Safe power utilization management cloud server

Abstract

The invention provides a safe power utilization management cloud server, and relates to the field of intelligent power utilization management, and the cloud server firstly receives real-time leakage current values of power utilization equipment in real time to obtain a time queue of the real-time leakage current values, and then extracts a leakage current safety threshold value from a background database; separating the current leakage current value in the time queue from the historical data, verifying the current leakage current value based on the historical data to evaluate whether the confidence coefficient of the current leakage current value meets a preset requirement, if so, calculating a difference value between the current leakage current value and a leakage current safety threshold value to obtain a leakage current offset value, and finally, judging whether the current leakage current value meets the preset requirement. And whether an early warning signal is sent to a factory administrator or not is determined by comparing the leakage current deviation value with a preset threshold value. Therefore, by improving the accuracy of electric leakage detection, electrical safety accidents can be better prevented.

Description

Technical Field

[0001] The present application relates to the field of intelligent electricity management, and more specifically, to a safe electricity management cloud server. Background Art

[0002] With the acceleration of industrial modernization, the number and types of various electrical equipment in factories have increased dramatically, and power safety has become a key factor in ensuring the smooth progress of production activities. In actual production scenarios, leakage is one of the major challenges facing power safety. Leakage can not only cause equipment damage and affect production efficiency, but more seriously, it can cause electric shock accidents, endanger people's lives, and may even cause electrical fires, causing huge property losses. Therefore, detecting and warning leakage is crucial to ensuring the safety of people and property.

[0003] However, traditional leakage detection and safety management methods have many limitations. Many factories rely on manual regular inspections, which is not only inefficient, but also difficult to monitor in real time and unable to detect sudden leakage during equipment operation. In addition, some early leakage protection devices have a single function and can only simply cut off the power supply when the leakage current exceeds a certain fixed threshold. They lack in-depth analysis and intelligent processing of historical leakage data. This single threshold method is easily affected by factors such as environmental noise and equipment aging, resulting in a high false alarm rate and missed alarm rate, thereby reducing the reliability and practicality of the device.

[0004] Therefore, a safe electricity management cloud server is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a safe electricity management cloud server.

[0006] According to one aspect of the present application, a safe electricity management cloud server is provided, which includes: A real-time leakage current value receiving module is used to receive the real-time leakage current value of the electric device to obtain a time queue of the real-time leakage current value; A leakage current safety threshold extraction module is used to extract the leakage current safety threshold from the background database; A current historical leakage current value extraction module is used to use the real-time leakage current value corresponding to the current time point in the time queue of the real-time leakage current value as the current leakage current value, and use the real-time leakage current values ​​corresponding to other time points in the time queue of the real-time leakage current value as the time queue of the historical leakage current value; A current leakage current value verification module, used to perform data verification on the current leakage current value based on the time queue of the historical leakage current value to obtain a verification result, wherein the verification result is used to indicate whether the confidence of the current leakage current value meets the preset requirements; a leakage current offset value calculation module, configured to calculate a difference between the leakage current safety threshold and the current leakage current value to obtain a leakage current offset value in response to the verification result indicating that the confidence level of the current leakage current value meets a preset requirement; The early warning module is used to determine whether to generate an early warning signal to the factory administrator client based on the comparison between the leakage current offset value and a preset threshold.

[0007] Compared with the prior art, the safe electricity management cloud server provided by the present application first receives the real-time leakage current value of the electrical equipment in real time to obtain the time queue of the real-time leakage current value, then extracts the leakage current safety threshold from the background database, and then separates the current leakage current value in the time queue from the historical data, verifies the current leakage current value based on the historical data to evaluate whether its confidence meets the preset requirements, and if it meets the preset requirements, calculates the difference between the current leakage current value and the leakage current safety threshold to obtain the leakage current offset value, and finally, determines whether to send an early warning signal to the factory manager by comparing the leakage current offset value with the preset threshold. In this way, by improving the accuracy of leakage detection, it helps to better prevent electrical safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 This is a block diagram of a safe electricity management cloud server according to an embodiment of the present application.

[0009] Figure 2 This is a block diagram of a current leakage current value verification module in a safe electricity management cloud server according to an embodiment of the present application.

[0010] Figure 3 This is a block diagram of a bidirectional encoding unit for historical leakage current values ​​in a safe electricity management cloud server according to an embodiment of the present application.

[0011] Figure 4 This is a block diagram of a historical leakage current bidirectional interaction unit in a safe electricity management cloud server according to an embodiment of the present application. DETAILED DESCRIPTION

[0012] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0013] With the continuous advancement of industrial modernization, the types and number of electrical equipment in factories have shown a rapid growth trend, and power safety has become a core link to ensure the smooth operation of production activities. In the actual production process, leakage is one of the main threats to power safety. Leakage may not only damage equipment and affect production efficiency, but more seriously, it may cause electric shock accidents, endanger personal safety, and even cause electrical fires, resulting in serious property losses. Therefore, how to detect and warn of leakage is of great significance to protecting people's lives and property safety.

[0014] However, traditional leakage detection methods and safety management methods have many shortcomings. At present, many factories still rely on manual regular inspections, which is not only inefficient and difficult to achieve real-time monitoring, but also often cannot quickly detect sudden leakage problems during equipment operation. In addition, traditional leakage protection devices have relatively simple functions, and their working principle is usually based on cutting off the power supply when the leakage current exceeds a certain fixed threshold. This method lacks in-depth analysis and intelligent processing of leakage data, and it is difficult to adapt to complex power consumption environments. Affected by environmental interference, equipment aging and other factors, such devices often have false alarms or missed alarms, which reduces the reliability and practicality of the device.

[0015] Based on this, this application proposes a safe electricity management cloud server, which can use the cloud server to monitor leakage current in real time and accurately, and can perform intelligent analysis and early warning based on equipment characteristics and operating conditions. Figure 1 FIG. 1 is a block diagram of a cloud server for safe electricity management according to an embodiment of the present application. Figure 1As shown, in the safe electricity management cloud server 100, it includes: a real-time leakage current value receiving module 110, which is used to receive the real-time leakage current value of the power-consuming equipment to obtain a time queue of the real-time leakage current value; a leakage current safety threshold extraction module 120, which is used to extract the leakage current safety threshold from the background database; a current historical leakage current value extraction module 130, which is used to use the real-time leakage current value corresponding to the current time point in the time queue of the real-time leakage current value as the current leakage current value, and use the real-time leakage current values ​​corresponding to other time points in the time queue of the real-time leakage current value as the time queue of the historical leakage current value; the current leakage current A value verification module 140 is used to perform data verification on the current leakage current value based on the time queue of the historical leakage current values ​​to obtain a verification result, and the verification result is used to indicate whether the confidence of the current leakage current value meets the preset requirements; a leakage current offset value calculation module 150 is used to calculate the difference between the leakage current safety threshold and the current leakage current value to obtain a leakage current offset value in response to the verification result that the confidence of the current leakage current value meets the preset requirements; an early warning module 160 is used to determine whether to generate an early warning signal to the factory administrator client based on a comparison between the leakage current offset value and the preset threshold.

[0016] In an embodiment of the present application, the real-time leakage current value receiving module 110 is used to receive the real-time leakage current value of the electrical equipment to obtain the time queue of the real-time leakage current value. It should be understood that the time queue of the real-time leakage current value is an ordered data set, which records the leakage current measurement value of the electrical equipment in chronological order. By analyzing this time series data, important features such as leakage trends, periodic patterns, and abnormal fluctuations of the electrical equipment can be identified. It should be noted that the safe electricity management cloud server receives the leakage current value data of the electrical equipment in real time through the Internet of Things technology. The Internet of Things technology is a technology that connects various physical devices through a network to achieve data exchange and remote control. The traditional manual inspection method has a time interval and cannot capture the sudden leakage of the equipment. Through the Internet of Things technology, the cloud server can receive the leakage current value in real time and quickly analyze the received data, which is conducive to improving the efficiency of leakage detection of electrical equipment.

[0017] The following is a detailed description of a specific implementation process of "receiving the real-time leakage current value of the electrical equipment to obtain the time queue of the real-time leakage current value": The first is hardware deployment. To accurately monitor the leakage current of electrical equipment, leakage current sensors must be reasonably installed on each electrical equipment. Due to the huge differences in the types and powers of electrical equipment in the factory, different equipment needs to use sensors that are compatible with them. For small production equipment, because of its low power and relatively low leakage current, it is necessary to use sensors with higher sensitivity and smaller range to accurately capture subtle leakage changes. For large-scale industrial production equipment, the electrical environment during operation is complex and the leakage current may be large, so it is necessary to equip it with sensors with a larger range and can adapt to complex environments. The installation location is also very critical. The sensor should be installed as close to the power access terminal of the electrical equipment as possible, so as to minimize interference during signal transmission and obtain the leakage current information of the equipment more accurately.

[0018] In order to effectively collect and process the data from each sensor, a data collector needs to be configured for each area or a group of electrical equipment. The data collector is a small data processing center that not only receives analog signals from each sensor, but also has the ability to convert these analog signals into digital form. At the same time, it needs to perform a preliminary check on the converted data to check whether the data is reasonable and whether there are abnormal fluctuations to ensure the quality of the data. In addition, the data collector must have a certain storage capacity to temporarily cache the data to prevent loss during data transmission. Moreover, the data collector must support multiple communication interfaces with the cloud server, such as Ethernet, Wi-Fi or GPRS, so as to select the most appropriate communication method according to different factory environments and actual needs, and reliably transmit the collected data to the cloud server.

[0019] The next step is data transmission. A stable network connection needs to be established between the data collector and the safe electricity management cloud server. If an Ethernet connection is used, the stability and security of the factory's internal network are particularly important. The data collector should be connected to the local area network through a switch, and parameters such as the IP address and gateway should be carefully set to ensure that the data collector can interact smoothly with the cloud server. If you choose a Wi-Fi or GPRS connection, you need to pay special attention to signal strength and network bandwidth. Insufficient signal strength may cause data transmission interruption or packet loss, while insufficient network bandwidth will affect the speed of data transmission and cannot guarantee real-time performance. Therefore, in actual applications, the most appropriate network connection method should be selected according to the specific environment and network conditions of the factory.

[0020] In order to ensure the accuracy and reliability of data transmission, it is necessary to select a suitable data transmission protocol. Common protocols such as Modbus and TCP / IP have their own characteristics. The Modbus protocol is simple and easy to use, and is very suitable for short-distance data transmission. Its implementation cost is relatively low, and it can meet some scenarios that do not require particularly high transmission speeds. The TCP / IP protocol has a wider range of applicability and better compatibility. It can run stably in both LAN and WAN environments, and can ensure reliable data transmission. In the process of data transmission, in order to prevent data from being stolen or tampered with, the data needs to be encrypted. Symmetric encryption or asymmetric encryption algorithms can be used to encrypt data before transmission, so that even if the data is intercepted during transmission, the attacker cannot obtain sensitive information.

[0021] The data collector will send the collected real-time leakage current values ​​of the electrical equipment to the cloud server at a certain time interval. This time interval can be flexibly adjusted according to actual needs, such as every 3 seconds, every 5 seconds or every 10 seconds. When sending data, the data collection timestamp will be attached at the same time. The timestamp is the "identity card" of the data. It enables the cloud server to accurately record the time information corresponding to each leakage current value, providing a key basis for the subsequent construction of the time queue.

[0022] The last part is data storage and organization. After receiving the real-time leakage current value and timestamp sent by the data collector, the safe power management cloud server will immediately conduct strict verification of the data. The verification process includes checking the integrity and accuracy of the data, and checking whether the data is missing, wrong or outliers. If there is a problem with the data, the cloud server will quickly send feedback to the data collector and ask it to resend the data to ensure that the data for subsequent analysis and processing is reliable.

[0023] The cloud server arranges the verified real-time leakage current values ​​in chronological order and builds a time queue of real-time leakage current values. The time queue can be stored in a database table or data structure (such as a queue, linked list, etc.). In the database table, special fields can be set to store leakage current values ​​and corresponding timestamps, and the data can be easily queried, sorted, and analyzed through SQL statements. When using data structure storage, data can be operated more flexibly, for example, data can be easily inserted, deleted, and searched. As new data is continuously received, the time queue will be continuously updated to always keep the latest leakage current data records, which can provide a solid and reliable data foundation for subsequent data analysis and processing.

[0024] In the embodiment of the present application, the leakage current safety threshold extraction module 120 is used to extract the leakage current safety threshold from the background database. It should be understood that the leakage current safety threshold is an important basis for judging whether the electrical equipment is in a safe operating state. Under normal circumstances, the electrical equipment will have a certain background leakage current due to various reasons (such as capacitive coupling, small leakage of insulating materials, etc.), but this current value is usually within a safe range. When the actual leakage current of the electrical equipment exceeds the leakage current safety threshold, it indicates that the equipment may have problems such as insulation damage and line failure, and there is a risk of leakage accidents. That is, by comprehensively analyzing the leakage current safety threshold and the current leakage current value, the cloud server can determine whether it is necessary to generate an early warning signal to remind the administrator to take corresponding measures to avoid the occurrence of leakage accidents. It should be noted that the leakage current safety threshold extracted from the background database will be transmitted to the cloud server via the network. On the cloud server side, all relevant data (including real-time leakage current value and safety threshold) are centrally stored in the same database, which can not only simplify the data management process, but also improve the efficiency of data analysis.

[0025] In an embodiment of the present application, the current historical leakage current value extraction module 130 is used to use the real-time leakage current value corresponding to the current time point in the time queue of the real-time leakage current value as the current leakage current value, and use the real-time leakage current values ​​corresponding to other time points in the time queue of the real-time leakage current value as the time queue of the historical leakage current value. It should be understood that the real-time leakage current value changes over time, and its change process contains information such as the operating status and potential faults of the electrical equipment. Historical data records the past operation of the equipment. By analyzing these data, the characteristics and laws such as the change mode and fluctuation range of the leakage current when the equipment is operating normally can be summarized. The current data reflects the latest status of the equipment at this moment. Distinguishing the two helps to use historical data to establish a normal model, identify immediate changes or long-term trends, and then compare them with current data, so that changes in the operating status of the equipment can be more accurately discovered.

[0026] In an embodiment of the present application, the current leakage current value verification module 140 is used to perform data verification on the current leakage current value based on the time queue of the historical leakage current value to obtain a verification result, and the verification result is used to indicate whether the confidence of the current leakage current value meets the preset requirements. Accordingly, it is considered that during the collection and transmission of leakage current data of electrical equipment, it will be interfered by various factors. For example, electromagnetic environment interference may cause deviations in sensor measurements, and sensor failures themselves may also cause inaccurate measurement results. Signal loss or distortion during data transmission will also affect data quality. Therefore, an indicator is needed to measure whether the currently measured leakage current value is accurate, stable, and not affected by noise or other interference factors, that is, confidence.

[0027] Based on this, in the current leakage current value verification module, the technical concept of the present application is to use AI-based data processing and timing feature extraction technology to perform forward and backward timing pattern analysis on the time queue of the historical leakage current value, respectively, so as to intelligently obtain the inferred leakage current value based on the core information heterogeneous correlation interaction between the forward timing pattern characteristics of the historical leakage current and the backward timing pattern characteristics of the historical leakage current, and generate the verification result based on the comparison between the offset between the inferred leakage current value and the current leakage current value and the preset confidence threshold. The present application extracts the time series characteristics of the historical leakage current from both the forward and backward directions and performs interactive analysis, which can automatically capture more complex timing patterns and trends, and provide more comprehensive data analysis results. Moreover, compared with the traditional single threshold, the present application can dynamically adjust the threshold according to the actual situation, and conduct a comprehensive evaluation in combination with the offset between the inferred leakage current value and the current leakage current value, thereby effectively reducing false alarms and missed alarms.

[0028] Specifically, Figure 2 FIG. 1 is a block diagram of a current leakage current value verification module in a safe power management cloud server according to an embodiment of the present application. Figure 2 As shown, the current leakage current value verification module 140 includes: a historical leakage current value bidirectional encoding unit 141, which is used to perform forward sequence encoding and backward sequence encoding on the time queue of the historical leakage current value respectively to obtain a historical leakage current forward time series pattern feature encoding vector and a historical leakage current backward time series pattern feature encoding vector; a historical leakage current bidirectional interaction unit 142, which is used to perform current time series core information heterogeneous correlation interaction anchoring analysis on the historical leakage current forward time series pattern feature encoding vector and the historical leakage current backward time series pattern feature encoding vector to obtain a historical leakage current bidirectional time series pattern interaction feature encoding vector; a leakage current value inference unit 143, which is used to obtain an inferred leakage current value based on the historical leakage current bidirectional time series pattern interaction feature encoding vector; and a verification result generation unit 144, which is used to generate the verification result based on the inferred leakage current value and the current leakage current value.

[0029] In the embodiment of the present application, the historical leakage current value bidirectional encoding unit 141 is used to perform forward sequence encoding and backward sequence encoding on the time queue of the historical leakage current value to obtain a historical leakage current forward time series pattern feature encoding vector and a historical leakage current backward time series pattern feature encoding vector. Specifically, Figure 3 FIG. 1 is a block diagram of a bidirectional encoding unit for historical leakage current values ​​in a safe electricity management cloud server according to an embodiment of the present application. Figure 3As shown, the historical leakage current value bidirectional encoding unit 141 includes: a historical leakage current forward time series pattern feature extraction subunit 1411, which is used to input the time queue of the historical leakage current value into a leakage current sequence encoder based on a forward LSTM model to obtain the historical leakage current forward time series pattern feature encoding vector; a historical leakage current backward time series pattern feature extraction subunit 1412, which is used to input the time queue of the historical leakage current value into a leakage current sequence encoder based on a backward LSTM model to obtain the historical leakage current backward time series pattern feature encoding vector.

[0030] In an embodiment of the present application, the historical leakage current forward time series pattern feature extraction subunit 1411 is used to input the time queue of the historical leakage current value into the leakage current sequence encoder based on the forward LSTM model to obtain the historical leakage current forward time series pattern feature encoding vector. It should be understood that the time queue of historical leakage current values ​​is time series data, and there is a temporal sequence and dependency relationship between its data points. For example, the leakage current value at a certain moment may be affected by factors such as the leakage situation at the previous moment and earlier moments, the operating status of the equipment, etc. Considering that the forward LSTM model processes the input sequence in chronological order, it controls the flow and memory of information through the internal gating mechanism (input gate, forget gate and output gate), and can effectively remember historical information and use it when needed. Based on this, the present application inputs the time queue of the historical leakage current value into the leakage current sequence encoder based on the forward LSTM model to use the forward LSTM to learn the change pattern and trend of the leakage current value from the past to the present, such as the periodic fluctuation of the current value, the trend of gradual increase or decrease, etc., to obtain the forward time series pattern feature encoding vector of the historical leakage current. In this way, the forward time series pattern features reflecting the change of the leakage current over time can be extracted from the data. These features contain the progressive relationship of the leakage situation of the equipment in time, such as the rate of change of the current value in different time periods, the fluctuation amplitude and other information, which is helpful for the subsequent in-depth analysis of the leakage law of the equipment.

[0031] The following is a detailed description of a specific implementation process of "inputting the time queue of the historical leakage current value into the leakage current sequence encoder based on the forward LSTM model to obtain the forward time series pattern feature encoding vector of the historical leakage current": First, the time queue of historical leakage current values ​​needs to be preprocessed, which generally refers to normalization. By normalizing the historical leakage current values ​​in the time queue of historical leakage current values, all data points can be within the same scale range. This step not only helps to improve the learning effect of the model, but also prevents the problem that the influence of other features is ignored due to the large value range of some features. Common normalization methods include minimum-maximum normalization and Z-score normalization. Minimum-maximum normalization scales the data to the [0,1] interval. This method is simple and intuitive, and can keep the data within a fixed range, which is convenient for model processing. Z-score normalization converts the data into a distribution with a mean of 0 and a standard deviation of 1. This method pays more attention to the distribution characteristics of the data and can make the data comparable.

[0032] The next step is to build a leakage current sequence encoder based on the forward LSTM model. The long short-term memory (LSTM) network is a special type of recurrent neural network that is particularly suitable for processing time series data. It can remember past information and affect the current output based on this information, which is very useful for analyzing the pattern of leakage current changes over time. Specifically, when designing this model, you first need to define the input layer, which is responsible for accepting the preprocessed historical leakage current value time series as input. Next is the LSTM layer, which consists of multiple LSTM units, each of which is responsible for capturing the dependencies between different time points in the input sequence. The forward LSTM model processes the input sequence in chronological order, that is, from the earliest time point to the latest time point. Finally, there is the output layer, whose task is to generate a feature encoding vector that represents the pattern and trend in the input sequence, which is the so-called historical leakage current forward time series pattern feature encoding vector.

[0033] Once the model structure is determined, a large amount of labeled historical leakage current data is needed to train the LSTM model. During the training process, the model parameters are adjusted through the back-propagation algorithm so that the model can accurately extract meaningful feature representations from the input time series. It should be noted here that choosing the right loss function and optimization algorithm is crucial to the success of the model. Commonly used loss functions include mean square error (MSE), while optimization algorithms such as Adam and SGD are effective tools for adjusting model parameters to minimize the loss function. Through repeated iterative training processes, the model gradually learns to identify important patterns and trends in the time series of historical leakage current values, thus laying a solid foundation for subsequent applications.

[0034] Once the model training is complete, it can be applied to new historical leakage current value time series to extract its forward time series pattern feature encoding vector. Specifically, the historical leakage current value time series to be analyzed is input into the trained forward LSTM model, and the model will output a feature encoding vector. This vector contains important information about the change of leakage current over time, such as periodic fluctuations, gradual increase or decrease trends, etc.

[0035] In an embodiment of the present application, the historical leakage current backward time series pattern feature extraction subunit 1412 is used to input the time queue of the historical leakage current value into the leakage current sequence encoder based on the backward LSTM model to obtain the historical leakage current backward time series pattern feature encoding vector. It should be understood that the time queue of the historical leakage current value not only has a time dependency from front to back, but also may have a dependency from back to front. Conventional forward analysis is to explore the development law in chronological order, but in some cases, the leakage condition in the future (relative to the current analysis time point) may have a potential impact on the current state. For example, early signals of certain equipment failures may gradually appear in subsequent leakage data. Therefore, in order to mine this potential dependency from the reverse time sequence and supplement the information not captured by the forward analysis, in the technical solution of the present application, the time queue of the historical leakage current values ​​is input into the leakage current sequence encoder based on the backward LSTM model to utilize the unique gating mechanism of the backward LSTM model to capture the complex reverse timing characteristics, comprehensively understand the various change patterns of the leakage current in the time series, and obtain the historical leakage current backward timing pattern feature encoding vector.

[0036] In the embodiment of the present application, the historical leakage current bidirectional interaction unit 142 is used to perform current time series core information heterogeneous correlation interaction anchoring analysis on the historical leakage current forward time series pattern feature coding vector and the historical leakage current backward time series pattern feature coding vector to obtain the historical leakage current bidirectional time series pattern interaction feature coding vector. Specifically, Figure 4 FIG. 1 is a block diagram of a historical leakage current bidirectional interaction unit in a safe power management cloud server according to an embodiment of the present application. Figure 4As shown, the historical leakage current bidirectional interaction unit 142 includes: a historical leakage current core information calculation subunit 1421, which is used to construct the core information of the historical leakage current forward time series pattern feature coding vector and the historical leakage current backward time series pattern feature coding vector to obtain the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector; a historical leakage current interaction subunit 1422, which is used to respectively calculate the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector. The feature granularity response interaction features and the feature value granularity response interaction features between the pattern feature core information anchor coding vectors are used to obtain the historical leakage current forward-backward time series pattern feature granularity response interaction coding vector and the historical leakage current forward-backward time series pattern feature value granularity response interaction coding vector; the historical leakage current bidirectional time series fusion subunit 1423 is used to fuse the historical leakage current forward-backward time series pattern feature granularity response interaction coding vector and the historical leakage current forward-backward time series pattern feature value granularity response interaction coding vector to obtain the historical leakage current bidirectional time series pattern interaction feature coding vector.

[0037] It should be understood that the forward time series pattern feature encoding vector and the backward time series pattern feature encoding vector of the historical leakage current capture the change pattern of the leakage current from different time directions respectively. The forward vector focuses on the evolution from the past to the present, while the backward vector focuses on the features from the present to the past. There are implicit interactions and feature dependency structures between these features. Therefore, in order to integrate the information of these two dimensions and more comprehensively present the complex pattern of leakage current changing over time, the present application performs an anisotropic correlation interaction anchoring analysis of the core information of the current time series on the forward time series pattern feature encoding vector of the historical leakage current and the backward time series pattern feature encoding vector of the historical leakage current to extract the core anchor point information using the autocorrelation decoupling technology, and obtains a more representative historical leakage current bidirectional time series pattern interaction feature encoding vector through the dual-level interaction modeling of feature granularity and feature value granularity, so as to provide more efficient input for subsequent leakage current value reasoning and verification.

[0038] Specifically, in the embodiment of the present application, the historical leakage current core information calculation subunit is used to: construct the characteristic autocorrelation association matrix of the historical leakage current forward time series pattern characteristic coding vector and the historical leakage current backward time series pattern characteristic coding vector to obtain the historical leakage current forward time series pattern characteristic autocorrelation association matrix and the historical leakage current backward time series pattern characteristic autocorrelation association matrix. The process can be expressed by the formula:

[0039] in, is the forward time series pattern feature encoding vector of the historical leakage current, is the historical leakage current backward time series pattern feature encoding vector, is the transpose operation, and is a feature mapping function, such as a linear mapping or a nonlinear kernel function, is the autocorrelation matrix of the historical leakage current forward time series pattern characteristics, is the autocorrelation matrix of the characteristic backward time series pattern of historical leakage current; The historical leakage current forward time series pattern feature autocorrelation matrix and the historical leakage current backward time series pattern feature autocorrelation matrix are respectively input into the core information anchoring network based on autocorrelation decoupling to obtain the historical leakage current forward time series pattern feature core information anchoring coding vector and the historical leakage current backward time series pattern feature core information anchoring coding vector. The process can be expressed by the formula:

[0040] in, is the autocorrelation matrix of the historical leakage current forward time series pattern characteristics, is the autocorrelation matrix of the historical leakage current backward time series pattern characteristics, To anchor the core information, For feature decoupling operations, , , and They are respectively Each row vector of and They are The corresponding forward weight matrix and forward bias vector, is matrix multiplication, is the forward rating vector, It is the first in the set of historical leakage current forward significant factors. The forward significant factor of the historical leakage current, is the normalization function, is the first in the set of normalized historical leakage current forward significant factors The normalized historical leakage current forward significant factor, is the The number of row vectors in is the anchor encoding vector of the core information of the forward time series pattern feature of the historical leakage current, , , and They are respectively Each row vector of and They are The corresponding backward weight matrix and backward bias vector, is the backward rating vector, It is the first in the set of historical leakage current backward significant factors. The backward significant factor of the historical leakage current, is the first in the set of normalized historical leakage current backward significant factors The normalized historical leakage current backward significant factor, is the The number of row vectors in , and and The number of It is the anchor encoding vector of the core information of the backward temporal pattern characteristics of the historical leakage current.

[0041] Among them, the forward score vector With the backward score vector It is dynamically generated through the self-attention mechanism. Specifically, during the model training phase, the forward and backward time series feature encoding vectors of the historical leakage current are input, and the multi-head attention mechanism is used to calculate the associated weights of the features of each time step to generate a scoring vector. The scoring vector reflects the core contribution of the features of different time steps to the current analysis. Its parameters are optimized through the back-propagation algorithm, with the goal of minimizing the mean square error (MSE) between the inferred leakage current value and the true value. Forward weight matrix With the forward bias vector In the core information anchoring network of autocorrelation decoupling, each row vector The corresponding weights and bias parameters are obtained through model training optimization. Specifically, during the training phase, the historical leakage current data set (including normal and abnormal samples) is used as input, and the mean square error (MSE) between the inferred leakage current value and the true value is minimized through the back propagation algorithm, and the weights and bias parameters are dynamically adjusted. Backward weight matrix and the back bias vector The setting method of is consistent with the forward parameters, but is optimized independently to capture the core information of the backward timing characteristics.

[0042] Accordingly, considering that the components of the forward and backward time series pattern feature encoding vectors of the historical leakage current are not independent of each other, but have complex correlations. By constructing the feature autocorrelation association matrix of these two time series pattern feature encoding vectors, these correlations hidden in the feature vectors can be made explicit, so that the model can analyze and understand the relationship between the features more deeply, which helps to dig out more potential information about the change of leakage current. In other words, the generated feature autocorrelation association matrix of the forward time series pattern of the historical leakage current and the feature autocorrelation association matrix of the backward time series pattern of the historical leakage current can clearly show the correlations between the originally implicit components in their corresponding feature vectors, so that the model can intuitively "see" the connection between different features, which helps the model to more accurately grasp the overall picture of the leakage current characteristics, and thus provides a clear direction for further analysis and processing.

[0043] It should be understood that the core information anchoring network based on autocorrelation decoupling is similar to a filter. Through techniques such as feature distillation, it can identify and remove redundant information in the original autocorrelation association matrix. Specifically, the two feature data containing leakage current, the autocorrelation association matrix of the forward time series pattern characteristics of the historical leakage current and the autocorrelation association matrix of the backward time series pattern characteristics of the historical leakage current, will contain some noise or repeated information that is irrelevant to the core leakage mode. This information will interfere with subsequent analysis. For example, some random data changes caused by slight fluctuations in the measurement environment are not the part that truly reflects the core feature mode of the leakage current. By inputting these two autocorrelation association matrices into the network, the noise reduction processing of the original matrix can be achieved, making the processed feature information purer. And while removing redundant feature information, the network can also focus on highlighting the parts that are highly correlated with the core feature mode. The core feature mode of leakage current data specifically refers to those information that can truly reflect the law and essential characteristics of leakage change. By highlighting these core parts, the model can more accurately grasp the key features of the leakage current, and thus provide more targeted information for subsequent analysis and decision-making.

[0044] Specifically, in the embodiment of the present application, the historical leakage current interaction subunit is used to: input the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector into the feature granularity response interaction encoder to obtain the historical leakage current forward-backward time series pattern feature granularity response interaction coding vector. The process can be expressed by the formula:

[0045] in, is the anchor encoding vector of the core information of the forward time series pattern feature of the historical leakage current, is the anchor encoding vector of the core information of the backward temporal pattern feature of the historical leakage current, It is added by location point. and are the response weight matrix and response bias vector, respectively. is the hyperbolic tangent function, is the interaction encoding vector of the characteristic granularity response of the forward-backward temporal pattern of the historical leakage current; Among them, the response weight matrix With the response bias vector It can be dynamically optimized through supervised learning. In the model training stage, the historical leakage current bidirectional time series pattern interactive feature encoding vector is used as input, and the actual leakage current value is used as the label. By minimizing the mean square error (MSE) between the predicted value and the actual value, the weight matrix and bias vector are updated using a gradient descent algorithm (such as the Adam optimizer) until the model converges.

[0046] The anchor coding vector of the feature core information of the forward time series pattern of the historical leakage current and the anchor coding vector of the feature core information of the backward time series pattern of the historical leakage current are input into the feature value granularity response interactive encoder to obtain the feature value granularity response interactive coding vector of the forward-backward time series pattern of the historical leakage current. The process can be expressed as:

[0047] in, is the anchor encoding vector of the core information of the forward time series pattern feature of the historical leakage current, is the anchor encoding vector of the core information of the backward temporal pattern feature of the historical leakage current, is the granular response interaction encoding vector of the eigenvalues ​​of the forward-backward time series pattern of the historical leakage current.

[0048] It should be understood that through the feature granularity response interactive encoder, the model can deeply explore the correlation and dependency between the anchor encoding vectors of the core information of the forward and backward time series pattern features of the historical leakage current at the micro level. For example, in the leakage current data, the current change trend at different time points may contain specific characteristic patterns. The forward feature may reflect the process of gradual development from the initial state, while the backward feature traces back the past changes from the current state. The feature granularity response interactive encoder can capture these subtle feature relationships, thereby more comprehensively understanding the pattern of leakage current changes. In other words, the generated historical leakage current forward-backward time series pattern feature granularity response interactive encoding vector can enable the model to have a more refined understanding of the leakage current characteristics, which helps to discover the laws hidden in the data details, such as the feature association of a specific leakage pattern in different time directions, thereby providing a basis for more accurate analysis of the leakage situation.

[0049] Accordingly, in order to achieve a comprehensive analysis and understanding of the leakage current data, it is necessary to connect the feature abstraction layer with the data numerical layer. In the previous steps, some analysis and processing have been performed at the feature pattern level, but only by combining it with the underlying numerical information can a complete data link be established, enabling the model to process and analyze the data from multiple angles. The present application is a key step in achieving this connection by inputting the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector into the eigenvalue granularity response interactive encoder, which can link the feature information with specific numerical values ​​and provide support for more accurate leakage current analysis. Specifically, the interdependence between the eigenvalues ​​of the historical leakage current forward and backward time series pattern feature core information anchor coding vectors is accurately characterized by element-by-element operations. It is also worth mentioning that by interactive modeling at the eigenvalue granularity, the model can learn more basic and more general leakage current data features, thereby enhancing the generalization ability of the model. This means that the model can better adapt and process leakage current data of different types and sizes, thereby improving the stability and reliability of the model.

[0050] Finally, the historical leakage current forward-backward time series pattern feature granularity response interactive coding vector and the historical leakage current forward-backward time series pattern feature value granularity response interactive coding vector are fused to obtain the historical leakage current bidirectional time series pattern interactive feature coding vector. The above process can be expressed as:

[0051] in, is the interactive encoding vector of the characteristic granularity response of the forward-backward temporal pattern of the historical leakage current, is the granular response interaction encoding vector of the forward-backward time series pattern characteristic value of the historical leakage current, For cascade operation, It is the interactive characteristic encoding vector of the historical leakage current bidirectional timing pattern.

[0052] It should be understood that the historical leakage current forward-backward time series pattern feature granularity response interactive coding vector focuses on capturing feature coupling at the micro level, and pays attention to the response relationship of features at the feature pattern level; while the historical leakage current forward-backward time series pattern feature value granularity response interactive coding vector goes deep into the bottom layer of the data, and characterizes the interdependence between feature values ​​through element-by-element operations. By combining the two, the final historical leakage current bidirectional time series pattern interactive feature coding vector can simultaneously cover multi-granularity information from feature patterns to values. In other words, the historical leakage current bidirectional time series pattern interactive feature coding vector contains both the changing trend information of the leakage pattern at the feature pattern level and the information on the mutual influence between feature values ​​at the numerical level, which can present the leakage current characteristics more comprehensively.

[0053] Preferably, in another embodiment of the present application, the historical leakage current bidirectional timing fusion subunit is used to: perform leakage current growth pattern gradient perturbation correction on the historical leakage current forward-backward timing pattern feature granularity response interaction coding vector and the historical leakage current forward-backward timing pattern feature value granularity response interaction coding vector to obtain a corrected historical leakage current forward-backward timing pattern feature granularity response interaction coding vector composed of multiple corrected historical leakage current forward eigenvalues ​​and a corrected historical leakage current forward-backward timing pattern feature value granularity response interaction coding vector composed of multiple corrected historical leakage current backward eigenvalues; fuse the corrected historical leakage current forward-backward timing pattern feature granularity response interaction coding vector and the corrected historical leakage current forward-backward timing pattern feature value granularity response interaction coding vector to obtain the historical leakage current bidirectional timing pattern interactive feature coding vector, and the process is expressed by the formula:

[0054] in, yes Middle The eigenvalues ​​at the positions, yes Middle The eigenvalues ​​at the positions, is the cosine function, yes The corrected eigenvalues, yes The corrected eigenvalues, is the corrected historical leakage current forward-backward time series pattern characteristic granularity response interaction encoding vector, is the granular response interaction encoding vector of the corrected historical leakage current forward-backward time series pattern feature value, It is the interactive characteristic encoding vector of the historical leakage current bidirectional timing pattern.

[0055] Here, considering that in the actual leakage current data analysis, the difference between the modeling representation of the characteristic granularity interaction and the eigenvalue granularity interaction of the historical leakage current forward-backward time series pattern may cause unstable perturbations in the characteristic manifold interface of the fused historical leakage current bidirectional time series pattern interaction characteristic coding vector with multi-granularity information expression, the eigenvalue granularity interaction representation of the historical leakage current forward-backward time series pattern is used as an extensibility constraint representation, so as to grow the overall characteristic interaction distribution under the eigenvalue diffusion process, and model the interface shape perturbation deviation of the overall distribution of the characteristic granularity interaction of the historical leakage current forward-backward time series pattern based on the growth exponential representation under the extensibility constraint, that is, the interaction interface gradient under the eigenvalue granularity is used as the perturbation contribution factor, and the growth exponential stabilization contribution of the gradient diffusion under the eigenvalue mutual dependence is determined, so as to realize the growth pattern gradient correction under the dominance of perturbation stabilization, and improve the manifold interface stability of the historical leakage current bidirectional time series pattern interaction characteristic coding vector.

[0056] In an embodiment of the present application, the leakage current value inference unit 143 is used to obtain the inferred leakage current value based on the historical leakage current bidirectional timing pattern interactive feature coding vector. Specifically, in an embodiment of the present application, the leakage current value inference unit is used to: input the historical leakage current bidirectional timing pattern interactive feature coding vector into a leakage current inference unit based on a decoder to obtain the inferred leakage current value. It should be understood that the decoder usually has a powerful reasoning ability in the deep learning model, and can reconstruct or predict related data based on the input feature coding. The leakage current inference unit is built based on the decoder, and uses its understanding and mapping capabilities of the input features to convert the complex feature information contained in the historical leakage current bidirectional timing pattern interactive feature coding vector into a prediction of the current possible leakage current value, providing a benchmark for subsequent comparison with the current leakage current value. In particular, in one embodiment of the present application, the historical leakage current bidirectional time series pattern interaction feature coding vector is input into a decoder-based leakage current inference device to obtain the inferred leakage current value, including: multiplying the decoding weight matrix of the decoder with the historical leakage current bidirectional time series pattern interaction feature coding vector to obtain the decoded historical leakage current bidirectional time series pattern interaction feature coding vector, and accumulating and summing all eigenvalues ​​of the decoded historical leakage current bidirectional time series pattern interaction feature coding vector to obtain the inferred leakage current value. Wherein, the decoding weight matrix is ​​obtained through supervised learning training. In the training stage, the historical leakage current bidirectional time series pattern interaction feature coding vector is used as input, and the true leakage current value at the corresponding time point is used as a label. By minimizing the mean square error (MSE) between the predicted value and the true value, the decoder parameters (for example, the decoding weight matrix) are optimized using the gradient descent algorithm until the model converges. After the training is completed, the decoding weight matrix is ​​fixed for the inference stage.

[0057] In an embodiment of the present application, the verification result generating unit 144 is used to generate the verification result based on the inferred leakage current value and the current leakage current value. Specifically, in an embodiment of the present application, the verification result generating unit is used to: divide the inferred leakage current value by the current leakage current value to obtain a leakage current offset amplitude value; and generate the verification result based on the comparison between the leakage current offset amplitude value and a preset confidence threshold. It should be understood that the magnitude of the leakage current may be different in consideration of different power usage scenarios and equipment characteristics. For example, the normal leakage current value of some large industrial equipment is relatively large, while the leakage current value of small equipment is relatively small. Simply calculating the difference between the inferred leakage current value and the current leakage current value can only reflect the absolute difference in the values ​​of the two. In order to more accurately show the degree of deviation of the current measured value from the expected value, the present application obtains the leakage current offset amplitude value by dividing the inferred leakage current value by the current leakage current value. That is to say, by dividing the leakage current offset amplitude value, the deviation between the two can be quantified in the form of a relative proportion, eliminating the influence caused by the different current magnitudes, and more accurately reflecting the degree of deviation of the current leakage current value relative to the value inferred based on the historical pattern. Then, based on the comparison between the leakage current offset amplitude value and the preset confidence threshold, the verification result is generated. That is, the leakage current offset amplitude value only reflects the degree of deviation between the current leakage current value and the inference value, but the isolated offset amplitude value cannot directly determine whether the current leakage current data is reliable. The preset confidence threshold is determined based on a large amount of historical data, equipment safety standards, actual operating experience and other factors. It can provide a clear reference standard for judging whether the leakage current offset amplitude is within a reasonable range. Specifically, comparing the leakage current offset amplitude value with the preset confidence threshold can intuitively judge the reliability of the current leakage current value. If the offset amplitude value is less than or equal to the preset confidence threshold, it means that the current leakage current value is close to the value inferred based on historical data, the confidence meets the preset requirements, and the verification result is that the current leakage current value is reliable; conversely, if the offset amplitude value is greater than the preset confidence threshold, it indicates that the current leakage current value deviates greatly from the historical pattern and there may be an abnormality, and the verification result is that the current leakage current value is unreliable.

[0058] In particular, the present application also considers that when the leakage current of the electrical equipment is monitored, various emergencies may occur, such as the failure of other electrical equipment causing the leakage current value of the equipment connected to it to suddenly increase. In this case, the current leakage current value may deviate greatly from the historical pattern, thereby affecting the reliability of the reasoning result. In order to ensure the feasibility of reasoning, the present application takes the following measures: the current leakage current value is verified by historical data to evaluate whether its confidence meets the preset requirements. If the current leakage current value deviates greatly from the historical pattern due to an emergency (such as a failure of other equipment), the leakage current offset amplitude value between the inferred leakage current value obtained by reasoning through the historical leakage current value and the current leakage current value will exceed the preset confidence threshold. At this time, the verification result is that the current leakage current value is unreliable, indicating that the leakage current value at the current time point may be an abnormal value. After finding that the current leakage current value is unreliable, the system will continue to monitor the leakage current in real time and reason about the next time series. If the emergency situation persists, as the abnormal current value data in the historical data increases, the system will gradually incorporate these abnormal values ​​into the historical data set. As time goes by, the inference model will readjust the expected leakage current value based on the new historical data, so that the inferred leakage current value is closer to the actual measured value. When the new historical data contains enough abnormal values, the leakage current offset amplitude value between the inferred leakage current value and the current leakage current value will also return to below the preset confidence threshold, and the current leakage current value collected is considered reliable. Once the current leakage current value is confirmed to be reliable, the system will continue to calculate the leakage current offset value and decide whether to generate a warning signal based on the comparison between this value and the preset threshold.

[0059] In summary, the current leakage current value verification module 140 is clearly explained, and it uses AI-based data processing and timing feature extraction technology to perform forward and backward timing pattern analysis on the time queue of historical leakage current values, respectively, so as to intelligently obtain the inferred leakage current value based on the core information heterogeneous correlation interaction between the forward timing pattern characteristics of the historical leakage current and the backward timing pattern characteristics of the historical leakage current, and generate the verification result based on the comparison between the offset between the inferred leakage current value and the current leakage current value and the preset confidence threshold. In this way, by extracting the time series characteristics of the historical leakage current from both the forward and backward directions and performing interactive analysis, more complex timing patterns and trends can be automatically captured, providing more comprehensive data analysis results. Moreover, compared with the traditional single threshold, the threshold can be dynamically adjusted according to the actual situation, and a comprehensive evaluation can be performed in combination with the offset between the inferred leakage current value and the current leakage current value, thereby effectively reducing false alarms and missed alarms.

[0060] In an embodiment of the present application, the leakage current offset value calculation module 150 is used to calculate the difference between the leakage current safety threshold and the current leakage current value to obtain the leakage current offset value in response to the verification result that the confidence of the current leakage current value meets the preset requirements. It should be understood that the current leakage current value obtained in real time will be interfered by various factors, such as errors in measuring instruments, environmental electromagnetic interference, etc. Through the previous verification analysis, the data reliability of the current leakage current value can be guaranteed. After confirming that the current leakage current value is reliable, paying attention to the difference between it and the leakage current safety threshold can directly reflect the degree of deviation between the leakage situation of the current electrical equipment and the safety standard. This is a key analysis point in safe electricity management, which helps to clarify whether the leakage state of the equipment is close to or exceeds the safety range, and thus can achieve a more accurate assessment of the leakage risk of the electrical equipment.

[0061] In an embodiment of the present application, the early warning module 160 is used to determine whether to generate an early warning signal to the factory administrator client based on the comparison between the leakage current offset value and the preset threshold value. It should be understood that by comparing with the preset threshold value, the risk can be accurately judged to ensure that the early warning signal is issued only when it is really needed. That is, when the leakage current offset value exceeds the preset threshold value, it means that the leakage of the equipment may cause safety accidents, such as electric shock, electrical fire, etc. Generating an early warning signal to the factory administrator client allows the administrator to know the hidden dangers at the first time and take corresponding measures, such as arranging equipment maintenance, stopping equipment operation, etc., so as to prevent accidents.

[0062] The following is a detailed description of a specific implementation process of "determining whether to generate a warning signal to the factory administrator client based on the comparison between the leakage current offset value and the preset threshold value": It should be noted that the reasonable setting of the preset threshold is the key basis for determining whether to issue an early warning. When setting the preset threshold, professionals need to comprehensively weigh multiple factors. Different types of electrical equipment have different tolerances to leakage due to differences in their internal structure, working principle and insulation performance. For high-precision detection instruments that are extremely sensitive to electromagnetic interference, even extremely small fluctuations in leakage current may affect their measurement accuracy. Therefore, the preset threshold of such equipment is usually set lower to ensure the stable operation of the equipment and the accuracy of the measurement results. However, for large industrial power equipment, leakage within a certain range is acceptable due to the characteristics of its own structure and protective measures, so the preset threshold will be relatively high. In addition, relevant electrical safety standards and specifications must be strictly followed to ensure that the preset threshold is set within the framework of laws and industry standards. These carefully set preset thresholds are stored in the background database of the cloud server so that they can be called at any time to provide accurate reference standards for subsequent comparative judgments.

[0063] First, the cloud server will make a real-time and rigorous comparison and judgment between the leakage current offset value and the preset threshold. The comparison process adopts an intuitive numerical comparison method, that is, directly judging whether the leakage current offset value is greater than the preset threshold. If the leakage current offset value is greater than the preset threshold, it means that the leakage of the current power-consuming equipment has exceeded the safety range, and there is a potential leakage risk, which needs to be paid attention to immediately; on the contrary, if the leakage current offset value is less than or equal to the preset threshold, it indicates that the equipment is in a normal and safe operating state. In order to further improve the accuracy and precision of the judgment, a hierarchical comparison strategy can also be adopted. For example, multiple preset thresholds of different levels are set, corresponding to different degrees of leakage risks. When the leakage current offset value exceeds the first-level warning threshold but does not reach the second-level warning threshold, it indicates that the equipment has a certain degree of leakage abnormality, but the situation is relatively mild. At this time, the cloud server will issue a mild warning signal; when the leakage current offset value exceeds the second-level warning threshold, it means that the leakage situation is more serious, which may pose a greater threat to the operation of the equipment and the safety of personnel. The cloud server will immediately issue a serious warning signal. This hierarchical comparison method enables factory managers to take appropriate measures based on different warning levels, thereby improving the efficiency and specificity of safety management.

[0064] Once it is determined that the leakage current offset value is greater than the preset threshold, the cloud server quickly generates a detailed and clear warning signal. The warning signal contains a wealth of key information, such as the specific name of the electrical equipment where the leakage occurs, the exact location, the actual value of the current leakage current, and the calculated leakage current offset value. This information is crucial for factory managers to quickly understand the overall picture of the leakage situation and help them make accurate decisions quickly. The warning signal is also sent in a variety of ways to ensure that the factory manager can be notified. Common methods include SMS, email, and special security management APP push. For mild warning signals, SMS or APP push is usually used. This method can quickly convey information to the administrator, reminding them to pay attention to the operating status of the equipment and conduct preliminary inspections and troubleshooting. For serious warning signals, in order to ensure the comprehensive communication of information and attract the administrator's attention, in addition to SMS and APP push, a detailed email will be sent to explain the situation, which will contain more in-depth data analysis and processing suggestions. When sending the warning signal, the cloud server will accurately transmit the warning signal to the factory administrator client through a stable network communication connection. In order to avoid the administrator failing to receive the warning signal due to various reasons, a multiple notification mechanism is also set up. For example, if no confirmation feedback is received from the administrator within a certain period of time after the first notification is issued, the system will automatically make a second notification, or even multiple notifications, to ensure that the warning information can be obtained by the administrator.

[0065] When the factory administrator receives the warning signal, he needs to take appropriate measures immediately. The handling methods vary depending on the severity of the warning signal. For mild warning signals, the administrator may first conduct a simple appearance inspection of the equipment to check whether there are obvious abnormal phenomena such as line damage and overheating, or use professional detection tools to conduct preliminary leakage detection on the equipment to determine whether further maintenance is required. For serious warning signals, the administrator will immediately stop the operation of the equipment and organize professional maintenance personnel to conduct a comprehensive inspection and repair of the equipment to avoid more serious safety accidents caused by leakage. After the processing is completed, the administrator needs to feedback the processing results to the cloud server. The cloud server will update the operating status of the equipment based on the feedback information and re-evaluate the leakage risk. If the processing results show that the leakage problem has been properly resolved, the cloud server will stop sending warning signals; if the problem still exists, the cloud server will continue to closely monitor the leakage situation and decide whether to issue a warning signal again based on the comparison result of the new leakage current offset value with the preset threshold, forming a complete closed-loop management process to ensure that the factory's power safety is always under control.

[0066] In summary, the safe electricity management cloud server 100 based on the embodiment of the present application is explained, which first receives the real-time leakage current value of the electrical equipment in real time to obtain the time queue of the real-time leakage current value, then extracts the leakage current safety threshold from the background database, and then separates the current leakage current value in the time queue from the historical data, verifies the current leakage current value based on the historical data to evaluate whether its confidence meets the preset requirements, and if it meets the preset requirements, calculates the difference between the current leakage current value and the leakage current safety threshold to obtain the leakage current offset value, and finally, determines whether to send an early warning signal to the factory manager by comparing the leakage current offset value with the preset threshold. In this way, by improving the accuracy of leakage detection, it helps to better prevent electrical safety accidents.

[0067] As described above, the safe power management cloud server 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a basic safe power management algorithm. In one possible implementation, the safe power management cloud server 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the safe power management cloud server 100 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the safe power management cloud server 100 can also be one of the many hardware modules of the wireless terminal.

[0068] Alternatively, in another example, the safe electricity management cloud server 100 and the wireless terminal may also be separate devices, and the safe electricity management cloud server 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

Claims

1. A safe electricity management cloud server, characterized in that: include: A real-time leakage current value receiving module is used to receive the real-time leakage current value of the electrical equipment to obtain a time queue of the real-time leakage current value; A leakage current safety threshold extraction module is used to extract the leakage current safety threshold from the background database; A current historical leakage current value extraction module is used to use the real-time leakage current value corresponding to the current time point in the time queue of the real-time leakage current value as the current leakage current value, and use the real-time leakage current values ​​corresponding to other time points in the time queue of the real-time leakage current value as the time queue of the historical leakage current value; A current leakage current value verification module, used to perform data verification on the current leakage current value based on the time queue of the historical leakage current value to obtain a verification result, wherein the verification result is used to indicate whether the confidence of the current leakage current value meets the preset requirements; a leakage current offset value calculation module, configured to calculate a difference between the leakage current safety threshold and the current leakage current value to obtain a leakage current offset value in response to the verification result indicating that the confidence level of the current leakage current value meets a preset requirement; The early warning module is used to determine whether to generate an early warning signal to the factory administrator client based on the comparison between the leakage current offset value and a preset threshold.

2. The safe electricity management cloud server according to claim 1, characterized in that: The current leakage current value verification module includes: A historical leakage current value bidirectional encoding unit, used for performing forward sequence encoding and backward sequence encoding on the time queue of the historical leakage current value to obtain a historical leakage current forward time series pattern feature encoding vector and a historical leakage current backward time series pattern feature encoding vector; A historical leakage current bidirectional interaction unit, used for performing current time series core information heterogeneous correlation interaction anchoring analysis on the historical leakage current forward time series pattern feature coding vector and the historical leakage current backward time series pattern feature coding vector to obtain the historical leakage current bidirectional time series pattern interaction feature coding vector; A leakage current value inference unit, used for obtaining an inferred leakage current value based on the historical leakage current bidirectional time series pattern interactive feature coding vector; A verification result generating unit is used to generate the verification result based on the inferred leakage current value and the current leakage current value.

3. The safe electricity management cloud server according to claim 2, characterized in that: The historical leakage current value bidirectional encoding unit comprises: A historical leakage current forward time series pattern feature extraction subunit, used for inputting the time queue of the historical leakage current value into a leakage current sequence encoder based on a forward LSTM model to obtain a historical leakage current forward time series pattern feature encoding vector; The historical leakage current backward time series pattern feature extraction subunit is used to input the time queue of the historical leakage current value into the leakage current sequence encoder based on the backward LSTM model to obtain the historical leakage current backward time series pattern feature encoding vector.

4. The safe electricity management cloud server according to claim 3, characterized in that: The historical leakage current bidirectional interaction unit comprises: A historical leakage current core information calculation subunit, used to construct the core information of the historical leakage current forward time series pattern feature coding vector and the historical leakage current backward time series pattern feature coding vector to obtain the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector; The historical leakage current interaction subunit is used to respectively calculate the characteristic granularity response interaction characteristics and the characteristic value granularity response interaction characteristics between the historical leakage current forward time series pattern characteristic core information anchor coding vector and the historical leakage current backward time series pattern characteristic core information anchor coding vector to obtain the historical leakage current forward-backward time series pattern characteristic granularity response interaction coding vector and the historical leakage current forward-backward time series pattern characteristic value granularity response interaction coding vector; The historical leakage current bidirectional timing fusion subunit is used to fuse the historical leakage current forward-backward timing pattern feature granularity response interactive coding vector and the historical leakage current forward-backward timing pattern feature value granularity response interactive coding vector to obtain the historical leakage current bidirectional timing pattern interactive feature coding vector.

5. The safe electricity management cloud server according to claim 4, characterized in that: The historical leakage current core information calculation subunit is used to: Constructing a characteristic autocorrelation association matrix of the historical leakage current forward time series pattern characteristic coding vector and the historical leakage current backward time series pattern characteristic coding vector to obtain a historical leakage current forward time series pattern characteristic autocorrelation association matrix and a historical leakage current backward time series pattern characteristic autocorrelation association matrix; The autocorrelation association matrix of the historical leakage current forward time series pattern characteristics and the autocorrelation association matrix of the historical leakage current backward time series pattern characteristics are respectively input into the core information anchoring network based on autocorrelation decoupling to obtain the core information anchoring coding vector of the historical leakage current forward time series pattern characteristics and the core information anchoring coding vector of the historical leakage current backward time series pattern characteristics.

6. The safe electricity management cloud server according to claim 5, characterized in that: The historical leakage current interaction subunit is used for: Inputting the historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector into a feature granularity response interactive encoder to obtain the historical leakage current forward-backward time series pattern feature granularity response interactive coding vector; The historical leakage current forward time series pattern feature core information anchor coding vector and the historical leakage current backward time series pattern feature core information anchor coding vector are input into the feature value granularity response interactive encoder to obtain the historical leakage current forward-backward time series pattern feature value granularity response interactive coding vector.

7. The safe electricity management cloud server according to claim 6, characterized in that: The historical leakage current bidirectional timing fusion subunit is used for: Performing leakage current growth pattern gradient perturbation correction on the historical leakage current forward-backward time series pattern characteristic granularity response interactive coding vector and the historical leakage current forward-backward time series pattern characteristic value granularity response interactive coding vector to obtain a corrected historical leakage current forward-backward time series pattern characteristic granularity response interactive coding vector composed of multiple corrected historical leakage current forward eigenvalues ​​and a corrected historical leakage current forward-backward time series pattern characteristic value granularity response interactive coding vector composed of multiple corrected historical leakage current backward eigenvalues; The corrected historical leakage current forward-backward timing pattern feature granularity response interactive coding vector and the corrected historical leakage current forward-backward timing pattern feature value granularity response interactive coding vector are fused to obtain the historical leakage current bidirectional timing pattern interactive feature coding vector.

8. The safe electricity management cloud server according to claim 7, characterized in that: The leakage current value inference unit is used to: input the historical leakage current bidirectional time series pattern interactive feature coding vector into a decoder-based leakage current inference unit to obtain the inferred leakage current value.

9. The safe electricity management cloud server according to claim 8, characterized in that: The verification result generating unit is used for: Dividing the inferred leakage current value by the current leakage current value to obtain a leakage current offset amplitude value; The verification result is generated based on a comparison between the leakage current offset amplitude value and a preset confidence threshold.

Images