Non-intrusive load monitoring method and system based on multi-dimensional features

Abstract

This invention discloses a non-intrusive load monitoring method and system based on multi-dimensional features. The method can determine the transient response window and steady-state response window of the load access by detecting load access events, and extract multi-dimensional steady-state features of the accessed load based on the steady-state response window and multi-dimensional transient features of the accessed load based on the transient response window. This can accurately determine the timing of transient feature extraction and steady-state feature extraction of the accessed load, and can also fully explore the multi-dimensional features such as steady-state, transient and quantization hidden in high-frequency data, which is beneficial to improving the accuracy of load identification.

Description

Technical Field

[0001] This invention relates to the field of load monitoring technology, and in particular to a non-intrusive load monitoring method and system based on multi-dimensional features, an electronic device, and a computer-readable storage medium. Background Technology

[0002] Non-intrusive load monitoring (NILM) involves installing a detection device at the user's power inlet to detect changes in the user's total current. Through information mining and data analysis, it enables real-time monitoring of load power consumption. It eliminates the need for sensors on each appliance; load information can be derived simply by analyzing the total data collected from the power inlet, reducing deployment costs and thus holding significant application value in fields such as smart grids. Furthermore, with the increasing maturity of high-frequency data acquisition technology, various transient details during load switching are being fully captured, providing new support for improving load identification accuracy. However, existing research utilizing high-frequency data has significant shortcomings: First, most methods focus only on the application of a single type of feature, such as using only transient current amplitude sequences or single steady-state waveforms for load identification. This fails to fully exploit the multi-dimensional features hidden in high-frequency data, including steady-state, transient, and quantization characteristics, resulting in low information utilization efficiency and insufficient accuracy in load identification. In addition, in non-intrusive load identification scenarios, the traditional 0-1 multidimensional knapsack algorithm is often used for load identification. However, the application of this algorithm has significant adaptability defects. This algorithm sets the power grid resource capacity (such as total power and total current threshold) to a fixed value, but ignores the actual scenario characteristics of non-intrusive monitoring. That is, the resource consumption caused by the switching of electrical appliances during residential electricity consumption is significantly uncertain, and the fixed capacity setting is difficult to adapt to the dynamically changing electricity consumption environment. Summary of the Invention

[0003] This invention provides a non-intrusive load monitoring method and system based on multi-dimensional features, an electronic device, and a computer-readable storage medium, which can improve the accuracy of load identification.

[0004] According to one aspect of the present invention, a non-invasive load monitoring method based on multi-dimensional features is provided, comprising the following: Collect electrical data of the current load operation and detect load access events based on the electrical data; The transient and steady-state response windows for load access are determined based on load access events; Multidimensional steady-state features of access load are extracted based on steady-state response window; multidimensional transient features of access load are extracted based on transient response window. Collect electrical data of various loads operating individually and label them accordingly. Extract multidimensional steady-state and multidimensional transient characteristics of various loads. Construct a load sample library based on the multidimensional steady-state, multidimensional transient characteristics and load labels of all loads. Based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, an improved 0-1 multidimensional knapsack problem is used to construct the objective function. The objective function is solved to obtain the load identification results of the access load.

[0005] Furthermore, the objective function is: ; in, Describe the objective function. Indicates the load quantity. Represents the weight of transient features. Represents the steady-state characteristic weights. Indicates load j The transient feature vector and the load samples in the load sample library j The similarity between transient feature vectors, Indicates load j The steady-state eigenvectors and load samples in the load sample library j The Euclidean distance between the steady-state eigenvectors Indicates load j The relevant 0-1 decision variables, if the load is selected j ,but It is 1 if it is true, otherwise it is 0. Indicates load j The transient eigenvectors, Represents the similarity function. This indicates the load samples in the load sample library. j The transient eigenvectors, Indicates load j The steady-state eigenvector, This indicates the load samples in the load sample library. j The steady-state eigenvector, Represents the Euclidean distance function. Represents the actual value of the transient characteristic. The actual value representing the steady-state characteristic. This indicates the load samples in the load sample library. j In the i The mean of transient-like features, This indicates the load samples in the load sample library. j In the The mean of the quasi-steady-state characteristics, Indicates the first i The inherent tolerance of transient-like features Indicates the first The inherent tolerance of quasi-steady-state characteristics This indicates the load samples in the load sample library. j In the i Standard deviation of transient features This indicates the load samples in the load sample library. j In the i Standard deviation of quasi-steady-state characteristics Dimensions representing transient features Dimensions representing steady-state characteristics.

[0006] Furthermore, the process of extracting multidimensional transient features of the access load based on the transient response window includes the following: Extract the transient response current curve within the transient response window, perform EMD decomposition on the transient response current to obtain multi-order harmonic components, perform discrete Hilbert transform on the multi-order harmonic components and construct an analytical signal, and extract the maximum amplitude of each harmonic component as a transient feature.

[0007] Furthermore, the process of determining the transient response window and steady-state response window for load access based on load access events includes the following: Using the detection time of the load connection event as the start time of the transient response window, a sliding window is constructed and a sliding step size is set. The sliding window is continuously slid across the acquired current and power curves. k If the current fluctuation amplitude is always less than the first threshold and the power fluctuation amplitude is always less than the second threshold within each sliding window, then the first... k The right boundary of the first sliding window is taken as the end time of the transient response window, thus determining the transient response window, and the first sliding window is set as the end time of the transient response window. k +1 sliding window as the steady-state response window.

[0008] Furthermore, the current fluctuation amplitude and power fluctuation amplitude are calculated based on the following formula: ; ; in, Indicates the amplitude of current fluctuation. This represents the maximum current within a sliding window. This represents the minimum current within a sliding window. This represents the average current within a sliding window. Indicates the power fluctuation range. This represents the maximum power within a sliding window. This represents the minimum power within a sliding window. This represents the average power within a sliding window.

[0009] Furthermore, the process of detecting load access events based on electrical data specifically involves: determining whether the change in current intensity is greater than a set threshold, where the threshold is 1 / 20 of the current intensity in the previous cycle; if the change is greater than the set threshold, then a load access event is determined to have occurred.

[0010] Furthermore, the multidimensional steady-state characteristics include active power, power factor, correlation coefficient, and effective current value.

[0011] In addition, the present invention also provides a non-intrusive load monitoring system based on multi-dimensional features, comprising: The event detection module is used to collect electrical data of the current load operation and detect load access events based on the electrical data; The response window determination module is used to determine the transient response window and steady-state response window of load access based on the load access event; The feature extraction module is used to extract multi-dimensional steady-state features of the access load based on the steady-state response window and multi-dimensional transient features of the access load based on the transient response window. The sample library construction module is used to collect electrical data of various loads operating individually and affix corresponding load labels, extract multidimensional steady-state characteristics and multidimensional transient characteristics of various loads, and construct a load sample library based on the multidimensional steady-state characteristics, multidimensional transient characteristics and load labels of all loads; The load identification module is used to construct an objective function based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, and to solve the objective function to obtain the load identification results of the access load.

[0012] In addition, the present invention also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the steps of the method described above by calling the computer program stored in the memory.

[0013] In addition, the present invention provides a computer-readable storage medium for storing a computer program for non-intrusive load monitoring based on multi-dimensional features, wherein the computer program executes the steps of the method described above when running on a computer.

[0014] The present invention has the following beneficial effects: The non-intrusive load monitoring method based on multi-dimensional features of the present invention can determine the transient response window and steady-state response window of the load access by detecting load access events, and extract multi-dimensional steady-state features of the accessed load based on the steady-state response window and multi-dimensional transient features of the accessed load based on the transient response window. This method can accurately determine the timing of transient feature extraction and steady-state feature extraction of the accessed load, and can also fully explore the multi-dimensional features such as steady-state, transient and quantization hidden in high-frequency data, which is beneficial to improving the accuracy of load identification.

[0015] In addition, the non-intrusive load monitoring system based on multi-dimensional features of the present invention also has the above-mentioned advantages.

[0016] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating a preferred embodiment of the non-intrusive load monitoring method based on multi-dimensional features of this application. Figure 2 This is a schematic diagram of the module structure of a non-intrusive load monitoring system based on multi-dimensional features, according to another embodiment of this application. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] Reference Figure 1 A preferred embodiment of this application provides a non-intrusive load monitoring method based on multi-dimensional features, including the following: Step S1: Collect electrical data of the current load operation and detect load access events based on the electrical data; Step S2: Determine the transient response window and steady-state response window for load access based on the load access event; Step S3: Extract multi-dimensional steady-state features of the access load based on the steady-state response window, and extract multi-dimensional transient features of the access load based on the transient response window; Step S4: Collect electrical data of various loads operating individually and affix load labels accordingly. Extract multidimensional steady-state and multidimensional transient characteristics of various loads. Construct a load sample library based on the multidimensional steady-state, multidimensional transient characteristics and load labels of all loads. Step S5: Based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, construct the objective function using the improved 0-1 multidimensional knapsack problem, solve the objective function, and obtain the load identification results of the access load.

[0020] It is understood that the non-intrusive load monitoring method based on multi-dimensional features in this embodiment can determine the transient response window and steady-state response window of the load access by detecting load access events, and extract multi-dimensional steady-state features of the accessed load based on the steady-state response window and multi-dimensional transient features of the accessed load based on the transient response window. This can accurately determine the timing of transient feature extraction and steady-state feature extraction of the accessed load, and can also fully explore the multi-dimensional features such as steady-state, transient and quantization hidden in high-frequency data, which is conducive to improving the accuracy of load identification.

[0021] This invention uses residential users as the monitoring object for illustrative purposes, but is not specifically limited thereto. In other embodiments of this invention, the monitoring object can also be a commercial or industrial electricity user. In step S1, a smart meter or monitoring device at the main power line of the household collects electrical data of the current load operation at high frequency. The electrical data includes current, power, and voltage, etc., and load access events are detected based on the collected electrical data. Specifically, it is determined whether the change in current intensity is greater than a set threshold, which is 1 / 20 of the current intensity in the previous cycle. If the change is greater than the set threshold, a load access event is determined to have occurred. It can be understood that when a residential user turns on a new household appliance, it means that a new load has been connected, and the collected current intensity will increase. To avoid false detections caused by current fluctuations, environmental interference, etc., this invention can determine whether a new load has been connected by judging whether the change in current intensity exceeds 1 / 20 of the current intensity in the previous cycle, thereby quickly and accurately detecting load access events.

[0022] Furthermore, in step S2, after detecting a new load connection, to extract the steady-state and transient characteristics of the connected load, it is first necessary to accurately identify when it is transient and when it is steady-state. Therefore, this invention determines the transient response window and steady-state response window of the load connection based on the load connection event, which can accurately determine the timing for extracting the transient and steady-state characteristics of the connected load. The process of determining the transient response window and steady-state response window of the load connection based on the load connection event includes the following: Using the detection time of the load connection event as the start time of the transient response window, a sliding window is constructed and a sliding step size is set. The sliding window is continuously slid across the acquired current and power curves. k If the current fluctuation amplitude is always less than the first threshold and the power fluctuation amplitude is always less than the second threshold within each sliding window, then the first...k The right boundary of the first sliding window is taken as the end time of the transient response window, thus determining the transient response window, and the first sliding window is set as the end time of the transient response window. k +1 sliding window as the steady-state response window.

[0023] Specifically, when a load access event is detected, this moment is taken as the start time T-start of the transient response window. Then, a sliding window is constructed and a sliding step size is set. The length of the sliding window and the sliding step size can be set according to actual needs. The sliding window is used to continuously slide on the high-frequency acquired current curve and power curve, respectively. If in continuous k If the current fluctuation amplitude is always less than the first threshold and the power fluctuation amplitude is always less than the second threshold within each sliding window, then the first... k The right boundary of the first sliding window is taken as the end time T-end of the transient response window, thus determining the transient response window as [T-start, T-end], and the first sliding window is set as [T-start, T-end]. k +1 sliding window is used as the steady-state response window. The values ​​of the first and second thresholds can be adjusted according to the required detection accuracy; preferably, the first threshold is 2% and the second threshold is 3%. Furthermore, the current fluctuation amplitude and power fluctuation amplitude are calculated based on the following formula: ; ; in, Indicates the amplitude of current fluctuation. This represents the maximum current within a sliding window. This represents the minimum current within a sliding window. This represents the average current within a sliding window. Indicates the power fluctuation range. This represents the maximum power within a sliding window. This represents the minimum power within a sliding window. This represents the average power within a sliding window.

[0024] It is understood that this invention comprehensively evaluates the degree of change in transient response from two dimensions: current fluctuation amplitude and power fluctuation amplitude. When both are less than the corresponding preset threshold, it can be regarded as the end of transient response and the entry into steady-state response. The end time of transient response can be accurately determined, thereby accurately determining the transient response window and steady-state response window, which is beneficial to improving the extraction accuracy of transient and steady-state features.

[0025] Furthermore, in step S3, the steady-state characteristics are used to describe the single-cycle current variation characteristics during single-load operation. The multi-dimensional steady-state characteristics extracted by this invention include active power, power factor, correlation coefficient, and effective current value, i.e., from the first...k The above four steady-state features are extracted from the electrical data within a sliding window. The specific extraction process is existing technology and will not be elaborated here. Additionally, transient features describe the dynamic characteristics of drastic parameter changes over a short period. Transient features are typically determined by the internal circuit structure and energy storage characteristics of the load, thus providing effective load identification. Therefore, after determining the transient response window, it is necessary to extract multi-dimensional transient features of the connected load based on the transient response window. The process of extracting multi-dimensional transient features of the connected load based on the transient response window includes the following: Extract the transient response current curve within the transient response window, perform EMD decomposition on the transient response current to obtain multi-order harmonic components, perform discrete Hilbert transform on the multi-order harmonic components and construct an analytical signal, and extract the maximum amplitude of each harmonic component as a transient feature.

[0026] Specifically, first extract the transient response current curve within the transient response window, and then analyze the transient response current... Perform EMD decomposition. The specific EMD decomposition process involves: first finding... Find all the maxima and minima, and fit them with curves to obtain... The upper and lower envelopes are denoted as . and The first of the upper and lower envelopes a The average value of the nth iteration is denoted as , , n This represents the number of samples, and the iterative process can be represented as: , Indicates the first a The approximate IMF function for each iteration is iterated until... Stop when the IMF condition is met. At this point, the first-order IMF function (i.e., the first-order harmonic component) is obtained, denoted as . The IMF conditions include: (1) the number of extreme points and the number of zero crossings are equal or differ by one throughout the entire data sequence; (2) at any time of the signal, the average value of the upper and lower envelopes determined by the local maxima and local minima is zero, i.e., the upper and lower envelopes are locally symmetrical about the time axis; then, As a new Repeat the above steps to obtain the second-order harmonic component. Continue repeating this process to obtain the third through eighth-order harmonic components, which are denoted as follows: Next, the decomposed 1st to 8th order harmonic components were analyzed. Discrete Hilbert transforms are performed on each harmonic component to construct an analytic signal, and the maximum amplitude of each harmonic component is extracted as a transient feature. For the th harmonic component... k First harmonic components Its Hilbert transform is: , , No. k Discrete analytic signal of first harmonic components It can be represented as: The modulus of the analytic signal is the first... k The first harmonic component is in the 1st order. n The transient amplitude of each sampling point can be expressed as: Then, the maximum amplitude of each harmonic component is extracted as a transient feature.

[0027] It is understood that by extracting active power, power factor, correlation coefficient and current RMS value as steady-state features, and extracting the maximum amplitude of the 1st to 8th order harmonic components as transient features, this invention can fully explore the multi-dimensional features such as steady-state, transient and quantization hidden in high-frequency data, which is beneficial to improving the accuracy of load identification.

[0028] In addition, in step S4, electrical data of various loads operating independently are collected using a 10kHz high-frequency acquisition device, and load labels are assigned to the corresponding electrical data. The same feature extraction method as in step S3 is used to extract multidimensional steady-state and multidimensional transient features of various loads, and a load sample library is constructed, which can be represented as follows: ,in, N This indicates the number of load samples, with each load sample representing a specific number of samples. This includes the multidimensional steady-state characteristics and multidimensional transient characteristics corresponding to the load.

[0029] Furthermore, in step S5, the traditional 0-1 multidimensional knapsack problem is defined as: given n1 profit... q j ( j Items with values ​​= 1, 2, ..., n1, m The amount of resources is d i ( i Resources of type 1, 2, ..., n1), the th j The item accounts for the first i The amount of each resource is v ij To maximize the total profit from the loading, the objective function can be expressed as: ,in, Describe the objective function. n Indicates the item number, m Indicates the number of dimensions. Representation Object j Profits Represents a container i capacity, Representation Object jFor containers i Consumption, Representation and object j Related decision variables, if the object is selected j ,but The value is 1 if the condition is met, and 0 otherwise. In the load identification problem, the object is the load. j , For load j Features Is with load j The associated 0-1 decision variables, if the load is selected j ,but It is 1 if it is true, otherwise it is 0. m Each resource constraint corresponds to a number of features. However, the traditional 0-1 multidimensional knapsack problem treats resources in different dimensions as fixed values, ignoring the actual characteristics of non-intrusive monitoring scenarios. Specifically, the resource consumption caused by appliance switching during residential electricity use is significantly uncertain, and fixed resource settings are difficult to adapt to dynamically changing electricity environments, resulting in poor load identification accuracy. Therefore, this invention improves the traditional 0-1 multidimensional knapsack problem by constructing an objective function using an improved 0-1 multidimensional knapsack problem, where the objective function is expressed as: ; in, Describe the objective function. Indicates the load quantity. Represents the weight of transient features. Represents the steady-state characteristic weights. Indicates load j The transient feature vector and the load samples in the load sample library j The similarity between transient feature vectors, Indicates load j The steady-state eigenvectors and load samples in the load sample library j The Euclidean distance between the steady-state eigenvectors Indicates load j The relevant 0-1 decision variables, if the load is selected j ,but It is 1 if it is true, otherwise it is 0. Indicates load j The transient eigenvectors, Represents the similarity function. This indicates the load samples in the load sample library. j The transient eigenvectors, Indicates load j The steady-state eigenvector, This indicates the load samples in the load sample library. j The steady-state eigenvector, Represents the Euclidean distance function. Represents the actual value of the transient characteristic. The actual value representing the steady-state characteristic. This indicates the load samples in the load sample library. j In the i The mean of transient-like features, This indicates the load samples in the load sample library. j In the The mean of the quasi-steady-state characteristics, Indicates the first i The inherent tolerance of transient-like features Indicates the first The inherent tolerance of the quasi-steady-state characteristic is usually a constant, set empirically. This indicates the load samples in the load sample library. j In the i Standard deviation of transient features This indicates the load samples in the load sample library. j In the i Standard deviation of quasi-steady-state characteristics Dimensions representing transient features Dimensions representing steady-state characteristics.

[0030] It is understood that, compared with the traditional 0-1 multidimensional knapsack problem, the improved 0-1 multidimensional knapsack problem of this invention introduces an inherent tolerance term and standard deviation, realizes dynamic adjustment of resource constraints in different dimensions, and combines the weighted fusion of transient and steady-state characteristics, which can adapt to the dynamically changing power environment and improve the accuracy of load identification.

[0031] In addition, after constructing the objective function, a bee colony algorithm that integrates genetic algorithms is used to search for the load combination with the maximum total revenue under feature constraints to complete load identification. The specific process is as follows: 1) Initialization parameters: randomly generated SN A honey source of 0-1 vectors , SN The number of nectar sources, and the length of each vector is [value missing]. n 2 The first vector j* Position x j* This indicates whether to select a load. j* If it is 1, it means the load is selected. j* Otherwise, it will not be selected, and the vector should satisfy all resource constraints, i.e. The control parameters of the bee colony algorithm are initialized, including the colony size, maximum number of loops, maximum number of iterations, and genetic algorithm fusion parameters (crossover probability). Pc Probability of mutation Pm Elite retention rate α For each initial solution, verify whether the constraints are met. If not, regenerate until the constraints are met.

[0032] 2) Set the fitness function in the bee colony algorithm. The fitness function is as follows: Meanwhile, the adaptation rules for genetic operations are defined: the new solution after crossover / mutation must meet the resource constraints, otherwise the constraint repair mechanism is triggered, that is, the load combination with high resources and low returns is eliminated first.

[0033] 3) The three types of bees perform optimization tasks on the 0-1 vector of nectar sources. They calculate fitness values ​​using a fitness function and embed a genetic algorithm into the optimization process to select the optimal solution from all possible solutions, maximizing the fitness function. The optimization process includes the following steps: a) Selection operation: From the current nectar source solution set of the bee colony, select high-quality nectar sources as parent solutions using the roulette wheel method according to fitness value. At the same time, directly retain the nectar sources with the top α% fitness according to the elite retention ratio α to avoid the loss of high-quality solutions; b) Crossover operation: For the selected parent honey source, crossover probability is applied. Pc Perform a single-point crossover (randomly select one load bit as the crossover point, and swap the 0-1 vectors after the crossover points of the two parent honey sources) to generate a child honey source; c) Mutation operation: For the offspring honey sources after crossover, the mutation probability is... Pm Execute bit mutation (randomly select several) x j (Reverse the 0 / 1 state) to generate a new honey source; d) Constraint verification and repair: Verify resource constraints for new nectar sources after crossover and mutation. If the constraints are not met, remove the overload until the constraints are met. Merge the original optimal solution of the bee colony with the new solution generated by genetic operations, calculate the fitness value of all solutions using the fitness function, and select the solution with the largest fitness function as the current optimal solution.

[0034] 4) Determine if the maximum number of iterations for a nectar source has been exceeded. If the number of iterations exceeds the maximum, perform a genetic algorithm to reconstruct the abandoned nectar source: select the best historical nectar source as the parent, perform crossover and mutation operations to generate a new nectar source, replacing the original one. The currently obtained nectar source is the best found nectar source, and determine whether it meets the termination condition based on the loop conditions.

[0035] 5) When the number of iterations reaches the set maximum number of iterations or the optimal solution has not improved for several consecutive times, perform a local optimization of the genetic algorithm on the final candidate optimal solution (i.e., only retain low probability mutation operations) to further improve the benefit level of the solution. After completion, stop the algorithm, output the current optimal load combination, and thus obtain the load identification result.

[0036] In addition, such as Figure 2 As shown, another embodiment of the present invention also provides a non-intrusive load monitoring system based on multi-dimensional features, preferably employing the non-intrusive load monitoring method based on multi-dimensional features as described above, including: The event detection module is used to collect electrical data of the current load operation and detect load access events based on the electrical data; The response window determination module is used to determine the transient response window and steady-state response window of load access based on the load access event; The feature extraction module is used to extract multi-dimensional steady-state features of the access load based on the steady-state response window and multi-dimensional transient features of the access load based on the transient response window. The sample library construction module is used to collect electrical data of various loads operating individually and affix corresponding load labels, extract multidimensional steady-state characteristics and multidimensional transient characteristics of various loads, and construct a load sample library based on the multidimensional steady-state characteristics, multidimensional transient characteristics and load labels of all loads; The load identification module is used to construct an objective function based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, and to solve the objective function to obtain the load identification results of the access load.

[0037] In addition, another embodiment of the present invention provides an electronic device including a processor and a memory, wherein the memory stores a computer program, and the processor executes the steps of the method described above by calling the computer program stored in the memory.

[0038] In addition, another embodiment of the present invention provides a computer-readable storage medium for storing a computer program for non-intrusive load monitoring based on multi-dimensional features, wherein the computer program executes the steps of the method described above when running on a computer.

[0039] Common computer-readable storage media include: floppy disks, flexible disks, hard disks, magnetic tapes, any other magnetic media, CD-ROMs, any other optical media, punch cards, paper tape, any other physical media with perforated patterns, random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), flash erasable programmable read-only memory (FLASH-EPROM), any other memory chips or cartridges, or any other media readable by a computer. Instructions may further be transmitted or received by a transmission medium. The term transmission medium can include any tangible or intangible medium used to store, encode, or carry instructions for execution by a machine, and includes digital or analog carrier communication signals or intangible media that facilitate communication of such instructions. Transmission media include coaxial cables, copper wires, and optical fibers, which contain conductors for transmitting a bus of computer data signals.

[0040] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this 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 this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0041] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0042] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0043] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0044] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0045] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

[0046] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A non-intrusive load monitoring method based on multi-dimensional features, characterized in that, Includes the following: Collect electrical data of the current load operation and detect load access events based on the electrical data; The transient and steady-state response windows for load access are determined based on load access events; Multidimensional steady-state features of access load are extracted based on steady-state response window; multidimensional transient features of access load are extracted based on transient response window. Collect electrical data of various loads operating individually and label them accordingly. Extract multidimensional steady-state and multidimensional transient characteristics of various loads. Construct a load sample library based on the multidimensional steady-state, multidimensional transient characteristics and load labels of all loads. Based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, an improved 0-1 multidimensional knapsack problem is used to construct the objective function. The objective function is solved to obtain the load identification results of the access load.

2. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 1, characterized in that, The objective function is: ； in, Describe the objective function. Indicates the load quantity. Represents the weight of transient features. Represents the steady-state characteristic weights. Indicates load j The transient feature vector and the load samples in the load sample library j The similarity between transient feature vectors, Indicates load j The steady-state eigenvectors and load samples in the load sample library j The Euclidean distance between the steady-state eigenvectors Indicates load j The relevant 0-1 decision variables, if the load is selected j ,but It is 1 if it is true, otherwise it is 0. Indicates load j The transient eigenvectors, Represents the similarity function. This indicates the load samples in the load sample library. j The transient eigenvectors, Indicates load j The steady-state eigenvector, This indicates the load samples in the load sample library. j The steady-state eigenvector, Represents the Euclidean distance function. Represents the actual value of the transient characteristic. The actual value representing the steady-state characteristic. This indicates the load samples in the load sample library. j In the i The mean of transient-like features, This indicates the load samples in the load sample library. j In the The mean of the quasi-steady-state characteristics, Indicates the first i The inherent tolerance of transient-like features Indicates the first The inherent tolerance of quasi-steady-state characteristics This indicates the load samples in the load sample library. j In the i Standard deviation of transient features This indicates the load samples in the load sample library. j In the i Standard deviation of quasi-steady-state characteristics Dimensions representing transient features Dimensions representing steady-state characteristics.

3. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 1, characterized in that, The process of extracting multidimensional transient features of access load based on transient response window includes the following: Extract the transient response current curve within the transient response window, perform EMD decomposition on the transient response current to obtain multi-order harmonic components, perform discrete Hilbert transform on the multi-order harmonic components and construct an analytical signal, and extract the maximum amplitude of each harmonic component as a transient feature.

4. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 1, characterized in that, The process of determining the transient response window and steady-state response window for load access based on load access events includes the following: Using the detection time of the load connection event as the start time of the transient response window, a sliding window is constructed and a sliding step size is set. The sliding window is continuously slid across the acquired current and power curves. k If the current fluctuation amplitude is always less than the first threshold and the power fluctuation amplitude is always less than the second threshold within each sliding window, then the first... k The right boundary of the first sliding window is taken as the end time of the transient response window, thus determining the transient response window, and the first sliding window is set as the end time of the transient response window. k +1 sliding window as the steady-state response window.

5. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 4, characterized in that, The current fluctuation amplitude and power fluctuation amplitude are calculated based on the following formula: ； ； in, Indicates the amplitude of current fluctuation. This represents the maximum current within a sliding window. This represents the minimum current within a sliding window. This represents the average current within a sliding window. Indicates the power fluctuation range. This represents the maximum power within a sliding window. This represents the minimum power within a sliding window. This represents the average power within a sliding window.

6. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 1, characterized in that, The process of detecting load access events based on electrical data is as follows: determine whether the change in current intensity is greater than a set threshold. The threshold is 1 / 20 of the current intensity in the previous cycle. If the change is greater than the set threshold, it is determined that a load access event has occurred.

7. The non-intrusive load monitoring method based on multi-dimensional features as described in claim 1, characterized in that, The multidimensional steady-state characteristics include active power, power factor, correlation coefficient, and effective current value.

8. A non-intrusive load monitoring system based on multi-dimensional features, characterized in that, include: The event detection module is used to collect electrical data of the current load operation and detect load access events based on the electrical data; The response window determination module is used to determine the transient response window and steady-state response window of load access based on the load access event; The feature extraction module is used to extract multi-dimensional steady-state features of the access load based on the steady-state response window and multi-dimensional transient features of the access load based on the transient response window. The sample library construction module is used to collect electrical data of various loads operating individually and affix corresponding load labels, extract multidimensional steady-state characteristics and multidimensional transient characteristics of various loads, and construct a load sample library based on the multidimensional steady-state characteristics, multidimensional transient characteristics and load labels of all loads; The load identification module is used to construct an objective function based on the multidimensional steady-state and multidimensional transient characteristics of the access load and the load sample library, and to solve the objective function to obtain the load identification results of the access load.

9. An electronic device, characterized in that, The method includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the steps of the method as described in any one of claims 1 to 7 by calling the computer program stored in the memory.

10. A computer-readable storage medium for storing a computer program for non-intrusive load monitoring based on multi-dimensional features, characterized in that, The computer program, when run on a computer, performs the steps of the method as described in any one of claims 1 to 7.

Images