Collaborative detection method and device for power internet of things

Abstract

The invention discloses a cooperative detection method and device for a power internet of things. The method comprises the following steps: acquiring equipment power associated data of multiple pieces of to-be-detected edge end equipment in the power Internet of Things, and extracting power feature data in the equipment power associated data based on a preset data analysis method; for each to-be-detected edge end device, determining a first detection result corresponding to the to-be-detected edge end device through a preset anomaly detection algorithm; performing cooperative detection on the first detection results of the plurality of to-be-detected edge end devices connected with the cooperative detection edge end device through the cooperative detection edge end device so as to obtain second detection results corresponding to the plurality of to-be-detected edge end devices; and determining a target detection result corresponding to the power Internet of Things based on the first detection result and the second detection result corresponding to each to-be-detected edge end device. The detection accuracy of the electric power Internet of Things and the reliability of a target detection result are improved.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things detection technology, and in particular to a collaborative detection method and device for an electric power Internet of Things. Background Art

[0002] In modern power systems, the application of power Internet of Things technology is becoming more and more extensive, and edge devices, as key components for data collection and processing, play an important role.

[0003] Edge devices are usually installed at key nodes of the power system to monitor and collect various power parameters including voltage, current, power, etc. to ensure the stable operation and reliability of the system. However, with the increasing number of power IoT devices and the increasing complexity of the power system, it is difficult to fully and effectively identify abnormal behaviors in the system, resulting in low accuracy of power IoT detection. Summary of the invention

[0004] The present invention provides a collaborative detection method and device for an electric power Internet of Things to solve the problem of low accuracy in detection of the electric power Internet of Things.

[0005] According to one aspect of the present invention, a collaborative detection method for an electric power Internet of Things is provided, the method comprising:

[0006] Collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data from the device power-related data based on a preset data analysis method;

[0007] For each edge device to be detected, determine a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm;

[0008] The collaborative detection edge device performs collaborative detection on the first detection results of multiple edge devices to be detected connected to the collaborative detection edge device to obtain second detection results corresponding to the multiple edge devices to be detected;

[0009] Determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0010] According to another aspect of the present invention, a collaborative detection device for an electric power Internet of Things is provided, the device comprising:

[0011] A data acquisition module is used to collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data from the device power-related data based on a preset data analysis method;

[0012] A local detection module, configured to determine, for each edge device to be detected, a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm;

[0013] A collaborative detection module, configured to collaboratively detect first detection results of a plurality of edge devices to be detected connected to the collaborative detection edge device, so as to obtain second detection results corresponding to the plurality of edge devices to be detected;

[0014] A detection result determination module is used to determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the collaborative detection method of the electric power Internet of Things described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the collaborative detection method of the electric power Internet of Things described in any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention is to collect the device power-related data of multiple edge devices to be detected in the power Internet of Things, extract the power feature data in the device power-related data based on the preset data analysis method; accurately extract the power feature data in the device power-related data; then, for each edge device to be detected, determine the first detection result corresponding to the edge device to be detected through a preset abnormality detection algorithm; preliminarily obtain the first detection result by performing local detection on the edge device to be detected; then, perform collaborative detection on the first detection results of multiple edge devices to be detected connected to it through the collaborative detection edge device, so as to obtain the second detection result corresponding to the multiple edge devices to be detected; through collaborative detection, further obtain the second detection result corresponding to the multiple edge devices to be detected. Finally, based on the first detection result and the second detection result corresponding to each edge device to be detected, determine the target detection result corresponding to the power Internet of Things, solve the problem of low accuracy of power Internet of Things detection, and improve the accuracy of power Internet of Things detection and the reliability of target detection results.

[0021] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 This is a flow chart of a collaborative detection method of the power Internet of Things provided according to the first embodiment of the present invention;

[0024] Figure 2 This is a flow chart of a collaborative detection method of the power Internet of Things provided according to the second embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of the structure of a collaborative detection device for an electric power Internet of Things provided according to a third embodiment of the present invention;

[0026] Figure 4 It is a structural schematic diagram of an electronic device that implements the collaborative detection method of the power Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 A flowchart of a collaborative detection method of the power Internet of Things is provided for the first embodiment of the present invention. This embodiment is applicable to the collaborative detection of the power Internet of Things. The method can be executed by a collaborative detection device of the power Internet of Things. The collaborative detection device of the power Internet of Things can be implemented in the form of hardware and / or software. The collaborative detection device of the power Internet of Things can be configured in an electronic device. Figure 1 As shown, the method includes:

[0031] S110, collecting device power-related data of multiple edge devices to be detected in the power Internet of Things, and extracting power feature data from the device power-related data based on a preset data analysis method.

[0032] Among them, edge devices can be understood as physical devices located at the edge network of the power Internet of Things. Device power-related data can be understood as relevant data of the power system collected in real time by edge devices. Power characteristic data can be understood as key characteristic parameters that reflect the status of edge devices.

[0033] Specifically, determine the types of edge devices that need to be detected (such as smart meters, transformers, generators, etc.) and their distribution locations. Collect the device power-related data of multiple edge devices to be detected in real time through sensors installed on the edge devices. And extract the power feature data in the device power-related data based on a preset data analysis method. Exemplarily, the device power-related data includes parameters such as voltage, current, power, frequency, and temperature.

[0034] Optionally, after collecting the device power related data of multiple edge devices to be detected in the power Internet of Things, it also includes: preprocessing the device power related data to obtain preprocessed device power related data, wherein the preprocessing includes at least one of data cleaning processing, missing value filling processing, noise filtering processing, normalization processing, smoothing processing and noise reduction processing.

[0035] Specifically, invalid data and erroneous data in the device power-related data are removed through data cleaning. Missing values ​​in the device power-related data are filled through missing value filling. Missing value filling can be done by interpolation or average value method. Noise in the device power-related data is removed through noise filtering. Noise filtering can use methods such as sliding average and Kalman filtering. The device power-related data is normalized by normalizing the device power-related data to eliminate the differences between different data dimensions. High-frequency noise is removed by smoothing the device power-related data.

[0036] Exemplarily, the smoothing process uses an adaptive sliding window averaging method, and the window size is adaptively adjusted according to the data change rate:

[0037]

[0038] Among them, x smooth (t) is the smoothed device power-related data; x(t) represents the device power-related data; n(t) represents the adaptive window size, which is adaptively adjusted according to the local fluctuation of the device power-related data.

[0039] The specific method of adaptively adjusting the window size according to the local fluctuation of the data is as follows:

[0040]

[0041] Where n(t) represents the adaptive window size at the current time t; C represents a constant coefficient, which is used to adjust the reference value of the window size; It represents the rate of change of the device power-related data at time t, that is, the first-order derivative of the device power-related data; Represents the ceiling function to ensure that the window size is an integer.

[0042] Optionally, the power characteristic data includes time domain characteristic data and frequency domain characteristic data; the extracting the power characteristic data from the device power-related data based on a preset data analysis method includes: extracting the time domain characteristic data from the device power-related data based on an adaptive multi-scale decomposition method; and extracting the frequency domain characteristic data from the device power-related data based on a variational mode decomposition method.

[0043] Specifically, the time domain feature data in the preprocessed device power-related data is extracted by an adaptive multi-scale decomposition method. The frequency domain feature data in the preprocessed device power-related data is extracted by a variational mode decomposition method.

[0044] Optionally, extracting the time domain feature data from the device power-related data based on an adaptive multi-scale decomposition method includes: performing multi-scale decomposition on the device power-related data by the following formula to obtain power feature components at multiple decomposition scales:

[0045]

[0046] Where x(t) represents the equipment power related data; x j (t) represents the power characteristic component at the jth scale; r(t) represents the residual term; J represents the number of decomposition scales;

[0047] Time domain characteristic data is extracted from the power characteristic component, wherein the time domain characteristic data includes at least one of a mean, a variance, a kurtosis, and a skewness.

[0048] Optionally, the decomposition process uses an adaptive filter bank to select the best decomposition scale by minimizing the error criterion. The calculation formula for selecting the best decomposition scale is as follows:

[0049]

[0050] Where J represents the number of decomposition scales; T represents the time length corresponding to the device power-related data, and there are T time points in total; x(t) represents the value of the device power-related data at time point t; x j (t) represents the value of the characteristic component at the jth scale at time t; λ represents the regularization parameter; Ω(J) represents the regularization term of the decomposition scale complexity, and ΩJ=J; represents the error term, which indicates the error between the device power-related data and the reconstructed device power-related data after decomposition; λ·Ω(J) represents the regularization term, which is used to control the number of decomposition scales.

[0051] Optionally, the extracting the frequency domain feature data from the device power related data based on a variational mode decomposition method includes:

[0052] Performing variational mode decomposition on the device power-related data to obtain a plurality of intrinsic mode functions;

[0053] The frequency domain feature data is extracted from the decomposed intrinsic mode function using the following formula:

[0054]

[0055] FrequencyEnergy=∑w | U(w)| 2 ;

[0056] FrequencyEntropy=-∑w P(w)logP(w);

[0057] Among them, U(w) represents the frequency spectrum of the local oscillator mode function, P(w) represents the probability density function of the frequency distribution, Dominant Frequency represents the dominant frequency, Frequency Energy represents the frequency energy, and Frequency Entropy represents the frequency entropy.

[0058] Specifically, variational mode decomposition is applied to the preprocessed power data to obtain multiple intrinsic mode functions, each of which represents the vibration mode of the signal within a specific frequency bandwidth:

[0059]

[0060] Among them, u k (t) represents the kth mode function; w k represents the center frequency of the kth mode function; δ(t) represents the Dirac function; * represents the convolution operation; j represents the imaginary unit; It represents the derivative operator with respect to time t, and is used to represent the rate of change with respect to time; is a complex exponential function used to frequency modulate the signal.

[0061] In an embodiment of the present invention, by combining adaptive multi-scale decomposition and variational modal decomposition, the power-related data of the equipment can be comprehensively analyzed from both the time domain and the frequency domain, and the state characteristics of the edge equipment to be detected can be fully mined; adaptive multi-scale decomposition can flexibly process non-stationary data, and variational modal decomposition can effectively separate different frequency components; multi-scale feature extraction and frequency domain modal analysis are used to improve the detection accuracy of abnormal conditions, and subtle anomalies and complex faults of the edge equipment to be detected can be accurately identified.

[0062] S120. For each edge device to be detected, determine a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm.

[0063] Among them, the first detection result can be understood as a local detection result.

[0064] Specifically, each edge device to be detected is locally detected by using a preset anomaly detection algorithm to obtain a local detection result.

[0065] S130. Perform collaborative detection on the first detection results of multiple edge devices to be detected connected to the collaborative detection edge device to obtain second detection results corresponding to the multiple edge devices to be detected.

[0066] The second detection result may be understood as a global detection result obtained through collaborative detection.

[0067] Specifically, the first detection result of the edge device to be detected is sent to the collaborative detection edge device adjacent to it, and the collaborative detection edge device performs collaborative detection on the first detection results of multiple edge devices to be detected connected to it, and the first detection results of multiple edge devices to be detected are data fused to obtain a consistent global detection result.

[0068] S140. Determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0069] Specifically, the target detection result corresponding to the electric power Internet of Things is determined based on the local detection result and the global detection result corresponding to each edge device to be detected.

[0070] The technical solution of the embodiment of the present invention is to collect the device power-related data of multiple edge devices to be detected in the power Internet of Things, extract the power feature data in the device power-related data based on the preset data analysis method; accurately extract the power feature data in the device power-related data; then, for each edge device to be detected, determine the first detection result corresponding to the edge device to be detected through a preset abnormality detection algorithm; preliminarily obtain the first detection result by performing local detection on the edge device to be detected; then, perform collaborative detection on the first detection results of multiple edge devices to be detected connected to it through the collaborative detection edge device, so as to obtain the second detection result corresponding to the multiple edge devices to be detected; through collaborative detection, further obtain the second detection result corresponding to the multiple edge devices to be detected. Finally, based on the first detection result and the second detection result corresponding to each edge device to be detected, determine the target detection result corresponding to the power Internet of Things, solve the problem of low accuracy of power Internet of Things detection, and improve the accuracy of power Internet of Things detection and the reliability of target detection results.

[0071] Embodiment 2

[0072] Figure 2 A flowchart of a collaborative detection method for an electric power Internet of Things provided in Embodiment 2 of the present invention, this embodiment is a further optimization of how to determine the first detection result corresponding to the edge device to be detected by a preset anomaly detection algorithm in the above embodiment. Optionally, the preset anomaly detection algorithm includes an isolation forest algorithm; the determination of the first detection result corresponding to the edge device to be detected by the preset anomaly detection algorithm includes: determining a first abnormal value corresponding to the edge device to be detected based on the isolation forest algorithm, and determining the first abnormal value as the first detection result corresponding to the edge device to be detected.

[0073] like Figure 2 As shown, the method includes:

[0074] S210: Collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data from the device power-related data based on a preset data analysis method.

[0075] S220. For each edge device to be detected, determine a first abnormal value corresponding to the edge device to be detected based on the isolation forest algorithm, and determine the first abnormal value as a first detection result corresponding to the edge device to be detected.

[0076] The first outlier value may be understood as an anomaly score obtained by local detection.

[0077] Specifically, a set of random trees are constructed, and each tree splits the data by randomly selecting a power feature data and a value of the power feature data. As the depth of the tree increases, the data points are continuously divided until the point is isolated individually. The tree depth of the isolated tree is used to evaluate the degree of abnormality of the data point. Each tree in the isolation forest records the isolation path length of each data point. The shorter the path, the faster the point is isolated, which usually represents an abnormal point. On the contrary, a longer path means that the point is not easily isolated and represents a normal point. For each data point, its abnormality score is calculated based on the average path length of multiple isolated trees. If the path length is shorter, the abnormality score of the data point is higher. The abnormality score can be determined by calculating the average path length of the data point in all trees and comparing it with the expected path length. Multiple abnormality scores corresponding to each data point of the edge device to be detected are determined as the first detection result corresponding to the edge device to be detected.

[0078] Optionally, the loss function of the isolation forest algorithm is formulated as follows:

[0079]

[0080] Among them, s(x i ) represents the data point x iThe anomaly score is calculated by the isolation forest algorithm; τ is the threshold used to distinguish normal points from abnormal points; λ is the regularization parameter; max(0,s(x i )-τ) means that when the anomaly score of a data point exceeds the threshold, a penalty is imposed on the anomaly point. If the anomaly score does not reach the threshold, the loss is 0; It is used to penalize the abnormal scores of normal points to ensure that the normal abnormal scores are as low as possible; n represents the total number of data points.

[0081] Optionally, the first abnormal value corresponding to the edge device to be detected is determined by the following formula:

[0082]

[0083] Among them, s(x i ) represents the data point x i The first outlier, h(x i ) represents the data point x i The average path length of all trees in the isolation forest; E(h(n)) represents the target path length under the total number of data points n.

[0084] Exemplarily, when the first outlier value s(x)≈1, it indicates that the data point is an outlier because its path is shorter. When s(x)≈0, it indicates that the data point is a normal point because its path is longer. Assume that the edge device data set in the power system contains 5 data points A, B, C, D, and E. The isolation forest algorithm is applied to construct 10 isolated trees. The average path length of data point A is 2.5, while the average path lengths of other points B, C, D, and E are 6.7, 5.8, 6.3, and 5.9, respectively. Through calculation, it is found that the average path length of A is significantly shorter than that of other data points, so A is considered an outlier and its anomaly score is higher. An example of anomaly score calculated by the isolation forest based on the path length is as follows: s(A)=0.95 (high anomaly score, indicating an outlier) s(B)=0.25 (low anomaly score, indicating a normal point) s(C)=0.3s(D)=0.28s(E)=0.32. According to the preset score threshold, if s(x) is greater than the preset score threshold (eg, 0.8), the data point is determined to be an abnormal point. The preset score threshold can be preset based on experience, and this embodiment does not limit it.

[0085] S230: Perform collaborative detection on the first detection results of multiple edge devices to be detected connected to the collaborative detection edge device to obtain second detection results corresponding to the multiple edge devices to be detected.

[0086] Specifically, after completing local detection, each edge device to be detected packages the first detection result and transmits it to the adjacent collaborative detection edge device through a low-latency network protocol between devices. Among them, the collaborative detection edge device can be understood as an edge device adjacent to the edge device to be detected in the power grid network topology. After receiving the detection results of multiple edge devices to be detected, the multiple first detection results are data fused to obtain a second detection result corresponding to the multiple edge devices to be detected.

[0087] Optionally, the collaborative detection edge device performs collaborative detection on the first detection results of multiple edge devices to be detected connected to it to obtain second detection results corresponding to the multiple edge devices to be detected, including: determining the timestamp information corresponding to each of the first detection results, and based on the timestamp information, weighted averaging the multiple first anomaly values ​​in different time windows to obtain the second anomaly value corresponding to each time window, and determining the second anomaly value corresponding to each time window as the second detection result corresponding to the multiple edge devices to be detected.

[0088] The timestamp information can be understood as a number or a string representing a certain moment, and the time window can be understood as a time interval.

[0089] Specifically, the timestamp information is used to time-align the detection results of different devices to ensure that the first detection results in the same time window are fused. The first outlier in each time window is fused using the weighted average method:

[0090]

[0091] Among them, S global S (t) represents the second abnormal value corresponding to the time window t; i (t) represents the first abnormal value of the i-th edge device to be detected in time window t; w i Represents the weight of the i-th device.

[0092] S240. Determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0093] Optionally, determining the target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected includes:

[0094] Determine an abnormal edge device detection result based on the first detection result and a preset first abnormal threshold;

[0095] Determine an abnormal time window detection result based on the second detection result and a preset second abnormal threshold;

[0096] Based on the abnormal event detection result and the abnormal time window detection result, a target detection result corresponding to the electric power Internet of Things is determined.

[0097] Specifically, within the same time window, determine the number of abnormal values ​​whose first abnormal value in the first detection result exceeds the preset first abnormal threshold. When the number of abnormal values ​​reaches the preset abnormal number threshold, it is determined that an abnormal event exists in the edge device to be detected, and the edge device to be detected with the abnormal event is determined as an abnormal edge device. Determine whether the second abnormal value corresponding to the different time windows exceeds the preset second abnormal threshold. If the second abnormal value exceeds the preset second abnormal threshold, it is determined that an abnormal event exists in the time window. Based on the abnormal edge device and the abnormal event window result, determine the target detection result corresponding to the electric power Internet of Things. Among them, the preset first abnormal threshold, the preset second abnormal threshold and the preset abnormal number threshold can be pre-set based on experience, and this embodiment does not limit them.

[0098] Abnormal events may include a failure or abnormal performance of a power device (such as a transformer, circuit breaker, sensor, etc.) in the power system. Key parameters of the equipment (such as voltage, current, temperature, etc.) show abnormal changes, exceeding the normal operating range. The communication link between edge devices is interrupted or unstable, resulting in data transmission delays or failures. The network signal strength fluctuates or the packet loss rate is too high, etc.

[0099] Optionally, within the target time window, if the first abnormal values ​​of more than half of the edge devices to be detected exceed the preset first abnormal threshold, it is considered that an abnormal event exists in the time window, and the edge devices to be detected with abnormal events are determined as abnormal edge devices. It can be expressed as follows:

[0100]

[0101] Among them, Anomaly Event indicates whether there is an abnormal event; S i (t) represents the first abnormal value, τ represents the preset first abnormal threshold, Represents an exponential function, which is 1 if the condition is met and 0 otherwise.

[0102] Furthermore, the locations of all edge devices to be detected that participate in the collaborative detection and their first detection results are analyzed by the following formula to determine the impact area of ​​the abnormal event:

[0103] Impact Area = Convex Hull ({(x i ,y i )|Si (t)>τ});

[0104] Among them, Impact Area represents the impact area of ​​the abnormal event; Convex Hull (·) represents the minimum convex hull function of the calculation point set, which represents the minimum boundary covering all abnormal edge devices; x i ,y i Respectively represent the horizontal and vertical coordinates of the position of the i-th edge device.

[0105] The severity of the abnormal event is evaluated by combining the second detection result and the number of abnormal edge devices through the following formula:

[0106]

[0107] Among them, Severity indicates the severity level of the abnormal event; Indicates the number of abnormal edge devices participating in collaborative detection.

[0108] The technical solution of the embodiment of the present invention performs local detection through an isolation forest algorithm to accurately obtain a first detection result of local detection, thereby improving the accuracy of local detection and the accuracy of the detection results of subsequent collaborative detection based on the first detection result.

[0109] As an optional example of the first embodiment of the present invention, the collaborative detection method of the power Internet of Things in this embodiment specifically includes the following steps:

[0110] Step 1: Collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and preprocess the device power-related data to obtain preprocessed device power-related data, wherein the preprocessing includes at least one of data cleaning processing, missing value filling processing, noise filtering processing, normalization processing, smoothing processing and noise reduction processing. Extract power feature data from the preprocessed device power-related data based on a preset data analysis method.

[0111] Exemplarily, the smoothing process uses an adaptive sliding window averaging method, and the window size is adaptively adjusted according to the data change rate:

[0112]

[0113] Among them, x smooth (t) is the smoothed device power-related data; x(t) represents the device power-related data; n(t) represents the adaptive window size, which is adaptively adjusted according to the local fluctuation of the device power-related data.

[0114] The specific method of adaptively adjusting the window size according to the local fluctuation of the data is as follows:

[0115]

[0116] Where n(t) represents the adaptive window size at the current time t; C represents the constant coefficient, which is used to adjust the reference value of the window size; It represents the rate of change of the device power-related data at time t, that is, the first-order derivative of the device power-related data; Represents the ceiling function to ensure that the window size is an integer.

[0117] Step 2: extracting the time domain feature data from the device power-related data based on an adaptive multi-scale decomposition method; and extracting the frequency domain feature data from the device power-related data based on a variational mode decomposition method.

[0118] Optionally, the power characteristic data includes time domain characteristic data and frequency domain characteristic data; the extracting the power characteristic data from the device power-related data based on a preset data analysis method includes: extracting the time domain characteristic data from the device power-related data based on an adaptive multi-scale decomposition method; and extracting the frequency domain characteristic data from the device power-related data based on a variational mode decomposition method.

[0119] Optionally, extracting the time domain feature data from the device power-related data based on an adaptive multi-scale decomposition method includes: performing multi-scale decomposition on the device power-related data by the following formula to obtain power feature components at multiple decomposition scales:

[0120]

[0121] Where x(t) represents the equipment power related data; x j (t) represents the power characteristic component at the jth scale; r(t) represents the residual term; J represents the number of decomposition scales;

[0122] Time domain characteristic data is extracted from the power characteristic component, wherein the time domain characteristic data includes at least one of a mean, a variance, a kurtosis, and a skewness.

[0123] Optionally, the decomposition process uses an adaptive filter bank to select the best decomposition scale by minimizing the error criterion. The calculation formula for selecting the best decomposition scale is as follows:

[0124]

[0125] Where J represents the number of decomposition scales; T represents the time length corresponding to the device power-related data, and there are T time points in total; x(t) represents the value of the device power-related data at time point t; x j(t) represents the value of the characteristic component at the jth scale at time t; λ represents the regularization parameter; Ω(J) represents the regularization term of the decomposition scale complexity, and Ω(J) = J; represents the error term, which indicates the error between the device power-related data and the reconstructed device power-related data after decomposition; λ·Ω(J) represents the regularization term, which is used to control the number of decomposition scales.

[0126] Specifically, variational mode decomposition is applied to the preprocessed power data to obtain multiple intrinsic mode functions, each of which represents the vibration mode of the signal within a specific frequency bandwidth:

[0127]

[0128] Among them, u k (t) represents the kth mode function; w k represents the center frequency of the kth mode function; δ(t) represents the Dirac function; * represents the convolution operation; j represents the imaginary unit; It represents the derivative operator with respect to time t, and is used to represent the rate of change with respect to time; is a complex exponential function used to frequency modulate the signal.

[0129] The frequency domain feature data is extracted from the decomposed intrinsic mode function using the following formula:

[0130]

[0131] FrequencyEnergy=∑w | U(w)| 2 ;

[0132] FrequencyEntropy=-∑w P(w)logP(w);

[0133] Among them, U(w) represents the frequency spectrum of the local oscillator mode function, P(w) represents the probability density function of the frequency distribution, Dominant Frequency represents the dominant frequency, Frequency Energy represents the frequency energy, and Frequency Entropy represents the frequency entropy.

[0134] In this embodiment, by combining adaptive multi-scale decomposition and variational modal decomposition, power data can be comprehensively analyzed from both time domain and frequency domain, and the equipment status characteristics can be fully mined; adaptive multi-scale decomposition can flexibly process non-stationary data, and variational modal decomposition can effectively separate different frequency components, which is suitable for a variety of power anomaly detection scenarios; multi-scale feature extraction and frequency domain modal analysis are used to improve the detection accuracy of abnormal states, and can accurately identify subtle anomalies and complex faults of equipment.

[0135] Step 3: For each edge device to be detected, determine a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm.

[0136] The anomaly detection algorithm adopts the isolation forest algorithm; since the isolation forest algorithm is an anomaly detection algorithm based on unsupervised learning, a loss function based on the isolation forest detection results is designed here to measure the anomaly detection performance of the model. This loss function combines the scores of the detected anomaly points with the separation of normal points to ensure that the model can better distinguish between normal points and anomalies.

[0137] Among them, the loss function of the isolation forest algorithm is as follows:

[0138]

[0139] Among them, s(x i ) represents the data point x i The anomaly score is calculated by the isolation forest algorithm; τ represents the threshold used to distinguish normal points from abnormal points; λ represents the regularization parameter; max(0, s(x i )-τ) means that when the anomaly score of a data point exceeds the threshold, a penalty is imposed on the anomaly point. If the anomaly score does not reach the threshold, the loss is 0. It is used to penalize the abnormal scores of normal points to ensure that the normal abnormal scores are as low as possible; n represents the total number of data points.

[0140] Optionally, the first abnormal value corresponding to the edge device to be detected is determined by the following formula:

[0141]

[0142] Among them, s(x i ) represents the data point x i The first outlier, h(x i ) represents the data point x i The average path length of all trees in the isolation forest; E(h(n)) represents the target path length under the total number of data points n.

[0143] In this embodiment, (max(0, s(x i )-τ)) 2 It is used to control the detection performance of outliers and ensure that the scores of outliers are higher than the threshold so that they can be successfully identified; It is used to suppress the abnormal scores of normal points, because the normal scores should be close to 0. This item can help the isolation forest model reduce false positives and improve the ability to distinguish normal points. It can enable the edge device to be detected to effectively balance the separation ability of abnormal points and normal points in the abnormal detection scenario, and adapt to different application scenarios and detection requirements by adjusting the threshold and regularization coefficient.

[0144] Step 4: The collaborative detection edge device performs collaborative detection on the first detection results of multiple edge devices to be detected connected to it to obtain second detection results corresponding to the multiple edge devices to be detected.

[0145] Communication and transmission of local detection results: After completing local anomaly detection, each edge device packages the detection results and transmits them to the nearby collaborative detection edge devices through a low-latency network protocol between devices.

[0146] Multi-device data fusion: After receiving the first detection results of multiple edge devices to be detected, data fusion is performed, and the detection results of different devices are time-aligned using timestamp information to ensure that the first detection results in the same time window are fused. For the first outlier in each time window, the weighted average method is used for fusion:

[0147]

[0148] Among them, S global S (t) represents the second abnormal value corresponding to the time window t; i (t) represents the first abnormal value of the i-th edge device to be detected in time window t; w i Represents the weight of the i-th device.

[0149] Step 5: Determine the target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0150] Optionally, determining the target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected includes:

[0151] Determine an abnormal edge device detection result based on the first detection result and a preset first abnormal threshold;

[0152] Determine an abnormal time window detection result based on the second detection result and a preset second abnormal threshold;

[0153] Based on the abnormal event detection result and the abnormal time window detection result, a target detection result corresponding to the electric power Internet of Things is determined.

[0154] Specifically, within the same time window, determine the number of abnormal values ​​whose first abnormal value in the first detection result exceeds the preset first abnormal threshold. When the number of abnormal values ​​reaches the preset abnormal number threshold, it is determined that an abnormal event exists in the edge device to be detected, and the edge device to be detected with the abnormal event is determined as an abnormal edge device. Determine whether the second abnormal value corresponding to the different time windows exceeds the preset second abnormal threshold. If the second abnormal value exceeds the preset second abnormal threshold, it is determined that an abnormal event exists in the time window. Based on the abnormal edge device and the abnormal event window result, determine the target detection result corresponding to the electric power Internet of Things. Among them, the preset first abnormal threshold, the preset second abnormal threshold and the preset abnormal number threshold can be pre-set based on experience, and this embodiment does not limit them.

[0155] Optionally, within the target time window, if the first abnormal values ​​of more than half of the edge devices to be detected exceed the preset first abnormal threshold, it is considered that an abnormal event exists in the time window, and the edge devices to be detected with abnormal events are determined as abnormal edge devices. It can be expressed as follows:

[0156]

[0157] Among them, Anomaly Event indicates whether there is an abnormal event; S i (t) represents the first abnormal value, τ represents the preset first abnormal threshold, Represents an exponential function, which is 1 if the condition is met and 0 otherwise.

[0158] Furthermore, the locations of all edge devices to be detected that participate in the collaborative detection and their first detection results are analyzed by the following formula to determine the impact area of ​​the abnormal event:

[0159] Impact Area = Convex Hull ({(x i ,y i )|S i (t)>τ});

[0160] Among them, Impact Area represents the impact area of ​​the abnormal event; Convex Hull (·) represents the minimum convex hull function of the calculation point set, which represents the minimum boundary covering all abnormal edge devices; x i ,y i Respectively represent the horizontal and vertical coordinates of the position of the i-th edge device.

[0161] The severity of the abnormal event is evaluated by combining the second detection result and the number of abnormal edge devices through the following formula:

[0162]

[0163] Among them, Severity indicates the severity level of the abnormal event; Indicates the number of abnormal edge devices participating in collaborative detection.

[0164] In this embodiment, multiple edge devices to be detected can collaborate to detect and fuse data to form a consistent second detection result. This ensures that abnormal events can be detected and confirmed in a timely and accurate manner in a complex power Internet of Things environment, and that the impact area and severity level of abnormal events can be assessed, which helps to take timely response measures and ensure the safe and stable operation of the power system.

[0165] The technical solution of the embodiment of the present invention effectively solves the problem that it is difficult for a single device to fully identify system anomalies through local detection by introducing a collaborative anomaly detection method for edge devices. First, the accuracy of the data is ensured by real-time collection of device power-related data and preprocessing. Next, the power feature data is extracted using time series analysis and frequency domain analysis to accurately capture changes in the state of the edge device to be detected. Then, each edge device to be detected performs local anomaly detection based on the power feature data to preliminarily identify potential abnormal data points. Finally, the detection results are shared between the edge devices to be detected, and collaborative detection is performed to form a consistent second detection result. Through the second detection result, not only can it be confirmed whether an abnormal event has occurred and when it occurred, but also its affected area and severity can be accurately assessed. The accuracy and reliability of anomaly detection are improved, the limitations of single device detection are overcome, and more comprehensive and effective anomaly monitoring is achieved.

[0166] Embodiment 3

[0167] Figure 3 This is a schematic diagram of the structure of a collaborative detection device for the power Internet of Things provided by the third embodiment of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310, a local detection module 320, a collaborative detection module 330 and a detection result determination module 340.

[0168] Among them, the data acquisition module 310 is used to collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data in the device power-related data based on a preset data analysis method; the local detection module 320 is used to determine, for each edge device to be detected, a first detection result corresponding to the edge device to be detected through a preset abnormality detection algorithm; the collaborative detection module 330 is used to perform collaborative detection on the first detection results of multiple edge devices to be detected connected to it through a collaborative detection edge device, so as to obtain a second detection result corresponding to the multiple edge devices to be detected; the detection result determination module 340 is used to determine the target detection result corresponding to the power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

[0169] The technical solution of the embodiment of the present invention is to collect the device power-related data of multiple edge devices to be detected in the power Internet of Things through a data acquisition module, extract the power feature data in the device power-related data based on a preset data analysis method; accurately extract the power feature data in the device power-related data; then, through the local detection module, for each edge device to be detected, determine the first detection result corresponding to the edge device to be detected through a preset abnormality detection algorithm; by performing local detection on the edge device to be detected, initially obtain the first detection result; then, through the collaborative detection module, the first detection result of multiple edge devices to be detected connected to it is collaboratively detected by the collaborative detection edge device to obtain the second detection result corresponding to the multiple edge devices to be detected; through collaborative detection, further obtain the second detection result corresponding to the multiple edge devices to be detected. Finally, through the detection result determination module, the target detection result corresponding to the power Internet of Things is determined based on the first detection result and the second detection result corresponding to each edge device to be detected, which solves the problem of low accuracy of power Internet of Things detection and improves the accuracy of power Internet of Things detection and the reliability of target detection results.

[0170] Optionally, the power characteristic data includes time domain characteristic data and frequency domain characteristic data; accordingly, the data acquisition module includes:

[0171] a time domain feature extraction unit, configured to extract the time domain feature data from the device power related data based on an adaptive multi-scale decomposition method; and

[0172] A frequency domain feature extraction unit is used to extract the frequency domain feature data from the device power related data based on a variational mode decomposition method.

[0173] Optionally, the time domain feature extraction unit includes:

[0174] The first decomposition subunit is used to perform multi-scale decomposition on the device power-related data by the following formula to obtain power characteristic components at multiple decomposition scales:

[0175]

[0176] Where x(t) represents the equipment power related data; x j (t) represents the power characteristic component at the jth scale; r(t) represents the residual term; J represents the number of decomposition scales;

[0177] The time domain feature extraction subunit is used to extract time domain feature data from the power feature component, wherein the time domain feature data includes at least one of mean, variance, kurtosis and skewness.

[0178] Optionally, the frequency domain feature extraction unit includes:

[0179] A second decomposition subunit is used to perform variational mode decomposition on the device power related data to obtain a plurality of intrinsic mode functions;

[0180] The frequency domain feature extraction subunit is used to extract frequency domain feature data from the decomposed intrinsic mode function using the following formula:

[0181]

[0182] Frequency Energy=∑w|U(w)| 2 ;

[0183] Frequency Entropy=-∑ w P(w)logP(w);

[0184] Among them, U(w) represents the frequency spectrum of the local oscillator mode function, P(w) represents the probability density function of the frequency distribution, Dominant Frequency represents the dominant frequency, Frequency Energy represents the frequency energy, and Frequency Entropy represents the frequency entropy.

[0185] Optionally, the local detection module is specifically used to:

[0186] A first abnormal value corresponding to the edge device to be detected is determined based on an isolation forest algorithm, and the first abnormal value is determined as a first detection result corresponding to the edge device to be detected.

[0187] Optionally, the local detection module is specifically used to:

[0188] The first abnormal value corresponding to the edge device to be detected is determined by the following formula:

[0189]

[0190] Among them, s(x i ) represents the data point x i The first outlier, h(x i ) represents the data point x i The average path length of all trees in the isolation forest; E(h(n)) represents the target path length under the total number of data points n.

[0191] Optionally, the collaborative detection module is specifically used to:

[0192] Determine the timestamp information corresponding to each of the first detection results, and based on the timestamp information, perform weighted averaging on the multiple first anomaly values ​​in different time windows to obtain the second anomaly value corresponding to each time window, and determine the second anomaly value corresponding to each time window as the second detection result corresponding to the multiple edge devices to be detected.

[0193] Optionally, the detection result determination module includes:

[0194] A device detection result determining unit, configured to determine an abnormal edge device detection result based on the first detection result and a preset first abnormal threshold;

[0195] a time window detection result determining unit, configured to determine an abnormal time window detection result based on the second detection result and a preset second abnormal threshold;

[0196] A target detection result determination unit is used to determine a target detection result corresponding to the electric power Internet of Things based on the abnormal event detection result and the abnormal time window detection result.

[0197] The device also includes a data preprocessing module.

[0198] The data preprocessing module is used to preprocess the device power-related data of multiple edge devices to be detected in the power Internet of Things after collecting the device power-related data to obtain preprocessed device power-related data, wherein the preprocessing includes at least one of data cleaning processing, missing value filling processing, noise filtering processing, normalization processing, smoothing processing and noise reduction processing.

[0199] The collaborative detection device for the electric power Internet of Things provided in the embodiment of the present invention can execute the collaborative detection method for the electric power Internet of Things provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method.

[0200] Embodiment 4

[0201] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0202] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0203] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0204] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as the collaborative detection of the power Internet of Things.

[0205] In some embodiments, the method for collaborative detection of the power Internet of Things may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for collaborative detection of the power Internet of Things described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the method for collaborative detection of the power Internet of Things in any other appropriate manner (e.g., by means of firmware).

[0206] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0207] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0208] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0209] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0210] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0211] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0212] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0213] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A collaborative detection method for power Internet of Things, characterized in that: include: Collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data from the device power-related data based on a preset data analysis method; For each edge device to be detected, determine a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm; The collaborative detection edge device performs collaborative detection on the first detection results of multiple edge devices to be detected connected to the collaborative detection edge device to obtain second detection results corresponding to the multiple edge devices to be detected; Determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

2. The method according to claim 1, characterized in that The power characteristic data includes time domain characteristic data and frequency domain characteristic data; The extracting the power characteristic data from the device power-related data based on a preset data analysis method includes: Extracting the time domain feature data from the device power related data based on an adaptive multi-scale decomposition method; as well as, The frequency domain feature data in the equipment power related data is extracted based on a variational mode decomposition method.

3. The method according to claim 2, characterized in that The extracting the time domain feature data from the device power related data based on the adaptive multi-scale decomposition method includes: The device power-related data is decomposed at multiple scales using the following formula to obtain power characteristic components at multiple decomposition scales: Where x(t) represents the equipment power related data; x j (t) represents the power characteristic component at the jth scale; r(t) represents the residual term; J represents the number of decomposition scales; Time domain characteristic data is extracted from the power characteristic component, wherein the time domain characteristic data includes at least one of a mean, a variance, a kurtosis, and a skewness.

4. The method according to claim 2, characterized in that: The extracting the frequency domain feature data from the equipment power related data based on the variational mode decomposition method includes: Performing variational mode decomposition on the device power-related data to obtain a plurality of intrinsic mode functions; The frequency domain feature data is extracted from the decomposed intrinsic mode function using the following formula: Frequency Energy=Σ w |U(w)| 2 ; Frequency Entropy=-∑ w P(w)logP(w); Among them, U(w) represents the frequency spectrum of the local oscillator mode function, P(w) represents the probability density function of the frequency distribution, DominantFrequency represents the dominant frequency, Frequency Enerogy represents the frequency energy, and Frequency Entropy represents the frequency entropy.

5. The method according to claim 1, characterized in that The preset anomaly detection algorithm includes an isolation forest algorithm; and determining the first detection result corresponding to the edge device to be detected by the preset anomaly detection algorithm includes: A first abnormal value corresponding to the edge device to be detected is determined based on an isolation forest algorithm, and the first abnormal value is determined as a first detection result corresponding to the edge device to be detected.

6. The method according to claim 5, characterized in that The first abnormal value corresponding to the edge device to be detected is determined by the following formula: Among them, s(x i ) represents the data point x i The first outlier, h(x i ) represents the data point x i The average path length of all trees in the isolation forest; E(h(n)) represents the target path length under the total number of data points n.

7. The method according to claim 5, characterized in that The collaborative detection of the first detection results of the plurality of edge devices to be detected connected to the collaborative detection edge device to obtain second detection results corresponding to the plurality of edge devices to be detected includes: Determine the timestamp information corresponding to each of the first detection results, and based on the timestamp information, perform weighted averaging on the multiple first anomaly values ​​in different time windows to obtain the second anomaly value corresponding to each time window, and determine the second anomaly value corresponding to each time window as the second detection result corresponding to the multiple edge devices to be detected.

8. The method according to claim 1, characterized in that The determining, based on the first detection result and the second detection result corresponding to each edge device to be detected, a target detection result corresponding to the electric power Internet of Things includes: Determine an abnormal edge device detection result based on the first detection result and a preset first abnormal threshold; Determine an abnormal time window detection result based on the second detection result and a preset second abnormal threshold; Based on the abnormal event detection result and the abnormal time window detection result, a target detection result corresponding to the electric power Internet of Things is determined.

9. The method according to claim 1, characterized in that: After collecting the device power related data of multiple edge devices to be detected in the power Internet of Things, it also includes: The device power-related data is preprocessed to obtain preprocessed device power-related data, wherein the preprocessing includes at least one of data cleaning processing, missing value filling processing, noise filtering processing, normalization processing, smoothing processing and noise reduction processing.

10. A collaborative detection device for power Internet of Things, characterized in that: include: A data acquisition module is used to collect device power-related data of multiple edge devices to be detected in the power Internet of Things, and extract power feature data from the device power-related data based on a preset data analysis method; A local detection module, configured to determine, for each edge device to be detected, a first detection result corresponding to the edge device to be detected by using a preset anomaly detection algorithm; A collaborative detection module, configured to collaboratively detect first detection results of a plurality of edge devices to be detected connected to the collaborative detection edge device, so as to obtain second detection results corresponding to the plurality of edge devices to be detected; A detection result determination module is used to determine a target detection result corresponding to the electric power Internet of Things based on the first detection result and the second detection result corresponding to each edge device to be detected.

Images