An adaptive noise threshold event detection method, system, device, and medium

Abstract

The application discloses a kind of self-adapting noise threshold event detection method, system, equipment and medium, it is related to non-invasive load identification field, the method comprises: obtaining acquisition data sequence;According to acquisition data sequence, calculate active power sequence;Mean value window, event detection window and variance detection window are added to active power sequence;Mean value window and event detection window are moved, and self-adapting noise threshold is calculated once every time it moves, based on self-adapting noise threshold, determine whether there is event occurrence time, when detecting event occurrence time, variance detection window moves from event occurrence time, and the variance of power point in current variance detection window is calculated once every time it moves, and it is determined as steady state time node according to the variance and variance detection threshold value calculated;According to the power point before event occurrence time and the power point after steady state time node, feedback verification is carried out, and it is determined whether event occurrence time is event switching point.The application can improve the accuracy of event detection.

Description

Technical Field

[0001] This invention relates to the field of non-invasive load identification, and in particular to an adaptive noise threshold event detection method, system, device, and medium. Background Technology

[0002] To promote intelligent demand-side management of the power system, it is necessary to monitor the usage of each electrical device at the user end. Traditional intrusive identification methods require connecting a sensor to each electrical device at the user end, which consumes a lot of manpower and resources and has low applicability. Non-intrusive load monitoring technology does not require installing sensors on each device. It only needs to collect the total power consumption information at the user entrance and analyze the load characteristic information to achieve load identification.

[0003] In non-intrusive load identification technology, event detection is a crucial component. The accuracy of detecting event occurrence points and determining when the system enters a steady state directly impacts the accuracy of subsequent load identification. Existing event detection methods in the identification process employ a Cumulative Sum (CUSUM) detection algorithm. This algorithm requires pre-setting the noise figure and is prone to false detections for devices with large power fluctuations. Therefore, the accuracy of existing event detection methods needs improvement. Summary of the Invention

[0004] Based on this, embodiments of the present invention provide an adaptive noise threshold event detection method, system, device, and medium to improve the accuracy of event detection.

[0005] To achieve the above objectives, embodiments of the present invention provide the following solutions:

[0006] An adaptive noise threshold event detection method includes:

[0007] Acquire the data sequence; the data sequence includes a voltage sequence and a current sequence;

[0008] The active power sequence is calculated based on the collected data sequence; the active power sequence includes: multiple power points;

[0009] A first sliding window is added to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window;

[0010] The average value window and the event detection window are moved according to a first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When the event occurrence time is detected, the variance detection window moves from the event occurrence time according to the first set step size.

[0011] Each time the variance detection window is moved, the variance of the power point in the current variance detection window is calculated, and when the variance of the power point in the current variance detection window is less than the variance detection threshold, the time corresponding to the power point in the current variance detection window is determined as the steady-state time node.

[0012] Feedback verification is performed based on the power point before the event occurrence and the power point after the steady-state time node to determine whether the event occurrence is an event switching point;

[0013] The event detection and judgment process includes:

[0014] Calculate the current adaptive noise threshold based on the power points within the current average window;

[0015] Calculate the statistics of each power point within the current event detection window based on the current adaptive noise threshold, and determine whether there is an event occurrence time within the current event detection window based on the statistics.

[0016] Optionally, the current adaptive noise threshold is calculated based on the power points within the current average window, specifically including:

[0017] The mean power within the current average window is calculated based on the power points within the average window, thus obtaining the first average power.

[0018] The current adaptive noise threshold is calculated based on the maximum power point within the current average window, the minimum power point within the current average window, the first average power, and the noise level coefficient.

[0019] Optionally, the presence of an event occurrence time within the current event detection window can be determined based on statistics, specifically including:

[0020] If the statistic of the target power point is less than the time alarm value, then the detection delay of the previous power point is increased by 1 and determined as the detection delay of the target power point; the target power point is any power point within the current event detection window;

[0021] If the statistic of the target power point is greater than the time alarm value, the time of the event is determined based on the sequence number of the target power point and the detection delay of the previous power point.

[0022] If the statistics of all power points in the current event detection window are 0, then proceed to the next event detection window.

[0023] Optionally, feedback verification is performed based on the power point before the event occurrence and the power point after the steady-state time node to determine whether the event occurrence is an event switching point, specifically including:

[0024] The switching point judgment value is calculated based on the power point before the event occurrence and the power point after the steady-state time node; the calculation formula for the switching point judgment value is as follows:

[0025]

[0026] Where γ represents the switching point judgment value; μ0 represents the power mean of the power points before the event occurs; μ1 represents the power mean of the power points after the steady-state time node; P min This represents the minimum power point of the detected portion of the active power sequence;

[0027] If the cut-off point judgment value is 1, the time when the event occurs is determined to be the event cut-off point; if the cut-off point judgment value is 0, the time when the event occurs is determined not to be the event cut-off point.

[0028] Optionally, the current formula for calculating the adaptive noise threshold is:

[0029]

[0030] Where, p max p represents the maximum power point within the current average value window; min The minimum power point within the current average window is represented by ε; the noise level coefficient is represented by μ. p1 β represents the first mean power; β represents the current adaptive noise threshold.

[0031] Optionally, the collected data sequence is obtained, specifically including:

[0032] Obtain an initial data sequence; the initial data sequence includes: an initial voltage sequence and an initial current sequence;

[0033] The initial data sequence is smoothed to obtain the collected data sequence.

[0034] Optionally, the initial data sequence is smoothed to obtain the acquired data sequence, specifically including:

[0035] A second sliding window is added to the initial data sequence, and the second sliding window is moved according to a second set step size. The initial data sequence is smoothed using a moving average filtering algorithm to obtain the collected data sequence.

[0036] The present invention also provides an adaptive noise threshold event detection system, comprising:

[0037] A sequence acquisition module is used to acquire a data sequence; the data sequence includes a voltage sequence and a current sequence.

[0038] A power calculation module is used to calculate an active power sequence based on the collected data sequence; the active power sequence includes multiple power points;

[0039] A windowing module is used to apply a first sliding window to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window;

[0040] The event detection module is used to move the average value window and the event detection window according to a first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When the event occurrence time is detected, the variance detection window moves from the event occurrence time according to the first set step size.

[0041] The steady-state detection module is used to calculate the variance of the power point in the current variance detection window each time the variance detection window is moved, and to determine the time corresponding to the power point in the current variance detection window as the steady-state time node when the variance of the power point in the current variance detection window is less than the variance detection threshold.

[0042] The feedback verification module is used to perform feedback verification based on the power point before the event occurrence time and the power point after the steady-state time node to determine whether the event occurrence time is an event switching point;

[0043] The event detection and judgment process in the event detection module includes:

[0044] Calculate the current adaptive noise threshold based on the power points within the current average window;

[0045] Calculate the statistics of each power point within the current event detection window based on the current adaptive noise threshold, and determine whether there is an event occurrence time within the current event detection window based on the statistics.

[0046] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the above-described adaptive noise threshold event detection method.

[0047] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described adaptive noise threshold event detection method.

[0048] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0049] This invention proposes an adaptive noise threshold event detection method, system, device, and medium. It abandons the traditional method of pre-setting the noise level in the detection algorithm and adopts an adaptive noise threshold. The adaptive noise threshold is dynamically selected according to the current fluctuation of the system, avoiding false detection. After detecting the change point, a variance detection window is added to detect when the system enters a steady state. A feedback verification mechanism is introduced to determine whether the point belongs to the event switching point based on the difference between the mean before and after the detected change point. This provides a guarantee for the subsequent extraction of steady-state waveforms and has outstanding anti-interference ability. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A flowchart of the adaptive noise threshold event detection method provided in an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram illustrating the evolution of a detection event provided in an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram illustrating the modular implementation process of the adaptive noise threshold event detection method provided in this embodiment of the invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] Example 1

[0057] See Figure 1 The adaptive noise threshold event detection method of this embodiment includes:

[0058] Step 101: Obtain the data sequence; the data sequence includes a voltage sequence and a current sequence.

[0059] Step 101 specifically includes:

[0060] 1) An initial data sequence is acquired using a high-frequency electrical acquisition device; the initial data sequence includes: an initial voltage sequence and an initial current sequence.

[0061] 2) The high-frequency electrical acquisition device may be affected by noise during the measurement process. In order to make the acquired data closer to the standard template waveform, the initial data sequence is smoothed to obtain the acquired data sequence.

[0062] The smoothing process is as follows: a second sliding window is added to the initial data sequence, the second sliding window is moved according to a second set step size, and a moving average filtering algorithm is used to smooth the initial data sequence to obtain the collected data sequence. Specifically: a second sliding window with a width of 5 is selected and slids on the initial data sequence with a step size of 1. A moving average filtering algorithm is used within the window, and the calculation method is as follows:

[0063] yy0=y0

[0064]

[0065]

[0066]

[0067] …

[0068] Among them, [y0,y1,…y n ] represents the initial data sequence, n is the length of the initial data sequence, and yy i This represents the i-th data value in the acquired data sequence after smoothing. The voltage sequence is represented by [u1, u2, ..., u...]. n The current sequence is represented by [i1, i2, ..., i...]. n ]express.

[0069] Step 102: Calculate the active power sequence based on the collected data sequence; the active power sequence includes multiple power points. Specifically:

[0070] The initial data sequence acquired by the high-frequency electrical acquisition device is smoothed in step 101 to obtain the voltage sequence [u1, u2, ..., u]. n ] and current sequence [i1,i2,…,i n Because the measured voltage and current signal data is too large, directly using the collected data for detection would increase the computational load. Therefore, active power is calculated based on the voltage and current sequences. Since the voltage and current sequences fluctuate periodically, to make the data smoother and the detection more accurate, while reducing the subsequent computational load, the active power calculation formula is as follows:

[0071]

[0072] Where N represents the number of sampling points within one cycle, i a For the a-th sampling point of the current within one cycle, u a Let a be the voltage sampling point a within one cycle. The active power sequence [p1, p2, ..., p] is calculated. m ], where m = n / N, thus compressing the original sampling sequence by a factor of N, reducing the amount of subsequent calculations.

[0073] Step 103: Add a first sliding window to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window.

[0074] In this step, initial parameters are first set. The width of the average value window WD1 and the width of the event detection window WD2 are set, and the event alarm value h is set. An event occurs when the cumulative sum exceeds h. The first step size is set to l, representing the distance the window slides forward each time. The width of the variance detection window is set to WD3, and the variance detection threshold is set to δ. A variance within the variance detection window less than this value indicates that the system has entered a steady state. The original CUSUM detection algorithm pre-sets a noise figure, which can easily lead to false detections for devices with large power fluctuations. Therefore, this embodiment does not consider the determination of the noise figure when setting the initial parameters.

[0075] In one example, the width of the average value window WD1 is set to 10, the degree of the event detection window WD2 is set to 10, and the event alarm value h is set to the minimum detection power P. min Half of the first step size l is set to 10, the width of the variance detection window WD3 is set to 10, and the variance detection threshold δ is set to 0.1.

[0076] Step 104: Move the average value window and the event detection window according to the first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When the event occurrence time is detected, the average value window and the event detection window stop sliding, and the variance detection window moves from the event occurrence time according to the first set step size. Figure 2 r0 represents the initial time r k Indicates the time when the event occurred.

[0077] The event detection and judgment process includes:

[0078] 1) Calculate the current adaptive noise threshold based on the power points within the current average window. Specifically: Calculate the mean power within the current average window based on the power points within the current average window to obtain the first average power; calculate the current adaptive noise threshold based on the maximum power point, the minimum power point, the first average power, and the noise level coefficient within the current average window.

[0079] The formula for calculating the first average power is:

[0080]

[0081] Where, p k This represents the k-th power point within the window.

[0082] The current formula for calculating the adaptive noise threshold is:

[0083]

[0084] Where, p max p represents the maximum power point within the current average value window; min The value represents the minimum power point within the current average window; ε represents the noise level coefficient, typically set to 1; μ p1 β represents the first mean power; β represents the current adaptive noise threshold.

[0085] 2) Calculate the statistics of each power point within the current event detection window based on the current adaptive noise threshold, and determine whether an event occurs within the current event detection window based on the statistics. Specifically: If the statistics of the target power point are less than the time alarm value, the detection delay of the previous power point is incremented by 1 and determined as the detection delay of the target power point; the target power point is any power point within the current event detection window. If the statistics of the target power point are greater than the time alarm value, the event occurrence time is determined based on the sequence number of the target power point and the detection delay of the previous power point. If the statistics of all power points within the current event detection window are 0, the detection of the next event detection window is performed.

[0086] Furthermore, statistics include positive and negative statistics. The formula for calculating a statistic is:

[0087]

[0088]

[0089] in, This is the positive statistic for the k-th power point. p is the negative statistic for the k-th power point. k For the k-th power point, This is the positive statistic for the (k-1)th power point. This is the negative statistic for the (k-1)th power point. This is the positive statistic for the first power point within the current average window. It is the negative statistic for the first power point within the current average window.

[0090] Compare the calculated statistic with the set event alarm value h. When a positive offset is detected but the event alarm value h has not yet been reached, the detection delay d is recorded. + =d + +1, similarly, when When a negative offset is detected but the event alarm value h has not yet been reached, the detection delay d is recorded. - =d - +1. When or When a change in the system is detected, the event time τ = kd is calculated by working backwards. + Or τ=kd - If the detection statistic remains 0, the sliding window will be moved by a step size to the next window for detection, until the data block detection is complete.

[0091] Step 105: Each time the variance detection window is moved, the variance of the power point in the current variance detection window is calculated. When the variance of the power point in the current variance detection window is less than the variance detection threshold, the variance detection window stops sliding and the time corresponding to the power point in the current variance detection window is determined as the steady-state time node.

[0092] When the variance of the power points within the current variance detection window is less than the variance detection threshold, it indicates that the transient event has ended and the system has entered a steady state. Figure 2 Chinese r k +d t This indicates the steady-state time point. When the variance of the power points within the variance detection window is greater than or equal to the variance detection threshold, it indicates that the transient event has not yet ended and the system has not yet entered steady state. The variance detection window continues to slide to the next detection window.

[0093] The formula for calculating the variance of power points within the current variance detection window is:

[0094]

[0095] Where, μ p2 This represents the mean power within the variance detection window.

[0096] Step 106: Perform feedback verification based on the power point before the event occurrence time and the power point after the steady-state time node to determine whether the event occurrence time is the event switching point.

[0097] Step 106 specifically includes:

[0098] 1) Calculate the switching point judgment value based on the power point before the event occurrence and the power point after the steady-state time node; the calculation formula for the switching point judgment value is:

[0099]

[0100] Where γ represents the switching point judgment value; μ0 represents the power mean of the power points before the event occurs; μ1 represents the power mean of the power points after the steady-state time node; P min This represents the minimum power point of the detected portion of the active power sequence.

[0101] In practical applications, when calculating the switching point judgment value, power points within a certain range are selected. For example, the average power of ten power points before the event occurs and the average power of ten power points after the steady-state time node are used.

[0102] 2) If the switching point judgment value is 1, the time of the event occurrence is determined to be the event switching point; if the switching point judgment value is 0, the time of the event occurrence is determined to be not the event switching point, but only affected by interference signals. Figure 2 As shown, the time r of the event occurrence is... k Previous power point and steady-state time node r k +d t The power points are then compared to determine the event switching point. Figure 2 In this context, x(n) represents the active power sequence point.

[0103] Based on the above steps, it is possible to detect whether there are events of electrical equipment being turned on and off indoors without entering the residential building, accurately pinpoint the time of the event, and prepare for subsequent extraction of event characteristics and identification of load types, ultimately achieving demand-side electricity behavior management.

[0104] The modular implementation process of the above adaptive noise threshold event detection method is as follows: Figure 3 As shown, it mainly includes a data loading module, an event detection module, a steady-state detection module, and a feedback verification module. Figure 3 In this context, k represents the k-th point of the active power sequence to which the sliding window moves, and N... m N is the width of the average window WD1. d Where WD2 is the width of the event detection window, N is the length of the active power sequence, and μ is the value of WD2. pk is the mean of the power points within the average window. D Z represents the position of the current detection point within the event detection window. + and Z - d represents the positive and negative detection statistics, respectively. + and d - These represent the detection delays for positive and negative directions, respectively. h is the event alarm value (the threshold for event detection; reaching this value is considered an event occurring), and τ is the event occurrence time (the detection algorithm has a delay; the event occurrence time is the detection point minus the delay).

[0105] This embodiment addresses the requirement for high accuracy in non-intrusive load identification. To prevent false detections in the event detection phase of the identification process, it proposes an adaptive noise threshold event detection method with feedback verification and a variance detection window. This method improves upon existing event detection algorithms by employing an adaptive noise threshold and introducing a feedback verification mechanism to prevent false detections and accurately detect load switching events. Furthermore, the introduction of a sliding variance detection window detects when the system enters a steady state, ensuring the extraction of the steady-state waveform and providing excellent anti-interference capabilities.

[0106] Example 2

[0107] In order to implement the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, an adaptive noise threshold event detection system is provided below.

[0108] The system includes:

[0109] The sequence acquisition module is used to acquire the collected data sequence; the collected data sequence includes a voltage sequence and a current sequence.

[0110] The power calculation module is used to calculate the active power sequence based on the collected data sequence; the active power sequence includes multiple power points.

[0111] A windowing module is used to apply a first sliding window to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window.

[0112] The event detection module is used to move the average value window and the event detection window according to a first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When an event is detected, the variance detection window moves from the time of the event occurrence according to the first set step size.

[0113] The steady-state detection module is used to calculate the variance of the power points in the current variance detection window each time the variance detection window is moved, and to determine the time corresponding to the power points in the current variance detection window as the steady-state time node when the variance of the power points in the current variance detection window is less than the variance detection threshold.

[0114] The feedback verification module is used to perform feedback verification based on the power point before the event occurrence time and the power point after the steady-state time node to determine whether the event occurrence time is the event switching point.

[0115] The event detection judgment process in the event detection module includes: calculating the current adaptive noise threshold based on the power points in the current average window; calculating the statistics of each power point in the current event detection window based on the current adaptive noise threshold; and determining whether there is an event occurrence time in the current event detection window based on the statistics.

[0116] Example 3

[0117] This embodiment provides an electronic device, including a memory and a processor. The memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the adaptive noise threshold event detection method of Embodiment 1.

[0118] Alternatively, the aforementioned electronic device may be a server.

[0119] In addition, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the adaptive noise threshold event detection method of Embodiment 1.

[0120] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0121] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An adaptive noise threshold event detection method, characterized in that, include: Obtain the collected data sequence; The acquired data sequence includes a voltage sequence and a current sequence; Calculate the active power sequence based on the collected data sequence; The active power sequence includes: multiple power points; A first sliding window is added to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window; The average value window and the event detection window are moved according to a first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When the event occurrence time is detected, the variance detection window moves from the event occurrence time according to the first set step size. Each time the variance detection window is moved, the variance of the power point in the current variance detection window is calculated, and when the variance of the power point in the current variance detection window is less than the variance detection threshold, the time corresponding to the power point in the current variance detection window is determined as the steady-state time node. Feedback verification is performed based on the power point before the event occurrence and the power point after the steady-state time node to determine whether the event occurrence is an event switching point; The event detection and judgment process includes: Calculate the current adaptive noise threshold based on the power points within the current average window; Calculate the statistics of each power point within the current event detection window based on the current adaptive noise threshold, and determine whether there is an event occurrence time within the current event detection window based on the statistics.

2. The adaptive noise threshold event detection method according to claim 1, characterized in that, The current adaptive noise threshold is calculated based on the power points within the current average window, specifically including: The mean power within the current average window is calculated based on the power points within the average window, thus obtaining the first average power. The current adaptive noise threshold is calculated based on the maximum power point within the current average window, the minimum power point within the current average window, the first average power, and the noise level coefficient.

3. The adaptive noise threshold event detection method according to claim 1, characterized in that, The statistical measures are used to determine whether an event occurs within the current event detection window. Specifically, this includes: If the statistic of the target power point is less than the time alarm value, then the detection delay of the previous power point is increased by 1 and determined as the detection delay of the target power point; the target power point is any power point within the current event detection window; If the statistic of the target power point is greater than the time alarm value, the time of the event is determined based on the sequence number of the target power point and the detection delay of the previous power point. If the statistics of all power points in the current event detection window are 0, then proceed to the next event detection window.

4. The adaptive noise threshold event detection method according to claim 1, characterized in that, Feedback verification is performed based on the power point before the event occurrence and the power point after the steady-state time node to determine whether the event occurrence time is an event switching point, specifically including: The switching point judgment value is calculated based on the power point before the event occurrence and the power point after the steady-state time node; the calculation formula for the switching point judgment value is as follows: Where γ represents the switching point judgment value; μ0 represents the power mean of the power points before the event occurs; μ1 represents the power mean of the power points after the steady-state time node; P min This represents the minimum power point of the detected portion of the active power sequence; If the cut-off point judgment value is 1, the time when the event occurs is determined to be the event cut-off point; if the cut-off point judgment value is 0, the time when the event occurs is determined not to be the event cut-off point.

5. The adaptive noise threshold event detection method according to claim 2, characterized in that, The current formula for calculating the adaptive noise threshold is: Where, p max p represents the maximum power point within the current average value window; min The minimum power point within the current average window is represented by ε; the noise level coefficient is represented by μ. p1 β represents the first mean power; β represents the current adaptive noise threshold.

6. The adaptive noise threshold event detection method according to claim 1, characterized in that, The data sequence was acquired, specifically including: Obtain an initial data sequence; the initial data sequence includes: an initial voltage sequence and an initial current sequence; The initial data sequence is smoothed to obtain the collected data sequence.

7. The adaptive noise threshold event detection method according to claim 6, characterized in that, The initial data sequence is smoothed to obtain the collected data sequence, specifically including: A second sliding window is added to the initial data sequence, and the second sliding window is moved according to a second set step size. The initial data sequence is smoothed using a moving average filtering algorithm to obtain the collected data sequence.

8. An adaptive noise threshold event detection system, characterized in that, include: A sequence acquisition module is used to acquire a data sequence; the data sequence includes a voltage sequence and a current sequence. A power calculation module is used to calculate an active power sequence based on the collected data sequence; The active power sequence includes: multiple power points; A windowing module is used to apply a first sliding window to the active power sequence; the first sliding window includes: an average value window, an event detection window, and a variance detection window; The event detection module is used to move the average value window and the event detection window according to a first set step size. Each time the average value window and the event detection window are moved, an event detection judgment process is performed. When the event occurrence time is detected, the variance detection window moves from the event occurrence time according to the first set step size. The steady-state detection module is used to calculate the variance of the power point in the current variance detection window each time the variance detection window is moved, and to determine the time corresponding to the power point in the current variance detection window as the steady-state time node when the variance of the power point in the current variance detection window is less than the variance detection threshold. The feedback verification module is used to perform feedback verification based on the power point before the event occurrence time and the power point after the steady-state time node to determine whether the event occurrence time is an event switching point; The event detection and judgment process in the event detection module includes: Calculate the current adaptive noise threshold based on the power points within the current average window; Calculate the statistics of each power point within the current event detection window based on the current adaptive noise threshold, and determine whether there is an event occurrence time within the current event detection window based on the statistics.

9. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the adaptive noise threshold event detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the adaptive noise threshold event detection method as described in any one of claims 1 to 7.

Images