Cooperative control method and system based on power distribution internet of things low-voltage intelligent switch

Abstract

The invention discloses a cooperative control method and system based on a power distribution Internet of Things low-voltage intelligent switch, and relates to the technical field of intelligent power distribution. The multi-dimensional data is acquired through the sensor, the low-dimensional state representation is generated, local processing is carried out by utilizing the pre-trained neural network model, quick response is realized by combining a low-delay communication technology, the response speed and the control precision of the intelligent power distribution system are remarkably improved, and the system reliability is improved. Power line communication, a time-sensitive network protocol, a symmetric encryption algorithm and a distributed consistency algorithm are adopted, low time delay, high reliability and communication safety of data transmission are ensured, the overall collaboration and stability of the system are enhanced, millisecond-level fault isolation is achieved through a fast switch control algorithm, and the reliability of the system is improved. Fault classification and positioning are performed in combination with a support vector machine algorithm, and a protection strategy is dynamically optimized by using an incremental learning algorithm, so that the adaptive ability and fault response ability of the system are improved, and the stability and continuity of power supply are ensured.

Description

Technical Field

[0001] This invention relates to the field of intelligent power distribution technology, specifically a collaborative control method and system based on low-voltage intelligent switches for power distribution IoT. Background Technology

[0002] Intelligent power distribution systems are crucial in modern energy management, as their stability and efficiency directly impact the reliability and economic benefits of power supply. With the increasing prevalence of distributed energy resources and electric vehicles, low-voltage power distribution networks face complex and ever-changing operating environments, necessitating precise control and rapid response. Currently, most solutions rely on centralized control or static rules, which struggle to cope with dynamic load changes and real-time demands in extreme environments. For example, traditional systems typically have protection action times on the order of seconds when facing sudden faults, leading to an expansion of the fault's scope. Furthermore, existing methods lack efficient encryption mechanisms and precise timing control in inter-device communication, easily causing command conflicts or response delays. Summary of the Invention

[0003] The purpose of this invention is to provide a collaborative control method and system based on low-voltage intelligent switches of power distribution IoT. Through multi-dimensional data acquisition and processing, low-latency communication and encryption mechanisms, and rapid fault isolation and dynamic optimization strategies, the response speed, control accuracy, collaborative capability and adaptability of the intelligent power distribution system are significantly improved, ensuring the reliability and efficiency of power supply.

[0004] The objective of this invention can be achieved through the following technical solutions: This application provides a collaborative control method based on low-voltage intelligent switches in power distribution IoT, including the following steps: S1. Obtain multi-dimensional data of real-time environmental parameters from the distributed devices deployed in the low-voltage smart switch, process the multi-dimensional data through a data compression algorithm to reduce the data dimension of the multi-dimensional data, and then input it into a pre-constructed dynamic digital model. Use the dynamic digital model to classify the state of the low-voltage smart switch and obtain the state classification result. S2. The distributed device uses a pre-trained neural network model to perform local processing on high-frequency data, generate local control strategies, and obtain autonomous decision-making instructions for the distributed device; wherein, the high-frequency data refers to data whose frequency of occurrence is greater than a preset frequency threshold; S3. Obtain key features from the local control strategy, transmit them to the cloud via power line communication, process the transmitted data using time-sensitive networking protocol, and obtain cloud-optimized input data. S4. If the input data meets the preset global optimization conditions, a distributed consensus algorithm is adopted, combined with real-time environmental parameters, to generate global collaborative optimization logic and obtain unified control instructions. S5. The unified control command is encrypted using a symmetric encryption algorithm to obtain an encrypted synchronization command set, which is then sent to the distributed device via power line communication. S6. In response to the distributed device receiving the synchronization instruction set, a timestamp verification mechanism is used to determine whether the timing of the synchronization instruction set meets the microsecond-level synchronization requirement, and a synchronization execution sequence is obtained. S7. Based on the synchronous execution sequence, the distributed device adopts a fast switching control algorithm to perform millisecond-level fault isolation and obtain protection actions under load change scenarios.

[0005] Furthermore, step S1 specifically includes: Real-time environmental parameters are acquired from the distributed device by sensors. The real-time environmental parameters include a multi-dimensional data set of voltage data, current data, and temperature data. The multi-dimensional data set is then dimensionality-reduced by principal component analysis algorithm to generate a low-dimensional state representation. The missing features of the low-dimensional state representation are supplemented by interpolation to obtain a complete low-dimensional state representation. Then, a long short-term memory network model is used to construct a dynamic digital model to generate time series state predictions. The prediction results of the time series state predictions are smoothed by a sliding window algorithm to obtain a stable state prediction sequence. Clustering algorithms are used to classify the operating states of the low-voltage smart switch, determine the operating state categories, generate dynamic digital model input data for the low-voltage smart switch, and store it in a distributed database.

[0006] Furthermore, step S2 specifically includes: A pre-trained neural network model is used to acquire high-frequency data from distributed devices. Local processing is performed on the high-frequency data to obtain processed feature data. The feature data is then analyzed using the neural network model to generate a local control strategy and determine the device control parameters. If the control parameters meet the preset threshold, an autonomous decision-making instruction is generated, a device control command is output, the status data of the device after executing the control command is obtained, and the status features are extracted using data processing methods to obtain the device operation trend. The operating trend is predicted by a neural network model, an updated control strategy is generated, the optimization direction of the strategy is determined, and if the deviation between the optimization direction and the current strategy exceeds a preset threshold, the decision command is adjusted and the device autonomous decision command is output.

[0007] Furthermore, step S3 specifically includes: Key features are obtained from the local control strategy, a key feature set is determined by a feature extraction algorithm, the key feature set is transmitted to the cloud via power line communication, and a compressed data packet is generated by a data compression algorithm. If the compressed data packet meets the requirements of the Time-Sensitive Networking Protocol (TSN), the compressed data packet is transmitted through the TSN to obtain the cloud-received data. The data decompression algorithm is then used to restore the key feature set and generate decompressed feature data. The integrity of the decompressed feature data is determined by a machine learning classification algorithm to obtain a complete feature dataset. Then, cloud optimization processing is performed to generate cloud-optimized input data, which is used to update the local control strategy and generate updated control parameters.

[0008] Furthermore, step S4 specifically includes: If the input data meets the preset conditions, the input data is processed by a distributed consensus algorithm to obtain a consistency verification result. Then, real-time environmental parameters are obtained, and the parameters and verification results are fused to generate an environment adaptation dataset. If the environment adaptation dataset meets the global optimization criteria, then the convex optimization algorithm is used to process the environment adaptation dataset to determine the global optimization solution, generate collaborative optimization logic, and obtain the logical reasoning result. If the logical reasoning result meets the unified instruction generation condition, then the result is mapped through the rule engine to generate a unified control instruction. The unified control instructions are distributed to each node using a message queue mechanism to determine the instruction execution sequence. If the feedback data of the instruction execution sequence is consistent with the expectation, the integrity of the sequence is verified by a data verification algorithm to obtain the final execution confirmation.

[0009] Furthermore, step S5 specifically includes: A unified control command set is obtained through the power line communication channel, encrypted using a preset symmetric encryption algorithm, and an encrypted synchronization command set is generated. This command set is then distributed to distributed devices through the power line communication network to ensure the integrity of the command set. A message confirmation mechanism is used to determine whether the instruction has been received correctly. If the instruction has been received correctly, the decryption module uses a preset symmetric encryption algorithm key to obtain the decrypted control instruction. Distributed devices perform synchronization operations, generate execution status data, acquire the execution status data through the power line communication network, use a data verification algorithm to determine the consistency of the execution results, and generate a synchronization instruction execution confirmation signal to confirm that the distributed device operation is complete.

[0010] Furthermore, step S6 specifically includes: The distributed device receives an encrypted instruction set, parses the instruction content using a symmetric decryption algorithm to obtain a plaintext instruction set, and uses a timestamp extraction algorithm to obtain the timestamps of each instruction from the plaintext instruction set to generate a timestamp sequence. If the difference between timestamps in the timestamp sequence is less than a preset microsecond threshold, the instructions are sorted by a time-series verification algorithm, and the instructions are distributed to each distributed device using a consistent hashing algorithm to generate a device instruction mapping. By using the device instruction mapping, it is detected whether the timestamps of each device instruction meet the synchronization accuracy requirements, and a synchronization verification result is obtained. If the synchronization verification result meets the preset synchronization accuracy threshold, a synchronization execution sequence is generated according to the device instruction mapping, and the instructions are executed in parallel by distributed devices to obtain the data consistency status.

[0011] Furthermore, step S7 specifically includes: The system acquires status data of distributed devices by using sensors to collect voltage, current and switch status information. Real-time device operating parameters are obtained through data preprocessing. If the real-time device operating parameters exceed the preset threshold, a fault detection mechanism is triggered. The support vector machine algorithm is used to classify parameter anomalies, determine the fault type and location, and generate fast switching control commands. A pre-established sequence synchronization model is used to calculate the timing of the switching actions of each device to obtain a synchronization control scheme. The switching states of distributed devices are adjusted by the synchronous control scheme, control commands are sent to each device using a real-time communication protocol, millisecond-level fault isolation is completed, load change data is extracted from the results of millisecond-level fault isolation, and time series analysis is used to detect load change characteristics and generate a description of the change scenario. The protection action strategy is updated based on the description of the mutation scenario. Conditional judgment logic is used. If the mutation magnitude exceeds a preset threshold, the protection parameters are adjusted to obtain the protection action under the load mutation scenario. This is used to dynamically adjust the operating status of distributed devices. A status feedback mechanism is used to obtain system stability indicators and determine whether the system has returned to normal operation.

[0012] Furthermore, in step S7, after obtaining the protection actions under the load change scenario, the method further includes: obtaining feedback data from the protection actions, updating the state representation through a dynamic digital model, optimizing local autonomous decision-making using an incremental learning algorithm, and obtaining a new control strategy.

[0013] This application provides a collaborative control system based on a low-voltage intelligent switch for power distribution IoT, used to implement a collaborative control method based on a low-voltage intelligent switch for power distribution IoT, including: The data acquisition and preprocessing module obtains real-time environmental parameters from distributed devices deployed in low-voltage smart switches, performs dimensionality reduction processing through principal component analysis algorithm to generate low-dimensional state representations, and uses interpolation to supplement missing features to determine the input data for constructing a dynamic digital model. The local control strategy generation module uses a pre-trained neural network model to process the high-frequency data collected by the distributed devices locally, generate local control strategies, dynamically adjust the control strategies according to the device's operating trends, and generate autonomous decision-making instructions for the devices. The data transmission and communication module extracts key features from the local control strategy and transmits them to the cloud via power line communication, while using time-sensitive network protocol to process the transmitted data. The cloud optimization module performs integrity checks and optimization on data transmitted from distributed devices. It uses a distributed consensus algorithm combined with real-time environmental parameters to generate global collaborative optimization logic and generate unified control commands. The instruction issuance and execution module issues unified control instructions generated in the cloud to distributed devices via power line communication, and encrypts the unified control instructions using a symmetric encryption algorithm. The distributed devices then receive the instructions, decrypt them, verify them, and execute them synchronously. The synchronous execution and fault isolation module uses a fast switching control algorithm to perform millisecond-level fault isolation on the synchronous execution sequence, and generates protection actions in the event of sudden load changes. It also uses a timestamp verification mechanism to ensure microsecond-level synchronous execution of instructions. The dynamic feedback and optimization module obtains feedback data from protection actions, updates the state representation through a dynamic digital model, and uses an incremental learning algorithm to optimize local autonomous decisions and generate new control strategies.

[0014] The beneficial effects of this invention are as follows: By collecting multidimensional data from sensors and generating low-dimensional state representations, local processing is performed using a pre-trained neural network model to generate local control strategies. Combined with low-latency communication technology, rapid response is achieved, solving the problems of slow response speed and low control accuracy of traditional systems under dynamic load changes and extreme environments. It can quickly adapt to complex operating environments, significantly improve the response speed and control accuracy of intelligent power distribution systems, and ensure the reliability and efficiency of power supply. By employing power line communication, time-sensitive networking protocol, symmetric encryption algorithm, and distributed consensus algorithm, low latency, high reliability, and communication security of data transmission are ensured. This solves the problem of existing methods lacking efficient encryption mechanisms and precise timing control when devices communicate collaboratively. Through low-latency communication and encryption mechanisms, command conflicts and response delays are avoided, enhancing the overall collaboration and communication security of the system. This ensures that distributed devices can cooperate efficiently, improving the system's operating efficiency and stability. A fast switching control algorithm is adopted to achieve millisecond-level fault isolation. The support vector machine algorithm is combined with the fault classification and location. The protection strategy is dynamically optimized through incremental learning algorithm. This solves the problems of long protection action time and widening fault range in traditional systems during sudden faults. It can quickly isolate faults and reduce the impact of faults on the system. At the same time, the dynamic optimization of the protection strategy improves the system's adaptability and ensures the stability and continuity of power supply. Attached Figure Description

[0015] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0016] Figure 1 A flowchart illustrating a collaborative control method based on a low-voltage intelligent switch of power distribution IoT provided in Embodiment 1 of this application; Figure 2 This is a flowchart illustrating step S5 in a collaborative control method based on a low-voltage smart switch of power distribution IoT provided in Embodiment 1 of this application. Figure 3 This is a flowchart illustrating step S6 in a collaborative control method based on a low-voltage smart switch of power distribution IoT provided in Embodiment 1 of this application. Figure 4 This is a schematic diagram of the structure of a collaborative control system based on a low-voltage intelligent switch of power distribution IoT, provided in Embodiment 2 of this application. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0018] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0019] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0020] Example 1 Please see Figures 1-3 This embodiment provides a collaborative control method based on low-voltage intelligent switches in power distribution IoT, including the following steps: S1. Obtain multi-dimensional data of real-time environmental parameters from the distributed devices deployed in the low-voltage smart switch, process the multi-dimensional data through a data compression algorithm to reduce the data dimension of the multi-dimensional data, and then input it into a pre-constructed dynamic digital model. Use the dynamic digital model to classify the state of the low-voltage smart switch and obtain the state classification result.

[0021] Furthermore, step S1 specifically includes: Real-time environmental parameters are acquired from the distributed device by sensors. The real-time environmental parameters include a multi-dimensional data set of voltage data, current data, and temperature data. The multi-dimensional data set is then dimensionality-reduced by principal component analysis algorithm to generate a low-dimensional state representation. The missing features of the low-dimensional state representation are supplemented by interpolation to obtain a complete low-dimensional state representation. Then, a long short-term memory network model is used to construct a dynamic digital model to generate time series state predictions. The prediction results of the time series state predictions are smoothed by a sliding window algorithm to obtain a stable state prediction sequence. Based on a stable state prediction sequence, a clustering algorithm is used to classify the operating states of the low-voltage smart switch, determine the operating state category, generate dynamic digital model input data for the low-voltage smart switch, and store it in a distributed database.

[0022] Specifically, by collecting multidimensional data from low-voltage smart switches and performing dimensionality reduction, feature supplementation, construction of dynamic digital models, smoothing of prediction results, and classification of operating states, the problems of complex data processing, unstable prediction, and inaccurate state classification in traditional systems are solved. This significantly improves data processing efficiency and accuracy, provides reliable support for real-time monitoring and precise control of smart power distribution systems, and enhances the stability and reliability of the system.

[0023] S2. The distributed device uses a pre-trained neural network model to perform local processing on high-frequency data, generate local control strategies, and obtain autonomous decision-making instructions for the distributed device; wherein, the high-frequency data refers to data whose frequency of occurrence is greater than a preset frequency threshold.

[0024] Furthermore, autonomous decision-making instructions are obtained from the equipment, specifically including: A pre-trained neural network model is used to acquire high-frequency data from distributed devices. Local processing is performed on the high-frequency data to obtain processed feature data. The feature data is then analyzed using the neural network model to generate a local control strategy and determine the device control parameters. If the control parameters meet the preset threshold, an autonomous decision-making instruction is generated, a device control command is output, the status data of the device after executing the control command is obtained, and the status features are extracted using data processing methods to obtain the device operation trend. The operating trend is predicted by a neural network model, an updated control strategy is generated, the optimization direction of the strategy is determined, and if the deviation between the optimization direction and the current strategy exceeds a preset threshold, the decision command is adjusted and the device autonomous decision command is output.

[0025] The neural network model is trained using a large amount of historical data. Specifically, multi-dimensional data related to low-voltage smart switches, including voltage, current, and temperature, are first collected as a training dataset. This training dataset is then input into the neural network, and the weights and bias parameters are adjusted using the backpropagation algorithm. This allows the network to learn the features and patterns in the data. After multiple iterations of training, the neural network gradually optimizes and eventually becomes capable of accurately classifying, predicting, or making decisions based on the input data. The pre-trained neural network model can be directly applied to the local processing of distributed devices to quickly generate local control strategies and autonomous decision-making instructions.

[0026] Specifically, by using a pre-trained neural network model to locally process high-frequency data from distributed devices, local control strategies are generated and device control parameters are determined, thereby generating autonomous decision-making instructions. This solves the problems of slow response and inaccurate decision-making in traditional systems when processing high-frequency data, significantly improving the autonomous decision-making capability and response speed of the devices, and enhancing the adaptability and stability of distributed devices in complex operating environments.

[0027] More specifically, transfer learning techniques are introduced to optimize pre-trained neural network models, enabling them to better adapt to different operating environments and fault scenarios. By pre-training on multiple similar tasks, the model can learn broader knowledge, improving its performance on specific tasks. This involves collecting multiple task datasets similar to those used in smart power distribution systems, including fault detection and load prediction under different environments. A general neural network architecture (such as convolutional neural networks or recurrent neural networks) is then selected and pre-trained on these similar task datasets, allowing the model to learn common features and patterns across different tasks. After pre-training, the model is transferred to specific smart power distribution system tasks, and the last few layers are fine-tuned to adapt to the specific task requirements. During fine-tuning, a small amount of actual smart power distribution system data is used to update the model parameters, further improving the model's performance on specific tasks. In this way, the model not only utilizes the extensive knowledge learned from multiple similar tasks but also quickly adapts to new operating environments and fault scenarios, significantly improving its performance and adaptability on specific tasks.

[0028] S3. Obtain key features from the local control strategy, transmit them to the cloud via power line communication, process the transmitted data using time-sensitive network protocol, and obtain cloud-optimized input data.

[0029] Furthermore, step S3 specifically includes: Key features are obtained from the local control strategy, a key feature set is determined by a feature extraction algorithm, the key feature set is transmitted to the cloud via power line communication, and a compressed data packet is generated by a data compression algorithm. If the compressed data packet meets the requirements of the Time-Sensitive Networking Protocol (TSN), the compressed data packet is transmitted through the TSN to obtain the cloud-received data. The data decompression algorithm is then used to restore the key feature set and generate decompressed feature data. Based on the decompression feature data, a machine learning classification algorithm is used to determine the completeness of the decompression feature data to obtain a complete feature dataset. Then, cloud optimization processing is performed to generate cloud-optimized input data, which is used to update the local control strategy and generate updated control parameters.

[0030] Specifically, key features are extracted from local control strategies and transmitted via power line communication. Combined with data compression, time-sensitive network protocol transmission, decompression and restoration, and machine learning classification to determine data integrity, this solves problems such as latency and insufficient integrity in data transmission between devices and the cloud. Ultimately, it generates cloud-optimized input data, effectively supporting the updating of local control strategies and improving the accuracy and adaptability of system control.

[0031] S4. If the input data meets the preset global optimization conditions, a distributed consensus algorithm is adopted, combined with real-time environmental parameters, to generate global collaborative optimization logic and obtain unified control instructions.

[0032] Furthermore, step S4 specifically includes: If the input data meets the preset conditions, the input data is processed by a distributed consensus algorithm to obtain a consistency verification result. Then, real-time environmental parameters are obtained, and the parameters and verification results are fused to generate an environment adaptation dataset. If the environment adaptation dataset meets the global optimization criteria, then the convex optimization algorithm is used to process the environment adaptation dataset to determine the global optimization solution, generate collaborative optimization logic, and obtain the logical reasoning result. If the logical reasoning result meets the unified instruction generation condition, then the result is mapped through the rule engine to generate a unified control instruction. According to the unified control instructions, the unified control instructions are distributed to each node using a message queue mechanism to determine the instruction execution sequence. If the feedback data of the instruction execution sequence is consistent with the expectation, the integrity of the sequence is verified by a data verification algorithm to obtain the final execution confirmation.

[0033] Specifically, by employing distributed consensus algorithms and convex optimization algorithms to generate global collaborative optimization logic and unified control instructions, the problems of low global optimization efficiency and weak collaborative capabilities in traditional systems are solved. This significantly improves the global optimization efficiency of the system and the collaborative capabilities between distributed devices, ensuring the accuracy and integrity of instruction execution. As a result, the overall performance and stability of the system are enhanced, enabling the system to cope more efficiently with complex and ever-changing operating environments.

[0034] S5. The unified control command is encrypted using a symmetric encryption algorithm to obtain an encrypted synchronization command set, which is then sent to the distributed device via power line communication.

[0035] Furthermore, step S5 specifically includes: S51. Obtain a unified control command set through the power line communication channel, encrypt it using a preset symmetric encryption algorithm, generate an encrypted synchronization command set, and then distribute it to distributed devices through the power line communication network to ensure the integrity of the command set. S52. Based on the receiving status of the distributed device, a message confirmation mechanism is used to determine whether the instruction has been received correctly. If the instruction has been received correctly, the decryption module uses a preset symmetric encryption algorithm key to obtain the decrypted control instruction. S53. According to the decrypted control command, the distributed device performs a synchronization operation, generates execution status data, obtains the execution status data through the power line communication network, uses a data verification algorithm to determine the consistency of the execution result, and generates a synchronization command execution confirmation signal to confirm that the distributed device operation is complete.

[0036] The unified control command set is encrypted using a pre-defined symmetric encryption algorithm (such as AES or DES) to generate an encrypted synchronization command set, ensuring the security of commands during transmission. The specific operations include selecting a suitable symmetric encryption algorithm, generating and securely distributing a key, using this key to encrypt the command set, and then distributing the encrypted command set to distributed devices via a power line communication network. Upon receiving the command, the devices verify its integrity and decrypt it using the same key to obtain the original control commands, effectively preventing data tampering or theft and ensuring the system's communication security and command execution reliability.

[0037] Specifically, a symmetric encryption algorithm is used to encrypt unified control commands and send them to distributed devices via power line communication to ensure communication security. After receiving the commands, the devices verify the integrity of the commands through a message confirmation mechanism and decrypt and execute them. Finally, a data verification algorithm is used to confirm the consistency of the execution results. This process effectively solves the problems of communication insecurity and inconsistent execution results, and significantly improves the communication security and command execution reliability of the system.

[0038] S6. In response to the distributed device receiving the synchronization instruction set, a timestamp verification mechanism is used to determine whether the timing of the synchronization instruction set meets the microsecond-level synchronization requirement, and a synchronization execution sequence is obtained.

[0039] Furthermore, step S6 specifically includes: S61. Receive the encrypted instruction set through the distributed device, parse the instruction content using a symmetric decryption algorithm to obtain the plaintext instruction set, and use a timestamp extraction algorithm to obtain the timestamp of each instruction from the plaintext instruction set to generate a timestamp sequence. S62. If the difference between timestamps in the timestamp sequence is less than a preset microsecond threshold, then the instructions are sorted by the time sequence verification algorithm, and the instructions are distributed to each distributed device by the consistent hashing algorithm to generate a device instruction mapping. S63. Through the device instruction mapping, detect whether the timestamp of each device instruction meets the synchronization accuracy requirements and obtain the synchronization verification result. If the synchronization verification result meets the preset synchronization accuracy threshold, generate a synchronization execution sequence according to the device instruction mapping, and execute the instructions in parallel through distributed devices to obtain the data consistency status.

[0040] Specifically, a timestamp verification mechanism and a consistent hashing algorithm are adopted to ensure that the instructions received by the distributed devices meet the microsecond-level synchronization requirements and generate a synchronized execution sequence. This solves the problems of inaccurate instruction timing and insufficient synchronization accuracy in traditional systems, achieves microsecond-level synchronization accuracy, and significantly improves the collaborative capabilities between distributed devices and the overall operating efficiency of the system.

[0041] S7. Based on the synchronous execution sequence, the distributed device adopts a fast switching control algorithm to perform millisecond-level fault isolation and obtain protection actions under load change scenarios.

[0042] Furthermore, step S7 specifically includes: The system acquires status data of distributed devices by using sensors to collect voltage, current and switch status information. Real-time device operating parameters are obtained through data preprocessing. If the real-time device operating parameters exceed the preset threshold, a fault detection mechanism is triggered. The support vector machine algorithm is used to classify parameter anomalies, determine the fault type and location, and generate fast switching control commands. A pre-established sequence synchronization model is used to calculate the timing of the switching actions of each device to obtain a synchronization control scheme. The sequence synchronization model is established by analyzing and modeling the switching timing of distributed devices. First, switching action data of distributed devices under normal operation and fault conditions are collected, including timestamps of switching actions and device status. Based on this data, mathematical modeling methods (such as time series analysis and system dynamics modeling) are used to construct a model to describe the timing relationships and dependencies between switching actions of devices. Through simulation and optimization, model parameters are adjusted to ensure that the model accurately reflects the synchronization requirements of devices in different scenarios. Finally, after verification and testing, a sequence synchronization model that can accurately calculate the switching timing of each device is obtained, which is used to guide actual synchronization control operations.

[0043] The switching states of distributed devices are adjusted by the synchronous control scheme, control commands are sent to each device using a real-time communication protocol, millisecond-level fault isolation is completed, load change data is extracted from the results of millisecond-level fault isolation, and time series analysis is used to detect load change characteristics and generate a description of the change scenario. After completing millisecond-level fault isolation, load change data is extracted from the isolation results. This data reflects the dynamic changes in load before and after the fault. Time series analysis is used to process this load change data. By analyzing the trends, periodicity, and abnormal fluctuations in the data, specific characteristics of load mutations are detected, such as the magnitude, duration, and frequency of the mutations. Based on these detected characteristics, a detailed description of the mutation scenario is generated, providing a basis for subsequent protection action strategy adjustments and system optimization, thereby achieving accurate identification and description of load mutation scenarios.

[0044] The protection action strategy is updated based on the description of the mutation scenario. Conditional judgment logic is used. If the mutation magnitude exceeds a preset threshold, the protection parameters are adjusted to obtain the protection action under the load mutation scenario. This is used to dynamically adjust the operating status of distributed devices. A status feedback mechanism is used to obtain system stability indicators and determine whether the system has returned to normal operation.

[0045] Specifically, by using a fast switching control algorithm and millisecond-level fault isolation technology, combined with the support vector machine (SVM) algorithm for fault classification and location, and time series analysis to detect load change characteristics, the system solves the problems of long fault isolation time and untimely protection action in traditional systems under load change scenarios. This significantly improves the system's fault response speed and adaptability, ensuring the stability and continuity of power supply.

[0046] More specifically, a deep reinforcement learning (DRL) algorithm is introduced to further optimize the accuracy and speed of fault isolation. The specific implementation process is as follows: A DRL-based intelligent agent is constructed. This agent can receive real-time status data from distributed devices, including information such as voltage, current, and switch status. Through interaction with the environment, the agent learns the optimal fault isolation strategy. During the training phase, the agent takes actions based on the current state (such as adjusting the switch status) and updates its policy network based on reward signals from the environment (such as the success or failure of fault isolation, isolation time, etc.). Through continuous trial and error and learning, the agent can gradually master the method of quickly and accurately isolating faults in different fault scenarios. In actual operation, the DRL agent dynamically adjusts the control strategy based on real-time data to ensure rapid fault response in complex and ever-changing operating environments, achieving millisecond-level fault isolation and significantly improving the stability and reliability of the system.

[0047] Furthermore, in step S7, after obtaining the protection actions under the load change scenario, the method further includes: obtaining feedback data from the protection actions, updating the state representation through a dynamic digital model, optimizing local autonomous decision-making using an incremental learning algorithm, and obtaining a new control strategy.

[0048] Furthermore, new control strategies are derived, specifically including: Feedback data is obtained from the protection action. Real-time feedback data is collected by sensors and stored in a preset database to obtain a structured feedback dataset. The structured feedback dataset is processed by a dynamic digital model to update the state representation and generate the current state vector. If the deviation between the current state vector and the preset state threshold exceeds a predetermined range, the incremental learning algorithm is used to optimize the autonomous decision-making model, obtain the optimized decision parameters, generate a new control strategy, and output it to the execution module. New protection action data is obtained from the execution module, real-time feedback data is updated and stored in the preset database to obtain the updated feedback dataset. The accuracy of the dynamic digital model is then verified, the state representation is adjusted, and the verified state vector is generated for iterative optimization of the control strategy, and the final control strategy is output.

[0049] Specifically, feedback data is obtained from protection actions, collected by sensors, and stored in a database to form a structured dataset. The state representation is updated through a dynamic digital model, and the autonomous decision-making model is optimized by combining incremental learning algorithms. New control strategies are generated and iteratively verified and adjusted. This solves the problem that control strategies are difficult to adapt to dynamic changes in the system, realizes continuous optimization of control strategies, and improves the flexibility and reliability of the system in dealing with scenarios such as sudden load changes.

[0050] Example 2 Please see Figure 4This embodiment provides a collaborative control system based on a low-voltage intelligent switch of power distribution IoT, used to implement a collaborative control method based on a low-voltage intelligent switch of power distribution IoT, including: The data acquisition and preprocessing module obtains real-time environmental parameters, including multi-dimensional data such as voltage, current, and temperature, from the distributed devices deployed in the low-voltage intelligent switch. It performs dimensionality reduction on these data using principal component analysis algorithm to generate low-dimensional state representations and uses interpolation to supplement missing features, thereby determining the input data for constructing a dynamic digital model. This provides an accurate and concise data foundation for subsequent equipment state analysis and control strategy generation. The local control strategy generation module uses a pre-trained neural network model to process the high-frequency data collected by the distributed devices locally, generate local control strategies, dynamically adjust the control strategies according to the device operation trends, and generate autonomous decision-making instructions for the devices. This ensures that the distributed devices can make autonomous decisions and respond quickly, improving the system's flexibility and response speed. The data transmission and communication module extracts key features from the local control strategy and transmits these features to the cloud via power line communication. At the same time, it uses time-sensitive networking protocol to process the transmitted data, efficiently and securely transmitting the key information of the distributed devices to the cloud, providing data support for global optimization. The cloud optimization module performs integrity checks and optimization on data transmitted from distributed devices. It uses a distributed consensus algorithm combined with real-time environmental parameters to generate global collaborative optimization logic and generate unified control commands, thus realizing global-level optimization decisions and ensuring the overall performance and collaboration of the system. The instruction issuance and execution module sends unified control instructions generated in the cloud to distributed devices via power line communication. The unified control instructions are encrypted using a symmetric encryption algorithm. After receiving the instructions, the distributed devices decrypt, verify, and execute them synchronously, ensuring the accurate issuance and execution of control instructions and enabling collaborative operation of distributed devices. The synchronous execution and fault isolation module employs a fast switching control algorithm to perform millisecond-level fault isolation for synchronous execution sequences, and generates protection actions in load change scenarios. It ensures microsecond-level synchronous execution of instructions through a timestamp verification mechanism, which can quickly respond to faults, protect the safe operation of the system, and dynamically adjust the control strategy when the load changes. The dynamic feedback and optimization module acquires feedback data from protection actions, updates the state representation through a dynamic digital model, and uses an incremental learning algorithm to optimize local autonomous decisions, generating new control strategies. This achieves dynamic feedback and continuous optimization of the system, improving its adaptability and stability.

[0051] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A collaborative control method based on low-voltage intelligent switches in power distribution IoT, characterized in that: Includes the following steps: S1. Obtain multi-dimensional data of real-time environmental parameters from the distributed devices deployed in the low-voltage smart switch, process the multi-dimensional data through a data compression algorithm to reduce the data dimension of the multi-dimensional data, and then input it into a pre-constructed dynamic digital model. Use the dynamic digital model to classify the state of the low-voltage smart switch and obtain the state classification result. S2. The distributed device uses a pre-trained neural network model to perform local processing on high-frequency data, generate local control strategies, and obtain autonomous decision-making instructions for the distributed device; wherein, the high-frequency data refers to data whose frequency of occurrence is greater than a preset frequency threshold; S3. Obtain key features from the local control strategy, transmit them to the cloud via power line communication, process the transmitted data using time-sensitive networking protocol, and obtain cloud-optimized input data. S4. If the input data meets the preset global optimization conditions, a distributed consensus algorithm is adopted, combined with real-time environmental parameters, to generate global collaborative optimization logic and obtain unified control instructions. S5. The unified control command is encrypted using a symmetric encryption algorithm to obtain an encrypted synchronization command set, which is then sent to the distributed device via power line communication. S6. In response to the distributed device receiving the synchronization instruction set, a timestamp verification mechanism is used to determine whether the timing of the synchronization instruction set meets the microsecond-level synchronization requirement, and a synchronization execution sequence is obtained. S7. Based on the synchronous execution sequence, the distributed device adopts a fast switching control algorithm to perform millisecond-level fault isolation and obtain protection actions under load change scenarios.

2. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S1 specifically includes: Real-time environmental parameters are acquired from the distributed device by sensors. The real-time environmental parameters include a multi-dimensional data set of voltage data, current data and temperature data. The multi-dimensional data set is then dimensionality-reduced by principal component analysis algorithm to generate a low-dimensional state representation. The missing features of the low-dimensional state representation are supplemented by interpolation to obtain a complete low-dimensional state representation. Then, a long short-term memory network model is used to construct a dynamic digital model to generate time series state predictions. The prediction results of the time series state predictions are smoothed by a sliding window algorithm to obtain a stable state prediction sequence. Clustering algorithms are used to classify the operating states of the low-voltage smart switch, determine the operating state categories, generate dynamic digital model input data for the low-voltage smart switch, and store it in a distributed database.

3. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S2 specifically includes: A pre-trained neural network model is used to acquire high-frequency data from distributed devices. Local processing is performed on the high-frequency data to obtain processed feature data. The neural network model is then used to analyze the feature data to generate a local control strategy and determine the device control parameters. If the control parameters meet the preset threshold, an autonomous decision-making instruction is generated, a device control command is output, the status data of the device after executing the control command is obtained, and the status features are extracted using data processing methods to obtain the device operation trend. The operating trend is predicted by a neural network model, an updated control strategy is generated, and the strategy optimization direction is determined. If the deviation between the strategy optimization direction and the current strategy exceeds a preset threshold, the decision command is adjusted and the device autonomous decision command is output.

4. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S3 specifically includes: Key features are obtained from the local control strategy, a key feature set is determined by a feature extraction algorithm, the key feature set is transmitted to the cloud via power line communication, and a compressed data packet is generated by a data compression algorithm. If the compressed data packet meets the requirements of the Time-Sensitive Networking Protocol (TSN), the compressed data packet is transmitted through the TSN to obtain the cloud-received data. The data decompression algorithm is then used to restore the key feature set and generate decompressed feature data. The integrity of the decompressed feature data is determined by a machine learning classification algorithm to obtain a complete feature dataset. Then, cloud optimization processing is performed to generate cloud-optimized input data, which is used to update the local control strategy and generate updated control parameters.

5. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S4 specifically includes: If the input data meets the preset conditions, the input data is processed by a distributed consensus algorithm to obtain a consistency verification result. Then, real-time environment parameters are obtained, and the real-time environment parameters and the consistency verification result are fused to generate an environment adaptation dataset. If the environment adaptation dataset meets the global optimization criteria, then the convex optimization algorithm is used to process the environment adaptation dataset to determine the global optimization solution, generate collaborative optimization logic, and obtain the logical reasoning result. If the logical reasoning result meets the unified instruction generation condition, then the logical reasoning result is mapped through the rule engine to generate a unified control instruction. The unified control instructions are distributed to each node using a message queue mechanism to determine the instruction execution sequence. If the feedback data of the instruction execution sequence is consistent with the expected data, the integrity of the sequence is verified by a data verification algorithm to obtain the final execution confirmation.

6. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S5 specifically includes: A unified control command set is obtained through the power line communication channel, encrypted using a preset symmetric encryption algorithm, and an encrypted synchronization command set is generated. This encrypted synchronization command set is then distributed to the distributed devices through the power line communication network to ensure the integrity of the synchronization command set. A message confirmation mechanism is used to determine whether the synchronization instruction set has been received correctly. If the instruction has been received correctly, the decryption module uses a preset symmetric encryption algorithm key to obtain the decrypted control instruction. The distributed device performs synchronization operations and generates execution status data according to the received control instructions. The execution status data is transmitted through the power line communication network. A data verification algorithm is used to determine the consistency of the execution results and generate a synchronization instruction execution confirmation signal to confirm that the distributed device operation is complete.

7. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S6 specifically includes: The distributed device receives the synchronization instruction set, uses a symmetric decryption algorithm to parse the contents of the synchronization instruction set to obtain a plaintext instruction set, and uses a timestamp extraction algorithm to obtain the timestamps of each instruction from the plaintext instruction set to generate a timestamp sequence. If the difference between timestamps in the timestamp sequence is less than a preset microsecond threshold, the instructions are sorted by a time-series verification algorithm, and the instructions are distributed to each distributed device using a consistent hashing algorithm to generate a device instruction mapping. By using the device instruction mapping, it is detected whether the timestamps of each device instruction meet the synchronization accuracy requirements, and a synchronization verification result is obtained. If the synchronization verification result meets the preset synchronization accuracy threshold, a synchronization execution sequence is generated according to the device instruction mapping, and the instructions are executed in parallel by distributed devices to obtain the data consistency status.

8. The collaborative control method based on a low-voltage intelligent switch of power distribution IoT according to claim 1, characterized in that: Step S7 specifically includes: The system acquires status data of distributed devices by using sensors to collect voltage, current and switch status information. Real-time device operating parameters are obtained through data preprocessing. If the real-time device operating parameters exceed the preset threshold, a fault detection mechanism is triggered. The fault detection mechanism includes: using a support vector machine algorithm to classify parameter anomalies, determine the fault type and location, and generate fast switching control commands; using a pre-established sequence synchronization model to calculate the timing of the switching actions of each distributed device, and obtain a synchronization control scheme. The switching states of distributed devices are adjusted by the synchronous control scheme, and control commands are sent to each distributed device using a real-time communication protocol to complete millisecond-level fault isolation. Load change data is extracted from the results of millisecond-level fault isolation, and time series analysis is used to detect load change characteristics and generate a description of the change scenario. The protection action strategy is updated based on the description of the mutation scenario. Conditional judgment logic is used. If the mutation magnitude exceeds a preset threshold, the protection parameters are adjusted to obtain the protection action under the load mutation scenario. This is used to dynamically adjust the operating status of distributed devices. A status feedback mechanism is used to obtain system stability indicators and determine the operating status of each distributed device.

9. A collaborative control method based on a low-voltage intelligent switch for power distribution IoT according to claim 8, characterized in that: In step S7, after obtaining the protection actions under the load change scenario, the method further includes: obtaining feedback data from the protection actions, updating the state representation through a dynamic digital model, optimizing local autonomous decision-making using an incremental learning algorithm, and obtaining a new control strategy.

10. A collaborative control system based on a low-voltage intelligent switch for power distribution IoT, used to implement the collaborative control method based on a low-voltage intelligent switch for power distribution IoT as described in any one of claims 1-9, characterized in that: include: The data acquisition and preprocessing module obtains real-time environmental parameters from distributed devices deployed in low-voltage smart switches, performs dimensionality reduction processing through principal component analysis algorithm to generate low-dimensional state representations, and uses interpolation to supplement missing features to determine the input data for constructing a dynamic digital model. The local control strategy generation module uses a pre-trained neural network model to process the high-frequency data collected by the distributed devices locally, generate local control strategies, dynamically adjust the control strategies according to the device's operating trends, and generate autonomous decision-making instructions for the devices. The data transmission and communication module extracts key features from the local control strategy and transmits them to the cloud via power line communication, while using time-sensitive network protocol to process the transmitted data. The cloud optimization module performs integrity checks and optimization on data transmitted from distributed devices. It uses a distributed consensus algorithm combined with real-time environmental parameters to generate global collaborative optimization logic and generate unified control commands. The instruction issuance and execution module issues unified control instructions generated in the cloud to distributed devices via power line communication, and encrypts the unified control instructions using a symmetric encryption algorithm. The distributed devices then receive the instructions, decrypt them, verify them, and execute them synchronously. The synchronous execution and fault isolation module uses a fast switching control algorithm to perform millisecond-level fault isolation on the synchronous execution sequence, and generates protection actions in the event of sudden load changes. It also uses a timestamp verification mechanism to ensure microsecond-level synchronous execution of instructions. The dynamic feedback and optimization module obtains feedback data from protection actions, updates the state representation through a dynamic digital model, and uses an incremental learning algorithm to optimize local autonomous decisions and generate new control strategies.

Images