A mixed electric signal disassembling method and system based on resident electricity load

Abstract

A mixed electric signal disassembling method and system based on residential electricity load, characterized in that the method comprises the following steps: step 1, collecting the mixed electric signal of the residential electricity node in the power distribution network through the load monitoring device, and adding random noise to the mixed electric signal; step 2, extracting signal fluctuation trend information from the mixed electric signal with random noise added, and obtaining a pre-disassembled signal; step 3, determining whether the pre-disassembled signal meets the general index of alternating current signal, thereby obtaining a first disassembled signal; step 4, sequentially extracting a second disassembled signal and subsequent disassembled signals by using the methods in steps 2 and 3; step 5, extracting first and second characteristic values, and obtaining effective disassembled signals; step 6, using a characteristic separation matrix to segment each of the effective disassembled signals to generate a plurality of single load electric signals, thereby completing the disassembly of the mixed electric signal.

Description

Technical Field

[0001] This invention relates to the field of power systems, and more specifically, to a method and system for disassembling hybrid electrical signals based on residential electricity load. Background Technology

[0002] Residential load is a crucial component of urban load, and investigating and forecasting its characteristics is essential for urban power grid planning, economic operation, and electricity market marketing. In-depth analysis of regional load characteristics allows for effective load monitoring, improves demand-side electricity consumption, and stabilizes the overall power grid load, thereby enhancing social benefits.

[0003] Currently, with the large-scale development of power distribution units within residential areas, residential electricity load will have a significant impact on the safe and economical operation of the power distribution network. The increasing number of residential communities and the variability in electricity consumption patterns have led to numerous unsolvable problems in the design of power supply and distribution systems within residential areas using existing technologies. How to provide residents with a safer, more stable, and reliable power supply and distribution system, as well as high-quality power supply, has become an urgent issue to be addressed in engineering practice.

[0004] Numerous studies have yielded fruitful results regarding load characteristic analysis of power systems. While these findings offer valuable insights for analyzing load characteristics in residential areas, they also have their limitations and shortcomings, leading to generally inflated load indices. If the electricity consumption of each individual load can be effectively and accurately separated from residential electricity load, the overall electricity consumption of the residential area can be predicted more accurately, thereby ensuring the goodness of fit and fitness of the prediction model.

[0005] However, existing technologies still lack an effective algorithm that can accurately break down residential electricity loads, and cannot delve into the level of breaking down residential electricity loads down to the load of each electricity meter and each actual electrical device.

[0006] To address the aforementioned issues, a new method and system for disassembling hybrid electrical signals based on residential electricity load is proposed. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a method and system for splitting mixed electrical signals based on residential electricity load. By collecting mixed electrical signals and obtaining multiple split signals through iterative processing, the validity of the split signals is determined and further segmented to ultimately obtain the single load electrical signals from the residential electricity load.

[0008] The present invention adopts the following technical solution.

[0009] The first aspect of this invention relates to a method for disassembling a mixed electrical signal based on residential electricity load. The method includes the following steps: Step 1, collecting mixed electrical signals from residential electricity nodes in a distribution network using load monitoring equipment, and adding random noise to the mixed electrical signals; Step 2, extracting signal fluctuation trend information from the mixed electrical signals with added random noise, and removing the signal fluctuation trend information from the mixed electrical signals to obtain a pre-disassembled signal; Step 3, determining whether the pre-disassembled signal meets the general indicators of AC electrical signals. If it does not meet the general indicators, re-extracting the signal fluctuation trend information based on the pre-disassembled signal, and iteratively updating the pre-disassembled signal until the pre-disassembled signal meets the general indicators of AC electrical signals. Step 4: Remove the first disassembly signal from the mixed electrical signal with added random noise, and extract the second disassembly signal and subsequent disassembly signals sequentially using the methods in Steps 2 and 3, until the mixed electrical signal with added random noise can no longer be disassembled; Step 5: Extract the first feature value and the second feature value from the multiple disassembly signals, and first remove the interference terms in the disassembly signal based on the first feature value, and then obtain the effective disassembly signal from the disassembly signal after removing the interference terms based on the second feature value; Step 6: Use the feature separation matrix to segment each effective disassembly signal to generate multiple single-load electrical signals, thereby completing the disassembly of the mixed electrical signal.

[0010] Preferably, the method for extracting signal fluctuation trend information is as follows: extracting local extrema of the fluctuation signal from the mixed electrical signal with added random noise; extracting the time point corresponding to the local extrema, and determining the shape of the local extrema based on the curve slope before the time point and the curve slope after the time point; adding the local extrema to the first morphology set or the second morphology set based on the determination, and generating signal fluctuation trend information based on the correspondence between the first morphology set and the second morphology set respectively.

[0011] Preferably, the general indicators of the AC signal include a first general indicator and a second general indicator; wherein, the first general indicator is used to determine whether the phase of the pre-disassembly signal is opposite between adjacent local extreme points, and if so, the pre-disassembly signal is determined to meet the first general indicator; the second general indicator is used to determine whether the upper and lower envelopes of the pre-disassembly signal are symmetrical, and if so, the pre-disassembly signal is determined to meet the second general indicator.

[0012] Preferably, adding random noise to the mixed electrical signal further includes: adding different random noises to the mixed electrical signal multiple times to generate multiple different mixed electrical signals with added random noise; the method further includes: using steps 2-4 to decompose the mixed electrical signal based on multiple different mixed electrical signals with added random noise; and the multiple decomposed signals further include generating multiple corresponding decomposed signals based on the decomposition of multiple different mixed electrical signals with added random noise; summing and averaging each set of corresponding decomposed signals to generate multiple decomposed signals.

[0013] Preferably, a correlation coefficient is obtained between any one of the multiple disassembled signals and the mixed electrical signal; if the correlation coefficient is less than the first characteristic value, then any one of the signals is determined to be an interference term.

[0014] Preferably, a disassembly signal dataset is constructed based on the disassembly signals after excluding interference terms. After removing the central feature from each disassembly signal, the feature vector of the disassembly signal dataset is calculated. The feature vectors are sorted according to their contribution rate, and effective feature vectors with a cumulative contribution rate greater than the second feature value are extracted. Each effective disassembly signal is generated based on each effective feature vector.

[0015] Preferably, an original separation matrix is ​​constructed based on the effective decomposition signal, and the original separation matrix is ​​input into the fast ICA algorithm to complete the segmentation of the effective decomposition signal.

[0016] Preferably, a median filtering algorithm is used to smooth the mixed electrical signal with added random noise first, and then the signal fluctuation trend information is extracted.

[0017] Preferably, an adaptive sliding window algorithm is used to customize the window size of the median filtering algorithm.

[0018] A second aspect of this invention relates to a hybrid electrical signal decomposition system based on residential electricity load. The system implements the hybrid electrical signal decomposition method based on residential electricity load described in the first aspect of this invention. The system includes a noise addition unit, a pre-decomposition unit, a cyclic decomposition unit, a decomposition termination unit, a validity determination unit, and a segmentation unit. The noise addition unit collects hybrid electrical signals from an electricity information acquisition system and adds random noise to the hybrid electrical signals. The pre-decomposition unit extracts signal fluctuation trend information from the hybrid electrical signals with added random noise and removes the signal fluctuation trend information from the hybrid electrical signals to obtain a pre-decomposition signal. The cyclic decomposition unit determines whether the pre-decomposition signal meets the general indicators of AC electrical signals; if not, it re-extracts the signal fluctuation trend based on the pre-decomposition signal. The system extracts information and iteratively updates the pre-disassembly signal until it meets the general indicators of AC signals, thereby obtaining the first disassembly signal; a disassembly termination unit is used to remove the first disassembly signal from the mixed electrical signal with added random noise, and sequentially extracts the second disassembly signal and subsequent disassembly signals using the methods in steps 2 and 3, until the mixed electrical signal with added random noise can no longer be disassembled; a valid determination unit is used to extract the first feature value and the second feature value from multiple disassembly signals, and firstly removes interference terms from the disassembly signal based on the first feature value, and then obtains the valid disassembly signal from the disassembly signal after removing interference terms based on the second feature value; a segmentation unit is used to segment each valid disassembly signal using a feature separation matrix to generate multiple single-load electrical signals, thereby completing the disassembly of the mixed electrical signal.

[0019] The beneficial effects of this invention are that, compared with the prior art, the hybrid electrical signal decomposition method and system based on residential electricity load of this invention can obtain multiple decomposition signals by collecting hybrid electrical signals and using a cyclic iterative approach, determine the validity of the decomposition signals, and further segment the decomposition signals to ultimately obtain the single load electrical signals in the residential electricity load. This invention is effective and reliable, enabling accurate and reasonable allocation of the overall load from the perspective of individual user-side devices, providing more possibilities for demand-side energy management in smart grids. Attached Figure Description

[0020] Figure 1 This is a schematic diagram illustrating the steps of a hybrid electrical signal decomposition method based on residential electricity load according to the present invention;

[0021] Figure 2 This is a schematic diagram of the architecture of a hybrid electrical signal splitting system based on residential electricity load according to the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this invention are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments not described in this invention obtained by those skilled in the art based on the embodiments described in this invention without creative effort should fall within the protection scope of this invention.

[0023] Figure 1 This is a schematic diagram illustrating the steps of a hybrid electrical signal decomposition method based on residential electricity load according to the present invention. Figure 1 As shown, the first aspect of the present invention relates to a method for disassembling mixed electrical signals based on residential electricity load, the method comprising steps 1 to 6.

[0024] Step 1: Collect mixed electrical signals from residential electricity consumption nodes in the power distribution network using load monitoring equipment, and add random noise to the mixed electrical signals.

[0025] In this invention, various existing load monitoring devices can be used to collect mixed electrical signals. For example, existing non-intrusive load identification devices can monitor and analyze data such as voltage and current at the load inlet of power users to identify the type of user's electrical equipment and perform sub-item energy calculations. Non-intrusive load monitoring devices can analyze and process the data from the total load meter to identify each electrical device and its operating status. The load monitoring devices mentioned in this invention are not limited to the above-mentioned devices; existing data analysis platforms and systems in the power system, such as electricity information acquisition systems, can also be used to collect mixed electrical signals at a certain node of the distribution network to be analyzed.

[0026] A node here can be a node equipped with load monitoring equipment, such as a regional node in a residential power grid, or a lower-level node within the residential power grid. Due to limitations imposed by the location of the load monitoring equipment, the power consumption data collected at the nodes in this invention typically cannot represent the power consumption of just one device. Generally speaking, the power consumption here will vary periodically or non-periodically at different points in time, depending on factors such as the power consumption characteristics of multiple devices and user habits.

[0027] Therefore, if the situation of multiple electrical loads under a node can be accurately broken down, the inherent patterns of electricity consumption for each load can be analyzed individually. In this way, through these rationally extracted inherent patterns, the power system can comprehensively analyze and predict the electricity consumption characteristics of each load at future times under different environmental, climatic, economic and other objective factors, and obtain the comprehensive electricity consumption situation of the node in the future time frame of the power grid by simple summation.

[0028] As mentioned above, in order to reasonably predict the future electricity consumption characteristics of residential electricity consumption nodes, it is necessary to collect complex mixed electrical signals at the nodes and obtain the individual electricity consumption of each individual load through reasonable decomposition methods.

[0029] In this invention, after acquiring the aforementioned mixed electrical signal, accurate decomposition requires first performing reasonable analysis and removal of the noise components. However, the load decomposition desired in this invention will not involve excessive acquisition of other information at the nodes, such as introducing complex distribution network topology into the algorithm. Therefore, this load decomposition method is quite indiscriminate before decomposition, both for the load signal and the noise signal. Therefore, this invention employs multiple additions of random noise to achieve reasonable and effective filtering of the noise components. The specific implementation of this part will be described in detail later; here, it is only stated that the method of this invention first adds a portion of noise to the mixed electrical signal, and only on this basis is the signal feature extraction achieved from the mixed electrical signal with added noise.

[0030] Preferably, a median filtering algorithm is used to smooth the mixed electrical signal with added random noise first, and then the signal fluctuation trend information is extracted.

[0031] It should be noted that in this invention, after adding noise, the mixed electrical signal can be further preprocessed, such as smoothing the acquired discrete digital electrical signal. Various filtering methods can be used to smooth the electrical signal in this invention, but the median filtering algorithm provides the best smoothing effect.

[0032] In one embodiment of the present invention, an 11th-order one-dimensional median filter is selected.

[0033] Preferably, an adaptive sliding window algorithm is used to customize the window size of the median filtering algorithm.

[0034] The median filtering algorithm in this invention does not simply determine the window size. To improve the smoothing effect while minimizing changes to the original mixed electrical signal, the window size used in this invention is adaptively adjustable. Specifically, this adjustment method can be found in existing technologies.

[0035] It should be noted that in one embodiment of the present invention, the initial size of the window is 3600 pixels, and the sliding step size of the window adjustment is also 3600 pixels. That is, based on the initial size of the window, the actual size of the window can be increased or decreased by multiples according to the value of the pixel grayscale value within the window, with 3600 as the minimum unit.

[0036] Step 2: Extract signal fluctuation trend information from the mixed electrical signal with added random noise, and remove the signal fluctuation trend information from the mixed electrical signal to obtain the pre-disassembled signal.

[0037] After adding noise and smoothing the original signal, this invention can extract signal fluctuation trend information from the pre-processed mixed electrical signal. This signal fluctuation trend information is feature data extracted from the signal pattern of the mixed electrical signal. This information can serve as feature data distinguishing it from individual electrical loads and as a general information feature of a large number of superimposed loads, providing a basis for the actual signal decomposition in subsequent steps.

[0038] Preferably, the method for extracting signal fluctuation trend information is as follows: extracting local extrema of the fluctuation signal from the mixed electrical signal with added random noise; extracting the time point corresponding to the local extrema, and determining the shape of the local extrema based on the curve slope before the time point and the curve slope after the time point; adding the local extrema to the first morphology set or the second morphology set based on the determination, and generating signal fluctuation trend information based on the correspondence between the first morphology set and the second morphology set respectively.

[0039] Understandably, in order to obtain general information about the wave signal, this invention first extracts local extrema from the signal. These local extrema can be local maxima or local minima of the wave signal. Once the shape of the local extrema is determined, the method of this invention can also perform preliminary classification of the local extrema, for example, obtaining a first set of shapes and a second set of shapes.

[0040] Generally, since each local extremum in the first and second morphological sets corresponds to the original sampling time of the extremum data, this invention can construct a smooth signal waveform from all the original data in the first morphological set using methods in the prior art, similar to filtering or analog-to-digital conversion. Similarly, the second morphological set can also be reproduced as a continuous, smooth signal waveform.

[0041] By comparing the two signal waveforms, we can obtain signal fluctuation trend information jointly constructed by the two signals by adding weights and superimposing them. For example, by averaging the values ​​of the two signal waveforms at each time step, we can obtain an intermediate signal, which is a special case of signal fluctuation trend information in this invention.

[0042] Generally speaking, signal fluctuation trend information can characterize the power consumption of a typical node after the superposition of numerous loads. Therefore, after extracting this general information, removing this general information from the original signal superimposed with noise can initially realize a characteristic signal, which is the signal to be decomposed mentioned earlier.

[0043] Step 3: Determine whether the pre-disassembly signal meets the general indicators of AC signals. If it does not meet the general indicators of AC signals, extract the signal fluctuation trend information again based on the pre-disassembly signal and iteratively update the pre-disassembly signal until the pre-disassembly signal meets the general indicators of AC signals, thereby obtaining the first disassembly signal.

[0044] Preferably, the general indicators of the AC signal include a first general indicator and a second general indicator; wherein, the first general indicator is used to determine whether the phase of the pre-disassembly signal is opposite between adjacent local extreme points, and if so, the pre-disassembly signal is determined to meet the first general indicator; the second general indicator is used to determine whether the upper and lower envelopes of the pre-disassembly signal are symmetrical, and if so, the pre-disassembly signal is determined to meet the second general indicator.

[0045] It is understood that the general indicators of the AC signal in this invention may include a variety of different indicators. This invention only uses the first and second general indicators as examples to illustrate the determination process of this pre-disassembled signal.

[0046] Specifically, the first general indicator is used to determine whether, in this pre-disassembled signal, it can cross the zero point of the signal between two consecutive monotonic extrema. In other words, although the form of load power consumption data in the power system varies greatly, for each load power consumption data point, its power form should be able to meet the basic AC characteristics at the grid connection node.

[0047] On the other hand, the second general index is used to determine the shape of the upper and lower envelopes of the pre-disassembled signal. That is, when the load actually consumes electricity, the electricity consumption of a single load should also be an effective power output. In other words, regardless of the form of power modulation the signal is subjected to, the upper and lower envelopes in its power output characteristics should be balanced and the same.

[0048] It should be noted that, in each acquisition of a disassembly signal, the present invention requires the aforementioned steps to determine whether the disassembly signal can be actually used in the next step. This determination is based on the general specifications of AC signals.

[0049] Step 4: Remove the first disassembly signal from the mixed electrical signal with added random noise, and use the methods in Steps 2 and 3 to extract the second disassembly signal and subsequent disassembly signals in sequence until the mixed electrical signal with added random noise can no longer be disassembled.

[0050] In this invention, a variable number of disassembly signals are actually extracted through a continuous loop of steps 2-4. After each disassembly is completed and verified, the method of this invention also supports extracting signal fluctuation trend information from the remaining signals, obtaining a pre-disassembled signal, verifying the pre-disassembled signal, and obtaining a correct disassembly signal. When the signal is disassembled to the final state, it is actually impossible to extract signal fluctuation trend information, or for the pre-disassembled signal, it is impossible to verify that it can become a disassembly signal.

[0051] It should be noted that steps 1 to 4 in this invention can be executed repeatedly. Specifically, more accurate noise prediction and signal decomposition can be achieved by adding different noises multiple times to the original mixed electrical signal and performing different decompositions on the same mixed electrical signal multiple times.

[0052] Preferably, adding random noise to the mixed electrical signal further includes: adding different random noises to the mixed electrical signal multiple times to generate multiple different mixed electrical signals with added random noise; the method further includes: using steps 2-4 to decompose the mixed electrical signal based on multiple different mixed electrical signals with added random noise; and the multiple decomposed signals further include generating multiple corresponding decomposed signals based on the decomposition of multiple different mixed electrical signals with added random noise; summing and averaging each set of corresponding decomposed signals to generate multiple decomposed signals.

[0053] The method described in this invention allows for the cancellation of random noise added multiple times to a mixed electrical signal to a certain extent, thus offsetting the actual noise. Furthermore, by averaging the results of multiple disassembly operations, the true state of the disassembled signal under the influence of noise can be obtained more accurately and macroscopically.

[0054] Step 5: Extract the first feature value and the second feature value from multiple disassembly signals. First, remove interference items from the disassembly signals based on the first feature value, and then obtain the valid disassembly signals from the disassembly signals after removing interference items based on the second feature value.

[0055] In this invention, considering that the methods in steps 1 to 4 above are difficult to truly achieve the acquisition of an effective disassembly signal, two feature values ​​are constructed to further analyze the effectiveness of the disassembly signal.

[0056] Preferably, a correlation coefficient is obtained between any one of the multiple disassembled signals and the mixed electrical signal; if the correlation coefficient is less than the first characteristic value, then any one of the signals is determined to be an interference term.

[0057] It should be noted that the correlation coefficient mentioned in this invention can be obtained using any existing correlation calculation method. In one embodiment of this invention, the effect is better when the correlation coefficient is calculated based on the Pearson correlation coefficient.

[0058] In this invention, the first feature value is typically used to initially verify the validity of the disassembly signal. For example, it is usually used to initially remove highly irrelevant information by eliminating interference. The second feature value, however, is used after eliminating obvious interference, and then after a more careful sorting and verification of the remaining disassembly signals, to extract the valid disassembly signal.

[0059] Preferably, a disassembly signal dataset is constructed based on the disassembly signals after excluding interference terms. After removing the central feature from each disassembly signal, the feature vector of the disassembly signal dataset is calculated. The feature vectors are sorted according to their contribution rate, and effective feature vectors with a cumulative contribution rate greater than the second feature value are extracted. Each effective disassembly signal is generated based on each effective feature vector.

[0060] It is understood that, in this invention, the disassembled signals after excluding interference terms can be used to construct a joint matrix based on the time order and the disassembled signal number. In this matrix, each horizontal vector can represent a single disassembled signal. Then, the central feature can be extracted for each disassembled signal. This invention can achieve the extraction of the signal central feature through various methods such as taking the signal median or mean. The removal of the central feature can be achieved by subtracting the central feature from each value in the disassembled signal vector.

[0061] In this way, multiple different decomposed signals can be normalized into a form that removes the central feature, making it easier to extract the eigenvectors of the joint matrix. Using existing matrix algorithms, multiple eigenvectors and corresponding eigenvalues ​​can be extracted from this joint matrix. In one embodiment of the present invention, the eigenvectors are represented as e1, e2, ..., e... M The corresponding eigenvalues ​​are λ1, λ2, ..., λ. M .

[0062] This invention also supports solving for the contribution rate of eigenvectors. This contribution rate can be obtained by assessing the importance of a particular eigenvector in the joint matrix, or it can be obtained based on the magnitude of the eigenvalues ​​corresponding to that eigenvector. For example, the proportion of an eigenvalue of a particular eigenvector to all eigenvalues ​​can be considered a contribution of that vector. This invention supports sorting by contribution magnitude; the greater the contribution, the higher the vector ranks, as it is more likely to represent the actual electricity consumption of the most important loads at a node.

[0063] In this invention, after sorting, multiple vectors with the highest contribution rates can be extracted. If the total contribution of these vectors is less than the second eigenvalue, then these vectors can be considered as feature vectors to be adopted. If the contribution rate of the next vector increases to a level greater than the second eigenvalue, then this vector is no longer considered as a feature vector to be adopted. Furthermore, each valid feature vector to be adopted can generate a valid decomposed signal. For example, the original signal dataset can be multiplied by the feature vectors to be adopted, etc.

[0064] Step 6: Use a feature separation matrix to segment each effective decomposed signal to generate multiple single-load electrical signals, thereby completing the decomposition of the mixed electrical signal.

[0065] Preferably, an original separation matrix is ​​constructed based on the effective decomposition signal, and the original separation matrix is ​​input into the fast ICA algorithm to complete the segmentation of the effective decomposition signal.

[0066] In this invention, after obtaining the effective decomposition signal, the effective decomposition signal can be further segmented. This segmentation is achieved using an invertible matrix. The fast ICA algorithm in this invention differs from ordinary neural network algorithms in that it can add a large amount of sample data to the computation in each iteration.

[0067] Specifically, this invention first decenters the original separation matrix, ensuring that the mean of each vector in each matrix is ​​0. Then, whitening is performed on the data. The number of iterations is appropriately set, and a random initial vector is selected. In this invention, the value of this initial vector is updated during each iteration, resulting in a converged vector value. Finally, after the iteration ends, the converged vector value is used as a parameter to multiply the transpose of the vector with the whitened matrix to obtain the result. Finally, using methods from Fast ICA, the feature separation matrix is ​​obtained through predefined nonlinear functions, whitening matrices, and mean calculation.

[0068] Finally, by multiplying the feature separation matrix with the original separation matrix formed by the effective decomposed signal, the segmentation of the effective decomposed signal can be achieved.

[0069] In this invention, to analyze the effectiveness of load decomposition, the correlation index between signals is used to evaluate its superiority. Generally, when the correlation index is greater than or equal to 0.8, the recovered signal is considered to be highly correlated with the decomposed signal.

[0070] This invention can use historical electricity consumption data to conduct experimental simulations to verify the confidence level of the algorithm. In one embodiment of this invention, the simulation experiment selected three common household appliances from low-frequency data: lighting equipment, computers, and refrigerators. Lighting equipment is a resistive load with fast dynamic response and relatively stable power waveforms, making it easy to monitor and decompose. Computers are nonlinear loads with relatively large randomness in their power waveforms, resulting in distortion and making decomposition difficult. Refrigerators are continuously variable loads, making them easier to decompose compared to computers.

[0071] Table 1 shows the performance of the algorithm after performing steps 1-4 multiple times, compared to performing the calculation once. As shown in Table 1, the algorithm's performance is significantly improved after adding random noise multiple times. Although adding noise multiple times increases the computation time by nearly two times, it improves the decomposition effect for lighting equipment by 5.31%, the decomposition effect for computers by 40.79%, and the decomposition effect for the total load by 86.63%.

[0072]

[0073] Table 1 Algorithm Performance Table

[0074] Figure 2 This is a schematic diagram of the architecture of a hybrid electrical signal splitting system based on residential electricity load according to the present invention. Figure 2As shown, a second aspect of the present invention relates to a hybrid electrical signal dismantling system based on residential electricity load. The system is used to implement a hybrid electrical signal dismantling method based on residential electricity load as described in any of the first aspects of the present invention. The system includes a noise addition unit, a pre-dismantling unit, a cyclic dismantling unit, a dismantling termination unit, a validity determination unit, and a segmentation unit. The noise addition unit is used to collect hybrid electrical signals from an electricity information acquisition system and add random noise to the hybrid electrical signals. The pre-dismantling unit is used to extract signal fluctuation trend information from the hybrid electrical signals with added random noise and remove the signal fluctuation trend information from the hybrid electrical signals to obtain a pre-dismantled signal. The cyclic dismantling unit is used to determine whether the pre-dismantled signal meets the general indicators of AC electrical signals; if not, it re-extracts signal fluctuation trend information based on the pre-dismantled signal and iteratively updates the pre-dismantled signal until the pre-dismantled signal meets the general indicators of AC electrical signals. The system employs a first disassembly unit to remove the first disassembly signal from the mixed electrical signal with added random noise, and sequentially extracts the second disassembly signal and subsequent disassembly signals using the methods in steps 2 and 3, until the mixed electrical signal with added random noise can no longer be disassembled; a valid determination unit extracts the first feature value and the second feature value from the multiple disassembly signals, firstly removes interference terms from the disassembly signals based on the first feature value, and then obtains valid disassembly signals from the disassembly signals after removing interference terms based on the second feature value; and a segmentation unit uses a feature separation matrix to segment each valid disassembly signal to generate multiple single-load electrical signals, thereby completing the disassembly of the mixed electrical signal.

[0075] The beneficial effects of this invention are that, compared with the prior art, the hybrid electrical signal decomposition method and system based on residential electricity load of this invention can obtain multiple decomposition signals by collecting hybrid electrical signals and using a cyclic iterative approach, determine the validity of the decomposition signals, and further segment the decomposition signals to ultimately obtain the single load electrical signals in the residential electricity load. This invention is effective and reliable, enabling accurate and reasonable allocation of the overall load from the perspective of individual user-side devices, providing more possibilities for demand-side energy management in smart grids.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A mixed electrical signal disassembling method based on residential electricity load, characterized in that, The method comprises the following steps: Step 1, collecting the mixed electrical signal of the residential electricity node in the power distribution network through the load monitoring device, and adding random noise to the mixed electrical signal; Step 2, extracting signal fluctuation trend information from the mixed electrical signal with random noise added, and removing the signal fluctuation trend information from the mixed electrical signal to obtain a pre-disassembly signal; Step 3, determining whether the pre-disassembly signal meets the general index of alternating current signal, if not, re-extracting the signal fluctuation trend information based on the pre-disassembly signal, and iteratively updating the pre-disassembly signal until the pre-disassembly signal meets the general index of alternating current signal, thereby obtaining a first disassembly signal; the general index of alternating current signal includes a first general index and a second general index; wherein the first general index is used to determine whether the pre-disassembly signal is opposite in phase between adjacent local extreme points, if yes, it is determined that the pre-disassembly signal meets the first general index; the second general index is judged based on the upper and lower envelope lines of the pre-disassembly signal; Step 4, removing the first disassembly signal from the mixed electrical signal with random noise added, and sequentially extracting a second disassembly signal and subsequent disassembly signals using the methods in steps 2 and 3, until the mixed electrical signal with random noise added cannot be disassembled; Step 5, extracting first and second characteristic values from a plurality of disassembly signals, and first removing interference terms in the disassembly signals based on the first characteristic value, constructing a disassembly signal dataset based on the disassembly signals after excluding the interference terms, removing the central feature from each of the disassembly signals in the dataset, and calculating the feature vectors of the disassembly signal dataset; sorting the feature vectors according to their contribution rates, and extracting effective feature vectors with a cumulative contribution rate greater than the second characteristic value; generating each effective disassembly signal based on each effective feature vector; Step 6, segmenting each of the effective disassembly signals using a feature separation matrix to generate a plurality of single-load electrical signals, thereby completing the disassembly of the mixed electrical signal.

2. The mixed electrical signal disassembly method based on residential electricity load according to claim 1, characterized in that: The method for extracting signal fluctuation trend information is: Extracting local extreme values of fluctuation signals from the mixed electrical signal with random noise added; Extracting the time points corresponding to the local extreme values, and determining the shape of the local extreme values based on the curve slope before the time points and the curve slope after the time points; Based on the determination, adding the local extreme values to the first shape set or the second shape set, and generating signal fluctuation trend information based on the corresponding relationship of the first shape set and the second shape set, respectively.

3. The mixed electrical signal disassembly method based on residential electricity load according to claim 2, characterized in that: The second general index is used to determine whether the upper and lower envelope lines of the pre-disassembly signal are symmetrical, if yes, it is determined that the pre-disassembly signal meets the second general index.

4. The method of claim 3, wherein adding random noise to the mixed electrical signal further comprises: adding different random noise to the mixed electrical signal multiple times to generate multiple different mixed electrical signals with added random noise; the method further comprises: implementing the decomposition of the mixed electrical signal based on the multiple different mixed electrical signals with added random noise using steps 2-4; and, the multiple decomposed signals further comprise generating multiple corresponding decomposed signals based on the decomposition of the multiple different mixed electrical signals with added random noise; summing and averaging each set of corresponding decomposed signals to generate the multiple decomposed signals.

5. The method of claim 4, wherein: obtaining a correlation coefficient between any one of the multiple decomposed signals and the mixed electrical signal; when the correlation coefficient is less than the first eigenvalue, determining that the any one signal is an interference term.

6. The method of claim 5, wherein: sorting the eigenvectors according to their contribution rates and extracting effective eigenvectors with a cumulative contribution rate greater than the second eigenvalue, comprises: sorting the eigenvectors according to their contribution rates, with a greater contribution rate resulting in a higher ranking; after sorting, extracting multiple eigenvectors with high contribution rates; if the total contribution of the multiple eigenvectors is less than the second eigenvalue, these eigenvectors can be considered as the effective eigenvectors; if the contribution value of the next eigenvector increases, making the cumulative contribution rate greater than the second eigenvalue, then this eigenvector is not considered as an effective eigenvector.

7. The method of claim 6, wherein: constructing a raw separation matrix based on the effective decomposed signals and inputting the raw separation matrix into a fast ICA algorithm to complete the segmentation of the effective decomposed signals.

8. The method of claim 1, wherein: applying a median filter algorithm to smooth the mixed electrical signal with added random noise, and then extracting the signal fluctuation trend information.

9. The method of claim 8, wherein: applying an adaptive sliding window algorithm to self-adjust the window size of the median filter algorithm.

10. A system for decomposing a mixed electrical signal based on residential electricity load, wherein: the system is configured to implement the method of any one of claims 1-9; and the system comprises a noise adding unit, a pre-decomposition unit, a cyclic decomposition unit, a decomposition interruption unit, an effective determination unit, and a segmentation unit; wherein: the noise adding unit is configured to add random noise to the mixed electrical signal to generate a mixed electrical signal with added random noise; the pre-decomposition unit is configured to implement the decomposition of the mixed electrical signal with added random noise to generate multiple decomposed signals; the cyclic decomposition unit is configured to implement the decomposition of the mixed electrical signal with added random noise based on the multiple decomposed signals using the pre-decomposition unit; the decomposition interruption unit is configured to determine whether the decomposition of the mixed electrical signal with added random noise is interrupted based on the multiple decomposed signals; the effective determination unit is configured to determine whether the multiple decomposed signals are effective based on the multiple decomposed signals; the segmentation unit is configured to segment the effective decomposed signals based on the effective decomposed signals. The noise adding unit is configured to collect a mixed electrical signal from the power utilization information collection system and add random noise to the mixed electrical signal. The pre-disassembly unit is configured to extract signal fluctuation trend information from the mixed electrical signal to which the random noise is added and cut the signal fluctuation trend information from the mixed electrical signal to obtain a pre-disassembly signal. The cyclic disassembly unit is configured to determine whether the pre-disassembly signal meets a general index of an alternating current electrical signal, re-extract signal fluctuation trend information based on the pre-disassembly signal if the pre-disassembly signal does not meet the general index of the alternating current electrical signal, and iteratively update the pre-disassembly signal until the pre-disassembly signal meets the general index of the alternating current electrical signal, thereby obtaining a first disassembly signal. The disassembly intermediate unit is configured to remove the first disassembly signal from the mixed electrical signal to which the random noise is added and sequentially extract a second disassembly signal and subsequent disassembly signals by using the method in steps 2 and 3 until the mixed electrical signal to which the random noise is added cannot be disassembled. The effective determination unit is configured to extract a first characteristic value and a second characteristic value from a plurality of the disassembly signals, remove interference terms in the disassembly signals based on the first characteristic value, and obtain effective disassembly signals from the disassembly signals from which the interference terms are removed based on the second characteristic value. The segmentation unit is configured to segment each of the effective disassembly signals by using a characteristic separation matrix to generate a plurality of single-load electrical signals, thereby completing disassembly of the mixed electrical signal.

Images