A power distribution network line risk prediction method and system

Abstract

The present application relates to a kind of distribution network line risk prediction method and system, it is related to electric power system monitoring and intelligent operation and maintenance technical field, specific steps include: the voltage time series data of voltage monitoring point and the three-phase voltage time series data of transformer output end are collected;With the arithmetic mean of each time series data as turning point identification is carried out to turning point reference line;According to the time interval of each turning point and previous turning point, relative distance is calculated, and the size of the data of each turning point and arithmetic mean is combined to carry out time series dimension reduction;Based on the final time series dimension reduction result, the DTW distance between the voltage time series of voltage monitoring point and the three-phase voltage time series of transformer output end is calculated using DTW algorithm, and the target phase of the minimum DTW distance is selected as the judgment result;According to predetermined rule, analyze DTW distance change rate and change trend, and predict the abnormal type of target phase line.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring and intelligent operation and maintenance technology, specifically to a method and system for predicting risks in distribution network lines. Background Technology

[0002] In the intelligent operation and maintenance of power distribution networks, accurate phase identification and early prediction of line operation risks are two core and closely related links to ensure power supply reliability and security. However, existing technical solutions typically treat these two tasks separately, resulting in low operation and maintenance efficiency and insufficient early warning capabilities. For phase identification, current methods mainly rely on regular manual inspections or the deployment of high-precision synchronous phase measurement devices. The former is inefficient, lacks real-time performance, and is prone to errors, while the latter is difficult to apply on a large scale in numerous cost-sensitive distribution substations due to the high cost of the equipment and the complexity of deployment. Meanwhile, the prediction of risks such as line aging, load fluctuations, short circuits, and phase loss often relies on another independent monitoring system or complex analysis model, failing to achieve deep coupling and linkage analysis with the basic topology and real-time phase information of the lines. This results in risk warnings lacking precise line targeting and timeliness.

[0003] To process massive amounts of voltage monitoring data, time series dimensionality reduction techniques have been introduced to improve computational efficiency. Piecewise aggregation approximation (PAA) is one commonly used method, but its method of compressing data through piecewise averaging has significant drawbacks: it over-smooths the data, losing key fluctuation and transition features in the voltage series, which are crucial for accurately identifying phases and sensing abnormal changes. Furthermore, the PAA method has poor adaptability to nonlinear deformations of the series, and its fixed piecewise strategy lacks flexibility, making it difficult to adapt to voltage data under different operating conditions, thus limiting its effectiveness in precise matching and trend analysis.

[0004] Dynamic Time Warping (DTW) algorithm can effectively measure the similarity between two time series and handle sequence matching problems with different rates, making it very suitable for phase identification and sequence trend comparison based on waveform similarity. However, directly applying the DTW algorithm to raw long-sequence voltage data results in extremely high computational complexity, making it difficult to meet the needs of real-time monitoring and online analysis of distribution networks.

[0005] The method for assessing abnormal transformer substation conditions disclosed in Chinese invention patent CN116014886A, while achieving condition monitoring through multi-source data acquisition, suffers from several drawbacks: reliance on numerous terminal devices, high data redundancy, insensitivity to capturing instantaneous dynamic changes in lines, and a lack of integrated modeling and analysis of phase identification and risk evolution trends. Therefore, there is an urgent need for a low-cost, highly practical solution that can deeply integrate phase identification and line risk prediction based on efficient dimensionality reduction, to fully utilize the widely deployed, simplified voltage monitoring data and achieve precise and intelligent operation and maintenance of the distribution network. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention proposes a method and system for predicting risks in power distribution network lines.

[0007] The technical solution of the present invention is as follows:

[0008] On the one hand, this invention proposes a method for predicting risks in power distribution network lines, the specific steps of which include:

[0009] Collect voltage time series data from voltage monitoring points and three-phase voltage time series data from the transformer output terminal;

[0010] The arithmetic mean of each time series data is used as the baseline for identifying turning points.

[0011] The relative distance between each inflection point and the previous inflection point is calculated, and the sign is determined by comparing the data of each inflection point with the arithmetic mean. The time series with fewer than a preset threshold is segmented for dimensionality reduction, and the final time series dimensionality reduction result is output.

[0012] Based on the final time series dimensionality reduction result, the DTW algorithm is used to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and the target phase corresponding to the minimum DTW distance is selected as the judgment result.

[0013] Under the same phase, DTW distance is calculated by sliding time windows with minute-level data windows as the unit. The DTW distance change rate at different times between the same time windows in two adjacent days is calculated. The DTW distance change rate and trend are analyzed according to predetermined rules to determine the type of anomaly in the target line.

[0014] Based on the anomaly type and DTW distance change trend, the real-time risk level of the target line is quantified. Based on historical DTW distance change rate data, the development trend of anomalies of the target line is predicted by a time series prediction model, and the line risk prediction result is generated.

[0015] In a preferred embodiment, the step of extracting turning points using the arithmetic mean of each time series data as a turning point baseline specifically includes:

[0016] Calculate the arithmetic mean of each time series data;

[0017] Record the data points in the time series that change from below the arithmetic mean to above the arithmetic mean, or from above the arithmetic mean to below the arithmetic mean, as turning points;

[0018] Adding the first and last points of the time series forms a complete turning point sequence.

[0019] As a preferred implementation, when a data point in the time series changes from below the arithmetic mean to above the arithmetic mean, or changes from above the arithmetic mean to below the arithmetic mean, as an inflection point, a dynamic threshold is set to retain only the first data point that crosses the arithmetic mean point.

[0020] In a preferred embodiment, the rule for determining the sign by comparing the data at each inflection point with the arithmetic mean is specifically as follows:

[0021] When the data at each inflection point is greater than the arithmetic mean, the relative distance ratio is multiplied by a positive sign; when the data at each inflection point is less than the arithmetic mean, it is multiplied by a negative sign. The positive and negative signs are used to distinguish the direction of each inflection point: a positive sign indicates that the data changes from below the arithmetic mean to above it, and a negative sign indicates that the data changes from above the arithmetic mean to below it.

[0022] In a preferred embodiment, the step of segmenting and reducing the dimensionality of time series with fewer than a preset threshold specifically includes:

[0023] When the number of inflection points extracted from the time series is less than the preset threshold, the number of original data between adjacent original inflection points is calculated, and the interval with the most data is selected as the sub-segment.

[0024] New turning points are extracted from the data within the sub-segment, and the relative distance and sign between the turning points are calculated. At the same time, the relative distance is normalized to the length of the original time series.

[0025] The extracted new inflection points are added to the original inflection point sequence. The interval with the most data is selected as the sub-segment to extract new inflection points and add them to the original inflection point sequence. The iteration stops when the number of inflection points in the original inflection point sequence meets the preset threshold. The final time series dimensionality reduction result is output, and the segmented dimensionality reduction is completed.

[0026] In a preferred embodiment, the rule for determining the phase of the voltage monitoring point is as follows:

[0027] If DTW_UA_X < DTW_UB_X and DTW_UA_X < DTW_UC_X, then it is determined that X is the A-phase of the transformer;

[0028] If DTW_UB_X < DTW_UA_X and DTW_UB_X < DTW_UC_X, then it is determined that X is the B-phase of the transformer;

[0029] If DTW_UC_X < DTW_UA_X and DTW_UC_X < DTW_UB_X, then it is determined that X is the C-phase of the transformer;

[0030] Wherein, X is the voltage monitoring point, DTW_UA_X is the distance between the A-phase of the transformer and the dimensionality reduction result of the time series of the voltage monitoring point, DTW_UB_X is the distance between the B-phase of the transformer and the dimensionality reduction result of the time series of the voltage monitoring point, and DTW_UC_X is the distance between the C-phase of the transformer and the dimensionality reduction result of the time series between the voltage monitoring points.

[0031] As a preferred implementation manner, the step of analyzing the DTW distance change rate and change trend according to a predetermined rule to judge the abnormal type existing in the target line is specifically as follows:

[0032] Set a DTW distance change rate threshold. When the DTW distance change rate exceeds the threshold, a warning signal is triggered;

[0033] If the DTW distance change rate suddenly changes and exceeds the preset minimum mutation threshold, it is judged that the target line has load fluctuations or a short circuit;

[0034] If the DTW distance value continuously increases or decreases, it is judged that the target line has aging or a broken phase.

[0035] As a preferred implementation manner, the step of further quantifying the real-time risk level of the target line according to the abnormal type and the DTW distance change trend, predicting the development trend of the abnormality of the target line based on the historical DTW distance change rate data, and generating a line risk prediction result is specifically as follows:

[0036] For the target line, based on the DTW distance change rate data during its historical normal operation period, a line operation baseline model is established;

[0037] For the abnormal target line judged to have load fluctuations or a short circuit, combine the mutation amplitude and the deviation from the historical baseline to quantify its high risk level, and predict the development trend of the abnormality of the target line based on the historical DTW distance change rate data;

[0038] For anomalies identified as aging or phase loss, the medium- to long-term risk level is quantified by combining the duration and rate of change of the trend. Based on historical DTW distance change rate data, a time series prediction model is used to predict the development trend of the anomaly on the target line.

[0039] On the other hand, the present invention proposes a distribution network line risk prediction system, comprising:

[0040] The data acquisition module collects voltage time series data from voltage monitoring points and three-phase voltage time series data from the transformer output terminal;

[0041] The inflection point extraction module uses the arithmetic mean of each time series data as the standard, retaining all data points that cross the arithmetic mean within the time series, as well as the first and last points of the time series as inflection points;

[0042] The time series dimensionality reduction module calculates the relative distance between each inflection point and the previous inflection point, and determines the sign by comparing the data at each inflection point with the arithmetic mean. It performs segmented dimensionality reduction on time series with fewer than a preset threshold and outputs the final time series dimensionality reduction result.

[0043] The phase identification module, based on the final time series dimensionality reduction result, uses the DTW algorithm to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and selects the target phase corresponding to the minimum DTW distance as the identification result;

[0044] The anomaly detection module calculates DTW distance using a sliding time window with minute-level data windows under the same phase. It calculates the DTW distance change rate at different times within the same time window over two adjacent days, analyzes the DTW distance change rate and trend according to predetermined rules, and determines the anomaly type of the target line.

[0045] The risk prediction module then quantifies the real-time risk level of the target line based on the anomaly type and DTW distance change trend. Based on historical DTW distance change rate data, it uses a time series prediction model to predict the development trend of anomalies on the target line and generates line risk prediction results.

[0046] On the other hand, the present invention proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a power distribution line risk prediction method according to any embodiment of the present invention.

[0047] The present invention has the following beneficial effects:

[0048] 1. This invention extracts key turning points of change through dimensionality reduction, significantly reducing the amount of data and computational complexity. Traditional methods typically use raw voltage data directly, resulting in large amounts of data and redundant information, leading to high computational complexity.

[0049] 2. This invention, through mean calculation and inflection point identification, can clearly reflect the fluctuation trend of voltage data, improving the accuracy of identification and outperforming the extreme point dimensionality reduction method. Existing methods may struggle to effectively capture the effective changes and noise effects in voltage data.

[0050] 3. The DTW method has a large computational cost when processing long sequence data, but the length of the data sequence after dimensionality reduction is significantly reduced, which greatly improves the efficiency of DTW computation and makes the method more practical.

[0051] 4. This invention is highly adaptable to data fluctuation ranges and sampling frequencies, suitable for voltage data sequences of varying scales and complexities, and can be widely applied in practical scenarios. It can efficiently process voltage data and accurately identify phases. Attached Figure Description

[0052] Figure 1 This is a flowchart of the steps of the present invention;

[0053] Figure 2 This is a distribution diagram of the original voltage data. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0056] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0057] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0058] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0059] Example 1:

[0060] See Figure 1 A method for predicting risks in power distribution network lines, comprising the following steps:

[0061] Step S1: Collect voltage time series data from voltage monitoring points and three-phase voltage time series data from the transformer output terminal;

[0062] To reduce data complexity and extract key features, this approach first collects time-series data from voltage monitoring points and three-phase voltage data from the transformer output, and then performs dimensionality reduction processing.

[0063] Step S2: Using the arithmetic mean of each time series data as the standard, retain all data points in the time series that cross the arithmetic mean, as well as the first and last points of the time series as turning points;

[0064] Step S201: Calculate the arithmetic mean

[0065] First, calculate the arithmetic mean of the data sequence. This arithmetic mean serves as the basis for subsequent feature determination. The specific calculation formula is as follows:

[0066]

[0067] In the formula, The length of the data sequence. For the first Data points.

[0068] Step S202: Identify the turning point

[0069] First, retain the first point of the data sequence as the initial point. Then, analyze the relationship between each data point and the mean. When the data changes from below the mean to above the mean, or from above the mean to below the mean, record that point as a turning point. For data segments with consecutive values ​​above or below the mean, retain only the first data point to avoid redundancy. Finally, ensure that the last point of the data sequence is retained as the ending point.

[0070] Step S3: Calculate the relative distance between each inflection point and the previous inflection point, and determine the sign by comparing the data of each inflection point with the arithmetic mean. Perform segmented dimensionality reduction on time series with fewer than a preset threshold, and output the final time series dimensionality reduction result.

[0071] Step S301: Relative distance standardization

[0072] The algorithm first calculates the time interval between each inflection point and the previous inflection point as the relative distance. Then, it standardizes this distance by dividing it by the total sequence length, obtaining the relative distance ratio. This method effectively compresses the data size while preserving important trend change information. The standardized relative distance ratio reflects the rhythm and magnitude of trend changes, providing a more compact and representative data representation for subsequent DTW similarity calculations.

[0073] The standardized formula for calculating relative distance is as follows:

[0074]

[0075] In the formula, The length of the time series. For the turning point index, where This represents the number of turning points.

[0076] Step S302: Symbol Conversion

[0077] When the data is higher than the arithmetic mean, the phase position is multiplied by 1;

[0078] When the data is below the arithmetic mean, the relative position is multiplied by -1.

[0079] Step S303: Segmentation Processing

[0080] When the number of turning points extracted from the time series is less than the preset threshold (10 points in this scheme), the number of original data between adjacent original turning points is calculated, and the interval with the most data is selected as the sub-segment.

[0081] New turning points are extracted from the data within the sub-segment, and the relative distance and sign between the turning points are calculated. At the same time, the relative distance is normalized to the length of the original time series.

[0082] The extracted new inflection points are added to the original inflection point sequence. The interval with the most data is selected as the sub-segment to extract new inflection points and add them to the original inflection point sequence. The iteration stops when the number of inflection points in the original inflection point sequence meets the preset threshold. The final time series dimensionality reduction result is output, and the segmented dimensionality reduction is completed.

[0083] Step S4: Based on the final time series dimensionality reduction result, the DTW algorithm is used to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and the target phase corresponding to the minimum DTW distance is selected as the judgment result.

[0084] Step S401: Calculate the similarity between the final time series dimensionality reduction results using Euclidean distance. Based on the dynamic programming principle of the dynamic time programming algorithm, construct a cumulative distance matrix and store the cumulative similarity value between the two final time series dimensionality reduction results.

[0085] Based on the dynamic programming principle of the DTW algorithm, a cumulative distance matrix is ​​constructed. Each element C[i,j] in the matrix represents the minimum cumulative distance between the first i data points of the voltage time series of phase A and the first j data points of the voltage time series of voltage monitoring point X.

[0086] Step S402: In each cumulative distance matrix, the minimum cumulative distance value corresponding to the last data point is used as the similarity index between the two voltage effective value sequences.

[0087] In the cumulative distance matrix, find the minimum cumulative distance value corresponding to the last data point, such as C[N,N], as the similarity index between the two sequences.

[0088] Step S403: Calculate the minimum cumulative distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, which is the DTW distance value, and determine the phase of the voltage monitoring point;

[0089] Step S5: Under the same phase, DTW distance is calculated by sliding time window with minute-level data window as unit. The DTW distance change rate at different times between the same time window in two adjacent days is calculated. The DTW distance change rate and trend are analyzed according to the predetermined rules to determine the type of anomaly in the target line.

[0090] In this step, the sliding time window means: using minute-level data windows as units, the DTW distance is calculated by sliding the time window, and the trend of DTW distance change is analyzed.

[0091] The rate of change is calculated as follows: The rate of change of the DTW distance between two adjacent days within the same time window is calculated using the following formula:

[0092] Rate of change = DTW_day1 / DTW_day0

[0093] In the formula, DTW_day0 is the DTW distance between the target phase at the transformer output terminal and the user terminal at the same time on the previous day, and DTW_day1 is the DTW distance at the same time on the current day.

[0094] A preset change rate threshold is set. When the change rate exceeds the threshold, a fault warning mechanism is triggered, and a warning signal is sent.

[0095] Based on the changing trend of DTW distance, further analyze the specific types of line anomalies:

[0096] When the DTW distance change exceeds the set threshold, it is determined that the line may have load fluctuations or short circuit problems;

[0097] When the DTW distance continues to increase or decrease, it is determined that the line may have aging or phase loss problems.

[0098] Step S6: Based on the anomaly type and DTW distance change trend, quantify the real-time risk level of the target line. Based on historical DTW distance change rate data, use a time series prediction model to predict the development trend of anomalies on the target line and generate line risk prediction results.

[0099] In this step, a baseline model for the operation of the target line is established based on the DTW distance change rate data during its historical normal operation period.

[0100] For abnormal target lines identified as load fluctuations or short circuits, their high-risk level is quantified by combining the magnitude of the sudden change with the historical baseline deviation. Based on historical DTW distance change rate data, the development trend of the abnormality of the target line is predicted by a time series prediction model.

[0101] High-amplitude abrupt changes are often precursors or early signs of faults. By comparing the current abrupt change magnitude (DTW distance change rate) with a baseline model of the DTW change rate established during the line's historical normal operation, the range of the abrupt change is determined. A larger deviation indicates a greater deviation from normal operating conditions and a higher risk level. Several deviation thresholds are pre-set to classify line risks into multiple levels. Based on the changing trends of line risk levels, the time series prediction model estimates the potential time range from the current abnormal state to a critical fault state, according to the speed and intensity of the abrupt change, thereby predicting the risks present in the line.

[0102] For anomalies identified as aging or phase loss, the medium- to long-term risk level is quantified by combining the duration and rate of change of the trend. Based on historical DTW distance change rate data, a time series prediction model is used to predict the development trend of the anomaly on the target line.

[0103] Based on the duration and rate of change of line aging or phase loss, a time series prediction model can be established to predict the abnormal development trend of the target line, determine the lifespan and operational safety of the target line, and make advance judgments on line maintenance and safety risks.

[0104] Based on steps S1-S3 above, a data dimensionality reduction demonstration is performed using the original data sequence [10, 15, 18, 12, 20, 25, 16, 22] as an example:

[0105] 1) Calculation of the arithmetic mean:

[0106] 2) Turning point identification:

[0107] Point 1 (10): Below the mean, recorded as a turning point;

[0108] Point 2 (15): Keep below the mean, the same as the previous point, do not record;

[0109] Point 3 (18): Above the mean, recorded as a turning point;

[0110] Point 4 (12): Below the mean, recorded as a turning point;

[0111] Point 5 (20): Above the mean, recorded as a turning point;

[0112] Point 6 (25): Keep above the mean, the same as the previous point, do not record;

[0113] Point 7 (16): Below the mean, recorded as a turning point;

[0114] Point 8 (22): Above the mean, recorded as a turning point.

[0115] The turning point index is [0, 2, 3, 4, 6, 7], corresponding to the 1st, 3rd, 4th, 5th, 7th, and 8th points.

[0116] 3) Calculate the relative distance:

[0117] The total number of data points is 8, and the index range is from 0 to 7.

[0118] Total length: n-1 = 7

[0119] Calculate each inflection point using indexed data:

[0120] Turning point 0: Distance = 0, relative distance = 0 / 7 = 0;

[0121] Turning point 2: Distance = 2 - 0 = 2, relative distance = 2 / 7 ≈ 0.2857;

[0122] Turning point 3: Distance = 3 - 2 = 1, relative distance = 1 / 7 ≈ 0.1429;

[0123] Turning point 4: Distance = 4 - 3 = 1, relative distance = 1 / 7 ≈ 0.1429;

[0124] Turning point 6: Distance = 6 - 4 = 2, relative distance = 2 / 7 ≈ 0.2857;

[0125] Turning point 7: Distance = 7 - 6 = 1, relative distance = 1 / 7 ≈ 0.1429;

[0126] 4) Symbol conversion:

[0127] Determine the symbol at each turning point:

[0128] Index 0: Value = 10 < Mean → Sign = -1 → Final value = -0.0000;

[0129] Index 2: Value = 18 > Mean → Sign = +1 → Final value = +0.2857;

[0130] Index 3: Value = 12 < Mean → Sign = -1 → Final value = -0.1429;

[0131] Index 4: Value = 20 > Mean → Sign = +1 → Final Value = +0.1429;

[0132] Index 6: Value = 16 < Mean → Sign = -1 → Final value = -0.2857;

[0133] Index 7: Value = 22 > Mean → Sign = +1 → Final value = +0.1429.

[0134] 5) Final dimensionality reduction result:

[0135] Original data: [10, 15, 8, 12, 20, 25, 16, 22] (8 points);

[0136] The dimensionality-reduced sequence is: [-0.0000, +0.2857, -0.1429, +0.1429, -0.2857, +0.1429] (6 points in total).

[0137] According to steps S4-S5, the raw data of a certain transformer area (e.g.) Figure 2 Based on the data shown, the calculation is performed using the first 180 points of voltage data (one point per second, three minutes of data):

[0138] 1) Calculate the inflection point index of the three-phase voltages UA, UB, UC and voltage monitoring point X:

[0139] Three-phase voltage UA: Original data point = 180, critical inflection point = 12

[0140] Mean = 233.6631, Turning point index = [0, 36, 37, 42, 43, 50, 103, 105, 122, 151, 173, 179], Data compression ratio = 15.00:1.

[0141] Three-phase voltage UB: Original data point = 180, critical inflection point = 24

[0142] Mean = 235.4593, Turning point index = [0, 14, 16, 17, 18, 19, 20, 21, 44, 51, 67, 69, 75, 79, 102, 105, 122, 158, 161, 162, 164, 170, 172, 179], Data compression ratio = 7.50:1.

[0143] Three-phase voltage UC: Original data point = 180, critical inflection point = 14

[0144] Mean = 235.1332, Turning point index = [0, 25, 28, 29, 41, 52, 65, 80, 103, 105, 123, 151, 178, 179], Data compression ratio = 12.86:1.

[0145] Voltage monitoring point X: Raw data points = 180, critical inflection point = 15

[0146] Mean = 234.6703, Turning point index = [0, 26, 29, 30, 41, 52, 66, 81, 101, 102, 103, 105, 123, 152, 179], Data compression ratio = 12.00:1.

[0147] 2) Calculate the inflection point values ​​of the three-phase voltages UA, UB, UC and voltage monitoring point X.

[0148] UA=[0, -0.201, 0.006, -0.028, 0.006, -0.039, 0.296, -0.011, 0.095, -0.162, 0.123, 0.034];

[0149] UB=[0, -0.078, 0.011, -0.006, 0.006, -0.006, 0.006, -0.006, 0.128, -0.039, 0.089, -0.011, 0.034, -0.022, 0.128, -0.017, 0.095, -0.201, 0.017, -0.006, 0.011, -0.034, 0.011, 0.039];

[0150] UC=[0, -0.14, 0.017, -0.006, 0.067, -0.061, 0.073, -0.084, 0.128, -0.011, 0.101, -0.156, 0.151, 0.006];

[0151] X=[0, -0.145, 0.017, -0.006, 0.061, -0.061, 0.078, -0.084, 0.112, -0.006, 0.006, -0.011, 0.101, -0.162, 0.151];

[0152] 3) Calculate the DTW distance separately.

[0153] UA,X: DTW distance = 0.6425;

[0154] UB,X: DTW distance = 1.1397;

[0155] UC,X: DTW distance = 0.2067;

[0156] 4) Output the phase identification result of X.

[0157] The minimum distance between X and the three-phase data sequence is 0.2067, indicating that X is phase C.

[0158] Example 2

[0159] A power distribution network line risk prediction system, comprising:

[0160] The data acquisition module collects voltage time series data from voltage monitoring points and three-phase voltage time series data from the transformer output terminal;

[0161] The inflection point extraction module uses the arithmetic mean of each time series data as the standard, retaining all data points that cross the arithmetic mean within the time series, as well as the first and last points of the time series as inflection points;

[0162] The time series dimensionality reduction module calculates the relative distance between each inflection point and the previous inflection point, and determines the sign by comparing the data at each inflection point with the arithmetic mean. It performs segmented dimensionality reduction on time series with fewer than a preset threshold and outputs the final time series dimensionality reduction result.

[0163] The phase identification module, based on the final time series dimensionality reduction result, uses the DTW algorithm to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and selects the target phase corresponding to the minimum DTW distance as the identification result;

[0164] The anomaly detection module calculates DTW distance using a sliding time window with minute-level data windows under the same phase. It calculates the DTW distance change rate at different times within the same time window over two adjacent days, analyzes the DTW distance change rate and trend according to predetermined rules, and predicts the risk type of the target phase line.

[0165] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for predicting risks in power distribution network lines, characterized in that, The specific steps include: Collect voltage time series data from voltage monitoring points and three-phase voltage time series data from the transformer output terminal; The arithmetic mean of each time series data is used as the baseline for identifying turning points. The relative distance between each inflection point and the previous inflection point is calculated, and the sign is determined by comparing the data of each inflection point with the arithmetic mean. The time series with fewer than a preset threshold is segmented for dimensionality reduction, and the final time series dimensionality reduction result is output. The specific steps for segmenting and reducing the dimensionality of time series with fewer than a preset threshold are as follows: When the number of inflection points extracted from the time series is less than the preset threshold, the number of original data between adjacent original inflection points is calculated, and the interval with the most data is selected as the sub-segment. New turning points are extracted from the data within the sub-segment, and the relative distance and sign between the turning points are calculated. At the same time, the relative distance is normalized to the length of the original time series. The extracted new inflection points are added to the original inflection point sequence. The interval with the most data is selected as the sub-segment to extract new inflection points and add them to the original inflection point sequence. The iteration stops when the number of inflection points in the original inflection point sequence meets the preset threshold. The final time series dimensionality reduction result is output, and the segmented dimensionality reduction is completed. Based on the final time series dimensionality reduction result, the DTW algorithm is used to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and the target phase corresponding to the minimum DTW distance is selected as the judgment result. Under the same phase, DTW distance is calculated by sliding time windows with minute-level data windows as the unit. The DTW distance change rate at different times between the same time windows in two adjacent days is calculated. The DTW distance change rate and trend are analyzed according to predetermined rules to determine the type of anomaly in the target line. Based on the anomaly type and DTW distance change trend, the real-time risk level of the target line is quantified. Based on historical DTW distance change rate data, the development trend of anomalies of the target line is predicted by a time series prediction model, and the line risk prediction result is generated.

2. The method for predicting the risk of a power distribution network line according to claim 1, characterized in that, The specific steps for extracting turning points using the arithmetic mean of each time series data as a turning point baseline are as follows: Calculate the arithmetic mean of each time series data; Record the data points in the time series that change from below the arithmetic mean to above the arithmetic mean, or from above the arithmetic mean to below the arithmetic mean, as turning points; Adding the first and last points of the time series forms a complete turning point sequence.

3. The method for predicting the risk of a power distribution network line according to claim 2, characterized in that, When a data point in a time series changes from below the arithmetic mean to above the arithmetic mean, or from above the arithmetic mean to below the arithmetic mean, as an inflection point, a dynamic threshold is set to retain only the first data point that crosses the arithmetic mean.

4. The method for predicting the risk of a power distribution network line according to claim 1, characterized in that, The specific rule for determining the sign by comparing the data at each inflection point with the arithmetic mean is as follows: When the data at each turning point is greater than the arithmetic mean, the relative distance ratio is multiplied by a positive sign; when the data at each turning point is less than the arithmetic mean, it is multiplied by a negative sign; the positive and negative signs are used to distinguish the direction of each turning point. The positive sign indicates that the data changes from below the arithmetic mean to above it, and the negative sign indicates that the data changes from above the arithmetic mean to below it.

5. The method for predicting the risk of a power distribution network line according to claim 1, characterized in that, The judgment rule for the phase of the voltage monitoring point is as follows: If DTW_UA_X < DTW_UB_X and DTW_UA_X < DTW_UC_X, then it is determined that X is phase A of the transformer; If DTW_UB_X < DTW_UA_X and DTW_UB_X < DTW_UC_X, then it is determined that X is phase B of the transformer; If DTW_UC_X < DTW_UA_X and DTW_UC_X < DTW_UB_X, then it is determined that X is phase C of the transformer; Where X is the voltage monitoring point, DTW_UA_X is the distance between phase A of the transformer and the dimensionality reduction result of the voltage time series at the voltage monitoring point, DTW_UB_X is the distance between phase B of the transformer and the dimensionality reduction result of the voltage time series at the voltage monitoring point, and DTW_UC_X is the distance between phase C of the transformer and the dimensionality reduction result of the voltage time series at the voltage monitoring point.

6. The method for predicting the risk of a power distribution network line according to claim 1, characterized in that, The steps for analyzing the DTW distance change rate and change trend according to the predetermined rules to determine the abnormal type existing in the target line are specifically as follows: Set a DTW distance change rate threshold. When the DTW distance change rate exceeds the threshold, a warning signal is triggered; If the DTW distance change rate undergoes a mutation and exceeds the preset minimum mutation threshold, it is determined that the target line has a load fluctuation or a short circuit; If the DTW distance value continuously increases or decreases, it is determined that the target line has aging or a broken phase.

7. The method for predicting the risk of a power distribution network line according to claim 1, characterized in that, The steps for further quantifying the real-time risk level of the target line according to the abnormal type and the DTW distance change trend, and predicting the development trend of the abnormality of the target line through a time series prediction model based on the historical DTW distance change rate data to generate a line risk prediction result are specifically as follows: For the target line, based on the DTW distance change rate data during its historical normal operation period, a line operation baseline model is established; For the abnormal target line judged to have a load fluctuation or a short circuit, combine the mutation amplitude and the deviation from the historical baseline to quantify its high risk level, and predict the development trend of the abnormality of the target line through a time series prediction model based on the historical DTW distance change rate data; For the judged abnormality of aging or a broken phase, combine the duration of the change trend and the change rate to quantify its medium- and long-term risk level, and predict the development trend of the abnormality of the target line through a time series prediction model based on the historical DTW distance change rate data.

8. A risk prediction system for power distribution network lines, characterized in that, Including: A data acquisition module that acquires the voltage time series data of the voltage monitoring points and the three-phase voltage time series data at the output end of the transformer; A turning point extraction module that uses the arithmetic mean of each time series data as a standard, retains all the data points in the time series that cross the arithmetic mean, as well as the first point and the last point of the time series as turning points; The time series dimensionality reduction module calculates the relative distance between each inflection point and the previous inflection point, and determines the sign by comparing the data at each inflection point with the arithmetic mean. It performs segmented dimensionality reduction on time series with fewer than a preset threshold and outputs the final time series dimensionality reduction result. The specific steps for segmenting and reducing the dimensionality of time series with fewer than a preset threshold are as follows: When the number of inflection points extracted from the time series is less than the preset threshold, the number of original data between adjacent original inflection points is calculated, and the interval with the most data is selected as the sub-segment. New turning points are extracted from the data within the sub-segment, and the relative distance and sign between the turning points are calculated. At the same time, the relative distance is normalized to the length of the original time series. The extracted new inflection points are added to the original inflection point sequence. The interval with the most data is selected as the sub-segment to extract new inflection points and add them to the original inflection point sequence. The iteration stops when the number of inflection points in the original inflection point sequence meets the preset threshold. The final time series dimensionality reduction result is output, and the segmented dimensionality reduction is completed. The phase identification module, based on the final time series dimensionality reduction result, uses the DTW algorithm to calculate the DTW distance between the voltage time series of the voltage monitoring point and the three-phase voltage time series of the transformer output terminal, and selects the target phase corresponding to the minimum DTW distance as the identification result; The anomaly detection module calculates DTW distance using a sliding time window with minute-level data windows under the same phase. It calculates the DTW distance change rate at different times within the same time window over two adjacent days, analyzes the DTW distance change rate and trend according to predetermined rules, and determines the anomaly type of the target line. The risk prediction module then quantifies the real-time risk level of the target line based on the anomaly type and DTW distance change trend. Based on historical DTW distance change rate data, it uses a time series prediction model to predict the development trend of anomalies on the target line and generates line risk prediction results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements a method for predicting the risk of a power distribution network line as described in any one of claims 1 to 7.

Images