Multi-stage load event detection method and system based on time domain analysis

Abstract

The invention discloses a multi-stage load event detection method and system based on time domain analysis, and relates to the technical field of non-intrusive power load monitoring, and the method comprises the steps: calculating an adaptive threshold value for each sampling point, scanning a total power sequence through a sliding window, and determining the position of a load event; determining a search interval of a transient process of the load event by taking the position of the sliding window of the detected load event as a starting point; and in the search interval, detecting a complete transient process of the load event based on a change point detection principle. According to the multi-stage load event detection method based on time domain analysis, the flexibility and accuracy of load event detection are improved by calculating the self-adaptive power threshold value for each sampling point, the detection precision of the starting and ending time points of the load event is improved through the determination strategy of the starting and ending time points of the load event, and the detection accuracy of the load event is improved. The method has better effects in the aspects of accuracy, response speed and flexibility.

Description

Technical Field

[0001] The present invention relates to the technical field of non-intrusive power load monitoring, and in particular to a multi-stage load event detection method and system based on time domain analysis. Background Art

[0002] Improving the power sector's ability to perceive electricity consumption behavior on the electricity consumption side can provide data support for the formulation of energy-saving and emission reduction policies. However, as more and more electrical appliances with complex working principles and different load characteristics are becoming widely popular in user scenarios, how to accurately detect and perceive the operating mode of electrical appliances has become a research focus in the field of smart grids.

[0003] The non-intrusive load monitoring (NILM) technology proposed by the Hart team at the Massachusetts Institute of Technology only requires the installation of a sensor with communication function at the user's power supply entrance. By analyzing the total power consumption data, the power consumption information of each (main) electrical appliance in the user can be obtained. NILM can be divided into event-based NILM and non-event-based NILM according to whether its work depends on load event detection. The former has become the mainstream engineering solution due to its obvious advantages such as high computing efficiency and easy realization of full unsupervised operation.

[0004] Load event detection is a key step in the event-based NILM method. Whether the complete transient process of load events with different amplitudes, time scales and complex power signal waveforms can be accurately detected directly determines the final effect of NILM. However, most of the methods proposed in existing studies have a single principle and fixed parameter settings, and are often unable to effectively cope with the detection task of the complete transient process from different electrical load events. In addition, the widespread background noise also has a negative impact on the detection effect. Summary of the invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that the existing load event detection method is sensitive to background noise, lacks adaptive threshold setting, has difficulty in identifying steady-state periods, and has problems in accurately detecting transient processes of different load events.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a multi-stage load event detection method based on time domain analysis, comprising calculating an adaptive threshold for each sampling point, scanning the total power sequence through a sliding window, and determining the load event location; using the sliding window position where the load event is detected as the starting point to determine the search interval of the transient process of the load event; in the search interval, detecting the complete transient process of the load event based on the change point detection principle.

[0008] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the calculation of the adaptive threshold for each sampling point includes, for each sampling point, performing statistical analysis on a set of power data within a specific time period to determine a dynamic threshold through the relationship between the proportional coefficient and the power data, quantifying the volatility of the data, extracting the change trend of the data, and combining the preset coefficient to form a dynamically adjustable reference value, which is adjusted with the change of time and power data, and scanning the total active power sequence through a sliding window with a length of L, and calculating the adaptive power threshold at the middle time point of the sliding window, which is expressed as:

[0009]

[0010] in, is the adaptive power threshold, P(i t ) is the total power sampling point in the i-th sliding window, σ() is the standard deviation calculation, α is the proportional coefficient used to calculate the adaptive threshold setting, and the upper limit P is set for the elements in the adaptive power threshold. th,max and the lower limit P th,min , and obtain the adaptive threshold sequence.

[0011] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the load event location is determined in the time series of power data, the extreme value difference reflects the fluctuation amplitude of power data in a certain time period, and by comparing the fluctuation amplitude with the set maximum threshold, it is judged whether there is an abnormal power change in the time period, that is, a load event, and the power data at consecutive time points are analyzed to find the most significant variation amplitude in the data. When the variation amplitude exceeds the preset maximum allowable range, it indicates that there is abnormal load behavior in this time period, and a load event is generated by a time series of power data with a length of w. s The sliding window scans the total active power sequence again to determine whether there is a load event, which is expressed as:

[0012] Range(P(t i ))>max(P th (t i ))

[0013] Among them, Range() is a range calculation operation. If the total power sampling point in the i-th sliding window satisfies the judgment formula, a load event occurs in the window.

[0014] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the search interval for determining the transient process of the load event includes analyzing the change trend of the power data in a specific time window, extracting a straight trend line representing the change of the data in the time period, the slope of the trend line reflects the change rate of the power data in the time period, combining the degree of deviation of the power data at each time point in the time window from the average value and the influence of the time deviation from the average value, quantifying the change trend of the power data, comparing it with a specific threshold, and judging whether the current time window is in a stable state. If the change trend of the power data reaches or exceeds the set value, the power data has changed in the time window, indicating that the load event has occurred. Starting from the two ends of the window where the load event occurs, the length of each window is w. l and w r The sliding window slides forward and backward to scan the total active power sequence, which can be expressed as:

[0015]

[0016]

[0017] Among them, k th1 and k th2 They are respectively the judgment sliding window w l and w r The threshold for sliding to stop, and They are sliding windows w l and w r The average value of the timestamp in and is the average value of the total active power sampling samples in the two sliding windows. When the judgment formula is satisfied, the sliding window w l and w r Stop sliding and define the middle area as the search interval for the transient process of the load event.

[0018] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the complete transient process of detecting load events based on the change point detection principle includes judging the type of load events, calculating the average value of power data in two different time periods, which are two stable states before and after the load event, respectively, and judging whether the load event is on or off by comparing the two average values. When the power average value before the event is lower than the power average value after the event, the load changes from a lower power state to a higher power state, indicating that the load is on. If the power average value after the event is lower than the power average value before the event, it indicates that the load is off. According to the sliding window w l and w rThe total active power signal in each steady-state period determines the type of load event, which is expressed as:

[0019]

[0020] Where type is the type of load event.

[0021] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the complete transient process of detecting load events based on the change point detection principle also includes determining the starting point of the transient process of the load event according to the change point detection principle, by comparing the power changes at the current time point and the subsequent time point, combined with the preset range value, judging whether the power change occurs, when the load event is determined to be in the on state, if the power value at the current moment increases compared with the next moment, and the power change at the previous moment is small, the starting point of the load event is determined, when the load event is determined to be in the off state, if the power value at the current moment decreases compared with the next moment, and the power change at the previous moment is also small, it indicates the termination of the load event, which is expressed as:

[0022]

[0023] Among them, {P tran} is the total active power sequence in the search interval, and the sampling point that satisfies the judgment formula is defined as the starting point of the transient process of the load event t start .

[0024] As a preferred solution of the multi-stage load event detection method based on time domain analysis described in the present invention, wherein: the complete transient process of detecting the load event based on the change point detection principle also includes determining the end point of the transient process of the load event according to the change point detection principle. At a specific moment, the difference between the current power value and the reference value does not exceed a preset range, and the power change amplitude between the current power value and the previous moment is within a preset range. The reference value represents the stable state of the power data in the transient process, which is expressed as:

[0025] |u2(t)-P(t)|<λ*range({P tran}&|P(t)-P(t-1)|<γ*range({P tran}

[0026] Among them, λ and γ are the proportional coefficients of the start and end time point detection.

[0027] Another object of the present invention is to provide a multi-stage load event detection system based on time domain analysis, which can determine the search interval of the transient process of the load event through the search interval module, solving the current problem of being unable to accurately identify the transient process of the load event.

[0028] As a preferred solution of the multi-stage load event detection system based on time domain analysis described in the present invention, it includes: an event determination module, a search interval module, and a process detection module; the event determination module is used to calculate an adaptive threshold for each sampling point, and determine the load event position by scanning the total power sequence through a sliding window; the search interval module is used to use the sliding window position where the load event is detected as the starting point to determine the search interval of the transient process of the load event; the process detection module is used to detect the complete transient process of the load event in the search interval based on the change point detection principle.

[0029] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a multi-stage load event detection method based on time domain analysis.

[0030] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a multi-stage load event detection method based on time domain analysis.

[0031] Beneficial effects of the present invention: The multi-stage load event detection method based on time domain analysis provided by the present invention improves the flexibility and accuracy of load event detection by calculating an adaptive power threshold for each sampling point, improves the integrity and accuracy of load event detection through precise search and positioning of the steady-state period, improves the response speed and reliability of the overall load monitoring system through the change point detection method, improves the accuracy of event identification by comparing the average values ​​of power data in two time windows before and after the event, and improves the detection accuracy of the start and end time points of the load event through the strategy for determining the start and end time points of the load event. The present invention achieves better results in terms of accuracy, response speed and flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0033] Figure 1 An overall flow chart of a multi-stage load event detection method based on time domain analysis provided in the first embodiment of the present invention.

[0034] Figure 2 An overall module diagram of a multi-stage load event detection system based on time domain analysis provided in the third embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0036] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a multi-stage load event detection method based on time domain analysis, comprising:

[0037] S1: Calculate the adaptive threshold for each sampling point, scan the total power sequence through the sliding window, and determine the load event location.

[0038] Furthermore, calculating an adaptive threshold for each sampling point includes calculating an adaptive threshold for each sampling point.

[0039] It should be noted that, through the relationship between the proportional coefficient and the power data, a dynamic threshold is determined by statistical analysis of a set of power data within a specific time. By quantifying the volatility of the data, the change trend of the data is extracted, and combined with the preset coefficient, a dynamically adjustable reference value is formed, which is adjusted with the change of time and power data. The total active power sequence is scanned through a sliding window with a length of L, and the adaptive power threshold at the middle time point of the sliding window is calculated and expressed as:

[0040]

[0041] in, is the adaptive power threshold, P(i t ) is the total power sampling point in the i-th sliding window, σ() is the standard deviation calculation, α is the proportional coefficient used to calculate the adaptive threshold setting, and the upper limit P is set for the elements in the adaptive power threshold. th , max and the lower limit P th , min , and obtain the adaptive threshold sequence.

[0042] Furthermore, determining the location of the load events is included in the time series of the power data.

[0043] It should be noted that the extreme value difference reflects the fluctuation range of power data within a certain period of time. By comparing the fluctuation range with the set maximum threshold, it is determined whether there is an abnormal power change in the time period, that is, a load event. The power data at consecutive time points are analyzed to find the most significant change range in the data. When the change range exceeds the preset maximum allowable range, it means that there is an abnormal load behavior in this time period. sThe sliding window scans the total active power sequence again to determine whether there is a load event, which is expressed as:

[0044] Range(P(t i ))>max(P th (t i ))

[0045] Among them, Range() is a range calculation operation. If the total power sampling point in the i-th sliding window satisfies the judgment formula, a load event occurs in the window.

[0046] It should also be noted that the calculation of adaptive thresholds allows the system to accurately respond to the initial changes in load events and can capture signs of load events earlier. Through this early capture capability, the system can respond at the very beginning of the load event, thereby having more time and information for subsequent processing. Since the threshold is calculated dynamically, it can filter out non-event changes caused by accidental fluctuations or background noise. The flexibility of this threshold reduces the false alarm rate of the system and makes the detection of load events more reliable. Different electrical equipment may have different power characteristics. Through the setting of adaptive thresholds, the system can automatically adjust to adapt to these different characteristics. There is no need to set specific detection parameters for each device, which greatly simplifies the deployment and use of the system.

[0047] S2: Taking the sliding window position where the load event is detected as the starting point, determine the search interval of the transient process of the load event.

[0048] Furthermore, determining the search interval of the load event transient process includes analyzing the change trend of the power data within a specific time window.

[0049] It should be noted that a straight trend line representing the change of data in a time period is extracted. The slope of the trend line reflects the rate of change of power data in a time period. The degree of deviation of power data at each time point in the time window from the average value and the influence of time deviation from the average value are combined to quantify the change trend of power data. The trend line is compared with a specific threshold to determine whether the current time window is in a stable state. If the change trend of power data reaches or exceeds the set value, the power data has changed in the time window, indicating that a load event has occurred. Starting from the two ends of the window where the load event occurs, the length of each window is w. l and w r The sliding window slides forward and backward to scan the total active power sequence, which can be expressed as:

[0050]

[0051]

[0052] Among them, kth1 and k th2 They are respectively the judgment sliding window w l and w r The threshold for sliding to stop, and They are sliding windows w l and w r The average value of the timestamp in and is the average value of the total active power samples in the two sliding windows. When the judgment formula is met, the sliding window w l and w r Stop sliding and define the middle area as the search interval for the transient process of the load event.

[0053] It should also be noted that sliding window scanning allows the system to gradually define the impact range of the event forward and backward starting from the detected load event location. Through this gradually narrowing search interval, the system can accurately locate the impact range of the load event, thereby reducing the interference of irrelevant data. By determining an accurate search interval, the system only needs to concentrate on processing the data within this interval, rather than the entire power sequence. This centralized processing method significantly improves the processing efficiency of the system, especially in scenarios with large data sets and real-time detection, it can speed up the detection speed. Through sliding window scanning, this step can accurately distinguish between steady-state periods and transient periods, ensuring that the data in the search interval mainly reflects the transient process, thereby providing a more accurate data basis for subsequent detection. This improvement in accuracy enables the system to maintain a high level of detection performance in a complex power environment.

[0054] S3: In the search interval, the complete transient process of the load event is detected based on the change point detection principle.

[0055] Furthermore, the complete transient process of detecting a load event based on the change point detection principle includes determining the type of the load event.

[0056] It should be noted that the average value of power data in two different time periods is calculated, which are the two stable states before and after the load event. By comparing the two average values, the load event is judged to be on or off. When the power average value before the event is lower than the power average value after the event, the load changes from a lower power state to a higher power state, indicating that the load is on. If the power average value after the event is lower than the power average value before the event, it indicates that the load is off. According to the sliding window w l and w r The total active power signal in each steady-state period determines the type of load event, which is expressed as:

[0057]

[0058] Where type is the type of load event.

[0059] Furthermore, detecting the complete transient process of the load event based on the change point detection principle also includes determining the starting point of the transient process of the load event based on the change point detection principle.

[0060] It should be noted that by comparing the power changes at the current time point with the power changes at the subsequent time points, combined with the preset range value, it is determined whether a power change occurs. When the load event is determined to be in the on state, if the power value at the current moment increases compared to the subsequent moment, and the power change at the previous moment is small, the starting point of the load event is determined. When the load event is determined to be in the off state, if the power value at the current moment decreases compared to the subsequent moment, and the power change at the previous moment is also small, it indicates the termination of the load event, which is expressed as:

[0061]

[0062] Among them, {P tran} is the total active power sequence in the search interval, and the sampling point that satisfies the judgment formula is defined as the starting point of the transient process of the load event t start .

[0063] Furthermore, detecting the complete transient process of the load event based on the change point detection principle also includes determining the end point of the transient process of the load event based on the change point detection principle.

[0064] It should be noted that at a specific moment, the difference between the current power value and the reference value does not exceed the preset range, and the power change amplitude between the current power value and the previous moment is within the preset range. The reference value represents the stable state of the power data in the transient process, which is expressed as:

[0065] |u2(t)-P(t)|<λ*range({P tran}&|P(t)-P(t-1)|<γ*range({P tran}

[0066] Among them, λ and γ are the proportional coefficients of the start and end time point detection.

[0067] It should also be noted that the change point detection method can keenly identify the time points when significant changes occur in the power data. These change points usually correspond to the beginning or end of load events. By accurately capturing these key nodes, the system can accurately locate the starting and ending points of the transient process, thereby fully describing the entire event process. By accurately locating the start and end time points of the load event, this step ensures a complete record of the event process, which not only helps to understand and analyze the overall picture of the load event, but also provides reliable data support for subsequent load modeling and prediction. Change point detection can effectively distinguish between real load events and false changes caused by noise. This capability enables the system to provide accurate detection results in a noisy power environment, reduce the possibility of misjudgment, and improve the robustness of the system.

[0068] Example 2, an embodiment of the present invention, provides a multi-stage load event detection method based on time domain analysis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0069] The total power data of phase A and phase B are selected from the public data set BLUED for testing load events for testing. The evaluation indicators are the state recognition accuracy indicators commonly used in the field of non-intrusive load monitoring, including precision, recall and F1 score.

[0070] In the embodiment, the key parameters are set as follows: the sliding window length for calculating the adaptive threshold is 30 seconds; the upper limit of the adaptive threshold is set to 60W, and the lower limit is set to 10W; the sliding window length for load event positioning is set to 10 seconds; the sliding window length for multi-stage steady-state search is set to 5 seconds; the sliding window length for determining the sliding window w l and w r The sliding stop threshold k th1 Set to 10, k th2 Set to 0.5.

[0071] Table 1 Test results

[0072] Scenario Accuracy Recall F1 score BLUEDA 99.3％ 87.9％ 93.2％ BLUEDB 94.6％ 56.3％ 70.6％

[0073] As shown in Table 1, the detection results of the multi-stage load event detection technology based on time domain analysis in different scenarios described in the present invention are ahead of the field of non-intrusive load monitoring.

[0074] Example 3, reference Figure 2 , as an embodiment of the present invention, provides a multi-stage load event detection system based on time domain analysis, including an event determination module, a search interval module, and a process detection module.

[0075] The event determination module is used to calculate the adaptive threshold for each sampling point, and determine the load event location by scanning the total power sequence through a sliding window; the search interval module is used to take the sliding window position where the load event is detected as the starting point to determine the search interval of the transient process of the load event; the process detection module is used to detect the complete transient process of the load event in the search interval based on the change point detection principle.

[0076] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0077] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0078] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0079] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A multi-stage load event detection method based on time domain analysis, characterized in that: include: Calculate the adaptive threshold for each sampling point, scan the total power sequence through the sliding window, and determine the load event location; Taking the sliding window position where the load event is detected as the starting point, the search interval of the transient process of the load event is determined; In the search interval, the complete transient process of the load event is detected based on the change point detection principle.

2. The multi-stage load event detection method based on time domain analysis according to claim 1, characterized in that: The calculation of the adaptive threshold for each sampling point includes, for each sampling point, performing statistical analysis on a set of power data within a specific time period to determine a dynamic threshold through the relationship between the proportional coefficient and the power data, extracting the change trend of the data by quantifying the volatility of the data, and combining the preset coefficient to form a dynamically adjustable reference value, which is adjusted with the change of time and power data, and the total active power sequence is scanned through a sliding window with a length of L, and the adaptive power threshold at the middle time point of the sliding window is calculated as: in, is the adaptive power threshold, P(i t ) is the total power sampling point in the i-th sliding window, σ() is the standard deviation calculation, α is the proportional coefficient used to calculate the adaptive threshold setting, and the upper limit P is set for the elements in the adaptive power threshold. th,max and the lower limit P th,min , and obtain the adaptive threshold sequence.

3. The multi-stage load event detection method based on time domain analysis according to claim 2, characterized in that: The load event location is determined in the time series of the power data. The extreme value difference reflects the fluctuation range of the power data in a certain time period. By comparing the fluctuation range with the set maximum threshold, it is determined whether there is an abnormal power change in the time period, that is, a load event. The power data at consecutive time points are analyzed to find the most significant change range in the data. When the change range exceeds the preset maximum allowable range, it means that there is abnormal load behavior in this time period. s The sliding window scans the total active power sequence again to determine whether there is a load event, which is expressed as: Range(P(t i ))>max(P th (t i )) Among them, Range() is a range calculation operation. If the total power sampling point in the i-th sliding window satisfies the judgment formula, a load event occurs in the window.

4. The multi-stage load event detection method based on time domain analysis according to claim 3 is characterized in that: The search interval for determining the transient process of the load event includes analyzing the change trend of the power data in a specific time window, extracting a straight trend line representing the change of the data in the time period, the slope of the trend line reflects the change rate of the power data in the time period, combining the degree of deviation of the power data at each time point in the time window from the average value and the influence of the time deviation from the average value, quantifying the change trend of the power data, comparing it with a specific threshold, and judging whether the current time window is in a stable state. If the change trend of the power data reaches or exceeds the set value, the power data has changed in the time window, indicating that the load event has occurred. Starting from the two ends of the window where the load event occurs, the length of each window is w. l and w r The sliding window slides forward and backward to scan the total active power sequence, which can be expressed as: Among them, k th1 and k th2 They are respectively the judgment sliding window w l and w r The threshold for sliding to stop, and They are sliding windows w l and w r The average value of the timestamp in and is the average value of the total active power samples in the two sliding windows. When the judgment formula is met, the sliding window w l and w r Stop sliding and define the middle area as the search interval for the transient process of the load event.

5. The multi-stage load event detection method based on time domain analysis according to claim 4, characterized in that: The complete transient process of detecting load events based on the change point detection principle includes determining the type of load event, calculating the average value of power data in two different time periods, which are two stable states before and after the load event, and determining whether the load event is on or off by comparing the two average values. When the power average value before the event is lower than the power average value after the event, the load changes from a lower power state to a higher power state, indicating that the load is on. If the power average value after the event is lower than the power average value before the event, it indicates that the load is off. According to the sliding window w l and w r The total active power signal in each steady-state period determines the type of load event, which is expressed as: Where type is the type of load event.

6. The multi-stage load event detection method based on time domain analysis according to claim 5, characterized in that: The complete transient process of detecting load events based on the change point detection principle also includes determining the starting point of the transient process of the load event according to the change point detection principle, and judging whether a power change occurs by comparing the power change between the current time point and the later time point in combination with the preset range value. When the load event is determined to be in the on state, if the power value at the current moment increases compared with the later moment, and the power change at the previous moment is small, the starting point of the load event is determined. When the load event is determined to be in the off state, if the power value at the current moment decreases compared with the later moment, and the power change at the previous moment is also small, it indicates the termination of the load event, which is expressed as: Among them, {P tran } is the total active power sequence in the search interval, and the sampling point that satisfies the judgment formula is defined as the starting point of the transient process of the load event t start .

7. The multi-stage load event detection method based on time domain analysis according to claim 6, characterized in that: The complete transient process of detecting load events based on the change point detection principle also includes determining the end point of the transient process of the load event according to the change point detection principle. At a specific moment, the difference between the current power value and the reference value does not exceed a preset range, and the power change amplitude between the current power value and the previous moment is within a preset range. The reference value represents the stable state of the power data in the transient process, which is expressed as: |u2(t)-P(t)|<λ*range({P tran }&|P(t)-P(t-1)|<γ*range({P tran } Among them, λ and γ are the proportional coefficients of the start and end time point detection.

8. A system using the multi-stage load event detection method based on time domain analysis as claimed in any one of claims 1 to 7, characterized in that: It includes an event determination module, a search interval module, and a process detection module; The event determination module is used to calculate an adaptive threshold for each sampling point and determine the load event position by scanning the total power sequence through a sliding window; The search interval module is used to determine the search interval of the transient process of the load event by taking the sliding window position where the load event is detected as the starting point; The process detection module is used to detect the complete transient process of the load event in the search interval based on the change point detection principle.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-stage load event detection method based on time domain analysis according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-stage load event detection method based on time domain analysis according to any one of claims 1 to 7 are implemented.

Images