Gas relay quick response method based on Internet of Things

Abstract

The invention relates to the field of gas relays, and particularly discloses a gas relay quick response method based on the Internet of Things, which ensures the accuracy and reliability of event data through front-end signal digitization and state machine verification, and lays a foundation for analysis. The cloud abandons static threshold judgment, introduces dynamic time sequence analysis, and constructs an enhanced feature vector capable of reflecting the fault development trend by calculating the change rate and acceleration of environmental parameters. And further, fault probability distribution is generated by using Bayesian inference to quantify uncertainty, decision optimization is performed based on an expected risk minimization principle, and the lowest risk of a response instruction is ensured. According to the scheme, the analysis dimension is improved from a static snapshot process to a dynamic process, decision logic is upgraded from a deterministic rule to probabilistic reasoning, the problem of misjudgment caused by analysis superficial layer is solved, and the response accuracy and reliability are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of gas relay, and more specifically, to a gas relay fast response method based on Internet of Things. BACKGROUND

[0002] As a key protection device for ensuring the safe operation of power transformers, the core function of gas relay is to monitor the gas generated by the decomposition of insulating oil or electrical faults in the transformer in real time, and trigger an alarm or trip signal when the preset threshold is reached, so as to prevent the expansion of faults and avoid causing major economic losses and power grid safety accidents. In the traditional operation and maintenance mode, the alarm signal of the gas relay is usually triggered locally, relying on the on-site patrol of the operation and maintenance personnel or the simple forwarding of remote signals, the response chain is long, the processing efficiency is low, and it is difficult to meet the urgent needs of modern power grids for fast and accurate fault disposal. Therefore, combining Internet of Things technology with gas relay monitoring to build an intelligent solution that can realize remote monitoring and fast response has become an important development direction for the intelligent upgrading of power systems.

[0003] However, in the prior art, although the Internet of Things application scheme realizes the remote transmission of the alarm signal, there are still significant limitations in data processing and decision analysis. The cloud platform research and judgment mechanism of these schemes has defects in the shallow analysis dimension and rigid logic judgment when processing the alarm data uploaded by the gas relay. Specifically, the existing technology is difficult to accurately distinguish between real emergency faults and alarms caused by environmental factors interference or non-emergency conditions under complex working conditions, resulting in high false alarm rate and missed alarm rate, affecting the reliability and effectiveness of fast response. This is because the existing cloud platform analysis logic tends to use environmental temperature, humidity and other context information as isolated and static parameters for threshold judgment to correct the authenticity of the alarm, and fails to deeply mine the internal relationship between the dynamic process of these data over time and the alarm event itself. For example, the system cannot effectively distinguish between the alarm triggered by the continuous abnormal rise in temperature caused by the sudden increase in transformer load and the alarm triggered by the stable high temperature caused by the strong midday sunlight. This snapshot analysis of data ignores the trend information of the development of physical quantities before the occurrence of faults. The judgment logic based on fixed rules or simple models will greatly reduce the accuracy and reliability of diagnosis when facing complex or atypical working conditions, and is likely to cause misjudgment or underestimate the severity of potential risks.

[0004] Therefore, an optimized gas relay fast response scheme based on Internet of Things is expected. SUMMARY

[0005] In order to solve the above technical problems, the present application is proposed. Embodiments of the present application provide a gas relay fast response method based on Internet of Things.

[0006] According to an aspect of the present application, a gas relay fast response method based on Internet of Things is provided, which comprises:

[0007] Collecting an original electrical signal in response to detecting an internal physical obstacle of a transformer;

[0008] Digitizing the original electrical signal and encapsulating initial data to obtain an initial data packet;

[0009] Based on a local environment data stream, locally enhancing and preprocessing the initial data packet to obtain a preprocessed data load;

[0010] Encrypting the preprocessed data load to obtain an encrypted data stream, and transmitting the encrypted data stream to a cloud platform;

[0011] The cloud platform parses the encrypted data stream to obtain the preprocessed data load, and intelligently judges the preprocessed data load to obtain a judgment result;

[0012] Based on the judgment result, a fast response control instruction is generated.

[0013] In the above gas relay fast response method based on Internet of Things, digitizing the original electrical signal and encapsulating initial data to obtain an initial data packet comprises: sampling the original electrical signal at a preset sampling rate to obtain an original electrical level sequence; performing software debouncing on the original electrical level sequence to obtain a debounced signal level; a state machine mode continuously monitors the debounced signal level, wherein when the debounced signal level is detected to transition from one stable state to another stable state, the state machine mode determines that a verified state change occurs and triggers an effective state change trigger signal; the verified state change is matched with a preset event mapping table to obtain a digital state value; in response to the effective state change trigger signal, a timestamp and a device ID are collected; the timestamp, the device ID and the digital state value are structured and encapsulated to obtain the initial data packet.

[0014] In the above gas relay fast response method based on Internet of Things, based on a local environment data stream, locally enhancing and preprocessing the initial data packet to obtain a preprocessed data load comprises: obtaining a local environment data stream, the local environment data stream comprising environmental temperature, humidity and GPS coordinates of the device; recombining the local environment data stream and the initial data packet to obtain the preprocessed data load, the preprocessed data load comprising identity information, event data and environmental context information.

[0015] In the above gas relay fast response method based on the Internet of Things, the identity information is a device ID, the event data includes a timestamp and a digital state value, and the environmental context information includes environmental temperature, humidity, and GPS coordinates of the device.

[0016] In the above gas relay fast response method based on the Internet of Things, the data encryption is performed on the pre-processed data load to obtain an encrypted data stream, and the encrypted data stream is transmitted to the cloud platform, including: serializing the pre-processed data load to obtain serialized data; and encrypting the serialized data to obtain the encrypted data stream.

[0017] In the above gas relay fast response method based on the Internet of Things, the cloud platform analyzes the encrypted data stream to obtain the pre-processed data load, and intelligently judges the pre-processed data load to obtain a judgment result, including: extracting environmental context information from the pre-processed data load, and calculating a temperature change rate, a humidity change rate, a temperature change acceleration, and a humidity change acceleration based on the environmental context information; extracting a digital state value from the pre-processed data load; concatenating the digital state value, the environmental temperature, the humidity, the temperature change rate, the humidity change rate, the temperature change acceleration, and the humidity change acceleration to obtain an enhanced feature vector; generating a fault probability distribution based on Bayesian inference on the enhanced feature vector to obtain a fault probability distribution; and based on a cost matrix, performing decision optimization based on expected risk minimization on the fault probability distribution to obtain the judgment result.

[0018] In the above gas relay fast response method based on the Internet of Things, the fault probability distribution is generated based on Bayesian inference on the enhanced feature vector to obtain a fault probability distribution, including: generating the fault probability distribution based on Bayesian inference on the enhanced feature vector according to the following formula:

[0019]

[0020] where V is the enhanced feature vector, P(V|F k ) is the likelihood, and P(F k ) is the prior probability of the kth fault.

[0021] In the above gas relay fast response method based on the Internet of Things, based on the cost matrix, the decision optimization based on the expected risk minimization is performed on the fault probability distribution to obtain the judgment result, including: based on the cost matrix, the decision optimization based on the expected risk minimization is performed on the fault probability distribution according to the following formula to obtain the judgment result:

[0022]

[0023] wherein, A j is a cost matrix, P(F k |V) is a failure probability distribution, F k is a prior probability of the kth failure, A opt is a judgment result.

[0024] Compared with the prior art, the gas relay rapid response method based on the Internet of Things provided by the application lays a foundation for analysis by ensuring the accuracy and reliability of event data through front-end signal digitization and state machine verification. The cloud discards static threshold judgment and introduces dynamic timing analysis to construct an enhanced feature vector that can reflect the failure development trend by calculating the rate of change and acceleration of environmental parameters. Furthermore, the failure probability distribution is generated by using Bayesian inference to quantify uncertainty, and decision optimization is performed based on the principle of minimum expected risk to ensure the lowest risk of response instructions. The scheme improves the analysis dimension from a static snapshot to a dynamic process and upgrades the decision logic from deterministic rules to probabilistic reasoning, solving the misjudgment problem caused by shallow analysis and improving the accuracy and reliability of the response. BRIEF DESCRIPTION OF DRAWINGS

[0025] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application, when taken in conjunction with the accompanying drawings. The drawings provided in the present application are used to provide further understanding of embodiments of the present application, and constitute a part of the specification, and are used to explain the present application together with embodiments of the present application, and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 is a flowchart of the gas relay rapid response method based on the Internet of Things according to the embodiments of the present application;

[0027] Figure 2 is a data flow diagram of the gas relay rapid response method based on the Internet of Things according to the embodiments of the present application;

[0028] Figure 3 is a flowchart of signal digitization and initial data packaging of the original electrical signal to obtain an initial data packet of the gas relay rapid response method based on the Internet of Things according to the embodiments of the present application;

[0029] Figure 4 is a flowchart of local data enhancement and preprocessing of the initial data packet to obtain a preprocessed data load based on the local environmental data flow of the gas relay rapid response method based on the Internet of Things according to the embodiments of the present application;

[0030] Figure 5A flowchart of a process of analyzing encrypted data stream to obtain pre-processed data load and intelligently judging the pre-processed data load to obtain a judgment result by a cloud platform of a gas relay fast response method based on Internet of Things according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Hereinafter, example embodiments according to the present application will be described in detail with reference to accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application. It should be understood that the present application is not limited to the described embodiments.

[0032] As shown in the present application and claims, unless the context clearly indicates otherwise, "one", "an", "a", and / or "the" do not mean to specify a single number, but also include a plurality. Generally, the terms "comprise" and "include" only indicate the inclusion of the steps and elements explicitly identified, and these steps and elements do not constitute an exclusive list, and the method or device can also include other steps or elements.

[0033] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0034] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in sequence. Instead, various steps can be processed in reverse order or simultaneously, as needed. Meanwhile, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0035] Hereinafter, example embodiments according to the present application will be described in detail with reference to accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application. It should be understood that the present application is not limited to the described example embodiments.

[0036] The prior art cloud platform analysis mechanism has a shallow analysis dimension and rigid logic for gas relay alarm, and cannot effectively distinguish false alarms caused by environmental factors from real emergency failures, thereby affecting the reliability of rapid response. In order to solve the above technical problems, in the technical scheme of the present application, a gas relay rapid response method based on Internet of Things is proposed. Specifically, first, at the device end, the original electrical signal is finely digitized, including software debouncing and state machine modal verification, to ensure that the captured state change is unambiguous and highly accurate, and an initial data packet containing an accurate timestamp and device ID is immediately encapsulated, laying a reliable data foundation for subsequent analysis. Subsequently, the data packet is securely transmitted to the cloud platform after being fused with environmental data streams (such as temperature and humidity) locally. In the cloud, the present scheme discards the traditional static threshold judgment and instead performs deep time series mining on the environmental data, constructing an enhanced feature vector that can depict the dynamic trend of fault development by calculating its change rate and acceleration. Based on this high-dimensional feature, the system uses a Bayesian inference framework to generate a probability distribution that quantifies the likelihood of each type of fault, rather than giving a single, arbitrary conclusion. Finally, by combining a pre-set cost matrix, the probability distribution is optimized through an expected risk minimization decision model to generate the most scientific and risk-averse rapid response control instruction, realizing a fundamental change from simple alarm reporting to intelligent and forward-looking fault disposal.

[0037] Figure 1 A flowchart of the gas relay rapid response method based on Internet of Things according to an embodiment of the present application. Figure 2 A data flow diagram of the gas relay rapid response method based on Internet of Things according to an embodiment of the present application is shown in Figure 1 and Figure 2 The gas relay rapid response method based on Internet of Things according to an embodiment of the present application includes: S100, in response to detecting an internal physical obstacle of a transformer, collecting an original electrical signal; S200, digitizing the original electrical signal and encapsulating the initial data to obtain an initial data packet; S300, based on local environmental data streams, performing local data enhancement and preprocessing on the initial data packet to obtain a preprocessed data load; S400, encrypting the preprocessed data load to obtain an encrypted data stream, and transmitting the encrypted data stream to a cloud platform; S500, the cloud platform analyzes the encrypted data stream to obtain the preprocessed data load, and intelligently analyzes the preprocessed data load to obtain an analysis result; S600, based on the analysis result, a rapid response control instruction is generated.

[0038] Specifically, in step S100, the original electrical signal is collected in response to detecting the physical barrier inside the transformer. It should be understood that the physical barrier inside the transformer, such as the accumulation of gas due to the decomposition of insulating oil or the action of the baffle due to the impact of oil flow, is a direct physical representation of a potential fault, and the mechanical action of the gas relay is a direct response to this physical representation. Therefore, in the technical solution of the present application, the original electrical signal is collected in response to detecting the physical barrier inside the transformer, so as to capture the most original and direct mapping of the fault event in the electrical layer. That is, specifically, the physical barrier inside the transformer, such as the accumulation of gas or the impact of oil flow, will drive the mechanical components inside the gas relay, such as the float or the baffle, to displace. The mechanical displacement will link one or more electrical switching elements, such as reed switches or micro switches, to change the contact state, i.e., from the open state to the closed state, or from the closed state to the open state. The change of the switch state will directly cause the electrical characteristics of the monitoring circuit where the switch is located to change, thereby generating an original electrical signal that can be collected by the electronic device, which is manifested as a transition of voltage or current level. In this way, it can be ensured that the trigger source of the entire fast response process is real and timely, providing an untreated and high-fidelity initial input for all subsequent data processing and intelligent research.

[0039] More specifically, in one specific example of the present application, when a slight fault occurs inside the transformer and slowly generates gas, the gas accumulates at the top of the relay, causing the upper float to sink due to the decrease in buoyancy. The permanent magnet linked to the upper float approaches the reed switch, and when it reaches the preset alarm position, the magnetic field strength is sufficient to cause the internal contacts of the reed switch to attract. The closure of this contact will ground a signal collection circuit that was originally in a high resistance state, causing the output voltage of the circuit to quickly drop from a stable high level (e.g., 3.3V) to a stable low level (e.g., 0V). The data collection unit continuously monitors the voltage of the signal collection circuit, thereby collecting the complete process of this level change, which constitutes the original electrical signal in response to the light gas alarm.

[0040] Specifically, in step S200, the original electrical signal is signal digitized and initial data is packaged to obtain an initial data packet. It should be understood that, since the original electrical signal generated by the mechanical action of the gas relay contact is inevitably accompanied by noise interference such as level jitter, and it is only an analog physical quantity without specific meaning. Therefore, in the technical solution of the present application, the original electrical signal is further signal digitized and initial data is packaged to obtain an initial data packet, so as to convert the unstable analog signal into a structured digital event with clear spatiotemporal context information. In this way, the uncertainty can be eliminated from the data source, providing a high-fidelity, standardized data basis for subsequent transmission and cloud intelligent research and judgment, and ensuring the accuracy and reliability of the entire response system.

[0041] Figure 3 A flowchart of the process of signal digitization and initial data packaging of the original electrical signal to obtain an initial data packet according to the gas relay fast response method based on the Internet of Things according to the embodiments of the present application. As shown in Figure 3 S200, it includes: S210, sampling the original electrical signal at a preset sampling rate to obtain an original level sequence; S220, software de-bouncing the original level sequence to obtain a de-bounced signal level; S230, a state machine mode continuously monitors the de-bounced signal level, wherein when it is detected that the de-bounced signal level transitions from one stable state to another stable state, the state machine mode determines that a verified state change occurs and triggers an effective state change trigger signal; S240, matching the verified state change with a preset event mapping table to obtain a digital state value; S250, in response to the effective state change trigger signal, collecting a time stamp and a device ID; S260, structuring and packaging the time stamp, the device ID and the digital state value to obtain the initial data packet.

[0042] That is, the implementation of this step consists of first, the data acquisition unit continuously samples the raw electrical signal at a pre-set, sufficiently high sampling rate, converting it from a continuous analog waveform to a discrete raw level sequence. Next, the raw level sequence is processed by a software debouncing algorithm, which filters out the short-lived, spurious level fluctuations caused by the mechanical bounce of the contact by setting a stabilization period judgment window in the time dimension, thus outputting a stable, glitch-free debounced signal level. Subsequently, the state machine mode continuously monitors the debounced signal level, and only when it detects a clear transition of the level from one known stable state to another and remains for a sufficiently long time, the state machine determines that a verified state change has occurred and triggers a unique valid state change trigger signal. In response to this trigger signal, two parallel operations are immediately performed: on the one hand, the verified state change (e.g. a transition from high to low) is matched with a pre-set event mapping table to obtain the digital state value it represents; on the other hand, the current timestamp is obtained from the system real-time clock and the unique device ID of the device is read from the non-volatile memory. Finally, the obtained timestamp, device ID and digital state value, these three key information elements, are organized and packaged according to a pre-defined structure to form the final initial data packet.

[0043] More specifically, in one specific example of the present application, when a voltage transition from 3.3V to 0V representing a light gas alarm is collected, the data acquisition unit samples it at a sampling rate of 1kHz, obtaining a raw level sequence containing jitter noise. The software debouncing program, after detecting that the level first falls below a certain threshold, waits and confirms that the level remains in the low level range for the next 20 milliseconds, thus generating a stable debounced signal level of 0V. The state machine mode detects that the signal transitions from a stable high level state to a stable low level state, and immediately determines that a verified state change has occurred and issues a valid state change trigger signal. The system, in response to this signal, matches this high-to-low level change with the event mapping table to obtain the digital state value 1 representing the light gas alarm. At the same time, the current timestamp is collected and the device ID is read. Finally, these three data are packaged into a structured initial data packet, ready for subsequent processing.

[0044] Specifically, in step S300, based on the local environment data stream, the initial data packet is locally data enhanced and preprocessed to obtain a preprocessed data load. It should be understood that since an initial data packet only contains events, time and identity, it cannot reflect the physical environment conditions at the time of alarm occurrence by itself, and these environmental factors are crucial for accurately determining the authenticity and severity of the fault. Therefore, in the technical solution of the present application, based on the local environment data stream, the initial data packet is locally data enhanced and preprocessed to obtain a preprocessed data load, so as to bind the core event data with the real-time environmental context information occurring at the time of data leaving the occurrence site. In this way, an information dimension richer and context more complete preprocessed data load can be provided for intelligent research and judgment of the cloud platform, laying a foundation for subsequent complex correlation analysis and trend prediction, thereby greatly improving the accuracy of fault diagnosis.

[0045] Figure 4 For the method for quickly responding to the gas relay based on the Internet of Things according to the embodiment of the present application, the flowchart of locally data enhancing and preprocessing the initial data packet based on the local environment data stream to obtain a preprocessed data load. As shown in Figure 4 S300, it includes: S310, acquiring a local environment data stream, the local environment data stream including environmental temperature, humidity and GPS coordinates of the device; S320, recombining the local environment data stream with the initial data packet to obtain the preprocessed data load, the preprocessed data load including identity information, event data and environmental context information. In particular, it is worth mentioning that here, the identity information is a device ID, the event data includes a timestamp and a digital state value, and the environmental context information includes environmental temperature, humidity and GPS coordinates of the device.

[0046] That is, specifically, the implementation process of this step is: first, after generating the initial data packet, the local data acquisition unit immediately issues a data acquisition instruction to the integrated or connected environmental sensors (such as digital temperature and humidity sensors) and positioning modules (such as GPS receivers). Second, the environmental sensors and positioning modules respond to the instruction and measure the current environmental temperature, humidity and geographical position coordinates of the device, respectively, and return these data to the data acquisition unit to form a local environment data stream. Finally, the data acquisition unit creates a new data entity with a more complex structure, i.e., a preprocessed data load, and fills the identity information and event data extracted from the initial data packet and the newly acquired environmental context information into the corresponding fields of the data entity, respectively, to complete the recombination and enhancement of the data.

[0047] More specifically, in one specific example of the present application, after the data acquisition unit of the device generates an initial data packet containing a light gas alarm (digital state value 1), the microcontroller inside it immediately initiates a query to the on-board temperature and humidity sensor and GPS module. The sensor returns the current ambient temperature as 28.5°C and the humidity as 75% RH. The GPS module returns the current coordinates. The microcontroller then constructs a pre-processed data load, in which the identity information field is assigned to the device ID; the event data field contains the timestamp and the digital state value; and the environmental context information field contains the temperature 28.5°C, humidity 75% RH and the above GPS coordinates. This complete pre-processed data load is then submitted to the encryption transmission module.

[0048] Specifically, in step S400, the pre-processed data load is data-encrypted to obtain an encrypted data stream, and the encrypted data stream is transmitted to the cloud platform. It should be understood that, since the pre-processed data load contains sensitive information related to the state and location of the key equipment of the power system, when transmitted through an open network such as the Internet of Things, it faces the risk of eavesdropping, tampering or forgery. Therefore, in the technical solution of the present application, the pre-processed data load is further data-encrypted to obtain an encrypted data stream, and the encrypted data stream is transmitted to the cloud platform, so as to impose a solid security barrier on the data before it leaves the local terminal, and ensure its confidentiality and integrity in the transmission link. In this way, unauthorized third parties can be effectively prevented from obtaining or tampering with the key alarm data, ensuring that the data received by the cloud platform is real and reliable, and providing a fundamental security guarantee for the effectiveness of subsequent intelligent research and judgment.

[0049] More specifically, in the embodiment of the present application, data-encrypting the pre-processed data load to obtain an encrypted data stream and transmitting the encrypted data stream to the cloud platform comprises: serializing the pre-processed data load to obtain serialized data; and encrypting the serialized data to obtain the encrypted data stream. That is, specifically, the implementation process of this step is to first convert the structured pre-processed data load in the memory into a continuous byte stream, i.e. serialized data, through a serialization protocol such as JSON or Protobuf, so that it is suitable for network transmission. Next, the local data acquisition unit calls its internal encryption engine, uses a pre-set or negotiated device-specific key with the cloud platform, applies a high-strength symmetric encryption algorithm such as AES-256 to the serialized data for encryption processing, thereby generating the final encrypted data stream. Finally, the encrypted data stream is handed over to the wireless communication module of the device, and sent to the designated receiving address of the cloud platform through the established secure transmission channel based on TLS / DTLS, for example.

[0050] Specifically, in step S500, the cloud platform parses the encrypted data stream to obtain a preprocessed data load, and intelligently judges the preprocessed data load to obtain a judgment result. It should be understood that the existing cloud platform judgment mechanism has the core defects of shallow analysis dimension and rigid logic judgment when processing the alarm data uploaded by the gas relay. This mechanism tends to use environmental temperature, humidity and other context information as isolated parameters to correct the authenticity of the alarm, and fails to deeply mine the internal relationship between the dynamic process of these data evolving over time and the alarm event itself. This snapshot analysis of data ignores the development trend of physical quantities before the fault occurs. For example, it cannot distinguish between alarms triggered by a continuously rising temperature and alarms triggered by a stable high temperature. Therefore, this method is difficult to capture the complete picture of fault evolution, and its judgment logic based on fixed rules or simple models will greatly reduce the accuracy and reliability of the diagnosis when facing complex or atypical working conditions, which may lead to misjudgment or insufficient assessment of the severity of potential risks. That is, although the preprocessed data load uploaded from the front-end device contains snapshot information of the event and the environment, it does not have diagnostic capabilities and cannot reveal the fault development trend and potential risks hidden behind the data. Therefore, in the technical solution of the present application, the cloud platform parses the encrypted data stream to obtain a preprocessed data load, and intelligently judges the preprocessed data load to obtain a judgment result, so as to convert discrete and static data points into deep insights that can represent the dynamic behavior of the system, and make optimal decisions based on probability reasoning and risk assessment. In this way, the limitations of traditional static threshold-based judgment can be fundamentally overcome, and accurate, forward-looking and risk-controllable intelligent diagnosis of transformer faults can be achieved.

[0051] Figure 5 A flowchart of the cloud platform of the gas relay fast response method based on the Internet of Things according to the embodiments of the present application parsing the encrypted data stream to obtain a preprocessed data load, and intelligently judging the preprocessed data load to obtain a judgment result. As shown in Figure 5 S500, it includes: S510, extracting environmental context information from the preprocessed data load, and calculating a temperature change rate, a humidity change rate, a temperature change acceleration, and a humidity change acceleration based on the environmental context information; S520, extracting a digital state value from the preprocessed data load; S530, concatenating the digital state value, the environmental temperature, the humidity, the temperature change rate, the humidity change rate, the temperature change acceleration, and the humidity change acceleration to obtain an enhanced feature vector; S540, generating a fault probability distribution based on Bayesian inference on the enhanced feature vector to obtain a fault probability distribution; S550, based on a cost matrix, performing decision optimization based on expected risk minimization on the fault probability distribution to obtain the judgment result.

[0052] More specifically, in the technical solution of the present application, an intelligent research and judgment process based on dynamic timing characteristics and probability reasoning is proposed. First, dynamic timing characteristic vectorization is performed. Considering that original, isolated data points, such as current temperature values, cannot fully represent a complex physical process. The incubation and development of a fault is a dynamic process, and the key information is contained in the trend of data change. Therefore, discrete and static data points are upgraded to a high-dimensional feature vector that can describe the dynamic behavior of the system. Specifically, this goal is achieved by introducing calculus methods, specifically, calculating the first and second derivatives of key environmental parameters at the time of alarm occurrence to obtain temperature change rate, humidity change rate, temperature change acceleration, and humidity change acceleration.

[0053] In the scenario of transformer monitoring, the first derivative represents the instantaneous change rate of parameters such as temperature, reflecting the rapidity of fault development; the second derivative represents the acceleration of this trend, revealing whether the fault is accelerating or slowing down. Then, a digital state value is extracted from the preprocessed data load, and by concatenating the digital state value, environmental temperature, humidity, temperature change rate, humidity change rate, temperature change acceleration, and humidity change acceleration, an enhanced feature vector is generated that includes the current state, change rate, and change acceleration. This provides a highly information-dense data basis for subsequent accurate research and judgment, equivalent to drawing a unique dynamic fingerprint for each alarm event.

[0054] Next, fault probability distribution generation based on Bayesian inference is performed. Considering the complexity and uncertainty of the failure mechanism of power equipment, a single feature combination may correspond to multiple potential fault modes. Traditional deterministic logic judgment is too arbitrary and ignores this inherent ambiguity. Therefore, a probability reasoning framework is used to quantify the confidence of each possible fault occurrence under the current observed dynamic features. Specifically, Bayesian theorem is used to construct this reasoning capability, that is, the enhanced feature vector is subjected to fault probability distribution generation based on Bayesian inference to obtain the fault probability distribution, using the following formula:

[0055]

[0056] where V is the enhanced feature vector, P(V|F k ) is the likelihood, P(F k ) is the prior probability of the kth fault, and P(F k |V) is the fault probability distribution.

[0057] Correspondingly, it can calculate the probability P(F kposterior probability of the fault. This calculation process integrates the historical occurrence probability of the fault itself (prior probability P(F k )) and the correlation strength (likelihood P(V|F k )) between the specific fault and the specific dynamic feature fingerprint learned by the pre-trained model. In this way, the output of the system is no longer a black-or-white diagnostic conclusion, but a more detailed and realistic fault probability distribution, which presents the likelihood of all potential faults in the form of a probability list, providing a rich and comprehensive perspective for subsequent decision-making.

[0058] Finally, decision optimization based on expected risk minimization is implemented. The value of a diagnostic conclusion ultimately lies in whether it can guide the correct response measures, and different measures have completely different consequences (i.e., risks or costs) under different real faults. An optimal decision must systematically weigh all possibilities and their corresponding consequences. Therefore, the probability distribution obtained in the previous step is converted into specific, risk-optimal action instructions. For this purpose, the expected risk calculation model in decision theory is introduced, that is, based on the cost matrix, the fault probability distribution is optimized for decision-making based on expected risk minimization to obtain the said judgment result, which is as follows:

[0059]

[0060] where A j is the cost matrix, P(F k |V) is the fault probability distribution, F k is the prior probability of the kth fault, and A opt is the judgment result.

[0061] Specifically, in the technical solution of the present application, the expected risk of each alternative action (such as immediate tripping or sending an inspector) is calculated, that is, the cost of all possible fault scenarios is considered and weighted by the probability of the occurrence of the fault scenario. Then, the system will automatically select the action that minimizes the total expected risk as the final recommendation. In this way, the system output disposal suggestion is no longer based on simple alarm levels, but on careful and quantitative assessment of future risks, realizing the leap from alarm to intelligent decision-making, ensuring that each response is the most stable and rational choice under current information.

[0062] In summary, by deeply mining the time dimension information of the data, the accuracy of fault diagnosis is improved, and the false positives and false negatives caused by environmental interference and other factors are effectively reduced. More importantly, it improves the analysis capability of the system from simple state monitoring to the prediction of fault development trend and the cognitive level of probability modeling of multiple possibilities. The output is no longer a single and rigid alarm signal, but a comprehensive intelligent research and judgment result containing probability diagnosis, risk assessment and optimized disposal suggestions. In this way, it provides a strong technical support for the safe operation and intelligent operation and maintenance of the power system. By realizing more accurate and forward-looking fault diagnosis, it can effectively prevent the occurrence of malignant accidents and ensure the safety and stability of the power grid. At the same time, by providing the optimal disposal suggestion of risk, it can help the operation and maintenance department to allocate human and material resources more efficiently and accurately, avoid unnecessary power outages and equipment maintenance, thereby prolonging the life cycle of equipment assets, reducing the total cost of operation and maintenance, and promoting the power grid to a higher level of intelligent and automated level.

[0063] Specifically, in step S600, a quick response control instruction is generated based on the research and judgment result. It should be understood that the research and judgment result generated by intelligent research and judgment is itself a decision-making information, which cannot directly affect the physical world or operation and maintenance process, and must be converted into an executable instruction to complete the entire response loop. Therefore, in the technical solution of the present application, a quick response control instruction is further generated based on the research and judgment result, so as to convert the abstract and intelligent analysis conclusion into specific actions that can be directly understood and executed by automated equipment or operation and maintenance personnel. In this way, it can ensure that the optimal disposal strategy derived by intelligent research and judgment can be quickly and accurately implemented, thereby truly realizing the full-process automation and efficiency from fault perception to disposal completion.

[0064] That is, specifically, the instruction generation module of the cloud platform receives and analyzes the research and judgment result output by the intelligent research and judgment module. Secondly, according to the disposal suggestion type, severity level and target device information contained in the research and judgment result, the module matches in a pre-set instruction mapping library. If the research and judgment result points to automatic disposal, a standardized remote control instruction conforming to the communication protocol of the target device is generated; if it points to manual disposal, a structured manual alarm notification containing complete event details and disposal suggestions is generated. Finally, the generated control instruction or alarm notification is distributed to the corresponding execution channel, such as being issued to the control unit of the intelligent circuit breaker, or being pushed to the mobile terminal of the operation and maintenance personnel.

[0065] More specifically, in one specific example of the present application, the instruction generation module receives a diagnosis result for the device as a suggestion to immediately send an inspector for inspection. The module determines that this is a manual handling task, and then generates an alarm notification package. The notification package contains the following information: the ID of the target device, the event type is a light gas alarm, the severity level is medium, the geographic coordinates, and the handling suggestion is "immediately send an inspector to the scene for verification, and focus on the transformer body and oil level". Subsequently, the alarm notification package is distributed to the SMS gateway and the mobile application push server, and an alarm SMS and App push are sent to the mobile phone of the preset operation team leader, to ensure that the operation personnel can obtain clear and complete handling instructions at the first time.

[0066] In summary, the gas relay fast response method based on the Internet of Things according to the embodiments of the present application is illustrated, which ensures the accuracy and reliability of event data through front-end signal digitization and state machine verification, laying a foundation for analysis. The cloud discards static threshold judgment and introduces dynamic time series analysis, constructs an enhanced feature vector that can reflect the failure development trend by calculating the rate of change and acceleration of environmental parameters. Further, the Bayesian inference is used to generate a failure probability distribution to quantify uncertainty, and the decision optimization is made based on the principle of minimum expected risk, to ensure the lowest risk of response instructions. The scheme improves the analysis dimension from static snapshot to dynamic process, and upgrades the decision logic from deterministic rule to probabilistic reasoning, solves the misjudgment problem caused by shallow analysis, and improves the accuracy and reliability of response.

[0067] As described above, the data collaborative management platform according to the embodiments of the present application can be implemented in various wireless terminals, such as a server with an electric vehicle transmission downshift control algorithm, etc. In one possible implementation, the data collaborative management platform according to the embodiments of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the data collaborative management platform can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the data collaborative management platform can also be one of the many hardware modules of the wireless terminal.

[0068] The above has described the embodiments of the present disclosure, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles, practical applications, or improvements to the technology in the market of the embodiments, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein.

Claims

1. A gas relay fast response method based on Internet of Things, characterized in that, The method comprises: collecting original electrical signals in response to detecting an internal physical barrier of a transformer; digitizing the original electrical signals and encapsulating initial data to obtain initial data packets; based on a local environment data stream, performing local data enhancement and preprocessing on the initial data packets to obtain preprocessed data loads; encrypting the preprocessed data loads to obtain encrypted data streams, and transmitting the encrypted data streams to a cloud platform; the cloud platform parses the encrypted data streams to obtain the preprocessed data loads, and performs intelligent research and judgment on the preprocessed data loads to obtain research and judgment results; based on the research and judgment results, generating a quick response control instruction.

2. The IoT-based gas relay fast response method according to claim 1, wherein, digitizing the original electrical signals and encapsulating initial data to obtain initial data packets, comprising: sampling the original electrical signals at a preset sampling rate to obtain original electrical level sequences; performing software debouncing on the original electrical level sequences to obtain debounced signal levels; a state machine mode continuously monitors the debounced signal levels, wherein when it is detected that the debounced signal levels transition from one stable state to another stable state, the state machine mode determines that a verified state change has occurred and triggers an effective state change trigger signal; matching the verified state change with a preset event mapping table to obtain a digital state value; in response to the effective state change trigger signal, collecting a timestamp and a device ID; structurally encapsulating the timestamp, the device ID, and the digital state value to obtain the initial data packets. 3.The IoT-based gas relay fast response method according to claim 1, wherein, based on a local environment data stream, performing local data enhancement and preprocessing on the initial data packets to obtain preprocessed data loads, comprising: obtaining a local environment data stream, which includes environmental temperature, humidity, and GPS coordinates of the device; recombining the local environment data stream with the initial data packets to obtain the preprocessed data loads, which include identity information, event data, and environmental context information.

4. The IoT-based gas relay fast response method according to claim 3, wherein, The identity information is the device ID, the event data includes the timestamp and the digital state value, and the environmental context information includes the environmental temperature, humidity, and GPS coordinates of the device.

5. The IoT-based gas relay fast response method according to claim 1, wherein, encrypting the preprocessed data loads to obtain encrypted data streams, and transmitting the encrypted data streams to a cloud platform, comprising: serializing the preprocessed data loads to obtain serialized data; encrypting the serialized data to obtain the encrypted data streams.

6. The IoT-based gas relay fast response method according to claim 1, wherein, the cloud platform parses the encrypted data streams to obtain the preprocessed data loads, and performs intelligent research and judgment on the preprocessed data loads to obtain research and judgment results, comprising: extracting environmental context information from the preprocessed data loads, and calculating a temperature change rate, a humidity change rate, a temperature change acceleration, and a humidity change acceleration based on the environmental context information; extracting the digital state value from the preprocessed data loads; concatenating the digital state value, the environmental temperature, the humidity, the temperature change rate, the humidity change rate, the temperature change acceleration, and the humidity change acceleration to obtain an enhanced feature vector; generating a fault probability distribution based on Bayesian inference on the enhanced feature vector to obtain a fault probability distribution; Based on the cost matrix, decision optimization of the fault probability distribution based on expected risk minimization is performed to obtain the judgment result.

7. The IoT-based gas relay fast response method according to claim 6, wherein, The enhanced feature vector is subjected to Bayesian inference-based fault probability distribution generation to obtain a fault probability distribution, including: the enhanced feature vector is subjected to Bayesian inference-based fault probability distribution generation to obtain the fault probability distribution according to the following formula: where V is the enhanced feature vector, P(V|F k ) is the likelihood, P(F k ) is the prior probability of the kth fault.

8. The IoT-based gas relay fast response method according to claim 6, wherein, Based on the cost matrix, decision optimization of the fault probability distribution based on expected risk minimization is performed to obtain the judgment result, including: based on the cost matrix, decision optimization of the fault probability distribution based on expected risk minimization is performed to obtain the judgment result according to the following formula: where A j is the cost matrix, P(F k |V) is the failure probability distribution, F k is the kth failure, A opt is the assessment result.

Images