Non-invasive real-time load monitoring method and storage medium

Abstract

This invention provides a non-intrusive real-time load monitoring method and storage medium. By analyzing corresponding power consumption patterns, it extracts load feature libraries for the on / off states of various electrical devices and generates a load feature library for combinations of two electrical devices. It extracts a certain time-length segment of total load patterns from real-time data, filters them based on their relative values, and calculates the distance between each segment and a standard load pattern. If the distance is within a certain threshold, the collected pattern segment is considered to match the device. This invention can achieve real-time determination of different electrical device on / off events, enabling rapid monitoring of these events with quick response; it also requires minimal investment and is highly practical.

Description

Technical Field

[0001] This invention relates to the field of electrical technology, and in particular to a non-intrusive real-time load monitoring method and storage medium. Background Technology

[0002] Smart grids represent the future direction of global power grid development, and smart electricity consumption is a crucial component of smart grids. Key technologies primarily lie in Advanced Metering Infrastructure (AMI) standards, systems, and terminal technologies; smart electricity consumption's two-way interactive operation mode and supporting technologies; and the interaction between user electricity consumption environment and consumption patterns. Load monitoring, as one of the most important components of AMI, is the first step in realizing smart grids.

[0003] Load monitoring provides insights into the specific operational status and energy consumption of residential users' electrical equipment, offering comprehensive information on household appliance power consumption and promoting potential energy savings. Furthermore, the availability of electricity consumption information offers significant benefits for power system operation scheduling and grid planning. Therefore, it can improve asset utilization and energy efficiency. Traditional load monitoring employs an intrusive approach, installing sensors on all electrical equipment to record usage. This method suffers from poor practicality, high implementation costs, high economic costs, and privacy concerns. In contrast, non-intrusive load monitoring requires less investment and is highly practical.

[0004] Therefore, this invention proposes a non-intrusive load monitoring method to extract load characteristics of different electrical appliances from electrical load information, so as to monitor electrical events and determine whether the appliances are turned on or off. Summary of the Invention

[0005] The purpose of this invention is to provide a non-intrusive real-time load monitoring method to solve the problems of poor practical operability, high implementation cost, high economic cost, and user privacy issues existing intrusive load monitoring technologies.

[0006] To achieve the above objectives, the present invention provides a non-intrusive real-time load monitoring method, comprising: an acquisition step, acquiring standard load imprints of multiple electrical devices, wherein the standard load imprint is a waveform of power consumption change of an electrical device during a period of being turned on or off; a feature library establishment step, sorting and storing the standard load imprints of the electrical devices in a load feature library; a first acquisition step, acquiring the real-time load imprint of a user during a specific period, wherein the real-time load imprint is a waveform of power consumption change of the user during the specific period; a calculation step, sequentially calculating the matching degree between each electrical device in the load feature library and the user, wherein the matching degree is the distance value between the standard load imprint of each electrical device and the real-time load imprint; and a judgment step, if the matching degree between an electrical device and the user is less than a threshold, then determining that the user is matched with the electrical device during the specific period.

[0007] Furthermore, the acquisition step specifically includes: a measurement step, which measures the load signature of each electrical device, which is the power change waveform of the electrical device; and a search step, which extracts the power change waveform of each electrical device during the period when it is started or turned off from the measured load signature, and uses it as the standard load signature.

[0008] Furthermore, in the search step, the standard load information includes power change waveforms for three consecutive time periods, namely, a stable power waveform for the first time period, a power abrupt change waveform for the second time period, and a stable power waveform for the third time period.

[0009] Furthermore, in the search step, each instantaneous power value in the standard load imprint is subtracted from an initial value; the initial value is the initial data value of the monitoring equipment used to monitor each electrical device.

[0010] Furthermore, in the feature library establishment step, the sorting rule is arranged from longest to shortest according to the time length of the standard load imprint.

[0011] Furthermore, after the first acquisition step, the method further includes: a first determination step, which determines whether the difference between the maximum and minimum instantaneous power in the real-time load imprint is greater than a threshold; if yes, proceed to the next step; if no, return to the first acquisition step.

[0012] Furthermore, if the duration of the specific time period is longer than the duration of the period when the load is turned on or off, then after the first acquisition step, the method further includes: a truncation step, which truncates a load imprint segment from the real-time load imprint, wherein the time period of the segment is the same as that of the standard load imprint; and an update step, which updates the real-time load imprint to the load imprint segment.

[0013] Furthermore, after the interception step, the method further includes: a second determination step, wherein the load imprint segment includes a continuous first time period, a second time period, and a third time period, determining whether a stable waveform of one segment is located in the first time period of the load imprint segment, and determining whether a stable waveform of another segment is located in the third time period of the load imprint segment; if yes, proceed to the next step; if no, return to the calculation step and perform matching calculation with the next standard load imprint.

[0014] Furthermore, in the calculation step, the distance value includes Euclidean distance or dynamic time-normalized distance.

[0015] Furthermore, in the feature library establishment step, the load feature library includes a load feature library for single electrical appliances and a load feature library for mixed electrical appliances; the load feature library for single electrical appliances is formed by sorting and storing the load imprints of individual electrical devices into the load feature library.

[0016] Further, the hybrid electrical load feature library is obtained through the following processing steps: A second acquisition step, acquiring the standard load imprints of the first and second electrical devices, where the electrical types of the first and second electrical devices are different; a first overlay step, overlaying the standard load imprints of the first and second electrical devices and storing them in the hybrid electrical load feature library (the overlay method is based on overlaying at the same time); a second overlay step, shifting the start time of the standard load imprint of the first electrical device backward by a time interval ∆t, updating the start time of the standard load imprint of the first electrical device, and then overlaying it again with the standard load imprint of the second electrical device and storing it in the hybrid electrical load feature library; a loop step, repeating the second overlay step, the number of repetitions being int(max(la,lb)-min(la,lb)) / ∆t), where la is the time length of the standard load imprint of the first electrical device, and lb is the time length of the standard load imprint of the second electrical device.

[0017] The present invention provides a storage medium storing computer-readable instructions that, when read by at least one processor, cause the at least one processor to perform at least one step in the non-intrusive real-time load monitoring method.

[0018] The beneficial effects of this invention are as follows: This invention provides a non-intrusive real-time load monitoring method and storage medium. By analyzing the corresponding power consumption records, it extracts the on / off load feature databases of each electrical device and generates a load feature database of two combined electrical devices. It extracts a certain time-length segment of total load records from real-time data, and after filtering based on their relative values, calculates the distance between each segment and a standard load record. If the distance is within a certain threshold, the collected record segment is determined to match the device. This invention can achieve real-time determination of different electrical device on / off events, enabling rapid monitoring of these events with quick response; simultaneously, it requires minimal investment and is highly practical. Attached Figure Description

[0019] The present invention will now be further described with reference to the accompanying drawings and embodiments.

[0020] Figure 1 A flowchart of the non-intrusive real-time load monitoring method provided by the present invention;

[0021] Figure 2 A flowchart of the acquisition steps provided by the present invention;

[0022] Figure 3 A flowchart for establishing a hybrid electrical load feature library provided by the present invention;

[0023] Figure 4 A graph showing the load imprint of a refrigerator being turned on, provided as an embodiment of the present invention.

[0024] Figure 5 This is a graph showing the superposition of microwave oven opening marks and refrigerator opening marks provided in the data embodiment of the present invention;

[0025] Figure 6 A graph showing the opening marks of a microwave oven and a refrigerator, provided for a data embodiment of the present invention;

[0026] Figure 7 The graph is a cut-out of the imprint provided in the data embodiment of the present invention. Detailed Implementation

[0027] To better understand the content of this invention, specific embodiments are provided below to further illustrate the invention, but the implementation and protection scope of this invention are not limited thereto.

[0028] The following description of the embodiments is made with reference to the accompanying drawings, which illustrate specific embodiments in which the invention can be implemented.

[0029] This invention provides a non-intrusive real-time load monitoring method. This method monitors and determines electrical appliance events from power load imprint (power) curves. By analyzing the corresponding electrical power quantities, it extracts the load characteristic databases of each electrical appliance's on / off state and generates a load characteristic database of combined loads of two appliances. It then extracts load imprints for a specific time period from real-time data, calculates the distance (degree of difference) between these imprints and standard load imprints, and determines whether the two match.

[0030] like Figure 1 As shown, the non-intrusive real-time load monitoring method includes the following steps S1 to S9.

[0031] S1. Acquisition Step: Acquire standard load signatures for multiple electrical devices. The standard load signature is the power consumption change waveform of an electrical device during the period it is turned on or off. The length of the standard load signature for each appliance may be different. The standard load imprint length for opening and closing is denoted as and .

[0032] like Figure 2 As shown, the acquisition steps specifically include S101~S102.

[0033] S101. Measurement steps: Measure the load signature of each electrical device, which is the power change waveform of the electrical device.

[0034] S102, Searching step: Extract the power change waveform of each electrical device during the period when it is started or turned off from the measured load imprint, and use it as the standard load imprint.

[0035] In the search step, each instantaneous power value in the standard load imprint is subtracted from the initial value; the initial value is the data initial value of the monitoring equipment used to monitor each electrical device.

[0036] In the search step, the standard load information includes power change waveforms for three consecutive time periods: a stable power waveform in the first time period, a sudden power change waveform in the second time period, and a stable power waveform in the third time period. That is, the standard load imprint format is: off-peak - sudden change - off-peak, with each of the first and third time periods lasting 0.2 seconds.

[0037] S2. Feature library establishment step: Sort and store the standard load imprints of the electrical equipment into a load feature library. In the feature library establishment step, the sorting rule is based on the time length of the standard load imprints from longest to shortest.

[0038] In the feature library establishment step, the load feature library includes a load feature library for single electrical appliances and a load feature library for mixed electrical appliances.

[0039] The load feature library for a single electrical appliance is formed by storing the load imprints of individual electrical devices in a sorted manner.

[0040]

[0041]

[0042] in, A load characteristic library for individual electrical devices; and Electrical equipment Standard load markings for opening and closing; A collection of all electrical appliances; Record each moment within the time period Corresponding power value .

[0043] like Figure 3 As shown, the hybrid electrical load feature library is obtained through the following processing steps S201 to S204.

[0044] S201, Second acquisition step: Acquire the standard load imprints of the first electrical device and the second electrical device, wherein the electrical types of the first electrical device and the second electrical device are different.

[0045] Specifically, and Standard load markings for the on / off states of two different electrical devices.

[0046] S202, First superposition step: Superposition of the standard load imprint of the first electrical equipment and the standard load imprint of the second electrical equipment into the hybrid electrical load feature library (the superposition method is to superimpose based on the same time).

[0047] S203, Second superposition step: Shift the start time of the standard load imprint of the first electrical device backward by a time interval ∆t, update the start time of the standard load imprint of the first electrical device, and then superimpose it with the standard load imprint of the second electrical device again and store it in the hybrid electrical load feature library.

[0048] S204, Repeat the second superposition step, repeating it a certain number of times. Where, la is the time length of the standard load imprint of the first electrical equipment, and lb is the time length of the standard load imprint of the second electrical equipment.

[0049] Finally, the following load characteristic library of hybrid electrical appliances was obtained.

[0050]

[0051]

[0052] in, For the load characteristic library of hybrid electrical appliances; for two electrical appliances A set of superimposed combinations of devices being turned on or off; For combination The superimposed imprints of all different misaligned positions; similarly, For each moment within the time period The corresponding superimposed power value .

[0053] S3. First data acquisition step: Acquire a user's real-time load profile over a specific time period. The real-time load profile is the waveform of the user's power consumption change during that specific time period. The acquired load profile power must be initialized to zero.

[0054] If the duration of the specific time period is greater than the duration of the period when the device is turned on or off, then after the first acquisition step, the method further includes: overlaying.

[0055] S4. First determination step: Determine whether the difference between the maximum and minimum instantaneous power values ​​in the real-time load imprint is greater than a threshold; if yes, proceed to the next step; otherwise, return to the first acquisition step. This step is used to reduce computational load and reduce the false judgment rate.

[0056] S5. Extraction Step: Extract a segment of the load imprint from the real-time load imprint, denoted as data s. The time period of this segment is the same as that of the standard load imprint; its duration is... It needs to be longer than the standard load imprint length of all feature libraries.

[0057] S6. Second determination step: The load imprint segment includes a continuous first time period, a second time period, and a third time period. Determine whether the stable waveform of one segment is located in the first time period of the load imprint segment, and determine whether the stable waveform of another segment is located in the third time period of the load imprint segment. If yes, proceed to the next step; if no, return to the calculation step and perform matching calculation with the next standard load imprint.

[0058] S7. Update step: Update the real-time load imprint to a load imprint fragment, denoted as... .

[0059] S8. Calculation steps: Calculate the matching degree between each electrical device in the load feature database and the user in sequence. The matching degree is the distance between the standard load imprint of each electrical device and the real-time load imprint.

[0060] In the calculation step, the distance value includes Euclidean distance or dynamic time-normalized distance.

[0061] The matching method used in this example for electrical appliance events is Dynamic Time Warping (DTW). In time series analysis, DTW is one of the algorithms used to measure the similarity between two time series, and the speed at which they match may vary. During each matching process, the DTW distance between the standard sequence and the truncated sequence is calculated. If the distance is less than a certain threshold, the sequences are considered a match.

[0062] Euclidean distance, also known as Euclidean distance, is the most common distance metric, measuring the absolute distance between two points in multidimensional space. It can also be understood as the true distance between two points in m-dimensional space, or the natural length of a vector (i.e., the distance from that point to the origin). In two-dimensional and three-dimensional space, Euclidean distance is the actual distance between two points. The specific calculation formula is as follows:

[0063] .

[0064] S9. Judgment step: If the matching degree between an electrical device and the user is less than a threshold, then it is determined that the user is matched with the electrical device in the specific time period.

[0065] The threshold is determined by: using historical data, calculating each standard load imprint. Corresponding threshold The specific method is as follows: For each moment in the historical data, a truncated length of [length missing] is extracted. A piece of data, calculate its relationship with The distance value. Adjust the threshold. This makes all electrical equipment The information on whether the load is turned on or off can be correctly filtered.

[0066] The present invention provides a storage medium storing computer-readable instructions that, when read by at least one processor, cause the at least one processor to perform at least one step in the non-intrusive real-time load monitoring method.

[0067] This invention provides a non-intrusive real-time load monitoring method and storage medium. By analyzing the corresponding power consumption marks, the method extracts the load feature databases of each electrical device's on / off state and generates a load feature database of two electrical device combinations. The method extracts the total load marks of a certain time length from the real-time data, and after judging and filtering their relative values, it calculates the distance between them and the standard load marks one by one. If the distance is within a certain threshold, it is determined that the collected mark segment matches the device.

[0068] In summary, this method can achieve real-time determination of different electrical equipment opening and closing events, and can quickly monitor the opening and closing events of electrical equipment based on real-time data with a rapid response; at the same time, it requires little economic investment and is highly practical.

[0069] The present invention also provides a data embodiment to support the non-intrusive real-time load monitoring method of the present invention.

[0070] The first step is to obtain standard load signatures. This involves extracting the load signatures for the start-up and shutdown of individual electrical devices. The load signature used in this method represents the power. The electrical device... The standard load imprint length for opening and closing is denoted as and .

[0071] The second step includes a first sub-step and a second sub-step. The first sub-step is as follows: Prepare a single electrical load characteristic library. Sort the standard load imprints of each electrical device during startup and shutdown according to the length of the load imprints from longest to shortest, and record their labels. This constitutes the load characteristic library.

[0072] In this embodiment,

[0073] ;

[0074] ;

[0075] in, This is a collection of all electrical appliances (m-microwave oven, c-rice cooker, f-refrigerator). A library of load characteristics for individual electrical appliances. The order of arrangement is based on the length of the imprints, from longest to shortest; and Electrical appliances The standard load markings for opening and closing. The lengths of the middle imprints are: 25, 18, 12, 12, 8, 8.

[0076] If the refrigerator's standard load rating is: , Figure 4The corresponding power curve is shown, where the load imprint sampling rate is 10Hz, i.e., the time axis unit is 0.1s, and the power... The value is in units of 0.1w.

[0077] The second sub-step: Prepare a mixed electrical load feature library. Actual electrical load imprints may show situations where two different electrical devices are simultaneously turned on or off within a short period. This method only considers the simultaneous on / off of two devices and cannot match them to the on / off of a single device. This feature library is used to identify mixed on / off situations of electrical devices.

[0078] and Standard load imprints for the on / off switching of two different electrical devices. Based on the smallest time unit. By continuously changing the misalignment length, and By performing staggered stacking, a set of stacked products produces a quantity of... The mixed standard load imprints. The mixed standard load imprints generated by combinations of non-repeating electrical equipment constitute the mixed electrical load characteristic library.

[0079]

[0080]

[0081]

[0082]

[0083] in, A load characteristic library for hybrid electrical appliances; This is a set of superimposed combinations of two electrical appliances being turned on or off (m-microwave oven, c-rice cooker, f-refrigerator; s-start, e-end). For combination The superimposed imprints of all different misaligned positions; similarly, For each moment within the time period The corresponding superimposed power value .

[0084] Example: One example of data from the group where microwave oven opening marks and refrigerator opening marks are superimposed: . Figure 5 For the corresponding curve image, Figure 6 For the load imprints of two separate devices used for overlay.

[0085] The third step involves monitoring real-time power data. At each moment, the longest time frame is extracted. The total load information, denoted as this data segment. , It includes the power value for each moment within the time period. In this example... .

[0086] Example: Extract the following data:

[0087] , Figure 7 This is the corresponding curve image.

[0088] Step 4, make a judgment Is the difference between the maximum and minimum values ​​greater than 100w? Initial screening is performed to reduce computational load and false positive rate. If If a candidate is retained after the initial screening, proceed to the next step; otherwise, skip to the next step.

[0089] Continue collecting data immediately.

[0090] In this example: The difference between the maximum and minimum values ​​is 13481, or 1348.1w, which is greater than the threshold of 100w, so proceed to the fifth step.

[0091] Fifth step, regarding the imprint If the electrical equipment is already turned on, its on / off mark will not be matched; conversely, the same applies if it is not turned on. Because... The length may differ from the standard load mark. Perform a second truncation. Calculate. When calculating the matching degree, truncate The length of the front end is The load imprint segment is denoted as .

[0092] Step 6: Judgment Whether the first time period at the very beginning and the third time period at the very end are stable waveforms, in order to avoid truncating them as part of a long load imprint.

[0093] The specific judgment method is as follows: Is the difference between the maximum and minimum power values ​​within 0.2 seconds at the very beginning and the very end less than 5W? If a sample is retained after screening, proceed to the next step; if it fails screening, proceed to the next standard load imprint.

[0094] Step 7: Calculate the distance value. After secondary filtering, a specific method is used to calculate the distance. and The distance value measures the difference between the target segment and the standard imprint.

[0095] Step 8: Judgment step. If the distance value is less than a certain threshold, then a judgment is made. This refers to the device's startup or shutdown event, i.e., matching the two.

[0096] The matching method used in this example for electrical appliance events is Dynamic Time Warping (DTW). In time series analysis, DTW is one of the algorithms used to measure the similarity between two time series, and the speed at which they match may vary. During each matching process, the DTW distance between the standard sequence and the truncated sequence is calculated. If the distance is less than a certain threshold, the sequences are considered a match.

[0097] In this embodiment:

[0098] for :

[0099] Secondary screening: The front and back ends are in a steady state.

[0100] Distance calculation: ,because Therefore, the next load imprint is determined.

[0101] for :

[0102] Secondary screening: The front and back ends are in a steady state.

[0103] Distance calculation: ,because Therefore, the next load imprint is determined.

[0104] for :

[0105] Secondary screening: The front and back ends are in a steady state.

[0106] Distance calculation: ,because Therefore, the next load imprint is determined.

[0107] for :

[0108] Secondary screening: The front and back ends are in a steady state.

[0109] Distance calculation: ,because The value is less than the threshold, so a refrigerator start-up event is determined to have occurred.

[0110] This invention provides a non-intrusive real-time load monitoring method and storage medium. By analyzing corresponding power consumption patterns, it extracts load feature libraries for the on / off states of various electrical devices and generates a load feature library for combinations of two electrical devices. It extracts a certain time-length segment of total load patterns from real-time data, filters them based on their relative values, and calculates the distance between each segment and a standard load pattern. If the distance is within a certain threshold, the collected pattern segment is considered to match the device. This invention can achieve real-time determination of different electrical device on / off events, enabling rapid monitoring of these events with quick response; it also requires minimal investment and is highly practical.

[0111] It should be noted that the present invention, as fully described, can have various modifications and variations, and is not limited to the specific embodiments described above. The above embodiments are merely illustrative and not intended to limit the scope of the invention. In short, the scope of protection of the present invention should include those modifications, substitutions, and alterations that are obvious to those skilled in the art.

Claims

1. A non-intrusive load monitoring method, characterized in that, include: The acquisition step involves acquiring standard load signatures for multiple electrical devices, where the standard load signature is a waveform representing the power consumption change of an electrical device during the period it is turned on or off. The acquisition steps specifically include: a measurement step, which measures the load signature of each electrical device, which is the power change waveform of the electrical device; and a search step, which extracts the power change waveform of each electrical device during the period when it is started or turned off from the measured load signature, as the standard load signature; in the search step, the standard load signature includes the power change waveforms of three consecutive time periods, namely the power stable waveform of the first time period, the power sudden change waveform of the second time period, and the power stable waveform of the third time period; The feature library establishment step involves sorting and storing the standard load imprints of the electrical equipment into a load feature library. The load feature library includes a single-device load feature library and a mixed-device load feature library. The single-device load feature library is formed by sorting and storing the load imprints of individual electrical equipment into the load feature library. The mixed-device load feature library is obtained through the following processing steps: a second acquisition step, acquiring the standard load imprints of a first electrical equipment and a second electrical equipment, where the electrical types of the first and second electrical equipment are different; a first overlay step, overlaying the standard load imprints of the first and second electrical equipment and storing them in the mixed-device load feature library; a second overlay step, shifting the start time of the standard load imprint of the first electrical equipment backward by a time interval ∆t, updating the start time of the standard load imprint of the first electrical equipment, and then overlaying it again with the standard load imprint of the second electrical equipment and storing it in the mixed-device load feature library; and a loop step, repeating the second overlay step, the number of repetitions being int(max(la,lb)-min(la,lb)). ) / ∆t), where la is the time length of the standard load imprint of the first electrical equipment, and lb is the time length of the standard load imprint of the second electrical equipment; The first data acquisition step involves acquiring a user's real-time load profile over a specific time period, where the real-time load profile is the waveform of the user's power consumption change during that specific time period. The first determination step is to determine whether the difference between the maximum and minimum instantaneous power values ​​in the real-time load imprint is greater than a threshold; if yes, proceed to the next step; if no, return to the first acquisition step. If the duration of the specific time period is longer than the duration of the period during which the device is turned on or off, then after the first data collection step, the method further includes: The extraction step involves extracting a segment of the load imprint from the real-time load imprint, the time period of which is the same as that of the standard load imprint. The second determination step involves determining whether a stable waveform of one segment is located within the first time period of the load imprint segment, and whether a stable waveform of another segment is located within the third time period of the load imprint segment. If so, proceed to the next step. The update step involves updating the real-time load imprint to a load imprint fragment; The calculation steps involve sequentially calculating the matching degree between each electrical device in the load feature database and the user, where the matching degree is the distance between the standard load imprint of each electrical device and the real-time load imprint. The judgment step determines whether the matching degree is less than a threshold. If so, it determines that the user is matched with the electrical equipment in the specific time period; otherwise, it returns to the calculation step to perform matching calculation with the next standard load imprint.

2. The non-intrusive real-time load monitoring method according to claim 1, characterized in that, In the search step, the initial value is subtracted from each instantaneous power value in the standard load imprint; The initial value is the initial data value of the monitoring equipment used to monitor each of the electrical devices.

3. The non-invasive real-time load monitoring method according to claim 1, characterized in that, In the feature library establishment step, the sorting rule is arranged from longest to shortest according to the time length of the standard load imprint.

4. The non-invasive real-time load monitoring method according to claim 1, characterized in that, In the calculation step, the distance value includes Euclidean distance or dynamic time-normalized distance.

5. A storage medium storing computer-readable instructions that, when read by at least one processor, cause at least one processor to perform at least one step in the non-intrusive real-time load monitoring method as described in any one of claims 1 to 4.

Images