DE202022104934U1 - An IoT-enabled system to monitor, manage, and alert high-power consumers - Google Patents

Abstract

An IoT-enabled system for monitoring, managing, and alerting high-power loads, where the system monitors multiple loads to identify devices and estimate each device's load characteristics.

Description

Technical field of the invention:

The present invention relates to an IOT-enabled system for monitoring, managing and alerting the high power consumption utilities.

Background of the invention:

WO2011002735A1 - describes a load disaggregation system and method that measures electricity consumption metrics to identify appliances/devices in the building and estimate their corresponding consumption. This work uses machine learning algorithms to identify devices and estimate the power consumption of devices/household appliances in the building. The system consists of a single measurement system that is installed in a non-contact manner on the building's main power supply line. This module measures the voltage (V) and current (A) on the main power line that supplies power to the building. This measurement unit measures the metrics with a constant sample rate.

Groupings of alternative elements or embodiments of the invention disclosed herein are not to be construed as limitations. Each group member may be referenced and claimed individually or in any combination with other group members or other elements found herein. One or more members of a group may be included in or removed from a group for reasons of convenience and/or patentability. When such inclusion or deletion occurs, the specification is deemed to contain the group in modified form, thereby satisfying the written description of all Markush groups used in the appended claims.

Reference to ranges of values herein is intended solely as a shortcut and method of referring individually to each individual value that falls within the range. Unless otherwise noted herein, each individual value is incorporated into the specification as if individually cited herein. All of the methods described herein can be performed in any suitable order, unless otherwise noted herein or otherwise clearly contradicted by the context.

The use of any examples or exemplary phrases (e.g., "like") provided with respect to particular embodiments herein is intended only to better explain the invention and does not limit the scope of the invention otherwise claimed. No language in the description should be construed as indicating any non-claimed element essential to practice the invention.

The information disclosed above in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art.

aim of the invention

The main objective of the present invention is an IOT-enabled system for monitoring, managing and alerting the utilities with high power consumption.

Summary of the invention:

Accordingly, the present invention is an IOT-enabled system for monitoring, managing, and alerting the high power consumption utilities.

The system monitors multiple utilities to identify devices and estimate each device's load characteristics. The system effectively correlates and uses multiple utility consumption signals to identify and disaggregate equipment load information. The method is also adaptable and feasible when complex devices are added to the building, e.g. H. Devices such as cluster devices (devices with many load components), variable load devices, etc.

The system uses the utility consumption pattern from multiple utilities to improve the accuracy of the algorithm. The system also addresses the lack of information content in the power consumption signals and the similarity of the device's power consumption characteristics/patterns.

The system increases the accuracy/reliability of device detection and device category detection. The system also offers integrated multi-utility load disaggregation.

Detailed description of the invention:

The following description contains the preferred best mode of an embodiment of the present invention. From this description of the invention it is clear that the invention is not limited to these illustrated embodiments is limited, but that the invention also encompasses a large number of modifications and embodiments thereto. Therefore, the present description should be considered as illustrative and not restrictive. While the invention is susceptible to various modifications and alternative constructions, it should be understood that there is no intention to limit the invention to the specific form disclosed, but on the contrary the invention is intended to cover all modifications, alternative constructions and equivalents as may be mentioned in the patents The spirit and scope of the invention fall as defined in the claims.

In each embodiment described herein, the open-ended terms "comprise," "comprise," and the like (which are synonymous with "including," "having," and "characterized by") may be replaced by the respective partially closed phrases "consisting essentially of." consisting essentially of" and the like or the respective closed phrases "consisting of", "consists of, the like".

The present invention relates to an IOT-enabled system for monitoring, managing and alerting the high power consumption utilities.

The system non-intrusively monitors consumption from one or more utilities by one or more appliances/devices in a building and distributes consumption to individual utility or load consumer appliances. The system includes one or more monitoring and measurement modules, a combination module, a feature extraction module, an event recognition module, and a classification module.

The one or more monitoring and metering modules is a device used to monitor and meter usage of one or more utilities and generate a respective utility usage signal. A separate monitoring and measuring module is provided for each type of supply in the building/site. Each of the one or more monitoring and metering modules generates one or more operational consumption signals. The monitoring and measuring module is installed at the source of the respective power supply. Furthermore, each of the one or more monitoring and measuring modules has the same sampling frequency. Thus, the parameters monitored by the monitoring and measurement modules are measured at each sampling frequency. The combination module receives and unifies the utility signals from each of the one or more monitoring and metering modules with respect to time. In other words, the usage signals from each of the monitoring and metering modules are synchronized and vectored by the combination module to produce the unified usage signal. The one or more monitoring and measurement modules, together with the combination module, are referred to as a single-point non-intrusive measurement and monitoring module.

The feature extraction module extracts a plurality of features from the unified operational consumption signal. The feature includes, but is not limited to, one or more temporal, statistical, steady state, and transient features.

The event detection module detects each state change of one or more utility usages in the uniform utility usage signal by analyzing the extracted features. The event detection module sends information regarding a state change to the classification module for further processing.

The classification module receives the features from the feature extraction module where the event is recognized. The classification module identifies the one or more utility consuming devices, disaggregates the unified utility consumption signal, and calculates an amount of the utility used.

Alternatively, the classification module identifies a category of device and calculates the amount of utility used by the identified category of device.

The classification module first identifies a category of the device and then uses the identified category to determine the specific device.

Utility is a consumable that includes, but is not limited to, water, electricity, gas, and oil. The benefit also includes beverages such as milk, coffee, tea, juices, etc., which are dispensed through various types of vending machines.

Utilities include electricity, water, gas and other utilities that are comprehensively monitored by respective monitoring and metering modules.

The monitoring and measurement modules measure a variety of parameters or metrics for each of the utilities used. Those monitored and metered for electricity Parameters include but are not limited to voltage, current, real power, reactive power, apparent power, phase difference between voltage and angle, power factor, etc. Utility, the monitored and measured parameters are but not limited to valve pressure (PSI) and liters per hour. Similarly, for gas as a utility, the parameters are valve pressure (PSI) and one BTU per unit time, where BTU stands for British Thermal Unit.

Supply/energy consumption across multiple supply sources is measured and unified. The unified utility consumption signal includes multiple dimensions based on the number of utilities measured. The dimensions of some utilities include time, power consumption in KW, water consumption in 5 gallons/hour, gas consumption in cc/hour. The possible uses and corresponding dimensions are provided for illustrative purposes and should not be considered as limiting.

The uniform operational consumption signal includes time-synchronized operational consumption signals from each of the one or more monitoring and metering modules.

The unified utility consumption signal is a multi-dimensional (or high-dimensional) signal with different measurement metrics for the one or more utilities. The multi-dimensional signal or data includes more entropy or information than one-dimensional data. Having more information relies on the correlation properties that exist between each dimension of the monitored care data. Each appliance signature takes into account one or more supply parameters. For example: A lamp includes a non-zero current signature but has a dead signal to other utilities. Similarly, a gas stove has a non-zero gas signature, but electricity and other utilities include a zero value.

The system and method uses different cross-correlation properties between the different utility consumption signals in the uniform utility consumption signal. Cross-correlation and correlation properties are used to identify each device and disaggregate the load. Considering a supply load pattern for a lamp and a water heater, only one power consumption signal will experience a change when the lamp is turned on. In the case of a lamp, the cross-correlation of the power consumption signal with another supply signal is zero. A zero cross-correlation means that the lamp condition has no effect on the uniform consumption signal, except for electricity. When the water heater is turned on, a change is detected in both the electricity and water usage signals. In the case of a water heater, there is a non-zero cross-correlation between the power consumption signal and the water consumption signal. The cross-correlation properties are used to more reliably identify the device.

The merging and synchronization of multiple consumption signals into a unified consumption signal enables the derivation of the cross-correlation and correlation properties between different consumables that are detected by one or more monitoring and measurement modules. The cross-correlation and correlation properties increase the accuracy and reliability of the proposed system and method.

The unified utility consumption signal increases the reliability and accuracy of load sharing of those devices that use multiple utilities for their operations. For example, the devices that use multiple utilities include, but are not limited to, a washing machine, a water heater, a dishwasher, etc. that use both electricity and water for their respective operations. The properties also indirectly help to increase the reliability and accuracy of a device that uses only one utility. Because the lack of consumption signals in all but one utility increases the likelihood of correctly detecting the device with a utility load.

One or more auxiliary sensors are provided in the building to provide additional information such as humidity, light, weather, temperature, pressure, etc. The system aims to increase the accuracy of recording and breakdown of consumption. For example, when the natural light is less, there is a high probability that the lights in the building are on. This increases the likelihood that a device will be detected as a light. Similarly, an increase in ambient temperature increases the likelihood that the air conditioning (AC) is on.

The feature extraction module is customized to extract a variety of features from the in Extract real-time recorded data of the unified operational consumption signal. The plurality of features are extracted by applying one or more machine learning algorithms, pattern recognition algorithms, and so on. The temporal characteristics are properties of the unified utility signal over time. The statistical features are probabilistic properties of the unified utility signal.

The statistical characteristics include, but are not limited to, standard mean, deviation, and variance. When a device is turned on, the operational consumption signal has two components, namely a transient and a steady-state response. Transient behavior corresponds to the moment when the device is turned on, while steady-state behavior corresponds to the characteristics of the signal after the device has been turned on for a certain period of time. Furthermore, the transient and steady-state application patterns are multi-dimensional or more than two-dimensional. The temporal, statistical, transient, and steady state characteristics of each utility's consumption signal (metric) are captured using signal processing techniques. The multiple features are extracted either independently or in combination with each other.

The event detection module continuously checks for a state change of one or more utility usages in the extracted plurality of features. State change is defined as an event for a device or a device within a building or property. The event detection module includes a set of logic rules to detect the occurrence of the event. For example, if there is a change in power consumption greater than fifty watts, then that change is considered an event. Similarly, multiple logic rules define the occurrence of the event for different devices.

Once the event is detected, corresponding features are extracted from the recently recorded unified operational consumption signal. The features are extracted from a predefined time frame. In addition, the time frame is fixed and adjustable. An example is provided to illustrate the process of defining the timeframe and should not be taken as a limitation. If an event is detected at time t, then the signal start from (t - 20 seconds) to (t + 20 seconds) is defined as the time frame for the unified utility signal. The 40 second signal is used to extract the features for the detected event from the uniform operational consumption signal.

The classification module is used to split energy consumption among individual energy consuming devices or devices. The classification module includes a database storing known usage signal patterns or device signatures for devices to be used in the building. The known patterns are represented in the same way as those of the unified useful consumption signal. Thus, the known useful consumption signal patterns use cross-correlation properties. The signatures of a transient and stationary appliance are multidimensional (more than two dimensions). The device's signatures are defined in higher-dimensional data that spans the building's various utilities. All utilities defined for the building are taken into account when defining the application signature. The cross-correlation and correlation properties are captured by the classification module. Based on the known patterns, a classification model called a classifier is created.

The process of creating a classifier is called training, which involves retrieving known patterns from the database, followed by extracting features from the known patterns. Once the classification model and are created, the classification model is applied to real-time data to classify and identify the utility consuming devices. The trained classification model is built using any of a variety of pattern recognition algorithms and machine learning techniques. The pattern recognition algorithms include, but are not limited to, a cluster algorithm, Markov process-based algorithms, and so on.

The classification module creates one or more classification models for determining a category of the utility devices. The classification module creates one or more classification models for determining the individual utility device. In addition, one or more classification models are created and adjusted to first determine the category of high-consumption devices and then narrow down to identify the specific high-consumption device.

The classification model is used for reliable and efficient identification of devices/systems in the building and for load disaggregation of the identified devices. The properties of the useful consumption are defined by predefined flags male shown on different useful consumption signals. Once the devices or device category are identified, a device estimation module is used to calculate the amount of usage.

The database for the classification engine is provided (or manually updated) with known samples of the utility consuming devices before the system is deployed to a building or property. Alternatively, once the system is installed in the building, the database is automatically updated with the unknown (or new) payload patterns for one or more devices in real time. The unknown patterns are then further processed to identify the corresponding device and over time the one or more unknown devices in the building are identified and corresponding load profiles are recorded. The system is also able to identify small variations in existing load patterns from one or more utilities and update them in the database when detected. The known patterns are also referred to as appliance/device signatures. The known patterns are defined/represented in high dimensional data spanning multiple utilities of the system and before the application signature is defined, each metric of the one or more utilities is considered.

The feature extraction engine, event detection engine, and classification engine are collectively referred to as the intelligent disaggregation engine.

A building containing N devices and consuming M resources. The system estimates and identifies the N devices, a corresponding state (ON/OFF), and their consumption of utilities for each of the M utilities (or specific utilities of interest). For each of the M utilities, a metering and monitoring module is connected to the main supply/line of the utility. The measurement and monitoring module measures parameters related to useful consumption. The measuring and monitoring module attached to the power supply line measures, for example, the voltage (V), the current (A) and the power factor. Power consumption is derived from these three metrics. The one or more M&M modules have the same sampling frequency and are synchronized and vectored to generate the unified payload signal. Therefore, parameters related to M utilities are measured at each sampling period. The temporal, statistical, transient and steady-state properties of each measured signal for each utility considered in the methodology are calculated.

The properties are represented as features and these are extracted from the measured raw signal at the feature extraction module. The extracted features are then used by the event detection module to detect a change in utility load that corresponds to a state change of the building equipment or a state change of the building power supply. The classification engine uses a machine learning model previously modeled for device usage patterns based on current conditions such as time, current active/inactive devices, etc. This model is derived using a hidden Markov model modeled on time-series data of all known devices. Device category identification and device identification modules use a machine learning based classification model to identify device category and devices. The presented invention uses several models based on different machine learning techniques.

The non-contact load monitoring method includes the steps of monitoring and measuring consumption of one or more utilities and generating one or more corresponding utility consumption signals. The monitoring and measurement of the one or more utilities is performed by one or more monitoring and measurement modules.

The one or more useful consumption signals are combined by a combining module into a uniform useful consumption signal. Combining one or more payload signals includes synchronizing the one or more payload signals in time. A plurality of features, including temporal, statistical, steady-state, and transient features, are extracted from the uniform operational consumption signal by a feature extraction module. Extracting multiple features includes using correlation properties between multiple utility consumption signals in the unified utility consumption signal. The extracted features are then analyzed for any state change of a device or state change of use of the utility. When a state change is detected, the corresponding extracted features are associated with one or more devices using one or more classification models. If there is no state change, the step is repeated. The classification of the extracted char Maling involves applying a classification model to identify a device and a category of the device.

The amount of useful consumption by the one or more devices is calculated or billed. The calculation or calculation of the useful consumption by one or more devices is carried out using an estimation model.

The system monitors multiple utilities to identify devices and estimate each device's load characteristics.

The system effectively correlates and uses multiple utility consumption signals to identify and disaggregate equipment load information. The method is also adaptable and feasible when complex devices are added to the building, e.g. H. Devices such as cluster devices (devices with many load components), variable load devices, etc.

The system uses the utility consumption pattern from multiple utilities to improve the accuracy of the algorithm.

The system also addresses the lack of information content in the power consumption signals and the similarity of the device's power consumption characteristics/patterns.

The system increases the accuracy/reliability of device detection and device category detection. The system also offers integrated multi-utility load disaggregation.

Additional advantages and modifications will readily occur to those skilled in the art. Therefore, the invention in its broader aspects is not limited to the specific details and representative embodiments shown and described herein. Accordingly, various modifications can be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• WO 2011002735 A1 [0002]

Claims (5)

An IoT-enabled system for monitoring, managing, and alerting high-power loads, where the system monitors multiple loads to identify devices and estimate each device's load characteristics. IoT-enabled system to monitor, manage and alert high-power consumers claim 1 , where the system effectively correlates and uses multiple consumption signals to detect and disaggregate device load information. IoT-enabled system to monitor, manage and alert high-power consumers claim 1 , where the system evaluates usage patterns from multiple consumers to improve the accuracy of the algorithm. IoT-enabled system to monitor, manage and alert high-power consumers claim 1 , where the system handles the lack of information content in the power consumption signals and the similarity of the power consumption characteristics/patterns of the device. IoT-enabled system to monitor, manage and alert high-power consumers claim 1 , whereby the system increases the accuracy/reliability of device detection and device category detection.