Smart converter system with connected device monitoring and ai-driven optimization

Abstract

A device monitoring system is disclosed. The system may have a converter configured to deliver AC or DC power, a waveform monitoring module configured to capture electrical characteristics of connected devices, and an Al control unit configured to analyze waveform data. The Al control unit may apply machine-learning algorithms to identify device¬ specific signatures, determine device identity, and help optimize power delivery and safety. The system may further include a communication module for sharing inference models with a cloud server and receiving updated inference data for fleet-wide optimization and confirmation of device identity. The Al control unit may implement feedback loops to update optimization parameters based on device responses and performance over time, enabling predictive maintenance, load balancing among multiple converters, and configuration of device-specific charging profiles. Corresponding methods for waveform acquisition, device identification, adaptive optimization, and cloud-based refinement of inference models are also provided.

Description

SMART CONVERTER SYSTEM WITH CONNECTED DEVICE MONITORING AND AI-DRIVEN OPTIMIZATIONRELATED APPLICATIONS

[0001] This application claims the benefit of U.S. provisional patent application 63 / 748,669, filed January 23, 2025, titled “Smart Battery Inverter System with Connected Device Monitoring and AI-Driven Optimization,” the entirety of the disclosure of which is hereby incorporated by this reference.TECHNICAL FIELD

[0002] This document relates to an inverter / converter system with connected device monitoring and Al-driven optimization that draws power from a modular battery system or another power source.BACKGROUND

[0003] Electronic devices rely on electrical systems power. Modern electronic devices range from smartphones and laptops to electric vehicles and machinery and devices in industrial, aerospace, mining, military, medical, research, scientific, commercial, and consumer applications, all of which rely on electrical systems for power. There is a demand for improved inverters and converters that can supply power to these devices safely and efficiently.SUMMARY

[0004] The present disclosure relates to a device monitoring system comprising a converter electrically coupled to a power source and configured to convert DC power to AC power, a waveform monitoring module configured to detect unique electrical characteristics of connected devices to generate waveform data, and an Al control unit that generates an inference model by analyzing the waveform data, identifies each connected device, and optimizes system performance and safety based on the inference model and the identity of each connected device.

[0005] Particular embodiments may comprise one or more of the following features. The Al control unit may compare waveform data to predefined fault signatures. The Al control unit may be configured to create the inference model using machine learning algorithms to analyze the waveform data, the inference model then being uploaded to a cloud server, and the Al control unit later receiving updated inference data from the cloud server for fleet-wideoptimization. The Al control unit may be configured to update optimization parameters based on a comparison of the inference model with other cloud-shared inference models. The Al control unit may be configured to simultaneously manage multiple connected devices. The Al control unit may be configured to detect and respond to potential safety hazards. The device monitoring system may be coupled to two or more converters, each converter operating in a swarm configuration to provide load balanced power delivery to the connected devices. The Al control unit may implement feedback loops to refine optimization based on device performance over time.

[0006] The present disclosure relates to a device monitoring system comprising a converter electrically coupled to a power source and configured to convert DC or AC power, a waveform monitoring module configured to detect unique electrical characteristics of connected devices to generate waveform data, and an Al control unit that generates an inference model by analyzing the waveform data, the inference model including the identity of each connected device and updated optimization parameters based on connected device responses and performance over time, wherein the Al control unit optimizes system performance and safety based on the inference model

[0007] Particular embodiments may comprise one or more of the following features. The device monitoring system may further comprise a communication module configured to share the inference model with a cloud server and receive updated inference data from the cloud server for fleet-wide optimization. The Al control unit may be configured to update optimization parameters based on other cloud-shared inference models. The waveform monitoring module may record transient waveform anomalies for predictive maintenance. The device monitoring system may redistribute power using load balancing algorithms. The connected device identity in the inference model may be confirmed by separate identity data located in at least one of the other cloud-shared inference models. The device monitoring system may adjust charging profiles based on device identity determined by Al.

[0008] The present disclosure relates to a method for optimizing power delivery using a device monitoring system, comprising detecting and recording a waveform of a device plugged into the device monitoring system and generating waveform data, interpreting the waveform data using an Al control unit to determine the identity of the device, optimizing power delivered to the device based on the device identity, learning from device responses and performance over time to refine optimization parameters, thereby generating an inference model, and sharing the inference model with a cloud server.

[0009] Particular embodiments may comprise one or more of the following features. The method may further comprise receiving updated inference data from the cloud server to improve optimization across a fleet of battery systems or power systems. Learning from device responses may comprise implementing feedback loops to update optimization parameters. Interpreting the waveform data may include applying machine learning algorithms to identify device-specific signatures. The method may further comprise confirming the identity of the device by comparing the device-specific signatures to inference data derived from at least one other cloud-shared inference model.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Implementations will hereinafter be described in conjunction with the appended or included DRAWINGS, where like designations denote like elements, and:

[0011] FIG. 1 is a block diagram of a device monitoring system according to some embodiments with a plurality of devices connected; and

[0012] FIG. 2 is a flowchart of an ALdrive optimization process for a device monitoring system according to some embodiments.DETAILED DESCRIPTION

[0013] Detailed aspects and applications of the disclosure are described below in the following drawings and detailed description of the technology. Unless specifically noted, it is intended that the words and phrases in the specification and the claims be given their plain, ordinary, and accustomed meaning to those of ordinary skill in the applicable arts.

[0014] In the following description, and for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the various aspects of the disclosure. It will be understood, however, by those skilled in the relevant arts, that embodiments of the technology disclosed herein may be practiced without these specific details. It should be noted that there are many different and alternative configurations, devices and technologies to which the disclosed technologies may be applied. The full scope of the technology disclosed herein is not limited to the examples that are described below.

[0015] The singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to “a step” includes reference to one or more of such steps.

[0016] The word "exemplary," "example," or various forms thereof are used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as"exemplary" or as an “example” is not necessarily to be construed as preferred or advantageous over other aspects or designs. Furthermore, examples are provided solely for purposes of clarity and understanding and are not meant to limit or restrict the disclosed subject matter or relevant portions of this disclosure in any manner. It is to be appreciated that a myriad of additional or alternate examples of varying scope could have been presented but have been omitted for purposes of brevity.

[0017] The term “plurality”, as used herein, means more than one. When a range of values is expressed, another embodiment includes from the one particular value to the other particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another embodiment. All ranges are inclusive and combinable.

[0018] Throughout the description and claims of this specification, the words “comprise” and “contain” and variations of the words, for example “comprising” and “comprises”, mean “including but not limited to”, and are not intended to (and do not) exclude other components.

[0019] As required, detailed embodiments of the present disclosure are included herein. It is to be understood that the disclosed embodiments are merely exemplary of the invention that may be embodied in various forms. Therefore, specific structural and functional details disclosed herein are not to be interpreted as limits, but merely as a basis for teaching one skilled in the art to employ the present invention. The specific examples below will enable the disclosure to be better understood. However, they are given merely by way of guidance and do not imply any limitation.

[0020] The present disclosure may be understood more readily by reference to the following detailed description taken in connection with the accompanying figures and examples, which form a part of this disclosure. It is to be understood that this disclosure is not limited to the specific materials, devices, methods, applications, conditions, or parameters described or shown herein, and that the terminology used herein is for the purpose of describing particular embodiments by way of example only and is not intended to be limiting of the claimed inventions.

[0021] More specifically, this disclosure, its aspects and embodiments, are not limited to the specific material types, components, methods, or other examples disclosed herein. Many additional material types, components, methods, and procedures known in the art are contemplated for use with particular implementations from this disclosure. Accordingly, for example, although particular implementations are disclosed, such implementations and implementing components may comprise any components, models, types, materials, versions,quantities, or the like as is known in the art for such systems and implementing components, consistent with the intended operation.

[0022] Applicant notes that the rapid proliferation of electronic devices has significantly increased the demand for efficient and reliable electrical systems. Modem devices, ranging from smartphones and laptops to electric vehicles and industrial machinery, rely heavily on electrical systems for power. As the diversity and complexity of these devices grow, so does the need for electrical systems that can adapt to varying usage patterns, ensure safety, and optimize performance.

[0023] Conventional electrical systems typically supply power in the same way to all connected devices. Similarly, battery management systems focus on preventing overcharging, overheating, or deep discharge, but do so in the same way for all devices. While these systems provide a level of protection and efficiency, they lack the ability to dynamically adjust to the specific characteristics and demands of connected devices. As a result, devices may operate sub-optimally, leading to reduced efficiency, shortened lifespan, and potential safety risks.

[0024] Traditional battery and inverter systems lack the capability to dynamically adjust their performance based on the specific needs of connected devices. Instead, traditional systems perform the same regardless of what devices have been connected. This can lead to inefficiencies, potential safety hazards, and suboptimal performance.

[0025] The presently disclosed device monitoring system 100 not only monitors the connected devices 104 but also dynamically tunes the power output of one or more batteries 192 to optimize performance, enhance efficiency, and ensure safety and security. The presently disclosed device monitoring system 100 thus seamlessly integrates with a wide range of devices 104, adapts to diverse operational requirements, and addresses the unique challenges of modern battery 192 usage in interconnected ecosystems, such as a modular battery system 190.

[0026] FIG. 1 illustrates various embodiments of a modular battery system 190 that includes one or more battery units 192 with a battery management system 194 (BMS) in communication with a device monitoring system 100 electrically coupled to one or more electrical devices 104. Device monitoring system 100 comprises a converter 102, where converter 102 may be a converter, an inverter, a rectifier, or other power modifier for converting: DC to AC, AC to DC, DC to DC (e.g., buck, boost, buck-boost), or AC to AC (e.g., voltage or frequency conversion). The device monitoring system 100 can be grid-connected, utility-connected, or mobile. In some embodiments, the device monitoring system 100 monitors connected or plugged-in devices 104 for their unique waveforms 122 and usesartificial intelligence (Al) (e.g., using Al control unit 110) to tune the battery unit 192 for optimal performance, efficiency, safety, and security.

[0027] Various electrical devices 104 can electrically couple to the device monitoring system 100 via an array of connectors 106 (e.g., USB 3.0, USB-C, DC barrel, HDMI, NEMA 5-15, NEMA 5-20, NEMA 6-20, NEMA 14-30, J1772, CCS1 / CCS2, SAE J3400, and so forth). Connectors 106 can selectively be turned on or off by the device monitoring system 100 or modular battery system 190. For example, USB or low- voltage connectors 106 maybe on while NEMA 14-30 connector(s) 106 are turned off. Multiple converters 102 or multiple device monitoring systems 100 can swarm and operate together to provide uniform power monitoring and power delivery across multiple arrays of connectors 106. In some embodiments, a single device monitoring system 100 includes multiple converters 102, where the multiple converters 102 can be the same or different types of converters 102 (e.g., two inverters, one converter, and a buck-boost converter). In various embodiments, a single device monitoring system 100 is electrically coupled to one or more external converters 102. In some embodiments, multiple device monitoring systems 100 are electrically coupled to each other, and can operate independently in communication with each other or operate in a swarm system to coordinate at least one system or power output. The AC synchronization between multiple converters 102 allows a modular battery system 190 to distribute power in an optimal manner and also balance the power delivery of the modular battery system 190 and batteries 192. In some embodiments, device monitoring system 100 operates outside of and the modular battery system 190 by connecting to another power source like a generator, solar panels, AC main power grids, local power grids, and so forth. In further embodiments, device monitoring system 100 operates by connecting to both a modular battery system 190 (or another battery system) and another power source (e.g., AC main, generator, solar panels, and so forth).

[0028] Electrical devices 104 can be consumer electronics (e.g., television, laptop or personal computer, camera), communications devices (e.g., mobile phone, radio, modem, networking device, satellite transceiver), infrastructure devices (e.g., lighting, HVAC system, plumbing pump), military devices (e.g., encryption device, electronic warfare device, surveillance device, radar system, drone, GPS device), vehicles (e.g., electric vehicle, UAV, naval drone), emergency management devices (e.g., refrigerator, chemical sensor, water treatment system, fuel pump), healthcare devices (e.g., ventilator, diagnostic device, lifesupport system, surgical tool, medical theatres), household devices (e.g., hair dryer, microwave oven, vacuum, clothes washer), and so forth. Clearly, this illustrative inventory of electrical devices 104 is not exhaustive — a person of ordinary skill in the art will know that a litany ofelectrical devices 104 can be electrically coupled to modular battery system 190, device monitoring system 100, and converters 102.

[0029] In some embodiments where separate connectors 106 are used, each electrical device 104 can have a separate waveform 122 that connects to the device monitoring system 100. Thus, if five different electrical devices 104 are connected, then five different waveforms 122 are provided to the device monitoring system 100.

[0030] In alternative embodiments, multiple electrical devices 104 can connect to a single connector 106. For example, if five different electrical devices 104 are connected, then one waveform 122 is provided to the device monitoring system 100 for all five devices 104. Here, the waveform 122 will contain electrical interference because of multiple devices 104 switching on / off and many other behaviors. This interference caused by multiple devices 104 sharing a single waveform 122 is sensed by sensors 108 and analyzed by one or both of the waveform monitoring module 112 and Al control unit 110 to generate interference data characterizing these interference patterns. One or both of the interference patterns and interference data is included in the waveform data 120 for analysis by the device monitoring system 100. Interference-pattern deconvolution uses iterative optimization, reinforcement learning, or blind source separation to isolate device-specific signatures.

[0031] The device monitoring system 100 can also communicate with the devices 104. Some devices 104 like electric vehicles can negotiate power control via communication links with the device monitoring system 100. some devices 104 communicate via protocols like OpenADR and the like. Some large loads including, for example, devices 104 like HVAC units and water boilers can be indirectly controlled via their thermostats. Communication and controls allows the device monitoring system 100 to make sure critical loads, like a ventilator in an emergency center can keep running, while there might not be enough power left to run all the lights and the HVAC at maximum settings.

[0032] Waveforms 122 of electrical devices 104, which represent the time-varying electrical signals such as voltage, current, or power, can be utilized as diagnostic and optimization tools by analyzing the characteristics of waveforms 122 using waveform monitoring module 112. In some embodiments, the device monitoring system 100 comprises one or more sensors 108 configured to monitor the unique waveform characteristics of connected devices 104 by sensing waveforms 122. One or more of sensors 108 can be integrated into waveform monitoring module 112 or sensor(s) 108 can be separate from, and in communication with, waveform monitoring module 112. Waveform data 120 includes waveform 122 data sensed (e.g., by sensor(s) 108) or waveform 122 data extracted by orprovided to components of the device monitoring system 100 (e.g., Al control unit 110, waveform monitoring module 112). Current and voltage are waveform 122 characteristics captured directly (e.g., by sensor(s) 108), while the frequency and phase of waveform 122 are derived from the captured data (e.g., derived by waveform monitoring module 112). Waveform data 120 can include current, voltage, frequency, phase, and other helpful parameters and data (e.g., RMS, peak data, harmonics, wavelet domain features, transients, inrush, spikes, fault signals, topology, power factor, balance, flicker, polarity, skewness, and so forth). The device monitoring system 100 may comprise an Al-driven control unit 110 configured to process the waveform data 120 gathered by the sensors 108 and otherwise derived to optimize the device monitoring system’s 100 performance, efficiency, safety, and security. In some embodiments, the device monitoring system 100 simultaneously manages multiple connected devices 104, ensuring optimal operation for each device 104.

[0033] In some embodiments, the device monitoring system 100 comprises one or more of: a battery unit 192, a converter unit 102 (e.g., an inverter, converter, etc.), a waveform monitoring module 112, and an Al control unit 110. In some embodiments, the battery unit 192 is configured to store electrical energy and supply power to devices 104 connected to the device monitoring system 100. In some embodiments, the converter unit 102 is configured to convert DC power to AC power for use by connected devices 104. In some embodiments, the waveform monitoring module 112 comprises sensors 108 configured to detect the unique electrical characteristics (such as the waveforms 122) of connected devices 104 and generate waveform data 120. The sensors 108 may be high-precision sensors and may be configured to continuously monitor the waveforms 122 of the connected devices 104. The sensors 108 may capture waveform data 120 such as voltage, current, etc., which can be used to derive additional waveform data 120 (e.g., frequency, phase, and so on). In some embodiments, converter 102 has the waveform monitoring module 112 integrated into the converter 102. In some embodiments, converter 102 has the Al control unit 110 integrated into the converter 102.

[0034] In some embodiments, the sensors 108 comprise one or more of a current transformer, a voltage probe, shunt, Rogowski coil, fast current sense amplifier, differential operational amplifier, resistive divider, capacitive divider, high speed successive approximation register analog digital converter (SAR ADC), power-quality (PQ) meters, digital fault recorders, low-power instrument-class sensors (e.g., LPCT), electromagnetic field / flux sensors, optical sensors, temperature sensors, harmonic phasors, and so forth. These sensors 108 are configured to capture real-time waveform data 120 from connected devices 104. In some embodiments, the waveform data 120 captured by the sensors 108 or otherwisederived (e.g., derived by waveform monitoring module 112) is digitized, such as using analog-to-digital converters (ADCs), allowing the waveform data 120 to be processed.

[0035] In some embodiments, the Al control unit 110 is configured to process waveform data 120 and other data from the waveform monitoring module 112 to optimize system performance. In various embodiments or aspects the Al control unit will improve the functioning of converter hardware itself by dynamically modifying switching frequency, output regulation, and battery discharge profiles based on real-time waveform signatures, producing improved physical performance unavailable in prior systems. In some embodiments, the waveform data 120 includes one or more of the amplitude, frequency, phase, harmonic content, and transient behavior of each connected device 104. The Al control unit 110 may implement one or more algorithms such as Fast Fourier Transform (FFT), Wavelet Transform, and a time-domain analysis to process the waveform data 120. As is clear from the illustrative list of sensors 108 and waveform data 120 above, the different types of data included in waveform data 120 can be expansive depending on the system needs of the device monitoring system 100. In some embodiments, the waveform data 120 is generated by two or more of the sensors 108, waveform monitoring module 112, and Al control unit 110.

[0036] The Al control unit 110 may use one or more machine learning algorithms or statistical models to analyze data from the waveform monitoring module 112, extracting significant features, patterns, or both from the waveform data 120. In some embodiments, the Al control unit 110 uses a CNN-based classifier trained on waveform signatures comprising (voltage, current, harmonic spectrum, and transient feature vectors. In some embodiments, the Al control unit 110 is configured to identify patterns and anomalies in the waveforms 122. For example, deviations in harmonic distortion, unexpected transient spikes, or phase imbalances can indicate specific faults, such as insulation breakdown, overloading, or component degradation. The waveform data 120, once extracted, may be compared against baseline data or predefined fault signatures stored in a diagnostic database. In some embodiments, a fault detection algorithm identifies anomalies by calculating deviations or inconsistencies, thus enabling the classification of specific issues, such as short circuits, open circuits, or mechanical wear in motors. The Al control unit 110 and the waveform monitoring module 112 thus integrate to facilitate continuous waveform monitoring, diagnostics, and real-time optimization, through one or more automated controls and user alerts.

[0037] By generating high-quality waveform data 120 from waveform 122 sampled at sufficiently high rates, the ‘signature’ or ‘fingerprint’ of different devices 104 becomes evident to the Al control unit 110, thereby allowing the Al control unit 110 to accurately identify eachdevice 104 and generate an inference model of the device’s 104 behavior. Each electrical device 104 has a different waveform when operating — and, particularly, when connected or started. This is comparable to an electronic fingerprint / signature for each separate device 104. Devices 104 like communication equipment, an electric vehicle, and a hair dryer have very different fingerprints. Even HVAC equipment from different manufacturers have a different fingerprint. This fingerprint is derived from the current and voltage in relation to time. By using various sensors 108, it is possible to detect these unique fingerprints or signatures for each device 104. For example, a selection or combination of shunts, fast current-sense amplifiers, current transformers, Rogowski coils and high-speed comparators as sensors 108 provide the waveform data 120 that enables the Al control unit 110 to identify each device 104. Moreover, the device monitoring system 100 can upload the inference model for a device 104 to a cloud server and compare itself to other inference models stored in the cloud. By comparing data across multiple inference models, a device monitoring system 100 can access highly accurate identity and predictive behavior data for very specific devices 104 (e.g., distinguishing between different manufacturers of mobile phones or HVAC equipment.). Having sufficiently high sampling resolution and using Al-based analysis (e.g., Al control unit 110) of the waveform 122 allow for great accuracy in the inference models. These electronic fingerprints of devices 104 have at least three dimensions: current, voltage and time. Some waveform 122 sampling interval are as quick as 2.5 nanoseconds and zero / threshold detection within a minimum of 150 picoseconds (e.g., using High Resolution Pulse Width Modulation (HRPWM)). In some embodiments, the sampling rate of the waveform 122 is at least 350 microseconds (e.g., sampling at or faster than 250 ps, 200 ps, 125 ps, 50 ps, 20 ps, and so on). In certain embodiments, the sampling rate of the waveform 122 is at least 75 ps for short bursts, long bursts, or continuously (e.g., sampling at or faster than 75 ps, 50 ps, 35 ps, 20 ps, 10 ps, and so on). In further embodiments, the sampling rate of the waveform 122 is in sub-microsecond range for at least a burst duration (e.g., sampling at or faster than 950 ns, 800 ns, 700 ns, 500 ns, 300 ns, and so on) before returning to a lower sampling rate. In some instances, in burst-sampling modes, the system increases sampling frequency from 75 ps to 950 ns to detect device fingerprint transitions.

[0038] The patterns and anomalies extracted from the waveform data 120 may thus allow the device monitoring system 100 to optimize performance, enhance efficiency, ensure safety, and maintain security, as discussed above. For example, performance may be optimized by adjusting the power output to match specific needs of each connected device 104. By incorporating feedback loops, the Al control unit 110 can use the waveform data 120 todynamically optimize operating conditions. This allows the components of the connected devices 104 to last longer than they would if they were used without regard for optimal working conditions. As another example, efficiency may be enhanced by minimizing energy loss through fine-tuning the converter’s 102 operation. By monitoring real-time waveforms 122, parameters such as operating frequency, load balancing, and power factor can be adjusted to enhance efficiency. As another example, safety may be ensured by detecting and responding to potential hazards such as short circuits and overloads. In some embodiments, advanced analytics and machine learning models can predict future failures by identifying trends and gradual waveform 122 deviations that are indicative of progressive wear or aging. This enables proactive maintenance, reducing unplanned downtimes. As another example, security may be maintained by protecting against unauthorized access and tampering.

[0039] The Al control unit 110 may confirm the identity of a connected device 104 by comparing its locally generated waveform signature against inference data derived from other cloud-shared models. This cross-verification enhances accuracy and prevents misclassification of devices with similar electrical characteristics. For instance, two electric vehicles from different manufacturers may exhibit similar base waveforms, but subtle harmonic or other differences stored in cloud-shared inference models allow precise identification of a particular device 104 (here, two EVs).

[0040] The Al control unit 110 may implement feedback loops to refine optimization parameters over time. These loops analyze device 104 performance data and operational outcomes following each adjustment, enabling iterative improvements in power delivery strategies. For example, if a connected medical system exhibits reduced efficiency under certain load conditions, the Al control unit 110 can adjust converter 102 switching frequency, voltage, line frequency, battery 192 discharge rate, etc., and evaluate the resulting performance — thereby converging toward optimal settings. The back-and-forth interplay of the Al control unit 110 and the waveform monitoring module 112 during these feedback loops helps the device monitoring system 100 to more accurately pull apart and distinguish individual electrical devices 104 when multiple devices 104 share a single waveform 122 (e.g., when multiple devices 104 share a single connector 106). The combination of fast waveform 122 sampling rates (with optional bursts of even faster sampling) with the power of the feedback loops using the Al control unit 110 and the waveform monitoring module 112 results in improved ability to discern unique signatures or fingerprints of electrical devices 104 — even when the waveform 122 is cluttered with interference. Feedback loops improve the ability ofthe device monitoring system 100 to tease out the interference patterns and interference data from the waveform 122 to identify individual devices 104.

[0041] In some embodiments, the waveform monitoring module 112 records transient anomalies such as voltage spikes, harmonic distortion, or phase imbalance. These anomalies are analyzed by the Al control unit 110 to predict potential component failures or degradation. For example, repeated detection of high-frequency transients in a motor waveform may indicate bearing wear, prompting proactive maintenance before catastrophic failure occurs.

[0042] In some embodiments, the device monitoring system 100 redistributes power among connected devices 104 using load balancing algorithms. These algorithms consider factors such as device priority, real-time demand, and battery state-of-charge to allocate power efficiently. For example, during peak demand, the device monitoring system 100 may reduce power to non-critical loads while maintaining full power to life-support equipment. The battery management system 194 or device monitoring system 100 may adjust charging and discharging profiles based on device 104 identity determined by the Al control unit 110. For example, when an electric vehicle is connected, the device monitoring system 100 may apply a high-current fast-charging profile, whereas for sensitive medical equipment, the device monitoring system 100 may use a low-ripple, tightly regulated output to ensure operational safety.

[0043] As noted above, the device monitoring system 100 is configured to use Al and machine learning to improve the algorithms used and decisions made over time. For example, in some embodiments, such as the embodiment shown in FIG. 2, a device 104 may be plugged into 204 the device monitoring system 100 having a battery 192 that is ready 202. The device monitoring system 100 detects and records 206 the waveform 122 of the device 104 and generates waveform data 120. The Al control unit 110 interprets the waveform data 120 to determine the signature and identity of the connected device 104 and shares this information 208. The device monitoring system 100 may adjust 210 the battery management system 194 and other systems to optimize the power delivery to the connected device 104. Over time, the Al control unit 110 learns 212 from the response of the device 104 to the device monitoring system 100 and the performance of the device 104 over time. The device monitoring system 100 may be configured to share an inference model back to the cloud 214, and the cloud may share updated inferences 216 with all device monitoring systems 100 in a fleet of device monitoring systems 100.

[0044] The optimal operating conditions for a particular device 104 not only are affected by the type of device 104, but also by how the device 104 is used. Thus, the device monitoring system 100 may be configured to also monitor how each device 104 is used and how the device104 responds to that use. For example, if a particular device 104 is used for hours every day, the device monitoring system 100 can track that use and learn how the device 104 responds to that use, as well as the performance of the device 104 under the optimized conditions provided by the device monitoring system 100. This information can be shared with the cloud so that, if another user begins using a similar device 104 in a similar way, the device monitoring system 100 can take the use into account in optimizing performance over time. In this way, the Al and machine learning make it possible for the device monitoring system 100 to provide benefits in performance that are otherwise unavailable.

[0045] In some embodiments, the Al control unit 110 is configured to communicate with a remote cloud server to upload inference models generated from waveform data 120 of connected devices 104. The cloud server aggregates inference models from multiple device monitoring systems 100 and distributes updated optimization parameters back to each device monitoring system 100. This fleet-wide learning approach enables continuous improvement in device 104 identification accuracy and optimization strategies across geographically dispersed installations. For example, if a first device monitoring system 100 identifies a unique waveform signature for a new electric vehicle model, that signature can be shared with other device monitoring systems 100 to enhance recognition and performance optimization for similar devices 104.

[0046] In certain embodiments, two or more converters 102 operate in a coordinated swarm configuration. Each converter 102 communicates with others to balance load distribution and synchronize AC output parameters such as voltage and frequency. This swarm approach improves device monitoring system 100 resilience and efficiency, particularly in high-demand scenarios where multiple devices 104 require simultaneous power delivery. For instance, in a modular battery system 190 supporting both residential loads and electric vehicle charging, swarm coordination ensures stable power quality and prevents overloading of individual converters 102.

[0047] The present disclosure is also related to a method of managing electrical devices 104. The device monitoring system 100 may initialize and perform a self-check to ensure all components are functioning correctly. The waveform monitoring module 112 may continuously collect data from connected devices 104. Monitoring of the waveform 122 by the waveform monitoring module 112 may comprise monitoring the electrical grid to which the device 104, the waveform monitoring module 112, the device monitoring system 100, the Al control unit 110, the converter 102, sensors 108, or any other component is coupled. Monitoring the waveform 122 or other feature of the electrical grid may provide advantagesfor charging batteries 192 or for operating directly from a power grid and collecting data from the same, such as to detect, plan for, and use patterns, features, or aspects of the power grid. The Al control unit 110 may analyze the collected data to identify the unique waveform 122 characteristics of each device 104. The Al control unit 110 may, based on the analysis, adjust the battery 192 and converter 102 settings to optimize performance, efficiency, safety, and security. The device monitoring system 100 may simultaneously manage multiples devices 104, ensuring each operates at its optimal level. Many additional implementations are possible. Further implementations are within the CLAIMS.

[0048] It will be understood that implementations of the device monitoring system 100 include but are not limited to the specific components disclosed herein, as virtually any components consistent with the intended operation of various device monitoring systems may be utilized. Accordingly, for example, it should be understood that, while the drawings and accompanying text show and describe particular device monitoring system implementations, any such implementation may comprise one or more of any shape, size, style, type, model, version, class, grade, measurement, concentration, material, weight, quantity, and the like consistent with the intended operation of device monitoring systems.

[0049] The concepts disclosed herein are not limited to the specific device monitoring systems 100 shown herein. For example, it is specifically contemplated that the components included in particular device monitoring systems may be formed of any of many different types of materials or combinations that can readily be formed into shaped objects and that are consistent with the intended operation of the device monitoring system.

[0050] Furthermore, device monitoring systems may be manufactured separately and then assembled together, or any or all of the components may be manufactured simultaneously and integrally joined with one another. Manufacture of these components separately or simultaneously, as understood by those of ordinary skill in the art, may involve any suitable arrangement.

[0051] In places where the description above refers to particular device monitoring system implementations, it should be readily apparent that a number of modifications may be made without departing from the spirit thereof and that these implementations may be applied to other implementations disclosed or undisclosed. The presently disclosed device monitoring systems are, therefore, to be considered in all respects as illustrative and not restrictive.

Claims

1. CLAIMSWe claim:

1. A device monitoring system, comprising:a converter electrically coupled to a power source and configured to convert DC power to AC power;a waveform monitoring module configured to detect unique electrical characteristics of connected devices to generate waveform data; andan Al control unit that generates an inference model by analyzing the waveform data, identifies each connected device, and optimizes system performance and safety based on the inference model and the identity of each connected device.

2. The device monitoring system of claim 1, wherein the Al control unit compares waveform data to predefined fault signatures.

3. The device monitoring system of claim 1, wherein the Al control unit is configured to create the inference model using machine learning algorithms to analyze the waveform data, the inference model then being uploaded to a cloud server, and the Al control unit later receiving updated inference data from the cloud server for fleet-wide optimization.

4. The device monitoring system of claim 1, wherein the Al control unit is configured to update optimization parameters based on a comparison of the inference model with other cloud-shared inference models.

5. The device monitoring system of claim 4, wherein the Al control unit is configured to simultaneously manage multiple connected devices.

6. The device monitoring system of claim 4, wherein the Al control unit is configured to detect and respond to potential safety hazards.

7. The device monitoring system of claim 4, wherein the device monitoring system is coupled to two or more converters, each converter operating in a swarm configuration to provide load balanced power delivery to the connected devices.

8. The device monitoring system of claim 4, wherein the Al control unit implements feedback loops to refine optimization based on device performance over time.

9. A device monitoring system, comprising:a converter electrically coupled to a power source and configured to convert DC or AC power;a waveform monitoring module configured to detect unique electrical characteristics of connected devices to generate waveform data; andan Al control unit that generates an inference model by analyzing the waveform data, the inference model including the identity of each connected device and updated optimization parameters based on connected device responses and performance over time, wherein the Al control unit optimizes system performance and safety based on the inference model.

10. The device monitoring system of claim 9, further comprising a communication module configured to share the inference model with a cloud server and receive updated inference data from the cloud server for fleet-wide optimization.

11. The device monitoring system of claim 10, wherein the Al control unit is configured to update optimization parameters based on other cloud-shared inference models.

12. The device monitoring system of claim 11, wherein the waveform monitoring module records transient waveform anomalies for predictive maintenance.

13. The device monitoring system of claim 12, wherein the device monitoring system redistributes power using load balancing algorithms.

14. The device monitoring system of claim 11, wherein the connected device identity in the inference model is confirmed by separate identity data located in at least one of the other cloud-shared inference models.

15. The device monitoring system of claim 14, wherein the device monitoring system adjusts charging profiles based on device identity determined by Al.

16. A method for optimizing power delivery using a device monitoring system, comprising:detecting and recording a waveform of a device plugged into the device monitoring system and generating waveform data;interpreting the waveform data using an Al control unit to determine the identity of the device;optimizing power delivered to the device based on the device identity;learning from device responses and performance over time to refine optimization parameters, thereby generating an inference model; andsharing the inference model with a cloud server.

17. The method of claim 16, further comprising receiving updated inference data from the cloud server to improve optimization across a fleet of battery systems or power systems.

18. The method of claim 17, wherein learning from device responses comprises implementing feedback loops to update optimization parameters.

19. The method of claim 16, wherein interpreting the waveform data includes applying machine learning algorithms to identify device-specific signatures.

20. The method of claim 19, further comprising confirming the identity of the device by comparing the device-specific signatures to inference data derived from at least one other cloud-shared inference model.

Images