WO2009039500A1 - Procédé et appareil de détermination des économies d'énergie réalisées en employant un modèle de consommation d'énergie de référence qui comprend un algorithme d'intelligence artificielle - Google Patents

Procédé et appareil de détermination des économies d'énergie réalisées en employant un modèle de consommation d'énergie de référence qui comprend un algorithme d'intelligence artificielle Download PDF

Info

Publication number
WO2009039500A1
WO2009039500A1 PCT/US2008/077232 US2008077232W WO2009039500A1 WO 2009039500 A1 WO2009039500 A1 WO 2009039500A1 US 2008077232 W US2008077232 W US 2008077232W WO 2009039500 A1 WO2009039500 A1 WO 2009039500A1
Authority
WO
WIPO (PCT)
Prior art keywords
energy
facility
baseline
condition data
time interval
Prior art date
Application number
PCT/US2008/077232
Other languages
English (en)
Inventor
Kenneth B. Barclay
Timothy J. Mattison
Melvin A. Jones
Paul Macgregor
Original Assignee
Sterling Planet, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sterling Planet, Inc. filed Critical Sterling Planet, Inc.
Publication of WO2009039500A1 publication Critical patent/WO2009039500A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/04—Billing or invoicing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Definitions

  • the invention relates to determining energy savings and, more particularly, to determining energy savings by using an artificial intelligence-based energy use model to calculate a building's baseline energy use, and for determining energy savings from the baseline energy use calculation.
  • degree-day models may be implemented off-site with historical data consisting of only monthly degree-days and energy bills and utilizing statistical regression models, these models have proven to be fairly inaccurate.
  • facility models have proven to be very accurate, but these models, such as DOEII, are very complex and require an extensive on-site evaluation of building design parameters, such as, for example, window coverage, directional orientation, insulation, and equipment, such as chillers, boilers, HVAC systems, lighting and motors.
  • building design parameters such as, for example, window coverage, directional orientation, insulation, and equipment, such as chillers, boilers, HVAC systems, lighting and motors.
  • these models have been proven to be impractical in terms of time and cost for use with a portfolio of buildings, especially dispersed across a large geographic region.
  • traditional performance contracts and new tradable conservation attribute markets have been difficult to implement.
  • EEC Energy Efficiency Credit
  • Energy Savings Certificate represents the value of energy not used at a building through the implementation of energy efficiency and conservation projects.
  • EEC Energy Efficiency Credit
  • Energy Savings Certificate represents the value of energy not used at a building through the implementation of energy efficiency and conservation projects.
  • Several U.S. states have passed legislation specifying that tradable EECs may be used to meet mandates for reducing energy generated in their state.
  • the electricity suppliers may purchase EECs equivalent to a percentage of their total annual retail sales, such as 4% by 2010 in the state of Connecticut.
  • GFG Greenhouse Gas
  • an EEC Since an EEC has the environmental attributes of avoided air emissions including SO 2 , NOx and CO 2 associated with it in accordance with the location of the energy reduction, an EEC may be purchased to reduce indirect CO 2 emissions.
  • EECs In the case of the former, states with mandates, EECs are certified by the states, usually under the direction of the public utility commissions. In the case of the latter, voluntary transactions, EECs are certified by non-profit certification organizations such as Environmental Resources Trust, Inc. (ERT). In either case, the key issue for certification is Measurement and Verification (M&V) of the energy savings derived from the energy efficiency or conservation project.
  • M&V Measurement and Verification
  • the present invention relates to determining energy cost savings in an energy-consuming facility, such as a commercial building or group of such buildings, using an artificial intelligence model, for example a neural network model, that projects or estimates the amount of energy that would have been consumed by the facility but for the implementation of energy efficiency or conservation measures.
  • Energy savings are represented by the difference between the estimate of energy that would have been consumed but for the measures and the actual amount of energy consumed by the facility under actual conditions during a time interval after the measures have been implemented.
  • baseline facility condition data is input to an artificial intelligence model generator, for example a neural network model generator.
  • the baseline facility condition data represents baseline conditions experienced by the facility during a first time interval before energy conservation measures.
  • the baseline facility conditions include at least weather conditions experienced by the facility.
  • the baseline facility conditions can further include facility occupancy data, representing the extent to which the facility is fully or partially occupied, and production or manufacturing data, representing the extent to which the facility is fully or partially engaged in its normal operations.
  • Baseline energy consumption data is also input to the neural network model generator.
  • the baseline energy consumption data represents the amount of energy consumed by the facility during the first time interval.
  • such baseline facility condition data and corresponding baseline energy consumption data can be input for a plurality of such time intervals, such as on a per- month basis.
  • baseline facility condition data and corresponding baseline energy consumption can be input for each of 36 months.
  • the neural network model generator In response to the baseline facility condition data and corresponding baseline energy consumption data, the neural network model generator generates a neural network model.
  • the model is a neural network that represents or models how facility energy consumption responds to facility conditions.
  • the model is used to predict or estimate the amount of energy that would have been consumed by the facility but for the implementation of energy efficiency or conservation measures.
  • Actual facility condition data representing actual facility conditions during a second time interval after the energy conservation measures have been implemented, is input to the model.
  • the actual facility condition data can be of the same types as described above with regard to the baseline facility condition data.
  • the baseline facility condition data consists of weather data
  • the actual facility condition data can correspondingly consist of weather data.
  • the neural network model was generated based upon the baseline facility condition data and baseline energy consumption, then in response to the actual facility condition data the neural network model outputs an estimate of the amount of energy that would have been consumed during the second time interval (under the actual facility conditions) but for the energy conservation measures.
  • Energy savings can then be computed. Energy savings can be defined by the difference between the actual energy consumed during the second time interval and the estimate of energy that would have been consumed during the second time interval but for the energy conservation measures.
  • One embodiment includes data-expansion pre-processing that includes combining raw energy-consumption datums for individual time periods into a greater number of total energy-consumption datums. In this way, more training data is input, which results in a better-trained, more accurate neural network model.
  • Another embodiment includes model-updating post-processing that includes repeating the energy-savings-determining method with updated facility condition data. This permits energy savings to be recognized from additional energy-saving measures and at the same time more accurately generates EECs even when overall facility energy use increases due to expanded operations and usage creep.
  • Still another embodiment includes a system-based method for determining EECs that is similar to the facility- based method but instead is based on individual energy-using systems. With this method, small energy savings in a large facility can be more readily and accurately projected.
  • FIG. 1 is a block diagram of an exemplary computing system for determining energy cost savings using a neural network-based model.
  • FIG. 2 is a flow diagram, illustrating an exemplary computer-implemented method for determining energy cost savings using a neural network-based model.
  • Fig. 3 illustrates exemplary sinusoidal functions representing percentage of hours above and percentage of hours below the saturation temperature.
  • Fig. 4 is an exemplary table summarizing the baseline data that forms the input data for the neural network model generator.
  • Fig. 5 is an exemplary table summarizing neural network parameters and their selections in the exemplary embodiment.
  • Fig. 6 depicts an exemplary screen display of baseline facility condition data and baseline energy data.
  • Fig. 7 illustrates an exemplary database table structure.
  • Fig. 8 is a continuation sheet of Fig. 7.
  • Fig. 9 is an exemplary facility relationships diagram.
  • Fig. 10 is an exemplary vendor relationships diagram.
  • Fig. 11 is an exemplary forecast relationships diagram.
  • Fig. 12 is an exemplary table of baseline facility condition data and baseline energy consumption data.
  • Fig. 13 is an exemplary table of output data produced by the model.
  • Fig. 14 depicts an exemplary screen display of baseline energy consumption, actual energy consumption, and projected electrical energy savings.
  • Fig. 15 depicts an exemplary screen display of a "dashboard" or summary page, showing energy savings and related information.
  • Fig. 16 is a table illustrating four raw datums expanded to ten total datums, which include the raw datums and new combined datums generated from the raw datums by a data-expanding pre-process.
  • Fig. 17 graphically illustrates how a facility's actual energy usage creeps up over time due to expanded operations, while the projected energy usage based on outdated facility condition data does not.
  • Fig. 18 graphically illustrates the projected energy usage of the facility of
  • FIG. 17 increasing, as determined by a post-process that generates an updated energy use model based on updated facility condition data.
  • Fig. 19 is a flow diagram, illustrating an exemplary computer-implemented method for determining energy cost savings of an individual energy-consuming system using a neural network-based model.
  • an exemplary computing system 10 can be used to determine energy cost savings in an energy-consuming facility.
  • the term "facility" as used herein refers to any group of one or more commercial or residential buildings or other operations that consume energy (for example, for heating and cooling the space).
  • the invention can be used by, for example, owners or managers of the facility and related entities to assess whether energy-conserving or efficiency-enhancing measures that have been implemented are resulting in energy savings.
  • system, method and computer program product are described in relation to a standalone computing system 10 for purposes of illustration, in alternative embodiments they can relate to a World Wide Web-based arrangement in which the user operates a client computer that is located remotely from a server computer. In such embodiments, the combination of client and server computers defines a computing system similar to computing system 10.
  • Computing system 10 can comprise a general-purpose personal computer such as a desktop, laptop or handheld computer.
  • a computing system 10 includes a programmed processor system 12, a display 14, a keyboard 16, mouse 18 or similar pointing device, network interface 20, fixed-medium data storage device 22 such as a magnetic disk drive, and a removable-medium data storage device 24 such as a CD-ROM or DVD drive.
  • programmed processor system 12 includes a conventional arrangement of one or more processors, memories and other logic that together define the overall computational and data manipulation power of computing system 10.
  • 10 comprises a personal computer or similar general-purpose computer, in other embodiments it can comprise any other suitable system.
  • portions of such a computing system can be distributed among a number of networked computers, data storage devices, network devices, and other computing system elements.
  • software elements described below, can be stored in a distributed manner and retrieved via network interface 20 from multiple sources on an as-needed basis. Similarly, they can be stored on multiple disks or other data storage media and retrieved or otherwise loaded into computing system 10 on an as-needed basis.
  • the software elements or portions thereof can be retrieved on an as-needed basis from storage devices 22 or 24 or from a remote computer or storage device (not shown) via network interface 20.
  • the functions of software elements 26, 28, 30, 32, 34 and 36 can be distributed over a greater number of elements or, alternatively, combined or condensed into fewer elements.
  • Additional software elements commonly included in computing systems, such as an operating system (e.g., MICROSOFT WINDOWS), utilities, device drivers, etc., are included but not shown for purposes of clarity.
  • an operating system e.g., MICROSOFT WINDOWS
  • utilities e.g., device drivers, etc.
  • Model generator 28 and engine 30 can be portions or components of a commercially available software tool 38 for predicting outcomes using a neural network model.
  • a commercially available software tool 38 for predicting outcomes using a neural network model.
  • One such software tool that has been found to be suitable is NEURALWORKS PREDICT, which is produced by Neuralware of Carnegie, Pennsylvania. It should be note that the invention is not limited to using any particular software, and that persons skilled in the art to which the invention relates will readily be capable of providing suitable neural network software in view of the teachings herein.
  • a neural network is a non-linear estimation technique that replicates the function on neurons in the human brain through a collection of interconnected mathematical functions with dynamic weighting of connections enabling continuous "learning”.
  • Neural networks form these interconnected mathematical functions from the input pattern, not the input data, and apply continuously changing weights in response to the level of correlation.
  • neural network models are able to extract the essential characteristics from numerical data as opposed to memorizing all of the data. This reduces the amount of data needed and forms an implicit model without having to form a complex physical model of the underlying phenomenon such as in the case of a building.
  • the NEURALWORKS PREDICT package is specifically directed to the use of a neural network to predict outcomes for any of a wide range of problems. PREDICT can be used by software developers who have no expert knowledge of neural networks. With only minimal user involvement, PREDICT addresses the issues associated with building models from empirical data.
  • PREDICT analyzes input data to identify appropriate transforms, partitions the input data into training and test sets, selects relevant input variables, and then constructs, trains, and optimizes a neural network tailored to the problem.
  • the programmed processor system 12 is adapted to include a different artificial intelligence software tool.
  • the artificial intelligence software tool may include an adaptive system other than a neural network software tool.
  • Suitable artificial intelligence software may include artificial intelligence using relational database management techniques, web- enabled data capturing, visual monitoring, statistical reporting, and remote monitoring software tools.
  • Such artificial intelligence software's capabilities may include but are not limited to artificial, non-linear, statistical data-modeling tools, pattern matching and learning capabilities, recognizing locations of facilities, weather data, building usage, and statistical correlation. Persons of ordinary skill in the art will be able to readily select and configure such artificial intelligence software for use in projecting energy usage based on inputted actual facility condition data and based on certain energy- conserving or energy efficiency-enhancing measures having not been implemented.
  • the exemplary programmed processor system 12 described herein includes the neural network software tool 38.
  • GUI graphical user interface
  • the user can interact with computing system 10 through user interface 26 in a conventional manner.
  • User interface 26 can comprise, for example, a graphical user interface (GUI) that operates in accordance with standard windowing and graphical user interface protocols supported by MICROSOFT WINDOWS or similar operating system. That is, the user can manipulate (e.g., open, close, resize, minimize, etc.) windows on display 14, launch application software that executes within one or more windows, and interact with pictures, icons and graphical control structures (e.g., buttons, checkboxes, pull-down menus, etc.) on display 14 using mouse 18, keyboard 16 or other input devices. What is displayed within a window under control of an application program is generally referred to herein as a screen or screen display of the application program.
  • GUI graphical user interface
  • User interface 26 can include not only the logic through which screen displays are generated and made viewable but also computational logic that generates and organizes, tabulates, etc., numerical values to be displayed or otherwise output. Similarly, user interface 26 can include logic for importing, exporting, opening and closing data files.
  • a method for determining energy savings in an energy-consuming facility is illustrated by the steps shown in Fig. 2.
  • the facility can be, for example, a commercial or residential building or group of such buildings.
  • a facility is described that is involved in manufacturing, the facility can be involved in any sort of operations in which it is desirable to conserve energy or maximize energy use efficiency.
  • essentially all facilities that purchase electricity from utility companies for purposes such as heating, cooling and illuminating the facility desire to conserve energy or maximize efficiency.
  • the method is primarily effected through the operation of programmed processor system 12 (Fig. 1) operating under control of an application program (software).
  • the application program can thus comprise some or all of the software elements shown in Fig.
  • computing system 10 can be provided to computing system 10 via a network 40, such as the Internet, or via one or more removable disks 42, such as CD-ROMs, DVDs, etc.).
  • a network 40 such as the Internet
  • removable disks 42 such as CD-ROMs, DVDs, etc.
  • application program or other such software stored or otherwise carried on such media constitutes a "computer program product.”
  • the method begins when the user causes the application program to begin executing.
  • user interface 26 can generate a screen display with a main menu of options that allows a user to navigate to any selected step, such that the method begins or continues at that step.
  • steps can be performed in any other suitable order.
  • additional steps can be included. Steps along the lines of those shown in Fig. 2 can be combined with other such steps to define a method having a smaller number of steps and, conversely, the steps shown in Fig. 2 can be separated into a greater number of steps.
  • baseline facility condition data is input.
  • the baseline facility condition data represents facility conditions during a first time interval before the energy conservation measures whose effect is to be measured in terms of savings have been implemented.
  • Baseline facility condition data can include weather conditions experienced by the facility during the time interval as well as occupancy data and production data.
  • the baseline facility condition data includes at least weather data.
  • the user can be prompted through user interface 26 (Fig. 1) to load or otherwise select data files to input.
  • some types of data can be automatically collected and input without user interaction.
  • the inputting steps be performed largely by loading or downloading data files, some types of data can be input by the user manually typing in the data.
  • the facility condition data is stored in database 36 upon inputting it to computer system 10 and prior to further processing. Nevertheless, in other embodiments the input data can be received, stored and otherwise manipulated in any suitable manner.
  • the baseline facility condition data includes historical datasets representing data gathered over a time interval of at least about 24 months and preferably no more than about 36 months.
  • Historical weather datasets can include, for example, measurements of dry bulb temperature, wet bulb temperature, and solar radiation for each hour of the time interval.
  • computing system 10 or an associated data gathering system that in turn provides data to computing system 10 can electronically collect (e.g., via the Internet) measurements of dry bulb temperature, wet bulb temperature, solar radiation, and other weather-related conditions for the geographic location of the facility from a weather agency such as the National Oceanic & Atmospheric Administration (NOAA).
  • NOAA National Oceanic & Atmospheric Administration
  • Historical occupancy datasets can include, for example, the peak number of persons occupying the facility on each day of the time interval. In instances in which the facility comprises one or more buildings with large variations in occupancy among them, which is sometime the case in the lodging and healthcare industries, peak daily occupation can be utilized, when available.
  • historical production data can also be included in the baseline facility condition data.
  • Historical production data can include, for example, the number of product units manufactured on each day of the time interval.
  • the production data can include production levels for each line. For buildings with many independent production lines, production lines are preferably aggregated into a smaller number of lines, such as about three to five lines. Occupancy and production data can be input by the user filling out spreadsheet templates, which convert the data for automated input to database 36.
  • step 44 of inputting baseline facility condition data further comprises performing some pre-processing of that data (by means of pre-processing element 32 (Fig. 1)) before inputting it to neural network model generator 28, as described below.
  • pre-processing element 32 Fig. 1
  • two additional weather-based statistical variables are created from the dry bulb temperature data: hours above saturation temperature per billing month and hours below saturation temperature per billing month. These two additional variables incorporate the latency effects of extreme temperatures on the heating and cooling loads of a building and the resulting energy use.
  • Other pre-processing can include summing all hourly and daily data and converting them to average monthly values that correspond to energy billing periods so that energy savings can be correlated more readily with energy utility company billings.
  • the saturation temperature is the average of the maximum and minimum saturation temperatures.
  • the maximum saturation temperature can be found by an iterative trial process of calculating the percentage of hours for each month of the time interval (e.g., 36 months) that is above the trial temperature. On the initial iteration, the trial temperature begins at the minimum hourly temperature over the time interval. As the trial temperature increases (e.g., in increments of one degree), fewer months will have 100% of their temperatures above the trial temperature. The maximum saturation temperature is found when no month has 100% of the hours above the trial temperature. Conversely, the minimum saturation temperature is found by an iterative trial process of calculating the percentage of hours for each month of the time interval (e.g., 36 months) that is below the trial temperature.
  • the trial temperature begins at the maximum hourly temperature over the time interval. As the trial temperature decreases in increments of one-degree, fewer months will have 100% of their temperatures below the trial temperature. The minimum saturation temperature is found when no month has 100% of the hours below the trial temperature.
  • the above-described iterative process generates two sinusoidal functions 43 and 45, representing percentage of hours above and percentage of hours below the saturation temperature, respectively, in each billing month, varying from 0% to 100% (0 to 1 ), i.e., percentage of hours of extreme temperatures. From the illustration, it can be seen that the saturation temperature ensures a very useful representation of the data by preventing the saturation of the curves with multiple points above 100% creating a flat peak and loss of data.
  • step 46 baseline energy consumed by the facility during the time interval is input.
  • the user can be prompted through user interface 26 (Fig. 1) to load or otherwise select data files to input.
  • computing system 10 or an associated data gathering system that in turn provides data to computing system 10 can electronically collect energy billing information over the time interval (e.g., using Electronic Data Interchange (EDI) protocols via network 40 (Fig. 1)).
  • EDI Electronic Data Interchange
  • the user can manually enter energy data into a spreadsheet template, which converts the data for automated input to database 36.
  • the energy consumption data is stored in database 36 prior to further processing
  • the input data can be received, stored and otherwise manipulated in any suitable manner.
  • computing system 10 can obtain automatically from a remote source via network 40, it automatically updates database 36 with new weather and energy data on a periodic (e.g., monthly) basis to maintain the baseline data in a current state.
  • Step 46 can further include performing some pre-processing on the energy data (e.g., by means of pre-processing element 32 (Fig. 1)) before inputting it to neural network model generator 28, as described below.
  • the data can be converted, if not already in such a form, to monthly energy consumption values corresponding to utility company billing periods.
  • Monthly energy billing data will usually require conversion because they are typically based on the energy supplier's (e.g., utility company's) reading of the applicable meter at the facility, not the calendar month.
  • the number of days for each billing month and the starting day for each billing month can vary by several days.
  • energy data should generally be converted (i.e., normalized) from monthly totals to daily averages, such as electricity in terms of kWh/day, and natural gas in terms of Btu/day.
  • energy variables principally electricity and natural gas
  • the energy use can be aggregated for a single energy use model using either Btu or kWh.
  • a summary of the baseline data that forms the input data for neural network model generator 28 is shown in Fig. 4. Note that, with regard to the use of a neural network algorithm, energy use can be considered a dependent variable, and weather or other facility conditions can be considered independent variables.
  • the exemplary screen display shown in Fig. 6 illustrates the manner in which the baseline facility condition data and baseline energy data can be displayed after they have been input and loaded into database 36 (Fig. 1).
  • the humidity values shown in Fig. 6 are calculated by the system and displayed instead of wet bulb temperature.
  • Database 36 can be a standard relational database defined by tables and the data relationships.
  • An exemplary table structure of database 36 is shown in Figs. 7-8 with descriptive textual labels indicating the table contents.
  • a facility relationships diagram, illustrated in Fig. 9, represents how buildings or other facility units are related to clients, corporate divisions, addresses, owners (potentially shared facilities), contacts, vendor accounts (energy supply), and other entities.
  • a vendor relationships diagram illustrated in Fig.
  • model generator 28 generates a neural network-based model
  • Model 50 (Fig. 1) in response to the baseline energy consumption data and baseline facility condition data.
  • Model 50 represents a facility's baseline energy use. As described below, model 50 can be used as a tool for projecting or estimating the amount of energy that would have been consumed by the facility but for the implementation of the energy efficiency or conservation measures in question.
  • parameters for the neural network algorithm must be defined that are appropriate for the application. These parameters include the data variability or noise level, data transformation scope, variable selection scope, and network selection scope. The last three parameters refer to the scope or the range of options the algorithm evaluates in finding the best distributions of data, data subsets of variables and network types such as Multi-Layer Perception (MLP) and Generalized Regression (GR). Persons skilled in the art to which the invention relates will readily be capable of defining suitable parameters for neural network tool 38 or other such neural network element.
  • MLP Multi-Layer Perception
  • GR Generalized Regression
  • FIG. 5 A summary of suitable neural network parameters and their selections for execution of model 50 to determine a building's baseline energy use is shown in Fig. 5.
  • FIG. 12 A table of exemplary baseline facility condition data and baseline energy consumption data to be input to model generator 28 is shown in Fig. 12.
  • the model parameters generally must be defined in terms of data variability, data transformations, variable subsets, and network types for input to the neural network-based model 70.
  • model generator 28 In response to the baseline energy consumption data and baseline facility condition data (and further based upon the selected model parameters, as described above with regard to Fig. 5), model generator 28 produces model 50.
  • Model 50 represents a baseline energy use model for the building or other facility unit that can produce monthly forecasts or projections over a time interval.
  • a table of exemplary output data produced by model 50 in response to the baseline energy consumption data and baseline facility condition data (and for the time interval to which this baseline data corresponds) is illustrated in Fig. 13.
  • Such baseline monthly forecasts or predictions enable a measure of the accuracy of model 50.
  • An average monthly error and a weighted total error can be calculated to ensure that the forecast produced by model 50 is within acceptable error tolerances. For example, a weighted total error of less than 2% may be considered acceptable.
  • model 50 is sufficiently accurate to be used to predict or estimate energy savings after an energy- conserving or energy efficiency-enhancing measure is implemented.
  • actual facility condition data represents facility conditions during a second time interval, after the energy-conserving or energy efficiency-enhancing measures have been implemented.
  • the actual facility condition data can include weather conditions experienced by the facility during the time interval as well as, in some embodiments of the invention, occupancy data and production data, as described above with regard to the baseline facility condition data.
  • the actual facility condition data can be input in the same manner as described above with regard to the baseline facility condition data.
  • Step 52 can further include performing pre-processing on the data (by means of pre-processing element 34 (Fig. 1)), similar to the pre-processing described above.
  • model 50 in response to the actual facility condition data, produces a prediction or estimate of the amount of energy that the facility would have consumed had the energy-conserving or energy efficiency-enhancing measures not been implemented.
  • the prediction comprises monthly values for each month in the second time interval.
  • the predicted or estimated energy consumption is subtracted from the actual energy consumption for each month in the second time interval.
  • the difference represents the estimated energy savings that resulted from implementing the energy-conserving or energy efficiency-enhancing measures.
  • the system can automatically re-perform some or all of the above-described steps, especially steps 46 and 48, so as maintain a current baseline.
  • the predicted or estimated energy consumption for the current month can be compared with the actual energy consumption for the current month.
  • EECs corresponding to the computed energy savings can be generated, displayed and stored.
  • a renewable energy credit represents proof that one MegaWatt hour (MWh) of electricity has been generated from a renewable-fueled source
  • an EEC generated in accordance with the present invention can represent proof that, for example, one megawatt hour (MWh) of energy has been saved as a result of implementing an energy-conserving or efficiency-enhancing measure.
  • EECs are denominated in MWhs and are equal to the energy savings, thus requiring no conversions or calculations.
  • the avoidance of air emissions associated with the energy savings and the EECs are calculated by using the United States Environmental Protection Agency (EPA) conversion factors that are location specific.
  • EPA United States Environmental Protection Agency
  • the system locates in the database the appropriate conversion factors for SO 2 , NOx and CO 2 .
  • Database 36 can maintain the most current EPA e-Grid data on the conversion factors.
  • the computed energy savings value and any related data that will be required by the certifying agency can be stored in database 36 or other data storage area in a format suitable for transfer to the certifying agency via either a paper form or electronic means.
  • the system makes a distinction between EECs that have been certified for sale and those tags yet to be certified. Typically, government regulatory agencies perform certification quarterly.
  • EECs can be traded on the market.
  • the EECs are referred to as "WHITE TAGS," a Sterling Planet, Inc. brand name.
  • WHITE TAGS a Sterling Planet, Inc. brand name.
  • model 50 Given that a user may have hundreds to thousands of buildings, the system is designed to be as completely automated as possible and serve primarily as a monitoring and reporting tool employing a highly advanced analytical engine. The complexities and operation of model 50 are mostly hidden from the user. However, the user may execute model 50 to evaluate various scenarios to determine the impact on the energy savings, EECs, and avoidance of air emissions. Scenarios may be either a change in location of the building or a change in the temperature (dry bulb).
  • the user can specify a different location for the building and execute the model to see how the building would perform in terms of energy usage in different climates. Similarly, the user may add or subtracts degrees of temperature to the average monthly temperatures (dry bulb). Scenarios apply to the created baseline energy use model and thus effect only the time period after the energy efficiency or conservation measure became operational.
  • new computer-based systems, computer-implemented methods, and computer program products that provide for additional pre-processing and postprocessing of data.
  • additional pre-processing generates a larger, more sufficient amount of energy-consumption data for training the neural network model from a smaller, less sufficient amount of energy-consumption data.
  • Neural networks rely on training to enable them to make accurate projections based on input data. The training process exposes the neural network to a variety of inputs and corresponding outputs. The wideness of the variety determines the range over which the neural network can respond with predictable accuracy.
  • the amount and variety of training data determines whether the neural network will tend to memorize (simple over-fitting), as opposed to generalize (learning). The more data available, and the wider the variety of that data, the more testing and verification can be done to avoid over-fitting.
  • the data-expansion pre-processing can be performed at step 46 (Fig. 2), for example by means of pre-processing element 32 (Fig. 1), before inputting the energy-consumption data to the neural network model generator 28.
  • the data- expansion pre-processing includes combining inputted individual energy-consumption datums for individual time periods into a greater number of energy-consumption datums. For example, a first utility bill could list the energy consumed by the facility over a 28-day period and a second utility bill could list the energy consumed by the facility over a 31 -day period that immediately follows the first period. These utility bills provide two individual raw datums — energy used in period one and energy used in period two.
  • the data-expansion process includes generating a third combined datum based on the first two individual raw datums.
  • the third combined datum is the total energy consumed by the facility in the combined first and second time periods.
  • the total energy consumed is the energy used in period one plus the energy used in period two.
  • the combined time period is the amount of time (e.g., the number of days) in period one plus the amount of time in period two — in this case, 59 days.
  • Utility bill time periods are typically sequential, so the data-expansion preprocessing can be set up to generate the third combined datum by using the start date of the first period and end date of the second period to get the combined time period. But in that case only sequential time period datums could be used in the process.
  • the data-expansion pre-processing may additionally include the step of first matching up start and end dates of the individual raw energy-consumption datums to make sure that the time periods are sequential before combining them this way.
  • the data-expansion pre-process was illustrated for two raw datums. The same process can be expanded for use with any number N of individual raw datums to generate Vz (N 2 + N) total (raw plus combined) datums. Each raw datum is combined with each other raw datum, individually and collectively in every unique combination available, to generate the combined datums. For example, Fig.
  • the data-expansion pre-process expands four individual raw datums (left column) into ten total datums (right column, including the raw datums and the combined datums).
  • the data-expansion pre- process generates six combined datums (datums 2, 3, 4, 6, 7, and 9) for inputting, in addition to the four raw datums (datums 1 , 5, 8, and 10), to the neural network model generator 28.
  • the data-expansion pre-process will expand three individual raw datums into six total (raw and combined) datums, five individual raw datums into fifteen total datums, and so forth.
  • this data-expansion pre-processing enables the neural network to be better trained with only a relatively few raw datums.
  • the preprocessing generates a larger amount and greater variety of energy-consumption datums for better training the neural network model.
  • the variety is increased because the time periods of the combined datums are longer than those for the raw datums.
  • the neural network model tends to "learn" rather than merely over-fit when exposed to the datums.
  • the raw energy-consumption datums need not be normalized (e.g., converted to represent average daily energy consumption).
  • the neural network accounts for the differing time periods of the energy-consumption datums, without actually normalizing them.
  • day-normalized datums the significance of any erroneous data present can be amplified. This can be because for example there is typically some post-processing to expand the projected energy savings out to a longer period (de-normalizing), thereby magnifying the error. Also, day-normalizing can reduce the accuracy of the datums.
  • one time period might include more weekends (when some facilities consume less energy because they are not open or not fully staffed/operational) than another.
  • day-normalizing the raw energy-consumption datums there is no (or at least less) subsequent de-normalizing, so outlier datums are less significant.
  • the datums used to train the neural network tend to more accurately reflect the causal relationships between actual facility conditions and actual energy consumption.
  • workdays per period can be integrated into the neural network, whereas when day-normalized this data is not integratable.
  • the data-expansion pre-processing includes time- normalizing the raw energy-consumption datums (e.g., to a daily basis).
  • the data-expansion pre-processing need not expand the raw datums into exactly Vz (N 2 + N) total datums.
  • the preprocessing can be set up so that only the largest possible time periods are used in new combined datums. In the example of Fig. 16, this would result in the right column only having the raw and combined datums 1 , 4, 5, 7, 8, 9, and 10.
  • Another exemplary embodiment includes post-processing that permits energy savings to be recognized from additional energy-conserving or energy-efficient measures and at the same time more accurately generates EECs even when facility energy use creeps upward.
  • a building's energy performance profile is modeled once, before an energy-conserving or energy-efficient measure is implemented (i.e., for the first time interval).
  • the EECs are determined based on the difference between actual post-measure energy consumption (i.e., in the second time interval), as reported by utility bills, and the projected energy use had the measure not been implemented and given the same facility condition data (i.e., weather, occupancy, units manufactured, etc.) in the second time interval. But this is only accurate if the facility condition data actually stays the same throughout the second time interval, which is not usually the case.
  • the facility condition data, and thus the actual energy used is dynamic and changes due to the increased load resulting from expanded operations. Such expanded operations might include, for example, the facility being used more intensely, more equipment being installed, and/or the facility being enlarged.
  • the actual energy used will tend to creep upward due to small incremental changes in the facility condition data such as hiring a new staff person, adding another copying machine, etc. Because of all this, in many cases the facility's actual energy consumption will tend to increase over time. This energy usage creep is illustrated graphically in Fig. 17. As the facility expands operations, the actual post-measure energy consumption in the second time interval (t2), as reported by utility bills, also increases. So the difference between the projected energy consumption — which is based on the outdated facility condition data for the second time interval — and the actual energy consumption, as determined in step 56, is reduced. Accordingly, the EECs generated would be reduced by a corresponding amount.
  • the post-processing includes repeating the method of Fig. 2 to generate an updated neural network-based model of the facility based on updated facility condition data, and generating updated EECs using the updated model.
  • This post-processing is done after the method of Fig. 2 has been completed (at least after step 54 has been completed), but before any additional energy-conserving or energy-efficient measures are implemented.
  • the post-processing can be done as soon as sufficient new actual energy consumption data for the second time interval is available (i.e., after a few utility bills have been received). Automatic meter reading and other conventional methods and devices can be used to get more actual energy use data in a shorter period of time.
  • the post-processing can be deferred and done later, just before a subsequent energy saving measure is implemented. In this way, more actual energy use data is available for inputting and training the neural network model.
  • the post-processing includes repeating the method of
  • Fig. 2 using updated facility condition data.
  • the updated facility condition data for the second time interval is inputted. This is based on the facility as it stands and is used in the second time interval after the initial energy-saving measure has been implemented but before the subsequent energy-saving measure is implemented.
  • the updated actual energy consumption for the second time interval is inputted. This is taken from the utility bills in the second time interval after the initial energy-saving measure has been implemented but before the subsequent energy-saving measure is implemented.
  • the updated energy usage model for the facility is generated based on these inputs.
  • the updated facility condition data for the third time interval is inputted to the model.
  • the model outputs the projected energy consumption (but for the subsequent energy saving measure that has been implemented) for the third time interval.
  • this is subtracted from the updated actual energy consumption for the third time interval, which is taken from the utility bills in the third time interval after the subsequent energy- saving measure has been implemented.
  • the EECs are generated based on the projected energy savings determined by step 56.
  • the updated facility condition data which more accurately represents the expanded operations of the facility, results in a more-accurate, updated energy usage model being generated.
  • the updated energy usage model outputs a more-accurate and increased (typically, but not necessarily) projected energy consumption (but for the subsequent energy saving measure that has been implemented) for the third time interval.
  • the projected energy savings determined at step 56 are more accurate and not compressed by the outdated energy usage model.
  • This increased projected energy consumption is illustrated graphically in Fig. 18.
  • the projected energy savings are not compressed due to expanded operations and usage creep. Accordingly, the corresponding EECs are not lessened in the third time period, so the facility gets the EECs it has earned. This allows the facility to continue to expand operations while continuing to earn EECs from prior efficiency measures without being impacted by new equipment installations.
  • the post-processing is performed just before the subsequent energy-saving measure is implemented, at the end of the second time period (t2).
  • the post-processing could be performed after each time of expanded operations, or it could be performed routinely as a regularly scheduled action (e.g., monthly, quarterly, etc.).
  • a facility might implement two measures of expanded operations, first upgrading its lighting system and subsequently retrofitting a chiller. If the EECs are being determined based on an energy use model generated based on facility condition data including the lighting system upgrade, but not the chiller retrofit, then the facility is being short-changed on EECs. By performing the model-updating post-process, the EECs will be determined based on an updated energy use model generated based on updated facility condition data including the lighting system upgrade and the chiller retrofit. So the EECs awarded will more accurately reflect the true energy savings.
  • this model-updating post-process may be used to provide a baseline against which the improved facility is graded on an ongoing basis as to how well its own conservation and operation and maintenance (O&M) measures are going. That is, the facility managers can compare the updated energy usage projection for the second time interval (from repeated step 54) to the actual energy consumption in the second time interval (from original step 56) to see if the numbers match-up reasonably well. This could be useful to help determine if the facility is operating as expected or better than it has historically.
  • O&M conservation and operation and maintenance
  • performing this post-processing eliminates the possibility of over-crediting EECs. For example, if after some time the facility managers shut down a production line or shift, then the actual energy consumption would decrease. This results in a change to the facility condition data. Without the model-updating post- process to reflect this change in the model, it will appear that the facility is saving more energy that is really is, and the facility could be awarded increased EECs that it has not earned. By performing the model-updating post-process, the energy use model is updated with the updated facility condition data, so the energy savings and corresponding EECs are more accurately determined on an on-going basis.
  • new computer-based systems, computer-implemented methods, and computer program products that provide for determining energy savings for an individual or group of individual energy-consuming systems within a facility or group of facilities.
  • the previously described embodiments provide for determining energy savings for an entire facility or group of facilities. But sometimes, instead of relying on high-level facility energy usage and savings data, is it desirable to be able to use system-specific energy usage and savings data.
  • a facility-based model is more of conservation measure than a system efficiency measure.
  • Energy efficiency may, at its most basic level, deal with the efficiency of a single system, be it lighting, refrigeration, or cooling, etc.
  • the facility may end up adding more equipment. This raises the energy usage overall, but nevertheless increases the efficiency.
  • the whole system is more taxed than before, which is the opposite of the intended effort, or at least counterproductive, if the intent is to just retard energy use growth as opposed to stopping it.
  • Conservation on the other hand, would likely look to stop energy use growth, or perhaps reduce usage overall.
  • the facility-based model rewards facilities as long as their total usage is below the baseline projected usage, but does not reward them if they use the energy savings from installing more-efficient equipment to justify installing more actual equipment. Second, in some situations it may be desirable to reward the efficiency of a specific system by itself, as opposed to rewarding only overall energy conservation.
  • the computer-based systems, computer- implemented methods, and computer program products of this embodiment are similar to those of the above-described embodiments, except that they are adapted for modeling energy use and determining energy savings for only one or a group of individual energy-consuming systems. This system-based modeling and EEC- determining process provides a lower-level granularity for individual systems relative to the facility-based approach of the other embodiments.
  • this system-based modeling and EEC-determining process includes the following steps, which are illustrated in Fig. 19.
  • baseline system condition data is generated and input to the neural network generator.
  • the system condition data does not include weather data, occupancy, or other most other of the factors included in the facility condition data. Rather, the system condition data includes the amount of time the system (or each system within the group) is operated in a given time period. This can be based on the average load level of the system during that period. Or this can be broken down into the amount of time the system is operated at each of several load levels (e.g., 25%, 50%, 75%, and 100% of capacity) during that period.
  • Pre-processing of the data can be done, for example, to time-normalize it.
  • energy consumption data is determined and input to the neural network generator.
  • the energy consumption data is collected while subjecting the energy-using system (or group of systems) to measurable loads.
  • the primary loads are the CPU, disk system, RAM system, network bandwidth, and fan speed.
  • the CPU is the largest load.
  • the fan is likely directly related to the other variables and so typically it can be ignored. This can be done over a short period of time by artificially operating the system at different load levels and measuring the energy consumption at each load level.
  • the load levels can be selected to correspond to the loadings used in step 60.
  • the system energy consumption at each load level can be measured by individually metering the system (i.e., the system is metered separately from the overall facility).
  • Pre-processing of the data can be done, for example, to time-normalize it.
  • this step can include factoring in the indirect energy usage for the system. For example, when the computer system is running it generates heat, and the cooling system then has to use energy to remove that amount of heat from the building. So the computer system itself may run at 5OW, but the cooling system might need to run at an additional 25W to cool the computer system.
  • this indirect energy usage by the cooling system can be added to and included in the system energy consumption data. This can be done by measuring or collecting the efficiency ratings of the indirect energy using equipment, for example, the Coefficient of Performance (COP), SEER, or EER of the cooling equipment. In alternative embodiments, for convenience this step is not included in the method, but in that case the energy savings determined by the method will not be as accurate.
  • a neural network model is generated based on these inputs.
  • the computer servers can now be consolidated through virtualization.
  • the updated system condition data for the virtualized machine is determined and input to the neural network model.
  • the model outputs a projected energy usage "but for" the server consolidation, at step 70 this projection is subtracted from the actual energy used by the virtualized machine after the consolidation, and at step 72 EECs are generated based on the projected energy savings.
  • the updated system condition data is determined similarly to what was done at step 60 except it would be based on the virtualized machine.
  • the actual energy used by the virtualized machine after the consolidation can be readily determined by individually metering the system.
  • the energy savings may be much smaller than the standard error for the model of an entire facility.
  • a typical data center might use about 5 MW (3,650 MWh per month), and a group of virtualized servers might save about 0.00057 MW (0.414 MWh per month). This is about a 0.01 % difference and within the standard error of a typical model, and therefore undetectable when looking at facility data.
  • this system-based modeling and EEC-determining process avoids the potential problems associated with small overall efficiency improvements and those having a negligible impact on large facilities with single meters.
  • the process captures savings where further construction or new installations will again raise overall consumption (i.e., by using the model-updating post-process described above). It also enables the near-instantaneous modeling of a system so that historical data stretching long periods into the past is not necessary.

Abstract

La présente invention concerne un système à base informatique, un procédé informatique et un produit programme informatique qui facilitent la détermination des économies en frais d'énergie dans une installation consommant de l'énergie telle qu'un bâtiment commercial, au moyen d'un modèle d'intelligence artificielle, par exemple d'un modèle de réseau neuronal, qui projette ou estime la quantité d'énergie qui aurait été consommée par l'installation sans l'application de mesures concernant le rendement énergétique ou la conservation de l'énergie. Les économies d'énergie sont représentées par la différence entre l'estimation de l'énergie qui aurait été consommée sans l'application des mesures et la quantité d'énergie réellement consommée par l'installation sans les conditions réelles au cours d'un intervalle de temps suivant l'application des mesures.
PCT/US2008/077232 2007-09-20 2008-09-22 Procédé et appareil de détermination des économies d'énergie réalisées en employant un modèle de consommation d'énergie de référence qui comprend un algorithme d'intelligence artificielle WO2009039500A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US97398407P 2007-09-20 2007-09-20
US60/973,984 2007-09-20

Publications (1)

Publication Number Publication Date
WO2009039500A1 true WO2009039500A1 (fr) 2009-03-26

Family

ID=40468433

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2008/077232 WO2009039500A1 (fr) 2007-09-20 2008-09-22 Procédé et appareil de détermination des économies d'énergie réalisées en employant un modèle de consommation d'énergie de référence qui comprend un algorithme d'intelligence artificielle

Country Status (1)

Country Link
WO (1) WO2009039500A1 (fr)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013138526A1 (fr) * 2012-03-13 2013-09-19 Lutron Electronics Co., Inc. Outil mobile et/ou à base de nuage pour permettre des informations précises sur des projets nouveaux et de rénovation
US8781767B2 (en) 2011-08-19 2014-07-15 General Electric Company Systems and methods for data anomaly detection
EP2457177A4 (fr) * 2009-07-23 2017-10-18 Siemens Industry, Inc. Système et procédé de qualification du fonctionnement de groupes à eau glacée
EP2457037A4 (fr) * 2009-07-23 2017-12-13 Siemens Industry, Inc. Pompage pour débit à la demande
US9927467B2 (en) 2014-10-14 2018-03-27 International Business Machines Corporation Estimating energy savings from building management system point lists
US20200098058A1 (en) * 2018-05-06 2020-03-26 Strong Force TX Portfolio 2018, LLC Systems and methods for fleet energy acquisition on a spot market
CN114456862A (zh) * 2020-11-04 2022-05-10 中国石油化工股份有限公司 一种天然气深冷脱氮预处理工艺参数的优化方法
US11567478B2 (en) 2020-02-03 2023-01-31 Strong Force TX Portfolio 2018, LLC Selection and configuration of an automated robotic process
US11599940B2 (en) 2018-05-06 2023-03-07 Strong Force TX Portfolio 2018, LLC System and method of automated debt management with machine learning
EP4283541A1 (fr) * 2022-05-27 2023-11-29 Honeywell International Inc. Prédiction basée sur l'apprentissage automatique de la ligne de base de consommation d'énergie
CN117436848A (zh) * 2023-12-21 2024-01-23 山东康吉诺技术有限公司 基于人工智能的电机能耗监测方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040225649A1 (en) * 2003-02-07 2004-11-11 Yeo Jeffrey W. Identifying energy drivers in an energy management system
US20050043862A1 (en) * 2002-03-08 2005-02-24 Brickfield Peter J. Automatic energy management and energy consumption reduction, especially in commercial and multi-building systems
US20070143045A1 (en) * 2005-12-21 2007-06-21 Sterling Planet Inc. Method and apparatus for determining energy savings by using a baseline energy use model that incorporates a neural network algorithm

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050043862A1 (en) * 2002-03-08 2005-02-24 Brickfield Peter J. Automatic energy management and energy consumption reduction, especially in commercial and multi-building systems
US20040225649A1 (en) * 2003-02-07 2004-11-11 Yeo Jeffrey W. Identifying energy drivers in an energy management system
US20070143045A1 (en) * 2005-12-21 2007-06-21 Sterling Planet Inc. Method and apparatus for determining energy savings by using a baseline energy use model that incorporates a neural network algorithm

Cited By (66)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2457177A4 (fr) * 2009-07-23 2017-10-18 Siemens Industry, Inc. Système et procédé de qualification du fonctionnement de groupes à eau glacée
EP2457037A4 (fr) * 2009-07-23 2017-12-13 Siemens Industry, Inc. Pompage pour débit à la demande
US8781767B2 (en) 2011-08-19 2014-07-15 General Electric Company Systems and methods for data anomaly detection
WO2013138526A1 (fr) * 2012-03-13 2013-09-19 Lutron Electronics Co., Inc. Outil mobile et/ou à base de nuage pour permettre des informations précises sur des projets nouveaux et de rénovation
US9927467B2 (en) 2014-10-14 2018-03-27 International Business Machines Corporation Estimating energy savings from building management system point lists
US11715164B2 (en) 2018-05-06 2023-08-01 Strong Force TX Portfolio 2018, LLC Robotic process automation system for negotiation
US11605124B2 (en) 2018-05-06 2023-03-14 Strong Force TX Portfolio 2018, LLC Systems and methods of smart contract and distributed ledger platform with blockchain authenticity verification
US11586994B2 (en) 2018-05-06 2023-02-21 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods for providing provable access to a distributed ledger with serverless code logic
US11928747B2 (en) 2018-05-06 2024-03-12 Strong Force TX Portfolio 2018, LLC System and method of an automated agent to automatically implement loan activities based on loan status
US11727506B2 (en) 2018-05-06 2023-08-15 Strong Force TX Portfolio 2018, LLC Systems and methods for automated loan management based on crowdsourced entity information
US11599940B2 (en) 2018-05-06 2023-03-07 Strong Force TX Portfolio 2018, LLC System and method of automated debt management with machine learning
US11599941B2 (en) 2018-05-06 2023-03-07 Strong Force TX Portfolio 2018, LLC System and method of a smart contract that automatically restructures debt loan
US11727319B2 (en) 2018-05-06 2023-08-15 Strong Force TX Portfolio 2018, LLC Systems and methods for improving resource utilization for a fleet of machines
US11605127B2 (en) 2018-05-06 2023-03-14 Strong Force TX Portfolio 2018, LLC Systems and methods for automatic consideration of jurisdiction in loan related actions
US11605125B2 (en) 2018-05-06 2023-03-14 Strong Force TX Portfolio 2018, LLC System and method of varied terms and conditions of a subsidized loan
US11610261B2 (en) 2018-05-06 2023-03-21 Strong Force TX Portfolio 2018, LLC System that varies the terms and conditions of a subsidized loan
US11609788B2 (en) 2018-05-06 2023-03-21 Strong Force TX Portfolio 2018, LLC Systems and methods related to resource distribution for a fleet of machines
US11620702B2 (en) 2018-05-06 2023-04-04 Strong Force TX Portfolio 2018, LLC Systems and methods for crowdsourcing information on a guarantor for a loan
US11625792B2 (en) 2018-05-06 2023-04-11 Strong Force TX Portfolio 2018, LLC System and method for automated blockchain custody service for managing a set of custodial assets
US11631145B2 (en) 2018-05-06 2023-04-18 Strong Force TX Portfolio 2018, LLC Systems and methods for automatic loan classification
US11636555B2 (en) 2018-05-06 2023-04-25 Strong Force TX Portfolio 2018, LLC Systems and methods for crowdsourcing condition of guarantor
US11645724B2 (en) 2018-05-06 2023-05-09 Strong Force TX Portfolio 2018, LLC Systems and methods for crowdsourcing information on loan collateral
US11657340B2 (en) 2018-05-06 2023-05-23 Strong Force TX Portfolio 2018, LLC Transaction-enabled methods for providing provable access to a distributed ledger with a tokenized instruction set for a biological production process
US11657339B2 (en) 2018-05-06 2023-05-23 Strong Force TX Portfolio 2018, LLC Transaction-enabled methods for providing provable access to a distributed ledger with a tokenized instruction set for a semiconductor fabrication process
US11657461B2 (en) 2018-05-06 2023-05-23 Strong Force TX Portfolio 2018, LLC System and method of initiating a collateral action based on a smart lending contract
US11669914B2 (en) 2018-05-06 2023-06-06 Strong Force TX Portfolio 2018, LLC Adaptive intelligence and shared infrastructure lending transaction enablement platform responsive to crowd sourced information
US11676219B2 (en) 2018-05-06 2023-06-13 Strong Force TX Portfolio 2018, LLC Systems and methods for leveraging internet of things data to validate an entity
US11681958B2 (en) 2018-05-06 2023-06-20 Strong Force TX Portfolio 2018, LLC Forward market renewable energy credit prediction from human behavioral data
US11687846B2 (en) 2018-05-06 2023-06-27 Strong Force TX Portfolio 2018, LLC Forward market renewable energy credit prediction from automated agent behavioral data
US11688023B2 (en) 2018-05-06 2023-06-27 Strong Force TX Portfolio 2018, LLC System and method of event processing with machine learning
US11710084B2 (en) 2018-05-06 2023-07-25 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods for resource acquisition for a fleet of machines
US11715163B2 (en) 2018-05-06 2023-08-01 Strong Force TX Portfolio 2018, LLC Systems and methods for using social network data to validate a loan guarantee
US20200098058A1 (en) * 2018-05-06 2020-03-26 Strong Force TX Portfolio 2018, LLC Systems and methods for fleet energy acquisition on a spot market
US11720978B2 (en) 2018-05-06 2023-08-08 Strong Force TX Portfolio 2018, LLC Systems and methods for crowdsourcing a condition of collateral
US11727504B2 (en) 2018-05-06 2023-08-15 Strong Force TX Portfolio 2018, LLC System and method for automated blockchain custody service for managing a set of custodial assets with block chain authenticity verification
US11727320B2 (en) 2018-05-06 2023-08-15 Strong Force TX Portfolio 2018, LLC Transaction-enabled methods for providing provable access to a distributed ledger with a tokenized instruction set
US11829907B2 (en) 2018-05-06 2023-11-28 Strong Force TX Portfolio 2018, LLC Systems and methods for aggregating transactions and optimization data related to energy and energy credits
US11580448B2 (en) 2018-05-06 2023-02-14 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods for royalty apportionment and stacking
US11727505B2 (en) 2018-05-06 2023-08-15 Strong Force TX Portfolio 2018, LLC Systems, methods, and apparatus for consolidating a set of loans
US11734774B2 (en) 2018-05-06 2023-08-22 Strong Force TX Portfolio 2018, LLC Systems and methods for crowdsourcing data collection for condition classification of bond entities
US11734620B2 (en) 2018-05-06 2023-08-22 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods for identifying and acquiring machine resources on a forward resource market
US11734619B2 (en) 2018-05-06 2023-08-22 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods for predicting a forward market price utilizing external data sources and resource utilization requirements
US11741553B2 (en) 2018-05-06 2023-08-29 Strong Force TX Portfolio 2018, LLC Systems and methods for automatic classification of loan refinancing interactions and outcomes
US11741552B2 (en) 2018-05-06 2023-08-29 Strong Force TX Portfolio 2018, LLC Systems and methods for automatic classification of loan collection actions
US11741402B2 (en) 2018-05-06 2023-08-29 Strong Force TX Portfolio 2018, LLC Systems and methods for forward market purchase of machine resources
US11741401B2 (en) 2018-05-06 2023-08-29 Strong Force TX Portfolio 2018, LLC Systems and methods for enabling machine resource transactions for a fleet of machines
US11748822B2 (en) 2018-05-06 2023-09-05 Strong Force TX Portfolio 2018, LLC Systems and methods for automatically restructuring debt
US11763214B2 (en) 2018-05-06 2023-09-19 Strong Force TX Portfolio 2018, LLC Systems and methods for machine forward energy and energy credit purchase
US11763213B2 (en) 2018-05-06 2023-09-19 Strong Force TX Portfolio 2018, LLC Systems and methods for forward market price prediction and sale of energy credits
US11769217B2 (en) 2018-05-06 2023-09-26 Strong Force TX Portfolio 2018, LLC Systems, methods and apparatus for automatic entity classification based on social media data
US11776069B2 (en) 2018-05-06 2023-10-03 Strong Force TX Portfolio 2018, LLC Systems and methods using IoT input to validate a loan guarantee
US11790288B2 (en) 2018-05-06 2023-10-17 Strong Force TX Portfolio 2018, LLC Systems and methods for machine forward energy transactions optimization
US11790286B2 (en) 2018-05-06 2023-10-17 Strong Force TX Portfolio 2018, LLC Systems and methods for fleet forward energy and energy credits purchase
US11790287B2 (en) 2018-05-06 2023-10-17 Strong Force TX Portfolio 2018, LLC Systems and methods for machine forward energy and energy storage transactions
US11810027B2 (en) 2018-05-06 2023-11-07 Strong Force TX Portfolio 2018, LLC Systems and methods for enabling machine resource transactions
US11816604B2 (en) 2018-05-06 2023-11-14 Strong Force TX Portfolio 2018, LLC Systems and methods for forward market price prediction and sale of energy storage capacity
US11823098B2 (en) 2018-05-06 2023-11-21 Strong Force TX Portfolio 2018, LLC Transaction-enabled systems and methods to utilize a transaction location in implementing a transaction request
US11829906B2 (en) 2018-05-06 2023-11-28 Strong Force TX Portfolio 2018, LLC System and method for adjusting a facility configuration based on detected conditions
US11586177B2 (en) 2020-02-03 2023-02-21 Strong Force TX Portfolio 2018, LLC Robotic process selection and configuration
US11586178B2 (en) 2020-02-03 2023-02-21 Strong Force TX Portfolio 2018, LLC AI solution selection for an automated robotic process
US11567478B2 (en) 2020-02-03 2023-01-31 Strong Force TX Portfolio 2018, LLC Selection and configuration of an automated robotic process
CN114456862A (zh) * 2020-11-04 2022-05-10 中国石油化工股份有限公司 一种天然气深冷脱氮预处理工艺参数的优化方法
CN114456862B (zh) * 2020-11-04 2022-10-21 中国石油化工股份有限公司 一种天然气深冷脱氮预处理工艺参数的优化方法
EP4283541A1 (fr) * 2022-05-27 2023-11-29 Honeywell International Inc. Prédiction basée sur l'apprentissage automatique de la ligne de base de consommation d'énergie
CN117436848A (zh) * 2023-12-21 2024-01-23 山东康吉诺技术有限公司 基于人工智能的电机能耗监测方法
CN117436848B (zh) * 2023-12-21 2024-03-08 山东康吉诺技术有限公司 基于人工智能的电机能耗监测方法

Similar Documents

Publication Publication Date Title
US7881889B2 (en) Method and apparatus for determining energy savings by using a baseline energy use model that incorporates an artificial intelligence algorithm
US8543343B2 (en) Method and apparatus for determining energy savings by using a baseline energy use model that incorporates an artificial intelligence algorithm
US7519485B2 (en) Method and apparatus for determining energy savings by using a baseline energy use model that incorporates a neural network algorithm
WO2009039500A1 (fr) Procédé et appareil de détermination des économies d'énergie réalisées en employant un modèle de consommation d'énergie de référence qui comprend un algorithme d'intelligence artificielle
Hong et al. Probabilistic electric load forecasting: A tutorial review
Xu et al. Discrete time–cost–environment trade-off problem for large-scale construction systems with multiple modes under fuzzy uncertainty and its application to Jinping-II Hydroelectric Project
CN105976258B (zh) 能量节约计量系统和方法
Bruton et al. Review of automated fault detection and diagnostic tools in air handling units
US8340831B2 (en) Non-intrusive load monitoring system and method
Armel et al. Is disaggregation the holy grail of energy efficiency? The case of electricity
Keirstead et al. A review of urban energy system models: Approaches, challenges and opportunities
US20120239164A1 (en) Graphical language for optimization and use
Mansouri et al. A robust optimisation framework in composite generation and transmission expansion planning considering inherent uncertainties
US20150347943A1 (en) Load profile management and cost sensitivity analysis
Shao et al. Multivariate statistical and similarity measure based semiparametric modeling of the probability distribution: A novel approach to the case study of mid-long term electricity consumption forecasting in China
Mařík et al. Decision support tools for advanced energy management
US10928784B2 (en) Central plant optimization system with streamlined data linkage of design and operational data
Lazzari et al. User behaviour models to forecast electricity consumption of residential customers based on smart metering data
El Kontar et al. Profiling occupancy patterns to calibrate Urban Building Energy Models (UBEMs) using measured data clustering
Gallagher et al. IntelliMaV: A cloud computing measurement and verification 2.0 application for automated, near real-time energy savings quantification and performance deviation detection
Rachunok et al. Assessment of wind power scenario creation methods for stochastic power systems operations
Mahbub et al. A neural approach to product demand forecasting
WO2013032754A1 (fr) Moteur de rapport avec modèle
Li et al. Analyzing heating equipment's operations based on measured data
Ambad et al. A methodology for design for warranty with focus on reliability and warranty policies

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 08831900

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 08831900

Country of ref document: EP

Kind code of ref document: A1