Edge device for induction motor fault diagnosis

IN604703BActive Publication Date: 2026-09-30INDIAN INSTITUTE OF TECHNOLOGYKHARAGPUR +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
IN202431073896
Authority / Receiving Office
IN · IN
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-12-30
Publication Date
2026-09-30
Estimated Expiration
2044-12-30

AI Technical Summary

Technical Problem

Existing induction motor fault diagnosis systems face challenges with complex hardware setups, high costs, and signal degradation due to noise and dynamic range issues, particularly in variable-speed environments, which affect diagnostic accuracy and scalability.

Method used

A sensor-based edge device with a streamlined configuration using a single current sensor, dynamic gain control, and advanced noise cancellation techniques to enhance signal integrity and accuracy, incorporating a transducer with galvanic isolation, common-mode signal clamping, and adaptive gain adjustment for efficient fault detection in induction motors.

Benefits of technology

The solution simplifies hardware, reduces costs, and enhances diagnostic accuracy by maximizing signal utilization and noise mitigation, enabling reliable fault detection in variable-speed induction motors with real-time processing and user-friendly data analysis tools.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

An edge device for induction motor fault diagnosis is disclosed. The edge device comprises a sensor configured to obtain a current signal from an induction motor and convert the current signal to a voltage signal using a transducer with galvanic isolation. A signal conditioning circuit is coupled to the sensor and configured to process the voltage signal to produce a clamped signal with reduced common mode noise. An automatic gain adjuster is coupled to the signal conditioning circuit and configured to dynamically adjust amplification of the signal. A processor is configured to convert the amplified signal to a digital format using an analog to digital converter. The processor conditions the digital signal to suppress noise using a noise rejecter and a low-pass filter. The processor detects fault frequency components and diagnoses motor faults by performance of spectral analysis and comparison of spectral peak amplitudes with threshold values.
Need to check novelty before this filing date? Find Prior Art

Description

FIELD OF INVENTIONThe present disclosure relates to induction motor fault diagnosis, and more particularly to a sensor-based edge device for induction motor fault diagnosis.BACKGROUNDInduction motors are widely utilized across numerous industrial applications due to their robust construction, reliability, and efficiency. These motors serve as workhorses in manufacturing facilities, processing plants, and various automated systems where continuous operation is expected. The operational health of induction motors directly impacts productivity, safety, and operational costs in industrial environments.Monitoring the condition of induction motors has become an area of increasing interest as industries seek to minimize unplanned downtime and optimize maintenance schedules. Various approaches have been developed to assess motor health, including vibration analysis, thermal imaging, acoustic monitoring, and current signature analysis. Among these techniques, current signature analysis has gained attention because it can detect a range of mechanical and electrical faults by examining the spectral characteristics of the motor stator current.Existing fault diagnosis systems for induction motors often employ multiple sensors to capture different operational parameters. Such multi-sensor configurations can increase system complexity, installation costs, and maintenance requirements. The hardware overhead associated with deploying extensive sensor arrays may present challenges for widespread adoption, particularly in facilities with numerous motors requiring monitoring.Signal acquisition in motor monitoring applications presents technical challenges related to dynamic range utilization and noise management. When motor current varies across different operating conditions, fixed-gain amplification approaches may not fully utilize the available range of analog-to-digital converters, potentially limiting the detection of low-amplitude fault signatures. Additionally, the amplification of weak signals can simultaneously amplify noise present in the measurement system, which may affect the accuracy of subsequent diagnostic analysis. Various automatic gain control (AGC) approaches have been explored in related fields, including AGC circuits for visible light communication systems, analog AGC circuitry for complementary metal-oxide-semiconductor (CMOS) wireless local area network WLAN receivers, digital automatic gain control techniques, AGC circuits for quadrature signal paths, linear gain code interleaved AGC circuits, and automatic gain control systems for analog-to-digital converters.Power supply irregularities and common-mode interference represent additional sources of signal degradation in industrial measurement environments. These noise sources can introduce artifacts into acquired signals, potentially obscuring fault-related frequency components that are relevant for diagnostic purposes. Approaches to address noise in signal acquisition have included noise-cancelling techniques for embedded analog-to-digital converters (ADCs), auto-adjustable low-signal processing techniques using programmable mixed-signal systems-on-chip (SoCs), improved fully differential analog filters, and modeling of power supply noise imbalance in differential low noise amplifiers. Filter design techniques have also been investigated, including methods for improving stop-band attenuation in Sallen-Key low-pass filters, ringing test methods for second-order Sallen-Key filters, and practical methods for designing resistor-capacitor (RC) active filters. Additional noise mitigation approaches have included overvoltage clamping methods for AC systems, signal filtering techniques, background noise removal systems for industrial environments, and power supply transient reduction methods using low dropout voltage regulators with improved power supply rejection.Edge computing architectures have emerged as an approach for processing data locally rather than transmitting raw measurements to centralized servers. Such architectures can reduce latency in fault detection and decrease dependence on network connectivity. However, implementing sophisticated signal processing algorithms on resource-constrained embedded platforms presents design considerations related to computational efficiency and power consumption. Microcontrollers such as the ESP32 have been explored for data processing applications in embedded systems. Industrial data collection and monitoring systems have been developed that address signal conditioning, input channel routing, and intelligent data management for industrial Internet of Things environments.Variable-speed induction motor drives, which are increasingly common in industrial applications, introduce additional complexity to fault diagnosis because the fundamental frequency and associated fault signatures vary with operating speed. Diagnostic systems intended for such applications benefit from adaptive signal processing capabilities that can accommodate changing operational parameters.The field of induction motor fault diagnosis has been extensively studied. Comprehensive reviews and bibliographies have documented fault detection and diagnosis techniques for induction motors. Advanced fault detection techniques for three-phase induction motors have been reviewed, and artificial intelligence applications in fault diagnostics of rotating industrial machines have been surveyed. Acoustic emission-based early fault diagnosis approaches have also been reviewed.Various signal processing techniques have been applied to motor fault detection. These include high-resolution rotor fault diagnosis based on stator current signature analysis, condition monitoring using discrete wavelet transform and artificial neural networks, and bearing fault diagnosis based on singular value decomposition of stator current. Broken rotor bar detection has been investigated using discrete-time Fourier transform (DTFT) and autoregressive-based (AR-based) spectrum methods at lower sampling rates, as well as through density of maxima approaches for low slip and variable load conditions. Comparative studies have examined DWT and MUSIC techniques for broken rotor bar detection. Additional approaches have included inverse thresholding applied to spectrograms, adaptive wavelet packet transform for multi-fault diagnosis, successive variational mode decomposition for rotor broken bar fault diagnosis, and matrix pencil methods for bearing fault detection.Spectral estimation techniques have been applied to motor current analysis, including covariance matrix fitting methods, random matrix theory-based approaches, improved multiple-Toeplitz matrices reconstruction methods, MUSIC algorithm variants, and sinusoidal parameter estimation techniques. Machine learning methods have been applied for rotor fault characterization under different load levels, and selective features combined with machine learning have been used for bearing fault diagnosis.Current signature analysis methods for motor fault detection have been developed, including approaches for fault detection in induction machines, rotor bar fault detection, generator fault detection, and diagnostic methods for rotary machines. Specific techniques have addressed fault detection in induction motors, methods for determining AC motor faults in non-variable frequency devices, broken rotor bar detection based on current signature analysis, and graph-based motor current signature analysis for broken-rotor bar detection in inverter-fed induction motors.OBJECTIVE OF THE INVENTIONOne object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that includes a streamlined sensor configuration and leverages sophisticated algorithms and signal processing techniques to effectively monitor and diagnose faults in the induction motor using data from a single current sensor.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that simplifies the hardware setup and reduces costs, making the edge device a more affordable solution for industrial applications.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that enhances signal strength for precise diagnostics and condition monitoring and includes real-time processing to identify faults in variable speed induction motor drives.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that implements an optimized common mode signal clamping technique to minimize common mode voltage variations and maintain signal integrity for accurate data capture.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that develops an adaptive gain control mechanism that dynamically adjusts the gain to maximize the analog to digital converter's range utilization.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that identifies and characterizes various sources of noise that can impact the accuracy of the current signature measurement and implements advanced noise cancellation techniques to enhance diagnostic accuracy.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that integrates automatic gain stage and noise mitigation techniques into a cohesive data acquisition system that can amplify weak signals while effectively mitigating noise.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that validates performance in real-world industrial settings under various operating conditions to ensure reliability and accuracy.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that develops user-friendly software tools for data analysis and visualization that provide industrial operators with intuitive interfaces for monitoring motor conditions, analyzing captured data, and generating diagnostic reports.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that ensures compatibility with existing industrial infrastructure and scalability to different types of induction motors and operational environments.Another object of the present disclosure is to provide a sensor-based edge device for induction motor fault diagnosis that includes advanced communication capabilities including both Wi-Fi and Ethernet to enable efficient function in a wide range of IoT applications and industrial environments.SUMMARYThis summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.According to an aspect of the present disclosure, an edge device for induction motor fault diagnosis is provided. The edge device includes a sensor configured to obtain a current signal from an induction motor and convert the current signal to a voltage signal using a transducer with galvanic isolation. The edge device further includes a signal conditioning circuit coupled to the sensor and configured to process the voltage signal to produce a clamped signal with reduced common mode noise. The edge device further includes an automatic gain adjuster coupled to the signal conditioning circuit and configured to dynamically adjust amplification of the signal. The edge device further includes a processor configured to convert the amplified signal to a digital format using an analog to digital converter, condition the digital signal to suppress noise using a noise rejecter and a low-pass filter, and detect fault frequency components and diagnose motor faults by performance of spectral analysis and comparison of spectral peak amplitudes with threshold values.According to other aspects of the present disclosure, the edge device may include one or more of the following features. The transducer may comprise a split core transducer configured to effectively separate a primary circuit responsible for power transmission from a secondary circuit dedicated to the measurement process. The sensor may further comprise a resistor with selective value configured to perform current-to-voltage conversion and establish a desired amplitude for output voltage signal according to the current drawn by the motor.According to other aspects of the present disclosure, the signal conditioning circuit may comprise a common-mode signal clamping circuit including an operational amplifier based peak detector configured to detect a peak of the voltage signal and store the positive peak as a voltage value in a capacitor, and a summer configured to generate a positively clamped signal that involves the common mode signal component upon detection of the peak by the peak detector. A diode may be connected to the capacitor and configured to become active with rise of amplitude of the signal from the sensor to cause the capacitor to charge and capture the larger peak amplitude, and when the peak amplitude decreases, the diode may remain inactive until the capacitor discharges to the decreased level, whereby the operational amplifier based peak detector serves as a comparator to facilitate the process. The capacitor may be configured to undergo discharge via an input bias current of a unity gain buffer operational amplifier configured to remove loading effect of the capacitor and a summer operational amplifier configured to generate the positively clamped signal, whereby the operational amplifiers set a total discharging current dependent on capacitance value of the capacitor and a maximum rate of voltage reduction to fix the device ability to detect a maximum motor current degradation rate.According to other aspects of the present disclosure, the automatic gain adjuster may comprise a digital potentiometer, a microcontroller configured to operate the digital potentiometer, and operational amplifiers configured to maximize utilization of converting range of the analog to digital converter. The automatic gain adjuster may further comprise an operational amplifier based inverting amplifier configured to receive the clamped signal and output an amplified inverting version of the clamped signal with a controllable gain via the microcontroller operated digital potentiometer, an additional inverting amplifier with unity gain configured to convert the amplified inverting version to a positive form, and a peak detector configured to monitor the amplified signal for the purpose of control of the digital potentiometer and management of discharge of a second capacitor through a transistor. The microcontroller may be configured to control the gain in accordance with a range of the analog to digital converter by initiation of a gain control routine with the digital potentiometer set to an initial value, setting of a digital logic signal at a transistor base resistance to a low state to place the transistor in cutoff mode and enable a peak detector capacitor to charge during a signal cycle, sampling of a peak voltage across the peak detector capacitor to obtain a feedback value representative of a maximum amplitude of an input signal, comparison of the sampled peak voltage against an upper threshold and a lower threshold to determine a gain adjustment, discharge of the peak detector capacitor by setting of the digital logic signal at the transistor base resistance to a high state to turn on the transistor followed by return of the logic signal to the low state to reset the capacitor for the next measurement cycle, and repetition of the process in a continuous loop within firmware to maintain the peak voltage within a hysteresis band defined by the upper and lower thresholds and thereby stabilize automatic gain control while prevention of oscillations due to noise.According to other aspects of the present disclosure, the noise rejecter may comprise an operational amplifier based power supply noise rejecter configured to cancel common-mode noise and power supply noise, and the low-pass filter may comprise a Sallen-Key low-pass filter configured to pass primary frequency components below 159 Hz to examine the current signature of a variable-speed induction motor. The processor may be configured to detect sideband fault frequency components of healthy as well as faulty motor current using at least one of Fast Fourier Transform, Power Spectral Density, or MUSIC algorithm and calculate location of characteristic fault frequency components in stator current spectrum of the motor, and the processor may be configured to identify faulty or healthy motor by determination of peak value in the frequency spectrum and comparison with a threshold value.According to other aspects of the present disclosure, the edge device may further comprise a dual-rail power supply including a switched-mode power supply and low-dropout regulators configured to minimize electrical noise, and a communication interface including at least one of a Wi-Fi unit for wireless communication over a local network or internet, or a TCP / IP Ethernet unit for wired communication, whereby the Wi-Fi and / or TCP / IP Ethernet units meet diverse communication needs of modern IoT and industrial applications.According to other aspects of the present disclosure, the processor may comprise a single-board computer configured to process stator current signal as acquired from the signal conditioner by estimation of rotor speed from the stator current, estimation of fundamental frequency in the stator current signal and removal of fundamental component from the stator current signal by use of Extended Kalman Filter to ensure that remaining stator current signal primarily contains harmonic components that may indicate the faults, calculation of slip using the estimated rotor speed and the fundamental frequency, determination of autocorrelation matrix size based on the fundamental frequency and slip, formation of data matrix using the stator current and the autocorrelation matrix size, determination of signal and noise subspace sizes using eigen values of signal and noise subspaces and their respective eigenvalue matrices, estimation of frequencies and magnitudes of remaining harmonic components including closely spaced harmonics linked with incipient faults by use of the MUSIC algorithm, and classification of fault types based on detected harmonic components and their magnitudes.According to another aspect of the present disclosure, a method for induction motor fault diagnosis is provided. The method includes obtaining a current signal from an induction motor using a sensor comprising a transducer 103 with galvanic isolation and converting the current signal to a voltage signal. The method further includes processing the voltage signal using a signal conditioning circuit to produce a clamped signal with reduced common mode noise. The method further includes dynamically adjusting amplification of the signal using an automatic gain adjuster to maximize utilization of converting range of an analog to digital converter. The method further includes converting the amplified signal to a digital format using the analog to digital converter. The method further includes conditioning the digital signal to suppress noise using a noise rejecter and a low-pass filter. The method further includes detecting fault frequency components to diagnose motor faults by performing spectral analysis and comparing spectral peak amplitudes with threshold values.The foregoing general description of the illustrative embodiments and the following detailed description thereof are merely exemplary aspects of the teachings of this disclosure and are not restrictive.BRIEF DESCRIPTION OF FIGURESThese and other features, aspects, and advantages of the present invention will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:FIG. 1 illustrates a block diagram of an edge device for induction motor fault diagnosis, according to aspects of the present disclosure;FIG. 2 illustrates a graphical depiction of signal synthesis for the edge device for induction motor fault diagnosis, according to aspects of the present disclosure;FIG. 3 illustrates a current sensor configuration comprising a schematic circuit diagram and a transducer, according to aspects of the present disclosure;FIG. 4 illustrates a block diagram of the process of generating a positive clamped signal from an input sinusoidal waveform, according to aspects of the present disclosure;FIG. 5 illustrates a schematic of a common mode signal clamping circuit for the edge device, according to aspects of the present disclosure;FIG. 6 illustrates a block diagram of an automatic gain adjuster, according to aspects of the present disclosure;FIG. 7 illustrates a schematic of the automatic gain adjuster for the edge device, according to aspects of the present disclosure;FIG. 8 illustrates a flowchart of a gain control mechanism implemented by a microcontroller, according to aspects of the present disclosure;FIG. 9 illustrates a block diagram of the operation of a differential amplifier for power supply noise rejection, according to aspects of the present disclosure;FIG. 10 illustrates a schematic of a signal conditioning circuit for the edge device, according to aspects of the present disclosure; FIG. 11 illustrates a comprehensive schematic of an enhanced current signature acquisition system for the edge device, according to aspects of the present disclosure;FIG. 12 illustrates a frequency spectrum graph showing detection of broken rotor bar fault signatures, according to aspects of the present disclosure;FIG. 13 illustrates multiple views of the edge device for induction motor fault diagnosis, according to aspects of the present disclosure;FIG. 14 illustrates a hardware communication interface between a single board computer and the microcontroller, according to aspects of the present disclosure;FIG. 15 illustrates oscilloscope waveforms showing signal characteristics at different stages of the current signature acquisition system, according to aspects of the present disclosure;FIG. 16 illustrates a logic analyzer capture showing serial communication waveforms for the edge device, according to aspects of the present disclosure;FIG. 17 illustrates stator current characteristics including time domain waveform and frequency domain analysis, according to aspects of the present disclosure;FIG. 18 illustrates oscilloscope waveforms and measurement data for motor current signatures under different fault conditions, according to aspects of the present disclosure;FIG. 19 illustrates experimental equipment and diagnostic output for the induction motor fault diagnosis, according to aspects of the present disclosure;FIG. 20 illustrates data logger validation showing sequential logging readings files, according to aspects of the present disclosure;FIG. 21 illustrates frequency domain analysis plots showing exact fundamental frequency estimation, according to aspects of the present disclosure;FIG. 22 illustrates a frequency spectrum plot showing slot harmonic estimation, according to aspects of the present disclosure; andFIG. 23 illustrates frequency spectrum plots showing broken rotor fault detection through current signature analysis, according to aspects of the present disclosure.DETAILED DESCRIPTIONFor the purpose of promoting an understanding of the principles of the present disclosure, reference will now be made to the various embodiments, and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the present disclosure is thereby intended, such alterations and further modifications in the illustrated system, and such further applications of the principles of the present disclosure as illustrated therein, being contemplated as would normally occur to one skilled in the art to which the present disclosure relates.It will be understood by those skilled in the art that the foregoing general description and the following detailed description are explanatory of the present disclosure and are not intended to be restrictive thereof. Whether or not a certain feature or element was limited to being used only once, it may still be referred to as "one or more features" or "one or more elements," "at least one feature," or "at least one element." Furthermore, the use of the terms "one or more" or "at least one" feature or element does not preclude there being none of that feature or element, unless otherwise specified by limiting language, including, but not limited to, "there needs to be one or more…" or "one or more elements are required."Reference is made herein to some "embodiments." It should be understood that an embodiment is an example of a possible implementation of any features and / or elements of the present disclosure. Some embodiments have been described for the purpose of explaining one or more of the potential ways in which the specific features and / or elements of the proposed disclosure fulfill the requirements of uniqueness, utility, and non-obviousness.Use of the phrases and / or terms including, but not limited to, "a first embodiment," "a further embodiment," "an alternate embodiment," "one embodiment," "an embodiment," "multiple embodiments," "some embodiments," "other embodiments," "further embodiment", "furthermore embodiment", "additional embodiment" or other variants thereof do not necessarily refer to the same embodiments. Unless otherwise specified, one or more particular features and / or elements described in connection with one or more embodiments may be found in one embodiment, or may be found in more than one embodiment, or may be found in all embodiments, or may be found in no embodiments. Although one or more features and / or elements may be described herein in the context of only a single embodiment, or in the context of more than one embodiment, or in the context of all embodiments, the features and / or elements may instead be provided separately or in any appropriate combination or not at all. Conversely, any features and / or elements described in the context of separate embodiments may alternatively be realized as existing together in the context of a single embodiment.Any particular and all details set forth herein are used in the context of some embodiments and therefore should not necessarily be taken as limiting factors to the proposed disclosure. The terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process or method that comprises a list of steps does not include only those steps but may include other steps not expressly listed or inherent to such process or method. Similarly, one or more devices or sub-systems or elements or structures or components proceeded by "comprises... a" does not, without more constraints, preclude the existence of other devices or other sub-systems or other elements or other structures or other components or additional devices or additional sub-systems or additional elements or additional structures or additional components.The term "couple" and the derivatives thereof refer to any direct or indirect communication between two or more elements, whether or not those elements are in physical contact with each other. The terms "transmit", "receive", and "communicate", as well as the derivatives thereof, encompass both direct and indirect communication. The term "or" is an inclusive term meaning "and / or". The phrase "associated with," as well as derivatives thereof, refer to include, be included within, interconnect with, contain, be contained within, connect to or with, couple to or with, be communicable with, cooperate with, interleave, juxtapose, be proximate to, be bound to or with, have, have a property of, have a relationship to or with, or the like. The term "controller" refers to any device, system, or part thereof that controls at least one operation. The functionality associated with any particular controller may be centralized or distributed, whether locally or remotely. The phrase "at least one of," when used with a list of items, means that different combinations of one or more of the listed items may be used, and only one item in the list may be needed. For example, "at least one of A, B, and C" includes any of the following combinations: A, B, C, A and B, A and C, B and C, and A and B and C, and any variations thereof. As an additional example, the expression "at least one of a, b, or c" may indicate only a, only b, only c, both a and b, both a and c, both b and c, all of a, b, and c, or variations thereof. Similarly, the term "set" means one or more. Accordingly, the set of items may be a single item or a collection of two or more items.Moreover, multiple functions described below may be implemented or supported by one or more computer programs, each of which is formed from computer-readable program code and embodied in a computer-readable medium. The terms "application" and "program" refer to one or more computer programs, software components, sets of instructions, procedures, functions, objects, classes, instances, related data, or a portion thereof adapted for implementation in a suitable computer-readable program code. The phrase "computer-readable program code" includes any type of computer code, including source code, object code, and executable code. The phrase "computer-readable medium" includes any type of medium capable of being accessed by a computer, such as Read Only Memory (ROM), Random Access Memory (RAM), a hard disk drive, a Compact Disc (CD), a Digital Video Disc (DVD), or any other type of memory. A "non-transitory" computer-readable medium excludes wired, wireless, optical, or other communication links that transport transitory electrical or other signals. A non-transitory computer-readable medium includes media where data may be permanently stored and media where data may be stored and later overwritten, such as a rewritable optical disc or an erasable memory device.As used herein, the term "processor" may include a unit implemented in hardware, software, or firmware, and may interchangeably be used with other terms, for example, "logic," "logic block," "part," or "circuitry." A processor may be a single integral component, or a minimum unit or part thereof, adapted to perform one or more functions. For example, according to an embodiment of the disclosure, the processor may be implemented in a form of an application-specific integrated circuit (ASIC), a single-board computer, a microcontroller, or a combination thereof.As used herein, the term "sensor" refers to a sensing configuration that utilizes one current sensor to obtain a current signal from an induction motor, as opposed to multi-sensor configurations that employ multiple sensors to capture different operational parameters.As used herein, the term "transducer " refers to a device configured to convert one form of energy to another, and in the context of the present disclosure, refers to a current transducer configured to convert motor current to a proportional output signal.As used herein, the term "galvanic isolation" refers to the separation of functional sections of electrical systems to prevent current flow between them while still permitting energy or information exchange, typically achieved through electromagnetic, optical, or other non-conductive means.As used herein, the term "signal conditioning circuit" refers to circuitry configured to process an electrical signal to prepare it for subsequent stages of processing, and may include clamping circuits, filters, amplifiers, and noise rejection circuits.As used herein, the term "automatic gain adjuster" refers to a circuit or system configured to dynamically adjust the amplification of a signal based on feedback, typically to maintain the signal within a desired amplitude range.As used herein, the term "diagnostic module" refers to a functional unit, implemented in hardware, software, firmware, or a combination thereof, configured to analyze processed signals and determine the presence or absence of fault conditions.As used herein, the term "sideband fault frequency components" refers to frequency components that appear in the stator current spectrum at frequencies offset from the fundamental supply frequency, which are indicative of various motor fault conditions such as broken rotor bars.Referring to FIG. 1, in an embodiment, a block diagram of the edge device 100 for induction motor fault diagnosis is illustrated, showing the interconnection of the main components. The edge device 100 includes a sensor 102 configured to obtain a current signal from an induction motor 101 and convert the current signal to a voltage signal using a transducer 103 with galvanic isolation. The edge device 100 further includes a signal conditioning circuit 104 coupled to the sensor 102 and configured to process the voltage signal to produce a clamped signal with reduced common mode noise. The edge device 100 further includes an automatic gain adjuster 105 coupled to the signal conditioning circuit 104 and configured to dynamically adjust amplification of the signal. The edge device 100 further includes a processor 106 configured to convert the amplified signal to a digital format using an analog to digital converter 107, condition the digital signal to suppress noise using a noise rejecter 108 and a low-pass filter 109, and detect fault frequency components and diagnose motor faults by performance of spectral analysis and comparison of spectral peak amplitudes with threshold values using a spectral analysis module 110.In an embodiment, the sensor 102 receives a current signal from the induction motor 101. The transducer 103 with galvanic isolation within the sensor 102 converts the current signal to a voltage signal. The voltage signal is output from the sensor 102 to the signal conditioning circuit 104.In an embodiment, the signal conditioning circuit 104 processes the voltage signal to produce a clamped signal with reduced common mode noise. The clamped signal is output from the signal conditioning circuit 104 to an automatic gain adjuster 105.In an embodiment, the automatic gain adjuster 105 dynamically adjusts amplification of the clamped signal to produce an amplified signal. The amplified signal is output from the automatic gain adjuster 105 to the processor 106.In an embodiment, the processor 106 receives the amplified signal and converts the amplified signal to a digital format using the analog to digital converter 107. The processor 106 conditions the digital signal to suppress noise using the noise rejecter 108 and the low-pass filter 109. The processor 106 detects fault frequency components and diagnoses motor faults by performance of spectral analysis and comparison of spectral peak amplitudes with threshold values using the spectral analysis module 110. The processor 106 outputs a fault diagnosis result indicating whether the induction motor 101 is healthy or faulty based on the detected fault frequency components.In an embodiment, the edge device 100 may further comprise a dual-rail power supply including a switched-mode power supply and low-dropout regulators configured to minimize electrical noise. In an embodiment, the edge device 100 may further comprise a communication interface 111 including at least one of a Wi-Fi unit for wireless communication over a local network or internet, or a TCP / IP Ethernet unit for wired communication, whereby the Wi-Fi and / or TCP / IP Ethernet units meet diverse communication needs of modern IoT and industrial applications.Induction motor monitoring through the analysis of current signature is a powerful tool for ensuring the reliability, efficiency, and safety of industrial operations. The amplitude of the current signature is contingent upon the electrical current consumed by the motor. As a result, utilizing the entire range of the analog to digital converter 107 with a fixed gain amplifier is not possible. By maximizing the analog to digital converter 107 range with the automatic gain adjuster 105, a broader spectrum of information can be captured, which is beneficial for diagnosing and monitoring electrical and mechanical systems. With a wider range, the system becomes more sensitive to variations in the current signature. Consequently, minor anomalies that occur in the early stages of a fault or degradation can be detected more effectively when using the full analog to digital converter 107 range.Amplifying the current signature through an automatic gain stage enhances the strength or amplitude of a weak signature. The amplification, however, comes with a caveat: not only does the amplification boost the desired signal, but the amplification also amplifies any accompanying noise that exists within the signal and system. The accuracy and dependability of the measured current can be impacted by a variety of sources of noise. The present disclosure addresses the challenge by denoising the signal to improve the accuracy of subsequent analysis or measurements.Referring to FIG. 2, in an embodiment, a graphical depiction of signal synthesis 200 for the edge device 100 for induction motor fault diagnosis illustrates the signal processing flow corresponding to the block diagram of FIG. 1. FIG. 2 shows a signal synthesis 200 that incorporates an automatic dynamic gain stage for signal amplification while effectively mitigating the accompanying noise.On the left side of FIG. 2, a current signal with varying amplitudes is shown, with amplitudes of current sensor signals indicated in millivolts, including values of 200 mV, 300 mV, and 150 mV relative to a ground level. The current signal represents the raw current signal obtained from the induction motor 101 by the sensor 102 shown in FIG. 1.In an embodiment, the signal then passes through a dynamic gain stage, corresponding to the automatic gain adjuster 105 shown in FIG. 1, which produces a 3.1 Volt fixed amplitude signal with noise. The dynamic gain stage adjusts the amplification of the signal to maximize utilization of the analog to digital converter 107 range. The amplitudes of auto clamped signals are shown as peak to peak values including 400 mV, 600 mV, and 300 mV.In an embodiment, following the dynamic gain stage, the signal enters an analog signal processing block, corresponding to the noise rejecter 108 and low-pass filter 109 shown in FIG. 1, where the filtered signal with amplitude is processed to use the entire range of the analog to digital converter 107. The analog signal processing block performs noise mitigation and signal conditioning to prepare the signal for digital conversion.In an embodiment, the processed signal then passes to the analog to digital converter 107 operating with a 3.3 Volt reference voltage. The output of the system is a digital signal that can be analyzed by the spectral analysis module 110 to detect sideband fault frequency components and identify motor faults based on the processed current signature.FIG. 3 illustrates a current sensor configuration 300 comprising a schematic circuit diagram and the transducer 103, according to aspects of the present disclosure. In particular, in an embodiment, an active current to voltage extractor 300 corresponding to the sensor 102 and the transducer 103 with galvanic isolation shown in FIG. 1 is illustrated. FIG. 3 depicts a current sensor configuration comprising two main elements: a schematic circuit diagram on the left and a physical representation of the transducer 103 on the right. In an embodiment, the transducer 103 may be a split-core current transducer.The schematic shows the current sensor output labeled as IOUT connected through a resistor R to an operational amplifier configured in an inverting arrangement, with the output designated as Vo. The operational amplifier symbol includes the standard triangular representation with positive and negative supply rails indicated.In an embodiment, the physical device shown is a TT 100-SD split-core transducer 103, which is a clamp-style current sensor with an aperture through which a conductor carrying the primary current can be passed. Below the image, an electrical data table specifies the primary nominal current rating as 100 amperes RMS and the corresponding output current as 33.33 milliamperes.In an embodiment, the transducer 103 with galvanic isolation shown in FIG. 1 may comprise the split core transducer 103 configured to effectively separate a primary circuit responsible for power transmission from a secondary circuit dedicated to the measurement process. The split core transducer 103 is specifically crafted for electronically measuring alternating motor current waveforms. The transducer 103 integrates galvanic isolation, effectively separating the primary circuit from the secondary circuit. The split-core design allows for non-invasive installation around existing conductors without requiring circuit interruption, making the split core transducer 103 suitable for industrial motor monitoring applications.In an embodiment, the sensor 102 further comprises a resistor R with selective value configured to perform current-to-voltage conversion and establish a desired amplitude for output voltage signal according to the current drawn by the motor. The output voltage VO of the extractor may be expressed as:VO = -IOUTR…(1)where IOUT is the current signal and R is the resistor for the extractor.In an embodiment, with the transducer 103 TT 100-SD having a maximum allowable output voltage amplitude of 7.5 volts, a resistor R = 100O is selected for the current-to-voltage conversion. When the motor draws a current of 100 amperes, the resulting maximum output voltage amplitude is 3.33 volts, well within the 7.5-volt limit. The design of the automatic gain adjuster 105 is intricately tied to the selection of resistor R, emphasizing the critical role of resistor R in maintaining the output voltage within the specified limits.Referring to FIG. 4, in an embodiment, a block diagram 400 illustrates the process of generating a positive clamped signal from an input sinusoidal waveform, which is performed by the signal conditioning circuit 104 shown in FIG. 1. The single-ended analog to digital converter 107 in use has a limitation, converting only positive analog signals within the 3.3V to 0V range. However, the current sensor generates an alternating analog current signal proportional to the motor's electrical consumption. Consequently, there is a need to convert the current signal into a voltage signal. The voltage signal requires clamping above ground level, with minimal common mode signal, as excessive common mode signal could be unintentionally amplified due to dynamic gain amplification.As shown in the FIG. 4, to diminish the common mode signal, a peak detector is employed. Upon detecting the peak of the voltage signal, as expressed by Eq.(1), a summer is used to generate a positively clamped signal (as expressed in Eq. 2), minimizing the common mode component. Since the optimized clamped signal lacks a negative half-cycle, an Automatic Gain Adjuster can be applied effectively. Vclamped = Vosin(ωt) + Vo (2)On the left side of FIG. 4, an input waveform graph shows a sinusoidal signal represented as Vosin(ωt) oscillating symmetrically above and below the zero reference line over time. The input signal is fed into two parallel paths.In an embodiment, the first path directs the signal through a Peak Detector block, which identifies and extracts the peak amplitude Vo of the sinusoidal waveform. The second path carries the original sinusoidal signal Vosin(ωt) directly forward.In an embodiment, both paths converge at a Summer block, which combines the detected peak value Vo with the original sinusoidal signal Vosin(ωt). The output of the Summer block produces the Positive Clamped Signal, mathematically expressed as:Vclamped = Vosin(ωt) + Vo...(2)The resulting output waveform is displayed on the right side of FIG. 4, showing a sinusoidal signal that has been shifted entirely above the zero reference line, with the waveform now oscillating between zero and twice the peak amplitude. The positive clamping technique ensures that the entire signal remains in the positive voltage domain, which is essential for subsequent processing by single-ended analog-to-digital converters that can only convert positive analog signals within a specified voltage range. The Positive Clamped Signal corresponds to the clamped signal shown in FIG. 1.Referring to FIG. 5, in an embodiment, a schematic of a common mode signal clamping circuit 500 implementing the signal conditioning circuit 104 and the block diagram of FIG. 4 is illustrated. The signal conditioning circuit 104 shown in FIG. 1 comprises a common-mode signal clamping circuit including an operational amplifier based peak detector configured to detect a peak of the voltage signal and store the positive peak as a voltage value in a capacitor C1, and a summer configured to generate a positively clamped signal that involves the common mode signal component upon detection of the peak by said peak detector.In an embodiment, the circuit includes a current transformer connector P2 that receives the current signal from the induction motor 101. The current transformer connector P2 is connected to a first operational amplifier U4.1, which actively converts the current to voltage. A calibration connector J6 and a calibration resistor R4 are provided in the input path for calibration purposes, along with a calibration connector J5 and resistor R5 in the feedback path of the circuit.In an embodiment, the first operational amplifier U4.1 outputs a voltage signal that is fed to a second operational amplifier U4.2, which functions as a peak detector corresponding to the Peak Detector block shown in FIG. 4. The second operational amplifier U4.2 has a non-inverting input terminal 5, an inverting input terminal 6, and an output terminal 7.In an embodiment, a diode D1 is connected between the output terminal 7 and a capacitor C1. The diode D1 connected to the capacitor C1 becomes active with rise of amplitude of the signal from the sensor 102 to cause the capacitor C1 to charge and capture the larger peak amplitude. When the peak amplitude decreases, the diode D1 remains inactive until the capacitor C1 discharges to the decreased level, whereby the operational amplifier based peak detector serves as a comparator to facilitate the process.In an embodiment, a resistor R6 is connected between the capacitor C1 and ground. The third operational amplifier U4.3 has a third operational amplifier positive input pin 10, a third operational amplifier negative input pin 9, and a third operational amplifier output terminal 8. The third operational amplifier U4.3 functions as a unity gain buffer that removes the loading effect of the capacitor C1.In an embodiment, the capacitor C1 is configured to undergo discharge via an input bias current of a unity gain buffer operational amplifier U4.3 configured to remove loading effect of the capacitor C1 and a summer operational amplifier U4.4 configured to generate the positively clamped signal, whereby the operational amplifiers U4.3 and U4.4 set a total discharging current dependent on capacitance value of the capacitor C1 and a maximum rate of voltage reduction to fix the device ability to detect a maximum motor current degradation rate.In an embodiment, the fourth operational amplifier U4.4 serves as a summer corresponding to the Summer block shown in FIG. 4 to generate a positively clamped signal by minimizing the common mode component upon detection of the peak. The fourth operational amplifier U4.4 has a fourth operational amplifier positive input pin 12, a fourth operational amplifier negative input pin 13, and a fourth operational amplifier output terminal 14. Resistors R7, R8, and R9 are connected in the summing configuration of the fourth operational amplifier U4.4. A zener diode ZD3 is connected at the output to provide voltage protection.In an embodiment, the common-mode signal clamping circuit 500 is configured to clamp the output voltage signal above a ground level with a common mode signal that avoids unintentional dynamic gain amplification of excessive common mode signal.FIG. 5 also shows waveforms illustrating the signal processing capabilities, including healthy motor waveform 1, half broken rotor bar waveform 2, and full broken rotor bar waveform 3, demonstrating the circuit's ability to process signals from motors in various health conditions.Accordingly, as shown in FIG. 5, when the peak amplitude of the CT rises, diode D1 becomes active, causing the capacitor to rapidly charge and capture the larger peak amplitude for an optimized clamping process. Conversely, when the peak amplitude of the CT decreases, diode D1 remains inactive until the capacitor discharges at the decreased level. In this scenario, the opamp2 serves as a comparator to facilitate this process.q=CV ,q is total capacitor charge (3)dq / dt=C dV / dt ,V is capacitor voltage (4)i=C dV / dt ,C is capacitor value (5)dV / dt=i / C ,i is sum of input bias current of opamp 2 3 (6)Referring to FIG. 6, in an embodiment, a block diagram 600 of the automatic gain adjuster corresponding to the automatic gain adjuster 105 shown in FIG. 1 is illustrated. The diagram shows the signal flow path beginning with a clamped voltage signal (VClamped) entering from the left side, corresponding to the clamped signal output from the signal conditioning circuit 104.In an embodiment, the clamped voltage signal is fed into a Digitally Controlled Potentiometer block, which receives control inputs from a Microcontroller Unit (MCU) for Control Potentiometer shown at the bottom of the diagram. The output from the Digitally Controlled Potentiometer proceeds to an Amplifier block, represented by a triangular symbol indicating signal amplification.In an embodiment, the Amplifier produces a fixed output voltage (VFixed) on the right side of the diagram, corresponding to the amplified signal shown in FIG. 1. A Feedback Loop is shown connecting from the Amplifier output back to the Digitally Controlled Potentiometer input, depicted by arrows flowing through a block labeled Feedback Loop at the top of the diagram.In an embodiment, the Amplifier output feeds into a Peak Detector block, which monitors the amplitude of the auto-gained signal. The Peak Detector output connects to the MCU for Control Potentiometer, completing the automatic gain control feedback mechanism. The configuration enables dynamic adjustment of signal amplification based on the detected peak voltage, ensuring optimal utilization of the analog to digital converter 107 range while maintaining signal integrity for accurate motor current signature analysis.Referring to FIG. 7, in an embodiment, a schematic 700 of the automatic gain adjuster implementing the block diagram of FIG. 6 and corresponding to the automatic gain adjuster 105 shown in FIG. 1 is illustrated. The automatic gain adjuster 105 comprises a digital potentiometer U5, a microcontroller U8 configured to operate the digital potentiometer U5, and operational amplifiers configured to maximize utilization of converting range of the analog to digital converter 107.In an embodiment, the digital potentiometer U5 is a microcontroller-operated component that serves as the core element for controlling the gain factor. The digital potentiometer U5 receives control signals from the microcontroller U8, which is connected to a crystal oscillator X1 for timing purposes. The microcontroller U8 includes multiple input and output pins for analog-to-digital conversion, serial peripheral interface communication, and digital logic control.In an embodiment, a first inverting amplifier U6.1 is configured to receive the clamped signal and output an amplified inverting version of the clamped signal with controllable gain via the digital potentiometer U5. The first inverting amplifier U6.1 is connected through a third resistor R12 at its input. A second inverting amplifier U6.2 with unity gain converts the auto-gained signal to a positive form, with a second resistor R11 connected at its input stage.In an embodiment, the circuit includes a first operational amplifier U7.1 and a peak detector operational amplifier U7.2. The peak detector operational amplifier U7.2 monitors the auto-gained signal for the purpose of controlling the digital potentiometer U5. A diode D2 is connected to a peak detector capacitor C2, which stores the peak voltage value. When the amplitude of the signal rises, the diode D2 becomes active, causing the peak detector capacitor C2 to rapidly charge and capture the larger peak amplitude.In an embodiment, a transistor Q1 manages the discharge of the peak detector capacitor C2, with a transistor base resistor R14 connected to the base of the transistor Q1. The microcontroller U8 controls the transistor Q1 through a digital logic signal to reset the peak detector capacitor C2 for subsequent measurement cycles. A zener diode ZD1 provides voltage regulation within the circuit.In an embodiment, the circuit further includes a first resistor R10, a fourth resistor R13, and a fifth resistor R23 for establishing appropriate signal levels and biasing conditions. Filter capacitors including a first filter capacitor C5, a second filter capacitor C6, and a third filter capacitor C7 provide signal conditioning and noise filtering throughout the circuit. The interconnection of these components enables the automatic gain adjuster 105 to dynamically adjust amplification based on the motor current level, ensuring optimal utilization of the analog to digital converter 107 dynamic range for enhanced fault signature detection.Referring to FIG. 8, in an embodiment, a flowchart of the gain control mechanism 800 implemented by the microcontroller U8 is illustrated. The flowchart shows the steps performed by the microcontroller U8 to control the gain in accordance with a range of the analog to digital converter 107.In an embodiment, the process begins at a Start block and proceeds to set the digital logic signal at the transistor base resistance to a low state to place the transistor Q1 in cutoff mode and enable the peak detector capacitor C2 to charge during a signal cycle.In an embodiment, the next step involves reading the peak voltage of the autogain signal by sampling a peak voltage across the peak detector capacitor C2 to obtain a feedback value representative of a maximum amplitude of an input signal.In an embodiment, a first decision diamond evaluates whether the peak voltage is between 0 Volt and 3.1 Volt. If the peak voltage is within the range of 0 to 3.1 Volt, the process proceeds to a second decision diamond that checks whether the digital potentiometer value is equal to zero.In an embodiment, if the digital potentiometer value equals zero at decision diamond, the process proceeds to set the digital potentiometer to zero to minimize the resistance and increase gain. If the digital potentiometer value does not equal zero, the process proceeds to decrement the digital potentiometer value to incrementally increase gain by one step.In an embodiment, if the peak voltage is not within the range of 0 to 3.1 Volt at decision diamond, the process proceeds to a third decision diamond that evaluates whether the peak voltage is between 3.1 Volt and 3.5 Volt. If the peak voltage falls within the hysteresis band of 3.1 to 3.5 Volt, the process proceeds directly to the capacitor discharge step.In an embodiment, if the peak voltage is not within the range of 3.1 to 3.5 Volt at decision diamond, the process proceeds to a fourth decision diamond that checks whether the peak voltage is greater than 3.5 Volt. If the peak voltage exceeds 3.5 Volt, the process proceeds to increment the digital potentiometer value to decrease gain.In an embodiment, following any gain adjustment or if no adjustment is needed, the process proceeds to discharge the peak detector capacitor by setting the digital logic signal at the transistor base resistance to a high state to turn on the transistor Q1, followed by returning the logic signal to the low state to reset the capacitor for the next measurement cycle.In an embodiment, the process then returns to the beginning to repeat the process in a continuous loop within firmware to maintain the peak voltage within a hysteresis band defined by the upper threshold of 3.5 Volt and lower threshold of 3.1 Volt and thereby stabilize automatic gain control while prevention of oscillations due to noise.Accordingly, mathematical Analysis to calculate the gain is provided here below:Assuming an inverting amplifier configuration, the gain is given by in Eq.(7):Equation (7)where Rf is the resistance from the MCP41010 digital potentiometer (wiper- to-end, ranging from 0 to 10 kΩ), and Rin is a fixed resistor (assumed to be 1 kΩ based on schematic inference).However, the accompanying text mentions a minimum gain of 1, which may suggest the use of a non-inverting configuration or that Rf and Rin are adjusted accordingly. Given 256 steps of resolution for the MCP41010, the step size is in Eq.(8):Equation (8)The corresponding gain resolution for each step is in Eq.(9):Equation (9)The maximum peak output is limited to 3.3 V due to a 100 Ω load resistor, so the automatic gain adjuster 105 primarily functions to amplify low-level signals to maximize ADC resolution. The automatic gain adjuster 105 represents a sophisticated yet cost-effective solution for dynamic signal scaling in current sensing systems. By leveraging digital control and feedback, it ensures optimal analog to digital conversion performance across varying loads, culminating the signal conditioning chain discussed in this thesis. Future work could involve ARM controller implementation for higher speeds or ML-based predictive gain adjustment.Referring to FIG. 9, in an embodiment, a block diagram of the operation of a differential amplifier 900 for power supply noise rejection corresponding to the noise rejecter 108 shown in FIG. 1 is illustrated. The differential amplifier configuration cancels common-mode noise and power supply noise by applying two opposite-phased signals.In an embodiment, the diagram shows a positive signal input and a negative signal input feeding into the differential amplifier. The differential amplifier subtracts the common-mode components present in both inputs, effectively rejecting power supply noise and other common-mode interference. The output of the differential amplifier is a clean signal with reduced noise content.In an embodiment, the noise rejecter 108 comprises an operational amplifier based power supply noise rejecter configured to cancel common-mode noise and power supply noise. The differential amplifier configuration is particularly effective in industrial environments where electromagnetic interference and power supply fluctuations are common sources of signal degradation.Referring to FIG. 10, in an embodiment, a schematic of a signal conditioning circuit 1000 implementing the noise rejecter 108 and low-pass filter 109 shown in FIG. 1 is illustrated. The circuit employs a differential amplifier configuration combined with a low-pass filter to cancel common-mode noise and power supply noise while eliminating undesirable high-frequency components.In an embodiment, the circuit receives input signals through first input resistor R15 and second input resistor R16, which connect to a negative signal path, and third input resistor R17 and fourth input resistor R18, which connect to a positive signal path and ground respectively. The fourth input resistor R18 is connected to ground to establish a reference potential. In an embodiment, the exemplary value of R15, R16, R17, and R18 may be 10 kΩ.In an embodiment, third operational amplifier U7.3 is configured as a differential amplifier with third operational amplifier negative input pin 9 receiving the signal through first input resistor R15 and third operational amplifier positive input pin 10 receiving the signal through third input resistor R17. Third operational amplifier output terminal 8 provides the processed signal to the subsequent filter stage through fifth resistor R19, such as value of 10 kΩ.In an embodiment, the low-pass filter 109 comprises a Sallen-Key low-pass filter configured to pass primary frequency components below 159 Hz to examine the current signature of a variable-speed induction motor. The low-pass filter section employs a Sallen-Key topology using fourth operational amplifier U7.4. Fifth resistor R19 and sixth resistor R20, such as value of 10 kΩ form the resistive elements of the filter, while first capacitor C3, such as value of 100nF and second capacitor C4, such as value of 100nF provide the capacitive elements. First potentiometer RP1, such as value of 100 kΩ is positioned between fifth resistor R19 and sixth resistor R20, allowing adjustment of the cutoff frequency.In an embodiment, fourth operational amplifier positive input pin 12 receives the filtered signal, while fourth operational amplifier negative input pin 13 is connected in a feedback configuration. Fourth operational amplifier output terminal 14 provides the filtered output signal.In an embodiment, the output section includes full range signal connector J7 for accessing the full range signal and ESP ADC input connector J9 for connection to the analog to digital converter 107. Seventh resistor R21 connects to full range signal connector J7, while eighth resistor R22 and second potentiometer RP2 form a voltage divider network connected to ESP ADC input connector J9. Zener diode ZD2 (e.g. 3.3V) provides overvoltage protection for the ADC input, clamping the signal to prevent damage to the converter.In an embodiment, the circuit processes current signature signals to pass primary frequency components below 159 Hz, which are relevant for examining the current signature of variable-speed induction motors. The Sallen-Key low-pass filter provides a quality factor of 0.5 with adjustable cutoff frequency through first potentiometer RP1, enabling optimization for different operating conditions.In an embodiment, the cutoff frequency is targeted at 159 Hz to retain primary motor current components while attenuating noise above the cutoff point. The quality factor Q is set to 0.5 for a smooth roll-off without peaking, ensuring stability and minimal ringing in the time domain.The transfer function of the Sallen-Key low-pass filter in Figure 9 is given by (10)The standard form of a second-order low-pass filter is in Eq.(11):Equation (11)where the natural frequency ωc = 2πfc and Q is the quality factor. In an embodiment, the cutoff frequency fc is derived from the component values according to:Equation (12)In an embodiment, substituting the values R19 = R20 = 10 kΩ, C3 = C4 = 100 nF yields a cutoff frequency of approximately 159 Hz, ensuring that frequencies below the cutoff (e.g., 50 / 60 Hz fundamentals and low harmonics) pass with near-unity gain, while higher frequencies are attenuated at -40 dB / decade (second-order roll-off), as shown below:Equation (13)In standard unity-gain Sallen-Key, the coefficient of s is , but RP1 modifies it to (approximate, depending on exact configuration). Adjusting RP1 alters Q and slightly shifts fc, providing flexibility for calibration against varying motor speeds or noise profiles. In the disclosed edge device 100, a dual-rail ±12 V supply is used to powers the operational amplifiers ensuring sufficient headroom for ±3.3 V signals and minimizing power supply noise injection.Post-filtering, the denoised motor current signatures show clearer harmonic content in the FFT, thereby improving fault detection. For instance, a sideband amplitude exceeding few percentages is indicative of broken rotor bars. The filter's stability ensures reliable performance under varying load conditions, with large gain and phase margins preventing oscillations that might otherwise mimic electrical faults.In summary, the Sallen-Key low-pass filter configured with Q = 0.5 effectively denoises AGC-processed motor currents. Its stable yet selective response makes it well-suited for enhanced motor health monitoring. Future improvements could include adaptive control of Q using a digital potentiometer to optimize filtering across variable-speed motor operations.FIG. 10 also shows waveforms including healthy motor waveform 1, half broken rotor bar waveform 2, full broken rotor bar waveform 3, and end ring fault waveform 4, demonstrating the circuit's ability to process signals from motors in various health conditions.Referring to FIG. 11, in an embodiment, a comprehensive schematic of an enhanced current signature acquisition system 1100 for the edge device 100 is illustrated, organized into multiple functional sections across the circuit board layout. The schematic is divided into five primary sections: a power supply section, an analog signal processing section, an auto gain feedback controller section, an ADC section, and a Raspberry Pi section.In an embodiment, the power supply section, located in the lower left portion of the schematic, includes multiple voltage regulators and connectors for providing stable power rails to the circuit. The section features connectors for receiving input power and generates regulated voltage outputs including positive and negative twelve volt rails and a five volt rail. The edge device 100 further comprises a dual-rail power supply including a switched-mode power supply and low-dropout regulators configured to minimize electrical noise.In an embodiment, the analog signal processing section occupies the right side of the schematic and contains operational amplifier circuits configured for signal conditioning. The section includes multiple operational amplifiers arranged in various configurations for noise rejection and low-pass filtering.In an embodiment, the auto gain feedback controller section is positioned in the upper central area and includes a microcontroller connected to a digital potentiometer for implementing automatic gain control. The microcontroller interfaces with the digital potentiometer through a serial peripheral interface connection, enabling dynamic adjustment of signal amplification based on the motor current level.In an embodiment, the ADC section interfaces with the analog signal processing circuitry and includes connections for analog-to-digital conversion. The Raspberry Pi section provides the main processing platform with a header connector for interfacing with a single-board computer. The edge device 100 further comprises a communication interface 111 including at least one of a Wi-Fi unit for wireless communication over a local network or internet, or a TCP / IP Ethernet unit for wired communication, whereby the Wi-Fi and / or TCP / IP Ethernet units meet diverse communication needs of modern IoT and industrial applications.In an embodiment, the processor 106 is configured to process stator current signal (Ia) by estimation of rotor speed (ωr) from the stator current. This is a critical step, as rotor speed influences slip and the frequencies of fault-related harmonics. The processor 106 may also estimate fundamental frequency (fs) in the stator current signal (Ia) and remove of fundamental component from the stator current signal (Ia) by use of Extended Kalman Filter (EKF) to ensure that remaining stator current signal primarily contains harmonic components that may indicate the faults. The processor 106 may also calculate slip (ωr) using the estimated rotor speed and the fundamental frequency (fs). The slot harmonic frequency and the speed are related as shown in Eq.(13): Equation (13)where fh is the slot harmonic frequency, the slip has been computed as shown in Eq.(14) Equation (14)The processor 106 may also determine autocorrelation matrix size based on the fundamental frequency and slip. In an embodiment, the autocorrelation matrix size N is determined by: Equation (15)where fs is the sampling frequency. The formula is fundamental to the spectral analysis module 110 operation, as the formula determines the appropriate size of the autocorrelation matrix based on the motor's operating parameters.The processor 106 may also form a data matrix (Xβ) using the stator current and the autocorrelation matrix (N) size. This matrix is essential for extracting statistical properties of the signal.The processor may also determine signal and noise subspace sizes using eigen values of signal, as shown in Eq. (16):Equation (16)where S and G represent the signal and noise subspaces and Δs & Δn are the respective eigenvalue matrices.The processor 106 may also estimate frequencies and magnitudes of remaining harmonic components including closely spaced harmonics linked with incipient faults by use of a Multiple Signal Classification (MUSIC) algorithm. This high-resolution method is effective in detecting closely spaced harmonics, especially those linked with incipient faults.The processor 106 may also classify fault types based on detected harmonic components and their magnitudes. The presence of specific harmonics (e.g., broken rotor bar, eccentricity) helps the processor 106 in classifying the fault types.Referring to FIG. 12, in an embodiment, a frequency spectrum graph 1200 illustrating detection of broken rotor bar fault signatures in an induction motor current signal is depicted. The horizontal axis represents frequency measured in Hertz, spanning a range from approximately 49.5 Hz to slightly above 50 Hz. The vertical axis represents magnitude of the frequency components.In an embodiment, the graph displays three distinct peaks corresponding to characteristic frequency components used in motor fault diagnosis. A central peak at the fundamental supply frequency of 49.97 Hz represents the dominant component of the motor current. A lower BRB sideband component appears at 49.71 Hz, positioned to the left of the fundamental frequency peak. An upper BRB sideband component appears at 50.23 Hz, positioned to the right of the fundamental frequency peak.In an embodiment, the presence of these sideband components on either side of the fundamental supply frequency indicates the characteristic fault signature associated with broken rotor bar conditions in induction motors, where the sidebands appear at frequencies offset from the supply frequency by twice the slip frequency multiplied by the supply frequency.As shown in FIG. 12, broken rotor bars introduce asymmetry in the rotor magnetic field, manifesting as sideband harmonics in the stator current spectrum. The characteristic fault frequencies are shown in Eq.(17)Equation (17)where Fs is the fundamental supply frequency (Hz), s is the slip, and k = 1, 2, . . . is the harmonic order. For early detection, the primary sidebands (k = 1) are analyzed post-fundamental removal via EKF, with amplitudes typically -40 to -60 dB below Fs in healthy motors but elevated in faulty cases.The fixed-point algorithms process the residual signal to identify these sidebands, thresholding peaks exceeding 3s above the noise floor. Validation on the 22 kW motor under broken bar fault shows clear sidebands, confirming algorithm efficacy.Referring to FIG. 13, in an embodiment, multiple views of the edge device 1300 for induction motor fault diagnosis are illustrated, labeled as views A, B, and C.In an embodiment, View A shows a perspective view of the assembled hardware system mounted on a platform with wheels, featuring a display screen on the left side and a main circuit board on the right. The main circuit board includes various electronic components, connectors, and an institutional logo. A cable connects the display unit to the main board.In an embodiment, View B shows a front view of the edge device 100 housed within an enclosure. The view reveals the internal arrangement of components including a Raspberry Pi single board computer positioned in the upper right portion, a current sensor located on the left side, an automatic gain adjuster circuit board in the center, a uniformed sampled MCU ADC module, and a signal conditioning circuit positioned in the lower portion of the enclosure. The Raspberry Pi displays a graphical user interface on the screen. Various circuit boards are interconnected and mounted within the protective housing.In an embodiment, View C presents a rear or back view of the edge device 100, showing the reverse side of the circuit boards and additional components. The view displays the Raspberry Pi from the back side, revealing the circuit board layout and connection ports. A dual rail switched-mode power supply module is visible in the lower portion of the enclosure. Multiple circuit boards with various electronic components and wireless communication modules are arranged within the housing. The enclosure provides protection for the internal electronics while allowing access to necessary connection points for power and data communication.Referring to FIG. 14, in an embodiment, a hardware communication interface 1400 between a single board computer and the microcontroller is illustrated. On the left side, the Raspberry Pi is shown with a 40-pin GPIO header clearly labeled, displaying the pin assignments including power pins (3.3V and 5V), ground pins, and various GPIO pins with alternate functions such as I2C (SDA, SCL), SPI (MOSI, MISO, SCLK, CE0, CE1), and UART (TXD, RXD).In an embodiment, the UART0 interface on the Raspberry Pi is specifically highlighted, indicating pins GPIO14 (UART TXD) and GPIO15 (UART RXD). On the right side, the ESP32-WROOM-32 module is depicted, which serves as a wireless-capable microcontroller.In an embodiment, the diagram indicates that UART0 on the ESP32 functions as the debugger and firmware uploader interface, while UART2 on the ESP32 is connected to the Raspberry Pi's UART0 for serial communication between the two devices. The configuration demonstrates the serial communication architecture employed in the edge device 100, where the ESP32 microcontroller handles wireless connectivity and low-level signal processing tasks while communicating with the Raspberry Pi single board computer for higher-level processing, data analysis, and network connectivity functions essential for the induction motor fault diagnosis system.Referring to FIG. 15, in an embodiment, oscilloscope waveforms 1500 showing signal characteristics at different stages of the current signature acquisition system are depicted. The display shows four panels arranged in a two-by-two grid, each displaying a waveform trace against a dark background with grid lines.In an embodiment, the waveforms demonstrate the signal transformation through the various processing stages of the edge device 100. The panels show the progression of the signal from the initial alternating form with negative and positive excursions to a positively clamped and amplified signal suitable for analog-to-digital conversion.In an embodiment, the frequency readings hovering around 50Hz correspond to the fundamental frequency of the induction motor supply, confirming proper signal acquisition for current signature analysis and fault diagnosis purposes.Referring to FIG. 16, in an embodiment, a logic analyzer capture 1600 showing serial communication waveforms for the edge device 100 is depicted. The display shows two digital channels labeled D0 and D1, corresponding to ESP32 Tx and ESP32 Rx respectively.In an embodiment, Channel D0 displays the ESP32 transmit signal configured as Async Serial, showing digital pulses with annotations indicating carriage return and line feed characters. Channel D1 displays the ESP32 receive signal configured as Async Serial, showing a constant high state during the captured time window.In an embodiment, the time scale indicates a capture duration starting at 981 ms with 300 microseconds resolution, with time markers showing increments of 10 microseconds and 20 microseconds along the horizontal axis. The waveform on Channel D0 exhibits transitions between high and low logic states representing serial data transmission, while Channel D1 remains at a steady high logic level indicating no received data during the capture period.In an embodiment, the logic analyzer capture demonstrates the serial communication functionality between the ESP32 microcontroller and other system components, which is essential for data transmission and for the Wi-Fi and TCP / IP Ethernet communication capabilities for IoT and industrial applications.Referring to FIG. 17, in an embodiment, stator current characteristics 1700 including time domain waveform and frequency domain analysis are presented.In an embodiment, the left plot displays the stator current waveform of the induction motor 101 in the time domain, with the horizontal axis representing time in milliseconds ranging from approximately 700 to 1700, and the vertical axis showing the current amplitude ranging from 0 to 3 units. The waveform exhibits a sinusoidal pattern with consistent amplitude oscillations between approximately 0 and 3, demonstrating the alternating nature of the motor current signal after processing through the signal acquisition system.In an embodiment, the right plot shows the Fast Fourier Transform (FFT) analysis of the R phase current of the induction motor 101, with the horizontal axis representing frequency in Hertz ranging from 0 to 150 Hz, and the vertical axis showing magnitude ranging from 0 to approximately 3500. The FFT spectrum displays a single dominant peak at approximately 50 Hz with a magnitude of roughly 3000, which corresponds to the fundamental supply frequency of the induction motor 101.The absence of significant sideband frequency components around the fundamental frequency in the spectrum indicates the current signature of a healthy motor operating under normal conditions. The plots demonstrate the signal processing capabilities of the edge device 100 in capturing high-fidelity current signals and performing frequency domain analysis essential for motor fault diagnosis through current signature analysis techniques.Referring to FIG. 18, in an embodiment, oscilloscope waveforms and measurement data 1800 for motor current signatures under different fault conditions are depicted, demonstrating the diagnostic capabilities of the edge device 100 for identifying various motor health states.In an embodiment, the upper portion shows four oscilloscope displays arranged in a two-by-two grid, each displaying a yellow waveform trace against a dark background with grid lines.In an embodiment, the healthy motor waveform 1 appears in the upper left quadrant and shows a relatively clean sinusoidal pattern characteristic of a motor without faults. The half broken rotor bar waveform 2 appears in the upper right quadrant and displays a sinusoidal pattern with visible amplitude modulation and distortion compared to the healthy motor waveform 1, indicating an incipient fault condition. The full broken rotor bar waveform 3 appears in the lower left quadrant and exhibits more pronounced amplitude variations and envelope modulation than the half broken rotor bar waveform 2, indicating a more severe fault condition. The end ring fault waveform 4 appears in the lower right quadrant and shows a cleaner sinusoidal pattern similar to the healthy motor waveform 1.In an embodiment, the lower portion of FIG. 18 presents measurement data tables corresponding to each waveform, labeled 1 through 4 from left to right. The measurement data for the healthy motor waveform 1 indicates a minimum value of -2.16V, a maximum value of 1.50V, an amplitude of 2.14V, and a frequency of 48.738Hz. The measurement data for the half broken rotor bar waveform 2 shows a minimum value of -300mV, a maximum value of 3.34V, an amplitude of 3.64V, and a frequency of 49.368Hz. The measurement data for the full broken rotor bar waveform 3 displays a minimum value of -440mV, a maximum value of 6.43V, an amplitude of 6.87V, and a frequency of 50.317Hz. The measurement data for the end ring fault waveform 4 indicates a minimum value of 240mV, a maximum value of 5.95V, an amplitude of 5.71V, and a frequency of 50.085Hz.Referring to FIG. 19, in an embodiment, experimental equipment and diagnostic output 1900 for the induction motor fault diagnosis system is depicted, demonstrating the validation of the edge device 100 in real-world operating conditions.In an embodiment, the upper left portion shows a frequency measurement device displaying a value of 500 on a digital screen, with control buttons including exit, enter, stop, and start functions. The frequency measurement device represents a variable frequency drive (VFD) controller used to set the operating frequency of the induction motor 101. Adjacent to the frequency measurement device is a digital display unit showing a reading of 0 166, representing an external ammeter measuring the phase current.In an embodiment, to the right of these devices is a terminal window showing shell output from the processor 106, with values including 0.93 and 0.99, along with text indicating phase current of 16.68 amperes, signal processing status, fundamental frequency of 50.020060803201276, and a completed status message. The terminal indicates running Local Python 3 from the usr / bin / python3 path, demonstrating the software implementation of the spectral analysis module 110.In an embodiment, below the equipment view, the figure shows a grid of six document icons representing faulty motor readings, each labeled with timestamps from November 17, 2023, with specific times ranging from 13:42:29 to 13:48:35. The document icons represent data files stored by the edge device 100 for offline analysis. To the right of the document icons is a vertical scale showing amplitude values ranging from 1.640000 at the bottom to 2.410000 at the top, with incremental values displayed.In an embodiment, the bottom portion of FIG. 19 presents two frequency domain analysis plots generated by the spectral analysis module 110. The left plot shows a frequency-domain analysis with a fundamental frequency of 49.06 Hz, displaying amplitude on the vertical axis ranging from 0.0 to 1.0 and frequency in Hz on the horizontal axis ranging from 0 to 100. The right plot shows a frequency search result indicating a fundamental frequency of 49.06 Hz with an exact value of 49.98 Hz, displaying magnitude on the vertical axis ranging from 0 to 3000 and frequency in Hz on the horizontal axis ranging from 40.0 to 60.0, with a prominent peak visible near the 50 Hz region.Referring to FIG. 20, in an embodiment, a file management interface showing data logger validation 2000 with sequential logging readings files is depicted, demonstrating the data storage capabilities of the edge device 100.In an embodiment, each file is represented by a standard document icon with horizontal lines indicating text content. The files are labeled with timestamps indicating the files contain faulty motor readings captured on November 17, 2023.In an embodiment, the top row shows three files with timestamps of 13-42-23.900151, 13-43-15.410284, and 13-44-02.290122 respectively. The bottom row displays three additional files with timestamps of 13-47-04.517886, 13-47-50.074644, and 13-48-35.647424.In an embodiment, on the right side of the image, a vertical color scale bar is visible, displaying numerical values ranging from approximately 1.640000 at the bottom to 2.410000 at the top, with intermediate values displayed. The color scale represents amplitude or intensity values associated with the motor current readings stored in the files.The data logger functionality enables the edge device 100 to store current signature data for offline analysis, supporting environments lacking real-time network connectivity such as remote industrial sites.Referring to FIG. 21, in an embodiment, frequency domain analysis plots 2100 showing exact fundamental frequency estimation by the spectral analysis module 110 are illustrated.In an embodiment, the left plot, titled "Frequency-Domain Analysis (Fundamental Frequency: 49.06 Hz)," displays amplitude on the vertical axis ranging from 0 to approximately 1.0 and frequency in Hertz on the horizontal axis spanning from 0 to beyond 100 Hz. The left plot shows a prominent peak centered near 50 Hz with a normalized amplitude of approximately 1.0, representing the fundamental frequency component of the motor current signal. The spectral response exhibits a relatively broad peak shape characteristic of standard FFT-based frequency analysis.In an embodiment, the right plot, titled "Frequency Search (Fundamental: 49.06 Hz, Exact: 49.98 Hz)," presents a higher-resolution frequency analysis with magnitude on the vertical axis extending from 0 to approximately 3000 and a narrower frequency range on the horizontal axis spanning from approximately 47 Hz to 60 Hz. The right plot demonstrates a significantly sharper and more precisely defined peak at approximately 50 Hz, with the magnitude reaching nearly 3000 units.The enhanced resolution and peak definition in the right plot illustrate the capability of advanced spectral estimation techniques, such as the MUSIC algorithm employed by the spectral analysis module 110, to achieve superior frequency resolution compared to conventional FFT methods. In an embodiment, the processor 106 is configured to estimate fundamental frequency in the stator current signal and remove fundamental component from the stator current signal by use of Extended Kalman Filter to ensure that remaining stator current signal primarily contains harmonic components that may indicate the faults.Referring to FIG. 22, in an embodiment, a frequency spectrum plot 2200 showing slot harmonic estimation by the spectral analysis module 110 is presented. The plot is titled "Slot Harmonic Frequency: 674.84 Hz" and displays the magnitude of signal components across a frequency range from approximately 670 Hz to 705 Hz on the horizontal axis, with magnitude values ranging from 0.0 to approximately 2.5 on the vertical axis.In an embodiment, the plot exhibits a prominent sharp peak at approximately 675 Hz with a magnitude reaching approximately 2.5, which corresponds to the slot harmonic frequency indicated in the title. The dominant peak represents the primary slot harmonic component in the motor current signature. Additionally, a secondary notable peak appears at approximately 705 Hz with a magnitude of roughly 1.6.In an embodiment, between the two prominent peaks, the spectrum shows relatively low-amplitude fluctuations and noise floor variations ranging between approximately 0.1 and 0.7 in magnitude.In an embodiment, the slot harmonic frequency is used by the processor 106 to calculate slip according to:Equation (18)where pp represents the number of pole pairs, NR denotes the number of rotor slots, fh is the slot harmonic frequency in Hertz, Fs is the fundamental frequency in Hertz, and vd represents an additional harmonic component offset. the slip may be computed as shown in Eq.(19) with k = 1 and nd = 0.Equation (19)For NR = 28 rotor slots, Pp = 2 pole pairs, and additional harmonic component offset vd = 0, the slip formula is applied. Given Fh = 674.84 Hz, Fs = 49.98 Hz, NR = 28, Pp = 2, and vd = 0, the resulting slip from Eq.(26) is s = 0.0355 and the estimated rotor speed from Eq.(20) is Nr = 1446.17 RPM, with an estimation error less than 0.5%.Equation (20)In an embodiment, the processor 106 is configured to calculate slip using the estimated rotor speed and the fundamental frequency, and determine autocorrelation matrix size based on the fundamental frequency and slip.Referring to FIG. 23, in an embodiment, frequency spectrum plots 2300 showing broken rotor fault detection through current signature analysis by the spectral analysis module 110 are presented, demonstrating the fault diagnosis result output by the processor 106 shown in FIG. 1.In an embodiment, the left plot shows a broken rotor frequency at 46.42 Hz representing a lower sideband component. The left plot displays magnitude on the vertical axis ranging from 0 to 300 and frequency in Hz on the horizontal axis ranging from 36 to 46 Hz. The left plot displays an oscillating waveform that represents the lower sideband fault frequency component detected in the stator current spectrum. The lower sideband component at 46.42 Hz appears below the fundamental supply frequency of approximately 50 Hz, indicating presence of a broken rotor bar fault condition.In an embodiment, the right plot shows a broken rotor frequency at 53.54 Hz representing an upper sideband component. The right plot displays magnitude on the vertical axis ranging from 0 to 250 and frequency in Hz on the horizontal axis ranging from 54 to 64 Hz. The right plot displays an oscillating waveform that represents the upper sideband fault frequency component detected in the stator current spectrum. The upper sideband component at 53.54 Hz appears above the fundamental supply frequency, providing a complementary indicator to the lower sideband component for broken rotor bar fault identification.In an embodiment, the lower sideband at 46.42 Hz and the upper sideband at 53.54 Hz represent characteristic fault frequency components that appear symmetrically around the fundamental supply frequency when a broken rotor bar condition exists in the induction motor 101. The frequency offset between the fundamental supply frequency and each sideband component corresponds to twice a slip frequency multiplied by the supply frequency.In an embodiment, the presence of both the lower sideband and the upper sideband in the stator current spectrum provides a diagnostic signature that the spectral analysis module 110 uses to identify broken rotor bar faults. The processor 106 is configured to identify faulty or healthy motor by determination of peak value in the frequency spectrum and comparison with a threshold value. The sideband fault frequency components shown in FIG. 23 demonstrate detection capability of the edge device 100 for identifying rotor faults under operational conditions.In an embodiment, for validation of broken rotor bar faults, with NR = 28 rotor slots, Pp = 2 pole pairs, and additional harmonic component offset vd = 0, with Fh = 674.84 Hz and Fs = 49.98 Hz, the resulting slip is s = 0.0355 and the estimated rotor speed is Nr = 1446.17 RPM, with an estimation error less than 0.5%. Overall, the algorithms achieved less than 0.5 Hz detection accuracy across sidebands, enabling real-time fault prognosis.In an embodiment, the processor 106 comprises a single-board computer configured to process stator current signal as acquired from the signal conditioning circuit 104. The single-board computer is a versatile computing platform widely used in embedded applications, serving as an ideal edge device for induction motor health monitoring due to the balance of computational power, GPIO interfaces for sensor integration, and support for real-time operating systems.In an embodiment, fixed-point arithmetic is chosen over floating-point for computational efficiency on ARM-based processors. By representing numbers with a fixed number of integer and fractional bits, operations such as multiplication and addition are executed faster, reducing power consumption and ensuring deterministic execution times essential for real-time systems. Floating-point operations on edge devices like the Raspberry Pi consume more cycles (e.g., 4-10x slower than integers) and power due to FPU overhead. Fixed-point uses scaled integers: a real number x is represented as xq = x · 2q, where q is the number of fractional bits. For motor currents (amplitudes ~0- 3.3 V), the Q15 format scales values to the range [-1, 1) with a resolution of 2-15 ~ 30 μV, which is sufficient for detecting -40 dB harmonics.Overflow is mitigated by using saturation arithmetic; multiplication results are shifted right by q bits after the operation. Quantization noise is modeled as uniformly distributed in the interval [-LSB / 2, LSB / 2], with variance.For Q15 representation, this yields s ~ 1.5 x 10-5, which is negligible when compared to typical sensor noise (approximately 0.1 mV RMS).Fixed-point implementations use Q15 format (16-bit integers with 15 fractional bits) for computational efficiency on edge devices such as the Raspberry Pi. Algorithms process stator current ia, sampled at 5 kS / s. NumPy simulates fixed-point arithmetic via int16 arrays with manual bit shifts.Notations:Fs: Fundamental frequency (Hz) from the Variable Frequency Drive (VFD).fs: Sampling frequency (Hz).fh: Slot harmonic frequency (Hz)ωr: Rotor speed (rad / s).k: Harmonic index (integer multiplier, typically k = 1 for principal slot harmonic)s: Slip (dimensionless), defined as the relative difference between synchronous speed and rotor speed.ia: Stator current.NR: Number of rotor slotsnd: Slot harmonic order (e.g., 1, 3, 5...), associated with spatial distributionpp: Number of pole pairsvd: Additional harmonic component offset (can be 0 or 1 depending on modulation asymmetry)N : Size of the autocorrelation matrix (number of samples used)Xβ: Data MatrixS: Signal subspaceG: Noise subspaceΔΔs: Eigenvalue matrix of SΔn: Eigenvalue matrix of GIn an embodiment, the disclosed edge device 100 includes advanced communication capabilities, including both Wi-Fi and Ethernet, which enable the edge device 100 to function efficiently in a wide range of IoT applications and industrial environments. The dual communication modes provide the edge device 100 with flexibility, reliability, and scalability, making the edge device 100 suitable for various operational needs.In an embodiment, Wi-Fi capability in the edge device 100 allows for wireless communication over a local network or the internet. The Wi-Fi feature is particularly advantageous in scenarios where wired connections are impractical, such as in remote monitoring systems and mobile IoT applications. The key benefits of Wi-Fi capability include flexibility in deployment, ease of connectivity, and remote management.In an embodiment, TCP / IP Ethernet capability enables wired communication, which is essential in environments where stable, high-speed, and low-latency data transfer is critical. The advantages of Ethernet connectivity include reliability and stability, high data throughput, and security.In an embodiment, the developed edge device 100, with Wi-Fi and TCP / IP Ethernet capabilities, is designed to meet the diverse communication needs of modern IoT and industrial applications. The dual-mode communication enables flexible deployment, robust performance, and secure connectivity, making the edge device 100 adaptable to a variety of environments and operational requirements.The disclosed device is capable of continuously monitoring the health of an induction motor directly at the edge, enabling real-time assessment without reliance on centralized processing. By performing signal acquisition, conditioning, and preliminary analysis locally, the edge device ensures fast detection of abnormal operating conditions while reducing latency and network bandwidth requirements. The device further supports offline data logging, allowing critical motor health parameters to be securely stored in on-board memory during periods of network unavailability, thereby ensuring uninterrupted monitoring and traceability of motor performance over time.In addition to health monitoring, the disclosed device is also capable of estimating the remaining useful life (RUL) of the induction motor directly on the edge. By analyzing degradation trends, operating stress, and historical condition indicators, the edge device can compute prognostic metrics that reflect the expected lifespan of the motor under current operating conditions. Performing RUL estimation locally enables timely maintenance decisions, early warning generation, and reduced dependence on continuous cloud connectivity, which is particularly beneficial in industrial environments with limited or intermittent network access.Furthermore, the disclosed device supports online data transmission to a remote server through advanced communication interfaces such as Wi-Fi and Ethernet, enabling seamless integration with IoT platforms and industrial networks. The logged and processed data, including health indicators and RUL estimates, can be periodically synchronized with the server for further statistical analysis, long-term trend evaluation, and fleet-level health assessment. This combined capability of edge-level intelligence, offline logging, and online reporting provides a robust framework for predictive maintenance and data-driven reliability management of induction motors.The present disclosure provides various advantages over conventional induction motor fault diagnosis systems. For example, unlike prior art that relies on multiple sensors and complex hardware setups, the disclosed edge device employs a streamlined single-sensor configuration, significantly reducing installation complexity and cost. The integration of an automatic gain adjuster ensures optimal utilization of the analog-to-digital converter range, enabling precise detection of low-amplitude fault signatures without sacrificing signal integrity. Advanced noise mitigation techniques, including common-mode clamping and differential filtering, enhance diagnostic accuracy even in electrically noisy industrial environments. Furthermore, the edge computing architecture allows real-time fault detection at the source, minimizing latency and reducing dependence on network connectivity. The system's adaptability to variable-speed drives, combined with sophisticated spectral analysis algorithms such as FFT and MUSIC, ensures accurate identification of incipient faults under diverse operating conditions. Additional benefits include compatibility with existing industrial infrastructure, scalability across motor types, and dual-mode communication (Wi-Fi and Ethernet) for seamless integration into IoT ecosystems. Collectively, these features make the proposed edge device a cost-effective, reliable, and future-ready solution for industrial motor health monitoring.A number of implementations have been described. Nevertheless, various modifications may be made without departing from the spirit and scope of the disclosure. Accordingly, other implementations are within the scope of the following claims.

Claims

1. An edge device for induction motor fault diagnosis, comprising: a sensor configured to obtain a current signal from an induction motor and convert it to a voltage signal using a transducer with galvanic isolation; a signal conditioning circuit coupled to the sensor and configured to process the voltage signal to produce a clamped signal with reduced common mode noise; an automatic gain adjuster coupled to the signal conditioning circuit and configured to dynamically adjust amplification of the signal; a processor configured to: convert the amplified signal to a digital format using an analog to digital converter; condition the digital signal to suppress noise using a noise rejecter and a low-pass filter; and detect fault frequency components and diagnose motor faults by performance of spectral analysis and comparison of spectral peak amplitudes with threshold values.

2. The device as claimed in claim 1, wherein the transducer comprises a split core transducer configured to effectively separate its primary circuit responsible for power transmission from its secondary circuit dedicated to the measurement process.

3. The device as claimed in claim 1, wherein the sensor further comprises a resistor with selective value configured to perform current-to-voltage conversion and establish a desired amplitude for output voltage signal according to the current drawn by the motor.

4. The device as claimed in claim 1, wherein the signal conditioning circuit comprises a common-mode signal clamping circuit including an operational amplifier based peak detector configured to detect a peak of the voltage signal and store the positive peak as a voltage value in a capacitor, and a summer configured to generate a positively clamped signal that involves the common mode signal component upon detection of the peak by said peak detector.

5. The device as claimed in claim 4, wherein the common-mode signal clamping circuit is configured to clamp the output voltage signal above a ground level with a common mode signal that avoids unintentional dynamic gain amplification of excessive common mode signal.

6. The device as claimed in claim 4, wherein a diode is connected to the capacitor and configured to become active with rise of amplitude of the signal from the sensor to cause the capacitor to charge and capture the larger peak amplitude, and wherein when the peak amplitude decreases, the diode remains inactive until the capacitor discharges to the decreased level, whereby the operational amplifier based peak detector serves as a comparator to facilitate this process.

7. The device as claimed in claim 4, wherein the capacitor is configured to undergo discharge via an input bias current of a unity gain buffer operational amplifier configured to remove loading effect of the capacitor and a summer operational amplifier configured to generate the positively clamped signal, whereby the operational amplifiers set a total discharging current dependent on capacitance value of the capacitor and a maximum rate of voltage reduction to fix the device ability to detect a maximum motor current degradation rate.

8. The device as claimed in claim 1, wherein the automatic gain adjuster comprises: a digital potentiometer; a microcontroller configured to operate the digital potentiometer; and an operational amplifiers configured to maximize utilization of converting range of the analog to digital converter.

9. The device as claimed in claim 8, wherein the automatic gain adjuster further comprises: an operational amplifier based inverting amplifier configured to receive the clamped signal and output an amplified inverting version of the clamped signal with a controllable gain via said microcontroller operated digital potentiometer; an additional inverting amplifier with unity gain configured to convert the amplified inverting version to a positive form; and a peak detector configured to monitor the amplified signal for the purpose of control of the digital potentiometer and management of discharge of a second capacitor through a transistor.

10. The device as claimed in claim 8, wherein the microcontroller is configured to control the gain in accordance with a range of the analog to digital converter by: initiation of a gain control routine with the digital potentiometer set to an initial value; setting of a digital logic signal at a transistor base resistance to a low state to place the transistor in cutoff mode and enable a peak detector capacitor to charge during a signal cycle; sampling of a peak voltage across the peak detector capacitor to obtain a feedback value representative of a maximum amplitude of an input signal; comparison of the sampled peak voltage against an upper threshold and a lower threshold to determine a gain adjustment; discharge of the peak detector capacitor by setting of the digital logic signal at the transistor base resistance to a high state to turn on the transistor, followed by return of the logic signal to the low state to reset the capacitor for the next measurement cycle; and repetition of the process in a continuous loop within firmware to maintain the peak voltage within a hysteresis band defined by the upper and lower thresholds and thereby stabilize automatic gain control while prevention of oscillations due to noise.

11. The device as claimed in claim 1, wherein the noise rejecter comprises an operational amplifier based power supply noise rejecter configured to cancel common-mode noise and power supply noise, and wherein the low-pass filter comprises a Sallen-Key low-pass filter configured to pass primary frequency components below 159 Hz to examine the current signature of a variable-speed induction motor.

12. The device as claimed in claim 1, wherein the processor is configured to detect sideband fault frequency components of healthy as well as faulty motor current using at least one of Fast Fourier Transform, Power Spectral Density, or MUSIC algorithm and calculate location of characteristic fault frequency components in stator current spectrum of the motor, and wherein the processor is configured to identify faulty or healthy motor by determination of peak value in the frequency spectrum and comparison with a threshold value.

13. The device as claimed in claim 1, further comprising: a dual-rail power supply including a switched-mode power supply and low-dropout regulators configured to minimize electrical noise; and a communication interface including at least one of a Wi-Fi unit for wireless communication over a local network or internet, or a TCP / IP Ethernet unit for wired communication, whereby said Wi-Fi and / or TCP / IP Ethernet units meet diverse communication needs of modern IoT and industrial applications.

14. The device as claimed in claim 1, wherein the processor comprises a single-board computer configured to process stator current signal as acquired from the signal conditioner by: estimation of rotor speed from the stator current; estimation of fundamental frequency in the stator current signal and removal of fundamental component from the stator current signal by use of Extended Kalman Filter to ensure that remaining stator current signal primarily contains harmonic components that may indicate the faults; calculation of slip using the estimated rotor speed and the fundamental frequency; determination of autocorrelation matrix size based on the fundamental frequency and slip; formation of data matrix using the stator current and the autocorrelation matrix size; determination of signal and noise subspace sizes using eigen values of signal and noise subspaces and their respective eigenvalue matrices; estimation of frequencies and magnitudes of remaining harmonic components including closely spaced harmonics linked with incipient faults by use of the MUSIC algorithm; and classification of fault types based on detected harmonic components and their magnitudes.