Test and measurement instrument and method for analyzing hardware under test

Abstract

To predict a future failure of hardware under test (HUT).SOLUTION: The test and measurement instrument includes one or more ports connected to the HUT31, sensors 30 connected to the HUT31 and the test and measurement instrument 32, a display for displaying representations of one or more signals from the HUT31 and the sensors 30, and one or more processors. The one or more processors are configured to execute a program that causes the one or more processors to obtain one or more datasets from the sensors 30, apply one or more machine-learning models 36 to the one or more datasets, and receive a predictive analysis result for the HUT31 from the one or more machine-learning models 36.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to testing motors and hardware having moving parts, and more particularly to using machine learning to predict future failures of the hardware under test. [Background technology]

[0002] Typically, designers test motors using Inverter Motor Drive Analysis (IMDA) software using voltage and current waveforms during the design and debug phases of the motor design workflow. The software typically measures power quality, direct-quadrature-zero (DQ0), and harmonics. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2024-151327 [Patent Document 2] Patent Publication No. 2021-012199 [Non-patent literature]

[0004] [Non-Patent Document 1] "MOTOR MONITORING MARKET SIZE & SHARE ANALYSIS - GROWTH TRENDS & FORECASTS (2025 - 2030)", Mordor Intelligence, [online], [Retrieved July 9, 2025], Internet<https: / / www.mordorintelligence.com / industry-reports / motor-monitoring-market> [Non-patent document 2] "Three-Phase Inverter / Motor / Drive Analysis," Tektronix, [online], [Retrieved November 4, 2025], Internet <https: / / www.tek.com / ja / datasheet / inverter-motor-drive-analysis-5series-mso-option-5-imda-application-datasheet> Summary of the Invention [Problem to be solved by the invention]

[0005] One common method for diagnosing motors, such as alternating current (AC) motors, brushless DC (BLDC) motors, single-phase or three-phase motors, is to use motor current signature analysis (MCSA) to detect motor faults. One major drawback of this type of motor analysis is that these motors cannot be analyzed while connected to the circuit; these tests are performed in controlled environments or lab conditions. Motor designers typically do not perform tests on motors in service.

[0006] Electric motors are exposed to several operational and environmental hazards that cause thermal stress in the windings, thereby reducing the motor's ability to insulate components from thermal stress and affecting its life expectancy. Therefore, motor monitoring has become a key aspect and the most effective tactic for identifying potential problems. Until recently, preventing motor failures was complex and costly. However, with the decline in the cost of submeter sensors and advancements in big data technology, motor monitoring has become accurate and affordable. All these factors mentioned above are expected to drive the growth of the global motor monitoring market over the forecast period.

[0007] As mentioned in the above-mentioned non-patent document 1, the motor monitoring market size is estimated to be US$ 2.67 billion in 2024 and is projected to reach US$ 4.4 billion by 2029, growing at a compound annual growth rate (CAGR) of 10.46% during the forecast period (2024-2029). The table below summarizes the survey results. [Table 1]

[0008] The major players in this market are ABB, National Instruments, Honeywell, Siemens, and General Electric. These motors consume 70% of the industry's electricity, demonstrating their importance. The market is expanding as consumer preferences move towards a "shift left" approach and seamless user experiences.

[0009] Companies that provide test and measurement solutions for this growing market will also experience growth. [Means for solving the problem]

[0010] Embodiments of the present application address the problem of predictive maintenance and fault diagnosis in an efficient manner, thereby improving the uptime of Hardware Under Test (HUT). HUTs may consist of various types of hardware, such as AC motors, which are integrated with electric vehicles (EVs) and industrial applications. The present embodiments apply to all types of motors, including, but not limited to, brushless DC motors, one-phase motors, and three-phase motors. Other types of HUTs may include devices with moving parts, whether the parts are moved by the motor or by other forces. For example, the present embodiments apply to the high-speed shaft of a wind turbine, which moves in response to wind pressure. The term HUT, as used herein, includes these examples and others.

[0011] The HUT's real-time sensors enable the collection of large amounts of data. Using machine learning, this data can be used to accurately predict future failures. In test and measurement systems, artificial intelligence (AI) and machine learning (ML) have emerged as powerful tools, revolutionizing data analysis and the extraction of insights from that analysis. ML modeling using AI has become a key data processing technique and a valuable tool for monitoring HUT systems in operation under dynamic load functions. For ease of explanation, the term machine learning in this application also includes aspects of generative AI.

[0012] Embodiments of the present application include HUT failure analysis based on sensors that monitor various sources of failure, such as transient voltages (surges, voltage and current imbalances, vibrations, high operating temperatures, HUT overloads, misalignment, etc.). ML models can be used for predictive maintenance of HUTs and for predicting HUT failures. While the following description focuses on motors, the embodiments apply to other types of systems operating on other types of forces, as discussed above. In the following process, the user would typically be a hardware designer, since the analysis performed will affect the design of the HUT. In this application, the term user applies not only to the designer who designs the hardware, but also to the user who monitors the operation of the HUT for maintenance and repair.

[0013] Various industrial and automotive sensors collect analog data to understand and predict the future health of the system. Collecting and analyzing data from these sensors using AI plays a key role in the HUT monitoring challenge. For example, this analysis can provide insights for predictive maintenance by analyzing vibration data from sensors installed on the HUT. Machine learning can detect anomalies and predict potential failures before they escalate. This proactive approach minimizes downtime and reduces maintenance costs. ML algorithms identify patterns in vibration signals and enable the diagnosis of problems such as broken rotor bars, defective bearings, and misalignment. Early detection of these issues allows for timely corrective action. ML leverages historical data to learn the HUT's behavior and provide insights into optimal operating conditions and failure trends. ML algorithms can run on embedded hardware such as field-programmable gate arrays (FPGAs) for real-time monitoring and continuous assessment of the HUT's health. Overall, ML and AI will improve HUT monitoring by predicting failures, optimizing maintenance, and ensuring reliable operation. [Brief explanation of the drawings]

[0014] [Figure 1] Figure 1 shows a block diagram of the motor design workflow. [Figure 2] FIG. 2 shows an embodiment of an experimental setup for testing hardware under test (HUT). [Figure 3] FIG. 3 shows an embodiment of a sensor mounted on an AC motor. [Figure 4] FIG. 4 shows a flow chart of an embodiment of a method for training and selecting a machine learning model. [Figure 5] FIG. 5 shows a flow chart of an embodiment of the feature extraction and model. [Figure 6] Figure 6 shows an example of a vibration signal captured from an AC motor in the time and frequency domains. [Figure 7] Figure 7 shows an embodiment of a user interface for measurement environment configuration. [Figure 8] Figure 8 shows an embodiment of a user interface for configuring a machine learning model. [Figure 9] FIG. 9 shows a flowchart of an embodiment for predicting hardware failures using machine learning. [Figure 10A] FIG. 10A shows a table of average measurements showing the relationship between faults and defects, bearing wear, misalignment, and imbalance. [Figure 10B] FIG. 10B shows that both horizontal and vertical vibrations have a direct relationship with speed, temperature, and armature current. [Figure 11A] FIG. 11A shows the correlation and confusion histograms. [Figure 11B] FIG. 11B shows an example of a three-dimensional histogram of a confusion matrix. [Figure 12] FIG. 12 shows an example of a data acquisition trend of a vibration plot. [Figure 13] FIG. 13 shows an example of a shaft centerline plot of vibration. [Figure 14]FIG. 14 shows an embodiment of a prediction plot. DETAILED DESCRIPTION OF THE INVENTION

[0015] As mentioned above, it is important to test the HUT while it is running on an actual test bench. Figure 1 shows an example of a HUT design workflow. The process begins at 10 by defining the requirements for the specific application of the HUT. The various components used in the HUT are characterized at 12. As mentioned above, the user (in this case, the designer) simulates and optimizes the HUT at 14. The portion of this process, included in box 22, includes live testing, allowing the designer to finalize the design and deploy and install the HUT. Electrical and mechanical testing is performed at 16, followed by design finalization at 18 and deployment at 20. The portion of the process, box 22, benefits from the application of ML.

[0016] For example, a HUT can capture key parameters such as vibration, voltage, current, temperature, acceleration, and strain from sensor outputs while the motor is running under load. By combining ML modeling with AI predictions of future failures, the health of the HUT can be visualized and dynamically analyzed. This represents a new opportunity for the test and measurement industry. Embodiments of the present application can be applied to several industries to improve HUT operation. These improvements include, but are not limited to, efficient and fail-safe industrial drives such as conveyor drives and pumps, electric mobility from cars to drones and electric aircraft, point-of-load (POL) converters for data centers, safe robots and mobile medical robots with long battery life, and efficient, predictive photovoltaic (PV) inverters.

[0017] As will be explained further, users can save these parameters as library files for other HUTs with similar specifications, building in-house knowledge of HUT functionality. An AI assistant (i.e., software employing machine learning) can run a series of operations on the equipment to help predict future failures and improve HUT performance through mechanical parameter analysis. This AI assistant provides the option to interface with an AI model that performs predictive analysis of the HUT.

[0018] This embodiment facilitates HUT testing in box 16 of the workflow in Figure 1. The ML model adapts based on the device's usage during debug / test. Users do not need large data sets to build the model. The data sets used here are much smaller than those typically used because the model dynamically learns from streaming sensor input as a learning process. After integration, users can test the operating HUT for mechanical and environmental parameters using test and measurement equipment such as an oscilloscope. In some embodiments, the test and measurement equipment captures signals from the HUT's sensors for vibration analysis, which is useful for identifying harmonics in the HUT's components. In some embodiments, the test and measurement equipment can also serve as a powerful digitizer for sensor data. Faulty waveforms are saved for ML modeling and AI prediction.

[0019] Various types of sensors, such as vibration sensors, play an important role in identifying potential faults before failure occurs. Vibration sensors come in many forms, including MEMS (microelectromechanical systems) sensors. These sensors are reliable and cost-effective, especially at low frequencies. Slow-rotating machinery benefits from low-frequency monitoring with MEMS sensors. Piezoelectric vibration sensors convert mechanical vibrations into electrical signals. A wide variety of Internet of Things (IoT) industrial sensors are available from a variety of vendors.

[0020] Creating an ML model based on sensor data collected from operational hardware such as a HUT involves several steps. This process involves analyzing the data with test and measurement equipment such as an oscilloscope. This process integrates hardware interfacing, data acquisition, data preprocessing, model development, visualization, and analysis.

[0021] Figure 2 shows an embodiment of a test environment. First, a sensor, such as sensor 30, must be mounted on the HUT 31. As shown by sensor 30, the sensor must be securely attached and positioned to capture the HUT's operating characteristics. The mounted sensor unit 30 provides sensor data, but system 35 can collect this sensor data in a variety of ways. For HUT 31, triaxial sensors, rather than single-axis or radial sensors, may provide more information and potentially lead to better fault detection. Figure 3 shows a triaxial sensor configuration consisting of axial sensor 44, radial sensor 42, and tangential sensor 46. Sensors, such as sensor 30, undergo conditioning to calibrate the sensor and enable detection of sensor failure. Test setups vary, and sensor 30 may capture data during sensor failure.

[0022] A data logger 34 (such as a dedicated data acquisition system or microcontroller) may collect the data. The data logger 34 may then transfer the data to some type of memory 33, such as an SD card or USB drive. The test and measurement equipment 32 may collect the data directly. Much of the data may consist of analog data, and many test and measurement equipment, such as oscilloscopes, may have an analog-to-digital converter to convert this analog data. The test and measurement equipment 32 or data logger 34 may collect or transfer the data to a remote store 40, such as a cloud-based memory. The remote store 40 may also receive data sets from the sensors 30 and results from other components to build the library described above.

[0023] The test and measurement instrument 32 may also include probes (not shown) that can connect to the HUT 31 directly or using a test fixture. The test and measurement instrument 32 may also include one or more processors that analyze data, receive user input, and act based on the user input. The presence of one or more processors and internal memory in the test and measurement instrument 32 also allows ML model and AI analysis functionality 36 to run on the test and measurement instrument 32. Alternatively, or in addition to the ML model and AI analysis functionality being stored in the test and measurement instrument 32, a computing device 38, such as a personal computer, may also store the ML model and AI analysis functionality. Various tasks and components described below may be distributed among the test and measurement instrument 32, the computing device 38, and devices connected via the cloud 40.

[0024] FIG. 4 illustrates the overall process for assembling and analyzing data collected from sensors. In FIG. 4, the HUT 31 is shown operating under normal load conditions and under specific conditions of interest, such as various loads or speeds, at 50. The system can convert the data provided by the sensors 30 into one or more data sets representing the operating range of the HUT 31. As mentioned above, the test and measurement instrument 32 may function as a real-time monitor or may log and store data either locally or remotely. Using the test and measurement instrument 32 as a real-time monitor, a user may benefit from viewing the time-domain waveform of the HUT 31 as it operates. Probes and sensors may be connected to the test and measurement instrument 32 via the cloud, an antenna, or directly.

[0025] After or during data collection, the system 35 (i.e., the combination of the test and measurement equipment 32 and / or computing device 38 and the remote store 40) aggregates the data into one or more data sets at 52, each of which may undergo preprocessing to normalize the data. The one or more data sets may be normalized to eliminate bias, improve feature extraction, and improve the training process. For example, the following normalization process may be applied to remove bias from the collected data sets:

number

[0026] A feature extraction process 54 extracts features from the data related to the operating characteristics of the HUT 31. This feature extraction process is described in more detail with respect to FIG. 5. Next, an ML model is applied to the data set at 56. The model applied to the data set at 45 is either ML model 36 or ML model 38 of FIG. 2, depending on where the model is operating. This model may consist of any of various types of predictive models, including random forests, decision trees, etc., and may be used for predictive maintenance of the HUT 31 to predict potential failures. Model selection may be performed by the user via user input on the test and measurement instrument 32 or computing device 38, or the user may select "AUTO" on the user interface, as described below. The AUTO option involves selecting an appropriate regression model with minimal error for the particular environment in which the HUT 31 operates. Model options include, but are not limited to, regression models such as linear, multivariate, polynomial, decision trees, neural networks, or clustering models.

[0027] This process may utilize various error measures to determine the best model. Two of these are the Mean Square Error (MSE) and the Root Mean Square Error (RMSE). The minimum or lowest error is the criterion used to select the appropriate model. MSE is calculated as follows:

number

number

[0028] Returning to Figure 4, the collected data set is used to train a model at 58. The model is trained to recognize patterns and predict outcomes based on sensor data. This process may occur dynamically as the model receives streaming data from the sensors. The resulting trained model may make predictions at 60, resulting in either a good HUT or a faulty HUT parameter. If the HUT parameters are faulty, maintenance may be performed or the HUT may be deemed to have failed. The model's output may include predictive analytics such as time to failure and predicted maintenance needs. During this process, a test data set is used to evaluate the model's performance. Model parameters may be adjusted as needed to improve accuracy and reduce overfitting (overfitting the training data, resulting in poor prediction accuracy for new data).

[0029] FIG. 5 illustrates an embodiment of the feature extraction and model evaluation process. Sensor data is converted into a training data set as described above. The first step is data preprocessing at 70, such as the normalization / standardization step described above. Features are extracted from the data at 72, and the resulting feature matrix is ​​used to train a model at 74. The upper process in the figure uses unsupervised learning, where the model essentially learns by itself. The lower process utilizes classification, where data is preprocessed at 76 and feature extraction is performed at 78. However, in a classification scheme, feature extraction results in a feature vector. The feature vector is then used to predict the performance of the HUT. These two processes can be used in conjunction to act as quality controls for each other.

[0030] In a self-supervised learning (SSL) setup for HUT fault detection, the model learns meaningful representations (embeddings) from raw sensor data without requiring labeled examples. These embeddings capture patterns in HUT behavior and serve as rich features for downstream tasks. To predict time to failure, a regression model is trained on these embeddings, along with some labeled failure data, to estimate the HUT's remaining useful life (RUL). This approach enables early detection of faults and real-time failure prediction, even in situations with limited labeled data. Failure prediction also takes into account transitions / drifts of condition data into failure / abnormal modes during operation.

[0031] When using sensor outputs to model the behavior or state of a running HUT, extracting appropriate features from data such as 72 or 78 in Figure 5 or 54 in Figure 4 is important for developing accurate and reliable machine learning models.

[0032] A "feature" can include almost anything extracted from the raw data: it can be the raw data itself, a combination of various sensor data, the result of a statistical analysis of multiple measurements such as the mean, variance, standard deviation, or a transformation such as a Fourier transform into the frequency domain.

[0033] Once the user selects features, the model is trained using those features. From then on, each time a process uses that model to make a prediction / classification, the process must extract the same features from new data. By way of example, features may consist of one or more of the following: time-domain measurements, frequency-domain parameters, harmonic analysis results (e.g., using a fast Fourier transform (FFT)), temperature rise (a gradual increase in temperature may indicate friction, misalignment, or electrical issues), and vibration patterns (changes in which may indicate imbalance, bearing failure, or misalignment). Vibration analysis results generally constitute one of the most important features for designers. Vibration analysis can detect various faults, including imbalance, bearing failure, and gear failure, at their early stages, among many other possibilities. Combining vibration data with other sensor outputs, such as temperature, current draw, and acoustic emissions, can provide a more comprehensive understanding of the HUT's condition. Integrating these data sources with advanced machine learning models can significantly enhance predictive maintenance strategies, reducing unplanned downtime and extending the life of HUTs.

[0034] Frequency domain parameters can provide a clearer understanding of complex vibration waveforms that are difficult to understand in the time domain. The test and measurement equipment can convert the model output into a form that the user can visualize on the test and measurement equipment. This conversion may include generating time series predictions or simulating the sensor output under hypothetical conditions. The sensor data can be displayed as a time domain voltage (units: volts / second) with an accompanying frequency domain plot. The time domain sensor information can be converted to frequency domain sensor information in the following manner, as shown in Equation 4: [Number 4] Main peak = max(ABS(FFT(sensor output data))) Figure 6 shows an example of a test and measurement instrument, such as an oscilloscope, showing the model's output in real time. Using the test and measurement instrument's analytical tools, users can gain insight into the HUT's or HUT's behavior and the model's predicted performance. The model can then be adjusted as needed based on these observations.

[0035] FIG. 7 shows a user interface for configuring measurements. User interface 82 on the left side of FIG. 7 shows the options available in one embodiment of this interface. The user selects the type of plot to use from the measurements. In this embodiment of user interface 82, supported plots may include vibration trend, feature trend, predicted trend, shaft centerline plot, correlation / confusion histogram plot, and correlation / confusion tables. User interface 84 on the right side of FIG. 7 shows an example of sensor configuration. In this particular embodiment, the user configures the nature of the shaft sensor used in the HUT analysis, as shown in FIG. 3.

[0036] Figure 8 shows the user interface 86 when a user selects settings for AI-ML modeling. In this example, the user selects time parameters and whether to use live data or data from a file. Generally, after selecting AI-ML modeling on a test and measurement instrument, the proposed application begins collecting data on key features, such as vibration and temperature, along with metadata. The acquired waveforms and data undergo preprocessing and cleaning processes, such as resampling, detrending, smoothing, and filtering, before extracting key features. The extracted features are used as a training set to build a model. Once a sufficient training set is collected, the model dynamically trains in the background. The model undergoes cross-validation with other data sets to verify accuracy and precision. If the model's accuracy is sufficiently high, the user can use it for predicting results in addition to measurement results. This process is repeated for each available model or runs simultaneously.

[0037] FIG. 9 shows a flowchart of an overall embodiment of an ML process for predictive HUT or HUT operation. The process begins at 90 by attaching sensors to the HUT. The HUT then powers up, and data is collected at 92 during operation. The data is then transferred to a test and measurement device at 94, either directly, through a data logger, or from the cloud, as described above. The data may also be sent to a store, such as the cloud, at 96. The data is pre-processed at 98. A model is selected at 100. As described above, the user may select the model, or the system may select the model if the user selects automatic. The model and associated data for the particular HUT and environment may be stored at 102 to build a HUT library. The selected model generates a prediction result at 104, either predictive maintenance or signs of a good or bad HUT. The sensor data used in this cycle is used to update the model and adjust its operation, as needed, at 106.

[0038] As mentioned above, one of the analysis steps for the model includes correlation and confusion matrices. The correlation matrix shows the interdependence of extracted features as scalar values. The correlation matrix is ​​useful for understanding the relationships between features and identifying potential issues such as multicollinearity. Figure 10A shows a table of average measurements showing the relationship between faults and defects, bearing wear, misalignment, and imbalance. Figure 10B shows that both horizontal and vertical vibrations have a direct relationship with speed, temperature, and armature current. The shaded boxes in both Figures 10A and 10B indicate possible causes of failure. For example, if vibration is highly correlated with temperature (i.e., close to 1), it means that higher temperatures increase vibration. In Figure 10B, for example, horizontal and vertical vibrations have a direct relationship with inertia.

[0039] Correlations can also be shown as bar graphs, which are useful for visualizing the interdependence of feature distributions. Figure 11A shows correlation and confusion histograms. Figure 11B shows an example of a 3D histogram of a confusion matrix. The Y-axis in these figures constitutes the magnitude of correlation.

[0040] The confusion histogram in Figure 11B can be used to create a confusion matrix, which generally compares the true negatives, true positives, false negatives, and false positives between predicted and actual results, showing how often the model gets confused between various classes. [Table 2]

[0041] Additionally, the Data Acquisition trend plot visualizes how measured parameters change over time based on the order in which the data was collected. Figure 12 shows a Data Acquisition trend plot of extracted features of vibration data. This plot also shows the user-defined limit lines / masks, which are useful for distinguishing between good and bad HUT behavior.

[0042] Figure 13 shows a centerline plot used to analyze the position of the rotor shaft relative to the bearings. Centerline plots can indicate important issues such as imbalance, bearing stiffness, and fluid instability. Data from the X and Y position sensors construct the plot over time. The centerline plot is an incremental plot, with crosshairs showing the shaft position during operation from a cross-sectional perspective. This plot is useful for understanding mechanical problems that can accumulate in a rotor over time as the shaft position shifts. The X and Y axes represent the output of the two position sensors. The circle represents the bearing gap diameter. The diameter of this circle is the maximum distance the rotor shaft can move horizontally or vertically. This diameter is entered by the user in the plot settings.

[0043] Figure 14 shows an example of a graphical pass / fail. The vibration values ​​against time show that the HUT is damaged after 8 hours of operation. Limit values ​​can be defined for scalar results to insert a pass / fail mask. The plot above shows an example of a dynamic plot showing the results resulting from the model.

[0044] In this way, sensors connected to the HUT can provide data that the ML model uses to predict and determine the HUT's behavior, such as pass / fail, normal or defective, or the need for maintenance. The ML model operates to perform predictive analysis of the motor's performance. Several ML models may be available, and the test and measurement equipment or computing device on which these ML models operate may select the model with the lowest error. These models may be supervised learning models or unsupervised learning models, including self-supervised learning models.

[0045] Aspects of the disclosed technology may operate on specially created hardware, firmware, digital signal processors, or specially programmed general-purpose computers, including processors that operate according to programmed instructions. The terms "controller" or "processor" herein contemplate microprocessors, microcomputers, ASICs, and dedicated hardware controllers, among others. Aspects of the disclosed technology may be implemented with computer-usable data and computer-executable instructions, such as one or more program modules, executed by one or more computers (including a monitoring module) or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc., which, when executed by a processor in a computer or other device, perform particular tasks or implement particular abstract data types. Computer-executable instructions may be stored in computer-readable storage media, such as hard disks, optical disks, removable storage media, solid-state memory, RAM, etc. Those skilled in the art will appreciate that the functionality of the program modules may be combined or distributed as desired in various embodiments. Furthermore, such functionality may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, field programmable gate arrays (FPGAs), etc. Certain data structures may be used to more effectively implement one or more aspects of the disclosed technology, and such data structures are considered within the scope of the computer-executable instructions and computer-usable data described herein.

[0046] The disclosed aspects may, in some cases, be implemented in hardware, firmware, software, or any combination thereof. The disclosed aspects may also be implemented as instructions carried by or stored on one or more computer-readable media, which may be read and executed by one or more processors. Such instructions may be referred to as a computer program product. As used herein, computer-readable media refers to any medium that can be accessed by a computing device. By way of example, and not limitation, computer-readable media may include computer storage media and communication media.

[0047] "Computer storage media" means any medium that can be used to store computer-readable information. By way of example and not limitation, computer storage media may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory and other memory technologies, compact disc read-only memory (CD-ROM), digital video disc (DVD) and other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage and other magnetic storage devices, and any other volatile or nonvolatile, removable or non-removable medium implemented in any technology. "Computer storage media" excludes signals themselves and transitory forms of signal transmission.

[0048] A communication medium means any medium usable for communicating computer-readable information. By way of example, and not limitation, communication media may include coaxial cable, fiber optic cable, air, or any other medium suitable for communicating electrical, optical, radio frequency (RF), infrared, acoustic, or other types of signals. Example

[0049] The following examples are provided to aid in understanding the technology disclosed in this application. Embodiments of the technology may include one or more of the examples described below, and any combination thereof.

[0050] Example 1 is a test and measurement device comprising one or more ports for connecting to hardware under test (HUT), a set of sensors connected to the HUT and the test and measurement device, a display for displaying representations of one or more signals from at least one of the HUTs and one or more sensors in the set of sensors, and one or more processors, the one or more processors being configured to execute a program that causes the one or more processors to perform the following processes: acquire data from the set of sensors; form one or more datasets from the data acquired from the set of sensors; apply one or more machine learning models to the one or more datasets; and receive predictive analysis results regarding the HUT from the one or more machine learning models.

[0051] Example 2 is the test and measurement device of Example 1, wherein the set of sensors connected to the test and measurement device are connected to the test and measurement device via a sensor data collector consisting of either a microcontroller or a data logger.

[0052] Example 3 is the test and measurement device of either Example 1 or Example 2, wherein the program causing the one or more processors to perform processing to acquire data includes a program causing the one or more processors to perform processing to acquire data from either the sensor or a remote store.

[0053] Example 4 is a test and measurement device of any of Examples 1 to 3, wherein the program causing the one or more processors to perform the process of forming one or more data sets includes a program causing the one or more processors to perform the processes of normalizing data and forming one or more training data sets, one or more test data sets, and one or more prediction data sets.

[0054] Example 5 is the test and measurement instrument of any of Examples 1 to 4, wherein the program causing the one or more processors to apply one or more machine learning models to one or more datasets includes a program causing the one or more processors to receive, via a user interface, an input identifying a selected model to apply to the one or more datasets, and apply the selected model to the one or more datasets.

[0055] Example 6 is the test and measurement device of Example 5, wherein the input specifying the selected model includes automatic selection, and the program causing the one or more processors to perform a process of applying a machine learning model includes a program causing the one or more processors to perform a process of applying a machine learning model with the lowest error rate.

[0056] Example 7 is the test and measurement instrument of any of Examples 1 to 6, wherein the one or more processors are further configured to execute a program that causes the one or more processors to dynamically train one or more of the machine learning models during operation using data streamed from the sensors.

[0057] Example 8 is a test and measurement device of any of Examples 1 to 7, wherein the program that causes the one or more processors to apply a machine learning model to the data set includes a program that causes the one or more processors to apply one or more machine learning models to one or more training data sets to train the one or more machine learning models, test the one or more machine learning models using one or more test data sets, and adjust parameters of the multiple machine learning models to improve accuracy of the multiple machine learning models.

[0058] Example 9 is the test and measurement device of Example 8, wherein the program that causes the one or more processors to apply the one or more machine learning models to one or more datasets and train the one or more machine learning models includes a program that causes the one or more processors to perform feature extraction for the one or more datasets of the sensor data.

[0059] Example 10 is the test and measurement system of any of Examples 1 to 9, wherein the one or more processors are further configured to execute a program that causes the one or more processors to generate a plot showing results from the one or more machine learning models.

[0060] Example 11 is a method comprising: acquiring data from a set of sensors connected to hardware under test (HUT); forming one or more datasets from the data acquired from the set of sensors; applying one or more machine learning models to the one or more datasets; and receiving one or more predictions from the one or more machine learning models regarding the likelihood of failure of the HUT.

[0061] Example 12 is the method of example 11, wherein receiving the data includes receiving the data from any of the sensors or from a remote store.

[0062] Example 13 is the method of either example 11 or example 12, wherein forming one or more datasets includes normalizing the data to form one or more training datasets, one or more testing datasets, and one or more prediction datasets.

[0063] Example 14 is the method of any of Examples 11 to 13, wherein applying one or more machine learning models to the one or more datasets includes receiving input via a user interface identifying selected models to apply to the one or more datasets; and applying the selected models to the one or more datasets.

[0064] Example 15 is the method of Example 14, wherein the input specifying the selected model includes automatic selection, and the process of applying the machine learning model includes applying the machine learning model with the lowest error rate.

[0065] Example 16 is the method of any of Examples 11 to 13, further comprising: dynamically training one or more of the machine learning models during operation using data streamed from the sensors.

[0066] Example 17 is any of the methods of Examples 11 to 13, wherein applying the machine learning models to the datasets includes applying the one or more machine learning models to one or more training datasets to train the one or more machine learning models; testing the one or more machine learning models using one or more test datasets; and adjusting parameters of the multiple models to improve accuracy of the multiple models.

[0067] Example 18 is the method of Example 17, wherein the process of applying the one or more machine learning models to the one or more datasets to train the one or more machine learning models includes a process of extracting features from one or more datasets of sensor data.

[0068] Example 19 is the method of any of Examples 11 to 18, wherein receiving one or more predictions from the one or more machine learning models includes converting the predictions from the one or more machine learning models into a format that can be displayed on a display.

[0069] Example 20 is the method of any of examples 11 to 19, further comprising generating a plot illustrating results from the one or more machine learning models.

[0070] All features disclosed in the specification, claims, abstract and drawings, and all steps in any disclosed method or process, may be combined in any combination, except where at least some of such features or steps are mutually exclusive combinations. Each feature disclosed in the specification, abstract, claims and drawings may be replaced by an alternative feature serving the same, equivalent or similar purpose, unless expressly stated otherwise.

[0071] Additionally, the description of this application refers to specific features. It should be understood that the disclosure herein includes all possible combinations of these specific features. When a specific feature is disclosed in connection with a particular aspect or example, that feature can also be used in connection with other aspects and examples, to the extent possible.

[0072] Furthermore, when this application refers to a method having two or more defined steps or processes, these defined steps or processes may be performed in any order or simultaneously, unless the circumstances do not preclude this possibility.

[0073] Although specific embodiments of the invention have been illustrated and described for purposes of illustration, it will be appreciated that various modifications can be made therein without departing from the spirit and scope of the invention. Accordingly, the invention should not be limited except as by the appended claims. [Explanation of symbols]

[0074] 30 sensors 31 Hardware Under Test (HUT) 32 Test and measurement equipment 33 Memory 34 Data Logger 35 Systems 36 ML Models and AI Analytics 38 Computing Devices 40 Cloud 42 Radiation Sensor 44 Axial Sensor 46 Tangential Sensor

Claims

1. 1. A test and measurement device comprising: one or more ports for connecting to hardware under test (HUT); a set of sensors connected to the HUT and the test and measurement device; a display that displays a representation of one or more signals from at least one of the HUTs and one or more sensors in the set of sensors; one or more processors Equipped with the one or more processors acquiring data from said set of sensors; forming one or more data sets from the data acquired from the set of sensors; applying one or more machine learning models to the one or more datasets; receiving predictive analysis results for the HUT from the one or more machine learning models; a test and measurement instrument configured to execute a program that causes the one or more processors to:

2. 10. The test and measurement instrument of claim 1, wherein the set of sensors connected to the test and measurement instrument are connected to the test and measurement instrument via a sensor data acquisition device comprising either a microcontroller or a data logger.

3. 2. The test and measurement instrument of claim 1, wherein the program causing the one or more processors to form one or more data sets includes a program causing the one or more processors to normalize data and to form one or more training data sets, one or more test data sets, and one or more prediction data sets.

4. The program causing the one or more processors to apply one or more machine learning models to one or more datasets comprises: receiving input via a user interface specifying a selected model to apply to one or more datasets; applying the selected model to the one or more datasets; 2. The test and measurement instrument of claim 1, further comprising a program that causes said one or more processors to:

5. 5. The test and measurement instrument of claim 4, wherein the input specifying the selected model includes automatic selection, and the program that causes the one or more processors to perform a process of applying a machine learning model includes a program that causes the one or more processors to perform a process of applying the machine learning model with the lowest error rate.

6. 10. The test and measurement instrument of claim 1, wherein the one or more processors are further configured to execute a program that causes the one or more processors to dynamically train one or more of the machine learning models during operation using data streamed from the sensors.

7. The program causing the one or more processors to apply a machine learning model to the dataset comprises: applying one or more machine learning models to one or more training datasets to train the one or more machine learning models; testing the one or more machine learning models using one or more test data sets; The process of adjusting the parameters of multiple machine learning models to improve their accuracy.

2. The test and measurement instrument of claim 1, further comprising a program that causes said one or more processors to:

8. acquiring data from a set of sensors connected to a hardware under test (HUT); forming one or more data sets from data acquired from the set of sensors; applying one or more machine learning models to the one or more datasets; receiving one or more predictions from the one or more machine learning models regarding the likelihood of failure of the HUT; A method for analyzing hardware under test (HUT), comprising:

9. 9. The method for analyzing hardware under test (HUT) of claim 8, wherein forming one or more data sets comprises normalizing the data and forming one or more training data sets, one or more test data sets, and one or more prediction data sets.

10. 9. The method for analyzing hardware under test (HUT) of claim 8, wherein the process of receiving one or more predictions from the one or more machine learning models includes converting the predictions from the one or more machine learning models into a format that can be displayed on a display.

Images