indicators, is a weight determined according to the importance of the indicator) to more scientifically evaluate the effect of transfer learning.

[0050] In the present invention, the control decision module adopts a multi-objective optimization algorithm to ensure the temperature stability of the mold while taking into account the quality and production efficiency of stamping. In addition to using the non-dominated sorting genetic algorithm (NSGA-II) for optimization, a quantum multi-objective optimization algorithm is introduced. The quantum multi-objective optimization algorithm uses the superposition and entanglement characteristics of quantum bits to search for the global optimal solution more quickly. Considering more objectives such as the service life of the mold and energy consumption, a multi-objective collaborative optimization model is established. The improved multi-objective optimization effect evaluation formula is adopted (in is the weight of the jth target, is the optimization effect index of the jth objective, is the synergistic correlation strength between the jth objective and other objectives, and m is the number of objectives), to more comprehensively evaluate the effect of multi-objective optimization.

[0051] In the present invention, the heating and cooling execution modules use intelligent flow control valves and variable frequency heating power supplies to achieve precise control of coolant flow and heating power. In addition to the traditional PID control algorithm, the model predictive control (MPC) algorithm is introduced. The MPC algorithm can predict and adjust the control parameters in advance according to the temperature model of the mold and future process requirements, thereby improving the control accuracy and response speed. Adopt an adaptive control strategy to automatically adjust the parameters of the control algorithm according to the real-time status of the mold and environmental changes. Use an improved control accuracy evaluation formula (in is the set control parameter value, is the actual control parameter value, is the number of control times, (which is the weight determined according to the control time and importance), to more reasonably evaluate the control accuracy.

[0052] In the present invention, the mold state detection module uses machine learning algorithms to classify and predict pressure and vibration data. In addition to using the method of combining support vector machine (SVM) and neural network, the convolutional neural network (CNN) and recurrent neural network (RNN) of deep learning are introduced. CNN can automatically extract the spatial features of data, and RNN can process the time features of sequence data to improve the accuracy of classification and prediction. By using the method based on ensemble learning, the results of multiple machine learning algorithms are fused to further enhance the reliability of classification and prediction. An improved classification accuracy evaluation formula is adopted (where TP is the number of true positive cases, TN is the number of true negative cases, FP is the number of false positive cases, FN is the number of false negative cases, is the confidence deviation of the nth classification,

[0053] is the total number of classifications), comprehensively considering the classification confidence deviation, to more accurately evaluate the accuracy of classification.

[0053] In the present invention, the fault diagnosis and early warning module uses a method combining expert system and machine learning for fault diagnosis. In addition to the expert system making a preliminary diagnosis according to pre-set rules and experience, and the machine learning algorithm learning and analyzing a large amount of fault data, a deep learning fault diagnosis model enhanced by knowledge graph is introduced. The knowledge graph can provide rich domain knowledge for the deep learning model to improve the accuracy and interpretability of fault diagnosis. By using the case-based reasoning fault diagnosis method, analogical reasoning is carried out using historical fault cases to quickly locate and solve newly emerging faults. An improved fault diagnosis accuracy evaluation formula is adopted (where is the number of faults correctly diagnosed, is the total number of diagnoses, is the deviation between the jth fault diagnosis time and the actual fault occurrence time, is the total fault diagnosis time), comprehensively considering the diagnosis time deviation, to more accurately evaluate the effect of fault diagnosis.

[0054] The above is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. An intelligent temperature control system for a stamping die of an aluminum alloy sheet, characterized in that, Including: Temperature acquisition module: Thermocouple sensors and fiber Bragg grating sensors are evenly distributed in the mold, and an adaptive layout algorithm is used to dynamically adjust the distribution positions of the sensors. Through the temperature uniformity evaluation formula Evaluate the uniformity of the mold temperature distribution, is the temperature collected by the th sensor, is the average temperature, n is the number of sensors, is the weight determined according to the sensor position and importance; Data transmission module: Utilize 5G communication technology and industrial Ethernet, adopt a blockchain-based data transmission authentication mechanism, and evaluate the data transmission stability through a data transmission stability evaluation formula Evaluate the data transmission stability, is the amount of correctly transmitted data, is the total amount of transmitted data, is the time of the j-th transmission interruption, and T is the total transmission time; Data analysis and processing module: Use noise reduction algorithms to denoise the collected data, and introduce quantum neural networks for feature extraction; Adopt the data feature effectiveness evaluation formula to evaluate the effectiveness of features, where $w_j$ is the weight of the $j$-th feature, $c_j$ is the contribution rate of the $j$-th feature, $r_{j}$ is the causal association strength between the $j$-th feature and the die temperature control target, and $m$ is the number of features; Control decision-making module: An intelligent decision-making model based on fuzzy logic control and particle swarm optimization algorithm is adopted to adjust the working parameters of the heating and cooling execution module according to the real-time state of the mold and the dynamic changes of the stamping process; A multi-parameter coupling control decision-making model is established, and the decision accuracy evaluation formula Evaluate the decision accuracy, is the number of correct decisions, is the total number of decisions, is the deviation between the mold temperature and the target temperature after the k-th decision, and L is the total number of decisions; Heating and cooling execution module: Equipped with an electromagnetic induction heating device and a microchannel liquid cooling device, and introducing intelligent phase change materials. When the mold temperature exceeds the threshold, the phase change material absorbs heat and undergoes a phase change. When the temperature is lower than the threshold, the phase change material releases heat. In the temperature acquisition module, self-calibration technology is adopted to calibrate the sensor regularly. A calibration algorithm based on Kalman filtering is used, and quantum calibration technology is introduced. A calibration method based on quantum bit error correction code is adopted to correct quantum noise and errors in the sensor measurement process. An adaptive calibration strategy is used to dynamically adjust the calibration period and calibration parameters according to the usage time and working environment factors of the sensor. Through the calibration error evaluation formula evaluate the calibration effect, is the true temperature value, is the temperature value measured by the sensor, is the number of calibrations, is the weight determined according to the calibration time and environmental conditions; In the data transmission module, data encryption technology is introduced to encrypt the transmitted data. The quantum encryption algorithm is adopted and combined with the homomorphic encryption technology. A hybrid encryption mechanism based on blockchain and quantum key distribution is used, which combines the distributed ledger of blockchain and the uncrackability of quantum keys. Through the data encryption security evaluation formula Evaluate the security of data encryption, where is the amount of securely transmitted data, is the total amount of data, is the probability that the encrypted data is attempted to be cracked for the th time, and is the total number of encrypted data transmissions; In the data analysis and processing module, transfer learning technology is introduced, and cross-domain transfer learning is carried out by using the temperature data and processing experience of the mold; the temperature control data and processing methods in the aerospace and automotive manufacturing fields are transferred to the temperature control system of the aluminum alloy sheet stamping die to broaden the data source and processing ideas; a transfer learning method based on meta-learning is adopted to learn the characteristics and processing modes of data in different fields; a transfer learning effect evaluation formula is used to evaluate the effect of transfer learning, where is the improvement value of the th data analysis index after transfer learning, is the value of the th data analysis index before transfer learning, n is the number of data analysis indexes, and In the control decision-making module, a multi-objective optimization algorithm is adopted to ensure the stable temperature of the mold, taking into account both the quality of stamping forming and production efficiency; the non-dominated sorting genetic algorithm is used for optimization, and the quantum multi-objective optimization algorithm is introduced; considering the service life and energy consumption of the mold, a multi-objective collaborative optimization model is established, and a multi-objective optimization effect evaluation formula is adopted Evaluate the effect of multi-objective optimization is the weight of the j-th objective is the optimization effect index of the j-th objective is the collaborative correlation strength between the j-th objective and other objectives, and m is the number of objectives 2. The intelligent temperature control system of the aluminum alloy sheet stamping die according to claim 1, characterized in that, It also includes: Mold status detection module: Monitor the pressure change during the stamping process through a pressure sensor, monitor the vibration of the mold using a vibration sensor, adopt an anomaly detection algorithm based on wavelet transform, and introduce an anomaly perception technology based on quantum entanglement states; use a multi-sensor data fusion algorithm to fuse and process the data of pressure, vibration, and quantum entanglement state perception. Adopt anomaly detection accuracy evaluation formula Evaluate the anomaly detection performance, is the number of correctly detected anomalies, is the actual number of anomalies, It is The delay time of anomaly detection, is the total monitoring time.

3. The intelligent temperature control system of the aluminum alloy sheet stamping die according to claim 1, characterized in that, It also includes: Fault diagnosis and early warning module: Using the fault tree analysis method and Bayesian network, analyze the operation data of each module, introduce the autoencoder of deep learning for fault feature extraction; adopt the fault reasoning mechanism based on the knowledge graph, structurally represent and associate fault knowledge to achieve fault diagnosis and reasoning; when potential faults are detected, display the fault location and maintenance guidance to the operator through virtual reality and augmented reality technologies; adopt the fault prediction accuracy evaluation formula Evaluate the fault prediction effect, is the number of faults correctly predicted, is the number of faults actually occurred, is the deviation between the j-th fault prediction time and the actual fault occurrence time, is the total fault monitoring time.

4. The intelligent temperature control system of the aluminum alloy sheet stamping die according to claim 1, characterized in that, In the heating and cooling execution module, an intelligent flow regulating valve and a variable frequency heating power supply are adopted to control the coolant flow rate and heating power. The traditional PID control algorithm is used, the model predictive control algorithm is introduced, and combined with the adaptive control strategy, the parameters of the control algorithm are automatically adjusted according to the real-time state of the mold and environmental changes. Through the control accuracy evaluation formula Evaluate the control accuracy, is the set control parameter value, is the actual control parameter value, is the number of control times, is the weight determined according to the control time and importance.

5. The intelligent temperature control system of the aluminum alloy sheet stamping mold according to claim 2, wherein, In the mold state detection module, a machine learning algorithm is used to classify and predict pressure and vibration data. The method of combining support vector machine and neural network is used to introduce deep learning convolutional neural network and recurrent neural network. The method based on ensemble learning is used to integrate the results of multiple machine learning algorithms and evaluate the classification accuracy through the formula. Evaluate the accuracy of classification, TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, and FN is the number of false negative examples. It is The confidence deviation of the sub-classification, is the total number of classifications.

6. The intelligent temperature control system of the aluminum alloy sheet stamping die according to claim 3, characterized in that, In the fault diagnosis and early warning module, a method combining an expert system and machine learning is used for fault diagnosis. The expert system conducts a preliminary diagnosis according to pre-set rules and experience, and the machine learning algorithm learns and analyzes the fault data; introduce a knowledge graph deep learning fault diagnosis model, adopt a fault diagnosis method based on case-based reasoning, and use historical fault cases for analogical reasoning to locate and solve newly emerging faults. Adopt the evaluation formula for the accuracy rate of fault diagnosis Evaluate the effect of fault diagnosis, is the number of faults correctly diagnosed, is the total number of diagnoses, is the deviation between the fault diagnosis time of the jth time and the actual fault occurrence time, is the total fault diagnosis time.

Images