An automated feature extraction method and system applicable to non-invasive load identification

Abstract

The present invention discloses an automated feature extraction method and system applicable to non-intrusive load identification, belonging to the technical field of industrial user power load power consumption monitoring and signal processing. The method of the present invention includes: acquiring the voltage and current waveforms of the non-intrusive load on the target power consumption load device side, and determining the start-stop flag time points of the target power consumption load device and the power fluctuation range during stable operation according to the voltage and current waveforms; determining the time-domain and frequency-domain feature difference sequences within a preset time before and after start-stop according to the start-stop flag time points; determining the time-domain and frequency-domain feature difference sequences within a preset time before and after the suspected start-stop flag time points; selecting the feature quantities with correlation coefficients greater than a preset value as identification features, and setting the reference coefficients of the identification features; establishing a feature library for non-intrusive load identification for the identification features with set reference coefficients. The present invention can realize the automatic extraction of the identification features and corresponding reference coefficients of the target device within the set feature thresholds.

Description

Technical Field

[0001] The present invention relates to the technical field of power load monitoring and signal processing for industrial users, and more specifically, to an automated feature extraction method and system suitable for non-intrusive load identification. Background Art

[0002] With the rapid development of the energy Internet, the power grid has accelerated the pace of digital transformation. The means of real-time monitoring and big data collection of users' power loads by the power grid are becoming increasingly rich. In order to provide more diverse energy efficiency services for various users, the feature analysis of users' power loads is the basis and core of various related researches. As an important part of power users, industrial users' electricity load characteristics show a development trend of high proportion and diversification. Implementing the operation state identification of the main electricity loads of industrial users based on non-intrusive load identification technology is an easy-to-maintain and cost-controllable method.

[0003] At present, the research and application of non-intrusive load identification technology in the industry mainly focus on residential users, and the formed feature extraction methods are also based on the electricity consumption characteristics of residential users. The electricity loads of industrial users are quite different from those of residents and are of various types. Therefore, an automated electricity load feature extraction method with high resolution needs to be formed according to the electricity consumption characteristics of industrial users. Summary of the Invention

[0004] In view of the above problems, the present invention proposes an automated feature extraction method suitable for non-intrusive load identification, including:

[0005] Obtain the voltage and current waveforms of the non-intrusive load on the target electricity load device side, and determine the start-stop flag time points of the target electricity load device and the power fluctuation range during stable operation according to the voltage and current waveforms;

[0006] According to the start-stop flag time points, determine the time-domain and frequency-domain feature difference sequences within a preset time before and after start-stop;

[0007] According to the start-stop flag time points and the power fluctuation range, determine the minimum power during stable operation of the target electricity load device. According to the minimum power, determine the suspected start-stop flag time points of the target electricity load device in the voltage and current waveforms of the metering point bus side of the target electricity load device, and determine the time-domain and frequency-domain feature difference sequences within a preset time before and after the suspected start-stop flag time points;

[0008] Perform correlation analysis on the time-domain and frequency-domain feature difference sequences of the start-stop flag time points and the time-domain and frequency-domain feature difference sequences of the suspected start-stop flag time points, determine the correlation coefficient, select the feature quantities with the correlation coefficient greater than the preset value as the identification features, and set the reference coefficients of the identification features;

[0009] Identify the characteristic of the set reference coefficient and establish a characteristic library for non-intrusive load identification.

[0010] Optionally, the preset value is 0.9.

[0011] Optionally, the voltage waveform of the non-intrusive load on the target electrical load device side is used to determine the start-stop flag time point of the target electrical load device, and the voltage waveform and current waveform of the non-intrusive load on the target electrical load device side are used to determine the power fluctuation range during stable operation.

[0012] Optionally, the number of elements in the time-domain and frequency-domain characteristic difference sequence of the start-stop flag time point is not equal to the number of elements in the time-domain and frequency-domain characteristic difference sequence of the suspected start-stop flag time point, and the elements at the missing time points need to be supplemented.

[0013] The present invention also proposes an automated feature extraction system suitable for non-intrusive load identification, including:

[0014] A data acquisition unit 201, which acquires the voltage and current waveforms of the non-intrusive load on the target electrical load device side, and determines the start-stop flag time point of the target electrical load device and the power fluctuation range during stable operation according to the voltage and current waveforms;

[0015] A first calculation unit 202, which determines the time-domain and frequency-domain characteristic difference sequence within a preset time before and after start-stop according to the start-stop flag time point;

[0016] A second calculation unit 203, which determines the minimum power value of the target electrical load device during stable operation according to the start-stop flag time point and the power fluctuation range, determines the suspected start-stop flag time point of the target electrical load device in the voltage and current waveforms of the metering point bus side of the target electrical load device according to the minimum power value, and determines the time-domain and frequency-domain characteristic difference sequence within a preset time before and after the suspected start-stop flag time point;

[0017] A feature extraction unit 204, which performs a correlation analysis on the time-domain and frequency-domain characteristic difference sequence of the start-stop flag time point and the time-domain and frequency-domain characteristic difference sequence of the suspected start-stop flag time point, determines the correlation coefficient, selects the feature quantity with the correlation coefficient greater than the preset value as the identification feature, and sets the reference coefficient of the identification feature;

[0018] A feature library establishment unit 205, which establishes a feature library for non-intrusive load identification for the identification feature with the set reference coefficient.

[0019] Optionally, the preset value is 0.9.

[0020] Optionally, the voltage waveform of the non-intrusive load on the target electrical load device side is used to determine the start-stop flag time point of the target electrical load device, and the voltage waveform and current waveform of the non-intrusive load on the target electrical load device side are used to determine the power fluctuation range during stable operation.

[0021] Optionally, the number of elements in the time-domain and frequency-domain feature difference sequence of the start-stop flag time point is not equal to the number of elements in the time-domain and frequency-domain feature difference sequence of the suspected start-stop flag time point, and elements at the missing time points need to be supplemented.

[0022] The present invention can realize the automatic extraction of the identification features and corresponding reference coefficients of the target device within the set feature threshold, filter out non-core feature information, and ensure the high identification performance and low resource occupancy rate of the subsequent identification algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart of the method of the present invention;

[0024] Figure 2 is a structural diagram of the system of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0025] Now, exemplary embodiments of the present invention will be introduced with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely, and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the drawings are not intended to limit the present invention. In the drawings, the same unit / element is denoted by the same reference numeral.

[0026] Unless otherwise specified, the terms (including scientific and technical terms) used herein have the ordinary meaning understood by those skilled in the art. In addition, it can be understood that the terms defined in the commonly used dictionary should be understood to have a meaning consistent with the context of their related fields, and should not be understood as idealized or overly formal meanings.

[0027] The present invention will be further described below in conjunction with the embodiments and the accompanying drawings:

[0028] The present invention proposes an automatic feature extraction method applicable to non-intrusive load identification, as Figure 1 shown, including:

[0029] Step 1: According to the voltage and current waveforms on the target electrical load device side, carry out corresponding research work to obtain the start-stop flag time point of the target load and the power fluctuation range during stable operation;

[0030] Step 2: According to the minimum power value of the target load during stable operation, obtain the start and stop time points of the suspected target load from the voltage and current waveforms on the bus side of the user metering point, and calculate the differences in time domain and frequency domain characteristics before and after the start and stop time points of the suspected target load;

[0031] Step 3: According to the start and stop flag time points of the target load, obtain the differences in time domain and frequency domain characteristics of the target load before and after start and stop from the voltage and current waveforms on the bus side of the user metering point;

[0032] Step 4: Conduct a correlation analysis on the difference sequences in Step 2 and Step 3, select the characteristic quantities with a correlation coefficient greater than 0.9 as the subsequent identification characteristics, and complete the setting of the reference coefficients of the identification characteristics;

[0033] Step 5: Establish an identification characteristic library for the target electrical equipment in the target industry.

[0034] Among them, Step 1 includes the following steps:

[0035] Step 1.1: According to the current waveform on the target electrical load side, obtain the specific start and stop time points of the target load:

[0036]

[0037]

[0038]

[0039] Among them, f0 is the sampling frequency, is the effective current value at time point t n i kn is the instantaneous current value collected in real time in the interval from t n to t n+1 When the current value at time point t n satisfies Equation (2), then this time point is the start point of the load. When the current value at time point t n satisfies Equation (3), then this time point is the stop point of the load.

[0040] Step 1.2: Obtain the steady-state operation time point of the target load and the time period of the transient process:

[0041]

[0042]

[0043]

[0044] In the formula, ΔI nThe difference in the effective value of the current between consecutive time points, M is the set determination threshold, where those satisfying Equation (5) and Equation (6) are respectively the steady-state entry point and the steady-state end point of the load. Subtract them from the start and stop time points of the load respectively to obtain the time period sequence T of the transient process n =[t1,t2,...,t n .

[0045] Step 1.2: According to the voltage and current waveforms of the target electrical load, obtain the real-time active power waveform of the target load:

[0046]

[0047] is the active power value in the time interval of t n in the transient process time series, i k and u k are respectively the instantaneous current and voltage values collected in real time in the time interval of t n in the transient process time series, f0 is the sampling frequency.

[0048] Step 1.3: According to the marked steady-state entry point and steady-state end point, obtain the power extreme values of the target electrical load in different steady-state operation intervals, and finally take the union to obtain the maximum steady-state operation power p smax and the minimum value p smin .

[0049] Furthermore, the said Step 2 includes the following steps:

[0050] Step 2.1: According to the voltage and current waveforms on the bus side, obtain the real-time active power waveform on the bus side, and perform power conversion according to the transformation ratios of PT and CT:

[0051]

[0052] In the formula, K pt and K ct are respectively the transformation ratios of PT and CT.

[0053] Step 2.2: Set the determination time window T0 according to the transient process time period of the target electrical load, is the power value sequence within this time window, calculate the power difference Δp′ n of t n within this time window, and refer to the power range of the stable operation of the target load. Obtain the start and stop time point sequence T n ′=[t1′,t2′,...,t m ′] according to the following determination conditions:

[0054]

[0055] k1×p smin <Δp′ n <k2×p smax (10)

[0056] Wherein, k1 and k2 are set parameters. It is recommended that the data of k1 be taken from the range (0, 1), and the data of k2 be taken from the range (1, 2), and p smin and p smax are the maximum and minimum steady-state operating powers respectively.

[0057] Step 2.3: Calculate the differences in time domain and frequency domain characteristics before and after the start-stop time points of the suspected target load to form a difference sequence ΔSP′ = [Δsp′1, Δsp′2,..., Δsp′ m :

[0058]

[0059] Among them, the characteristic quantity sequence sp at each moment is composed of K-dimensional characteristic quantities. Then, within the time window T0, and respectively take the extreme values of the corresponding characteristic quantities in each dimension within the time window T0:

[0060] sp = [sp1, sp2,..., sp K (12)

[0061] Among them, the calculation of the characteristic difference in Step 3 should be carried out according to the method described in Step 2.3 to form a characteristic difference sequence ΔSP = [Δsp1, Δsp2,..., Δsp n :

[0062]

[0063] Among them, Step 4 includes the following steps:

[0064] Step 4.1: Since Sort by the time in the start-stop time point sequence T n ′ of the suspected target load, and fill in zeros at the missing time points in the characteristic difference sequence ΔSP of the target load start-stop, so that the number of elements in the characteristic difference sequence ΔSP is equal to that of ΔSP′;

[0065] Step 4.2: Conduct a correlation analysis on the characteristic difference sequences ΔSP and ΔSP′ to obtain a correlation coefficient sequence ρ = [ρ1, ρ2,..., ρ K :

[0066]

[0067] Step 4.3: Sort the correlation coefficients of each dimensional feature quantity according to the correlation coefficient sequence ρ. Generally, select the feature quantities with a correlation coefficient greater than 0.9 as the subsequent identification features, and set the target feature quantity coefficients according to the following determination conditions:

[0068]

[0069] where A is the number of selected identification features, k i is the finally set target feature quantity coefficient, ρ i is the correlation coefficient corresponding to the selected identification feature, and a is the proportional parameter, which is obtained by calculating the above formula.

[0070] Among them, the feature library established in Step 5 needs to include at least the basic information of the target load and the electricity consumption characteristics. The basic information of the equipment needs to include the industry to which the equipment belongs, the production steps to which the equipment belongs, the equipment name and the equipment category. Among the electricity consumption characteristics, take the corresponding value of the target feature quantity coefficient as the identification feature, and take the corresponding value of the non-identification feature as zero.

[0071] The present invention also proposes an automated feature extraction system 200 applicable to non-intrusive load identification, as Figure 2 shown, including:

[0072] A data acquisition unit 201, which acquires the voltage and current waveforms of the non-intrusive load on the target power consumption load equipment side, and determines the start-stop flag time point of the target power consumption load equipment and the power fluctuation range during stable operation according to the voltage and current waveforms;

[0073] A first calculation unit 202, which determines the time-domain and frequency-domain feature difference sequences within a preset time before and after start-stop according to the start-stop flag time point;

[0074] A second calculation unit 203, which determines the minimum power value of the target power consumption load equipment during stable operation according to the start-stop flag time point and the power fluctuation range, determines the suspected start-stop flag time point of the target power consumption load equipment in the voltage and current waveforms of the metering point bus side of the target power consumption load equipment according to the minimum power value, and determines the time-domain and frequency-domain feature difference sequences within a preset time before and after the suspected start-stop flag time point;

[0075] A feature extraction unit 204, which performs a correlation analysis on the time-domain and frequency-domain feature difference sequences of the start-stop flag time point and the time-domain and frequency-domain feature difference sequences of the suspected start-stop flag time point, determines the correlation coefficient, selects the feature quantities with a correlation coefficient greater than the preset value as the identification features, and sets the reference coefficients of the identification features;

[0076] A feature library establishment unit 205, which establishes a feature library for non-intrusive load identification for the identification features with set reference coefficients.

[0077] Among them, the preset value is 0.9.

[0078] Among them, the voltage waveform of the non-intrusive load on the target electrical load device side is used to determine the start-stop flag time point of the target electrical load device, and the voltage waveform and current waveform of the non-intrusive load on the target electrical load device side are used to determine the power fluctuation range during stable operation.

[0079] Among them, the number of elements in the time-domain and frequency-domain feature difference sequence of the start-stop flag time point is not equal to the number of elements in the time-domain and frequency-domain feature difference sequence of the suspected start-stop flag time point, and the elements at the missing time points need to be supplemented.

[0080] The present invention can realize the automatic extraction of the identification features and corresponding reference coefficients of the target device within the set feature threshold, filter out non-core feature information, and ensure the high identification performance and low resource occupancy rate of the subsequent identification algorithm.

[0081] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0082] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0083] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions in Figure 1 one process or multiple processes and / or blocksFigure 1 The functions specified in one or more boxes.

[0084] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide for implementing the steps of the functions specified in one or more processes and / or boxes Figure 1 One process or more processes and / or boxes Figure 1 The steps of the functions specified in one or more boxes.

[0085] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0086] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. An automated feature extraction method applicable to non-intrusive load identification, the method comprising: Obtaining the voltage and current waveforms of the non-intrusive load on the target power consumption load device side, and determining the start-stop flag time points of the target power consumption load device and the power fluctuation range during stable operation according to the voltage and current waveforms; Determining the time-domain and frequency-domain feature difference sequences within a preset time before and after start-stop according to the start-stop flag time points; The voltage waveform of the non-intrusive load on the target power consumption load device side is used to determine the start-stop flag time points of the target power consumption load device, and the voltage waveform and current waveform of the non-intrusive load on the target power consumption load device side are used to determine the power fluctuation range during stable operation; The number of elements in the time-domain and frequency-domain feature difference sequences of the start-stop flag time points is not equal to the number of elements in the time-domain and frequency-domain feature difference sequences of the suspected start-stop flag time points, and elements at the missing time points need to be supplemented; According to the start-stop flag time points and the power fluctuation range, determining the minimum power during stable operation of the target power consumption load device, and according to the minimum power, determining the suspected start-stop flag time points of the target power consumption load device in the voltage and current waveforms of the metering point bus side of the target power consumption load device, and determining the time-domain and frequency-domain feature difference sequences within a preset time before and after the suspected start-stop flag time points, including: Obtaining the real-time active power waveform of the bus side according to the voltage and current waveforms of the bus side, and performing power conversion according to the transformation ratios of PT and CT; where K pt and K ct are the transformation ratios of the PT and CT, respectively; Set the determination time window T0 according to the transient process time period of the target electrical load. For the power value sequence within this time window, calculate t n The power difference Δp within this time window n ′. Refer to the power range of the stable operation of the target load, and obtain the start-stop time point sequence T n ′ = [t1′, t2′,..., t m ′]: k1×p smin <Δp n ′<k2×p smax Wherein, k1 and k2 are set parameters. It is recommended that the data of k1 be taken from the range (0, 1), and the data of k2 be taken from the range (1, 2). p smin and p smax are the maximum and minimum steady-state operating powers; Calculate the differences in time-domain and frequency-domain features before and after the start-stop time points of the suspected target load to form a difference sequence ΔSP′ = [Δsp1′, Δsp′2,..., Δsp′ m : Among them, the feature quantity sequence sp at each moment is composed of K-dimensional feature quantities, and the and respectively take the extreme values of the corresponding feature quantities in each dimension within the time window T0: sp = [sp1, sp2,..., sp K ​ Performing correlation analysis on the time-domain and frequency-domain feature difference sequences of the start-stop flag time points and the time-domain and frequency-domain feature difference sequences of the suspected start-stop flag time points, determining the correlation coefficient, selecting the feature quantities with the correlation coefficient greater than the preset value as the identification features, and setting the reference coefficients of the identification features; Establishing a feature library for non-intrusive load identification for the identification features with set reference coefficients.

2. The method according to claim 1, wherein the preset value is 0.

9.

3. An automated feature extraction system applicable to non-intrusive load identification, the system comprising: A data acquisition unit, which obtains the voltage and current waveforms of the non-intrusive load on the target power consumption load device side, and determines the start-stop flag time points of the target power consumption load device and the power fluctuation range during stable operation according to the voltage and current waveforms; A first calculation unit, which determines the time-domain and frequency-domain feature difference sequences within a preset time before and after start-stop according to the start-stop flag time points; The voltage waveform of the non-intrusive load on the target power consumption load device side is used to determine the start-stop flag time points of the target power consumption load device, and the voltage waveform and current waveform of the non-intrusive load on the target power consumption load device side are used to determine the power fluctuation range during stable operation; The number of elements in the time-domain and frequency-domain feature difference sequences of the start-stop flag time points is not equal to the number of elements in the time-domain and frequency-domain feature difference sequences of the suspected start-stop flag time points, and elements at the missing time points need to be supplemented; The second calculation unit determines the minimum power value of the target power consumption load device during stable operation according to the start-stop flag time point and the power fluctuation range, determines the suspected start-stop flag time point of the target power consumption load device in the voltage and current waveforms on the bus side of the metering point of the target power consumption load device according to the minimum power value, and determines the time domain and frequency domain feature difference sequences within a preset time before and after the suspected start-stop flag time point, including: Obtain the real-time active power waveform on the bus side according to the voltage and current waveforms on the bus side, and perform power conversion according to the transformation ratios of PT and CT: Where K pt and K ct are the transformation ratios of the PT and CT respectively; Set the determination time window T0 according to the transient process time period of the target electrical load. For the power value sequence within this time window, calculate t n The power difference Δp within this time window n ′. Refer to the power range of the stable operation of the target load, and obtain the start-stop time point sequence T n ′ = [t1′, t2′,..., t m ′] according to the following determination conditions: k1×p smin <Δp n ′<k2×p smax Wherein, k1 and k2 are set parameters. It is recommended that the data of k1 be taken from the range (0, 1), and the data of k2 be taken from the range (1, 2), p smin and p smax are the maximum and minimum steady-state operating powers; Calculate the differences in time-domain and frequency-domain features before and after the start-stop time points of the suspected target load to form a difference sequence ΔSP′ = [Δsp1′, Δsp′2,..., Δsp′ m : Among them, the feature quantity sequence sp at each moment is composed of K-dimensional feature quantities, so the and respectively take the extreme values of the corresponding feature quantities in each dimension within the time window T0: sp = [sp1, sp2,..., sp K ​ The feature extraction unit performs a correlation analysis on the time domain and frequency domain feature difference sequences at the start-stop flag time point and the time domain and frequency domain feature difference sequences at the suspected start-stop flag time point, determines the correlation coefficient, selects the feature quantity with the correlation coefficient greater than the preset value as the identification feature, and sets the reference coefficient of the identification feature; The feature library establishment unit establishes a feature library for non-intrusive load identification for the identification features with the set reference coefficient.

4. The system according to claim 3, wherein the preset value is 0.9.

Images