Injection mold temperature monitoring system using internet of things sensors

Abstract

This invention belongs to the field of intelligent sensors and industrial Internet of Things (IoT) technology. Specifically, it discloses an injection mold temperature monitoring system using IoT sensors. This system includes point-type temperature sensors deployed in key areas of the mold, an infrared thermal imaging device for scanning the cavity surface after mold opening, a data processing server, a machine learning modeling unit, and a mold temperature control interface unit. By fusing multi-source temperature data to construct a mapping model between temperature characteristics and product quality indicators, it predicts in real-time the quality status of the molded part, such as dimensional deviations, warpage, and shrinkage marks, and dynamically adjusts the mold temperature controller settings based on the prediction results. Through the above technical solution, this invention achieves high-dimensional perception of the mold temperature field, real-time prediction of product quality, and closed-loop optimization of mold temperature control, improving the stability, consistency, and intelligence level of the injection molding process.

Description

Technical Field

[0001] This invention belongs to the field of intelligent sensor and industrial Internet of Things technology, specifically relating to an injection mold temperature monitoring system using Internet of Things sensors. Background Technology

[0002] Injection molding, as a core process in modern manufacturing, is widely used in the automotive, electronics, and precision instrument industries. Its production efficiency and molding quality are directly affected by the condition of the mold. With the deep integration of intelligent manufacturing and industrial Internet of Things (IoT) technologies, real-time monitoring of the injection molding process through sensors has become a key link in ensuring process stability and improving product yield. Temperature, as the most sensitive physical parameter in injection molding, directly affects the melt flowability, cooling rate, and the physical and mechanical properties of the final molded part.

[0003] Mold temperature monitoring systems aim to collect data on the thermal cycle in real time through sensing units placed inside or around the mold, providing a basis for process optimization decisions. The core of this technology lies in revealing the thermodynamic evolution laws inside the mold cavity through precise capture of temperature characteristics, establishing a correlation mapping between physical parameters and molding quality in the complex injection molding cycle, and realizing digital characterization and automated control of the production process.

[0004] Existing temperature control solutions are mostly limited to monitoring the outlet temperature of the mold temperature controller, making it difficult to accurately reflect the true temperature field distribution and uniformity inside the mold. This results in quality defects caused by localized overheating or insufficient cooling being difficult to identify in advance. Traditional point-based temperature sensor layouts can only acquire limited local information, failing to comprehensively capture the spatiotemporal thermal characteristics changes within complex cavities, leading to one-sided data representation. Furthermore, due to the complex high-dimensional nonlinear coupling between injection molding process parameters and product quality indicators, existing analytical models and manual experience-based judgments are insufficient for accurate early warning of defects such as warpage and shrinkage marks. Moreover, the quality inspection process often exhibits lag, resulting in a lack of real-time feedback and correction capabilities in the production system, thus hindering yield control and the level of intelligence in injection molding production. Summary of the Invention

[0005] The purpose of this invention is to provide an injection mold temperature monitoring system using Internet of Things (IoT) sensors, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an injection mold temperature monitoring system employing an Internet of Things (IoT) sensor, comprising: A temperature sensing device is configured to deploy multiple point temperature sensors in key areas of the injection mold to collect temperature data of local locations inside the mold in real time. The key areas include the gate, the distal flow channel, the thin-walled area, the thick-walled area, and the inlet and outlet of the cooling water channel. An infrared thermal imaging device is configured to perform non-contact thermal imaging scanning on the surface of the mold cavity after each mold opening action to obtain a two-dimensional temperature field distribution image of the mold surface, so as to supplement the spatial information loss of point temperature sensing. A data processing server is configured to receive and synchronize multi-source temperature data from the temperature sensing device and the infrared thermal imaging device, and to clean, align and extract features from the raw data based on a preset time window. A machine learning modeling unit is configured to build a prediction model based on extracted temperature features. The prediction model can map temperature field features to product quality indicators, including dimensional deviations, weight fluctuations, warpage, and shrinkage grade of the molded part. The mold temperature control interface unit is configured to compare the quality prediction results output by the machine learning modeling unit with the preset quality target, and generate a feedback signal accordingly to send to the external mold temperature controller control system to dynamically adjust the medium set temperature of the mold temperature controller.

[0007] Preferably, the temperature sensing device includes a physical layer sensing component and a signal conditioning layer component; The physical layer sensing component includes multiple discrete detection points distributed inside the moving mold and the fixed mold of the injection mold. The discrete detection points are embedded to a predetermined depth from the surface of the cavity through a precision drilling process. The point temperature sensor is a thermocouple sensor or a platinum resistance temperature sensor; at the gate position, the point temperature sensor is configured to monitor the instantaneous thermal shock of the melt entering the cavity in the early stage, and capture the temperature jump at the moment of injection by setting the sampling frequency. In the thin-walled region and the thick-walled region, the point temperature sensor is configured to monitor the inconsistency in cooling rate caused by the difference in wall thickness; The signal conditioning layer component includes a multi-channel synchronous sampling circuit, which is responsible for amplifying and filtering the millivolt-level analog voltage signal generated by the sensor, and converting it into a digital sequence signal through an analog-to-digital converter circuit. Each node of the physical layer sensing component is encapsulated in a metal armored sleeve, the outer surface of which is coated with a ceramic material with high thermal conductivity. The sensor leads are made of heat-resistant compensation wires with a multi-layer shielding structure to resist electromagnetic pulse interference generated by the heating coil of the injection molding machine.

[0008] Preferably, the infrared thermal imaging device is physically deployed on a side-fixed observation frame during the mold opening stroke of the injection molding machine; The infrared thermal imaging device has a built-in uncooled focal plane array detector, which has long-wave infrared sensing capability and a preset pixel resolution. The infrared thermal imaging device communicates bidirectionally with the data processing server through an industrial communication interface. When the mold opening stroke of the injection molding machine reaches the preset end position, the infrared thermal imaging device is configured to receive a trigger pulse and perform a single or multiple shutter scans during the static window period before the mold cavity is fully exposed and the ejection mechanism is activated. The infrared thermal imaging device also includes an ambient temperature compensation module, which reads the ambient temperature parameters of the workshop in real time and corrects the radiation brightness of the acquired raw thermal image to eliminate temperature measurement errors caused by background radiation. The two-dimensional temperature field distribution image is stored in matrix form, and each matrix element corresponds to the temperature value of the geometric coordinates of the mold surface. The infrared thermal imaging device also integrates an image dust removal algorithm, which uses an airflow curtain to prevent workshop dust from contaminating the lens, and uses software algorithms to identify and filter out false hot spots caused by residual mold release agent on the mold surface.

[0009] Preferably, the data processing server has a computing core and a high-speed cache, and is equipped with a real-time data synchronization engine; The data synchronization engine precisely aligns the one-dimensional time-series temperature curve from the temperature sensing device with the two-dimensional transient thermal image from the infrared thermal imaging device in the time dimension, based on the time sequence label of the injection molding cycle. The data processing server is equipped with a feature extraction module, which calculates the peak temperature value, the integral value of the temperature evolution curve over time, the average cooling rate during the cooling stage, and the time to reach the peak temperature for one-dimensional time series data in each sampling period. For two-dimensional thermal image data, the feature extraction module uses an image segmentation algorithm to extract the temperature gradient distribution map of the mold surface, identify and calculate the area ratio of high-temperature accumulation areas and the degree of deviation of local hot spots from the overall average temperature; All extracted feature vectors are uniformly integrated into a high-dimensional feature tensor, which serves as the input benchmark for subsequent prediction models. The data processing server also has redundancy and fault tolerance capabilities. When a processing node experiences a hardware failure, its load will be automatically migrated to a backup node. It also supports long-term archiving of historical data and uses a time-series database to store temperature fluctuation data.

[0010] Preferably, the machine learning modeling unit is embedded with a pre-trained deep learning logic framework, which includes a feature mapping subunit and an adaptive weight adjustment subunit. The feature mapping subunit adopts a random forest algorithm or a multilayer perceptron neural network architecture and is configured to learn the nonlinear relationship between the intramolecular thermal field and product quality. During the offline training phase, the machine learning modeling unit optimizes its parameters by collecting data samples containing multi-source temperature features and corresponding measured quality indicators. During the online inference phase, the machine learning modeling unit will input the feature tensors extracted in real time into the model and output the predicted values ​​of the product quality indicators produced in the current cycle. The machine learning modeling unit also integrates a confidence assessment mechanism. When the confidence level of the predicted value is lower than a preset threshold, the system automatically triggers a manual intervention request and marks the batch of products as pending inspection. The machine learning modeling unit also integrates a transfer learning module. When the mold is changed or the batch of injection molding materials is changed, the transfer learning module uses a small amount of labeled data to fine-tune the parameters of the original model.

[0011] Preferably, the machine learning modeling unit is also configured with an anomaly detection mechanism, which uses a two-level control graph logic for real-time monitoring; The first level is a single-point deviation monitoring logic. When the temperature value of any measuring point in the temperature sensing device exceeds the preset statistical control upper or lower limit, the system determines that the local thermal balance is disrupted. The second level is the overall thermal field uniformity monitoring logic. By calculating the standard deviation of the temperature in each region of the infrared thermal image, when the standard deviation exceeds the preset uniformity tolerance, it is determined that the global cooling system has failed. This anomaly detection mechanism can identify thermodynamic trend deviations in the injection molding process before quality defects occur, and send a warning pulse to the mold temperature control interface unit. The anomaly detection mechanism also includes occlusion recognition logic. If a foreign object is detected obstructing the infrared field of view, the system will automatically discard the heat map data of the current period and replace it with the smoothed value of the adjacent period.

[0012] Preferably, the mold temperature control interface unit has standard industrial bus communication capabilities and supports multiple communication protocols; The mold temperature control interface unit internally stores a compensation algorithm library based on inversion logic; When the quality index predicted by the machine learning modeling unit shows a continuous deviation trend and reaches the tolerance boundary, the mold temperature control interface unit calculates the required medium temperature correction amount according to the inversion logic. The inversion logic is based on a preset sensitivity matrix, which describes the coupling relationship between the temperature setting of the mold temperature controller and the temperature changes of the characteristic temperature inside the mold. The generated feedback signal includes a new mold temperature setting recommendation, a heating or cooling rate requirement, and a duration command. The feedback signal is pushed to the controller of an external mold temperature controller to achieve closed-loop fine adjustment of the temperature of the mold cooling water or heating medium. The mold temperature control interface unit also has an energy efficiency monitoring function. Under the premise of ensuring product quality meets the standards, it can achieve green manufacturing by optimizing the output power of the mold temperature controller and reducing the power consumption of the chiller unit.

[0013] Preferably, the system further includes a process parameter association module, which is configured to acquire non-temperature process parameters in real time through the bus interface of the injection molding machine, including peak injection pressure, holding pressure curve, holding time length, and set temperatures of each section of the melt tube. The process parameter association module uses these non-temperature parameters as auxiliary feature variables and performs feature fusion with the temperature features provided by the temperature sensing device and the infrared thermal imaging device. By adding a non-temperature physical dimension, the machine learning modeling unit can more comprehensively perceive the energy balance and pressure balance in the injection molding process, thereby improving the generalization ability and prediction robustness of the quality prediction model when faced with different batches of raw materials or environmental temperature and humidity fluctuations.

[0014] Preferably, the data processing server and the machine learning modeling unit are integrated into an industrial edge computing all-in-one machine deployed next to the injection molding machine; The edge computing all-in-one machine is responsible for localized data processing and rapid feedback, and controls the entire process from sensor data collection to generating mold temperature control recommendations within a preset real-time period. The edge computing all-in-one machine is connected to the injection molding machine, mold temperature controller and various sensing units via shielded twisted pair cables. It has electromagnetic compatibility performance and can operate in environments with high-frequency injection molding machine motor interference. The system also includes a field gateway layer, which is responsible for aggregating and converting the data of all injection molding machines in the workshop and running real-time anomaly warning logic. The field gateway layer is responsible for binding the physical identity of the mold with the sensor data, and the thermal field record can be traced back to the specific mold number, machine position and maintenance history.

[0015] Preferably, the system further includes an ultra-sensitive thermal sensing network and an execution feedback interface; The ultra-sensitive thermal sensing network uses a detection unit based on fiber Bragg grating sensing technology to sense temperature by shifting the wavelength of light. It is deployed in the precision structure of the mold to sense localized minute thermal disturbances. The execution feedback interface is used to control an array of electric heating rods or a variable frequency electromagnetic induction heating device with millisecond-level response capability, and to precisely control the amount of heat compensation in the micro-region of the mold through pulse width modulation signal. When it is predicted that overheating will occur in a local area due to melt shearing, the execution feedback interface immediately cuts off the heating power supply at the corresponding location and triggers the micro-spray cooling mechanism to achieve isothermal molding control. The system also includes a cloud-based big data analysis platform that stores thermal data records of the entire life cycle of molds in the factory. The cloud-based big data analysis platform has a thermal field aging trend analysis function. By comparing the thermal characteristic drift of the same mold at different time periods, it can predict the degree of scaling in the internal water channels of the mold or the heating failure trend of the hot runner.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The injection mold temperature monitoring system using IoT sensors provided by this invention achieves multi-dimensional and high-resolution perception of the temperature field inside and on the surface of the mold by integrating point temperature sensing and infrared thermal imaging technology, overcoming the problem of insufficient thermal field characterization caused by traditional reliance on mold temperature controller outlet temperature or single measurement point data. 2. By constructing a machine learning-based "temperature feature-product quality" mapping model, key product quality indicators can be predicted in real time during the injection molding cycle, shortening the quality feedback delay and transforming process adjustment from "post-event correction" to "in-process intervention". 3. The system uses the mold temperature control interface unit to correlate the quality prediction results with the mold temperature controller parameters, forming a closed-loop control mechanism oriented towards product quality. This improves the stability, consistency, and intelligence of injection molding production, reduces the number of trial moldings and the scrap rate, and provides technical support for precision injection molding manufacturing. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention; Figure 2 This is a logical flowchart of the present invention based on the fusion of multi-source data from point temperature sensing and infrared thermal imaging and feature extraction. Figure 3 This is a schematic diagram of the core principle framework of the mapping from temperature field features to product quality indicators based on machine learning in this invention; Figure 4 This is a flowchart illustrating the anomaly detection mechanism combining single-point deviation and overall thermal field uniformity in this invention. Figure 5 This is a schematic diagram of the multi-level interaction and data flow between the prediction model output, quality target comparison, and external mold temperature controller in this invention. Detailed Implementation

[0018] Example 1: Reference Figures 1 to 5 The injection mold temperature monitoring system using IoT sensors includes a temperature sensing device, an infrared thermal imaging device, a data processing server, a machine learning modeling unit, and a mold temperature control interface unit. The temperature sensing device is used to deploy multiple point temperature sensors in the key areas of the injection mold to collect temperature data of local locations inside the mold in real time. The key areas include the gate, the far-end flow channel, the thin-walled area, the thick-walled area, and the inlet and outlet of the cooling water channel. The infrared thermal imaging device is used to perform non-contact thermal imaging scanning on the surface of the mold cavity after each mold opening action to obtain a two-dimensional temperature field distribution image of the mold surface, in order to supplement the information missing in the spatial dimension of point temperature sensing. The data processing server is used to receive and synchronize multi-source temperature data from the temperature sensing device and the infrared thermal imaging device, and to clean, align and extract features from the raw data based on a preset time window. The machine learning modeling unit is used to build a prediction model based on the extracted temperature features. The temperature feature prediction model can map temperature field features to product quality indicators, including the dimensional deviation, weight fluctuation, warpage degree and shrinkage grade of the molded part. The mold temperature control interface unit is used to compare the quality prediction results output by the machine learning modeling unit with the preset quality target, and generate a feedback signal accordingly to send to the external mold temperature controller control system to dynamically adjust the medium setting temperature of the mold temperature controller.

[0019] The temperature sensing device comprises a physical layer sensing component and a signal conditioning layer component. The physical layer sensing component includes multiple discrete detection points distributed within the moving and fixed molds of the injection mold. These points are embedded to a specific depth of 2 to 5 millimeters from the cavity surface using a precision drilling process. The point-type temperature sensor employs an industry-standard thermocouple sensor or a platinum resistance temperature sensor.

[0020] At the gate, the point temperature sensor is configured to monitor the instantaneous thermal shock of the melt entering the cavity in the early stage, and its real-time acquisition frequency is set to 100 times per second to capture the temperature jump at the moment of injection.

[0021] In both thin-walled and thick-walled regions, the point-type temperature sensor is configured to monitor inconsistent cooling rates due to differences in wall thickness. The signal conditioning layer assembly includes a multi-channel synchronous sampling circuit responsible for amplifying, filtering, and denoising the millivolt-level analog voltage signal generated by the sensor, and then converting it into a digital signal via an analog-to-digital converter.

[0022] The infrared thermal imaging device is physically deployed on a fixed observation frame on the side of the injection molding machine during the mold opening stroke. The device incorporates an uncooled focal plane array detector with long-wave infrared sensing capability and a pixel resolution of at least 640×480 pixels. The infrared thermal imaging device communicates bidirectionally with the data processing server via an industrial Ethernet interface.

[0023] When the injection molding machine's mold opening stroke reaches the preset end-point limit switch, the infrared thermal imaging device is configured to receive a trigger pulse and perform a single or multiple high-speed shutter scans within a static window period before the mold cavity is fully exposed and the ejection mechanism actuates. The infrared thermal imaging device also includes an ambient temperature compensation module, which corrects the radiance of the acquired raw thermal image by real-time reading of the workshop's ambient temperature parameters, eliminating temperature measurement errors caused by background radiation. The resulting two-dimensional temperature field distribution image is stored in matrix form, with each pixel corresponding to the temperature value of the mold surface's geometric coordinates.

[0024] The data processing server, equipped with a high-performance multi-core computing core and a large-capacity high-speed cache, is configured to run a real-time data synchronization engine. Based on the time-series labels of the injection molding cycle, the data synchronization engine precisely aligns the one-dimensional time-series temperature curve from the temperature sensing device with the two-dimensional transient thermal image from the infrared thermal imaging device in terms of time dimension.

[0025] The feature extraction module is configured to perform deep data mining tasks. For one-dimensional time series data, it calculates key parameters such as peak temperature values, integral values ​​of the temperature evolution curve over time, the slope of the cooling stage (i.e., the average cooling rate), and the time to reach peak temperature for each sampling period. For two-dimensional thermal image data, the feature extraction module uses an image segmentation algorithm to extract the temperature gradient distribution map of the mold surface, identify and calculate the area ratio of high-temperature clusters and the deviation of local hotspots from the overall average temperature. All extracted feature vectors are uniformly integrated into a high-dimensional feature tensor, which serves as the input benchmark for subsequent prediction models.

[0026] The machine learning modeling unit embeds a pre-trained deep learning logic framework. This anomaly detection mechanism includes a feature mapping subunit and an adaptive weight adjustment subunit. The feature mapping subunit employs a random forest algorithm or a multilayer perceptron neural network architecture and is configured to learn a complex nonlinear relationship between "intra-mold thermal field and product quality." During offline training, the modeling unit optimizes its parameters by collecting tens of thousands of data samples containing multi-source temperature features and corresponding laboratory-measured quality indicators (such as dimensions measured by a micrometer, weight measured by an electronic balance, and warpage acquired by an optical image measuring instrument).

[0027] During the online inference phase, the machine learning modeling unit inputs the feature tensors extracted in real time into the model and outputs predicted values ​​of product quality indicators for the current production cycle. The modeling unit also integrates a confidence assessment mechanism; when the confidence level of the predicted value falls below a preset threshold, the system automatically triggers a manual intervention request and marks the batch of products as pending inspection.

[0028] The machine learning modeling unit is also equipped with an anomaly detection mechanism. This mechanism employs a two-level control chart logic for real-time monitoring. The first level is a single-point deviation monitoring logic; when the temperature value at any measuring point in the temperature sensing device exceeds a preset statistical control upper or lower limit, the system determines it as a local thermal equilibrium disruption. The second level is an overall thermal field uniformity monitoring logic; by calculating the standard deviation of the temperature in each region of the infrared thermal image, when the standard deviation exceeds a preset uniformity tolerance, it is determined as a global cooling system failure. The anomaly detection mechanism can identify thermodynamic trend deviations in the injection molding process before quality defects occur and send warning pulses to the mold temperature control interface unit.

[0029] The mold temperature control interface unit possesses standard industrial bus communication capabilities, supporting multiple protocols including Modbus-TCP, Profinet, and EtherCAT. The interface unit internally stores a compensation algorithm library based on inversion logic. When the dimensional deviation or warpage predicted by the machine learning modeling unit shows a continuously increasing trend and reaches the tolerance boundary, the interface unit calculates the required medium temperature correction amount according to the inversion logic. This inversion logic is based on a preset sensitivity matrix, which describes the coupling relationship between changes in the mold temperature controller's set temperature and changes in the characteristic temperature within the mold. The generated feedback signals include new mold temperature setting recommendations, heating or cooling rate requirements, and duration commands. These signals are pushed to the programmable logic controller (PLC) of the external mold temperature controller to achieve closed-loop fine-tuning of the mold cooling water or heating medium temperature.

[0030] The data processing server also integrates a process parameter correlation module. This module is configured to acquire non-temperature-related process parameters in real time via the injection molding machine's bus interface, including peak injection pressure, holding pressure curve, holding time, and set temperatures for each section of the molten material tube. The module uses these non-temperature parameters as auxiliary feature variables and fuses them with the aforementioned temperature features. By adding these physical dimensions, the machine learning modeling unit can more comprehensively perceive the energy and pressure balance during the injection molding process, improving the generalization ability and predictive robustness of the quality prediction model when faced with fluctuations in raw material batches or environmental temperature and humidity.

[0031] Furthermore, the imaging timing of the infrared thermal imaging device is strictly constrained by the injection molding machine's action signals. The system precisely defines a static imaging window of 0.5 to 2 seconds by monitoring the level changes between the "mold opening complete" and "ejection start" signals of the injection molding machine. During this period, the mold cavity is in a static, exposed state and is not subject to mechanical vibration interference from the ejector rod's movement. The infrared thermal imaging device performs non-contact temperature measurement within this window, minimizing temperature measurement distortion caused by motion blur. Simultaneously, the system also features occlusion recognition logic. If a foreign object or operator's limb is detected obstructing the infrared field of view, the system automatically discards the thermal image data for the current cycle and replaces it with a smoothed value from the adjacent cycle, ensuring that the data source input to the machine learning modeling unit always maintains a high degree of authenticity and consistency.

[0032] The overall physical deployment of the system adopts a combination of edge computing and fieldbus. The data processing server and machine learning modeling unit are integrated into an industrial edge computing all-in-one machine deployed next to the injection molding machine. This deployment method ensures localized processing and rapid feedback of all data, controlling the entire process from sensor data acquisition to generating mold temperature control recommendations to within 0.5 seconds, fully meeting the millisecond-level response requirements of the injection molding production line. The edge computing all-in-one machine is connected to the injection molding machine, mold temperature controller, and various sensing units via shielded twisted-pair cables, possessing strong electromagnetic compatibility and enabling stable operation in environments with high-frequency injection molding machine motor interference.

[0033] The injection mold temperature monitoring system using IoT sensors provided in this embodiment achieves multi-dimensional, high-resolution perception of the temperature field inside and on the surface of the mold by integrating point-type temperature sensing and infrared thermal imaging technology. This overcomes the problem of insufficient thermal field characterization caused by traditional methods that rely solely on the outlet temperature of the mold temperature controller or data from a single measuring point. By constructing a machine learning-based "temperature feature-product quality" mapping model, the system can predict key product quality indicators in real time during the injection molding cycle, shortening the quality feedback delay and transforming process adjustments from "post-event correction" to "in-process intervention." The system uses a mold temperature control interface unit to correlate the quality prediction results back to the mold temperature controller control parameters, forming a closed-loop control mechanism oriented towards product quality. This effectively improves the stability, consistency, and intelligence level of injection molding production, reduces the number of trial moldings and the scrap rate, and provides reliable technical support for precision injection molding manufacturing.

[0034] Example 2: This example describes a variant of an injection mold temperature monitoring system based on a distributed Internet of Things architecture, which aims to solve the problems of high-concurrency data processing and refined temperature control in complex working conditions of ultra-large multi-cavity molds.

[0035] The injection mold temperature monitoring system using IoT sensors includes a distributed temperature sensor array, a multi-angle infrared scanning unit, a clustered edge processing server, an enhanced machine learning unit, and a multi-channel mold temperature linkage interface unit. The distributed temperature sensing array comprises dozens or even hundreds of intelligent temperature sensing nodes conforming to Industrial Internet of Things (IIoT) standards. These nodes are distributed at key points in each sub-cavity of the mold. Each sensor node is equipped with an independent micro data preprocessing chip, capable of performing preliminary linear compensation and unit conversion before data upload. The sensing array is networked via a high-bandwidth industrial bus topology, ensuring comprehensive thermal data coverage within the complex interior of large molds.

[0036] The multi-angle infrared scanning unit consists of multiple sets of infrared temperature measurement modules installed at different locations. For large molds with deep cavity structures or complex curved surfaces, a single-view infrared thermal image is insufficient to cover all surfaces. The multi-angle infrared scanning unit, through multi-view vision stitching technology, is configured to acquire the complete thermal distribution of the mold cavity from multiple dimensions, such as left, right, top, and bottom. Each infrared module includes an autofocus function, dynamically adjusting the focal length according to the mold's opening position to ensure the acquisition of a clear image of the temperature distribution characteristics.

[0037] The clustered edge processing server comprises multiple parallel-running processing blades, specifically designed to handle the computational pressure from ultra-large-scale sensor data. Internally, the server runs a distributed computing engine configured to break down massive temperature matrix operations into multiple subtasks. In the feature extraction stage, the server is configured to calculate thermal consistency characteristics between cavities, such as the temperature difference between adjacent cavities at the same time and the cross-correlation coefficients of cooling curves from different cavities. These characteristics are crucial for identifying single-cavity quality failures caused by cooling channel blockage.

[0038] The enhanced machine learning unit, based on the original random forest model, introduces a spatiotemporal graph convolutional network architecture. This architecture treats each measuring point within the mold as a node in a graph structure, and the heat conduction paths between these points as edges. The machine learning unit is configured to extract the evolution characteristics of the temperature field over time and its propagation characteristics in three-dimensional space. Through this graph neural network model, the system can identify minute heat conduction hysteresis phenomena and correlate them with the residual stress distribution within the product. The prediction result is no longer just a single quality value, but rather a distribution map containing the probability of quality risks in various parts of the product.

[0039] The multi-channel mold temperature linkage interface unit is configured to interface with an advanced mold temperature controller with multi-loop control capabilities. For different mold cavities or different areas of the mold, the interface unit can output independent temperature control commands. For example, when a persistently high temperature is detected in the left side of the fixed mold area, the interface unit sends a control signal to the corresponding branch of the mold temperature controller to increase the cooling medium flow rate or decrease the branch's set temperature, achieving true on-demand local thermal balance control.

[0040] Each node of the distributed temperature sensing array is encapsulated in a high-temperature, high-pressure resistant metal armored sleeve. The outer surface of the sleeve is coated with a ceramic material with high thermal conductivity to reduce the impact of the sensor's own thermal inertia on the response speed. The sensor leads use heat-resistant compensation wires with a multi-layer shielding structure to resist electromagnetic pulse interference generated by the high-power heating coil of the injection molding machine.

[0041] The multi-angle infrared scanning unit has a built-in automatic shutter calibration mechanism that automatically performs blackbody calibration on the detector between two injection molding cycles to ensure absolute accuracy of temperature measurement. The infrared scanning unit also integrates an image dust removal algorithm, which uses an airflow curtain to prevent workshop dust from contaminating the lens, and uses software algorithms to identify and filter out false hot spots caused by residual mold release agent on the mold surface.

[0042] The clustered edge processing server features redundancy and fault tolerance. When a processing node experiences a hardware failure, its load is automatically migrated to a backup node, ensuring uninterrupted operation of the monitoring system during 24-hour continuous production. The server also supports long-term archiving of historical data, utilizing a time-series database to store massive amounts of temperature fluctuation data, providing a digital twin foundation for subsequent process traceability and long-term yield analysis.

[0043] The enhanced machine learning unit also integrates a transfer learning module. When the mold is changed or the batch of injection molding materials is altered, this transfer learning module can quickly fine-tune the original model using a small amount of labeled data, without having to train from scratch. This improves the system's adaptability in multi-variety, small-batch production environments.

[0044] The multi-channel mold temperature linkage interface unit also features energy efficiency monitoring. While ensuring product quality meets standards, the interface unit is configured to optimize the output power of the mold temperature controller. For example, when the predictive model indicates sufficient temperature margin, the system will automatically and appropriately increase the cooling water set temperature to reduce the power consumption of the chiller unit, achieving green and intelligent manufacturing.

[0045] The system architecture in this embodiment enhances the system's ability to characterize complex, large molds through a distributed hardware and software design. The combination of multi-angle infrared sensing and graph neural networks enables temperature control accuracy to evolve from the "whole mold level" to the "mold cavity level" and even the "local feature level." This refined control method has a decisive technical advantage for producing high-precision, high-performance injection molded parts, while also enhancing the system's self-healing ability and robustness in the face of complex process fluctuations.

[0046] Example 3: This example focuses on describing the implementation of an ultra-high precision mold temperature monitoring system for the injection molding of precision medical devices. Its core lies in the ultimate capture and closed-loop feedback of minute temperature fluctuations and their dynamic evolution.

[0047] The temperature monitoring system for injection molds using IoT sensors includes an ultra-sensitive thermal sensing network, a high-speed dynamic infrared monitoring unit, a collaborative data fusion engine, a deep reinforcement learning control unit, and an execution feedback interface. The ultra-sensitive thermal sensing network employs a detection unit based on fiber Bragg grating sensing technology. Unlike traditional electrical signal sensors, fiber optic sensors sense temperature through wavelength shifts, exhibiting high measurement sensitivity and zero electromagnetic interference. By embedding hair-thin fiber optic sensing heads within the micron-level precision structure of a mold, the system can detect minute local thermal disturbances with a resolution of 0.01 degrees Celsius. This precision is crucial for controlling thermal shrinkage during the injection molding of precision lenses or microfluidic chips.

[0048] The high-speed dynamic infrared monitoring unit employs a high-frame-rate long-wave infrared camera, achieving a frame rate of up to 500 frames per second. Compared to ordinary static scanning, this high-speed dynamic infrared monitoring unit can record the complete dynamic process of heat dissipation from the mold surface at the moment of mold opening. The unit is configured to capture the non-steady-state changes in mold surface temperature within hundreds of milliseconds. These dynamic characteristics include information on the thermal diffusivity of the mold material and the degree of scaling in the internal cooling channels, which are key dimensions for establishing ultra-high-precision prediction models.

[0049] The collaborative data fusion engine runs on a hardware-accelerated field-programmable gate array (FPGA) architecture. The engine is configured to perform sub-millisecond data stream alignment. For high-frequency point data generated by fiber optic sensors and high-speed surface data generated by infrared scanning, the engine employs wavelet transform algorithms to extract thermal features of different frequency bands. The engine also integrates a thermal stress simulation calculation module, which uses real-time acquired temperature boundary conditions to synchronously run a thermoelastic mechanical model of the mold in memory, deriving the minute thermal deformation of the cavity wall under injection pressure.

[0050] The deep reinforcement learning control unit replaces the traditional static prediction logic. This control unit treats the injection molding process as a continuous decision-making Markov process. The control unit is configured to use current multi-source temperature and pressure characteristics, as well as historical production data, as state inputs, and generates a temperature control decision strategy through a deep Q-network algorithm. This temperature control decision strategy not only considers the quality prediction of the current cycle but also aims to optimize the stability of the entire production sequence. When the system detects a slight trend in the temperature field deviating in an unfavorable direction, the reinforcement learning algorithm issues a fine-tuning instruction in advance, achieving dynamic hedging of the process state through a "small steps, quick adjustments" strategy.

[0051] The execution feedback interface is specifically designed to control electric heating rod arrays or variable frequency electromagnetic induction heating devices with millisecond-level response capabilities. Compared to traditional medium circulation heating, this direct, localized heating method has low thermal inertia. The interface unit precisely controls the heat compensation amount in micro-regions of the mold through pulse width modulation signals. When the model predicts that a local area is overheating due to melt shearing, the interface unit will immediately cut off the heating power supply at the corresponding location and trigger a micro-spray cooling mechanism to achieve isothermal molding control.

[0052] The fiber optic demodulator of the ultra-sensitive thermal sensing network has multi-channel parallel processing capabilities, enabling simultaneous demodulation of hundreds of wavelength signals. This ensures that high-density measurement points can be deployed inside complex molds without signal delay. The optical fiber uses a special heat-resistant coating, which can withstand repeated periodic high-pressure impacts and high-temperature baking during injection molding.

[0053] The high-speed dynamic infrared monitoring unit is equipped with a high-pressure purge shield to ensure that the lens surface remains optically clean at all times. The monitoring unit also has an automatic background reduction function, which can remove infrared shadows caused by the movement of the injection molding machine's robotic arm in real time, ensuring that the extracted surface thermal field data only reflects the physical state of the mold itself.

[0054] The collaborative data fusion engine incorporates a self-calibration algorithm based on Kalman filtering. This algorithm automatically calculates and updates the thermal conductivity parameter model of the mold material by comparing the internal point temperature measured by the fiber optic sensor with the corresponding surface temperature measured by the infrared camera. This allows the system to automatically adapt to performance drift caused by material fatigue or waterway fouling during the mold's service life.

[0055] The deep reinforcement learning control unit possesses self-evolution capabilities. After each batch of injection molding is completed, the system automatically obtains the final quality inspection result as a reward signal and feeds it back to the control unit. Through this online learning mechanism, the control unit's decision-making model becomes increasingly accurate as the production cycle increases, ultimately reaching an optimal temperature control state that approaches or even surpasses that of human experts.

[0056] The execution feedback interface also features a robust failure protection mechanism. Upon detecting a fiber optic cable breakage or heating element damage, the interface unit immediately enters a safety mode, utilizing backup conventional sensor data to maintain basic production control, and simultaneously sending a detailed hardware fault coordinate diagnostic report to the central control room.

[0057] This embodiment introduces cutting-edge technologies such as fiber optic sensing, high-speed infrared, and reinforcement learning to push injection mold temperature monitoring to an extremely precise level. The system not only achieves accurate prediction of quality indicators but also eliminates microscopic thermal fluctuations that are difficult to overcome in traditional processes through microsecond-level localized thermal intervention, providing a solid technical guarantee for the production of medical, optical, and precision electronic devices with extremely high precision requirements. This all-weather, all-space precision sensing and closed-loop control marks a qualitative leap in injection molding processes, shifting from experience-driven to data-driven and model-driven approaches.

[0058] Example 4: This example illustrates a large-scale injection molding factory-level mold temperature monitoring system that combines IoT edge sensing with cloud collaboration.

[0059] The temperature monitoring system for injection molds using IoT sensors includes an intelligent sensing front-end, a field gateway layer, a cloud-based big data analysis platform, and a remote collaborative control terminal. The intelligent sensing front-end is deployed on every injection molding machine and its associated mold within the factory. Each front-end includes the aforementioned temperature sensing device and infrared thermal imaging device. These front-end components transmit the collected high-frequency thermodynamic data in real-time to the field gateway via low-power wide-area network technology or a 5G industrial dedicated network. The intelligent sensing front-end has a localized initial data compression algorithm, configured to significantly reduce the data traffic uploaded to the cloud while ensuring no feature loss.

[0060] The on-site gateway layer, composed of multiple high-performance edge gateways, is responsible for the initial aggregation and protocol conversion of data from all injection molding machines in the workshop. This gateway layer operates with real-time anomaly warning logic, maintaining basic local alarm functions even if the cloud connection is interrupted. The gateway layer also binds the physical identification of the mold to sensor data, ensuring that each thermal field record can be accurately traced to the specific mold number, injection molding machine location, and corresponding mold maintenance history.

[0061] The cloud-based big data analytics platform is built on a distributed computing cluster. The platform stores lifecycle thermal data records for all molds throughout the factory. The platform's machine learning modeling unit is configured to perform global model training tasks. By aggregating thermal data from different machines and molds producing similar products, the cloud platform can train a global prediction model with stronger generalization performance. The platform also features thermal field aging trend analysis capabilities. By comparing the thermal characteristic drift of the same mold over several months, it predicts the degree of scaling in the mold's internal water channels or the heating failure trend of hot runners, thus enabling a shift from preventative maintenance to predictive maintenance.

[0062] The remote collaborative control terminal is an interactive platform deployed in the factory dispatch center or on an expert's mobile device. The terminal is configured to visually display the thermal health distribution map of the entire workshop. When the cloud platform detects an anomaly in the quality prediction indicators of a particular injection molding machine, the terminal automatically pushes a detailed anomaly diagnostic report, including the in-mold thermal map at the time of the anomaly, a point-based temperature anomaly curve, and suggested process adjustment solutions. Factory technicians can remotely issue parameter correction commands through the terminal. After authorization, the commands are returned to the mold temperature control interface unit for execution via the cloud platform and the field gateway.

[0063] The infrared module of the intelligent sensing front end is configured with a low power consumption mode. When the injection molding machine is in the mold closing or preheating stage, the infrared module is in standby mode; it only wakes up instantaneously and performs imaging acquisition when it detects a mold opening proximity signal. This on-demand wake-up mechanism extends the lifespan of the sensing hardware and reduces the overall power consumption of the system.

[0064] The field gateway layer possesses robust electromagnetic interference isolation capabilities. It employs opto-isolation technology to protect the sensor input interfaces, ensuring that common-mode interference generated when the injection molding machine starts its high-power frequency converter does not affect the accuracy of data acquisition. The gateway also integrates time-sensitive networking technology, ensuring that the timestamp error of all sensor data is within the microsecond range, which is crucial for synchronous analysis of thermal characteristics in multi-machine collaborative operation.

[0065] The cloud-based big data analytics platform stores hundreds of preset models for different materials and mold structures in its predictive model library. When a new mold is launched, the platform automatically matches the closest initial predictive model based on the mold's CAD structural parameters, and continuously iterates the parameters using a self-balancing algorithm as production progresses. This "model-as-a-service" model simplifies the debugging workload for on-site process engineers.

[0066] The remote collaborative control terminal supports multi-dimensional visualization. In addition to conventional temperature curves and two-dimensional thermal maps, the terminal utilizes augmented reality technology to overlay real-time thermal field data onto the three-dimensional model of the mold. This allows maintenance engineers to clearly see which cooling water channel inside the mold has a flow imbalance, enabling precise disassembly and repair.

[0067] This embodiment elevates the intelligent perception of a single machine to a factory-level collaborative intelligence dimension through a cloud-edge collaborative architecture. By accumulating massive amounts of data and optimizing global models in the cloud, the system can detect long-cycle process fluctuations and potential risks that are difficult for a single machine system to capture, thereby improving the overall quality control level and comprehensive equipment efficiency in large-scale injection molding production. This digital management approach lays a crucial foundation for temperature control technology support in realizing a fully automated, lights-out factory.

[0068] In summary, this invention constructs an intelligent mold temperature monitoring system covering the entire injection molding process and possessing self-learning and evolution capabilities through multi-source sensor fusion sensing, machine learning quality mapping, and real-time closed-loop feedback control. This system solves the problems of traditional temperature control methods being lagging, one-sided, and lacking quality orientation, providing a disruptive process control solution for the fields of precision manufacturing and intelligent manufacturing.

[0069] Those skilled in the art should understand that the embodiments described above are merely for illustrating the technical principles and features of the present invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A temperature monitoring system for injection molds using IoT sensors, characterized in that, include: A temperature sensing device is configured to deploy multiple point temperature sensors in key areas of the injection mold to collect temperature data of local locations inside the mold in real time. The key areas include the gate, the distal runner, the thin-walled area, the thick-walled area, and the inlet and outlet of the cooling water channel. An infrared thermal imaging device is configured to perform non-contact thermal imaging scanning on the surface of the mold cavity after each mold opening action to obtain a two-dimensional temperature field distribution image of the mold surface, so as to supplement the spatial information loss of point temperature sensing. A data processing server is configured to receive and synchronize multi-source temperature data from the temperature sensing device and the infrared thermal imaging device, and to clean, align and extract features from the raw data based on a preset time window. The data processing server has a computing core and a high-speed cache, and is equipped with a real-time data synchronization engine; The data synchronization engine precisely aligns the one-dimensional time-series temperature curve from the temperature sensing device with the two-dimensional transient thermal image from the infrared thermal imaging device in terms of time dimension, based on the time sequence label of the injection molding cycle. The data processing server is equipped with a feature extraction module, which calculates the peak temperature value, the integral value of the temperature evolution curve over time, the average cooling rate during the cooling stage, and the time to reach the peak temperature for one-dimensional time series data in each sampling period. For two-dimensional thermal image data, the feature extraction module uses an image segmentation algorithm to extract the temperature gradient distribution map of the mold surface, identify and calculate the area ratio of high-temperature accumulation areas and the degree of deviation of local hot spots from the overall average temperature; All extracted feature vectors are uniformly integrated into a high-dimensional feature tensor, which serves as the input benchmark for subsequent prediction models. The machine learning modeling unit is configured to build a prediction model based on extracted temperature features. The prediction model can map temperature field features to product quality indicators, including dimensional deviation, weight fluctuation, warpage, and shrinkage grade of the molded part. The machine learning modeling unit is embedded with a pre-trained deep learning logic framework, which includes a feature mapping subunit and an adaptive weight adjustment subunit. The feature mapping subunit adopts a random forest algorithm or a multilayer perceptron neural network architecture and is configured to learn the nonlinear relationship between the intramolecular thermal field and product quality. During the offline training phase, the machine learning modeling unit optimizes its parameters by collecting data samples containing multi-source temperature features and corresponding measured quality indicators. During the online inference phase, the machine learning modeling unit will input the feature tensors extracted in real time into the model and output the predicted values ​​of the product quality indicators produced in the current cycle. The mold temperature control interface unit is configured to compare the quality prediction results output by the machine learning modeling unit with the preset quality target, and generate a feedback signal accordingly to send to the external mold temperature controller control system to dynamically adjust the medium set temperature of the mold temperature controller; the mold temperature control interface unit has standard industrial bus communication capabilities and supports multiple communication protocols. The mold temperature control interface unit internally stores a compensation algorithm library based on inversion logic; When the quality index predicted by the machine learning modeling unit shows a continuous deviation trend and reaches the tolerance boundary, the mold temperature control interface unit calculates the required medium temperature correction amount according to the inversion logic. The inversion logic is based on a preset sensitivity matrix, which describes the coupling relationship between the temperature setting of the mold temperature controller and the temperature changes of the characteristic temperature inside the mold. The generated feedback signal includes a new mold temperature setting recommendation, a heating or cooling rate requirement, and a duration command. The feedback signal is pushed to the controller of an external mold temperature controller to achieve closed-loop fine adjustment of the temperature of the mold cooling water or heating medium.

2. The injection mold temperature monitoring system using IoT sensors according to claim 1, characterized in that, The temperature sensing device includes a physical layer sensing component and a signal conditioning layer component; The physical layer sensing component includes multiple discrete detection points distributed inside the moving mold and the fixed mold of the injection mold. The discrete detection points are embedded to a predetermined depth from the surface of the cavity through a precision drilling process. The point temperature sensor is a thermocouple sensor or a platinum resistance temperature sensor; at the gate position, the point temperature sensor is configured to monitor the instantaneous thermal shock of the melt entering the cavity in the early stage, and capture the temperature jump at the moment of injection by setting the sampling frequency. In the thin-walled region and the thick-walled region, the point temperature sensor is configured to monitor the inconsistency in cooling rate caused by the difference in wall thickness; The signal conditioning layer component includes a multi-channel synchronous sampling circuit, which is responsible for amplifying and filtering the millivolt-level analog voltage signal generated by the sensor, and converting it into a digital sequence signal through an analog-to-digital converter circuit. Each node of the physical layer sensing component is encapsulated in a metal armored sleeve, the outer surface of which is coated with a ceramic material with high thermal conductivity. The sensor leads are made of heat-resistant compensation wires with a multi-layer shielding structure to resist electromagnetic pulse interference generated by the heating coil of the injection molding machine.

3. The injection mold temperature monitoring system using IoT sensors according to claim 2, characterized in that, The infrared thermal imaging device is physically deployed on a side-fixed observation frame during the mold opening stroke of the injection molding machine; The infrared thermal imaging device has a built-in uncooled focal plane array detector, which has long-wave infrared sensing capability and a preset pixel resolution. The infrared thermal imaging device communicates bidirectionally with the data processing server through an industrial communication interface. When the mold opening stroke of the injection molding machine reaches the preset end position, the infrared thermal imaging device is configured to receive a trigger pulse and perform a single or multiple shutter scans during the static window period before the mold cavity is fully exposed and the ejection mechanism is activated. The infrared thermal imaging device also includes an ambient temperature compensation module, which reads the ambient temperature parameters of the workshop in real time and corrects the radiation brightness of the acquired raw thermal image to eliminate temperature measurement errors caused by background radiation. The two-dimensional temperature field distribution image is stored in matrix form, and each matrix element corresponds to the temperature value of the geometric coordinates of the mold surface. The infrared thermal imaging device also integrates an image dust removal algorithm, which uses an airflow curtain to prevent workshop dust from contaminating the lens, and uses software algorithms to identify and filter out false hot spots caused by residual mold release agent on the mold surface.

4. The injection mold temperature monitoring system using an IoT sensor according to claim 3, characterized in that, The data processing server also has redundancy and fault tolerance capabilities. When a processing node experiences a hardware failure, its load will be automatically migrated to a backup node. It also supports long-term archiving of historical data and uses a time-series database to store temperature fluctuation data.

5. The injection mold temperature monitoring system using an IoT sensor according to claim 4, characterized in that, The machine learning modeling unit also integrates a confidence assessment mechanism. When the confidence level of the predicted value is lower than a preset threshold, the system automatically triggers a manual intervention request and marks the batch of products as pending inspection. The machine learning modeling unit also integrates a transfer learning module. When the mold is changed or the batch of injection molding materials is changed, the transfer learning module uses a small amount of labeled data to fine-tune the parameters of the original model.

6. The injection mold temperature monitoring system using IoT sensors according to claim 5, characterized in that, The machine learning modeling unit is also equipped with an anomaly detection mechanism, which uses a two-level control graph logic for real-time monitoring. The first level is a single-point deviation monitoring logic. When the temperature value of any measuring point in the temperature sensing device exceeds the preset statistical control upper or lower limit, the system determines that the local thermal balance is disrupted. The second level is the overall thermal field uniformity monitoring logic. By calculating the standard deviation of the temperature in each region of the infrared thermal image, when the standard deviation exceeds the preset uniformity tolerance, it is determined that the global cooling system has failed. This anomaly detection mechanism can identify thermodynamic trend deviations in the injection molding process before quality defects occur, and send a warning pulse to the mold temperature control interface unit. The anomaly detection mechanism also includes occlusion recognition logic. If a foreign object is detected obstructing the infrared field of view, the system will automatically discard the heat map data of the current period and replace it with the smoothed value of the adjacent period.

7. The injection mold temperature monitoring system using an IoT sensor according to claim 6, characterized in that, The mold temperature control interface unit also has an energy efficiency monitoring function. Under the premise of ensuring product quality meets the standards, it can achieve green manufacturing by optimizing the output power of the mold temperature controller and reducing the power consumption of the chiller unit.

8. The injection mold temperature monitoring system using an Internet of Things sensor according to claim 7, characterized in that, The system also includes a process parameter association module, which is configured to acquire non-temperature process parameters in real time through the injection molding machine's bus interface, including peak injection pressure, holding pressure curve, holding time length, and set temperatures for each section of the melt tube. The process parameter association module uses these non-temperature parameters as auxiliary feature variables and performs feature fusion with the temperature features provided by the temperature sensing device and the infrared thermal imaging device. By adding a non-temperature physical dimension, the machine learning modeling unit can more comprehensively perceive the energy balance and pressure balance in the injection molding process, thereby improving the generalization ability and prediction robustness of the quality prediction model when faced with different batches of raw materials or environmental temperature and humidity fluctuations.

9. The injection mold temperature monitoring system using an Internet of Things sensor according to claim 8, characterized in that, The data processing server and the machine learning modeling unit are integrated into an industrial edge computing all-in-one machine deployed next to the injection molding machine; The edge computing all-in-one machine is responsible for localized data processing and rapid feedback, and controls the entire process from sensor data collection to generating mold temperature control recommendations within a preset real-time period. The edge computing all-in-one machine is connected to the injection molding machine, mold temperature controller and various sensing units via shielded twisted pair cables. It has electromagnetic compatibility performance and can operate in environments with high-frequency injection molding machine motor interference. The system also includes a field gateway layer, which is responsible for aggregating and converting the data of all injection molding machines in the workshop and running real-time anomaly warning logic. The field gateway layer is responsible for binding the physical identity of the mold with the sensor data, and the thermal field record can be traced back to the specific mold number, machine position and maintenance history.

10. The injection mold temperature monitoring system using an Internet of Things sensor according to claim 9, characterized in that, The system also includes an ultrasensitive thermal sensing network and an execution feedback interface; The ultra-sensitive thermal sensing network uses a detection unit based on fiber Bragg grating sensing technology to sense temperature by shifting the wavelength of light. It is deployed in the precision structure of the mold to sense localized minute thermal disturbances. The execution feedback interface is used to control an array of electric heating rods or a variable frequency electromagnetic induction heating device with millisecond-level response capability, and to precisely control the amount of heat compensation in the micro-region of the mold through pulse width modulation signal. When it is predicted that overheating will occur in a local area due to melt shearing, the execution feedback interface immediately cuts off the heating power supply at the corresponding location and triggers the micro-spray cooling mechanism to achieve isothermal molding control. The system also includes a cloud-based big data analysis platform that stores thermal data records of the entire life cycle of molds in the factory. The cloud-based big data analysis platform has a thermal field aging trend analysis function. By comparing the thermal characteristic drift of the same mold at different time periods, it can predict the degree of scaling in the internal water channels of the mold or the heating failure trend of the hot runner.

Images