WO2022147299A1 - Adaptive power for ai/ml systems and methods in a microgrid - Google Patents

Abstract

A microgrid system including a first and second generating systems configured to generate electric energy, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing. The microgrid also includes an AI/ML module configured to train a control model using one or more AI/ML algorithms and the excess electric energy. The microgrid also includes a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module and configured to generate control signals for both the first and second generating systems based on at least the control model.

Description

ADAPTIVE POWER FOR AI/ML SYSTEMS AND METHODS IN A MICROGRID

CROSS-REFERENCE

[0001] This application claims the benefit of U.S. Provisional Applications No. 63/132,998, filed December 31, 2020, and U.S. Provisional Application No.: 63/133,108, filed December 31, 2020, all of which applications are incorporated herein by reference.

TECHNICAL FIELD

[0002] The disclosure generally relates to power and energy generation and control by a microgrid, and more particularly, to adaptive power control for AI/ML components and methods for distributed energy resources in a microgrid.

BACKGROUND

[0003] Microgrids that are designed to control a number of distributed energy resources provide an economical, resilient, and efficient source of power and energy to associated host facilities. There is a need for systems and methods to optimize the deployment and performance of microgrids and the controlled distributed energy resources.

BRIEF DESCRIPTION OF THE DRAWINGS

[0004] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized. Embodiments of the disclosure may be better understood by referencing the accompanying drawings.

[0005] FIG. 1 depicts a microgrid system and host facility.

[0006] FIG. 2 depicts a microgrid system and host facility that includes at least one AI/ML module, according to some embodiments.

[0007] FIG. 3 depicts a microgrid system and host facility that includes a number of AI/ML modules, according to some embodiments.

[0008] FIG. 4 depicts a microgrid system and host facility that includes a number of AI/ML modules, according to some embodiments. [0009] FIG. 5 depicts a microgrid system and host facility that includes a number of AI/ML modules, according to some embodiments.

[0010] FIG. 6 depicts a flowchart of example operations for deriving asset-specific training data to train and use an AI/ML module, according to some embodiments.

[0011] FIG. 7 depicts a flowchart of example operations for deriving asset-specific training data to train and use an AI/ML module, according to some embodiments.

[0012] FIG. 8 depicts a flowchart of example operations for deriving asset-specific training data to train and use an AI/ML module, according to some embodiments.

[0013] FIG. 9 depicts an example computer, according to some embodiments.

[0014] FIG. 10 depicts a flowchart of example operations to train and use an AI/ML module, according to some embodiments.

DETAILED DESCRIPTION OF EMBODIMENTS

[0015] The description that follows includes example systems, methods, techniques, and program flows that embody embodiments of the disclosure. However, it is understood that this disclosure may be practiced without these specific details. For instance, this disclosure refers to example distributed energy resources (DERs) at example locations for generating power.

However, such DERs can be located at other locations and different groupings of DERs may be implemented. For example, DERs can be positioned at any location on-site in a variety of different sizes and arrangements. In other instances, well-known instruction instances, protocols, structures, and techniques have not been shown in detail in order not to obfuscate the description.

[0016] Before the present invention is described in greater detail, it is to be understood that this invention is not limited to particular embodiments described, as such may, of course, vary. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting, since the scope of the present invention will be limited only by the appended claims.

[0017] Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limit of that range and any other stated or intervening value in that stated range, is encompassed within the invention. The upper and lower limits of these smaller ranges may independently be included in the smaller ranges and are also encompassed within the invention, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the invention.

[0018] Certain ranges are presented herein with numerical values being preceded by the term "about." The term "about" is used herein to provide literal support for the exact number that it precedes, as well as a number that is near to or approximately the number that the term precedes. In determining whether a number is near to or approximately a specifically recited number, the near or approximating unrecited number may be a number which, in the context in which it is presented, provides the substantial equivalent of the specifically recited number.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Unless otherwise indicated or apparent from context, percentages given herein are w/w. Although any methods and materials similar or equivalent to those described herein can also be used in the practice or testing of the present invention, representative illustrative methods and materials are now described.

[0020] All publications and patents cited in this specification are herein incorporated by reference as if each individual publication or patent were specifically and individually indicated to be incorporated by reference and are incorporated herein by reference to disclose and describe the methods and/or materials in connection with which the publications are cited. The citation of any publication is for its disclosure prior to the filing date and should not be construed as an admission that the present invention is not entitled to antedate such publication by virtue of prior invention. Further, the dates of publication provided may be different from the actual publication dates which may need to be independently confirmed.

[0021] It is noted that, as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise. It is further noted that the claims may be drafted to exclude any optional element. As such, this statement is intended to serve as antecedent basis for use of such exclusive terminology as "solely," "only" and the like in connection with the recitation of claim elements, or use of a "negative" limitation.

[0022] As will be apparent to those of skill in the art upon reading this disclosure, each of the individual embodiments described and illustrated herein has discrete components and features which may be readily separated from or combined with the features of any of the other several embodiments without departing from the scope or spirit of the present invention. Any recited method can be carried out in the order of events recited or in any other order which is logically possible.

[0023] Electric power can be generated at one or more central power stations and distributed regionally over great distances over a distribution network to the end-users of the electric power. The power stations come in a variety of different forms. For example, some power stations generate power using oil, natural gas, coal, or other fossil fuels which are burned releasing carbon dioxide (CO2) and other gases to create heat to turn a turbine that in turn runs an alternator or generator that generates electricity. Other heat sources for large scale power station includes geothermal heat and nuclear fuels and solar arrays. Hydroelectric power stations and wind are another form of large-scale power stations that can generate power for distribution on a large scale grid system. These large scale systems, although the norm, are subjected to a host of issues. For example, the distribution network itself is extremely costly to build and maintain. The distribution network also wastes vast amounts of land while endangering birds and other wildlife and habitats. Further, the distribution network itself is a huge source of lost power that dissipates during transmission. Additionally, the power stations themselves each carry inefficiencies and massive upfront costs. Therefore, the power stations and the associated distribution networks are difficult to fund and bring online to meet shifting energy needs. Further, by being centralized to a single location the geographical benefits found elsewhere are not harnessed for power. Further, having the power generating assets located at a single location also creates a larger susceptibility to localized weather patterns and security threats.

[0024] In response to the issues that the current power grid faces, a new era of distributed generation has sprouted up to meet the energy and environmental needs of all. These systems are sometimes found in the form of a microgrid which includes a number of distributed energy resources (DERs) and some form of control of those resources in an on-site or quasi-onsite manner. The DERs may include, but are not limited to, any number of sustainable and/or renewable power generation resources. Distributed generation systems provide an economical and efficient source of electrical power and thermal energy to associated host facilities. There is a need for systems and methods to optimize the deployment and performance of these distributed generation systems. [0025] A microgrid is a smart, small-scale version of the centralized electricity system which can generate, distribute, and regulate the flow of electricity at a local level. In addition to keeping the lights on when the central grid fails, microgrids are less polluting and more efficient than grid power because their energy is created close to where it is used. Delivering power from afar is inefficient in view of eight to fifteen percent of it dissipating in transit.

[0026] Microgrids employ a wide range of green power technologies including solar, wind, fuel cells, combined heat and power (CHP) plants, and energy storage. Intelligent microgrids integrate renewables to seamlessly balance the variable output of green power with traditional generation. In doing so, the microgrid overcomes the inconsistencies of solar and wind by automatically tapping into other resources when renewable energy isn’t available.

[0027] In some embodiments, the microgrid controller can be structured as a center point of the entire system. In some embodiments, the microgrid controller may take different forms. A microgrid can be tasked with managing power generating assets, batteries, associated switching and power management hardware, and even nearby building energy systems at a host facility. This can allow the microgrid the ability and capability of orchestrating multiple resources to meet the energy goals of a host facility and customer efficiently and resiliently.

[0028] In some embodiments, at the plant level, intelligent microgrids use machine learning (ML) and artificial intelligence (Al) to find opportunities to achieve better results and manage assets in real-time and predictively. At the grid level, intelligent systems can be networked together to quickly dispatch stored or generated power. The result of this AI/ML usage can provide a group of highly intelligent, disparate microgrid systems communicating to optimize their network of DERs.

[0029] In some cases, as a software-based system, the controller can manage energy supply in different ways. As electricity prices fluctuate throughout the day, the controller orchestrates the play of its assets. When demand for grid energy is high and prices go up, the controller signals the microgrid to use more of its own resources. If the microgrid has excess capacity, it may sell it to the grid. Working together, the microgrid’s resources are greater than the sum of its parts, driving system performance to a level of efficiency none could do alone. AI/ML Processing and Power Requirements

[0030] AI/ML systems and methods have large processing requirements. For example, AI/ML usage can be a computationally intensive endeavor that benefits most from large data sets to draw from that are free of downsampling, filtering, or selective precluding due to storage or transmission constraints on the system that exists between the data sources and the processing AI/ML unit. For example, the step of training an algorithm is a computationally-intensive process. Further, just as computationally demanding are the steps that come before algorithm training in order to provide the mass of data needed to train the algorithm. This data must be collected, stored, transmitted, categorized, processed, prepped, and then transmitted and stored again for training usage. Once trained, using the trained models can still present as a large computationally intense event. Further, events should be monitored to detect when a model should be retrained due to defined triggering events that would predicate such a need. Further, if customized AI/ML control is proposed for different assets, the processing and power needs are only multiplied to deal with the additional complexity of multiple trainings.

[0031] Another important requirement that will also be addressed in turn is the large energy /power requirements of an AI/ML system and method implementation. For example, these computationally intense systems and methods translate to large energy consumption events to facilitate the running of all the electrical processing hardware needed to accomplish the processing tasks of collecting, training, and using etc. AI/ML systems and methods which may be implemented using one or more AI/ML module(s). Accordingly, it is appreciated that AI/ML usage can be an energy and power intensive endeavor. For example, in some cases, training an algorithm is the most computationally and power intensive portion of the process as noted above. Specifically, some estimates suggest that the carbon footprint of training a single Al is as much as 284 tonnes of carbon dioxide equivalent which is roughly five times the lifetime emissions of an average car.

[0032] Further, before algorithm training, a mass of data needs to be collected, categorized, processed, prepped, and stored for training usage. Once trained, using the trained models can still present as a large energy and time consumption event. Further, events should be monitored to detected when a model should be retrained due to defined triggering events that would predicate such a need. Additionally, if customized AI/ML control is proposed for different assets, the processing and power needs are only multiplied to deal with the additional complexity. AI/ML Embodiments for Helping Processing and Power Requirements

[0033] One or more embodiments include a directly integrated AI/ML module in the microgrid system and/or controller that can turn on or off depending on sensor information received. This localized AI/ML module provides for an AL/ML mode that is case-specific and location deterministic. Further, localization allows for algorithmic parity based on what is best at each location and sensor data sources, and desired outcome and control. This can provide for more efficient and more accurately trained models providing faster control and system generation initially that is not present in the current state of the art. In some embodiments, the AL/ML module is integrated directly into a microcontroller that makes up a microgrid controller or other logic in the system. In some embodiments, the AI/ML module can take the form of an AL/ML firmware that is placed within each sensor device to allow for some sensor level processing before transmitting to the controller allowing for distributed processing to take advantage of the sensor processing power as a supplement to the controllers processing capacity which has not been done before.

[0034] The AI/ML integrated embodiments can provide a system that makes use of all computational capabilities already in the system and can also be provided as standalone new modules that are added to the microgrid. The new modules can be placed at the source of the data that is going to be used to train and also use the modules to help mitigate and eliminate many transmission step processing needs on the system.

[0035] One or more embodiments include a directly integrated AI/ML module in the microgrid controller that can turn on or off depending on sensor info received allowing for on-demand and real-time training and model usage when in an on-state that can be referred to as an AI/ML mode. This can provide resource-specific support as needed.

[0036] In some cases, a high usage of computation ability and time is a requirement of AI/ML processing. Having the ability to allow for the resources or other elements of the system to indicate when best to implement the AI/ML module and also to special tailor the usage allows for optimizations in computational usage, accuracy, and efficiency. For example, the AI/ML module can be used in an on-demand manner by a sensor based on sensor readings in a way that allows for situations where such processing is not needed to take advantage of that and run in a simpler faster state of operation. Also different sensor readings can precipitate different algorithm selection and training routines as well as the generation of different control outcomes using the AI/ML modules. [0037] In one or more embodiments, the novel ability to run in a multi-modal manner provides the microgrid an AI/ML mode that allows for the system to run in a simple, fast, or more responsive mode and then switch when needed to make smart AI/ML-driven adjustments and decisions based on the actual sensor feedback or other inputs to the AI/ML module. For example, when sensor reading and user and system needs and indicators show that nothing is occurring in the system that would require complex or even simple AI/ML support, the system can operate in a simple state saving computational cycles, power, etc. and when a change is detected the AI/ML module can be brought online to either train or used if the model is already trained depending on what is detected.

[0038] The unexpected results provided by this arrangement is a surprising time savings during simple operating conditions but also allows for complex AI/ML control when conditions incident would predicate such a need based on system feedback.

[0039] In addition, the AI/ML module, or a portion thereof, can be integrated directly as AL/ML firmware in each sensor device to allow for some sensor level processing before transmitting to the controller allowing for distributed processing to take advantage of the sensor processing power as a supplement to the controllers processing capacity.

[0040] In one or more embodiments, the sensor level processing allows for smart sensor data to be fed back allowing for smart AI/ML-driven control through sensors while the overall system may be operating in a simpler faster mode of operation.

[0041] In one or more embodiments, the distributed processing allows for the microgrid controller to focus on certain processing while also being able to delegate other items to sensor level AI/ML module processing.

[0042] In one or more embodiments, having the AI/ML directly running on the sensor allow for special tailoring of the module and also allows the module the ability to pull additional data as required from the sensor of which it is a part. Also, this allows the module to even request certain readings to further refine its iterative processing, etc.

[0043] In one or more embodiments, the on-sensor processing allows for any difficulty in transmission means to be accounted for by allowing for less data to be sent from a sensor to the controller in certain situations. In one or more embodiments, the microgrid controller is connected to a bunch of sensors. The sensors are placed at locations all over the system. For example, the sensors can be placed on, or in, each of the DER elements of the overall microgrid system. Sensors can also be placed along transmission lines. Further sensors could be placed at the receiving party’s host facility end as well.

[0044] In one or more embodiments, the motivating factors can include the fact that at times the processing takes too long to react to the immediate need to adjust to deal with large power fluctuations up and down. Also even in scenarios where time was available to process, the processing power of the microcontroller was lacking to complete some of the more advanced requests. Further, in one or more embodiments, the sensors overtaxed the system and communication channels by sending massive amounts of data overloading at times the microcontroller storage requiring far heavier controller level memory installs. Further, in one or more embodiments, there were limitations on sensor placement due to the need to make sure they could have the fidelity to transmit the sensor data.

[0045] In one or more embodiments, the advantages include cheaper microgrid controllers, more flexibility on sensor locations, more complex control, and processing possible. Some design considerations may include the increased load on sensors and the complexity of the sensor to microgrid controller processing and management schemes. Some alternatives, in accordance with one or more embodiments, include having a mix of sensor families where some do have AI/ML modules and others do not. In one or more embodiments, special purpose made modules could be included in sensors or elsewhere. In one or more embodiments, the system may have multi-mode sensors as well similar to the microgrid controller. Also for legacy systems, in accordance with one or more embodiments, a mini controller that handles the AI/ML sensors could be created that then translates for the already present legacy microgrid controller. In one or more embodiments, the multi-mode AI/ML microgrid controller is configured to handle a mix of both legacy sensors and AI/ML integrated sensors. The general utility of the AI/ML module is all its disclosed forms allow for a microgrid to have a bunch of options when it comes to controlling the system.

Example Microgrid Systems

[0046] FIG. 1 depicts a first example of microgrid 100 that includes a microgrid system and a host facility 122. In particular, FIG. 1 illustrates an example of a microgrid system 110.

[0047] As shown, in some cases, a microgrid system 110 may include a microgrid controller

112. The microgrid controller 112 can include electrical and communicative connections with all other elements of the microgrid 100. For example, the microgrid controller 112 can have direct connections with an energy storage system 114, a power generation system 116, and a host facility 122. The microgrid controller may also have connections with sensors 118, 120, and 124. The microgrid controller may receive signaling from any one of the connected elements at any given time. Further, the microgrid controller 112 can provide feedback and control signaling to the different elements 114, 116, 122, 118, 120, and 124 based on that which is received or has been received previously by the microgrid controller 112.

[0048] The microgrid system 110 may also include an energy storage system 114 as shown. In one or more embodiments, the energy storage system 114 may be any one from a group consisting of one or more of a battery, a gravity displacement system, a chemical conversion storage system, a compressed gas storage system, and/or a thermal energy storage system. In some embodiments, the energy storage system 114 may be a plurality of systems. For example, the energy storage system 114 could be a large array of batteries. In another example, the energy storage system 114 can be a combination of batteries and one or more thermal energy storage systems.

[0049] The microgrid system 110 may also include a power generation system 116 as shown. In one or more embodiments, the power generation system 116 may be one or more of a solar array, a wind turbine, a fuel cell, a cogeneration unit, a hydrogen generator, a natural gas generator, a gas turbine, a hydrogen turbine, a geothermal generator, a biogas generator, and/or a battery. In some cases, multiple power generation system 116 can be included. For example, a power generation system 116 may be made up of a solar array, a complement of batteries, and some cogeneration generators sized to provide a baseload as required by the host facility 122.

[0050] FIG. 2 depicts a second example microgrid 200 according to some embodiments. FIG. 2 illustrates an example of microgrid 200 that includes an example microgrid system 210 and host facility 222. Similar to the microgrid and host facility of FIG. 1, microgrid 200 includes the microgrid system 210 and host facility 222. Some of components and the configuration of the microgrid system 210 are similar to the microgrid 110. Specifically, microgrid 210 includes a microgrid controller 212, an energy storage system 214, a power generation system 216, and sensors 218, 220, and 224.

[0051] However, in contrast to the microgrid 100 of FIG. 1, a microgrid system 210 as shown in FIG. 2 further includes an AI/ML module 230 that can be on, in, or near the microgrid system 210 as shown. This AI/ML module 230 is designed and implemented to directly receive and train and use selected algorithms based on the microgrid system 210 and the host facility 222 capabilities and needs.

[0052] FIG. 3 depicts a second example microgrid 300 according to some embodiments. FIG. 3 illustrates an example of microgrid 300 that includes an example microgrid system 310 and host facility 322. Similar to the microgrid and host facility of FIG. 1, microgrid 300 includes the microgrid system 310 and host facility 322. Some of the components and the configuration of the microgrid system 310 are similar to the microgrid 110. Specifically, microgrid 310 includes a microgrid controller 312, an energy storage system 314, a power generation system 316, and sensors 318, 320, and 324.

[0053] However, in contrast to the microgrid 100 of FIG. 1, a microgrid system 310 as shown in FIG. 3 further includes a number of discrete localized AI/ML modules 332, 334, 336, and 338. Specifically, an AI/ML module 332 is provided that can be on, in, or near the microgrid controller 312. This AI/ML module 332 is designed and implemented to directly receive and train and use selected algorithms based on the microgrid controller 312 capabilities and needs.

[0054] Further, an AI/ML module 334 is provided that can be on, in, or near the energy storage system 314. This AI/ML module 334 is designed and implemented to directly receive and train and use selected algorithms based on the energy storage system 314 capabilities and needs. Further, as shown, the AI/ML module 334 can be placed within or near the sensor 318 that is located at the energy storage system 314. This can allow the AI/ML module 334 to directly receive sensor information without any latency or loss or the need to implement large bandwidth communication systems.

[0055] Another AI/ML module 336 is provided that can be on, in, or near the power generation system 316. This AI/ML module 336 is designed and implemented to directly receive and train and use selected algorithms based on the power generation system 316 capabilities and needs. Further, as shown, the AI/ML module 336 can be placed within or near the sensor 320 that is located at the power generation system 316. This can allow the AI/ML module 336 to directly receive sensor information without any latency or loss or the need to implement large bandwidth communication systems.

[0056] Similarly, AI/ML module 338 is provided that can be on, in, or near the host facility 322.

This AI/ML module 338 is designed and implemented to directly receive and train and use selected algorithms based on the host facility 322 capabilities and needs. Further, as shown, the AI/ML module 338 can be placed within or near the sensor 324 that is located at the host facility 322. This can allow the AI/ML module 338 to directly receive sensor information without any latency or loss or the need to implement large bandwidth communication systems.

[0057] FIG. 4 depicts a second example microgrid 400 according to some embodiments. FIG. 4 illustrates an example of microgrid 400 that includes an example microgrid system 410 and host facility 422. Similar to the microgrid and host facility of FIG. 1, the microgrid 400 includes the microgrid system 410 and host facility 422. Some of the components and the configuration of the microgrid system 410 are similar to the microgrid 110. Specifically, microgrid 410 includes an energy storage system 414, a power generation system 416, and sensors 418, 420, and 424.

[0058] However, in contrast to the microgrid 100 of FIG. 1, a microgrid system 410 as shown in FIG. 4 further includes a distributed microgrid controller network made up of disparate microgrid controllers 411, 413, and 415 that each includes their own AI/ML modules 431, 433, and 437, respectively. Additionally, sensors 420 and 424 also include localized AI/ML modules 436 and 438. These AI/ML modules that can be on, in, or near the associated elements are designed and implemented to directly receive and train and use selected algorithms based on the capabilities of the associated element and needs to which they are near or attached or within.

[0059] In some cases, an AI/ML module may be included outside the usual computational controller or sensor portions. For example, as shown an AI/ML module 435 may be provided as a standalone element within, for example, the energy storage system 414. This standalone element can provide additional and localized support as needed based on the specific element it is found within.

[0060] FIG. 5 depicts a second example microgrid 500 according to some embodiments. FIG. 5 illustrates an example of microgrid 500 that includes an example microgrid system 510 and host facility 522. Similar to the microgrid and host facility of FIG. 4, the microgrid 500 includes the microgrid system 510 and host facility 522. Some of the components and the configuration of the microgrid system 510 are similar to the microgrid 410. Specifically, microgrid 510 includes microgrid controllers 511,513, and 515, an energy storage system 514, a power generation system 516, and sensors 518, 520, and 524.

[0061] However, in contrast to the microgrid 400 of FIG. 4, a microgrid system 510 as shown in FIG. 5 further includes an AI/ML module 532 that can be on, in, or near a central microgrid controller 512 as shown, similar to FIG. 3. This AI/ML module 532 is designed and implemented to directly receive and train and use selected algorithms based on the microgrid system 510 and the host facility 522 capabilities and needs as a function of its central control operations.

[0062] As discussed, and provided in the above one or more embodiments, a first core requirements that should be provided for are the computational requirements of running an AI/ML module to provide the desired AI/ML support for the microgrid and associated assets. The other requirements that also needs to be specifically addressed, and which is the subject of the following embodiments, is providing systems and methods for providing for and handling of the power/energy requirements that are also needed for accomplishing the AI/ML processing.

AI/ML Adaptive Power Control for Microgrid

[0063] As noted above, according to one or more embodiments, a microgrid may use machine learning and artificial intelligence to train different models to help control more accurately what is going on in the microgrid. Training an Al module can require a large amount of processing capability as well as power. The power requirements to run a sufficient amount of processing to provide a trained model that can be used later to provide control can be a large amount of time and electrical power.

[0064] Therefore an overarching AI/ML structure may be required to control the microgrid at a macro level when looking to implement the different DER and the interplay between those when providing power to the host site/facility and what power can be used to for AI/ML processing.

[0065] In one or more embodiments, power from one or more DERs can be measured and monitored. Further, the host facility can be provided the requisite power leaving, at times leftover/unused power. When this power is detected, it can be used to for AI/ML processing. In one embodiment, for example, if excess power is identified for AI/ML processing at a first DER asset, that power can be routed to one or more AI/ML modules in the system that is in needs of power to complete a queued processing request. In one or more embodiments, if the power may be better fed directly to a local AI/ML module found at the first DER and the processing request and data may be transferred to that AI/ML module. This determination on whether to move the power to the AI/ML modules or to move the data and processing to the power source can be done based on a host of parameters. For example, some of these parameters may include thresholds determining if it is more efficient to use to power locally and move the data or if it makes more sense to instead transfer the power to where processing is already ongoing or where data is more difficult to move and transmit. For example, if communication channels and bus assets are being taxed it may make sense to move to power to where it is needed. In another case, when the power conduits are being taxes and the communication channels are running light on data, it may be better to move the data and processing requests.

AI/ML for DER

[0066] A microgrid may contain a number of different invariant or variant DERs. The individual DERs may require individual AI/ML trained models in order to control them accurately.

[0067] Further in an overall microgrid each of these DERs may each require their own specialized set of customized AI/ML modeling and trainings for those models based on the complexities and differences of those DERs.

[0068] As noted, in some cases DERs who are generating excess power can be the same that also require AI/ML processing. In this case the excess power can be directly applied locally to the AI/ML module in need of power to process data. In some cases, other AI/ML modules in the system may be helping process the data. In these cases power can be sent to them as well as permitted based on the amount of power available and the capacity to transmit the power without diluting the required load delivery to a host facility, storage device, or optimal grid connect sale of power, etc.

AI/ML for different states of a single DER

[0069] In some cases, a single DER may need to have multiple trained AI/ML modules to control the different facets of required use by a resource owner when different activities are required of the device. For example, the device may be called upon to generate power or two shift that power into a storage device. In other cases, the device may be called upon to react to host location activity in the form of ramping up or ramping down power needs at the host. In some cases, the DER may be acting instead as a energy storage apparatus, in the case of certain batteries or capacitors for example, then at other times they operate as power providing assets. A number of other factors may be occurring at a given DER that could be inputs and elements that may affect how that DER should respond and be controlled all of which should be taken into account by an AI/ML module or multiple AI/ML modules and power supplying and usage.

Layered AI/ML approach for microgrid [0070] Accordingly, in order to effectively control a microgrid with an AI/ML module a layered approach may be needed that includes local AI/ML as determined by each of the connected DERs as well as an overarching AI/ML modeling and control structure for the overall microgrid. In this case, power generation parameters can also be used to further determine which elements to power for processing based on one or more of the power available, the location of the power, the power conduit condition for transmitting, the data location, the communication transmission capacity, data storage capacity, and/or queue order of what AI/ML processing is recommended for processing.

Network layered AI/ML approach for network of Microgrids

[0071] Further in advanced network of microgrid scenarios a higher-level Al modeling may be required to allow for the interplay between multiple different micro grids with multiple different architectures. In such an environment, one or more cases may be provided where a first microgrid is generating excess power without a need for any further AI/ML processing where a second microgrid may be in the opposite position. In this case, the first microgrid may send power to the second to allow the second to process its AI/ML requests. Alternatively, in another embodiment, the second microgrid may send the data and AI/ML request to the first microgrid depending on communication transmission capabilities and processing capabilities of the two microgrids. For example, it is possible that the first microgrid not only had excess power but also is build with more processing capability due to it having more DERs etc. Accordingly, the second microgrid with its diminished processing and power capabilities, but still having a need for AI/ML processing, may send the data and request to the first microgrid for training of the AI/ML models etc.

[0072] Further, in some cases, once trained, the second microgrid may realize efficiency is power management and processing that leaves it with excess power and computing power of its own previously unrealized prior to AI/ML implementation gains. In this case, the second microgrid can now also provide power and computing resources to the first microgrid or other microgrids.

[0073] The training of an AI/ML model can often be the highest power requirements for the effective use of an Al model. Once trained the AI/ML model can be implemented and used in a repeatable fashion with significantly lower to minimal amounts of power and computing needs while providing an extremely complex an advanced control scheme based on the heavy frontend and processing that was required to train the model. This initial training of the model can be done at a local or aggregate scale on or off site.

[0074] In one or more scenarios and embodiments a microgrid may have a distributed and/or central processing capability. The distributed capability of processing may exist by having control systems and processing resources located at each of the different DERs and also at different elements within the microgrid. In some embodiments there may be a central controller as well or only a central controller in other cases. In all cases there is some computing capability present at the microgrid that could in some cases provide some of the computational needs for training an AI/ML model.

[0075] One of the major constraints for training an AI/ML model is often not only the processing needs required to do so which can be a function of both processing capability and time in which you have to process. For example if you have a long enough time span you can require less processing resources to achieve the training needed. Accordingly tradeoffs can be made as to the amount of computing power placed on site within the microgrid at the DERs at different elements etc. AI/ML models may also need to be routinely updated and retrained based on adaptations to the systems and other known changes that may affect the models.

[0076] Accordingly there is an obvious need to provide not only the computational power necessary to maintain a well trained modeling and Al portfolio but there is also a large power consumption need if the processing is to be done on site. Assuming the proper amount of processing resources provided within the microgrid in the form of a distributed computational system or a central controller or combination thereof then the next element that needs to be provided is the power necessary to run those processing capabilities.

[0077] In accordance with one or more embodiments, a host facility will have a demand curve of energy it requires from the microgrid. A perfectly efficient microgrid would be one that perfectly matches the demand curve provided by a host facility to the controller(s) of the microgrid. In practice, this is never the case for a number of design reasons and other system needs. For example, redundancy and resiliency sometimes require the system to be sized larger than the demand curve. In some examples, entire systems are installed that double the demand curse so that in the event the host facility’s other sources of power go down this backup microgrid system can be turned on to keep things running. Also some microgrids are sized up to provide the ability to actively rest or take certain assets offline for routine maintenance and cleaning. Also, when provided with a mix of power assets in a microgrid the different ramping up and down and power and load capabilities of each and the interplay between them can also create a need for further resiliency engineering.

[0078] Accordingly, there are naturally times when the microgrid is working below its max potential output, often well below. Some systems can simply shut off some of the DERs in this case but for some shutting them off actually amounts to just dumping and wasting the power. For example, for solar and wind and geothermal where the power sources continues to produce regardless, the best one can do is dump or open the circuit but the power is actually continuing to be generated to no one’s benefit. Some options to make use of this power is to store it for later. Another option includes selling it to a third party other than the host facility such as selling it back to the grid and/or large provider. Other than these, some of which aren’t always possible, the power can go unused.

[0079] Accordingly, in some cases, it is when this power is detected or predicted to be coming available, that such a AI/ML processing scheme can be implemented that makes use of this unused power. Accordingly, AI/ML processing can be selected based on the amount, location, and duration of power available in a DER, microgrid, or group of microgrids.

AI/ML Algorithm selection based on adaptive power control parameters

[0080] In one or more embodiments, an AI/ML algorithm can be selected based on the specific needs that the particular AI/ML module is situated to provide support for. For example, an AI/ML module that is placed in or near a sensor for a power storage element may require different control and monitoring that an AI/ML module that is tasked and situated near a microcontroller that is tasked with generating and disseminating broader level commands.

[0081] Accordingly, depending on the situation parameters, the location of the AI/ML module, and/or the surrounding elements and devices, different types of algorithms might be selected to better provide faster and more accurate training, control, and monitoring. Further, as power is often the main restrictive requirement to AI/ML processing, the power parameters: a) available amount of power, b) the location of the power, and c) the length of time that power can be provided can be the three power parameter elements used in some cases to determine what AI/ML processing, and further what algorithms, should be used. For example, over time the amount of power needed to accomplish training of the different algorithms can be determined and categorized generally for the different resources and processes. Based on this knowledge of power consumption metrics and the power available, in some cases, the selected of what algorithms to use can be determined based on the power parameters of the system. In some cases, a combination of the requested control accuracy of control needs of the system along with the power parameters can be looked at in conjunction to determined what algorithm to use. The unexpected results provided by this arrangement include component level customization and efficiencies that provide both computational savings as well as improved control of the system and components.

[0082] For example, in some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a regression algorithm from at least one from a group consisting of Ordinary Least Squares Regression (OLSR); Linear Regression; Logistic Regression; Stepwise Regression; Multivariate Adaptive Regression Splines (MARS); and/or Locally Estimated Scatterplot Smoothing (LOESS).

[0083] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include an instance-based algorithm from at least one from a group consisting of k-Nearest Neighbor (kNN); Learning Vector Quantization (LVQ); Self-Organizing Map (SOM); Locally Weighted Learning (LWL); and/or Support Vector Machines (SVM).

[0084] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a regularization algorithm from at least one from a group consisting of Ridge Regression; Least Absolute Shrinkage and Selection Operator (LASSO); Elastic Net; and/or Least- Angle Regression (LARS).

[0085] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a decision tree algorithm from at least one from a group consisting of: Classification and Regression Tree (CART); Iterative Dichotomiser 3 (ID3); C4.5 and C5.0 (different versions of a powerful approach); Chi-squared Automatic Interaction Detection (CHAID); Decision Stump; M5; and/or Conditional Decision Trees.

[0086] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a Bayesian algorithm from at least one from a group consisting of Naive Bayes; Gaussian Naive Bayes; Multinomial Naive Bayes; Averaged One-Dependence Estimators (AODE); Bayesian Belief Network (BBN); and/or Bayesian Network (BN).

[0087] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a clustering algorithm from at least one from a group consisting of k-Means; k-Medians; Expectation Maximisation (EM); and/or Hierarchical Clustering.

[0088] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include an association rule learning algorithm from at least one from a group consisting of Apriori algorithm; and/or Eclat algorithm.

[0089] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include an artificial neural network algorithm from at least one from a group consisting of: Perceptron; Multilayer Perceptrons (MLP); Back-Propagation; Stochastic Gradient Descent; Hopfield Network; and/or Radial Basis Function Network (RBFN).

[0090] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a deep learning algorithm from at least one from a group consisting of Convolutional Neural Network (CNN); Recurrent Neural Networks (RNNs); Long Short-Term Memory Networks (LSTMs); Stacked Auto-Encoders; Deep Boltzmann Machine (DBM); and/or Deep Belief Networks (DBN).

[0091] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include a dimensionality reduction algorithm from at least one from a group consisting of Principal Component Analysis (PCA); Principal Component Regression (PCR); Partial Least Squares Regression (PLSR); Sammon Mapping; Multidimensional Scaling (MDS); Projection Pursuit; Linear Discriminant Analysis (LDA); Mixture Discriminant Analysis (MDA); Quadratic Discriminant Analysis (QDA); and/or Flexible Discriminant Analysis (FDA).

[0092] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include an ensemble algorithm from at least one from a group consisting of Boosting; Bootstrapped Aggregation (Bagging); AdaBoost; Weighted Average (Blending); Stacked Generalization (Stacking);

Gradient Boosting Machines (GBM); Gradient Boosted Regression Trees (GBRT); and/or Random Forest.

[0093] In some cases, it can be determined based on at least the power parameters that an AI/ML module should include and make use of the AI/ML algorithms that include an algorithm from at least one from a group consisting of Feature selection algorithms; Algorithm accuracy evaluation; Performance measures; Optimization algorithms; Computational intelligence (evolutionary algorithms, etc.); Computer Vision (CV); Natural Language Processing (NLP); Recommender Systems; Reinforcement Learning; and/or Graphical Models.

Example Operations

[0094] FIG. 6 depicts a flowchart of example operations for applying an AI/ML module to a microgrid, according to some embodiments. Operations of a flowchart 600 of FIG. 6 can relate to any of the shown AI/ML modules from FIGs. 1 - 5. For example, flowchart 600 includes operations described as performed by an AI/ML Module 532, 531, 533, 535, 536, 537, and/or 538 for consistency with the earlier description. Such operations can be performed by hardware, firmware, software, or a combination thereof. However, assembly component naming, division, sub-section organization, program code naming, organization, and deployment can vary due to arbitrary operator choice, assembly ordering, programmer choice, programming language(s), platform, etc. Additionally, operations of the flowchart 600 are described in reference to the microgrids 100-500 of FIGs. 1-5.

[0095] FIG. 6 includes operations that apply in some implementations, with reference to FIGs. 1 - 5. Operations of the flowchart 600 begin at block 602. At block 602, an AI/ML module can be configured to receive local sensor data from local assets. The AI/ML module, at block 604, can further adjust and derive local training data from sensor data received locally to further customized localized Al-driven control of a local asset. Operation 606 can provide for the AI/ML module iteratively processing through a select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset. Further, the AI/ML module can, at block 608, create a control structure based on the trained Al model and the local asset control requirements. At operation block 610, the Al/module can apply the Al model based on the control structure thereby controlling the local asset. [0096] FIG. 7 depicts a flowchart of example operations for applying an AI/ML module to a microgrid, according to some embodiments. Operations of a flowchart 700 of FIG. 7 can relate to any of the shown AI/ML modules from FIGs. 1 - 5. For example, flowchart 700 includes operations described as performed by an AI/ML Module 532, 531, 533, 535, 536, 537, and/or 538 for consistency with the earlier description. Such operations can be performed by hardware, firmware, software, or a combination thereof. However, assembly component naming, division, sub-section organization, program code naming, organization, and deployment can vary due to arbitrary operator choice, assembly ordering, programmer choice, programming language(s), platform, etc. Additionally, operations of the flowchart 700 are described in reference to the microgrids 100-500 of FIGs. 1-5.

[0097] FIG. 7 includes operations that apply in some implementations, with reference to FIGs. 1 - 5. Operations of the flowchart 700 begin at block 702. At block 702, an AI/ML module can be configured to receive sensor data from local assets and other sensors elsewhere in the microgrid and host site. The AI/ML module, at block 704, can further adjust and derive local training data from sensor data received to further customized localized Al-driven control of local assets. Operation 706 can provide for the AI/ML module iteratively processing through select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset. Further, the AI/ML module can, at block 708 creates a control structure based on the local asset control requirements, the microgrid control requirements, and/or the host site control requirements. At operation block 710, the Al/module can apply the Al model based on the control structure thereby controlling the local asset.

[0098] FIG. 8 depicts a flowchart of operations that a microgrid as shown in any of FIGs. 1 - 5 may implement. Specifically, in accordance with one or more implementations, flowchart 800 includes operation 802, which provides for a microgrid that can generate electric energy for a host facility using first and second generating systems. At block 804, the microgrid detects properties of the first and second generating systems using a first and second sensor disposed within sensor range of the first and second generating systems respectively. At block 806, the microgrid can implement operations to generate corresponding sensor signaling corresponding to the detected properties using the first and second sensors. At block 807, the microgrid can implement operations to receive, at an artificial intelligence (Al) / Machine Learning (ML) module, the sensor signaling directly from at least one of the first and second sensors. Further, the microgrid can, at block 801, provide operations to train, using the AI/ML module, a control model using one or more AI/ML algorithms, and the received sensor signaling. The microgrid can also provide operations, at block 812, to generate control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

[0099] FIG. 10 depicts a flowchart of example operations to train and use an AI/ML module, according to some embodiments. The microgrid or a component of the microgrid can perform the operations. For example, at block/operation 1002, a microgrid or component can generate electric energy using a first and second generating systems, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing. At block/operation 1010, a microgrid or component can train, using an artificial intelligence (Al) / Machine Learning (ML) module, a control model using one or more AI/ML algorithms and the excess electric energy. Further, at block/operations 1012, a microgrid or component can generate control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

[0100] FIGS. 6, 7, 8, and 10 are annotated with a series of numbers. These numbers represent stages of operations. Although these stages are ordered for this example, the stages illustrate one example to aid in understanding this disclosure and should not be used to limit the claims. Subject matter falling within the scope of the claims can vary with respect to the order and some of the operations.

[0101] The flowcharts are provided to aid in understanding the illustrations and are not to be used to limit the scope of the claims. The flowcharts depict example operations that can vary within the scope of the claims. Additional operations may be performed; fewer operations may be performed; the operations may be performed in parallel, and the operations may be performed in a different order. For example, the operations depicted in some blocks can be performed in parallel or concurrently. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by program code. The program code may be provided to a processor of a general-purpose computer, special purpose computer, or other programmable machine or apparatus.

[0102] As will be appreciated, aspects of the disclosure may be embodied as a system, method, or program code/instructions stored in one or more machine-readable media. Accordingly, aspects may take the form of hardware, software (including firmware, resident software, micro- code, etc.), or a combination of software and hardware aspects that may all generally be referred to herein as a “circuit,” “module” or “system.” The functionality presented as individual modules/units in the example illustrations can be organized differently in accordance with any one of the platform (operating system and/or hardware), application ecosystem, interfaces, programmer preferences, programming language, administrator preferences, etc.

[0103] Any combination of one or more machine-readable medium(s) may be utilized. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable storage medium may be, for example, but not limited to, a system, apparatus, or device, that employs any one of or combination of electronic, magnetic, optical, electromagnetic, infrared, or semiconductor technology to store program code. More specific examples (a non-exhaustive list) of the machine-readable storage medium would include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a machine-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable storage medium is not a machine-readable signal medium.

[0104] A machine-readable signal medium may include a propagated data signal with machine- readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A machine-readable signal medium may be any machine-readable medium that is not a machine-readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0105] Program code embodied on a machine -readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0106] Computer program code for carrying out operations for aspects of the disclosure may be written in any combination of one or more programming languages, including an object-oriented programming language such as the Java® programming language, C++ or the like; a dynamic programming language such as Python; a scripting language such as Perl programming language or PowerShell script language; and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on a stand-alone machine, may execute in a distributed manner across multiple machines, and may execute on one machine while providing results and or accepting input on another machine.

[0107] The program code/instructions may also be stored in a machine-readable medium that can direct a machine to function in a particular manner, such that the instructions stored in the machine-readable medium produce an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

Example Computer

[0108] FIG. 9 depicts an example computer, according to some embodiments. A computer 900 of FIG. 9 can be representative of a computer or controller in the microgrid 100-500 of FIGs. 1- 5. For example, the computer 900 can be an example computer in the microgrid controller 112, AI/ML module 230, and/or sensor 518 (as described above). The computer 900 can also an example computer used to run normal microgrid operations (as described above). For example, the computer 900 can be an example computer positioned elsewhere in the microgrid or host facility.

[0109] The computer 900 includes a processor 901 (possibly including multiple processors, multiple cores, multiple nodes, and/or implementing multi-threading, etc.). The computer 900 includes a memory 907. The memory 907 may be system memory or any one or more of the above already described possible realizations of machine-readable media. The computer 900 also includes a bus 903 and a network interface 905.

[0110] The computer 900 also includes a controller 911 and a signal processor 912. The controller 911 and the signal processor 912 can be hardware, software, firmware, or a combination thereof. For example, the controller 911 and the signal processor 912 can be software executing on the processor 901. Any one of the previously described functionalities may be partially (or entirely) implemented in hardware and/or on the processor 901. For example, the functionality may be implemented with an application-specific integrated circuit, in logic implemented in the processor 901, in a co-processor on a peripheral device or card, etc. Further, realizations may include fewer or additional components not illustrated in FIG. 9 (e.g., video cards, audio cards, additional network interfaces, peripheral devices, etc.). The processor 901 and the network interface 905 are coupled to the bus 903. Although illustrated as being coupled to the bus 903, the memory 907 may be coupled to the processor 901.

[0111] While the aspects of the disclosure are described with reference to various implementations and exploitations, it will be understood that these aspects are illustrative and that the scope of the claims is not limited to them. In general, techniques for microgrid control based on AI/ML modules as described herein may be implemented with facilities consistent with any hardware system or hardware systems. Many variations, modifications, additions, and improvements are possible.

[0112] Plural instances may be provided for components, operations, or structures described herein as a single instance. Finally, boundaries between various components, operations, and data stores are somewhat arbitrary, and particular operations are illustrated in the context of specific illustrative configurations. Other allocations of functionality are envisioned and may fall within the scope of the disclosure. In general, structures and functionality presented as separate components in the example configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements may fall within the scope of the disclosure.

[0113] Use of the phrase “at least one of’ preceding a list with the conjunction “and” should not be treated as an exclusive list and should not be construed as a list of categories with one item from each category, unless specifically stated otherwise. A clause that recites “at least one of A, B, and C” can be infringed with only one of the listed items, multiple of the listed items, and one or more of the items in the list, and another item not listed.

Example Embodiments

[0114] Embodiment 1: A microgrid system comprising: a first and second generating systems configured to generate electric energy, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; a first and second sensor disposed within sensor range of the first and second generating systems respectively, and configured to detect properties of the first and second generating systems and generate corresponding sensor signaling corresponding to the detected properties; an artificial intelligence (Al) / Machine Learning (ML) module configured to receive the sensor signaling directly from at least one of the first and second sensor, and further configured to train a control model using one or more AI/ML algorithms and the received sensor signalingexcess electric energy; and a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module and configured to generate control signals for both the first and second generating systems based on at least the control model

[0115] Embodiment 2: The microgrid system of Embodiment 1, wherein the microgrid controller comprises: a plurality of microgrid controllers, wherein at least one localized microgrid controller from the plurality of microgrid controllers is disposed within, on, or near at least one of the first and second generating systems.

[0116] Embodiment 3: The microgrid system of Embodiment 2, wherein at least one from the plurality of microgrid controllers is configured as a master controller to centralize communication and control of the other localized microgrid controllers of the plurality of microgrid controllers.

[0117] Embodiment 4: The microgrid system of Embodiment 2, wherein the AI/ML module comprises a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of the microgrid controllers in the plurality of microgrid controllers.

[0118] Embodiment 5: The microgrid system of Embodiment 1, wherein the AI/ML module comprises a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near the microgrid controller.

[0119] Embodiment 6: The microgrid system of Embodiment 1, wherein the AI/ML module comprises a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of the first and second generating systems.

[0120] Embodiment 7: The microgrid system of Embodiment 1, wherein the AI/ML module comprises a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of the first and second sensors.

[0121] Embodiment 8: The microgrid system of Embodiment 1, wherein the AI/ML module comprises a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near the host facility. [0122] Embodiment 9: The microgrid system of Embodiment 1, wherein at least one of the first and second generating systems further comprising: an energy storage system configured to receive power from at least one of the first and second generating systems, and further configured to dispatch power to the host facility.

[0123] Embodiment 10: The microgrid system of Embodiment 9, wherein the microgrid controller comprises a plurality of microgrid controllers, wherein at least one from the plurality of microgrid controllers is disposed within, on, or near the energy storage system.

[0124] Embodiment 11: The microgrid system of Embodiment 9, wherein the energy storage system is at least one from a group consisting of one or more of a battery, a gravity displacement system, a chemical conversion storage system, a compressed gas storage system, and/or a thermal energy storage system.

[0125] Embodiment 12: The microgrid system of Embodiment 1, wherein the first and second generating systems are at least one from a group consisting of a solar array, a wind turbine, a fuel cell, a cogeneration unit, a hydrogen generator, a natural gas generator, a gas turbine, a hydrogen turbine, a geothermal generator, a biogas generator, and/or a battery.

[0126] Embodiment 13: The microgrid system of Embodiment 1, wherein the first and second sensor are at least one from a group consisting of a temperature sensor, a voltage sensor, a current sensor, a frequency sensor, an acoustic sensor, a light sensor, a vibration sensor, a flow sensor, a liquid sensor, and/or a gas sensor.

[0127] Embodiment 14: The microgrid system of Embodiment 1, wherein the detected properties include at least one from a group consisting of a temperature, a voltage, a current, a frequency, an acoustic patter, a light pattern, a vibration, a flow, a liquid type, and/or a gas type.

[0128] Embodiment 15: The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, a regression algorithm from at least one from a group consisting of Ordinary Least Squares Regression (OLSR); Linear Regression; Logistic Regression; Stepwise Regression; Multivariate Adaptive Regression Splines (MARS); and/or Locally Estimated Scatterplot Smoothing (LOESS).

[0129] Embodiment 16: The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, an instance-based algorithm from at least one from a group consisting of k-Nearest Neighbor (kNN); Learning Vector Quantization (LVQ); Self-Organizing Map (SOM); Locally Weighted Learning (LWL); and/or Support Vector Machines (SVM).

[0130] Embodiment 17: The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, a regularization algorithm from at least one from a group consisting of Ridge Regression; Least Absolute Shrinkage and Selection Operator (LASSO); Elastic Net; and/or Least-Angle Regression (LARS).

[0131] Embodiment 18: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, a decision tree algorithm from at least one from a group consisting of: Classification and Regression Tree (CART); Iterative Dichotomiser 3 (ID3); C4.5 and C5.0 (different versions of a powerful approach); Chi-squared Automatic Interaction Detection (CHAID); Decision Stump; M5; and/or Conditional Decision Trees.

[0132] Embodiment 19: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, a Bayesian algorithm from at least one from a group consisting of Naive Bayes; Gaussian Naive Bayes; Multinomial Naive Bayes; Averaged One- Dependence Estimators (AODE); Bayesian Belief Network (BBN); and/or Bayesian Network (BN).

[0133] Embodiment 20: The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, a clustering algorithm from at least one from a group consisting of k-Means; k-Medians; Expectation Maximisation (EM); and/or Hierarchical Clustering.

[0134] Embodiment 21 : The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, an association rule learning algorithm from at least one from a group consisting of Apriori algorithm; and/or Eclat algorithm.

[0135] Embodiment 22: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, an artificial neural network algorithm from at least one from a group consisting of: Perceptron; Multilayer Perceptrons (MLP); Back-Propagation;

Stochastic Gradient Descent; Hopfield Network; and/or Radial Basis Function Network (RBFN).

[0136] Embodiment 23: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, a deep learning algorithm from at least one from a group consisting of Convolutional Neural Network (CNN); Recurrent Neural Networks (RNNs); Long Short-Term Memory Networks (LSTMs); Stacked Auto-Encoders; Deep Boltzmann Machine (DBM); and/or Deep Belief Networks (DBN).

[0137] Embodiment 24: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, a dimensionality reduction algorithm from at least one from a group consisting of Principal Component Analysis (PCA); Principal Component Regression (PCR); Partial Least Squares Regression (PLSR); Sammon Mapping; Multidimensional Scaling (MDS); Projection Pursuit; Linear Discriminant Analysis (LDA); Mixture Discriminant Analysis (MDA); Quadratic Discriminant Analysis (QDA); and/or Flexible Discriminant Analysis (FDA).

[0138] Embodiment 25: The microgrid of Embodiment 1 wherein the AI/ML algorithms includes, based on at least the power parameters, an ensemble algorithm from at least one from a group consisting of: Boosting; Bootstrapped Aggregation (Bagging); AdaBoost; Weighted Average (Blending); Stacked Generalization (Stacking); Gradient Boosting Machines (GBM); Gradient Boosted Regression Trees (GBRT); and/or Random Forest.

[0139] Embodiment 26: The microgrid of Embodiment 1 wherein the AI/ML algorithms include, based on at least the power parameters, an algorithm from at least one from a group consisting of Feature selection algorithms; Algorithm accuracy evaluation; Performance measures;

Optimization algorithms; Computational intelligence (evolutionary algorithms, etc.); Computer Vision (CV); Natural Language Processing (NLP); Recommender Systems; Reinforcement Learning; and/or Graphical Models.

[0140] Embodiment 27: A method comprising: generating electric energy using a first and second generating systems wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; training, using the AI/ML module, a control model using one or more AI/ML algorithms and the excess electric energy; and generating control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

[0141] Embodiment 28: A method comprising: receiving local sensor data from the local asset; adjusting and Deriving local training data from sensor data received locally to further customized localized Al-driven control of local asset; iteratively processing through select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset; creating a control structure based on the trained Al model and the local asset control requirements; and applying the Al model based on the control structure thereby controlling the local asset.

[0142] Embodiment 29: A method comprising: receiving sensor data from local asset and other sensors elsewhere in the microgrid and host site; adjusting and Deriving local training data from sensor data received to further customized localized Al-driven control of local asset; iteratively processing through select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset; creating a control structure based on the local asset control requirements, the microgrid control requirements, and/or the host site control requirements; and applying the Al model based on the control structure thereby controlling the local asset.

[0143] Embodiment 30: One or more non-transitory machine-readable media comprising program code executable by a processor to cause the processor to: generate electric energy using a first and second generating systems wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; train, using the AI/ML module, a control model using one or more AI/ML algorithms and the excess electric energy; and generate control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

[0144] Embodiment 31 One or more non-transitory machine-readable media comprising program code executable by a processor to cause the processor to receive local sensor data from the local asset; adjust and Deriving local training data from sensor data received locally to further customized localized Al-driven control of local asset; iteratively process through select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset; create a control structure based on the trained Al model and the local asset control requirements, and apply the Al model based on the control structure thereby controlling the local asset.

[0145] Embodiment 32: One or more non-transitory machine-readable media comprising program code executable by a processor to cause the processor to receive sensor data from local asset and other sensors elsewhere in the microgrid and host site; adjust and Deriving local training data from sensor data received to further customized localized Al-driven control of local asset; iteratively process through select Al algorithm using the training data until an Al model is determined to have been trained for the custom use and control of the local asset; creating a control structure based on the local asset control requirements, the microgrid control requirements, and/or the host site control requirements; and applying the Al model based on the control structure thereby controlling the local asset. [0146] While some embodiments of the present invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the invention. It should be understood that different aspects of the invention can be appreciated individually, collectively, or in combination with each other. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in practicing the invention. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

WHAT IS CLAIMED IS:

1. A microgrid system comprising: a first and second generating systems configured to generate electric energy, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; an artificial intelligence (Al) / Machine Learning (ML) module configured to train a control model using one or more AI/ML algorithms and the excess electric energy; and a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module and configured to generate control signals for both the first and second generating systems based on at least the control model.

2. The microgrid system of claim 1, wherein the microgrid controller comprises: a plurality of microgrid controllers, wherein at least one localized microgrid controller from the plurality of microgrid controllers is disposed within, on, or near at least one of the first and second generating systems.

3. The microgrid system of claim 2, wherein at least one from the plurality of microgrid controllers is configured as a master controller to centralize communication and control of the other localized microgrid controllers of the plurality of microgrid controllers.

4. The microgrid system of claim 2, wherein the AI/ML module comprises: a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of the microgrid controllers in the plurality of microgrid controllers.

5. The microgrid system of claim 1, wherein the AI/ML module comprises: a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near the microgrid controller.

6. The microgrid system of claim 1, wherein the AI/ML module comprises: a plurality of AI/ML modules,

32 wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of the first and second generating systems.

7. The microgrid system of claim 1, wherein the AI/ML module comprises: a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near at least one of a first and second sensors.

8. The microgrid system of claim 1, wherein the AI/ML module comprises: a plurality of AI/ML modules, wherein at least one from the plurality of AI/ML modules is disposed within, on, or near the host facility.

9. The microgrid system of claim 1, wherein at least one of the first and second generating systems further comprising: an energy storage system configured to receive power from at least one of the first and second generating systems, and further configured to dispatch power to the host facility.

10. The microgrid system of claim 9, wherein the microgrid controller comprises: a plurality of microgrid controllers, wherein at least one from the plurality of microgrid controllers is disposed within, on, or near the energy storage system.

11. The microgrid system of claim 9, wherein the energy storage system is at least one from a group consisting of one or more of a battery, a gravity displacement system, a chemical conversion storage system, a compressed gas storage system, and/or a thermal energy storage system.

12. The microgrid system of claim 1, wherein the first and second generating systems are at least one from a group consisting of a solar array, a wind turbine, a fuel cell, a cogeneration unit, a hydrogen generator, a natural gas generator, a gas turbine, a hydrogen turbine, a geothermal generator, a biogas generator, and/or a battery.

13. The microgrid system of claim 7, wherein the first and second sensor are at least one from a group consisting of a temperature sensor, a voltage sensor, a current sensor, a frequency

33 sensor, an acoustic sensor, a light sensor, a vibration sensor, a flow sensor, a liquid sensor, and/or a gas sensor.

14. The microgrid system of claim 1, wherein the detected properties include at least one from a group consisting of a temperature, a voltage, a current, a frequency, an acoustic patter, a light pattern, a vibration, a flow, a liquid type, and/or a gas type.

15. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a regression algorithm from at least one from a group consisting of:

Ordinary Least Squares Regression (OLSR);

Linear Regression;

Logistic Regression;

Stepwise Regression;

Multivariate Adaptive Regression Splines (MARS); and/or Locally Estimated Scatterplot Smoothing (LOESS).

16. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, an instance-based algorithm from at least one from a group consisting of: k-Nearest Neighbor (kNN);

Learning Vector Quantization (LVQ);

Self-Organizing Map (SOM);

Locally Weighted Learning (LWL); and/or Support Vector Machines (SVM).

17. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a regularization algorithm from at least one from a group consisting of:

Ridge Regression;

Least Absolute Shrinkage and Selection Operator (LASSO);

Elastic Net; and/or

Least-Angle Regression (LARS).

18. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a decision tree algorithm from at least one from a group consisting of:

Classification and Regression Tree (CART);

Iterative Dichotomiser 3 (ID3);

C4.5 and C5.0 (different versions of a powerful approach); Chi-squared Automatic Interaction Detection (CHAID);

Decision Stump;

M5; and/or

Conditional Decision Trees.

19. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a Bayesian algorithm from at least one from a group consisting of:

Naive Bayes;

Gaussian Naive Bayes;

Multinomial Naive Bayes;

Averaged One -Dependence Estimators (AODE);

Bayesian Belief Network (BBN); and/or

Bayesian Network (BN).

20. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a clustering algorithm from at least one from a group consisting of: k-Means; k-Medians;

Expectation Maximisation (EM); and/or

Hierarchical Clustering.

21. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, an association rule learning algorithm from at least one from a group consisting of:

Apriori algorithm; and/or

Eclat algorithm.

22. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, an artificial neural network algorithm from at least one from a group consisting of:

Perceptron;

Multilayer Perceptrons (MLP);

Back-Propagation;

Stochastic Gradient Descent;

Hopfield Network; and/or Radial Basis Function Network (RBFN).

23. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a deep learning algorithm from at least one from a group consisting of:

Convolutional Neural Network (CNN);

Recurrent Neural Networks (RNNs);

Long Short-Term Memory Networks (LSTMs);

Stacked Auto-Encoders;

Deep Boltzmann Machine (DBM); and/or

Deep Belief Networks (DBN).

24. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, a dimensionality reduction algorithm from at least one from a group consisting of:

Principal Component Analysis (PCA);

Principal Component Regression (PCR);

Partial Least Squares Regression (PLSR);

Sammon Mapping;

Multidimensional Scaling (MDS);

Projection Pursuit;

Linear Discriminant Analysis (LDA);

Mixture Discriminant Analysis (MDA) ;

Quadratic Discriminant Analysis (QDA); and/or Flexible Discriminant Analysis (FDA).

25. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, an ensemble algorithm from at least one from a group consisting of:

Boosting;

Bootstrapped Aggregation (Bagging);

AdaBoost;

Weighted Average (Blending);

Stacked Generalization (Stacking);

Gradient Boosting Machines (GBM);

Gradient Boosted Regression Trees (GBRT); and/or

Random Forest.

36

26. The microgrid of claim 1 wherein the AI/ML algorithms include, based on at least the power parameters, an algorithm from at least one from a group consisting of:

Feature selection algorithms;

Algorithm accuracy evaluation;

Performance measures;

Optimization algorithms;

Computational intelligence (evolutionary algorithms, etc.);

Computer Vision (CV);

Natural Language Processing (NLP);

Recommender Systems;

Reinforcement Learning; and/or Graphical Models.

27. A method comprising: generating electric energy using a first and second generating systems, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; training, using an artificial intelligence (Al) / Machine Learning (ML) module, a control model using one or more AI/ML algorithms and the excess electric energy; and generating control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

28. One or more non-transitory machine-readable media comprising program code executable by a processor to cause the processor to: generate electric energy for a host facility using a first and second generating systems, wherein the electric energy includes excess electric energy based on power parameters that include power amount, power duration, and/or power timing; detect properties of the first and second generating systems using a first and second sensor disposed within sensor range of the first and second generating systems respectively; generate corresponding sensor signaling corresponding to the detected properties using the first and second sensors;

37 receive, at an artificial intelligence (Al) / Machine Learning (ML) module, the sensor signaling directly from at least one of the first and second sensor; train, using an artificial intelligence (Al) / Machine Learning (ML) module, a control model using one or more AI/ML algorithms and the excess electric energy; and generate control signals for both the first and second generating systems based on at least the control model using a microgrid controller communicatively connected to the first and second generating systems and the AI/ML module.

38