Non-intrusive load monitoring method and device

Abstract

The invention relates to the technical field of load monitoring, and discloses a non-intrusive load monitoring method and device. The method comprises the following steps: according to a total power sequence of a to-be-identified power grid, determining an event time point of a switching event of electric equipment in the to-be-identified power grid and a power feature corresponding to the switching event; performing clustering processing on the power features, and determining typical power features of the power grid to be identified; and carrying out classification processing on the typical power features, and determining electric equipment type information of the to-be-identified power grid according to a classification result. According to the method and the device, the use conditions of the electric equipment of different power types in the to-be-identified power grid in the preset sampling period can be determined only according to the power utilization total power data of the to-be-identified power grid without any intrusive monitoring load data.

Description

Technical Field

[0001] The present application relates to the technical field of load monitoring, and in particular to a non-invasive load monitoring method and device. Background Art

[0002] Existing load monitoring can be divided into intrusive load monitoring and non-intrusive load monitoring. Among them, intrusive load monitoring equips each electrical device with a data acquisition device, which can accurately obtain the load data of different electrical devices to analyze the behavior and energy consumption of different electrical devices. However, this method is costly, complicated to operate, and involves user privacy issues, which is not conducive to large-scale promotion. Non-intrusive load monitoring only installs a data acquisition device at the upstream electrical entrance. In theory, it only uses the total load data as information to monitor the power consumption of each internal electrical device, which can well overcome the shortcomings of intrusive load monitoring. Among them, non-intrusive load monitoring mainly includes two aspects: load identification and load decomposition. Load identification aims to extract the characteristics of different electrical equipment at the operating time, and identify the electrical equipment that the user is using through these characteristics; load decomposition is to decompose the total power into the power of each electrical equipment.

[0003] However, in the relevant non-invasive load monitoring solutions, it is still necessary to know the load data of each electrical equipment in advance, so it is inevitable to use the load data of intrusive monitoring, which is contrary to the original intention of non-invasive load monitoring, and cannot give full play to the advantages of non-invasive load monitoring, and cannot guarantee the effect of non-invasive load monitoring. Summary of the invention

[0004] In view of this, the present application provides a non-invasive load monitoring method and device, which can classify the electrical equipment used in the power grid to be identified within a preset sampling period into several power types based only on the total power consumption data of the power grid to be identified within the predicted sampling period, without any intrusive monitoring load data, so as to understand the usage of electrical equipment of different power types from a macro perspective.

[0005] According to one aspect of the present application, a non-intrusive load monitoring method is provided, comprising: determining, based on the total power sequence of the power grid to be identified, an event time point at which a switching event occurs on an electrical equipment in the power grid to be identified and a power feature corresponding to the switching event; wherein the power grid to be identified includes a plurality of the electrical equipment, the total power sequence includes the total power consumption of the plurality of electrical equipment at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the electrical equipment is connected or disconnected; clustering the power features to determine typical power features of the power grid to be identified; and classifying the typical power features to determine the electrical equipment type information of the power grid to be identified based on the classification results.

[0006] According to another aspect of the present application, a non-intrusive load monitoring device is provided, comprising: a determination module, for determining, based on the total power sequence of the power grid to be identified, an event time point at which a switching event occurs on an electrical equipment in the power grid to be identified and a power feature corresponding to the switching event; wherein the power grid to be identified includes a plurality of the electrical equipment, the total power sequence includes the total power consumption of the plurality of electrical equipment at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the electrical equipment is connected or disconnected; and, clustering the power features to determine typical power features of the power grid to be identified; and, classifying the typical power features to determine the electrical equipment type information of the power grid to be identified based on the classification results.

[0007] Through the above technical scheme, the present application provides a non-invasive load monitoring method and device, which does not require the use of invasively monitored load data, protects user privacy, and only monitors the switching events of electrical equipment based on the total power consumption data of multiple electrical equipment within the predicted sampling period, and extracts the power characteristics at the time point of the switching event, so as to cluster and classify all the extracted power characteristics. It does not focus on determining the power characteristics of each electrical equipment, but rather divides the electrical equipment used in the power grid to be identified within the preset sampling period into multiple power types from a macro perspective, determines the type information of the electrical equipment of the power grid to be identified, so as to understand the usage of electrical appliances of different power types, can be applied to various scenarios, improve the efficiency and versatility of non-invasive load monitoring, and help power companies to conduct overall load analysis, plan power grid capacity, and optimize power generation and distribution strategies.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0010] Figure 1 A schematic diagram of a flow chart of a non-intrusive load monitoring method provided in an embodiment of the present application is shown;

[0011] Figure 2 The structure block diagram of the non-intrusive load monitoring device provided in the embodiment of the present application is shown;

[0012] Figure 3The clustering result diagram provided by the embodiment of the present application is shown;

[0013] Figure 4 A prediction information diagram provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0014] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0015] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0016] Those skilled in the art will appreciate that, unless expressly stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, or there may be intermediate elements. In addition, the "connection" or "connection" used herein may include wireless connection or wireless fusion. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0017] Now, exemplary embodiments according to the present application will be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in a variety of different forms and should not be interpreted as being limited to the embodiments set forth herein. It should be understood that these embodiments are provided to make the disclosure of the present application thorough and complete, and to fully convey the concepts of these exemplary embodiments to those of ordinary skill in the art.

[0018] The non-intrusive load monitoring method provided in the embodiments of the present application can be applied to the terminal, can also be applied to the server, and can also be software running in the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the non-intrusive load monitoring method, etc., but is not limited to the above forms.

[0019] In this embodiment, a non-intrusive load monitoring method is provided. Figure 1 As shown, the method includes:

[0020] Step 101, according to the total power sequence of the power grid to be identified, determine the event time point of the switching event of the power consumption equipment in the power grid to be identified and the power characteristics corresponding to the switching event.

[0021] The power grid to be identified includes multiple power-consuming devices, the total power sequence includes the total power consumed by the multiple power-consuming devices at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the power-consuming device is connected or disconnected.

[0022] Specifically, the total power consumption at each sampling time point within a preset sampling period can be obtained from the power bus. The total power consumption at each sampling time point is the sum of the active power of multiple power-consuming devices connected to the power bus at the sampling time point. The total power is arranged according to the sampling time to form a total power sequence. The total power sequence can be obtained in real time or from historical records.

[0023] In this embodiment, there is no need for detailed load data of each electrical equipment. Only the switching events of each electrical equipment within a preset sampling period are monitored according to the total power sequence, the time when the switching event occurs is recorded, and the power characteristics corresponding to the switching event are extracted, so as to provide accurate analysis objects for subsequent non-invasive load monitoring and provide scientific data for subsequent analysis.

[0024] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, the steps of determining the event time point of the switching event of the electrical equipment in the power grid to be identified according to the total power sequence of the power grid to be identified specifically include: sliding a window of a preset length on the total power sequence to determine the window power parameters; determining the event time point of the switching event of the electrical equipment according to the window power parameters.

[0025] In this embodiment, a window of preset length is used to gradually scan the entire total power sequence to achieve continuous monitoring of the total power sequence and ensure the comprehensiveness of monitoring. The window power parameter is used as the representative value of the window to reflect the fluctuation of the total power in the window, and then the event time point of the switching event of the electrical equipment is determined according to the window power parameter, providing a basis for the subsequent determination of the power characteristics corresponding to the switching event.

[0026] Among them, the preset length of the window is determined according to the frequency of switching events of electrical equipment in the actual application scenario. For example, when analyzing residential electricity consumption data, the window length can be set to 3 to meet the situation where electrical appliance switching events in residential electricity consumption occur frequently and prevent missing short switching events.

[0027] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, a window of a preset length is slid on the total power sequence, and the steps of determining the window power parameters specifically include: calculating the power difference between the total powers corresponding to adjacent sampling time points in the window; sorting the power differences to determine the difference order; and determining the window power parameters according to the power difference that ranks first in the difference order.

[0028] In this embodiment, the maximum value of the power difference between the total powers corresponding to adjacent sampling time points in the window is used as the window power parameter to intuitively and accurately quantify the fluctuation of the total power in each window, reduce the complexity of data processing, and thus timely and effectively capture various power changes in the window.

[0029] Exemplarily, calculating the power difference between the total power corresponding to the next sampling time point and the previous sampling time point in the window is equivalent to determining the power change rate at each sampling time point. Sort the power differences from large to small, determine the difference order, and use the largest power difference at the first position in the difference order as the window power parameter to capture significant power changes in the window and provide an accurate basis for judging the occurrence of switching events.

[0030] For example, the length of the window is set to 3, the total power corresponding to the third sampling time point in the window is subtracted from the total power corresponding to the second sampling time point to determine the first power difference, and the total power corresponding to the second sampling time point in the window is subtracted from the total power corresponding to the first sampling time point to determine the second power difference. If the first power difference is greater than the second power difference, the first power difference is determined as the window power parameter of the window.

[0031] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiments, in order to fully illustrate the specific implementation process of this embodiment, the steps of determining the event time point of the switching event of the electrical equipment according to the window power parameter specifically include: if the window power parameter is greater than the first preset threshold value, it is determined that a switching event of the electrical equipment occurs in the window, and the sampling time points associated with the power difference values ​​corresponding to the window power parameter greater than the first preset threshold value are sorted to determine the time sequence; the event time point is determined according to the sampling time point that is first in the time sequence.

[0032] In this embodiment, a threshold is set to filter normal load fluctuations during the operation of electrical equipment, thereby preventing misjudgment and accurately capturing power changes when electrical equipment is switched on and off, thereby accurately locating the point at which the electrical equipment is switched on and off, and improving the reliability of non-invasive load monitoring.

[0033] Exemplarily, if the window power parameter is greater than the first preset threshold, it is determined that a switching event occurs in the window, and the sampling time point with a forward time sequence among the sampling time points used when calculating the power difference value as the window parameter value is used as the event occurrence point of the switching event. If the window power parameter is less than or equal to the first preset threshold, although power fluctuations may occur in the window, they are regarded as normal load fluctuations when the electrical appliance is running, and it is determined that no switching event occurs in the window.

[0034] For example, continuing to take the above-mentioned first power difference as an example, if the first power difference as a window power parameter is greater than the first preset threshold, it is determined that a switching event of an electric device occurs in the window. Further, the second sampling time point and the third sampling time point in the window used when calculating the first power difference, the second sampling time point with a forward time sequence is used as the event occurrence point of the switching event.

[0035] The first preset threshold is set according to the power of the electrical equipment in normal operation in the actual application scenario to reflect the normal power fluctuation of the electrical equipment. For example, in the residential power usage scenario, the first preset threshold is set to 200 to distinguish between normal power fluctuations and switching events in the residential power usage scenario.

[0036] Further, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, the power characteristics corresponding to the switching event are determined according to the total power sequence of the power grid to be identified, including: determining the active power characteristics of the switching event according to the difference between the total power corresponding to the event time point and the next sampling time point adjacent to the event time point in the total power sequence; determining the reactive power characteristics of the switching event according to the difference between the reactive power corresponding to the event time point and the next sampling time point adjacent to the event time point; and using the active power characteristics and the reactive power characteristics as the power characteristics corresponding to the switching event.

[0037] In this embodiment, the active power change value and the reactive power change value are used together as the power characteristics corresponding to the switching event, so as to provide more characteristic information for non-intrusive load monitoring and improve the accuracy of load monitoring.

[0038] Exemplarily, in the total power sequence, the total power corresponding to the next sampling time point adjacent to the event time point where the switching event occurs is subtracted from the total power corresponding to the event time point where the switching event occurs to determine the active power change value when the switching event occurs, and the first power difference is used as the active power feature of the switching event to reflect the active power fluctuation when the switching event occurs.

[0039] Similarly, the reactive power corresponding to each sampling time point within the preset sampling period is obtained, wherein the reactive power corresponding to the sampling time point is the sum of the reactive powers of multiple electrical devices at the sampling time point. The reactive power corresponding to the next sampling time point adjacent to the event time point where the switching event occurs is subtracted from the reactive power corresponding to the event time point where the switching event occurs, and the reactive power change value when the switching event occurs is determined, and the reactive power change value is used as the reactive power feature of the switching event to reflect the reactive power fluctuation when the switching event occurs. Thus, the active power feature and the reactive power feature are used together as the power feature corresponding to the switching event to comprehensively extract the load feature of the switching event and improve the reliability of load monitoring.

[0040] Step 102: cluster the power characteristics to determine the typical power characteristics of the power grid to be identified.

[0041] In this embodiment, power features corresponding to several switching events are clustered, and similar power fluctuations generated when switching events occur in electrical equipment with similar operating powers are classified into one category, thereby extracting typical power features corresponding to switching events of electrical equipment of several power types from multiple power features.

[0042] For example, the power fluctuation generated when a power consumption device with an operating power of about X is switched on is similar, but different from the power fluctuation generated when a power consumption device with an operating power of about Y is switched on. X and Y are not equal in size and have a certain numerical difference, for example, 800W and 1500W. After clustering the power features, the typical power features corresponding to the switching event of the power consumption device with an operating power of about X and the typical power features corresponding to the switching event of the power consumption device with an operating power of about Y can be obtained.

[0043] Further, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, step 102, that is, clustering the power characteristics to determine the typical power characteristics of the power grid to be identified, specifically includes: determining a data point set based on the power characteristics; randomly selecting any data point in the data point set as the first initial clustering center; removing the data points in the data point set from the initial clustering center as candidate data points; sorting the distances between the candidate data points and the initial clustering center to determine the distance order; taking the last distance in the distance order as the target distance of the candidate data point; determining the probability of the candidate data point being the next initial clustering center based on the target distance; determining the next initial clustering center based on the probability of the candidate data point being the next initial clustering center, and updating the candidate data points until the number of initial clustering centers is equal to the preset number; determining a preset number of clusters based on the initial clustering centers; determining the typical power characteristics based on the power characteristics of the cluster centers in the clusters.

[0044] In this embodiment, during the clustering process, the initial cluster centers are selected as dispersedly as possible so that the clusters obtained according to the initial cluster centers are more consistent with the actual distribution of power characteristics, so that the typical power characteristics obtained according to the clusters can accurately represent the power fluctuations generated when switching events occur in electrical equipment of different power types.

[0045] Exemplarily, the obtained power features are converted into a two-dimensional data point set, and each data point in the data point set can be marked in a two-dimensional coordinate. Further, a data point is randomly selected as the first initial cluster center, and the data points in the data point set excluding the initial cluster center are taken as candidate data points. Next, the minimum distance from the candidate data point to the selected initial cluster center is calculated as the target distance of the candidate data point, that is, for any candidate data point, the minimum distance from the candidate data point to each selected initial cluster center is determined as the target distance of the candidate data point. The distance can be calculated according to the Euclidean metric method.

[0046] Furthermore, the probability of a candidate data point being selected as the next initial cluster center is calculated based on the target distance of the candidate data point. This probability indicates that the farther the candidate data point is from the selected initial cluster center, the greater the possibility of being selected as the next initial cluster center. i The probability of being selected as the next initial cluster center is expressed as P(x i ), P(x i ) is calculated as x i is a candidate data point, n is the number of candidate data points, d(x i ) is the target distance.

[0047] Furthermore, according to the probability P(x i ), determine the cumulative probability of the candidate data points. Exemplarily, the candidate data point set D = {x1, x2, ... x n}The corresponding probability is P = {P(x1), P(x2), ... P(x n )}. Cumulative probability C = {C1, C2, ... C n}, where C1=P(x1), Ci=C i-1 +P(x i ).

[0048] Next, generate a random number r in the interval [0,1), and start from the first cumulative probability C1, and compare the random number r with each cumulative probability C1 in turn. i When we find the first <C i When the cumulative probability of i is selected as the next initial cluster center.

[0049] Furthermore, after obtaining a preset number of initial cluster centers according to the above process, for any candidate data point in the data point set excluding the initial cluster center, the distance from the candidate data point to the preset number of initial cluster centers is calculated, and the candidate data point is divided into clusters represented by the initial cluster center closest to it. For any cluster, the mean of all data points in the cluster is calculated, and the initial cluster center is replaced as the new cluster center. The above process of allocating data points and updating cluster centers is repeated until the preset conditions are met and the preset number of clusters is determined. Among them, the preset conditions can be set to the cluster center no longer changing, or reaching a preset number of iterations, or reaching a certain error threshold.

[0050] Furthermore, the power feature of the cluster center in a cluster is taken as a typical power feature to represent all similar power features included in the cluster, thereby obtaining a preset number of typical power features.

[0051] Step 103: classify the typical power characteristics and determine the type information of the power grid to be identified based on the classification result.

[0052] In this embodiment, typical power characteristics are classified and processed, and the power characteristics corresponding to the on and off events of electrical equipment of various power types are specifically determined according to the classification results, so that multiple electrical equipment used in the power grid to be identified within a preset sampling period are divided into multiple electrical appliances of known power types. The focus is not on determining the respective usage of each electrical equipment, but on determining the type information of electrical equipment in the power grid to be identified, understanding the usage of electrical appliances of different power types from a macro perspective, protecting user privacy, and improving the efficiency and versatility of non-invasive load monitoring. It can be applied to various scenarios and help power companies to conduct load forecasting and scheduling.

[0053] The on event is used to indicate that the power of the electrical device is turned on, and the off event is used to indicate that the power of the electrical device is turned off.

[0054] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiment, in order to fully illustrate the specific implementation process of this embodiment, the steps of classifying and processing the typical power characteristics specifically include: classifying the typical power characteristics according to the power direction of the typical power characteristics, and determining the first power characteristic and the second power characteristic.

[0055] The classification result includes a first power feature and a second power feature.

[0056] It should be noted that when an on event occurs in the electrical equipment, the power of the electrical equipment is connected, and when an off event occurs in the electrical equipment, the power of the electrical equipment is cut off. Therefore, when the performance of the electrical equipment is normal, the power fluctuations generated when the electrical equipment occurs an on event and an off event are in opposite directions and the absolute values ​​of the fluctuations are similar.

[0057] In this example, according to the power direction of the typical power characteristics, all typical power characteristics are divided into a first power characteristic corresponding to an on event and a second power characteristic corresponding to a off event, so that in subsequent steps, according to the combination of the first power characteristic and the second power characteristic, it is determined which specific power types the electrical equipment used in the power grid to be identified within the preset sampling period belongs to.

[0058] Exemplarily, based on the calculation method of the power feature in the above embodiment, that is, the power corresponding to the next sampling time point minus the power corresponding to the previous sampling time point, the power direction of the power feature corresponding to the on event in this embodiment should be positive, and the power direction of the power feature corresponding to the off event should be negative. Therefore, this embodiment divides the typical power feature into a first power feature with a positive power direction and a second power feature with a negative power direction according to the positive and negative power directions of the typical power feature.

[0059] Furthermore, as a refinement and extension of the specific implementation methods of the above-mentioned embodiments, in order to fully illustrate the specific implementation process of this embodiment, the steps of determining the type information of electrical equipment in the power grid to be identified based on the classification results specifically include: calculating the absolute value difference between the first power feature and the second power feature; determining the first power feature and the second power feature corresponding to the absolute value difference less than the second preset threshold as a power feature group; determining the power type to which the electrical equipment belongs based on the power feature group; and determining the number of different power types in the power grid to be identified based on the number of power feature groups.

[0060] In this embodiment, the first power feature and the second power feature whose absolute value difference is less than the second preset threshold are determined as a power feature group, so that the first power feature and the second power feature with similar absolute values ​​of power fluctuation values ​​are determined as power features corresponding to on events and off events of electrical equipment of the same power type, thereby determining the power type of the electrical equipment in the to-be-identified power grid and the number of power types according to the power feature group.

[0061] Exemplarily, the absolute value difference between the first power feature and the second power feature is calculated, and the power feature group is determined according to the first power feature and the second power feature corresponding to the absolute value difference less than the second preset threshold, so that the first power feature and the second power feature are combined into a plurality of power feature groups. Wherein, any power feature group represents a power type, therefore, the number of power feature groups is the number of power types, the first power feature in the power feature group represents the power feature corresponding to the on event of the power-consuming device of the power type, and the second power feature in the power feature group represents the power feature corresponding to the off event of the power-consuming device of the power type. Since when the electrical appliance has an on event, the power of the electrical appliance is turned on, that is, the power changes from 0 to the operating power, therefore, the specific power of the electrical appliance of the power type when it is running can be determined by the first power feature in the power feature group. Similarly, the specific power of the electrical appliance of the power type when it is running can also be determined according to the second power feature in the power feature group. Thus, the type information of the electrical equipment of the power grid to be identified can be determined according to the power feature group, such as dividing the electrical equipment used in the power grid to be identified within the preset sampling period into electrical appliances of several power types, such as low power, medium power, high power, and ultra-high power.

[0062] For example, in the clustering processing step, after obtaining 2K typical power features, the typical power features can be divided into k power feature groups according to the above rules in this embodiment, thereby dividing the multiple electrical devices used in the power grid to be identified within the preset sampling period into K power types of electrical appliances.

[0063] It should be noted that both the first power feature and the second power feature include an active power feature part and a reactive power feature part. According to the principle of electricity, when an electrical equipment is switched on or off, the reactive power of the electrical equipment will change accordingly with the active power. Therefore, when calculating the absolute value difference between the first power feature and the second power feature, the absolute value difference corresponding to the reactive power feature part and the absolute value difference corresponding to the active power feature part in the same power feature group are both less than the second preset threshold.

[0064] In one embodiment, the non-intrusive load monitoring method also includes: matching the first power feature in the power feature group with the power feature corresponding to the switching event, and determining the switching event matching the first power feature in the power feature group as an on event; wherein the on event is used to indicate that the power of the electrical equipment is turned on; matching the second power feature in the power feature group with the power feature corresponding to the switching event, and determining the switching event matching the second power feature in the power feature group as an off event; wherein the off event is used to indicate that the power of the electrical equipment is cut off; matching the on event and the off event according to the event time points at which the on event and the off event occur, and determining the usage information of the electrical equipment that meets the power type corresponding to the power feature group according to the matching results.

[0065] In this embodiment, after obtaining the power fluctuations corresponding to the on and off events of various power types of electrical equipment through the power feature group, the switching events corresponding to the various power types of electrical equipment are distinguished among all switching events, thereby determining the usage of the various power types of electrical equipment within a preset sampling period, providing macro power consumption information, and facilitating the analysis of the user's power consumption patterns.

[0066] Exemplarily, for any power type of electrical equipment, the first power feature in the power feature group corresponding to the power type is matched with the power feature corresponding to the switching event to obtain multiple on events that occurred in the electrical equipment of the power type. Similarly, the second power feature in the power feature group corresponding to the power type is matched with the power feature corresponding to the switching event to obtain multiple off events that occurred in the electrical equipment of the power type. Thus, the information of the first use of the electrical equipment of the power type can be determined through the first on event and the first off event that occurred in the electrical equipment of the power type, and then the information of several uses of the electrical equipment of the power type can be determined in sequence. Furthermore, the usage of electrical equipment of all power types within a preset sampling period can be obtained to analyze the power usage pattern.

[0067] It is worth mentioning that the total power consumption at a certain moment is determined by the working state of each electrical device at that moment and the corresponding power. To simplify the description, assume that there are M electrical devices under the total electric meter and each electrical device has only two simple states: on and off, and the power consumed in the on state is constant, then the total power measured by the electric meter at time t can be expressed as:

[0068]

[0069] where a m (t) represents the state of the electrical device m at time t. If the electrical device m is in the on state at time t, then a m (t)=1, otherwise a m (t) = 0. m It represents the power of the electrical equipment m in the on state. e(t) represents the error amount.

[0070] After the process of the above embodiment, it can be obtained that the M electrical devices include several types of electrical appliances, and the power of each type of electrical appliance in the on state can be obtained. Further, with the constraint that the sum of the number of electrical appliances of each power type is equal to M, a function between the number of electrical appliances of each power type, the power of each type of electrical appliance in the on state and the total power is constructed, and modeling optimization is performed. When the error at each moment is minimized, the number of electrical appliances of each power type is determined.

[0071] In one embodiment, an AMPds2 dataset is selected for research, which contains complete usage data of electricity, water and natural gas for a household in Burnaby for two years with a granularity of 1 minute. This embodiment selects the total active power and reactive power of 10080 minutes in one week in 2013 from the dataset for experiment.

[0072] Specifically, since the switching events of electrical appliances in residential electricity consumption occur frequently, the sliding window length is selected as 3. The first preset threshold is selected as 200, and the window power parameter is calculated. When the window power parameter is less than 200, the load does not change much, and it is determined that no event has occurred. A total of 2495 switching events were detected within one week, and the power features corresponding to the switching events were extracted.

[0073] Furthermore, the power features are clustered. The cluster center value is set to 18. The clustering results are as follows: Figure 3 shown.

[0074] Furthermore, according to the typical power features obtained by clustering, it is determined that the data set contains several types of electrical appliances of different power levels, and the usage information of each type of electrical appliance is determined. According to the usage information of each type of electrical appliance, the prediction information of the total power consumption of the week is determined, such as Figure 4 shown. Figure 4 The blue curve is the actual active power curve of the day, and the red curve is the prediction information calculated by this embodiment. It can be seen that the prediction information calculated by this embodiment is very close to the actual active power curve, which verifies the reliability of this embodiment.

[0075] It should be noted that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0076] Furthermore, if Figure 2 As shown, as a specific implementation of the above-mentioned non-intrusive load monitoring method, an embodiment of the present application provides a non-intrusive load monitoring device 200, and the non-intrusive load monitoring device 200 includes: a determination module 201.

[0077] Among them, the determination module 201 is used to determine the event time point of the switching event of the power equipment in the power grid to be identified and the power characteristics corresponding to the switching event according to the total power sequence of the power grid to be identified; wherein the power grid to be identified includes multiple power equipment, the total power sequence includes the total power consumption of multiple power equipment at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the power equipment is connected or cut off; and, clustering the power characteristics to determine the typical power characteristics of the power grid to be identified; and, classifying the typical power characteristics to determine the type information of the power equipment of the power grid to be identified according to the classification results

[0078] In one embodiment, the determination module 201 is specifically used to slide a window of a preset length on the total power sequence to determine the window power parameter; and determine the event time point when the switching event of the electric equipment occurs according to the window power parameter.

[0079] In one embodiment, the determination module 201 is specifically used to calculate the power difference between the total powers corresponding to adjacent sampling time points within the window; sort the power differences to determine the difference order; and determine the window power parameter based on the power difference that is ranked first in the difference order.

[0080] In one embodiment, the determination module 201 is specifically used to determine that a switching event of an electrical device occurs in the window if the window power parameter is greater than a first preset threshold value, and sort the sampling time points associated with the power difference values ​​corresponding to the window power parameters greater than the first preset threshold value to determine the time sequence; determine the event time point based on the sampling time point that is first in the time sequence.

[0081] In one embodiment, the determination module 201 is specifically used to determine the active power characteristics of the switching event based on the difference between the total power corresponding to the event time point and the next sampling time point adjacent to the event time point in the total power sequence; determine the reactive power characteristics of the switching event based on the difference between the reactive power corresponding to the event time point and the next sampling time point adjacent to the event time point; and use the active power characteristics and the reactive power characteristics as the power characteristics corresponding to the switching event.

[0082] In one embodiment, the determination module 201 is specifically used to determine a data point set according to a power feature; randomly select any data point in the data point set as the first initial cluster center; use the data points in the data point set excluding the initial cluster center as candidate data points; sort the distances between the candidate data points and the initial cluster center to determine a distance order; use the distance at the end of the distance order as the target distance of the candidate data point; determine the probability of the candidate data point being the next initial cluster center according to the target distance; determine the next initial cluster center according to the probability of the candidate data point being the next initial cluster center, and update the candidate data points until the number of initial cluster centers is equal to a preset number; determine a preset number of clusters according to the initial cluster centers; and determine a typical power feature according to the power features of the cluster centers in the clusters.

[0083] In one embodiment, the determination module 201 is specifically configured to classify the typical power features according to their power directions to determine the first power feature and the second power feature, wherein the classification result includes the first power feature and the second power feature.

[0084] In one embodiment, the determination module 201 is specifically used to calculate the absolute value difference between the first power feature and the second power feature; determine the first power feature and the second power feature corresponding to the absolute value difference less than the second preset threshold as a power feature group; determine the power type of the electrical equipment according to the power feature group; and determine the number of different power types in the power grid to be identified according to the number of power feature groups.

[0085] In one embodiment, the non-intrusive load monitoring device further comprises:

[0086] A matching module is used to match a first power feature in a power feature group with a power feature corresponding to a switching event, and determine the switching event matching the first power feature in the power feature group as an on event; wherein the on event is used to indicate that the power of the electrical equipment is turned on; match a second power feature in the power feature group with a power feature corresponding to the switching event, and determine the switching event matching the second power feature in the power feature group as an off event; wherein the off event is used to indicate that the power of the electrical equipment is cut off; match the on event and the off event according to the event time points at which the on event and the off event occur, and determine the usage information of the electrical equipment that meets the power type corresponding to the power feature group according to the matching results.

[0087] For the specific definition of the non-intrusive load monitoring device, please refer to the definition of the non-intrusive load monitoring method above, which will not be repeated here. Each module in the above non-intrusive load monitoring device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0088] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform, or the embodiments of the present application can be implemented by hardware.

[0089] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple submodules.

[0090] The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be thought of by technicians in this field should fall within the scope of protection of this application.

Claims

1. A non-intrusive load monitoring method, characterized in that: The method comprises: According to the total power sequence of the power grid to be identified, determine the event time point at which a switching event occurs for an electric device in the power grid to be identified and the power characteristics corresponding to the switching event; wherein the power grid to be identified includes a plurality of the electric devices, the total power sequence includes the total power consumption of the plurality of the electric devices at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the electric device is connected or disconnected; Performing clustering processing on the power characteristics to determine typical power characteristics of the power grid to be identified; The typical power characteristics are classified and processed, and the type information of the power consumption equipment of the power grid to be identified is determined according to the classification result.

2. The non-intrusive load monitoring method according to claim 1, characterized in that: The step of determining the event time point at which the switching event of the electric equipment in the power grid to be identified occurs according to the total power sequence of the power grid to be identified includes: Sliding a window of a preset length on the total power sequence to determine a window power parameter; An event time point at which a switching event occurs on the electric equipment is determined according to the window power parameter.

3. The non-intrusive load monitoring method according to claim 2, characterized in that: The sliding of a window of a preset length on the total power sequence to determine a window power parameter includes: Calculating a power difference between total powers corresponding to adjacent sampling time points within the window; Sorting the power differences to determine a difference order; The window power parameter is determined according to the power difference value that is at the first position in the difference sequence.

4. The non-intrusive load monitoring method according to claim 3, characterized in that: The step of determining the event time point at which the switching event of the electric equipment occurs according to the window power parameter includes: If the window power parameter is greater than a first preset threshold, it is determined that a switching event of the electric device occurs in the window, and the sampling time points associated with the power difference values ​​corresponding to the window power parameter greater than the first preset threshold are sorted to determine the time sequence; The event time point is determined according to the sampling time point that is first in the time sequence.

5. The non-intrusive load monitoring method according to claim 1, characterized in that: The determining, according to the total power sequence of the power grid to be identified, the power feature corresponding to the switching event includes: Determining the active power feature of the switching event according to the difference between the total power corresponding to the event time point and the next sampling time point adjacent to the event time point in the total power sequence; Determining the reactive power characteristic of the switching event according to the difference between the reactive power corresponding to the event time point and the next sampling time point adjacent to the event time point; The active power feature and the reactive power feature are used as power features corresponding to the switching event.

6. The non-intrusive load monitoring method according to claim 1, characterized in that: The clustering process of the power features to determine typical power features includes: determining a set of data points based on the power characteristics; Randomly select any data point in the data point set as the first initial cluster center; The data points in the data point set excluding the data points at the initial cluster center are taken as candidate data points; Sorting the distances between the candidate data points and the initial cluster centers to determine a distance order; The distance at the end of the distance sequence is used as the target distance of the candidate data point; Determine, according to the target distance, the probability of the candidate data point being the next initial cluster center; Determine the next initial cluster center according to the probability of the candidate data point being the next initial cluster center, and update the candidate data points until the number of the initial cluster centers is equal to a preset number; Determining the preset number of clusters according to the initial cluster centers; The typical power feature is determined according to the power features of the cluster centers in the clusters.

7. The non-intrusive load monitoring method according to claim 1, characterized in that: The classifying and processing the typical power characteristics includes: The typical power features are classified according to their power directions to determine a first power feature and a second power feature, wherein the classification result includes the first power feature and the second power feature.

8. The non-intrusive load monitoring method according to claim 7, characterized in that: The step of determining the type information of the electrical equipment of the to-be-identified power grid according to the classification result includes: calculating an absolute value difference between the first power characteristic and the second power characteristic; Determine the first power feature and the second power feature corresponding to the absolute value difference less than a second preset threshold as a power feature group; Determining the power type of the electrical equipment according to the power feature group; The number of different power types in the to-be-identified power grid is determined according to the number of the power feature groups.

9. The non-intrusive load monitoring method according to claim 8, characterized in that: The method further comprises: Matching the first power feature in the power feature group with the power feature corresponding to the switching event, and determining the switching event that matches the first power feature in the power feature group as an on event; wherein the on event is used to indicate that the power of the electrical device is turned on; Matching the second power feature in the power feature group with the power feature corresponding to the switching event, and determining the switching event matching the second power feature in the power feature group as a shutdown event; wherein the shutdown event is used to indicate that the power of the electrical device is cut off; The on event and the off event are matched according to the event time points at which the on event and the off event occur, and usage information of the electrical equipment that meets the power type corresponding to the power feature group is determined according to the matching result.

10. A non-intrusive load monitoring device, characterized in that: The device comprises: A determination module, configured to determine, based on the total power sequence of the power grid to be identified, an event time point at which a switching event occurs for an electric device in the power grid to be identified and a power feature corresponding to the switching event; wherein the power grid to be identified includes a plurality of the electric devices, the total power sequence includes the total power consumption of the plurality of the electric devices at each sampling time point within a preset sampling period, and the switching event is used to indicate that the power of the electric device is connected or disconnected; and, Performing clustering processing on the power characteristics to determine typical power characteristics of the power grid to be identified; as well as, The typical power characteristics are classified and processed, and the type information of the power consumption equipment of the power grid to be identified is determined according to the classification result.

Images