Machine learning techniques for assessing thermal durability characteristics of physical materials

Abstract

Systems and methods for assessing thermal durability characteristics of a material, such as relative thermal index (RTI), using machine learning are disclosed. According to certain aspects, an electronic device may train a machine learning model using a training data set that includes coordinates indicative of a plurality of characteristics of a plurality of materials and that is marked with at least one thermal durability characteristic for each of the plurality of materials. The trained machine learning model may analyze a set of coordinates associated with a candidate material and output at least one thermal durability characteristic of the candidate material. The machine learning model may also be continuously updated and used for subsequent analysis.

Description

Technical Field

[0001] This disclosure relates to improvements related to evaluating the properties of physical materials. More specifically, this disclosure relates to platforms and techniques for using machine learning techniques to evaluate the thermal endurance properties of physical materials. Background Art

[0002] Long-term thermal aging (LTTA) is a test process for studying the effects of long-term exposure to high temperatures on materials. It involves subjecting a material to a higher temperature for an extended period of time, typically several months or even years, to simulate the long-term aging that a material might experience under real-world conditions. This method helps researchers and engineers understand how the properties, performance, and structural integrity of a given material change over time due to thermal stress and environmental factors. Generally, LTTA is used to determine the thermal endurance properties of a given material, which quantify the material's resistance to heat aging.

[0003] However, LTTA is a time-consuming process that requires continuous resources and monitoring, typically for several months or years. This results in higher costs and longer research timelines, making it less feasible for various industries, especially those with fast-paced development cycles. Although LTTA is designed to simulate real-world conditions, sometimes certain accelerated aging methods that subject the material to higher temperatures for shorter periods of time are used to speed up the test process. However, these methods may over-accelerate the properties to be evaluated, leading to mechanisms that are not relevant to the thermal performance at the operating temperature, resulting in inconsistent and inaccurate results. Additionally, for some industries, subjecting materials to LTTA may be impractical due to ethical constraints or regulatory limitations, especially in cases where the materials being tested are intended for critical applications that require rapid development and testing.

[0004] Therefore, there is an opportunity for platforms and techniques to efficiently, accurately, and effectively evaluate the thermal endurance properties of physical materials. Summary of the Invention

[0005] In one embodiment, a computer-implemented method for evaluating the thermal durability characteristics of materials using machine learning is provided. The computer-implemented method may include: training, by at least one processor, a machine learning model using a training data set, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of multiple materials, and (ii) is labeled with at least one thermal durability characteristic of each of the multiple materials, wherein the at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; accessing, by the at least one processor, a coordinate set indicating the multiple characteristics of a candidate material; analyzing, by the at least one processor, the coordinate set indicating the multiple characteristics of the candidate material using the trained machine learning model; and outputting, based on the analysis, by the machine learning model, at least one predicted thermal durability characteristic of the candidate material.

[0006] In another embodiment, a system for evaluating the thermal durability characteristics of materials using machine learning is provided. The system includes: a memory that stores a set of computer-readable instructions; and one or more processors connected to the memory. The one or more processors may be configured to execute the set of computer-readable instructions to cause the one or more processors to: train a machine learning model using a training data set, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of multiple materials, and (ii) is labeled with at least one thermal durability characteristic of each of the multiple materials, wherein the at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; access a coordinate set indicating the multiple characteristics of a candidate material; analyze the coordinate set indicating the multiple characteristics of the candidate material using the trained machine learning model; and output, based on the analysis, by the machine learning model, at least one predicted thermal durability characteristic of the candidate material.

[0007] Further, in an embodiment, a non-transitory computer-readable storage medium is provided that is configured to store instructions executable by one or more processors. The instructions may include: instructions for training a machine learning model using a training data set, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of multiple materials, and (ii) is labeled with at least one thermal durability characteristic of each of the multiple materials, wherein the at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; instructions for accessing a coordinate set indicating the multiple characteristics of a candidate material; instructions for analyzing the coordinate set indicating the multiple characteristics of the candidate material using the trained machine learning model; and instructions for outputting, based on the analysis, by the machine learning model, at least one predicted thermal durability characteristic of the candidate material. Description of the Drawings

[0008] Figure 1 Shows an overview of components and entities associated with systems and methods according to some embodiments.

[0009] Figure 2 Shows an overview of certain components configured to facilitate the implementation of systems and methods according to some embodiments.

[0010] Figures 3A to 3C Shows an example graph indicating various properties of example materials according to some embodiments.

[0011] Figure 4 Shows an example schematic diagram indicating various functions of systems and methods according to some embodiments.

[0012] Figure 5 Shows an example flowchart of using machine learning to determine the RTI of a material according to some embodiments.

[0013] Figure 6 Is an example hardware schematic diagram of a server configured to perform various functions according to some embodiments. Detailed Description

[0014] The LTTA test method is a method for evaluating the stability and performance of materials (such as polymers) at high temperatures over a long period of time. This test is typically used in various industries, such as automotive, aerospace, electrical equipment, and electronics (such as household appliances, consumer electronics, industrial control, electrical fittings), and construction, where certain materials present in components and other assemblies need to maintain their properties upon long-term exposure to heat. The LTTA test is crucial for evaluating the long-term durability and reliability of materials and predicting potential changes in their mechanical, electrical, flammability, and other critical properties.

[0015] The relative thermal index (RTI) is a thermal durability characteristic of a material, which is an indication of the material's ability to maintain specific properties (physical, electrical, etc.) upon long-term exposure to high temperatures. It is the highest temperature below which the material maintains its characteristics over a reasonable period of time. For each material, multiple relative thermal indices can be established, where each index is related to a specific property and a specific thickness of the material. In other words, the relative thermal index is the highest temperature below which the material maintains its characteristics over a reasonable period of time.

[0016] Typically, LTTA testing involves subjecting test specimens or samples of a given material to high temperatures for an extended duration. A set of samples for the material under test is prepared and their uniformity and the presence of any defects are examined. These samples are usually in the form of standardized sample shapes, depending on the material type, the form in which it is provided (e.g., molded resin, sheet, film, or other relevant shapes), and its intended application. The selected temperature for the test is typically higher than the expected use temperature of the material, and the selected duration of the test can range from several months to several years, depending on the intended application and the desired level of aging to be simulated.

[0017] Typically, the samples are placed in a controlled environment, such as an oven or environmental chamber, that can maintain a constant temperature during the test. The samples are then subjected to the high temperature for a predetermined duration. This allows the material to undergo thermal aging and undergo related chemical reactions and physical changes. This experimental process simulates the effects of long-term exposure to high temperatures on the material under real-world conditions. During the entire test period, the samples are periodically evaluated to measure changes in their properties, which can include testing mechanical properties (such as tensile strength, elongation, and impact resistance), electrical properties (such as dielectric strength), and other properties (such as flammability).

[0018] The result output of LTTA testing includes data on how certain properties of the material change over time and under specified temperature conditions. This data can be used to evaluate the long-term stability of the material, predict its behavior in real-world applications, and make informed decisions about its suitability for specific use cases. The output can be presented as a graph showing the change of properties over time or a table summarizing the observed changes in mechanical strength, mechanical shock, electrical, and flammability characteristics.

[0019] In particular, the output of LTTA testing for a given material (such as a plastic or polymer material used in electrical equipment for electronic devices) is the RTI that quantifies the heat aging resistance of the material. Typically, the RTI is a measure of the maximum continuous use temperature of the material, and it is defined as the highest temperature at which the material can be used in a specific application while maintaining its basic properties (such as mechanical strength, electrical insulation ability, and other relevant characteristics). The RTI of a given material can be determined based on the retention of specific properties after thermal aging. For example, if the material is subjected to LTTA testing and its mechanical strength, electrical insulation ability, and other relevant properties remain within acceptable limits after the aging process, the RTI can be defined as the maximum temperature at which these properties are maintained. Thus, the RTI value serves as a key indicator for manufacturers and engineers to determine the appropriate maximum operating temperature of components (such as enclosures and switches) made of a specific material, and it helps ensure that the components will remain functional and safe even when exposed to high temperatures during their intended service life.

[0020] The RTI value is expressed as a temperature in degrees Celsius (°C) or degrees Fahrenheit (°F). Different materials and material types can be tested to determine the corresponding RTI, including, for example, thermoplastics, thermosets, reinforced plastics, elastomers, composite materials, etc.

[0021] Typically, the thermal endurance characteristics of a material can be divided into three subclasses: Electrical RTI, Mechanical Shock RTI, and Mechanical Strength RTI. Electrical RTI focuses on the ability of an insulating material to maintain its electrical insulating properties when exposed to high temperatures. This subclass considers factors such as dielectric strength, resistivity, and breakdown voltage, where the Electrical RTI indicates the highest temperature at which the material can be used as an effective electrical insulator without significant degradation of its electrical properties. It is a useful parameter for materials used in electrical devices and other components where reliable electrical insulation is necessary.

[0022] Mechanical Shock RTI evaluates the ability of a given material to withstand mechanical stresses, such as shock or vibration, at high temperatures. This subclass can be particularly relevant for materials used in applications where impact resistance is crucial, such as in enclosures, housings, or structural components, where Mechanical Shock RTI provides information about the toughness and durability of the material under dynamic conditions and high temperatures.

[0023] Mechanical Strength RTI focuses on the ability of a given material to maintain its mechanical strength and structural integrity when exposed to high temperatures. It measures properties such as tensile strength and flexural strength after thermal aging and indicates the highest temperature at which the material can maintain its structural integrity without significant loss of mechanical properties. This parameter is important for materials used in load-bearing applications where maintaining strength and structural stability is crucial.

[0024] However, as discussed herein, there are several disadvantages and challenges associated with LTTA testing. In particular, these tests are time-consuming and expensive, and require a large amount of material to be tested to be available over a long duration. Additionally, it is difficult to accurately predict actual real-world conditions because LTTA tests are typically conducted under controlled laboratory conditions, and the test conditions may not always accurately represent the multiple stresses that the material will face in actual applications.

[0025] Furthermore, LTTA tests may not always account for all factors that can affect material degradation because some degradation mechanisms may not become apparent until after several years of exposure, leading to inaccuracies in RTI assessment. Additionally, for newly developed materials, there may not be enough data available for their long-term behavior, and organizations may need to rely on accelerated aging tests or other methods for initial assessment, the results of which are inaccurate. Moreover, selecting appropriate aging conditions (e.g., temperature, humidity, etc.) for LTTA testing is challenging and requires a significant amount of prior knowledge or preliminary experiments to establish suitable conditions.

[0026] This embodiment uses machine learning techniques to determine the thermal durability characteristics of materials without the need for a full LTTA test. According to various aspects, an electronic device can use a training data set to train a machine learning model, the training data set including coordinates indicating multiple characteristics of multiple materials and labeled with at least one RTI for each of the multiple materials. Additionally, a coordinate test set of a candidate material is input into the machine learning model, which analyzes the coordinate test set and outputs at least one predicted RTI of the candidate material.

[0027] The described system and method improve the prior art, namely the techniques for evaluating the RTI of materials. In particular, the system and method can provide accurate results much faster than physical tests, which is particularly advantageous when making shorter design and material selection decisions. In addition, the system and method significantly reduce the costs associated with physical LTTA tests, which have significant duration, resource requirements, and equipment maintenance.

[0028] The system and method also improve existing machine learning techniques. Specifically, the training data for the machine learning model includes a coordinate training set indicating multiple characteristics of a given material, the multiple characteristics including infrared analysis, thermogravimetric analysis, and differential scanning calorimetry. These coordinate sets are typically used to evaluate various characteristics of materials, including chemical composition, quality control, thermal stability, compositional changes, thermal properties, and purity assessment; however, these coordinate sets are neither obtainable nor used in conventional LTTA tests. Thus, by using this unique data set to train the machine learning model, the system and method allow for the incorporation of a richer and wider data set, enabling the machine learning model to learn the complex mapping between multiple input data types and test results.

[0029] Additionally, the trained machine learning model is capable of inputting data in a similar format for a candidate material (i.e., coordinates related to infrared analysis, thermogravimetric analysis, and differential scanning calorimetry), analyzing the data, and outputting predicted test results, which are not possible to accurately obtain using conventional LTTA tests. By incorporating a multi-format data set, the system and method even improve conventional machine learning models trained on only limited test data. Furthermore, any modification to the data output by the machine learning model can be re-input into the machine learning model, which results in an improved output for subsequent analysis by the machine learning model.

[0030] Figure 1 An overview of a system 100 showing components configured to facilitate implementation of the described system and method is presented. It should be understood that system 100 is merely an example, and alternative or additional components may be envisioned.

[0031] As Figure 1As shown, system 100 may include a data source group 101, such as an electronic device, a database, or other types of data sources. According to an embodiment, the data source group 101 may be an electronic device, such as a mobile device (e.g., a smart phone), a desktop computer, a laptop computer, a tablet computer, a phablet, a GPS (Global Positioning System) or GPS-enabled device, a smart watch, smart glasses, a smart bracelet, a wearable electronic device, a PDA (Personal Digital Assistant), a pager, a computing device configured for wireless communication, etc., where the electronic device may be associated with an individual or entity (e.g., a server computer or machine) such as a company, enterprise, corporation, etc.

[0032] The data source group 101 may communicate with (one or more) server computers 115 via one or more networks 110. In an embodiment, the (one or more) networks 110 may support any type of data communication via any standard or technology (e.g., GSM, CDMA, VoIP, TDMA, WCDMA, LTE, EDGE, OFDM, GPRS, EV-DO, UWB, the Internet, IEEE 802 including Ethernet, WiMAX, Wi-Fi, Bluetooth, 4G / 5G / 6G, Edge, etc.). The (one or more) server computers 115 may be associated with an entity such as a company, enterprise, corporation, etc. (usually a company) and may be configured to employ techniques for training a machine learning model and use the machine learning model to determine the RTI of a material. The (one or more) server computers 115 may include various components that support communication with the data source group 101.

[0033] According to an embodiment, the (one or more) server computers 115 may have access to (one or more) data sets 116 that may include (one or more) training data sets and (one or more) validation data sets. The (one or more) server computers 115 may use the (one or more) data sets 116 to train a machine learning model. Specifically, the server computer 115 may first use the (one or more) training data sets to train a set of machine learning models, and then apply or input the (one or more) validation data sets into the generated set of machine learning models to determine which machine learning model in the set of machine learning models is the most accurate or can otherwise be used as the final or selected machine learning model.

[0034] Additionally, data source group 101 may compile, store, access, and / or utilize data or information associated with different materials. In particular, the data or information may indicate properties of different materials and may include coordinate sets indicating infrared analysis, thermogravimetric analysis, and / or differential scanning calorimetry of different materials. The server computer(s) 115 may analyze the data / information using a trained machine learning model according to the functionality described herein, which may result in one or more predicted RTIs for a given material. In some embodiments, the server computer 115 may access the raw data or information (and / or the data set(s) 116) from the data source group 101 or from another source.

[0035] The server computer(s) 115 may be configured to interface with or support a memory or storage device 113, 114 capable of storing various data (e.g., in one or more databases or other forms of storage). According to an embodiment, the storage devices 113, 114 may store data or information associated with any machine learning model(s) generated by the server computer(s) 115. Additionally, the server computer(s) 115 may access data associated with the stored machine learning model from the storage devices 113, 114 to input a set of inputs into the machine learning model. Further, the storage devices 113, 114 may store data associated with materials, such as data indicating properties of the materials.

[0036] Although shown as a single server computer 115 in Figure 1 it should be understood that the server computer(s) 115 may be in the form of a distributed cluster of computers, servers, machines, cloud-based services, etc. In this embodiment, an access entity may utilize the server computer(s) 115 as part of an on-demand cloud computing platform. Thus, when the data source group 101 is connected to the server computer(s) 115, the data source group 101 may actually be connected to one or more of a plurality of distributed computers, servers, machines, etc. to facilitate the described functionality. Additionally, it should be understood that variations in the amounts of the data source group 101 and the storage devices 113, 114 are envisioned. Figure 2 Illustrates more specific components associated with the system and method.

[0037] Figure 2 is an example environment 150 according to an embodiment, where input data 117 is processed into output data 151 via a material evaluation platform 155. The material evaluation platform 155 may be on any computing device or combination of computing devices (including as regarding Figure 1implemented on the (one or more) server computers 115 under discussion. The components of a computing device can include, but are not limited to, a processing unit (e.g., (one or more) processors 156), a system memory (e.g., memory 157), and a system bus 158 that couples the various system components including the memory 157 to the (one or more) processors 156. In some embodiments, the (one or more) processors 156 can include one or more parallel processing units capable of processing data in parallel with each other. The system bus 158 can be any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, or a local bus, and can use any suitable bus architecture. By way of example and not limitation, these architectures include Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MCA) bus, Enhanced ISA (EISA) bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus (also known as a mezzanine bus).

[0038] The material evaluation platform 155 can also include a user interface 153 configured to present content (e.g., input data, output data, processed data, and / or other information). Additionally, a user can inspect the results of a material evaluation and make selections on the presented content via the user interface 153, such as inspecting output data presented thereon, making selections, and / or performing other interactions. The user interface 153 can be embodied as part of a touch screen configured to sense a user's touch interaction and gestures. Although not shown, other system components communicatively coupled to the system bus 158 can include input devices such as a cursor control device (e.g., a mouse, trackball, touchpad, etc.) and a keyboard (not shown). A monitor or other type of display device can also be connected to the system bus 158 via an interface such as a video interface. In addition to the monitor, the computer can also include other peripheral output devices, such as a printer, which can be connected via an output peripheral interface (not shown).

[0039] The memory 157 may include various computer-readable media. Computer-readable media can be any available media that can be accessed by a computing device and can include volatile and non-volatile media, as well as removable and non-removable media. By way of non-limiting example, computer-readable media may include computer storage media, which can include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, routines, applications (e.g., the material evaluation application 160), data structures, program modules, or other data. Computer storage media may include, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by the processor 156 of the computing device.

[0040] The material evaluation platform 155 may operate in a networked environment and communicate with one or more remote platforms, such as the remote platform 165, via a network 162, such as a local area network (LAN), a wide area network (WAN), or other suitable network. The platform 165 may be implemented on any computing device and may include many or all of the elements described above with respect to the platform 155. In some embodiments, as will be further described herein, instead of or in addition to the platform 155, the material evaluation application 160 may be stored and executed by the remote platform 165.

[0041] Generally, each of the input data 117 and the output data 151 may be embodied as any type of electronic document, file, template, etc., which may include various graphical / visual and / or text content and may be stored as program data in the memory of a hard drive, magnetic disk, and / or optical disk drive in the material evaluation platform 155 and / or the remote platform 165. The material evaluation platform 155 may support one or more techniques, algorithms, etc. for analyzing the input data 117 to generate the output data 151. Specifically, the material evaluation application 160 may train and use a machine learning model to evaluate the RTI of a given candidate material. The memory 157 may store the output data 151 and other data generated or used by the material evaluation platform 155 in association with the analysis of the input data 117.

[0042] According to an embodiment, the material evaluation application 160 can employ machine learning and artificial intelligence techniques, such as, for example, regression analysis (e.g., logistic regression, linear regression, random forest regression, probit regression, or polynomial regression), entity resolution, classification analysis, k-nearest neighbors, decision trees, random forests, boosting, neural networks, support vector machines, deep learning, reinforcement learning, Bayesian networks, etc. When the input data 117 is a (one or more) training data set, the material evaluation application 160 can analyze / process the input data 117 to generate a (one or more) machine learning model for storage as part of the model data 163 that can be stored in the memory 157. In an embodiment, various output data 151 can be added to the (one or more) machine learning models stored as part of the model data 163. When analyzing or processing the input data 117, the material evaluation application 160 can use any output data 151 previously generated by the material evaluation platform 155.

[0043] The material evaluation application 160 (or another component) can cause the output data 151 (and in some cases the training or input data 117) to be displayed on the user interface 153 for inspection by a user of the material evaluation platform 155, such as to inspect specific model inputs and / or the results of machine learning analysis, as part of a dashboard, interface, etc. The user can select to inspect and / or modify the displayed data. For example, the user can inspect the output data 151 to manually override the model output, add additional outputs, and / or facilitate other functions.

[0044] Figures 3A to 3C Exemplary curves indicating various properties of exemplary materials are shown. It should be understood that Figures 3A to 3C each of the respective curves is exemplary, and different materials have different values for various properties.

[0045] Figure 3A A graph 305 depicting a differential scanning calorimetry (DSC) curve of a material is shown, which can measure the heat flow into or out of a material as the material undergoes a temperature change, revealing properties such as phase transitions, crystallization, and thermal properties.

[0046] In Figure 3A graph 305, the DSC results are plotted with temperature (°C) on the x-axis and heat flow (mW or mJ) on the y-axis. Endothermic and exothermic events associated with phase transitions, crystallization, and chemical reactions can be observed as peaks and valleys in the DSC curve. For example, as Figure 3A shown, at a temperature of around 160 °C, the material undergoes a phase transition.

[0047] Figure 3BFIG. 310 shows a thermogravimetric analysis (TGA) curve depicting a material, which can measure the weight change of the material when subjected to an increasing temperature. The TGA curve can enable the identification of the respective thermal stabilities and decomposition properties of the various components of the material formulation, as well as the presence of volatile components.

[0048] In Figure 3B the graph 310 of, the TGA results (i.e., the weight percentage or mass of the remaining material as a function of temperature) are plotted with temperature (°C) on the x-axis and weight percentage (or mass) on the y-axis. The TGA curve can show the weight loss corresponding to the release of volatile components of the material or the decomposition of the various components. For example, as Figure 3B shown, at a temperature of approximately 400 °C, the material begins to decompose rapidly.

[0049] Figure 3C FIG. 315 shows a graph depicting an infrared analysis (IR) curve of a material, which measures the absorption rate of the material to infrared light. This curve can provide information about the molecular vibrations and functional groups present in the material.

[0050] In Figure 3C the graph 315 of, the IR results are plotted with wavenumber (cm^-1) (which is inversely proportional to the wavelength of infrared light) on the x-axis and percent transmittance or percent absorbance on the y-axis, resulting in an IR spectrum. Different functional groups in the material can absorb infrared light at specific wavenumbers, generating characteristic peaks in the spectrum. The y-axis shows the intensity of the absorbed light, either expressed as percent transmittance (i.e., how much light passes through the material) or as percent absorbance (i.e., how much light is absorbed by the material).

[0051] Figure 4 FIG. 400 shows an example associated with the described embodiment. FIG. 400 includes one or more data sources 401 (such as one or more data sources 101 described with reference to Figure 1 and one or more servers 415 (such as one or more servers 115 described with reference to Figure 1 . It should be understood that these functions are merely exemplary, and additional or alternative functions in different orders can be envisioned.

[0052] As Figure 4 shown, one or more servers 415 can be configured to access (420) a training data set and a validation data set. According to an embodiment, each training and validation data set may include a set of coordinates respectively indicating a plurality of characteristics of a given material, where the plurality of characteristics may be infrared analysis, thermogravimetric analysis, and differential scanning calorimetry of the given material, as described with respect to Figures 3A to 3C . It should be understood that a given set of coordinates may be the (x,y) coordinates of a given curve, such as Figures 3A to 3CThe infrared analysis, thermogravimetric analysis, and differential scanning calorimetry curves described in

[0053] Additionally, each training and validation dataset can be labeled with at least one thermal durability characteristic of the relevant material. According to an embodiment, the at least one thermal durability characteristic can be one or more of electrical RTI, mechanical shock RTI, or mechanical strength RTI. For example, an exemplary coordinate set for a given plastic material can include three (3) separate (x,y) coordinate plots of infrared analysis, thermogravimetric analysis, and differential scanning calorimetry for the given plastic material, where the coordinate set can be labeled with each of electrical RTI, mechanical shock RTI, and mechanical strength RTI. It should be understood that a given coordinate set can be labeled with one or more of electrical RTI, mechanical shock RTI, or mechanical strength RTI.

[0054] (One or more) servers 415 can use the training and validation datasets to train (422) a machine learning model. According to an embodiment, (one or more) servers 415 can be configured to preprocess the training and validation datasets by cleaning the raw data, performing feature engineering on the raw data, and / or transforming the raw data into a format suitable for machine learning. Generally, (one or more) servers 415 can use the training dataset to train the model and use the validation dataset to fine-tune the model and monitor its performance.

[0055] In an embodiment, (one or more) servers 415 can determine or receive instructions related to the type of machine learning model to be used. For example, different machine learning models can be decision trees, neural networks, support vector machines, or any other suitable models. During training, the model can learn the underlying patterns and relationships in the training dataset. (One or more) servers 415 can iteratively update any parameters or weights of the model to minimize a selected loss function, which can measure any difference between the predictions of the model and the actual target values in the training dataset.

[0056] Additionally, (one or more) servers 415 can adjust hyperparameters that are not learned from the data, such as learning rate, depth of a decision tree, number of hidden layers in a neural network, etc. (One or more) servers 415 can use different hyperparameters when adjusting the model and can evaluate the performance of the model on the validation dataset to select the best hyperparameters.

[0057] Periodically or after a set number of training iterations (i.e., epochs), one or more servers 415 can use a validation dataset to evaluate the performance of the model, such as by evaluating accuracy, F1 score, mean squared error, etc. Generally, the performance of the model on the validation dataset helps to detect problems such as overfitting (i.e., the model is too complex and performs well on the training data but poorly on unseen data) or underfitting (i.e., the model is too simple and performs poorly on both the training and validation data).

[0058] Based on the results of the validation, one or more servers 415 can further fine-tune the model by adjusting hyperparameters or making changes to the model architecture. Generally, the steps of training, validating, and fine-tuning the model are repeated until the performance of the model on the validation dataset reaches a satisfactory level. In an embodiment, and at this point, one or more servers 415 can evaluate the performance of the model on a separate test dataset, which can more realistically estimate how well the model will perform on unseen data. After testing the machine learning model, the machine learning model can be used to analyze data associated with candidate materials.

[0059] In particular, one or more data sources 401 can provide (424) information related to candidate materials to one or more servers 415. It should be understood that one or more servers 415 can access the information locally or receive the information from another source. In an embodiment, the information can include a set of coordinates indicating infrared analysis, thermogravimetric analysis, and differential scanning calorimetry of the candidate material. For example, the information can include a first set of (x,y) coordinates for infrared analysis, a second set of (x,y) coordinates for thermogravimetric analysis, and a third set of (x,y) coordinates for differential scanning calorimetry, as discussed herein with respect to Figures 3A to 3C similarly discussed.

[0060] One or more servers 415 can use the machine learning model to analyze (426) the information. In particular, one or more servers 415 can input the set of coordinates indicating infrared analysis, thermogravimetric analysis, and differential scanning calorimetry of the candidate material into the trained machine learning model.

[0061] Based on the analysis in (426), one or more server computers 415 may output (428) at least one predicted thermal durability characteristic of a candidate material via a machine learning model. In an embodiment, the at least one predicted thermal durability characteristic may be at least one of the following: electrical relative thermal index (RTI), mechanical shock RTI, or mechanical strength RTI. Additionally, the machine learning model may output a confidence level for each of the at least one predicted thermal durability characteristics of the candidate material. In an embodiment, the confidence level may be a percentage ranging from 0% to 100%, or may be another metric that indicates the likelihood that a given predicted thermal durability characteristic accurately reflects the actual thermal durability characteristic of the candidate material.

[0062] One or more servers 415 may access (430) at least one actual RTI of a candidate material. In an embodiment, one or more servers 415 may receive or otherwise access at least one actual RTI as an input (e.g., via a user interface), which may be derived from a physical test of the candidate material or via another channel. One or more servers 415 may additionally use information associated with the candidate material (i.e., a set of coordinates indicating multiple characteristics of the candidate material) and at least one actual thermal durability characteristic of the candidate material to update (432) the machine learning model. Thus, one or more servers 415 may continuously update the machine learning model when accessing or determining new data or updated data. Additionally, one or more servers 415 may use the updated machine learning model in subsequent analyses, which improves the accuracy of the output of the machine learning model.

[0063] Figure 5 is a block diagram of an example method 500 for using machine learning to evaluate the thermal durability characteristics of materials. Method 500 may be implemented by one or more electronic devices (such as Figure 1 the server computer 115 depicted in

[0064] Method 500 may begin at block 505, where one or more electronic devices use a training data set to train a machine learning model. According to an embodiment, the training data set may (i) include a coordinate training set indicating multiple characteristics of a variety of materials, and (ii) be labeled with at least one thermal durability characteristic of each of the variety of materials, where the at least one thermal durability characteristic may be associated with each coordinate training set in the coordinate training set of the material.

[0065] In addition, in embodiments, a training dataset and a validation dataset can be used to train a machine learning model, where the validation dataset can (i) include a coordinate validation set indicating multiple characteristics of multiple materials, and (ii) be labeled with at least one thermal durability characteristic of each of the multiple materials, where at least one thermal durability characteristic is associated with each coordinate validation set in the coordinate validation set of the material. Additionally, in embodiments, at least one thermal durability characteristic can be one or more of electrical RTI, mechanical shock RTI, or mechanical strength RTI.

[0066] The electronic device can access (block 510) a coordinate set indicating multiple characteristics of a candidate material. In embodiments, the coordinate set can indicate at least one of the following: infrared analysis, thermogravimetric analysis, or differential scanning calorimetry of the candidate material.

[0067] The electronic device can use the trained machine learning model to analyze (block 515) the coordinate set indicating multiple characteristics of the candidate material, such as by inputting the coordinate set into the trained machine learning model. Additionally, at block 520, the machine learning model can output, based on the analysis, at least one predicted thermal durability characteristic of the candidate material, such as one or more of electrical RTI, mechanical shock RTI, or mechanical strength RTI. In embodiments, the machine learning model can additionally output a confidence level for each of at least one predicted thermal durability characteristic of the candidate material.

[0068] The electronic device can access (block 525) at least one actual thermal durability characteristic (e.g., actual RTI) of the candidate material, such as the result of physically testing the candidate material or via another channel. Additionally, the electronic device can update (block 530) the machine learning model with the coordinate set indicating multiple characteristics of the candidate material and at least one actual thermal durability characteristic of the candidate material.

[0069] Figure 6 An example server 615 (e.g., such as server 115 as described with respect to Figure 1 is shown in a hardware schematic, where the functions discussed herein can be implemented. It should be understood that the components of server 615 are merely exemplary, and additional or alternative components and arrangements can be envisioned.

[0070] Server 615 can include a processor 659 and a memory 656. Memory 656 can store an operating system 657 and an application set 651 (i.e., machine-readable instructions) that can facilitate the implementation of the functions discussed herein. For example, one of the application set 651 can be a material evaluation application 652, such as for training a machine learning model and using the machine learning model to evaluate the RTI of materials. It should be understood that one or more other applications 653 can be envisioned.

[0071] The processor 659 can be connected to the memory 656 to execute the operating system 657 and the application set 651. According to some embodiments, the memory 656 can also store other data 658, such as machine learning model data and / or other data, such as data related to materials that can be used in training, analysis, and determination as discussed herein. The memory 656 can include one or more forms of volatile and / or non-volatile, fixed and / or removable memory, such as read-only memory (ROM), electronically programmable read-only memory (EPROM), random access memory (RAM), erasable electronically programmable read-only memory (EEPROM), and / or other hard disk drives, flash memories, MicroSD cards, etc.

[0072] The server 615 can also include a communication module 655, which is configured to transmit data via one or more networks ( Figure 6 (not shown in the figure). According to some embodiments, the communication module 655 can include one or more transceivers (e.g., WAN, WWAN, WLAN, and / or WPAN transceivers), which operate according to IEEE standards, 3GPP standards, or other standards, and are configured to receive and send data via one or more external ports 654.

[0073] The server 615 can also include a user interface 662, which is configured to present information to the user and / or receive input from the user. As Figure 6 shown, the user interface 662 can include a display screen 663 and I / O components 664 (e.g., ports, capacitive or resistive touch-sensitive input panels, keys, buttons, lights, LEDs, external or built-in keyboards). According to some embodiments, the user can access the server 615 via the user interface 662 to check information, make selections, and / or perform other functions.

[0074] In some embodiments, the server 615 can perform the functions discussed herein as part of a "cloud" network, or can otherwise communicate with other hardware or software components within the cloud to send, retrieve, or otherwise analyze data.

[0075] Generally, a computer program product according to an embodiment can include a computer-usable storage medium (e.g., a standard random access memory (RAM), an optical disc, a universal serial bus (USB) drive, etc.) that contains computer-readable program code, where the computer-readable program code can be adapted to be executed by a processor 659 (e.g., working in conjunction with an operating system 657) to facilitate the implementation of the functions described herein. In this regard, the program code can be implemented in any desired language and can be implemented as machine code, assembly code, byte code, interpretable source code, etc. (e.g., via Golang, Python, Scala, C, C++, Java, Actionscript, Objective-C, Javascript, CSS, XML). In some embodiments, the computer program product can be part of a cloud network of resources.

[0076] Although the following presents a detailed description of multiple different embodiments, it should be understood that the legal scope of the present invention can be defined by the literal words of the claims appended to this patent. The detailed description should be construed as merely exemplary and does not describe every possible embodiment, because describing every possible embodiment is impracticable even if possible. Many alternative embodiments can be implemented using current technology or technology developed after the filing date of this patent, and these embodiments still fall within the scope of the claims.

[0077] Throughout the specification, multiple instances can implement components, operations, or structures described as a single instance. Although the individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations can be performed simultaneously, and the operations are not required to be performed in the illustrated order. Structures and functions presented as separate components in an example configuration can be implemented as a combined structure or component. Similarly, structures and functions presented as a single component can be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.

[0078] Additionally, certain embodiments are described herein as including logic or multiple routines, subroutines, applications, or instructions. These can constitute software (e.g., code embodied on a non-transitory machine-readable medium) or hardware. In hardware, a routine, etc. is a tangible unit capable of performing certain operations and can be configured or arranged in a certain manner. In an example embodiment, one or more computer systems (e.g., stand-alone client or server computer systems) or one or more hardware modules of a computer system (e.g., a processor or a group of processors) can be configured by software (e.g., an application or a portion of an application) to operate as a hardware module that performs certain operations described herein.

[0079] In various embodiments, a hardware module may be implemented mechanically or electronically. For example, a hardware module may include dedicated circuitry or logic that may be permanently configured (e.g., as a dedicated processor such as a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC)) to perform certain operations. A hardware module may also include programmable logic or circuitry (e.g., as included within a general-purpose processor or other programmable processor) that may be temporarily configured by software to perform certain operations. It should be understood that the decision to implement a hardware module mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0080] Accordingly, the term "hardware module" should be understood to encompass a tangible entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a particular manner or to perform particular operations described herein. Considering embodiments in which a hardware module is temporarily configured (e.g., programmed), each hardware module need not be configured or instantiated at any one instance in time. For example, in cases where a hardware module includes a general-purpose processor configured by software, the general-purpose processor may be configured to be respective different hardware modules at different times. The software may configure the processor accordingly, e.g., to constitute a particular hardware module at one instance in time and a different hardware module at a different instance in time.

[0081] A hardware module may provide information to, and receive information from, other hardware modules. Accordingly, the described hardware modules may be considered to be communicatively coupled. In cases where multiple such hardware modules exist simultaneously, communication may be achieved via signal transmission that connects the hardware modules (e.g., via appropriate circuitry and buses). In embodiments in which multiple hardware modules are configured or instantiated at different times, communication between such hardware modules may be achieved, for example, by storing and retrieving information in a memory structure accessible by the multiple hardware modules. For example, one hardware module may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. Then, at a later time, another hardware module may access the memory device to retrieve and process the stored output. A hardware module may also initiate communication with input or output devices and may operate on resources (e.g., a collection of information).

[0082] The various operations of the example methods described herein may be at least partially performed by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions. In some example embodiments, the modules referred to herein may include processor-implemented modules.

[0083] Similarly, the methods or routines described herein can be implemented, at least in part, by a processor. For example, at least some of the operations of the method can be performed by one or more processors or hardware modules implemented by a processor. The execution of certain operations can be distributed among one or more processors, which are not only resident within a single machine but also deployed across multiple machines. In some example embodiments, one or more processors can be located in a single location (e.g., in a home environment, an office environment, or as a server farm), while in other embodiments, the processors can be distributed across multiple locations.

[0084] The execution of certain operations can be distributed among one or more processors, which are not only resident within a single machine but also deployed across multiple machines. In some example embodiments, one or more processors or modules implemented by a processor can be located in a single geographical location (e.g., within a home environment, an office environment, or a server farm). In other example embodiments, one or more processors or modules implemented by a processor can be distributed across multiple geographical locations.

[0085] Unless otherwise specifically stated, discussions herein using terms such as "processing", "computing (computing and calculating)", "determining", "rendering", "displaying", etc. can refer to actions or processes of a machine (e.g., a computer) that manipulates or transforms data represented as physical (e.g., electronic, magnetic, or optical) quantities within one or more memories (e.g., volatile memory, non-volatile memory, or a combination thereof), registers, or other machine components that receive, store, transmit, or display information.

[0086] As used herein, any reference to "an embodiment" or "embodiments" means that a particular element, feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment. The phrase "in one embodiment" appearing in various places in the specification does not necessarily all refer to the same embodiment.

[0087] As used herein, the terms "comprises", "comprising", "includes", "including", "has", "having", or any other variation thereof are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. Further, unless expressly stated to the contrary, "or" refers to an inclusive "or" rather than an exclusive "or". For example, condition A or B is satisfied by any of the following: A is true (or present) and B is false (or absent), A is false (or absent) and B is true (or present), and both A and B are true (or present).

[0088] In addition, the articles "a" or "an" are used to describe elements and components of embodiments herein. This is done for convenience only and to provide a general description. The specification and the appended claims should be understood to include one or at least one, and the singular also includes the plural, unless clearly indicated otherwise.

[0089] The detailed description is to be construed as illustrative, and not to describe every possible embodiment, as it would be impractical to describe every possible embodiment.

Claims

1. A computer-implemented method for evaluating the thermal durability characteristics of a material using machine learning, the computer-implemented method comprising: training, by at least one processor, a machine learning model using a training data set, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of a plurality of materials, and (ii) is labeled with at least one thermal durability characteristic of each of the plurality of materials, wherein the at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; accessing, by the at least one processor, a coordinate set indicating the multiple characteristics of a candidate material; analyzing, by the at least one processor, the coordinate set indicating the multiple characteristics of the candidate material using the trained machine learning model; and outputting, based on the analysis, by the machine learning model, at least one predicted thermal durability characteristic of the candidate material.

2. The computer-implemented method according to claim 1, wherein, The training data set is labeled with at least one of an electrical relative thermal index, a mechanical shock relative thermal index, or a mechanical strength relative thermal index.

3. The computer-implemented method according to claim 1, wherein, Training the machine learning model includes: training, by the at least one processor, the machine learning model using the training data set and a validation data set, wherein the validation data set (i) includes a coordinate validation set indicating the multiple characteristics of the plurality of materials, and (ii) is labeled with the at least one thermal durability characteristic of each of the plurality of materials, wherein the at least one thermal durability characteristic is associated with each coordinate validation set in the coordinate validation set of the material.

4. The computer-implemented method according to claim 1, wherein, Accessing the coordinate set indicating the multiple characteristics of the candidate material includes: accessing, by the at least one processor, the coordinate set indicating at least one of the following of the candidate material: infrared analysis, thermogravimetric analysis, or differential scanning calorimetry.

5. The computer-implemented method according to claim 1, further comprising: accessing, by the at least one processor, at least one actual thermal durability characteristic of the candidate material; and updating, by the at least one processor, the machine learning model using (i) the coordinate set indicating the multiple characteristics of the candidate material and (ii) the at least one actual thermal durability characteristic of the candidate material.

6. The computer-implemented method according to claim 1, wherein, Outputting, by the machine learning model, the at least one predicted thermal durability characteristic of the candidate material includes: outputting, by the machine learning model, (i) the at least one predicted thermal durability characteristic of the candidate material and (ii) a confidence level for each of the at least one predicted thermal durability characteristic of the candidate material.

7. The computer-implemented method according to claim 1, wherein, Accessing the coordinate set indicating the multiple characteristics of the candidate material includes: accessing, by the at least one processor, the coordinate set, wherein each in the coordinate set includes (x, y) coordinates indicating a corresponding characteristic among the multiple characteristics.

8. A system for evaluating the thermal durability characteristics of a material using machine learning, comprising: a memory storing a set of computer-readable instructions; and One or more processors, connected to the memory and configured to execute the computer-readable instruction set to cause the one or more processors to: Use a training data set to train a machine learning model, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of multiple materials, and (ii) is labeled with at least one thermal durability characteristic of each of the multiple materials, wherein The at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; Access a coordinate set indicating the multiple characteristics of the candidate material; Use the trained machine learning model to analyze the coordinate set indicating the multiple characteristics of the candidate material; and Based on the analysis, output, by the machine learning model, at least one predicted thermal durability characteristic of the candidate material.

9. The system according to claim 8, wherein, The training data set is labeled with at least one of an electrical relative thermal index, a mechanical shock relative thermal index, or a mechanical strength relative thermal index.

10. The system according to claim 8, wherein, To train the machine learning model, the one or more processors are configured to: Use the training data set and a validation data set to train the machine learning model, wherein the validation data set (i) includes a coordinate validation set indicating the multiple characteristics of the multiple materials, and (ii) is labeled with the at least one thermal durability characteristic of each of the multiple materials, wherein the at least one thermal durability characteristic is associated with each coordinate validation set in the coordinate validation set of the material.

11. The system according to claim 8, wherein, The coordinate set indicates at least one of the following for the candidate material: infrared analysis, thermogravimetric analysis, or differential scanning calorimetry.

12. The system according to claim 8, wherein, The one or more processors are configured to execute the computer-readable instruction set to further cause the one or more processors to: Access at least one actual thermal durability characteristic of the candidate material; And Use (i) the coordinate set indicating the multiple characteristics of the candidate material and (ii) the at least one actual thermal durability characteristic of the candidate material to update the machine learning model.

13. The system according to claim 8, wherein, The machine learning model outputs (i) the at least one predicted thermal durability characteristic of the candidate material and (ii) a confidence level for each of the at least one predicted thermal durability characteristics of the candidate material.

14. The system according to claim 8, wherein Each in the coordinate set includes (x, y) coordinates indicating a corresponding characteristic among the multiple characteristics.

15. A non-transitory computer-readable storage medium configured to store instructions executable by one or more processors, the instructions including: Instructions for using a training data set to train a machine learning model, wherein the training data set (i) includes a coordinate training set indicating multiple characteristics of multiple materials, and (ii) is labeled with at least one thermal durability characteristic of each of the multiple materials, wherein the at least one thermal durability characteristic is associated with each coordinate training set in the coordinate training set of the material; Instructions for accessing a coordinate set indicating the multiple characteristics of the candidate material; Instructions for using the trained machine learning model to analyze the coordinate set indicating the multiple characteristics of the candidate material; and Instructions for outputting, by the machine learning model based on the analysis, at least one predicted thermal durability characteristic of the candidate material.

16. The non-transitory computer-readable storage medium according to claim 15, wherein, The training data set is labeled with at least one of an electrical relative thermal index, a mechanical shock relative thermal index, or a mechanical strength relative thermal index.

17. The non-transitory computer-readable storage medium according to claim 15, wherein, The instructions for training the machine learning model include: Instructions for training the machine learning model using the training data set and a validation data set, where the validation data set (i) includes a coordinate validation set indicating the multiple characteristics of the multiple materials, and (ii) is labeled with the at least one thermal durability characteristic of each of the multiple materials, where the at least one thermal durability characteristic is associated with each coordinate validation set in the coordinate validation set of the material.

18. The non-transitory computer-readable storage medium according to claim 15, wherein, The instructions for accessing the coordinate set indicating the multiple characteristics of the candidate material include: Instructions for accessing the coordinate set indicating at least one of the following for the candidate material: infrared analysis, thermogravimetric analysis, or differential scanning calorimetry.

19. The non-transitory computer-readable storage medium according to claim 15, wherein, The instructions further include: Instructions for accessing at least one actual thermal durability characteristic of the candidate material; and Instructions for updating the machine learning model using (i) the coordinate set indicating the multiple characteristics of the candidate material and (ii) the at least one actual thermal durability characteristic of the candidate material.

20. The non-transitory computer-readable storage medium according to claim 15, wherein, The instructions for outputting, by the machine learning model, the at least one predicted thermal durability characteristic of the candidate material include: Instructions for outputting, by the machine learning model, (i) the at least one predicted thermal durability characteristic of the candidate material and (ii) the confidence level of each of the at least one predicted thermal durability characteristics of the candidate material.

Images