Block algorithm for collecting and detecting abnormal electricity utilization information of smart power grid

Abstract

The invention discloses a block algorithm for collecting and detecting abnormal electricity utilization information of a smart grid, and relates to the field of smart grids. In order to solve the problem that abnormal working conditions in a field environment are difficult to reproduce in the prior art, the method comprises the following steps: firstly, sampling electricity consumption information of a user, dividing according to blocks, counting electricity consumption times, calculating uncertainty, and identifying the highest uncertainty; block retrieval and difference calculation are carried out by setting hyper-parameters and defining matrix blocks. In the hyper-parameter range, the similar blocks are combined according to the difference result, and the blocks before combination are recorded. And continuously updating the highest uncertainty block, and re-dividing the block. The relative uncertainty is used as a measurement index, a merging rule is stipulated, whether each re-divided block meets a condition or not is judged, and then the abnormal condition is merged or identified. According to the algorithm, data information can be effectively compressed, the structure is simplified, the intelligent power grid system is helped to identify and process abnormal power consumption conditions, and the data processing efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the field of smart grids, and in particular to a block algorithm for collecting and detecting abnormal electricity consumption information in smart grids. Background Art

[0004] With the rapid increase in electricity users, the popularization of smart meters, and the development strategy of smart grid proposed by State Grid Corporation of China, the construction of electricity information collection system has become a key component of smart grid. Power supply enterprises can collect massive operating data from distribution networks. Therefore, power companies need to organize and mine the data collected by user electricity information collection system to obtain the hidden information. The sources of electricity data are wide and the structure is complex, which makes it difficult for enterprises to analyze data. Data mining technology can realize functions such as data sorting, data classification, and abnormal data search, so as to promote the construction of ubiquitous power Internet of Things and enhance the holographic perception ability of the power grid.

[0005] my country also started early in the field of simulation detection, but due to the development of computers and electronic equipment, related research in the field of electricity is mainly concentrated on the power generation, transmission and transformation side. In 2014, State Grid Corporation of China, State Grid Qinghai Electric Power Company and State Grid Qinghai Electric Power Company Electric Power Research Institute jointly completed a simulation platform for power consumption information collection system, which solved the problems of low efficiency and inaccurate test results in indoor testing of collection equipment, and improved the efficiency of collection equipment testing and the accuracy of test results. In 2015, State Grid Jiangxi Electric Power Co., Ltd. Electric Power Research Institute and State Grid Corporation jointly completed a power consumption information collection simulation test platform, which is a power consumption information collection system including a test environment simulation module and a test statistical analysis module; it can simulate the existing power consumption information collection system put into formal production operation, and provide a platform and test objects for the performance and function detection of the system. In 2015, State Grid Jiangxi Electric Power Research Institute, State Grid Shanxi Company Measurement Center and Fuzhou Power Supply Company jointly completed a power consumption information collection system simulation test platform, which can simulate and replace the formal system to complete various performance and function tests. In 2017, the Electric Power Research Institute of Henan Electric Power Company of State Grid, Zhengzhou Wante Electric Co., Ltd. of Henan Electric Power Company of State Grid and State Grid Corporation of China jointly completed a power consumption information collection simulation training system, which can simulate the operating conditions of power consumption information collection equipment of various users on site, and can meet the business training of power consumption information collection personnel to the greatest extent, improve their business skills and work efficiency. With the continuous development of information technology, the informatization of power systems has become increasingly perfect, especially in the field of power distribution and consumption. A complete power consumption information collection system has been established, and related testing laboratories have been put into operation. The research on key technologies for simulation detection of power consumption information collection in power systems has also received more and more attention, but the technical level is still in the development stage.

[0006] From the perspective of existing acquisition equipment testing, the various functional test benches that have been built or are under construction exist as relatively independent equipment rather than a systematic capability verification system. In terms of practical application, the following problems still exist: after the terminal and module are tested and qualified, the problem of poor on-site operation needs to be solved; after proposing technical solutions for new business needs, laboratory verification in a simulated environment is required before pilot verification; the quality differences and operational risks of on-site operating equipment need to be analyzed in terms of indicators.

[0007] The testing methods provided by the existing test verification system need to be improved in terms of rationality, standardization, and usability: First, the existing laboratories are unable to simulate on-site operating environment failures and are difficult to correspond to actual conditions; second, the testing equipment is independent, and single-unit functional testing makes it difficult to discover hidden problems caused by the superposition of changes in environmental configurations and operating conditions under the overall combined operating conditions of the system.

[0008] In recent years, many provincial power science research institutes and the State Grid have successively carried out a number of power consumption information collection simulation test platforms, proposed methods, and developed test platforms for power consumption information collection simulation. Under the influence of multiple factors, many faults and new abnormal conditions that cannot be simulated by traditional simulation detection platforms have occurred, and traditional simulation detection platforms can no longer adapt to new detection needs.

[0009] In addition, with the application of various new products and technologies such as smart IoT electricity meters, substation smart fusion terminals (SCU), dual-mode communication units, smart circuit breakers, and various functional modules, existing laboratories are unable to achieve full coverage testing of protocols and standards, which will cause problems that are difficult to find and solve, such as on-site communication failures due to manufacturers' misunderstandings.

[0010] Therefore, it is urgent to establish a combination of digital technology and real equipment to conduct simulation tests on collection tasks, provide a complete, real and controllable verification method for problem diagnosis of existing collection equipment, operation of new standard equipment, operating system and business APP application, and provide a comprehensive and systematic verification method for the development of new technologies and new applications.

[0011] Abnormal working conditions on site are complex and diverse, and the failure of the power consumption information collection system is often caused by the combined effect of multiple factors. Therefore, how to reproduce abnormal working conditions in the field environment and how to build a field application environment in the laboratory are one of the key points and difficulties in this field. Summary of the invention

[0012] In order to solve the problem that it is difficult to reproduce abnormal working conditions in the field environment in the prior art, the present invention provides the following solution:

[0013] A block algorithm for abnormal power consumption information collection and detection in smart grids, the algorithm includes:

[0014] Step 1. Sampling electricity usage information;

[0015] Step 2. Divide the sampled electricity usage information into blocks, count the number of electricity usage in each block and calculate the uncertainty of each block;

[0016] Step 3. Find the highest uncertainty from all blocks, set the hyperparameter k1, search the matrix blocks defined by the hyperparameter k1, and calculate the block difference d;

[0017] Step 4. Set the hyperparameter k2, search in the matrix block defined by the hyperparameter k2, search based on the difference d, and calculate the highest uncertainty in the matrix search block defined by the hyperparameter k2 and the difference d of the current matrix element being searched. c , if d c ≥d, then ignore this block, if d c<d, then the block is merged with the current block with the highest uncertainty, and the merged block is used as the new block with the highest uncertainty. At the same time, the block before the merge is recorded and removed from the search object;

[0018] Step 5. Repeat step 4 and keep updating d and d c , re-dividing the blocks;

[0019] Step 6. Set the relative uncertainty C r Measure metrics for blocks and develop rules for merging blocks based on relative uncertainty;

[0020] Step 7: Determine whether each re-divided block meets the block merging rules. If the block merging rules are met, then the two blocks are deemed similar and merged into one block until all blocks are aggregated; if the block merging rules are not met, then it is determined that there is an abnormal power consumption condition in the block.

[0021] Furthermore, the uncertainty of a block is determined by:

[0022]

[0023] Get, where C er is the uncertainty of the block, n blk The point frequency of the power consumption area, n is the number of collection times, and a is the correction parameter.

[0024] Furthermore, in step 3, the block difference is

[0025] d=C max -C min

[0026] Get, where d is the block difference, C max is the maximum uncertainty, C min is the minimum degree of certainty.

[0027] Further, in step six, the relative uncertainty is calculated by:

[0028]

[0029] Get, where C r is the relative uncertainty, n blk The frequency of the power consumption block, n is the number of acquisitions, a is the correction parameter, n bmi is the frequency of the block with minimum uncertainty.

[0030] Furthermore, in step 6, the block merging rules are:

[0031] 1-δ c ≤C r ≤1+δ c

[0032] Among them, δ c is an indicator parameter used to measure the uncertainty similarity of blocks.

[0033] The beneficial effects of the present invention are:

[0034] The accuracy and reliability of identifying the user's power usage status are enhanced. By introducing hyperparameters and correction parameters, the algorithm can be flexibly adjusted according to different situations to meet different needs. Unlike traditional abnormal algorithms and hybrid algorithms, the present invention adopts a new detection idea and is unique. The unique iteration method of the algorithm can effectively improve the detection effect and reduce the false alarm rate. In general, the present invention brings significant beneficial effects to the identification of user power usage status. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Flowchart of the block algorithm for abnormal power consumption information collection and detection in smart grids;

[0036] Figure 2 This is a diagram showing the effect of detecting anomalies after 150 iterations of the algorithm described in the present invention;

[0037] Figure 3 This is the effect diagram of detecting anomalies after LOF algorithm iterates 150 times;

[0038] Figure 4 This is the effect diagram of detecting anomalies after 150 iterations of the clustering algorithm;

[0039] Figure 5 This is the schematic diagram of OFDM modulation reception. DETAILED DESCRIPTION

[0040] Embodiment 1, combination Figure 1-Figure 5 This embodiment describes a block algorithm for collecting and detecting abnormal power consumption information in a smart grid, the algorithm comprising:

[0041] Step 1. Sampling electricity usage information;

[0042] Step 2. Divide the sampled electricity usage information into blocks, count the number of electricity usage in each block and calculate the uncertainty of each block;

[0043] Step 3. Find the highest uncertainty from all blocks, set the hyperparameter k1, search the matrix blocks defined by the hyperparameter k1, and calculate the block difference d;

[0044] Step 4. Set the hyperparameter k2, search in the matrix block defined by the hyperparameter k2, search based on the difference d, and calculate the highest uncertainty in the matrix search block defined by the hyperparameter k2 and the difference d of the current matrix element being searched. c , if dc ≥d, then ignore this block, if d c <d, then the block is merged with the current block with the highest uncertainty, and the merged block is used as the new block with the highest uncertainty. At the same time, the block before the merge is recorded and removed from the search object;

[0045] Step 5. Repeat step 4 and keep updating d and d c , re-divide the blocks;

[0046] Step 6. Set the relative uncertainty C r Measure metrics for blocks and develop rules for merging blocks based on relative uncertainty;

[0047] Step 7: Determine whether each re-divided block meets the block merging rules. If the block merging rules are met, then the two blocks are deemed similar and merged into one block until all blocks are aggregated; if the block merging rules are not met, then it is determined that there is an abnormal power consumption condition in the block.

[0048] Specific, combined Figure 1 It can be concluded that the block algorithm for abnormal power consumption information collection and detection for smart grids first samples the user's power consumption information, then divides it into blocks and counts the number of power consumption in each block and calculates the uncertainty. Then find the highest uncertainty, set the hyperparameters, and delineate the matrix blocks according to the hyperparameters for retrieval and calculate the difference between the blocks. Retrieval is performed within the set hyperparameter range, and blocks are merged according to the difference results. When the difference between two blocks meets certain conditions, they are considered similar and merged into a new block, and the blocks before the merger are recorded. Then continue to repeat this process, update the highest uncertainty block, and re-divide the blocks. Set relative uncertainty as the block measurement indicator, and formulate merging rules. According to the merging rules, it is judged whether each re-divided block meets the conditions. If it meets the conditions, it is merged into a larger block; if it does not meet the conditions, it is determined that there is an abnormal power consumption condition in the block.

[0049] Combination Figure 2-Figure 4 It can be concluded that the algorithm of the present invention effectively improves the detection effect and reduces the false alarm rate compared with the traditional LOF algorithm and clustering algorithm.

[0050] The block algorithm for collecting and detecting abnormal electricity consumption information in smart grids first collects the imbalance characteristics of current and voltage.

[0051] It can be seen from the electric energy calculation formula W=UI cosφt that when the power factor cosφ and voltage U are relatively constant, when power consumption is abnormal and there is a problem with the measurement of current I, the daily power load curve is similar. This determines that the daily current curve and voltage curve of normal electricity users are also similar. Therefore, current and voltage can be used as basic characteristic indicators.

[0052] When users use electricity normally, they show the characteristics of balanced three-phase electricity use. The voltage data read by the meter fluctuates slightly above and below the rated value, and the current imbalance fluctuates around the value 0. The abnormal electricity use of users who use three-phase electricity can be reflected in the deviation of the voltage and current imbalance rate.

[0053] Voltage unbalance through

[0054]

[0055] Get, where U i,n represents the voltage value collected for phase i at time n, x V.n Indicates the voltage imbalance at time n.

[0056] Current imbalance through:

[0057]

[0058] Get, where I i,n is the current value collected by phase i at time n, x I.n Indicates the current imbalance at time n.

[0059] Then, the power imbalance characteristic matrix X is established n Characterizes the imbalance of electricity consumption of users at time n.

[0060] The power consumption unbalance characteristic matrix X n pass:

[0061]

[0062] Get, where X V =(x V.1 ,x V.2 ,...,x V.n ), X I =(x I.1 ,x I.2 ,...,x I.n ).

[0063] For the unbalanced power consumption characteristic matrix X in two-dimensional space n The sample points formed are subjected to outlier analysis.

[0064] Regarding the principle of collection, the communication of the power consumption information collection system is based on the theory of OFDM (Orthogonal Frequency Division Multiplexing). OFDM is a multi-carrier modulation scheme. Its basic principle is to decompose the transmission channel into several orthogonal sub-channels, convert the high-speed data signal to be transmitted into parallel low-speed data streams, and modulate them to the sub-carriers of each orthogonal channel for transmission. The superimposed and transmitted orthogonal signals are demodulated and separated in a certain way at the receiving end. For the specific principle, see Figure 5 .

[0065] In order to completely separate the N sub-channel signals during reception, it is necessary to satisfy the requirement that the signal is received within each symbol duration T. s Within the range, any two carriers are orthogonal to each other.

[0066] From the perspective of time domain, the superposition of the transmitter signal in space can be written as:

[0067] f(t)=∑A i sin(2πf i t)+∑B i cos(2πf i t)

[0068] Among them, f(t) represents the waveform after the superposition of the sending end signal, f i is the frequency of the i-th signal component, and the i-th signal component A i , B i Indicates the signal amplitude.

[0069] Writing in plural form:

[0070]

[0071] From the perspective of time domain, OFDM sends N subcarriers at the transmitter. These signals are superimposed in space. When the receiver receives the superimposed signal, the superimposed signal is coherently demodulated for each subcarrier to obtain the signal of each subcarrier. This is relatively complicated to implement. The physical meaning of the superposition formula of each subcarrier can be regarded as the signal B on different carriers. i Do the inverse Fourier transform.

[0072]

[0073] T s is the symbol duration mentioned above, f(kT s / N) is the signal waveform after superposition, B i Indicates the signal amplitude.

[0074] After normalization, it is the DFT operation:

[0075]

[0076] Therefore, OFDM can be implemented using IFFT. The IFFT module at the transmitting end calculates the waveform after the superposition of each carrier, so there is no need to send a large number of carriers; and the FFT module at the receiving end can directly calculate each signal without the need to perform coherence on each carrier.

[0077] The uncertainty of a block is determined by:

[0078]

[0079] Get, where C er is the uncertainty of the block, n blk The point frequency of the power consumption area, n is the number of collection times, and a is the correction parameter.

[0080] In step 3, the block difference is passed

[0081] d=C max -C min

[0082] Get, where d is the block difference, C max is the maximum uncertainty, C min is the minimum degree of certainty.

[0083] In step six, the relative uncertainty is calculated by:

[0084]

[0085] Get, where C r is the relative uncertainty, n blk The frequency of the power consumption block, n is the number of acquisitions, a is the correction parameter, n bmi is the frequency of the block with minimum uncertainty.

[0086] In step 6, the block merging rules are:

[0087] 1-δ c ≤C r ≤1+δ c

[0088] Among them, δ c is an indicator parameter used to measure the uncertainty similarity of blocks.

[0089] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments may be envisioned within the scope of the invention thus described. In addition, it should be noted that the language used in this specification is selected primarily for readability and teaching purposes, rather than for explaining or defining the subject matter of the present invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the present invention is illustrative, not restrictive, with respect to the scope of the present invention, which is defined by the appended claims.

Claims

1. The block algorithm for abnormal power consumption information collection and detection in smart grid is characterized by: The algorithm includes: Step 1. Sampling electricity usage information; Step 2. Divide the sampled electricity usage information into blocks, count the number of electricity usage in each block and calculate the uncertainty of each block; Step 3. Find the highest uncertainty from all blocks, set the hyperparameter k1, search the matrix blocks defined by the hyperparameter k1, and calculate the block difference d; Step 4. Set the hyperparameter k2, search in the matrix block defined by the hyperparameter k2, search based on the difference d, and calculate the highest uncertainty in the matrix search block defined by the hyperparameter k2 and the difference d of the current matrix element being searched. c , if d c ≥d, then ignore this block, if d c <d, then the block is merged with the current block with the highest uncertainty, and the merged block is used as the new block with the highest uncertainty. At the same time, the block before the merge is recorded and removed from the search object; Step 5. Repeat step 4 and keep updating d and d c , re-dividing the blocks; Step 6. Set the relative uncertainty C r Measure metrics for blocks and develop rules for merging blocks based on relative uncertainty; Step 7: Determine whether each re-divided block meets the block merging rules. If the block merging rules are met, then the two blocks are deemed similar and merged into one block until all blocks are aggregated; if the block merging rules are not met, then it is determined that there is an abnormal power consumption condition in the block.

2. The block algorithm for abnormal power consumption information collection and detection in smart grid according to claim 1 is characterized in that: In step 2, the uncertainty of the block is determined by: Get, where C er is the uncertainty of the block, n blk The point frequency of the power consumption area, n is the number of collection times, and a is the correction parameter.

3. The block algorithm for abnormal power consumption information collection and detection in smart grid according to claim 1 is characterized in that: In step 3, the block difference is passed d=C max -C min Get, where d is the block difference, C max is the maximum uncertainty, C min is the minimum degree of certainty.

4. The block algorithm for abnormal power consumption information collection and detection in smart grid according to claim 1 is characterized in that: In step six, the relative uncertainty is calculated by: Get, where C r is the relative uncertainty, n blk The frequency of the power consumption block, n is the number of acquisitions, a is the correction parameter, n bmi is the frequency of the block with minimum uncertainty.

5. The block algorithm for abnormal power consumption information collection and detection in smart grid according to claim 1 is characterized in that: In step 6, the block merging rule is: 1-d c ≤C r ≤1+δ c Among them, δ c is an indicator parameter used to measure the uncertainty similarity of blocks.

Images