GRU-based power internet of things time series data anomaly detection system

Abstract

The invention provides a GRU-based power internet of things time series data anomaly detection system, which is used for carrying out anomaly detection on a cable joint in a power distribution network and comprises a processor, the processor realizes anomaly detection through the following steps: receiving the instantaneous current of the cable joint, the real-time temperature of the cable joint and the environment temperature of the cable joint, and sampling the real-time temperature through a sliding window to obtain an original temperature sequence; according to the cable trench topology structure diagram and the real-time temperature of the power distribution network, obtaining the adjacent node temperature difference and the topology weighted temperature of the cable joint, and further obtaining the spatial comparison characteristics; according to the time domain feature, the frequency domain feature and the space comparison feature of the real-time temperature, the statistical feature of the cable joint is obtained; and in combination with the improved GRU model, obtaining an anomaly detection result of the cable joint. According to the method, through deep fusion of physical rules and data driving, early and accurate anomaly detection is realized compared with a traditional method.

Description

Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to a GRU-based electric power Internet of Things time series data anomaly detection system. Background Art

[0002] In recent years, with the large-scale development of distribution network planning projects and renovation projects, more and more newly manufactured cable joints have been put into use. However, due to the decline in material quality and the reduction of project budgets, as well as the uneven quality management and technical processes of cable joint production, the failure rate of cable joints is getting higher and higher.

[0003] Therefore, it is necessary to monitor whether the cable joints are abnormal to avoid failures of other components of the entire distribution network caused by abnormal cable joints. Summary of the Invention

[0004] The present invention provides a GRU-based power Internet of Things time series data anomaly detection system, the main purpose of which is to achieve earlier and more accurate anomaly detection than traditional methods through the deep integration of physical laws and data-driven.

[0005] An embodiment of the present invention provides a GRU-based power Internet of Things time series data anomaly detection system. The anomaly detection system is used to detect anomalies in cable joints in a distribution network. The anomaly detection system includes a processor that implements anomaly detection through the following steps:

[0006] receiving the instantaneous current of the cable joint, the real-time temperature of the cable joint, and the ambient temperature of the cable joint, and sampling the real-time temperature through a sliding window to obtain an original temperature sequence;

[0007] According to the cable trench topology structure diagram of the distribution network and the real-time temperature, the temperature difference and topology weighted temperature of adjacent nodes of the cable joint are obtained, and a spatial contrast feature is further obtained;

[0008] Obtaining statistical characteristics of the cable joint according to the time domain characteristics and frequency domain characteristics of the real-time temperature and the spatial contrast characteristics;

[0009] Obtaining an abnormality detection result of the cable joint based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence, and the statistical characteristics in combination with an improved GRU model;

[0010] In which, the improved GRU model is a dual-path heterogeneous GRU model including a physical path and a data path, the physical path is used to obtain the first temperature prediction value of the cable joint based on the thermal dynamic equation, and the data path is used to obtain the second temperature prediction value of the cable joint based on the temperature change acceleration, so that the improved GRU model obtains the anomaly detection result according to the first temperature prediction value and the second temperature prediction value.

[0011] Furthermore, the improved GRU model includes a physical path layer, a data path layer, a fusion layer, a residual layer, and a fault classification layer. The abnormality detection result of the cable joint is obtained based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence, and the statistical features in combination with the improved GRU model, and the steps include:

[0012] Inputting the instantaneous current and the ambient temperature into the physical path layer and combining them with a thermal dynamic equation to obtain the first temperature prediction value;

[0013] Inputting the original temperature sequence and the statistical features into the data path layer, and combining them with the temperature change acceleration to obtain the second temperature prediction value;

[0014] Inputting the first temperature prediction value and the second temperature prediction value into the fusion layer to obtain a temperature prediction value;

[0015] Inputting the real-time temperature and the temperature prediction value into the residual layer, and calculating the instantaneous residual and the cumulative residual;

[0016] The instantaneous residual and the cumulative residual are input into the fault classification layer to obtain the detection type of the cable joint, where the detection type includes normal, contact oxidation, insulation damage, and external heat source.

[0017] Furthermore, the instantaneous current is input into the physical path layer and combined with the thermal dynamic equation to obtain the first temperature prediction value, which is calculated as follows:

[0018]

[0019] Where, ΔT phys represents the first temperature prediction value, t represents the current time, τ represents the integral time variable, k represents the thermal resistance coefficient, I(τ) represents the instantaneous current at time τ, R represents the contact resistance of the cable joint, Δt represents the time step, α represents the heat dissipation coefficient, T(τ) represents the real-time temperature of the cable joint at time τ, T env (τ) represents the ambient temperature at time τ.

[0020] Furthermore, the original temperature sequence and the statistical features are input into the data path layer, and combined with the temperature change acceleration, the second temperature prediction value is obtained, and the calculation formula is as follows:

[0021]

[0022] r t =σ(W r ·[h t-1 ,x t ]+A·T neighbor );

[0023] Among them, z t represents the output vector of the update gate in the data path layer at time t, t represents the current time, σ represents the sigmoid function, W z represents the trainable weight matrix of the update gate, [h t-1 ,x t ] represents the concatenation of the hidden state vector at time t-1 and the input vector at time t, D z represents the physical coefficient matrix, T represents the original temperature sequence;

[0024] r t represents the output of the reset gate at time t in the data path layer, W r represents the trainable weight matrix of the reset gate, A represents the spatial attention matrix, T neighbor represents the temperature vector of the adjacent joints of the cable joint.

[0025] Furthermore, the element a in the i-th row and j-th column of the spatial attention matrix ij The calculation formula is as follows:

[0026]

[0027] Among them, a ij represents the contribution of the j-th cable joint to the state update of the i-th cable joint, W represents the feature transformation of the learning temperature correlation pattern, [T i ||T j ] represents the concatenated vector of the instantaneous temperature of the i-th cable joint and the j-th cable joint, LeakyReLU represents the leaky linear rectifier unit, and m represents the number of all adjacent joints of the i-th cable joint.

[0028] Furthermore, the adjacent joint temperature vector is calculated as follows:

[0029]

[0030] The initial value of the spatial attention matrix is as follows:

[0031]

[0032] Among them, T m (t) represents the real-time temperature of the mth adjacent joint of the cable joint at time t, represents the average real-time temperature of all adjacent joints, λ represents the thermal conductivity, and L represents the cable length.

[0033] Furthermore, the step of inputting the instantaneous residual and the cumulative residual into the fault classification layer to obtain the detection type of the cable joint includes:

[0034] If the cumulative residual is greater than the cumulative residual threshold, and the first-order derivative of the instantaneous residual is greater than the residual change rate, determining that the detection type of the cable joint is contact oxidation;

[0035] If the instantaneous residual at the current moment is greater than the instantaneous residual threshold, and the second-order derivative of the instantaneous residual is greater than 0, determining that the detection type of the cable joint is insulation damage;

[0036] If the temperature difference between the real-time temperature of the cable joint and the real-time temperature of the adjacent joint is within 2 degrees, then the detection type of the cable joint is determined to be an external heat source;

[0037] Otherwise, it is determined that the detection type of the cable connector is normal.

[0038] Furthermore, the processor communicates with the pre-buried micro-Rogowski coil, the platinum resistance temperature sensor, and the moisture-proof temperature sensor;

[0039] The micro Rogowski coil is pre-buried inside the cable connector and is used to collect the instantaneous current;

[0040] The platinum resistance temperature sensor is pre-buried inside the cable connector and is used to collect the real-time temperature;

[0041] The moisture-proof temperature sensor is installed in the cable trench of the cable connector and is used to collect the ambient temperature.

[0042] Furthermore, the temperature difference of adjacent nodes and the topology weighted temperature of the cable joint are obtained according to the cable trench topology structure diagram of the distribution network and the real-time temperature, and the calculation formula is as follows:

[0043]

[0044] T weighted =∑w ij T j ;

[0045] Where, ΔT neighborrepresents the temperature difference between adjacent nodes, T represents the real-time temperature of the cable joint, and T adjj represents the real-time temperature of the jth adjacent joint, m represents the number of all adjacent joints, T weighted represents the topological weighted temperature, w ij represents the weight coefficient between the i-th cable joint and the j-th adjacent joint, T j Indicates the real-time temperature of the j-th cable joint.

[0046] Furthermore, the time domain characteristics and frequency domain characteristics of the real-time temperature are obtained by the following steps:

[0047] Obtain multiple historical real-time temperatures through the historical sliding window;

[0048] Obtaining the time domain features according to the mean, sliding variance, first-order difference mean, and peak-to-peak value of the historical real-time temperature;

[0049] The amplitude of the main frequency component and the spectrum entropy of the historical real-time temperature are extracted to obtain the frequency domain features.

[0050] The present invention proposes a GRU-based power Internet of Things time series data anomaly detection system. Various types of sensors are pre-embedded in newly manufactured cable joints. The sensors send the collected data to a processor. The processor extracts the original temperature sequence and spatial contrast features based on the instantaneous current, real-time temperature, and ambient temperature. The processor then obtains statistical features based on the time domain features, frequency domain features, and spatial contrast features of the real-time temperature. Finally, the instantaneous current, real-time temperature, ambient temperature, original temperature sequence, and statistical features are input into an improved GRU model to obtain the anomaly detection results of the cable joint.

[0051] The advantages of the embodiments of the present invention are as follows:

[0052] 1. The GRU model is improved by using physical pathways and data pathways. The improved GRU model includes physical pathways and data pathways. The physical pathway can predict the theoretical temperature rise of the cable joint through thermodynamic equations, and the data pathway can capture the acceleration of the temperature change of the cable joint through the GRU model. The combination of the physical pathway and the data pathway makes the improved GRU model more sensitive to the temperature change of the cable joint, thereby being able to capture the abnormality of the cable joint more quickly and accurately.

[0053] 2. Using the cable trench topology diagram, the temperature difference and topology weighted temperature of adjacent nodes are calculated to establish shared information for a spatial attention mechanism. This spatial attention is then integrated into the reset gate of the data path. This reset gate blocks the propagation of false memories when the temperature of adjacent joints rises abnormally. Furthermore, when a cable joint is abnormal, the impact on the associated node model is automatically reduced. Therefore, this solution can reduce the false alarm rate of cable joint anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A schematic diagram of the structure of a GRU-based power Internet of Things time series data anomaly detection system provided by an embodiment of the present invention;

[0055] Figure 2 A flowchart of abnormality detection of a cable joint provided by an embodiment of the present invention;

[0056] Figure 3 A structural diagram of an improved GRU model provided in an embodiment of the present invention.

[0057] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0058] The embodiments of the present application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and are not to be construed as limiting the present application.

[0059] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0060] In the embodiments of the present application, at least one refers to one or more; a plurality refers to two or more. In the description of the present application, words such as "first", "second", and "third" are only used for the purpose of distinguishing descriptions, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0061] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, the terms "including," "comprising," "having," and their variations in this specification all mean "including but not limited to," unless otherwise specifically stated.

[0062] Distribution network cable joints are prone to failures, and their temperature time series data has the following characteristics:

[0063] (1) Multi-physics field coupling: Temperature is affected by multiple factors such as current, contact resistance, ambient temperature and humidity.

[0064] (2) Nonlinear hysteresis effect: Heat accumulation causes temperature changes to lag behind load changes by about 15-30 minutes.

[0065] (3) Spatial correlation: There is a conduction correlation between the temperatures of multiple joints in the same cable trench.

[0066] (4) Specific failure mode: Typical abnormalities include contact surface oxidation (slow temperature rise), insulation damage (sudden temperature rise), etc.

[0067] Figure 1 A schematic diagram of a GRU-based power IoT time series data anomaly detection system is provided in an embodiment of the present invention. Figure 1 As shown, the anomaly detection system is used to detect anomalies in cable joints in the distribution network. In newly manufactured cable joints, a micro Rogowski coil can be built into the cable joint, and the instantaneous current of the cable joint can be collected through the micro Rogowski coil. In addition, a moisture-proof temperature sensor is pre-buried in the cable trench of the cable joint. The moisture-proof temperature sensor can collect the air temperature within a range of 1 meter from the cable joint. The air temperature is the ambient temperature. Finally, a platinum resistance temperature sensor is embedded in the metal conductor of the cable joint to collect the real-time temperature of the cable joint.

[0068] In this solution, the processor of the abnormality detection system can communicate with the pre-buried micro Rogowski coil, the platinum resistance temperature sensor, and the moisture-proof temperature sensor, which can be wireless communication or wired communication.

[0069] Figure 2 A flow chart of abnormality detection of a cable joint provided by an embodiment of the present invention is as follows: Figure 2 As shown, the processor implements anomaly detection through the following steps:

[0070] S210, receiving the instantaneous current of the cable joint, the real-time temperature of the cable joint, and the ambient temperature of the cable joint, and sampling the real-time temperature through a sliding window to obtain an original temperature sequence;

[0071] Receive instantaneous current, real-time temperature and ambient temperature sent by each sensor.

[0072] After obtaining the real-time temperature corresponding to each historical moment and the current moment, the real-time temperature is sampled through a sliding window. For example, the length of the sliding window is 10 minutes, and sampling is performed every 10 seconds. The real-time temperature corresponding to 60 sampling points can be obtained. The real-time temperature corresponding to these 60 sampling points is the original temperature sequence.

[0073] S220, obtaining a temperature difference and a topology weighted temperature of adjacent nodes of the cable joint based on the cable trench topology structure diagram of the distribution network and the real-time temperature, and further obtaining a spatial contrast feature;

[0074] The cable trench topology map for the distribution network is created through multi-source data fusion and digital processing. The cable trench topology map is a topological relationship diagram consisting of nodes and edges. Nodes represent cable joints, and edges represent the connections between them. Each node has four key attributes: cable_id, length, material, and adj_nodes. cable_id represents the cable joint ID, which uniquely identifies the joint; length represents the cable length, which is used to calculate the thermal conductivity weight; material represents the material of the cable joint and determines the thermal conductivity k; and adj_nodes represents adjacent nodes, which are used for spatial attention calculation. This cable trench topology map is the basis for spatial correlation analysis in temperature anomaly detection, and its accuracy directly affects the reliability of the calculation of adjacent node temperature differences and topologically weighted temperatures.

[0075] Specifically, the calculation formulas for the temperature difference between adjacent nodes and the topology weighted temperature are as follows:

[0076]

[0077] T weighted =∑w ij T j ;

[0078] Where, ΔT neighbor represents the temperature difference between adjacent nodes, T represents the real-time temperature of the cable joint, and T adjj represents the real-time temperature of the jth adjacent joint, m represents the number of all adjacent joints, T weighted represents the topological weighted temperature, w ij represents the weight coefficient between the i-th cable joint and the j-th adjacent joint, T j Indicates the real-time temperature of the j-th cable joint.

[0079] It should be noted that when a cable joint experiences an anomaly, the source of the fault can be located by searching for the direction of maximum temperature gradient along the cable trench topology map and calculating the expected temperature rise based on the cable current carrying capacity. When a sensor in a cable joint fails, the data of the failed sensor can be reconstructed using the topological relationships of the cable trench topology map using the data of adjacent nodes. Embodiments of the present invention use the cable trench topology map to calculate the temperature relationship between the cable joint and adjacent joints, obtaining the temperature difference between adjacent nodes and the topologically weighted temperature, thereby generating spatial contrast features.

[0080] S230, obtaining statistical characteristics of the cable joint according to the time domain characteristics and frequency domain characteristics of the real-time temperature and the spatial contrast characteristics;

[0081] Specifically, multiple historical real-time temperatures are obtained through a historical sliding window;

[0082] Obtaining the time domain features according to the mean, sliding variance, first-order difference mean, and peak-to-peak value of the historical real-time temperature;

[0083] The amplitude of the main frequency component and the spectrum entropy of the historical real-time temperature are extracted to obtain the frequency domain features.

[0084] Through the historical sliding window, multiple historical real-time temperatures are collected; and the average value, sliding variance, first-order difference mean and peak-to-peak value of the historical real-time temperature are calculated. The average value, sliding variance, first-order difference mean and peak-to-peak value together constitute the time domain features. Among them:

[0085] average value

[0086] Sliding Variance

[0087] Mean of first-order differences

[0088] Peak-to-peak value max(T)-min(T).

[0089] Where n represents the number of historical real-time temperatures, p is a positive integer, T p represents the pth historical real-time temperature, max(T) represents the maximum value of all historical real-time temperatures, and min(T) represents the minimum value of all historical real-time temperatures.

[0090] Finally, the main frequency component amplitude and spectral entropy are extracted through FFT (Fast Fourier Transform). The main frequency component amplitude and spectral entropy constitute the frequency domain features. The main frequency component amplitude can reflect the load fluctuation frequency, and the spectral entropy can characterize the randomness of temperature changes.

[0091] S240, obtaining an abnormality detection result of the cable joint based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence, and the statistical features in combination with an improved GRU model;

[0092] In which, the improved GRU model is a dual-path heterogeneous GRU model including a physical path and a data path, the physical path is used to obtain the first temperature prediction value of the cable joint based on the thermal dynamic equation, and the data path is used to obtain the second temperature prediction value of the cable joint based on the temperature change acceleration, so that the improved GRU model obtains the anomaly detection result according to the first temperature prediction value and the second temperature prediction value.

[0093] By inputting the instantaneous current, real-time temperature, ambient temperature, original temperature sequence, and statistical features into an improved GRU (Gated Recurrent Units) model, anomaly detection results for the cable connector can be obtained. The improved GRU model is trained with samples and labels.

[0094] Specifically, Figure 3 A structural diagram of an improved GRU model provided by an embodiment of the present invention is shown as follows: Figure 3 As shown, the improved GRU model includes a physical path layer, a data path layer, a fusion layer, a residual layer, and a fault classification layer. The abnormality detection result of the cable joint is obtained based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence and the statistical features in combination with the improved GRU model, and the steps include:

[0095] Inputting the instantaneous current and the ambient temperature into the physical path layer and combining them with a thermal dynamic equation to obtain the first temperature prediction value;

[0096] Inputting the original temperature sequence and the statistical features into the data path layer, and combining them with the temperature change acceleration to obtain the second temperature prediction value;

[0097] Inputting the first temperature prediction value and the second temperature prediction value into the fusion layer to obtain a temperature prediction value;

[0098] Inputting the real-time temperature and the temperature prediction value into the residual layer, and calculating the instantaneous residual and the cumulative residual;

[0099] The instantaneous residual and the cumulative residual are input into the fault classification layer to obtain the detection type of the cable joint, where the detection type includes normal, contact oxidation, insulation damage, and external heat source.

[0100] First, the instantaneous current and ambient temperature are input to the physical path layer. Based on the thermal dynamics equations of electricity and physical rules, a first temperature prediction value for the cable joint is calculated. The original temperature series and statistical features are then input to the data path layer, where second-order derivative gating is introduced to capture the acceleration of temperature changes. Based on this temperature acceleration, a second temperature prediction value for the cable joint is calculated.

[0101] The data path layer includes an update gate and a reset gate. Second-order derivative gating is added to the update gate to capture the temperature acceleration of the cable joint. Spatial attention is introduced in the reset gate. When the temperature of the adjacent node rises abnormally, the propagation of false memory can be blocked. When the temperature of the adjacent node is normal, the behavior of the standard GRU model is maintained, thereby reducing the false alarm rate of cable joint anomaly detection.

[0102] In the update gate formula In, W z ·[h t-1 ,x t ] is the data-driven part, which inherits the time series modeling capability of the standard GRU model and captures the dynamic correlation of parameters such as current and temperature; For the physical enhancement part, when the temperature rises rapidly When the update gate output Z is increased t —> Reduce historical memory retention —> Faster response to sudden abnormalities; when the temperature changes steadily , the behavior of the standard GRU model is maintained.

[0103] The present invention is achieved through D z Thermodynamic laws are encoded into a neural network, and the second-order derivative term is more sensitive to sudden changes than the original temperature value, improving early warning capabilities. The influence of the physical term is automatically weakened when the temperature is stable, avoiding overfitting measurement noise. The second-order derivative term of this update gate enhances the traditional GRU model's ability to model the thermal dynamic characteristics of power equipment through a differential equation prior, making it more sensitive to temperature acceleration, providing early warning of cable joint anomalies and faster detection than traditional methods.

[0104] In the reset gate formula, W r ·[h t-1 ,x t ] is the data-driven part, which inherits the temporal memory management function of the standard GRU model; A·T neighbor For the spatial attention part, when the temperature of the adjacent node rises abnormally, the reset gate output r is increased. t—> Forced forgetting of abnormal historical states —> Blocking the propagation of false memories; when the temperature of adjacent nodes is normal, the standard GRU model behavior is maintained. This prevents the false propagation of abnormal temperature at adjacent joints. When a cable joint is abnormal, the impact on the associated node model is automatically reduced. It also transforms the physical connection relationships in the distribution network into an attention mechanism, reducing the false alarm rate due to heat conduction by 35% compared to traditional methods.

[0105] In the spatial attention matrix A, the calculation formula for each element is as follows:

[0106]

[0107] Among them, a ij represents the contribution of the j-th cable joint to the state update of the i-th cable joint, W represents the feature transformation of the learning temperature correlation pattern, [T i ||T j ] represents the concatenated vector of the instantaneous temperatures of the i-th cable joint and the j-th cable joint, which can represent the absolute temperature and relative temperature difference between the current node and the adjacent nodes. LeakyReLU represents a leaky linear rectifier unit, which can retain negative temperature difference information to prevent excessive suppression of normal temperature fluctuations. m represents traversing all adjacent joints of the i-th cable joint.

[0108] When T j Abnormally elevated—>α ij Increase—> Strengthen the abnormal propagation path. When T j Normal—>α ij Reduce->Weaken irrelevant interference. This solution prevents the false propagation of abnormal heat of adjacent nodes and automatically reduces the impact of abnormal cable joints on the associated node models when a cable joint is abnormal. It also transforms the physical connection relationship of the cable network into a trainable attention mechanism.

[0109] The first and second temperature prediction values are then input into the fusion layer for weighted fusion to produce a temperature prediction value. The weight of the first temperature prediction value is dynamically adjusted based on the current load change rate. The real-time temperature and the temperature prediction value are then input into the residual layer to calculate the instantaneous residual and cumulative residual. The instantaneous residual is the absolute difference between the real-time temperature and the temperature prediction value, and the cumulative residual is the sum of all instantaneous residuals. Finally, the instantaneous residual and cumulative residual are input into the fault classification layer to determine the cable joint detection type. This cable joint detection type includes four categories: normal, contact oxidation, insulation damage, and external heat source. Contact oxidation, insulation damage, and external heat source all indicate abnormal cable joints.

[0110] If the cumulative residual is greater than the cumulative residual threshold, and the first-order derivative of the instantaneous residual is greater than the residual change rate, then the detection type of the cable joint is determined to be contact oxidation;

[0111] If the instantaneous residual at the current moment is greater than the instantaneous residual threshold, and the second-order derivative of the instantaneous residual is greater than 0, determining that the detection type of the cable joint is insulation damage;

[0112] If the temperature difference between the real-time temperature of the cable joint and the real-time temperature of the adjacent joint is within 2 degrees, then the detection type of the cable joint is determined to be an external heat source;

[0113] Otherwise, it is determined that the detection type of the cable connector is normal.

[0114] Table 1

[0115]

[0116] Table 1 is a comparison table of fault type detection effects between the improved GRU model provided in an embodiment of the present invention and the traditional standard GRU model. As shown in Table 1, the improved GRU model in this solution has increased accuracy in the three abnormal types of increased contact resistance, partial discharge of insulation, and external heat source interference, and the improved GRU model has better effect.

[0117] Table 2

[0118]

[0119] Table 2 is a comparison table of comprehensive detection effects between the improved GRU model provided in an embodiment of the present invention and the traditional standard GRU model. As shown in Table 2, it can be seen that this solution has better performance in detection speed, false alarm rate and fault classification.

[0120] In summary, this solution achieves the following results through the deep coupling of physical laws and data-driven approaches:

[0121] Earlier: Physical equations provide prior knowledge, making it more accurate to amplify weak signals;

[0122] More accurate: spatial-temporal dual verification suppresses false positives;

[0123] Smarter: Adaptive mechanism reduces manual intervention.

[0124] Each module in the GRU-based power IoT time series data anomaly detection system can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0125] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0126] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0127] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A GRU-based power Internet of Things time series data anomaly detection system, characterized by: The anomaly detection system is used to detect anomalies in cable joints in a distribution network. The anomaly detection system includes a processor, and the processor implements anomaly detection through the following steps: receiving the instantaneous current of the cable joint, the real-time temperature of the cable joint, and the ambient temperature of the cable joint, and sampling the real-time temperature through a sliding window to obtain an original temperature sequence; According to the cable trench topology structure diagram of the distribution network and the real-time temperature, the temperature difference and topology weighted temperature of adjacent nodes of the cable joint are obtained, and a spatial contrast feature is further obtained; Obtaining statistical characteristics of the cable joint according to the time domain characteristics and frequency domain characteristics of the real-time temperature and the spatial contrast characteristics; Obtaining an abnormality detection result of the cable joint based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence, and the statistical characteristics in combination with an improved GRU model; In which, the improved GRU model is a dual-path heterogeneous GRU model including a physical path and a data path, the physical path is used to obtain the first temperature prediction value of the cable joint based on the thermal dynamic equation, and the data path is used to obtain the second temperature prediction value of the cable joint based on the temperature change acceleration, so that the improved GRU model obtains the anomaly detection result according to the first temperature prediction value and the second temperature prediction value.

2. The GRU-based power Internet of Things time series data anomaly detection system according to claim 1 is characterized in that: The improved GRU model includes a physical path layer, a data path layer, a fusion layer, a residual layer, and a fault classification layer. The abnormality detection result of the cable joint is obtained based on the instantaneous current, the real-time temperature, the ambient temperature, the original temperature sequence, and the statistical characteristics in combination with the improved GRU model, and the steps include: Inputting the instantaneous current and the ambient temperature into the physical path layer and combining them with a thermal dynamic equation to obtain the first temperature prediction value; Inputting the original temperature sequence and the statistical features into the data path layer, and combining them with the temperature change acceleration to obtain the second temperature prediction value; Inputting the first temperature prediction value and the second temperature prediction value into the fusion layer to obtain a temperature prediction value; Inputting the real-time temperature and the temperature prediction value into the residual layer, and calculating the instantaneous residual and the cumulative residual; The instantaneous residual and the cumulative residual are input into the fault classification layer to obtain the detection type of the cable joint, where the detection type includes normal, contact oxidation, insulation damage, and external heat source.

3. The GRU-based power Internet of Things time series data anomaly detection system according to claim 2 is characterized in that: The instantaneous current is input into the physical path layer and combined with the thermal dynamic equation to obtain the first temperature prediction value. The calculation formula is as follows: Where, ΔT phys represents the first temperature prediction value, t represents the current time, τ represents the integral time variable, k represents the thermal resistance coefficient, I(τ) represents the instantaneous current at time τ, R represents the contact resistance of the cable joint, Δt represents the time step, α represents the heat dissipation coefficient, T(τ) represents the real-time temperature of the cable joint at time τ, T env (τ) represents the ambient temperature at time τ.

4. The GRU-based power Internet of Things time series data anomaly detection system according to claim 2 is characterized in that: The original temperature sequence and the statistical features are input into the data path layer, and combined with the temperature change acceleration to obtain the second temperature prediction value. The calculation formula is as follows: r t =σ(W r ·[h t-1 ,x t ]+A·T neighbor ); Among them, z t represents the output vector of the update gate in the data path layer at time t, t represents the current time, σ represents the sigmoid function, W z represents the trainable weight matrix of the update gate, [h t-1 ,x t ] represents the concatenation of the hidden state vector at time t-1 and the input vector at time t, D z represents the physical coefficient matrix, T represents the original temperature sequence; r t represents the output of the reset gate at time t in the data path layer, W r represents the trainable weight matrix of the reset gate, A represents the spatial attention matrix, T neighbor represents the temperature vector of the adjacent joints of the cable joint.

5. The GRU-based power Internet of Things time series data anomaly detection system according to claim 4 is characterized in that: The element a in the i-th row and j-th column of the spatial attention matrix ij The calculation formula is as follows: Among them, a ij represents the contribution of the j-th cable joint to the state update of the i-th cable joint, W represents the feature transformation of the learning temperature correlation pattern, [T i ||T j ] represents the concatenated vector of the instantaneous temperature of the i-th cable joint and the j-th cable joint, LeakyReLU represents the leaky linear rectifier unit, and m represents the number of all adjacent joints of the i-th cable joint.

6. The GRU-based power Internet of Things time series data anomaly detection system according to claim 4 is characterized in that: The adjacent joint temperature vector is calculated as follows: Among them, T m (t) represents the real-time temperature of the mth adjacent joint of the cable joint at time t, Represents the average real-time temperature of all adjacent joints.

7. The GRU-based power Internet of Things time series data anomaly detection system according to claim 2 is characterized in that: The step of inputting the instantaneous residual and the cumulative residual into the fault classification layer to obtain the detection type of the cable joint includes: If the cumulative residual is greater than the cumulative residual threshold, and the first-order derivative of the instantaneous residual is greater than the residual change rate, determining that the detection type of the cable joint is contact oxidation; If the instantaneous residual at the current moment is greater than the instantaneous residual threshold, and the second-order derivative of the instantaneous residual is greater than 0, determining that the detection type of the cable joint is insulation damage; If the temperature difference between the real-time temperature of the cable joint and the real-time temperature of the adjacent joint is within 2 degrees, then the detection type of the cable joint is determined to be an external heat source; Otherwise, it is determined that the detection type of the cable connector is normal.

8. The GRU-based power Internet of Things time series data anomaly detection system according to claim 1 is characterized in that: The processor communicates with the pre-buried micro Rogowski coil, the platinum resistance temperature sensor, and the moisture-proof temperature sensor; The micro Rogowski coil is pre-buried inside the cable connector and is used to collect the instantaneous current; The platinum resistance temperature sensor is pre-buried inside the cable connector and is used to collect the real-time temperature; The moisture-proof temperature sensor is installed in the cable trench of the cable connector and is used to collect the ambient temperature.

9. The GRU-based power Internet of Things time series data anomaly detection system according to claim 1 is characterized in that: The temperature difference of the adjacent nodes and the topology weighted temperature of the cable joint are obtained according to the cable trench topology structure diagram of the distribution network and the real-time temperature. The calculation formula is as follows: T weighted =∑w ij T j ; Where, ΔT neighbor represents the temperature difference between adjacent nodes, T represents the real-time temperature of the cable joint, and T adjj represents the real-time temperature of the jth adjacent joint, m represents the number of all adjacent joints, T weighted represents the topological weighted temperature, w ij represents the weight coefficient between the i-th cable joint and the j-th adjacent joint, T j Indicates the real-time temperature of the j-th cable joint.

10. The GRU-based power Internet of Things time series data anomaly detection system according to claim 1 is characterized in that: The time domain characteristics and frequency domain characteristics of the real-time temperature are obtained by the following steps: Obtain multiple historical real-time temperatures through the historical sliding window; Obtaining the time domain features according to the mean, sliding variance, first-order difference mean, and peak-to-peak value of the historical real-time temperature; The amplitude of the main frequency component and the spectrum entropy of the historical real-time temperature are extracted to obtain the frequency domain features.

Images