A NILM event detection device and method based on adaptive composite sliding filtering

Abstract

The present invention relates to a NILM event detection device and method based on adaptive composite sliding filtering, wherein a data preprocessing module is used to calculate instantaneous power of voltage and current data and perform noise reduction processing on the data; a data judgment module generates a phase difference of voltage and current according to the voltage and current data, and judges the load type of this time period; an event detection module is used to adaptively select a power threshold bilateral sliding window detection module or a total harmonic distortion bilateral sliding window detection module according to the load type to detect transient events, and output the load switching interval and its characteristic information. In this way, the data preprocessing module eliminates noise from the voltage and current data and retains signal details, while the bilateral sliding window detection module introduces total harmonic distortion as the second basis for judgment in addition to power, and uses the complementary information of the two dimensions of harmonic distortion and power to adaptively judge the type of power load, thereby improving the reliability of the event detection device.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to a NILM event detection device and method based on adaptive composite sliding filtering. Background Art

[0002] Non-intrusive load monitoring technology can achieve more efficient power management and precise carbon emission monitoring, and its role in digital management of energy demand is becoming increasingly important. Event detection is an important part of non-intrusive load monitoring devices, which plays a role in determining the time point of power load input and output events.

[0003] There are two types of existing event detection methods, namely, change point detection and rule determination. Change point detection includes generalized likelihood ratio detection, chi-square goodness of fit, Bayesian information criterion, cumulative and equal probability methods; rule determination is to complete the detection and positioning of electrical switching events based on the change characteristics of a series of electrical quantity sequences of the equipment. Rule determination methods are usually based on specific rules and logic, and can be flexibly adjusted according to user needs and environmental changes, but the algorithm is not robust, and it is easy to miss and misdetect. For example, the sliding window detection based on power threshold has a difficult selection of reasonable event trigger thresholds, especially for the setting of event trigger thresholds for small power loads, which usually needs to be obtained in advance by experiments. At the same time, the unilateral sliding window filtering method used in this method is prone to signal feature loss, which reduces the accuracy and robustness of detection. The existing event detection method only monitors the difference in line power changes before and after the power grid is switched on, lacks analysis of the harmonic characteristic information of the load, resulting in poor identification accuracy. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and to provide a NILM event detection device and method based on adaptive composite sliding filtering, comprising a data preprocessing module, a data judgment module and an event detection module, wherein the data preprocessing module, the data judgment module and the event detection module are connected in sequence;

[0005] The data preprocessing module includes a data acquisition processing module and a data filtering module, and the data acquisition processing module is connected to the data filtering module;

[0006] The data acquisition and processing module is connected to the load device and is used to receive the voltage and current data of the load device and calculate the instantaneous power according to the voltage and current data;

[0007] The data filtering module is used to receive the voltage and current data and instantaneous power from the data acquisition and processing module, and perform noise reduction processing on the voltage and current data and instantaneous power, and output the voltage and current data and instantaneous power after noise reduction processing to the data judgment module;

[0008] The data judgment module is used to generate a phase difference between voltage and current based on the voltage and current data after noise reduction processing from the data preprocessing module, and judge the load type in this time period based on the phase difference, and output the load type to the event detection module;

[0009] The event detection module includes a selection module, a power threshold bilateral sliding window detection module, a total harmonic distortion bilateral sliding window detection module, and an output module. The selection module selects the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module to detect transient events according to the load type determined by the data judgment module. When the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module detects that a transient event has occurred in the device, the output module outputs the load switching interval and its characteristic information.

[0010] Preferably, the data filtering module includes a Butterworth filter, a sliding window averaging module and a Savitzky-Golay filter; the Butterworth filter is used to receive the voltage and current data from the data acquisition and processing module and to perform noise reduction on the voltage and current data; the sliding window averaging module is used to receive the instantaneous power from the data acquisition module and to perform noise reduction and smoothing on the instantaneous power; the Savitzky-Golay filter is used to receive the instantaneous power from the data acquisition module and to perform smoothing on the instantaneous power.

[0011] Preferably, the data judgment module includes a normalization processing module and a phase difference judgment module, the input end of the normalization processing module is electrically connected to the data filtering module, the output end of the normalization processing module is connected to the input end of the phase difference judgment module, and the output end of the phase difference judgment module is electrically connected to the input end of the event detection module;

[0012] The normalization processing module is used to perform normalization processing on the voltage and current data processed by the data filtering module, and transmit the processed voltage and current data to the phase difference judgment module;

[0013] The phase difference judgment module is used to calculate the phase difference of the voltage and current data according to the voltage and current data processed by the data filtering module, and to judge the load type in this time period, and output the load type to the event detection module.

[0014] Preferably, when the load type is a resistive load, the selection module selects a power threshold bilateral sliding window detection module to be electrically connected to the phase difference judgment module; when the load type is an inductive or capacitive load, the selection module selects a total harmonic distortion bilateral sliding window detection module to be electrically connected to the phase difference judgment module.

[0015] Preferably, the power threshold bilateral sliding window detection module and the total harmonic distortion bilateral sliding window detection module are both used to determine whether a transient event has occurred in the device according to whether the offset between the PM window and the LM window is greater than two set thresholds;

[0016] If a transient event is detected, the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module outputs the transient event interval, the transient event category, and the steady-state characteristics before and after the transient event interval.

[0017] Preferably, the power threshold bilateral sliding window detection module is used to use the difference between the mean values ​​of the sampling sequences of the calculation window as the offset between the PM window and the LM window; the total harmonic distortion bilateral sliding window detection module is used to use the difference between the THD values ​​of the calculation window as the offset between the PM window and the LM window.

[0018] Preferably, a NILM event detection method based on adaptive composite sliding filtering comprises: a data acquisition processing module calculates instantaneous power according to voltage and current data of a load device, and outputs the voltage and current data and instantaneous power to a data filtering module;

[0019] The data filtering module receives the voltage and current data and the instantaneous power, performs noise reduction processing on the voltage and current data and the instantaneous power, and outputs the voltage and current data and the instantaneous power after the noise reduction processing to the data judgment module;

[0020] The data judgment module generates a phase difference between voltage and current according to the voltage and current data after noise reduction processing, and judges the load type in this time period according to the phase difference, and outputs the load type to the selection module;

[0021] The selection module selects a power threshold bilateral sliding window detection module or a total harmonic distortion bilateral sliding window detection module according to the load type to detect transient events, and outputs the detection result to the output module;

[0022] When the detection result is that a transient event occurs in the load device, the output module outputs the load switching interval and its characteristic information.

[0023] The beneficial effects of the invention are as follows: the noise of current, voltage and instantaneous power is eliminated and the signal details are retained through the data preprocessing module, and the total harmonic distortion is introduced as the second judgment basis in addition to power through the bilateral sliding window detection module. The complementary information of the two dimensions of harmonic distortion and power is used to adaptively judge the type of power load, complete the judgment of the time point of the power load input and output events, and improve the reliability of the event detection device. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, without paying creative labor, other drawings obtained based on these drawings still belong to the scope of the present invention.

[0025] Figure 1 is a system block diagram of the present invention;

[0026] Figure 2 It is a flowchart of the event detection process of the present invention;

[0027] Figure 3 The detection process of the double-sided composite sliding window of the present invention;

[0028] Figure 4 The kettle event detection results of the present invention, wherein Figure (a) is the detection result of the bilateral sliding window event detection device, and Figure (b) is the detection result of the adaptive bilateral composite sliding filter event detection device;

[0029] Figure 5 The electric drill event detection results of the present invention, wherein Figure (a) is the detection result of the bilateral sliding window event detection device, and Figure (b) is the detection result of the adaptive bilateral composite sliding filter event detection device;

[0030] Figure 6 1 and 1 are the event detection results of the laptop computer of the present invention, wherein FIG. (a) is the detection result of the bilateral sliding window event detection device, and FIG. (b) is the detection result of the adaptive bilateral composite sliding filter event detection device. DETAILED DESCRIPTION

[0031] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings.

[0032] It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are for distinguishing two non-identical entities with the same name or non-identical parameters. It can be seen that "first" and "second" are only for the convenience of expression and should not be understood as limitations on the embodiments of the present invention. The subsequent embodiments will not explain this one by one.

[0033] The directions and positions mentioned in the present invention, such as "upper", "lower", "front", "back", "left", "right", "inside", "outside", "top", "bottom", "side", etc., are only for reference to the directions or positions of the drawings. Therefore, the directions and positions used are for explaining and understanding the present invention, but not for limiting the scope of protection of the present invention.

[0034] like Figures 1 to 6 As shown, this is an embodiment of the present invention. This embodiment first briefly introduces the load harmonic generation analysis and total harmonic distortion calculation. The current load equipment can be divided into three types according to electrical characteristics: resistive load, motor-driven load and power electronic load. The nonlinear devices in the power electronic load will cause the distortion of current and voltage, thereby significantly increasing the harmonic content. The total harmonic distortion rate of the current of household power electronic equipment even exceeds 160%. According to the actual measurement of the operating conditions of the residential distribution feeder, it is found that the total demand distortion coefficient of the power line harmonic current even reaches 12%. Table 1 summarizes the harmonic generation conditions of several commonly used household load devices.

[0035] Table 1 Harmonic generation status of electrical equipment

[0036] Total harmonic distortion (THD) is a quantification of the harmonic characteristics of a waveform, and is usually used to measure the degree of distortion of a signal waveform. A THD value that is too large indicates that the signal waveform is highly distorted, with abundant and significant high-order harmonic components. On the contrary, a smaller THD value means that the waveform distortion is lower and the high-order harmonic components are less. This indicator can effectively reflect the harmonic distortion characteristics of the signal and its potential impact on the performance of the device. The calculation formula for THD is as follows:

[0037] (1)

[0038] Among them, H1 represents the amplitude of the fundamental wave, H2, H3, etc. represent the amplitudes of the second and third harmonics, and n represents the highest harmonic order. The above are all common knowledge in the prior art.

[0039] This embodiment designs a NILM event detection device based on adaptive composite sliding filtering on the basis of the prior art, including a data preprocessing module, a data judgment module and an event detection module, wherein the data preprocessing module, the data judgment module and the event detection module are connected in sequence;

[0040] The data preprocessing module includes a data acquisition processing module and a data filtering module, and the data acquisition processing module is connected to the data filtering module;

[0041] The data acquisition and processing module is connected to the load device and is used to receive the voltage and current data of the load device and calculate the instantaneous power according to the voltage and current data;

[0042] The data filtering module is used to receive the voltage and current data and instantaneous power from the data acquisition and processing module, and perform noise reduction processing on the voltage and current data and instantaneous power, and output the voltage and current data and instantaneous power after noise reduction processing to the data judgment module;

[0043] The data judgment module is used to generate a phase difference between voltage and current based on the voltage and current data after noise reduction processing from the data preprocessing module, and judge the load type in this time period based on the phase difference, and output the load type to the event detection module;

[0044] The event detection module includes a selection module, a power threshold bilateral sliding window detection module, a total harmonic distortion bilateral sliding window detection module, and an output module. The selection module selects the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module to detect transient events according to the load type determined by the data judgment module. When the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module detects that a transient event has occurred in the device, the output module outputs the load switching interval and its characteristic information.

[0045] The data filtering module includes a Butterworth filter, a sliding window averaging module and a Savitzky-Golay filter; the Butterworth filter is used to receive the voltage and current data from the data acquisition and processing module and perform noise reduction on the voltage and current data;

[0046] The sliding window averaging module is used to receive the instantaneous power from the data acquisition module and perform noise reduction and smoothing on the instantaneous power;

[0047] The Savitzky-Golay filter is used to receive the instantaneous power from the data acquisition module and perform smoothing on the instantaneous power.

[0048] The data judgment module includes a normalization processing module and a phase difference judgment module, the input end of the normalization processing module is electrically connected to the data filtering module, the output end of the normalization processing module is connected to the input end of the phase difference judgment module, and the output end of the phase difference judgment module is electrically connected to the input end of the event detection module;

[0049] The normalization processing module is used to perform normalization processing on the voltage and current data processed by the data filtering module, and transmit the processed voltage and current data to the phase difference judgment module;

[0050] The phase difference judgment module is used to calculate the phase difference of the voltage and current data according to the voltage and current data processed by the data filtering module, and judge the load type of this time period, and output the load type to the event detection module. When the load type is a resistive load, the selection module selects the power threshold bilateral sliding window detection module to be electrically connected to the phase difference judgment module; when the load type is an inductive or capacitive load, the selection module selects the total harmonic distortion bilateral sliding window detection module to be electrically connected to the phase difference judgment module.

[0051] The power threshold bilateral sliding window detection module and the total harmonic distortion bilateral sliding window detection module are both used to determine whether a transient event has occurred in the device based on whether the offset between the PM window and the LM window is greater than the two set thresholds; if a transient event is detected, the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module outputs the transient event interval, the transient event category, and the steady-state characteristics before and after the transient event interval.

[0052] The power threshold bilateral sliding window detection module is used to use the difference between the mean values ​​of the sampling sequences of the calculation window as the offset between the PM window and the LM window; the total harmonic distortion bilateral sliding window detection module is used to use the difference between the THD values ​​of the calculation window as the offset between the PM window and the LM window.

[0053] This embodiment introduces a NILM event detection method based on adaptive composite sliding filtering, and the NILM event detection method based on adaptive composite sliding filtering is used to implement the above-mentioned NILM event detection device based on adaptive composite sliding filtering; it includes the following steps: the data acquisition processing module calculates the instantaneous power according to the voltage and current data of the load equipment, and outputs the voltage and current data and the instantaneous power to the data filtering module;

[0054] The data filtering module receives the voltage and current data and the instantaneous power, performs noise reduction processing on the voltage and current data and the instantaneous power, and outputs the voltage and current data and the instantaneous power after the noise reduction processing to the data judgment module;

[0055] The data judgment module generates a phase difference between voltage and current according to the voltage and current data after noise reduction processing, and judges the load type in this time period according to the phase difference, and outputs the load type to the selection module;

[0056] The selection module selects a power threshold bilateral sliding window detection module or a total harmonic distortion bilateral sliding window detection module according to the load type to detect transient events, and outputs the detection result to the output module;

[0057] When the detection result is that the load device has generated a transient event, the output module outputs the load switching interval and its characteristic information. It can be understood that the data acquisition processing module in this embodiment is any hardware with an acquisition function, such as an analog-to-digital converter (ADC); the sliding window averaging module can be a hardware that stores a sliding window averaging algorithm and can run the sliding window averaging algorithm, such as the TMS320 series chip of Texas Instruments (TI). The hardware with a sliding window averaging function can also have other functional modules, such as the normalization processing module and the phase difference judgment module in this embodiment, as well as a power threshold bilateral sliding window detection module or a total harmonic distortion bilateral sliding window detection module.

[0058] The bilateral sliding window detection of the prior art is to first set an array N of fixed length on a sequence S composed of multiple data sampling points, and perform a traversal operation on the sequence to be detected. The implementation effect can be vividly regarded as "sliding" on the sequence, and the length L of the array N is the overall size of the sliding window. The bilateral composite sliding window detection module of the present device further divides the sliding window into two parts, front and back, based on the prior art, and obtains the power average and THD value of the front and back windows respectively, and then compares the power and THD values ​​of the front and back windows to determine whether there is a switching event in the corresponding time period of the circuit, and further determines whether it is a switching event or a switching event. Among them, the data sampling point is composed of multiple voltages and currents.

[0059] It can be understood that the sampled data is filtered by the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module at the same time, and the total harmonic distortion (THD) information is incorporated. THD can reflect the switching of the load from the distortion of the waveform, and use the complementary information of the two dimensions of harmonic distortion and power to adaptively judge the type of power load, complete the judgment of the time point of the power load input and output events, and improve the reliability of the event detection module of the device. The change of THD is easier to capture than the change of active and reactive power, which makes the event detection module after adding THD more valuable in fault detection and predictive maintenance. The sampled data is voltage or voltage signal.

[0060] The following is the specific event detection process of this device: First, it is necessary to read the voltage and current data of the load equipment, calculate the instantaneous power, and perform data noise reduction. Among them, the voltage and current signals are denoised using the Butterworth filter. As a bidirectional filter, the Butterworth filter can effectively eliminate phase distortion and increase the accuracy of phase calculation. The instantaneous power signal is denoised and smoothed using a sliding window averaging module and a Savitzky-Golay filter. Compared with the Butterworth filter, they have less computational complexity and can retain more signal details.

[0061] After the voltage and current signals are processed for noise reduction, they will be normalized to reduce the impact of the absolute value difference. Normalization enables different data to be compared on the same scale to prevent some load characteristics from being submerged. Next, the phase difference of voltage and current is calculated by the cross-correlation method. Based on the phase difference information, the algorithm can determine the type of interval (resistive, inductive, capacitive) and adaptively select the main judgment basis for transient event switching detection. For resistive loads, the algorithm uses a bilateral power threshold sliding window for transient event detection. For inductive or capacitive loads, total harmonic distortion (THD) is used as the judgment basis for the bilateral sliding window.

[0062] The evolution process of the bilateral composite sliding window detection module is as follows Figure 3 As shown in the figure, the composite sliding window is composed of PM and LM. The data judgment module determines the type of interval (resistive, inductive, capacitive) to determine how to choose the mean value of the sampling sequence and the THD value of the calculation window in the PM and LM windows. Whether a transient event has occurred in the device is determined based on whether the offset between the PM window and the LM window is greater than the two set thresholds. If no transient event is detected,

[0063] The PM and LM windows will slide rightward to a new sampling point. If a transient event is detected, the transient event interval, transient event category (on / off), and steady-state characteristics (power, THD) before and after the transient event interval are output. This enables efficient detection and classification of load switching events, ensuring the reliability of load switching judgment and the accuracy of the switching interval.

[0064] The following experimental example illustrates the effectiveness of the crowdsourcing task recommendation device.

[0065] This experiment example uses the power load in Zhejiang Tianzheng Electric Laboratory to conduct a switching point event detection experiment. The sampling point X in an AC cycle is:

[0066] (2)

[0067] Wherein, C is the sampling frequency, Z is the AC frequency in the power device, and in this embodiment, C=6400 Hz, Z=50 Hz.

[0068] An adaptive bilateral composite sliding window detection module is used. The module consists of two windows, PM and LM, of the same length. The total length of the composite sliding window based on the power threshold is L1, and the total length of the sliding window for THD evaluation is L2. The relationship between L1 and L2 is:

[0069] (3)

[0070] In this experimental example, the instantaneous power threshold is set to 60W, the THD threshold is set to 0.8%, and the total sampling points of the event detection experiment for each load device are 12800. Through reasonable parameter settings, the feasibility verification of the adaptive composite sliding filter event detection device is completed.

[0071] In order to verify the effectiveness of this device in event detection, typical electrical loads in resistive load equipment, motor-driven load equipment and power electronic load equipment were selected in the laboratory, namely, kettle, electric drill and laptop computer. Event detection experiments were carried out through these three electrical loads to complete the feasibility verification of the adaptive composite sliding filter event detection device. Table 2 shows the characteristics of these typical electrical loads.

[0072] Table 2 Typical power load characteristics

[0073]

[0074] The experiment uses two event detection devices: one is the adaptive bilateral composite sliding filter event detection device proposed in this embodiment (hereinafter referred to as "this device"), which is based on power and THD; the other is a power-based bilateral sliding window event detection device (hereinafter referred to as "traditional device"), which is used for comparative experiments. Figure 4 , Figure 5 and Figure 6 In the figure, the round dots are the electrical appliance input points detected by each event detection device, and the square dots are the electrical appliance cut-out points detected.

[0075] Figure 4 (a) and Figure 4(b) shows the experimental results of the traditional device and the present device in event detection under the condition that the kettle is used as the switching appliance. There is no difference in the experimental results between the traditional device and the present device, and both successfully detected the switching point. The switching points detected by the two devices are the same, the input point is

[14239] , and the cut-out point is [104023]. This result shows that when facing the resistive load in event detection, the event detection task can be completed by only detecting the power.

[0076] Figure 5 (a) and Figure 5 (b) The experimental results of event detection of the traditional device and the present device under the condition of the electric drill as the switching electrical appliance are shown respectively. In the traditional device, the electrical appliance input point detection is

[7288] , and in the present device, the electrical appliance input point detection is

[6656] . The judgment of the switching interval is more accurate than that of the traditional device and closer to the actual switching point of the event. The reason for the inaccurate detection of the traditional device is that the first peak of the power after the electrical appliance is switched on fails to reach the set power threshold, and the specific switching point is not found until the second higher peak is reached, which results in the inaccurate switching point interval. However, by monitoring the more sensitive THD value, the present device can detect the electrical appliance switching point before the first peak is reached without resetting the discrimination threshold, avoiding the misjudgment and missed judgment that may be caused by resetting the threshold, and further improving the reliability and accuracy of event detection. The electrical appliance cut-out point detection of the traditional device is

[75941] , and the electrical appliance cut-out point detected by the present device is

[75940] , which is not much different from each other. In summary, this device is more accurate than traditional devices in detecting event intersection intervals.

[0077] Figure 6 (a) with Figure 6 (b) The experimental results of the event detection of the traditional device and the present device when the laptop is used as a switching electrical appliance. The traditional device successfully detects the event when the laptop is connected to the circuit, but due to the characteristics of the laptop itself, the current and power fluctuate frequently in the first few seconds, and the traditional device mistakenly determines it as a switching event during the current and power fluctuations. When the current changes back and forth in the subsequent period, the change never reaches the set power threshold. Especially after the laptop enters the steady state, due to its own characteristics, the transient characteristics when it is switched on and off are significantly different, making it impossible to detect the electrical switch-off point of the laptop through power monitoring. If the power threshold is lowered to detect the switch-off point, it will lead to the misjudgment of too many switch-on points. The present device accurately determines the event switching point by determining the THD change, and the output switching point determination results are

[10625] and

[74624] respectively, which proves that the adaptive composite sliding filter event detection device proposed by the present device maintains excellent accuracy and stability when detecting events for power electronic appliances with variable states.

[0078] In summary, the data obtained from the experimental results are summarized as shown in Table 3 below. In the event detection experiments of the kettle, electric drill, and laptop computer, the total accuracy of the bilateral sliding window event detection device based on the power threshold is only 71.42%, while the total accuracy of the adaptive bilateral composite sliding filter event detection device proposed in this embodiment is 100%, and no misjudgment occurs, proving that the device can accurately distinguish the input and output events of the three load types.

[0079] Table 3 Event detection experimental results

[0080]

[0081] Compared with the traditional power-based bilateral sliding window event detection device, this device uses instantaneous power and total harmonic distortion (THD) for event detection. It can accurately complete the event detection task without adjusting the power threshold, effectively solving the problem of missed and misjudgment caused by differences in power and transient characteristics during load switching. In addition, the NILM event detection device based on adaptive composite sliding filtering combines multiple noise reduction methods such as Butterworth filter, sliding window average and Savitzky-Golay filter, further improving the signal clarity and event detection accuracy. Compared with the traditional one-time noise reduction method, this device does not lose important signal characteristics while ensuring computational efficiency, while reducing the probability of false alarms and missed alarms, enhancing the robustness of the event detection device in complex environments, and improving its application potential in non-invasive load monitoring.

[0082] A person skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as ROM / RAM, disk, CD-ROM, etc.

[0083] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

[0084] It should be noted that the embodiments of the present invention can be implemented by hardware, software or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art will appreciate that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in processor control codes, such as such codes provided on a programmable memory or a data carrier such as an optical or electronic signal carrier.

[0085] In addition, although the operation of the method of the present invention is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired result. On the contrary, the steps described in the flow chart can change the order of execution. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step for execution, and / or one step can be decomposed into multiple steps for execution. It should also be noted that the features and functions of two or more systems according to the present invention can be embodied in one system. On the contrary, the features and functions of a system described above can be further divided into being embodied by multiple systems.

[0086] Although the invention has been described with reference to several specific embodiments, it should be understood that the invention is not limited to the specific embodiments disclosed. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A NILM event detection device based on adaptive composite sliding filtering, characterized in that: It includes a data preprocessing module, a data judgment module and an event detection module, wherein the data preprocessing module, the data judgment module and the event detection module are connected in sequence; The data preprocessing module includes a data acquisition processing module and a data filtering module, and the data acquisition processing module is connected to the data filtering module; The data acquisition and processing module is connected to the load device and is used to receive the voltage and current data of the load device and calculate the instantaneous power according to the voltage and current data; The data filtering module is used to receive the voltage and current data and instantaneous power from the data acquisition and processing module, and perform noise reduction processing on the voltage and current data and instantaneous power, and output the voltage and current data and instantaneous power after noise reduction processing to the data judgment module; The data judgment module is used to generate a phase difference between voltage and current based on the voltage and current data after noise reduction processing from the data preprocessing module, and judge the load type in this time period based on the phase difference, and output the load type to the event detection module; The event detection module includes a selection module, a power threshold bilateral sliding window detection module, a total harmonic distortion bilateral sliding window detection module, and an output module. The selection module is used to select the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module to detect transient events according to the load type determined by the data determination module. When the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module detects that a transient event has occurred in the device, the output module outputs the load switching interval and its characteristic information. Wherein, when the load type is a resistive load, the selection module selects the power threshold bilateral sliding window detection module to be electrically connected to the phase difference judgment module; When the load type is an inductive or capacitive load, the selection module selects the total harmonic distortion bilateral sliding window detection module to be electrically connected to the phase difference judgment module.

2. The NILM event detection device based on adaptive composite sliding filtering according to claim 1, characterized in that: The data filtering module includes a Butterworth filter, a sliding window average module and a Savitzky-Golay filter; The Butterworth filter is used to receive the voltage and current data from the data acquisition and processing module and to reduce noise on the voltage and current data; The sliding window averaging module is used to receive the instantaneous power from the data acquisition module and perform noise reduction and smoothing on the instantaneous power; The Savitzky-Golay filter is used to receive the instantaneous power from the data acquisition module and perform smoothing on the instantaneous power.

3. The NILM event detection device based on adaptive composite sliding filtering according to claim 1, characterized in that: The data judgment module includes a normalization processing module and a phase difference judgment module, the input end of the normalization processing module is electrically connected to the data filtering module, the output end of the normalization processing module is connected to the input end of the phase difference judgment module, and the output end of the phase difference judgment module is electrically connected to the input end of the event detection module; The normalization processing module is used to perform normalization processing on the voltage and current data processed by the data filtering module, and transmit the processed voltage and current data to the phase difference judgment module; The phase difference judgment module is used to calculate the phase difference of the voltage and current data according to the voltage and current data processed by the data filtering module, and to judge the load type in this time period, and output the load type to the event detection module.

4. The NILM event detection device based on adaptive composite sliding filtering according to claim 1, characterized in that: The power threshold bilateral sliding window detection module and the total harmonic distortion bilateral sliding window detection module are both used to determine whether a transient event has occurred in the device according to whether the offset between the PM window and the LM window is greater than two set thresholds; If a transient event is detected, the power threshold bilateral sliding window detection module or the total harmonic distortion bilateral sliding window detection module outputs the transient event interval, the transient event category, and the steady-state characteristics before and after the transient event interval.

5. The NILM event detection device based on adaptive composite sliding filtering according to claim 4 is characterized in that: The power threshold bilateral sliding window detection module is used to use the difference between the mean values ​​of the sampling sequences of the calculation window as the offset between the PM window and the LM window; The total harmonic distortion bilateral sliding window detection module is used to use the difference between the THD values ​​of the calculation windows as the offset between the PM window and the LM window.

6. A NILM event detection method based on adaptive composite sliding filtering, characterized in that: The NILM event detection method based on adaptive composite sliding filtering is used to implement the NILM event detection device based on adaptive composite sliding filtering as claimed in any one of claims 1 to 5; comprising the following steps: The data acquisition and processing module calculates the instantaneous power according to the voltage and current data of the load equipment, and outputs the voltage and current data and the instantaneous power to the data filtering module; The data filtering module receives the voltage and current data and the instantaneous power, performs noise reduction processing on the voltage and current data and the instantaneous power, and outputs the voltage and current data and the instantaneous power after the noise reduction processing to the data judgment module; The data judgment module generates a phase difference between voltage and current according to the voltage and current data after noise reduction processing, and judges the load type in this time period according to the phase difference, and outputs the load type to the selection module; The selection module selects a power threshold bilateral sliding window detection module or a total harmonic distortion bilateral sliding window detection module according to the load type to detect transient events, and outputs the detection result to the output module; When the detection result is that a transient event occurs in the load device, the output module outputs the load switching interval and its characteristic information.

Images