Adaptive Control System for Intelligent Power Supply and Distribution Substation Based on Edge Computing in Modular Prefabricated Tunnels

Abstract

This invention relates to the field of tunnel power supply and distribution technology, specifically to an adaptive control system for a modular prefabricated tunnel power supply and distribution intelligent substation based on edge computing. It includes: an edge computing and AI decision-making unit for rapid acquisition, processing, and real-time decision-making of multi-dimensional data on tunnel power supply and distribution; generating adaptive control strategies for power supply and distribution by deploying computing resources and machine learning algorithms at edge nodes close to the data source; and converting these strategies into executable control commands; a cloud platform collaborative management unit; and an intelligent sensing and IoT unit. This invention achieves hierarchical decision-making control of the tunnel power supply and distribution system by constructing a hybrid architecture of "edge computing and cloud platform collaboration" and a priority determination mechanism. Furthermore, this invention integrates multimodal data through a multi-protocol communication link module and a full-scenario data fusion analysis module, and performs data correlation analysis using algorithms such as Kalman filtering and D-S evidence theory.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel power supply and distribution, and more specifically, to an adaptive control system for an intelligent substation of modular prefabricated cabins for tunnel power supply and distribution based on edge computing. Background Art

[0002] With the development of tunnel infrastructure towards intelligence, the tunnel power supply and distribution system faces the dual challenges of real-time control and global energy efficiency optimization. The traditional centralized cloud platform control mode has problems such as high network latency and low equipment collaboration efficiency in the complex tunnel environment. Especially in the scenario of equipment failure emergency response, it is difficult to balance the local safety protection and cloud resource scheduling dynamically. The introduction of edge computing technology provides a distributed computing architecture for the tunnel power supply and distribution system. However, how to achieve the collaborative control of edge nodes and cloud platforms, multi-source data fusion analysis, and multi-system linkage decision-making remains a technical difficulty亟待解决 in this field.

[0003] For example, Chinese Patent CN202410055014.7 discloses a tunnel integrated cloud control system, which includes an application module, a data processing module, a communication module, a sensing module, and a control module; the sensing module is used to collect monitoring data of various main mechanical and electrical equipment in the tunnel; the communication module is used to upload the monitoring data; the data processing module is used to process the monitoring data to confirm the operation status of various main mechanical and electrical equipment in the tunnel; the application module is used to query the operation status of various main mechanical and electrical equipment in the tunnel and issue control instructions; the control module is used to transmit the control instructions to the corresponding main mechanical and electrical equipment. Through the above method, the operation status of various main mechanical and electrical equipment in the tunnel can be monitored with low monitoring difficulty, which is beneficial to the effect of tunnel integrated management. Another example is Chinese Patent CN202210826831.9, which discloses a tunnel management and control system based on cloud-edge-end collaboration. The system includes a tunnel edge controller and a tunnel terminal system and a cloud platform server connected to the tunnel edge controller; the tunnel terminal system is used to detect the lighting, fire protection, environment, traffic, and power distribution data in the tunnel and feedback it to the tunnel edge controller; the tunnel edge controller is used to process the data detected by the tunnel terminal system and transmit it to the cloud platform server; the cloud platform server is used to store the data in the tunnel edge controller. In the present invention, by setting up a tunnel edge controller for data processing, it has the ability to connect multiple devices with AI, access RSU, intelligent sensing terminal devices, and intelligent control terminal devices, and realize the access, aggregation, and management functions of monitoring data of tunnel main bodies and ancillary facilities, driving vehicle operation information, and roadside environment perception data.

[0004] While the aforementioned existing technical solutions all have their design advantages, they also suffer from the following technical defects: First, a lack of control decision-making hierarchy: the pure cloud architecture of Chinese patent CN202410055014.7 and the "edge-only forwarding" mode of Chinese patent CN202210826831.9 neither establish a "layered collaborative mechanism between local emergency decision-making (such as fault protection) and cloud-based global optimization (such as energy efficiency scheduling)." Short circuits and arc faults in tunnels require millisecond-level responses, but the instruction transmission delay of the pure cloud architecture or the logic of edge-only forwarding cannot prioritize triggering local protection at the moment of the fault, nor can it reconcile the conflict between safety instructions and energy efficiency scheduling instructions, making it difficult to balance the timeliness and rationality of the system response in emergency situations. Secondly, the depth of data fusion is insufficient: Chinese patent CN202410055014.7 focuses on the electrical parameters of electromechanical equipment, while Chinese patent CN202210826831.9 extends to data from multiple systems such as lighting and fire protection. However, both remain at the basic level of "data collection-aggregation-storage," failing to construct a multimodal correlation model encompassing "electrical parameters + environmental parameters + equipment lifecycle status." Due to the lack of deep data fusion and fault prediction capabilities, it is difficult to provide early warnings for complex operating conditions such as equipment aging and compound faults, thus hindering the effectiveness of preventive maintenance of the tunnel power supply and distribution system. Thirdly, the physical deployment adaptability is poor: both Chinese patents CN202410055014.7 and CN202210826831.9 focus on the system functional architecture and fail to design modular integration and physical protection schemes for equipment in light of the high humidity, dust, and narrow, dispersed environment of tunnels. Existing solutions often involve dispersed equipment installations, which neither enhances protection levels through enclosure isolation nor enables rapid deployment and fault isolation for segmented power supply and distribution within tunnels. This leads to significant issues such as equipment crosstalk and low maintenance efficiency during project implementation, increasing the complexity of system operation and maintenance throughout its entire lifecycle. Therefore, we propose an adaptive control system for intelligent substations with modular prefabricated tunnel power supply and distribution based on edge computing. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive control system for intelligent substation power supply and distribution based on edge computing modular prefabricated tunnel modules, in order to solve the problems mentioned in the background art.

[0006] To address the aforementioned technical problems, the present invention aims to provide an adaptive control system for a smart substation with modular prefabricated tunnel power supply and distribution based on edge computing, comprising:

[0007] The edge computing and AI decision-making unit is used to realize the rapid collection, processing and real-time decision-making of multi-dimensional data on power supply and distribution in tunnels. By deploying computing resources and machine learning algorithms at edge nodes close to the data source, it generates adaptive control strategies for power supply and distribution and transforms the strategies into executable control instructions. Based on the collaborative control instructions (including strategy type, parameter threshold and execution priority) issued by the cloud platform collaborative management unit, the final control instructions are generated through a priority determination mechanism (local emergency instructions > cloud optimization instructions).

[0008] The cloud platform collaborative management unit constructs a global data storage and analysis center to receive data from edge computing nodes and provide a global perspective on power grid operation. Through big data analysis and resource optimization algorithms, it issues collaborative control instructions to the edge computing and AI decision-making units to assist in optimizing local decision-making.

[0009] The intelligent sensing and IoT unit is used to monitor the operating parameters of the tunnel power supply and distribution equipment, the cabin environment and fire status in real time. It uses a multi-protocol communication link module to transmit data to the edge computing and AI decision-making unit and the cloud platform collaborative management unit, so as to build a full-scene monitoring system through the full-scene data fusion and analysis module.

[0010] Modular prefabricated cabin units integrate power supply and distribution equipment, fire protection systems and edge computing components. They adopt a separate steel structure design for high-pressure and low-pressure cabins, combined with fireproof and heat insulation technology, to achieve physical isolation, rapid installation and environmental protection.

[0011] The automated adaptive control unit executes the final control commands of the edge computing and AI decision-making unit. Through power supply and distribution circuit switching, micro-environment control, and multi-system linkage logic of decision and control modules, it realizes real-time power flow monitoring and emergency response to sudden events.

[0012] As a further improvement to this technical solution, the edge computing and AI decision-making unit includes a data processing module, a strategy generation module, and an instruction conversion module, wherein:

[0013] The data processing module is used to clean and normalize the raw data transmitted by the intelligent sensing and IoT units, and to remove noise using a sliding window filtering algorithm with a preset window width.

[0014] The strategy generation module generates an adaptive power supply and distribution control strategy based on a multi-layer neural network. The input of the neural network is historical load data, and the output is the power prediction value for future periods.

[0015] The instruction conversion module is used to convert the control strategy into executable control instructions, adopts an industrial control standard protocol, and the instruction generation delay meets the requirements of real-time control.

[0016] As a further improvement to this technical solution, the edge computing and AI decision-making unit generates the final control instruction through a priority determination mechanism, including the following steps:

[0017] S100.1. Parse the local real-time instructions (such as emergency power outage instructions triggered by equipment failure) and collaborative control instructions (such as peak and valley power supply optimization strategies) received by the edge computing and AI decision-making unit, and extract key features for priority calculation. The key features include instruction type, target object, parameter value range and time attribute.

[0018] Specifically:

[0019] Instruction types (divided into urgent and optimized categories);

[0020] Target (locating to a specific device or system module);

[0021] Parameter range (safety boundaries of key parameters in the acquisition command);

[0022] Timeliness attribute (records the time difference between the instruction generation time and the trigger time).

[0023] S100.2. Based on device status and command attributes, calculate the local command priority value using a two-factor weighting model. Priority value of coordinated control instructions The calculation formula is:

[0024]

[0025] in, Indicates the urgency factor; Indicates the current temperature rise of the equipment; This represents the average temperature rise under historical load conditions, used to reflect the degree of equipment abnormality.

[0026] ;

[0027] in, Indicates the optimization factor; This represents the difference between the real-time load rate and the rated load rate. This indicates the rated load rate, reflecting the urgency of system regulation.

[0028] S100.3, Based on local instruction priority value Priority value of coordinated control instructions The calculation results are used to execute hierarchical decision-making logic, which specifically includes:

[0029] First priority strategy (for emergency scenarios):

[0030] when Furthermore, when the instruction type is emergency (such as equipment short circuit or over-temperature alarm), the edge computing and AI decision-making unit generates the highest priority control instruction, and executes equipment protection actions (such as cutting off the fault circuit or starting the backup power supply) through the automated adaptive control unit. This is a threshold for equipment safety urgency (range 0-1), used to distinguish between normal operation and emergency failure scenarios. Equipment safety margin determined based on modular prefabricated cabin unit 400;

[0031] Specifically, the edge computing and AI decision-making unit immediately triggers the highest priority control process, interrupting the cloud instruction queue; the automated adaptive control unit performs fault isolation (such as cutting off the circuit or starting cooling) based on the device's safety response characteristics, and after the action is completed, it sends back the system clock synchronization status information through the intelligent sensing and IoT unit; the cloud platform collaborative management unit receives instruction conflict alarms, including priority differences ( ), Fault type (determined by local sensor data);

[0032] Parameter fusion strategy (corresponding to collaborative optimization scenarios):

[0033] when Alternatively, if the instruction type is optimization-related (such as peak-valley scheduling or reactive power compensation), the control parameters of the local real-time instructions and the coordinated control instructions are weighted and fused.

[0034] ;

[0035] in, This indicates the final control parameters that are fused and sent to the automated adaptive control unit, and Meet the electrical safety boundaries (such as voltage regulation range ± α% of rated value, where α is provided by the equipment manufacturer) and mechanical life constraints (such as tap changer interval ≥ minimum operating cycle of the equipment) of the modular prefabricated cabin unit. This represents the local control parameters generated by edge computing and AI decision-making units; This refers to the cloud-based control parameters generated by the cloud platform's collaborative management unit;

[0036] Second priority strategy (corresponding to global optimization scenarios):

[0037] when At this time, the edge computing and AI decision-making unit prioritizes executing cloud-based collaborative instructions, adjusting system operating parameters (such as load distribution and reactive power compensation) through an automated adaptive control unit, and thresholding... A global optimization priority threshold is used to balance local security and cloud energy efficiency goals, and the threshold... Determining the global energy efficiency model based on the cloud platform collaborative management unit.

[0038] Specifically, the fusion command must pass checks on parameter validity, equipment compatibility, and power grid stability; failure in any of these checks will trigger manual intervention.

[0039] Parameter validity: Verify whether the electrical / mechanical parameters exceed the rated range of the modular prefabricated cabin unit; if they exceed the limit, mark "parameter abnormal";

[0040] Equipment compatibility: Verify the number of operations / intervals allowed for the remaining lifespan of circuit breakers, capacitors, etc., and mark "lifespan warning" if the limit is exceeded;

[0041] Power grid stability: The power flow calculation is used to verify whether the operating status of the lines / buses meets the system safety boundaries (such as active power flow and voltage deviation meeting the power grid specifications). If the limits are exceeded, "power flow anomaly" is marked.

[0042] As a further improvement to this technical solution, the edge computing and AI decision-making unit also includes a strategy verification module, a parameter optimization module, and a version management module, wherein:

[0043] The strategy verification module verifies the effectiveness of the control strategy through the Monte Carlo simulation algorithm. The preset number of simulations is the industry standard value. Based on the device operating parameters collected by the intelligent sensing and IoT unit, it generates an evaluation report containing the risk probability distribution. The evaluation report is used to correct the priority judgment threshold parameters built into the edge computing and AI decision-making unit.

[0044] The parameter optimization module dynamically adjusts the strategy parameters based on a deep reinforcement learning framework, with the goal of minimizing grid loss. The optimization cycle is dynamically issued by the cloud platform collaborative management unit according to the load fluctuation characteristics, and directly outputs the two-factor weight parameters used for priority calculation in the edge computing and AI decision-making units.

[0045] The version management module uses blockchain hash storage technology to record historical versions of the strategy (including priority determination strategy configuration for edge computing and AI decision-making units). It supports one-click rollback to the previous valid version in case of anomalies. The rollback time meets the equipment fault tolerance requirements of the modular prefabricated cabin unit. The rollback trigger condition is linked to the security verification results of edge computing and AI decision-making units.

[0046] As a further improvement to this technical solution, the global data storage and analysis center of the cloud platform collaborative management unit adopts a distributed time-series database cluster architecture. It performs millisecond-level sampling and storage of device operation data (voltage, current, temperature, etc.) uploaded by the edge computing and AI decision-making units, and realizes global visualization of the power grid operation status. The distributed time-series database cluster architecture specifically includes:

[0047] A digital twin is constructed based on a 3D tunnel topology model and a power flow algorithm to render a load distribution heat map of the power supply and distribution network in real time.

[0048] By using the Long Short-Term Memory (LSTM) network to predict trends in historical data, a risk warning map of power grid operation for the next 24 hours is generated.

[0049] As a further improvement to this technical solution, the cloud platform collaborative management unit uses big data analysis and resource optimization algorithms to generate collaborative control instructions, including the following steps:

[0050] S200.1. A consistent hashing algorithm is used to manage the tunnel group power supply and distribution network by region, and the load rate of each region is monitored in real time. When the load rate of a certain region exceeds the preset overload threshold (e.g., 80%), the load transfer strategy of the adjacent region is triggered, and the transfer power threshold is calculated by the following formula. :

[0051] ;

[0052] in, For the area's rated capacity, The preset overload threshold is a dynamic adjustment factor (obtained by training the inter-regional line impedance matrix) and is the industry's conventional load rate critical value (such as the overload standard defined in GB / T50598-2010).

[0053] S200.2. Taking the minimization of grid losses, the maximization of new energy consumption, and the maximization of power supply reliability as objective functions, a non-dominated sorting genetic algorithm is used for iterative optimization to generate coordinated control instructions that include transformer tap adjustment levels (adjustment step size ≤ 1.25% of rated voltage), the number of capacitor switching groups (single switching capacity difference ≤ 500kvar), and the execution sequence of load transfer strategies (interval between adjacent areas ≥ 10 seconds). The coordinated control instructions are approximate Pareto front solutions, which are used by edge computing and AI decision-making units to select and execute based on local real-time status.

[0054] As a further improvement to this technical solution, the multi-protocol communication link module includes an industrial bus sub-link sub-module, a wireless sensing sub-link sub-module, and a 5G core sub-link sub-module, wherein:

[0055] The industrial bus sub-link sub-module uses the Modbus protocol to build a dedicated communication channel for power equipment, and performs millisecond-level periodic acquisition of voltage, current and temperature parameters of transformers and switchgear in the tunnel, adapting to the strong real-time monitoring requirements of power equipment.

[0056] The wireless sensing sub-link sub-module: builds an environmental monitoring network based on the LoRaWAN protocol, collects temperature, humidity and smoke concentration parameters, optimizes node energy consumption through a sleep-wake low-power communication mechanism, and adapts to tunnel cabling-restricted scenarios.

[0057] The 5G core sub-link sub-module adopts a 5GURLLC slicing channel to perform real-time priority transmission of fire alarm and gas concentration data, and the transmission latency meets the requirements of fire emergency response standards.

[0058] The full-scenario data fusion and analysis module includes a multimodal data fusion submodule, a device health assessment submodule, and an adaptive sampling control submodule, wherein:

[0059] The multimodal data fusion submodule is used to align electrical parameters (current, voltage), environmental parameters (temperature and humidity), and fire protection parameters (smoke concentration) in a spatiotemporal dimension, and uses a Kalman filter algorithm to eliminate data noise. The fusion accuracy meets the standards for power equipment condition monitoring.

[0060] The equipment health assessment submodule constructs a fault diagnosis model based on DS evidence theory, extracts the circuit breaker opening and closing coil current waveform and transformer oil temperature change rate as core feature quantities, and realizes early fault warning of equipment.

[0061] After receiving abnormal operating condition signals from the edge computing and AI decision-making unit, the adaptive sampling control submodule automatically adjusts the sampling frequency of key equipment and constructs a three-level sampling mechanism of normal cycle acquisition, abnormal trigger acquisition, and high-frequency tracking acquisition to meet the dynamic monitoring accuracy requirements under different operating conditions.

[0062] As a further improvement to this technical solution, the modular prefabricated cabin unit includes a high-pressure cabin module, a low-pressure cabin module, and a fire protection integrated module, wherein:

[0063] The high-voltage compartment module includes a switchgear submodule and a transformer submodule, wherein:

[0064] The switch cabinet sub-module has a built-in 10kV metal armored withdrawable switch cabinet with a Q355B steel plate frame and an arc light protection sensor installed inside the cabinet.

[0065] The transformer submodule is a dry-type transformer with an outer casing that has a protection level of not less than IP54. The cooling fan is linked to the temperature controller.

[0066] The low-pressure chamber module includes an intelligent power distribution module and an edge computing server sub-module, wherein:

[0067] The intelligent power distribution module integrates a molded case circuit breaker and an intelligent power meter, and supports Modbus protocol communication.

[0068] The edge computing server submodule adopts an industrial-grade server, is installed with a shock-absorbing bracket, and is suitable for wide-temperature operating environments.

[0069] The fire protection integrated module includes a fire extinguishing device submodule and a fire detection submodule, wherein:

[0070] The fire extinguishing device submodule integrates a heptafluoropropane gas fire extinguishing system, with nozzles arranged at preset intervals and equipped with a linkage triggering mechanism.

[0071] The fire detection submodule uses a smoke and heat sensor composite detector, and its detection sensitivity meets relevant national standards.

[0072] As a further improvement to this technical solution, the automated adaptive control unit includes an intelligent power distribution execution module and a micro-environment control module, wherein:

[0073] The intelligent power distribution execution module uses solid-state circuit breakers to construct an intelligent power distribution network. The intelligent power distribution execution module includes a fault isolation submodule and a dynamic load allocation submodule, wherein:

[0074] The fault isolation submodule is based on a high-speed current sensor and a solid-state circuit breaker to achieve rapid isolation of short-circuit faults.

[0075] The dynamic load allocation submodule dynamically adjusts the transformer tap position using a fuzzy PID controller based on the load prediction results from edge computing and AI decision-making units.

[0076] The microenvironment control module achieves adaptive adjustment of the tunnel microenvironment through an IoT edge controller, including a multi-parameter collaborative control submodule and a zoned temperature equalization submodule, wherein:

[0077] The multi-parameter collaborative control submodule dynamically adjusts the ventilation fan speed based on CO concentration, visibility, and traffic flow using a model predictive control algorithm.

[0078] The partitioned temperature equalization submodule adjusts the power of the tunnel lighting system through a PID controller to achieve cabin temperature difference control.

[0079] As a further improvement to this technical solution, the decision-making and control module includes a collaborative control submodule and an emergency response decision tree submodule, wherein:

[0080] The collaborative control submodule is based on an event bus architecture to build an inter-system communication mechanism, specifically including:

[0081] Assign dynamic priorities to systems such as fire protection, ventilation, and lighting (e.g., fire protection system priority is the highest level 1) to ensure that the response time of critical systems meets the real-time requirements of fire emergency response;

[0082] The Raft consensus algorithm is used to achieve state synchronization between control units, and the synchronization accuracy meets the standards for collaborative control of industrial-grade distributed systems.

[0083] The emergency response decision tree submodule constructs an emergency response model for sudden events based on a Bayesian network inference engine, specifically including:

[0084] A set of typical fault modes for the entire life cycle of a predefined tunnel power supply and distribution system (covering basic faults such as single-phase grounding and three-phase short circuits, as well as derivative composite faults).

[0085] By updating the fault probability distribution in real time through Bayesian inference, the optimal emergency response strategy is generated, and the timeliness of strategy generation meets the rapid response requirements for power system fault handling.

[0086] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0087] 1. This invention achieves hierarchical decision-making control of tunnel power supply and distribution systems by constructing a hybrid architecture and priority determination mechanism that integrates edge computing and cloud platform collaboration. When equipment experiences emergency faults such as short circuits or overheating, the edge computing unit can prioritize executing local protection actions (such as cutting off the faulty circuit), while the cloud platform provides global resource optimization assistance. This mechanism effectively reconciles the conflicts between local safety commands and cloud-based energy efficiency commands, ensuring the timeliness and rationality of system response in emergency situations and avoiding the command delay problem of traditional pure cloud architectures.

[0088] 2. This invention integrates multimodal data, including electrical parameters, environmental parameters, and fire protection parameters, through a multi-protocol communication link module and a full-scene data fusion analysis module. It then uses algorithms such as Kalman filtering and DS evidence theory to achieve data correlation analysis. This solution improves the accuracy of equipment fault diagnosis, enabling early prediction of complex faults such as insulation aging and surface discharge. Compared to traditional single-dimensional monitoring solutions, it effectively enhances the preventative maintenance capabilities of tunnel power supply and distribution systems.

[0089] 3. This invention utilizes a separate steel structure design for the high-pressure and low-pressure chambers, combined with fireproof and heat-insulating technology and an IP54 protection rating, making it suitable for the complex environment of tunnels characterized by high humidity and dust. This design achieves physical isolation of equipment and rapid installation, supports the deployment requirements of segmented power supply and distribution in tunnels, and reduces the risk of crosstalk through modular chamber design, improving later maintenance efficiency and effectively solving the adaptability problem of traditional substation cabinet designs in tunnel scenarios. Attached Figure Description

[0090] Figure 1 This is a schematic diagram of the system framework of the present invention;

[0091] The meanings of the labels in the diagram are as follows:

[0092] 100. Edge Computing and AI Decision-Making Unit; 110. Data Processing Module; 120. Policy Generation Module; 130. Instruction Conversion Module; 140. Policy Verification Module; 150. Parameter Optimization Module; 160. Version Management Module;

[0093] 200. Cloud Platform Collaborative Management Unit;

[0094] 300. Intelligent Sensing and IoT Unit; 310. Multi-Protocol Communication Link Module; 311. Industrial Bus Sub-Link Sub-Module; 312. Wireless Sensing Sub-Link Sub-Module; 313. 5G Core Sub-Link Sub-Module; 320. Full-Scenario Data Fusion and Analysis Module; 321. Multimodal Data Fusion Sub-Module; 322. Equipment Health Assessment Sub-Module; 323. Adaptive Sampling Control Sub-Module;

[0095] 400. Modular prefabricated cabin unit; 410. High-voltage cabin module; 411. Switchgear sub-module; 412. Transformer sub-module; 420. Low-voltage cabin module; 421. Intelligent power distribution module; 422. Edge computing server sub-module; 430. Fire protection integrated module; 431. Fire extinguishing device sub-module; 432. Fire detection sub-module;

[0096] 500. Automated Adaptive Control Unit; 510. Intelligent Power Distribution Execution Module; 511. Fault Isolation Submodule; 512. Dynamic Load Allocation Submodule; 520. Microenvironment Control Module; 521. Multi-Parameter Collaborative Control Submodule; 522. Zoned Temperature Equalization Submodule; 530. Decision and Control Module; 531. Collaborative Control Submodule; 532. Emergency Response Decision Tree Submodule. Detailed Implementation

[0097] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0098] like Figure 1 As shown, this embodiment provides an adaptive control system for a smart substation with modular prefabricated tunnel power supply and distribution based on edge computing, including:

[0099] The edge computing and AI decision-making unit 100 is used to realize the rapid collection, processing and real-time decision-making of multi-dimensional data of tunnel power supply and distribution. By deploying computing resources and machine learning algorithms at edge nodes close to the data source, it generates power supply and distribution adaptive control strategies and transforms the strategies into executable control instructions. Based on the collaborative control instructions (including strategy type, parameter threshold and execution priority) issued by the cloud platform collaborative management unit 200, the final control instructions are generated through a priority determination mechanism (local emergency instructions > cloud optimization instructions).

[0100] In this embodiment, the edge computing and AI decision-making unit 100 includes a data processing module 110, a policy generation module 120, and an instruction conversion module 130, wherein:

[0101] The data processing module 110 is used to clean and normalize the raw data transmitted by the intelligent sensing and IoT unit 300, and uses a sliding window filtering algorithm to remove noise, with the window width set to a preset value.

[0102] As a further explanation of this embodiment, the data processing module 110 is used to perform real-time filtering and preprocessing on the raw data transmitted by the intelligent sensing and IoT unit 300, and adopts a dynamic sliding window filtering algorithm to improve data reliability. Specifically, the algorithm adaptively adjusts the window width according to the device's operating status: under normal operating conditions, a 100-point window (corresponding to a 100ms sampling period and a sampling frequency of 1kHz) is used to remove random noise through mean filtering; when the device's temperature rise rate is detected to exceed a preset threshold (e.g., 0.5℃ / s), it automatically switches to a 50-point narrow window to improve the filtering accuracy of high-frequency noise. The sliding window filtering algorithm is implemented through the following mechanism: historical sampling data is stored using a circular buffer, and new data is written to overwrite the oldest data, and filtering is completed by summation and mean calculation. When the data missing rate within the window exceeds 10%, a forward interpolation completion strategy is triggered to ensure data continuity. At the hardware level, a parallel computing architecture can be built using a Xilinx Artix-7 FPGA, and pipelined design can be used to achieve zero-latency processing of 100Hz sampling data, meeting the requirements of the DL / T860 standard for real-time data processing of power systems. Furthermore, the data processing module 110 incorporates built-in data validity verification logic to identify and mark anomalies in the filtered data, providing a reliable data foundation for subsequent strategy generation. This design, through dynamic window adjustment and hardware acceleration, effectively suppresses data noise in the high electromagnetic interference environment of the tunnel, ensuring the accuracy of decision-making by the edge computing unit.

[0103] The strategy generation module 120 generates an adaptive control strategy for power supply and distribution based on a multi-layer neural network. The input of the neural network is historical load data, and the output is the power prediction value for the future period.

[0104] As a further explanation of this embodiment, the strategy generation module 120 constructs a load prediction model based on a multi-layer neural network to generate an adaptive power supply and distribution control strategy. First, the strategy generation module 120 adopts a three-layer fully connected neural network structure: the input layer contains 12 neurons, corresponding to 12 hours of historical load data; the hidden layer has 24 neurons and uses the ReLU activation function (the mathematical expression is...). , The gradient vanishing problem is avoided by preserving the linear response characteristics of non-negative inputs; the number of neurons is designed to be twice the expansion of the input dimension (12-dimensional historical load data), which strengthens the dimensional support capability of nonlinear feature mapping; the output layer has 1 neuron, which outputs the power prediction value for the next hour.

[0105] Secondly, the neural network is implemented using the TensorFlow framework and trained with the Adam optimizer to minimize grid losses. The initial learning rate is set to 0.001 and decays by 10% every 50 iterations. The loss function uses mean squared error (MSE) to ensure prediction stability.

[0106] Then, before model training, the historical load data was standardized, and the Z-Score method was used to map the data to a distribution interval with a mean of 0.8 and a standard deviation of 0.2, thereby improving the network convergence speed. The input sequence was constructed using a sliding window mechanism, with a step size of 1 hour to extract 12 consecutive hours of data as samples, and the batch size was set to 32. Iterative training was conducted for 200 rounds, and the validation set accounted for 20% to avoid overfitting.

[0107] Meanwhile, to address the unique load fluctuations characteristic of tunnel scenarios, the module integrates a residual correction mechanism: when the deviation between the predicted and actual values ​​exceeds 15% of the rated power, an LSTM-based residual network is automatically triggered. This network uses environmental parameters such as tunnel humidity and traffic flow, along with the initial prediction error, as input. It captures temporal features through a 16-unit LSTM layer. The LSTM layer internally employs the tanh activation function (mathematically expressed as: Used for cell state renewal, among which, The candidate values ​​of cell state correspond to the temporal feature inputs of the residual correction network: 8-dimensional data such as tunnel humidity, traffic flow, and initial prediction error, and the calculation results after weighting by the LSTM input gate; (representing state update magnitude adjustment) and the Sigmoid activation function (mathematical expression: ,in, This represents the weighted sum of the inputs to the gating mechanism; The combination of gated logic (for input gate, forget gate, and output gate) with 16 units adapts to the 8-dimensional input features composed of environmental parameters (humidity, traffic flow) and initial prediction error. It balances the accuracy and computational efficiency of low-dimensional time series modeling by using a 2x redundant dimension. At this time, the output residual correction value is superimposed on the initial prediction result to form the final power prediction value, providing a more accurate decision basis for subsequent control strategies.

[0108] In addition, at the hardware level, the policy generation module 120 can be deployed on the ARM Cortex-A72 architecture processor (such as NXPi.MX8MPlus) of the edge computing node, using its built-in NPU coprocessor to accelerate neural network calculations, and the time for a single prediction is controlled within 1 second, which meets the real-time requirements of the tunnel power supply and distribution system.

[0109] The instruction conversion module 130 is used to convert the control strategy into executable control instructions. It adopts the industrial control standard protocol, and the instruction generation delay meets the real-time control requirements.

[0110] In this embodiment, generating the final control instruction through the priority determination mechanism includes the following steps:

[0111] S100.1 The local real-time instructions (such as emergency power outage instructions triggered by equipment failure) and collaborative control instructions (such as peak and valley power supply optimization strategies) received by the edge computing and AI decision-making unit 100 are parsed, and key features for priority calculation are extracted. The key features include instruction type, target object, parameter value range and time attribute.

[0112] Instruction types (divided into urgent and optimized categories);

[0113] Target (locating to a specific device or system module);

[0114] Parameter range (safety boundaries of key parameters in the acquisition command);

[0115] Timeliness attribute (records the time difference between the instruction generation time and the trigger time).

[0116] S100.2. Based on device status and command attributes, calculate the local command priority value using a two-factor weighting model. Priority value of coordinated control instructions The calculation formula is:

[0117]

[0118] in, Indicates the urgency factor; Indicates the current temperature rise of the equipment; This represents the average temperature rise under historical load conditions, used to reflect the degree of equipment abnormality.

[0119] ;

[0120] in, Indicates the optimization factor; This represents the difference between the real-time load rate and the rated load rate. This indicates the rated load rate, reflecting the urgency of system regulation.

[0121] S100.3, Based on local instruction priority value Priority value of coordinated control instructions The calculation results are used to execute hierarchical decision-making logic, which specifically includes:

[0122] First priority strategy (for emergency scenarios):

[0123] when Furthermore, when the instruction type is emergency (such as equipment short circuit or over-temperature alarm), the edge computing and AI decision unit 100 generates the highest priority control instruction, which is then executed by the automated adaptive control unit 500 to perform equipment protection actions (such as cutting off the fault circuit or starting the backup power supply). This is a threshold for equipment safety urgency (range 0-1), used to distinguish between normal operation and emergency failure scenarios. Equipment safety margin determined based on modular prefabricated cabin unit 400;

[0124] Specifically, the edge computing and AI decision-making unit 100 immediately triggers the highest priority control process, interrupting the cloud command queue; the automated adaptive control unit 500 performs fault isolation based on the device's safety response characteristics, such as cutting off the circuit or starting cooling, and after the action is completed, it sends back the system clock synchronization status information through the intelligent sensing and IoT unit 300; the cloud platform collaborative management unit 200 receives command conflict alarms, including priority differences. The fault type is determined by local sensor data;

[0125] Parameter fusion strategy (corresponding to collaborative optimization scenarios):

[0126] when Alternatively, if the instruction type is optimization-related (such as peak-valley scheduling or reactive power compensation), the control parameters of the local real-time instructions and the coordinated control instructions are weighted and fused.

[0127] ;

[0128] in, This indicates the final control parameters that are fused and sent to the automated adaptive control unit 500, and Meets the electrical safety boundaries (such as voltage regulation range ± α% of rated value, where α is provided by the equipment manufacturer) and mechanical life constraints (such as tap change interval ≥ minimum operating cycle of the equipment) of the modular prefabricated cabin unit 400. This represents the local control parameters generated by the edge computing and AI decision-making unit 100; This refers to the cloud-based control parameters generated by the cloud platform collaborative management unit 200;

[0129] Second priority strategy (corresponding to global optimization scenarios):

[0130] when At the same time, the edge computing and AI decision-making unit 100 prioritizes executing cloud-based collaborative instructions and adjusts system operating parameters (such as load distribution and reactive power compensation) through the automated adaptive control unit 500. A global optimization priority threshold is used to balance local security and cloud energy efficiency goals, and the threshold... The global energy efficiency model is determined based on the cloud platform collaborative management unit 200.

[0131] Specifically, the fusion command must pass checks on parameter validity, equipment compatibility, and power grid stability; failure in any of these checks will trigger manual intervention.

[0132] Parameter validity: Verify whether the electrical / mechanical parameters exceed the rated range of the modular prefabricated cabin unit 400; if they exceed the limit, mark "parameter abnormal";

[0133] Equipment compatibility: Verify the number of operations / intervals allowed for the remaining lifespan of circuit breakers, capacitors, etc., and mark "lifespan warning" if the limit is exceeded;

[0134] Power grid stability: The power flow calculation is used to verify whether the operating status of the lines / buses meets the system safety boundaries (such as active power flow and voltage deviation meeting the power grid specifications). If the limits are exceeded, "power flow anomaly" is marked.

[0135] As a further explanation of this embodiment, the urgency factor Dynamic correction based on historical fault handling results: When a short circuit fault occurs in the equipment, if the protection action successfully eliminates the fault (confirmed by the status signal returned by the intelligent sensing unit). Keep the original value of 1.5; if a protection failure occurs (such as cascading trip), then The value is incremented by 0.1 (up to a maximum of 2.0), and the correction value is stored in the EEPROM of the edge computing and AI decision-making unit 100. This mechanism, through learning from multiple fault samples, can improve the success rate of handling similar faults; simultaneously, the optimization factor... This value is related to the number of global optimization iterations on the cloud platform: the initial value is 0.8 (load rate < 30%), and it is reassigned based on the current load rate range after each global energy efficiency optimization by the cloud platform. For example, when the load rate increases from 25% to 40%, The version was adjusted from 0.8 to 1.0 and synchronized to register 40023 of the edge computing and AI decision-making unit 100 via the Modbus protocol.

[0136] Furthermore, in this embodiment, for a 10kV dry-type transformer (rated temperature 105℃), when the current operating temperature is 80℃:

[0137] Safety margin factor = (10⁵ - 80) / 10⁵ = 0.238 =0.238×0.3≈0.07; For a circuit breaker (rated current 1250A), when the real-time current is 1000A:

[0138] The safety margin factor = (1250-1000) / 1250 = 0.2. Take a fixed value of 0.2 (because the safety margin factor is <0.6);

[0139] also, The initial value is 0.5 (cloud platform iteration count n=0), and n+1 is added after each optimization. When n=10, =0.5 + 10 × 0.05 = 1.0, and will not increase further after reaching the upper limit. This threshold is verified by the global energy efficiency model of the cloud platform collaborative management unit 200 to ensure a balance between local security and cloud optimization.

[0140] It is understood that the priority determination algorithm in this embodiment has a built-in fault tolerance mechanism for abnormal operating conditions. When the system detects the following operational abnormalities, it automatically initiates the fault tolerance process to ensure the continuity and reliability of the control strategy, specifically including:

[0141] (1) Fault tolerance handling for missing historical temperature rise data If the intelligent sensing and IoT unit 300 cannot obtain the historical average temperature rise of the device (In scenarios such as initial equipment commissioning or storage module failure), based on the safety redundancy principle of power equipment temperature rise design, 80% of the equipment's rated temperature rise is taken as a temporary replacement value. Taking a 10kV dry-type transformer as an example, when its rated temperature rise is 65℃, the temporary... The value is assigned to 52℃. This value is based on the "temperature rise margin under non-full load conditions" requirement in the thermal design code for power equipment to avoid misjudgment or failure of the protection strategy due to missing data;

[0142] (2) Fault-tolerant handling of unsynchronized target load rate in the cloud: When the cloud platform collaborative management unit 200 fails to issue the target load rate for peak and valley periods in a timely manner, the edge computing unit automatically switches to the preset economic operation mode, and the default load rate is 60% as the control benchmark. This load rate is derived from the engineering practice of economic operation of power supply and distribution system - most power equipment has the best overall performance in terms of energy efficiency ratio and equipment life when running within the 60% load range, ensuring that the system still maintains efficient operation when there is no global optimization instruction;

[0143] (3) Fault-tolerant handling of hardware computing failures If a hardware anomaly occurs in the ARM Cortex-A72 processor during priority calculation (such as register overflow, arithmetic unit error, etc.), the processor is reset through an independent watchdog module, and the version management module 160 is called to roll back to the previous valid policy version. The total time of the reset and rollback process is controlled within the hundreds of milliseconds to ensure that the system can quickly restart the control logic after the fault is recovered, and to avoid long-term loss of control.

[0144] In this embodiment, the edge computing and AI decision-making unit 100 further includes a policy verification module 140, a parameter optimization module 150, and a version management module 160, wherein:

[0145] The strategy verification module 140 verifies the effectiveness of the control strategy through a Monte Carlo simulation algorithm. The preset number of simulations is the industry standard value. Based on the equipment operating parameters collected by the intelligent sensing and IoT unit 300, it generates an evaluation report containing the risk probability distribution. The evaluation report is used to correct the priority judgment threshold parameters built into the edge computing and AI decision-making unit 100. The intelligent sensing and IoT unit 300 encapsulates parameters such as transformer load rate and circuit breaker temperature into Modbus frames and transmits them to the strategy verification module 140 through the industrial bus sub-link sub-module 311 of the multi-protocol communication link module 310. Registers 40011-40020 are used to transmit these parameters to the strategy verification module 140.

[0146] As a further explanation of this embodiment, the strategy verification module 140 in this embodiment verifies the effectiveness of the control strategy using a Monte Carlo simulation algorithm, setting 10,000 iterations of sampling (this design can cover a 99.7% confidence interval, meeting the requirements of industrial-grade simulation accuracy), and setting the voltage fluctuation range to ±15% of the rated value (based on the definition of extreme operating conditions in "GB / T12325-2008 Power Quality Supply Voltage Deviation"), and the equipment failure probability follows a Poisson distribution (parameter...). =0.02 times / hour, based on the statistical data of power equipment failures in the past 5 years), the convergence condition is that the deviation between two adjacent iterations is ≤0.5% (balancing computational efficiency and accuracy, single-thread simulation time ≤120 seconds); based on the equipment operation parameters transmitted by the intelligent sensing and IoT unit 300 through the industrial bus sub-link sub-module 311, random disturbances are injected into the voltage and load, and with the help of the parallel computing capability of Xilinx Artix-7 FPGA, a risk probability distribution assessment report containing a 95% confidence interval is generated; if the simulation finds that the probability of fault triggering exceeds 5%, the priority judgment threshold is automatically lowered by 10% (e.g., the equipment safety urgency threshold is corrected from 0.3 to 0.27), and the correction result is fed back to the threshold register of the strategy generation module 120.

[0147] The parameter optimization module 150 dynamically adjusts the strategy parameters based on the deep reinforcement learning framework, with the goal of minimizing grid loss. The optimization cycle is dynamically issued by the cloud platform collaborative management unit 200 according to the load fluctuation characteristics, and directly outputs the two-factor weight parameters used for priority calculation in the edge computing and AI decision-making unit 100.

[0148] As a further explanation of this embodiment, the parameter optimization module 150 in this embodiment dynamically adjusts the strategy parameters based on a deep reinforcement learning framework, and extracts the 3D state features (current load rate) of the intelligent power distribution module 421, temperature sensor, and energy efficiency monitoring equipment. Equipment temperature rise Grid losses Define a 6-dimensional action space for ±2-level adjustment of transformer taps and switching of capacitor banks 1-3, using a reward function. ,in The rate of change of loss, For temperature rise changes, To control the number of actions; training is performed using the Q-Learning algorithm (learning rate). Discount factor Each cycle completes 1000 iterations; the cloud platform collaborative management unit 200, through the 5G core sub-link sub-module 313, dynamically issues optimization cycles based on the load fluctuation standard deviation (set to 10 minutes when the load fluctuation standard deviation is >5%, and extended to 30 minutes when it is ≤5%), and directly outputs the dual-factor weight parameters. , The corrected value is sent to the parameter register of the strategy generation module 120 to achieve dynamic strategy optimization.

[0149] The version management module 160 uses blockchain hash storage technology to record historical versions of the strategy (including the priority judgment strategy configuration of the edge computing and AI decision unit 100). It supports one-click rollback to the previous valid version in case of an anomaly. The rollback time meets the equipment fault tolerance requirements of the modular prefabricated cabin unit 400. The rollback trigger condition is linked to the security verification result of the edge computing and AI decision unit 100.

[0150] As a further explanation of this embodiment, the version management module 160 in this embodiment integrates an STM32L4R9 encryption chip, and uses the SHA-256 algorithm to perform hash calculations on the policy version (priority determination policy, two-factor parameters). The block header encapsulates the version number, timestamp, and previous block hash, and the block body stores the parameter binary data. When the edge computing and AI decision-making unit 100 fails to execute the policy three times consecutively, or the security verification parameters exceed the limit (e.g., ... When a hardware calculation error occurs (triggered by Watchdog), the previous valid version of the blockchain is invoked, historical strategy loading is completed within 100ms, and the intelligent power distribution execution module 510 is synchronized via the industrial bus.

[0151] The cloud platform collaborative management unit 200 constructs a global data storage and analysis center to receive data from edge computing nodes and provide a global perspective on power grid operation. Through big data analysis and resource optimization algorithms, it issues collaborative control instructions to the edge computing and AI decision-making unit 100 to assist in optimizing local decision-making.

[0152] In this embodiment, the global data storage and analysis center of the cloud platform collaborative management unit 200 adopts a distributed time-series database cluster architecture to perform millisecond-level sampling and storage of device operation data (voltage, current, temperature, etc.) uploaded by the edge computing and AI decision-making unit 100 and realize global visualization of the power grid operation status. The distributed time-series database cluster architecture specifically includes:

[0153] A digital twin is constructed based on a 3D tunnel topology model and a power flow algorithm to render a load distribution heat map of the power supply and distribution network in real time.

[0154] As a further illustration of this embodiment, this embodiment can use Blender+GIS data to construct a 3D model of the tunnel with an accuracy of 0.5 meters, and integrate the following elements:

[0155] Spatial location of power supply and distribution equipment (transformers, switchgear, cable trenches);

[0156] Tunnel environmental parameters (humidity, temperature field distribution);

[0157] In addition, the OpenAPI provides an interface for querying geometric coordinates and device parameters.

[0158] As a further explanation of this embodiment, the Newton-Raphson method can be used to solve the power flow equations, and the specific parameters include:

[0159] Iteration convergence threshold: voltage amplitude error < 0.001 pu, phase angle error < 0.005 rad;

[0160] Node types: PV node (main transformer high voltage side), PQ node (load point), balance node (tunnel group grid connection side);

[0161] Furthermore, this embodiment can also utilize WebGL technology to render the load heatmap on the browser side, with the color mapping relationship as follows:

[0162] Load rate < 30%: Green (RGB0,255,0);

[0163] 30%-70%: Yellow (RGB255,255,0);

[0164] >70%: Red (RGB255,0,0).

[0165] By using the Long Short-Term Memory (LSTM) network to predict trends in historical data, a risk warning map of power grid operation for the next 24 hours is generated.

[0166] As a further explanation of this embodiment, in the trend prediction model constructed by the Long Short-Term Memory (LSTM) network in this embodiment, the input layer contains 24 neurons (corresponding to 8 time points each for voltage, current, and load rate in the past 24 hours), the hidden layer uses 2 layers of 64-unit LSTM and 1 layer of 32-unit fully connected layer, and the output layer has 3 neurons (predicting the operating parameters for the next 1 hour, 3 hours, and 6 hours). The training process uses the Adam optimizer (learning rate 0.001, decaying by 10% every 50 rounds), the loss function is a combination of mean squared error and mean absolute error, the batch size is 64, and the iteration is 200 rounds (validation set accounts for 20%). When the prediction error exceeds 15%, a secondary correction is triggered, and the correction value = initial prediction value + 0.3 × (actual value - initial prediction value).

[0167] In this embodiment, the cloud platform collaborative management unit 200 uses big data analysis and resource optimization algorithms to generate collaborative control instructions, including the following steps:

[0168] S200.1. A consistent hashing algorithm is used to manage the tunnel group power supply and distribution network by region, and the load rate of each region is monitored in real time. When the load rate of a certain region exceeds the preset overload threshold (e.g., 80%), the load transfer strategy of the adjacent region is triggered, and the transfer power threshold is calculated by the following formula. :

[0169] ;

[0170] in, For the area's rated capacity, The dynamic adjustment factor (obtained by training the inter-regional line impedance matrix) is preset to the industry's conventional load rate critical value (such as the heavy load standard defined in GB / T50598-2010).

[0171] As a further explanation of this embodiment, It is a dynamic adjustment factor, and ,in, This refers to the inter-regional line impedance. The average impedance of the entire network is used. The regional interconnection coefficient is determined by the topological connectivity of the distribution network and is a parameter used to quantify the degree of electrical connection between different areas in the tunnel group power supply and distribution network.

[0172] The regional interconnection coefficient is an indicator reflecting the "physical connection tightness" between two power supply and distribution areas (such as power supply areas in adjacent tunnel segments) based on network topology (such as line connection methods, distance, impedance, etc.). It is used to correct the load transfer power threshold to ensure that when a certain area is overloaded, the load transfer strategy of adjacent areas is implemented within the range of electrical safety and feasibility.

[0173] The value of the regional interconnection coefficient is determined by the following topological characteristics:

[0174] Line connection status: Whether the areas are directly connected by cables / buses, and the number of connected lines (e.g., single circuit / double circuit).

[0175] Line impedance: The impedance of the connecting lines between areas (the smaller the impedance, the tighter the electrical connection and the greater the interconnection factor).

[0176] Topological distance: The physical distance between regions in the power supply and distribution network (the closer the distance, the greater the interconnection coefficient is usually).

[0177] In the above formula for calculating the load transfer power threshold, the role of the regional interconnection coefficient is:

[0178] When inter-regional connections are tight (high interconnection coefficient), the actual power transfer threshold is lowered, preventing overload of connecting lines due to excessive power transfer. When inter-regional connections are loose (low interconnection coefficient), the power transfer threshold can be appropriately increased to improve resource utilization efficiency. The regional interconnection coefficient is a "quantitative index of connection tightness" based on the power supply and distribution network topology, used to dynamically balance the feasibility and electrical safety of load transfer.

[0179] S200.2. With the objective functions of minimizing grid losses, maximizing renewable energy consumption, and maximizing power supply reliability, a non-dominated sorting genetic algorithm is used for iterative optimization to generate coordinated control instructions that include transformer tap adjustment levels (adjustment step size ≤ 1.25% of rated voltage), the number of capacitor switching groups (single switching capacity difference ≤ 500kvar), and the execution sequence of load transfer strategies (interval between adjacent areas ≥ 10 seconds). The coordinated control instructions are approximate Pareto front solutions, which are selected and executed by the edge computing and AI decision-making unit 100 based on the local real-time status.

[0180] As a further explanation of this embodiment, the objective function in this embodiment specifically includes minimizing grid losses, maximizing renewable energy absorption, and maximizing power supply reliability, wherein:

[0181] Minimize grid losses Based on Kirchhoff's laws and Ohm's law, a network-wide loss calculation model is established:

[0182] ;

[0183] in, Indicates the first The real-time current of each line (uploaded by the edge computing and AI decision-making unit 100 with a sampling period of 100ms, obtained via MQTT protocol, and the data format conforms to the IEC61850 standard). Indicates the first The resistance value of the line (preset based on cable type and length, such as YJV22-8.7 / 10kV-3×240mm). 2 The cable resistance is 0.072 Ω / km, and its length is determined by the tunnel power distribution network topology.

[0184] Maximizing the absorption of new energy Quantitative Photovoltaics With wind power Grid-connected power output:

[0185] ;

[0186] in, This indicates the real-time output of the distributed photovoltaic power station in the tunnel cluster (monitored by the photovoltaic inverter, with a sampling period of 100ms, and uploaded via the Modbus protocol). This indicates the real-time power output of the wind farms surrounding the tunnel complex (provided by the wind farm's SCADA system and synchronized to the cloud platform via a 5G network).

[0187] Maximize power supply reliability Reliability metrics based on component failure rate and repair time:

[0188] ;

[0189] in, Indicates the first The failure rate of each component (refer to DL / T837-2012 "Code for Reliability Evaluation of Power Transmission and Transformation Facilities", such as the failure rate of 10kV dry-type transformer is 0.05 times / year, and the failure rate of 10kV circuit breaker is 0.08 times / year). Indicates the first The average repair time for each component (refer to DL / T837-2012, the repair time for circuit breaker faults is 4 hours, and the repair time for transformer faults is 8 hours).

[0190] Furthermore, the non-dominated sorting genetic algorithm in this embodiment iteratively optimizes the objective function, specifically including the following process:

[0191] First, an initial population of 100 is constructed. Each individual is mapped to power distribution control parameters through a real-number encoding (e.g., transformer tap position encoding is [-5, 5], corresponding to a ±2.5% rated voltage adjustment range, with a step size of 1.25% of the rated voltage). To avoid the initial solution being concentrated in a local area, Latin hypercube sampling is used to generate the population. Layered random sampling evenly covers the parameter space (e.g., combinations of tap positions and capacitor switching groups), ensuring the algorithm's global exploration capability for multiple objectives such as "grid loss, renewable energy consumption, and power supply reliability."

[0192] Subsequently, non-dominant relationship discrimination and crowding distance calculation are performed, where:

[0193] Domination relationship determination: Traverse the population individuals and screen the non-dominated solutions that are "not inferior to other individuals in all objective functions and are better in at least one objective function". These solutions are marked as Pareto level 1 (inferior solutions are marked as higher levels) to achieve preliminary screening under multi-objective conflict.

[0194] Crowding distance calculation: For each objective function (such as grid loss, renewable energy consumption), calculate the difference between adjacent individuals on the function value, quantify the "sparseness" of individuals in the objective space - retain solutions in sparse regions, avoid excessive aggregation of high-quality solutions, and ensure the balanced coverage of the solution set for multiple objectives;

[0195] Next, to simulate the recombination and mutation of biological genes, the algorithm drives population evolution through crossover and mutation operations:

[0196] Simulated binary crossover (probability 0.8, distribution index 15): For selected parent individuals, the "dispersion" of gene exchange is controlled by a distribution index of 15 (the larger the index, the closer the new solution is to the parent), while preserving the superior characteristics of the parent and exploring the neighborhood parameter space.

[0197] Multinomial mutation (probability 0.1, distribution exponent 20): The mutation amplitude is constrained by a distribution exponent of 20 (the larger the exponent, the smoother the mutation), which avoids destroying the local structure of high-quality solutions, and at the same time introduces global exploration capability to the algorithm to prevent getting trapped in local optima;

[0198] Finally, set up a double termination rule:

[0199] Maximum number of iterations (200): Matches the computational efficiency requirements of engineering scenarios (single iteration takes ≤3 seconds under GPU acceleration, total time for 200 iterations ≤10 minutes, meeting the hourly optimization cycle of cloud platforms);

[0200] Convergence determination (rate of change of optimal solution <5% over 20 consecutive iterations): When the rate of change of the optimal solution of the Pareto front on the objective function value is <5% over 20 consecutive iterations, the solution set is determined to be converged, the iteration is terminated early, and invalid calculations are avoided.

[0201] Through the above process, the non-dominated sorting genetic algorithm continuously generates a set of control strategies covering multiple objectives in the dynamic balance of "global exploration-local optimization-convergence determination", providing sufficient optimization scheme support for the decision output of the cloud platform collaborative management unit 200.

[0202] It should be added that after the algorithm iterations are complete, the top 20 optimal solutions are selected by crowding distance to form an approximate Pareto front solution set. Each solution contains:

[0203] Transformer tap adjustment parameters: such as adjustment range +2 (corresponds to a voltage increase of 2.5% of the rated value);

[0204] Capacitor switching parameters: e.g., 3 groups are switched on (compensation capacity 300kvar);

[0205] Load transfer strategy: For example, if area 1 transfers 2 MVA to area 2 at t=0s, area 2 will perform the receiving action at t=10s.

[0206] In summary, the edge computing and AI decision-making unit 100 selects execution strategies from the solution set based on local real-time status (such as equipment temperature rise and fault warning level): when the local safety margin coefficient is <0.6, the strategy with the least impact on equipment load is selected first; when the system is in normal operating condition, the strategy with the greatest reduction in power grid loss is selected, so as to achieve coordinated control of global optimization and local safety.

[0207] The intelligent sensing and IoT unit 300 is used to monitor the operating parameters of the tunnel power supply and distribution equipment, the cabin environment and fire status in real time, and uses the multi-protocol communication link module 310 to transmit the data to the edge computing and AI decision-making unit 100 and the cloud platform collaborative management unit 200, so as to build a full-scene monitoring system through the full-scene data fusion analysis module 320.

[0208] In this embodiment, the multi-protocol communication link module 310 includes an industrial bus sub-link sub-module 311, a wireless sensing sub-link sub-module 312, and a 5G core sub-link sub-module 313, wherein:

[0209] The Industrial Bus Sub-Link Sub-Module 311 uses the Modbus protocol to build a dedicated communication channel for power equipment, and performs millisecond-level periodic acquisition of voltage, current and temperature parameters of transformers and switchgear in the tunnel, adapting to the strong real-time monitoring requirements of power equipment.

[0210] As a further explanation of this embodiment, the industrial bus sub-link sub-module 311 in this embodiment uses a Siemens S7-1200 PLC as the main controller and connects to the transformer and switch cabinet in the tunnel through the Profibus DP bus; it is also equipped with a MOXAMGateMB3170 protocol converter to realize the conversion between Modbus RTU and TCP protocols, ensuring a 10ms response latency.

[0211] Wireless Sensing Sub-Link Sub-Module 312: Based on the LoRaWAN protocol, it builds an environmental monitoring network, collects temperature, humidity and smoke concentration parameters, optimizes node power consumption through a sleep-wake low-power communication mechanism, and is suitable for tunnel cabling-restricted scenarios.

[0212] As a further explanation of this embodiment, the wireless sensing sub-link sub-module 312 in this embodiment integrates a LoRaWAN wireless mesh network, the gateway adopts Dragino LG01-N, the terminal node is based on an STM32L071 microcontroller, and integrates an SHT31-D temperature and humidity sensor and an MQ-2 smoke sensor.

[0213] Furthermore, in this embodiment, the terminal node adopts a sleep-wake scheduling mechanism (wake-up cycle of 5 minutes), and only collects temperature and humidity (5-minute sampling cycle) and smoke concentration (2-minute sampling cycle) during the wake-up period; the gateway performs data aggregation every 30 minutes and transmits it to the RAK2245 receiving module of the edge computing unit through the LoRa link, reducing network power consumption to the μW level.

[0214] The 5G core sub-link sub-module 313 adopts a 5G URLLC slicing channel to perform real-time priority transmission of fire alarm and gas concentration data, and the transmission latency meets the requirements of fire emergency response standards.

[0215] As a further explanation of this embodiment, this embodiment can establish a URLLC slicing channel based on Huawei 5GCPEPro2, and the edge computing gateway can be Intel NUC11, which locally caches fire alarm and gas concentration data.

[0216] Furthermore, the priority rules for transmitting fire alarm and gas concentration data in this embodiment are as follows:

[0217] Fire alarm data: marked with QoS level 5 (highest priority), directly transmitted to the cloud platform collaborative management unit 200 via UDP protocol, ensuring that the end-to-end latency from sensor triggering to cloud reception is ≤50ms (compliant with the emergency response requirements of GB50116-2013 "Code for Design of Automatic Fire Alarm Systems").

[0218] Gas concentration data: Marked as QoS level 4, transmitted via TCP protocol to ensure data integrity.

[0219] The full-scenario data fusion and analysis module 320 includes a multimodal data fusion submodule 321, a device health assessment submodule 322, and an adaptive sampling control submodule 323, wherein:

[0220] The multimodal data fusion submodule 321 is used to align electrical parameters (current, voltage), environmental parameters (temperature and humidity), and fire protection parameters (smoke concentration) in a spatiotemporal dimension, and uses a Kalman filter algorithm to eliminate data noise. The fusion accuracy meets the standards for power equipment condition monitoring.

[0221] As a further explanation of this embodiment, when performing spatiotemporal alignment, the NTP network time protocol is used to synchronize the timestamps of each sensor. For data with different sampling rates (such as 10ms power parameters and 50ms oil temperature data), a cubic spline interpolation algorithm is used to fit them into a unified 10ms time series, thereby achieving spatiotemporal alignment of electrical, environmental, and fire protection parameters.

[0222] Meanwhile, the calculation formula for the Kalman filter algorithm in this embodiment is:

[0223] State prediction: ;in, Let be the state transition matrix, and , For process noise, covariance ; for Prior state estimation at time t; for Posterior state estimation at time 1;

[0224] Observation Update: ;in, Observation matrix , For Kalman gain, For the observed values, the measurement noise covariance R = 0.1; for The system state at any given moment;

[0225] In practice, the current state is first predicted by the state equation, and then the optimal estimate is obtained by combining the observed values. The noise of the current data is suppressed from ±5% to ±1%, and the noise of the oil temperature data is suppressed from ±0.5℃ to ±0.1℃. The fused data is transmitted to the edge computing and AI decision unit 100 via the SPI bus.

[0226] The equipment health assessment submodule 322 constructs a fault diagnosis model based on DS evidence theory, extracts the circuit breaker opening and closing coil current waveform and transformer oil temperature change rate as core feature quantities, and realizes early fault warning of equipment.

[0227] As a further explanation of this embodiment, the equipment health assessment submodule 322 constructs a fault diagnosis model based on DS evidence theory, which integrates multi-source features. It interacts in real time with the multimodal data fusion submodule 321 to obtain spatiotemporally aligned electrical and temperature data, thereby enabling early fault warnings for circuit breakers and transformers. The specific process is as follows:

[0228] First, establish data interaction with the multimodal data fusion submodule 321: receive time-space aligned monitoring data via the SPI bus, including circuit breaker opening and closing coil current waveforms (transmitted via industrial bus sub-link submodule 311) and transformer oil temperature timing data.

[0229] Simultaneously, feature parameters are extracted in real time:

[0230] For circuit breakers, analyze the peak value, rise time, and fall time of the coil current waveform, and calculate the waveform distortion rate (distortion rate = (actual waveform - standard waveform mean square error) / standard waveform energy).

[0231] For transformers, the oil temperature change rate is calculated using a 30-minute sliding window. The calculation formula is as follows:

[0232] ;

[0233] in, Indicates the rate of change of oil temperature; Indicates time The oil temperature; Indicates time The oil temperature 30 minutes prior; indicates; and is associated with the concurrent load rate (obtained from the multimodal data fusion submodule 321).

[0234] Subsequently, the feature parameters are mapped to the basic probability of failure:

[0235] On the circuit breaker side: when the waveform distortion rate is >15%, the relationship between the distortion degree and the fault probability is fitted using a Gaussian mixture model to output the basic fault probability. (Fault) (The higher the distortion rate, the better) Linearly increasing, under normal conditions (approaching 0)

[0236] Transformer side: Based on the oil temperature change rate and load rate, the basic probability of fault is mapped using the Sigmoid function. (Fault) (The higher the load rate, the faster the temperature rises.) The exponential rate increases under normal conditions. (approaching 0)

[0237] Then, the Dempster combination rule is used to fuse multi-source evidence to calculate the comprehensive probability of the target state (normal, warning, fault):

[0238] ;

[0239] Mole: Traversing the set of all states , The sum of their intersections is Product of evidence (e.g.) =Fault, When the intersection of the two faults is equal to the fault, it is included in the calculation. calculate;

[0240] Denominator: Conflict coefficient (quantifies the degree of contradiction between two pieces of evidence). If the conflict coefficient is greater than 0.5, the Yager rule softening calculation is automatically switched (reducing the weight of contradictory evidence) to avoid extreme conflicts from causing decision distortion.

[0241] Finally, execute the failure decision:

[0242] After fusion When the GPIO port outputs a high level, it triggers the interrupt response of the edge computing and AI decision-making unit 100;

[0243] Synchronous marking of fault type (coil malfunction / sudden oil temperature change), via I 2 The C-bus transmits the diagnostic results back to the multimodal data fusion submodule 321 for dynamic correction of the subsequent data denoising model;

[0244] like Only the warning log is triggered, which is then used by operations and maintenance personnel for retrospective analysis.

[0245] After receiving the abnormal operating condition signal sent by the edge computing and AI decision-making unit 100, the adaptive sampling control submodule 323 automatically adjusts the sampling frequency of key equipment and constructs a three-level sampling mechanism of normal cycle acquisition, abnormal trigger acquisition and high-frequency tracking acquisition to meet the dynamic monitoring accuracy requirements under different operating conditions.

[0246] As a further explanation of this embodiment, the three-level sampling parameters and triggering conditions in this embodiment are as follows:

[0247] Normal data collection (default mode):

[0248] Power parameters (transformer voltage / current, switchgear bus current): 100ms sampling period (configured via Modbus register 40001, transmitted via industrial bus sub-link sub-module 311); Environmental parameters (temperature, humidity, smoke concentration): 5-minute sampling period (configured via LoRa terminal node register 0x0010, transmitted via wireless sensor sub-link sub-module 312).

[0249] Triggering condition: The device health assessment submodule 322 outputs a normal signal ( ).

[0250] Abnormal Trigger Acquisition (Early Warning Response): Power Parameters: Sampling period compressed to 50ms (notified to the S7-1200 main controller via UART command (baud rate 9600); Oil Temperature / Contact Temperature: Sampling period compressed to 100ms (notified to the S7-1200 main controller via I... 2 (C bus rewriting sensor configuration register);

[0251] Triggering condition: The equipment health assessment submodule 322 outputs a warning signal ( ), or an anomaly flag indicating that the load rate of the edge computing and AI decision-making unit 100 is >80%.

[0252] High-frequency tracking and acquisition (fault response): Full equipment parameters: power parameters are sampled at 10ms and environmental parameters at 100ms (received from the cloud platform temporary configuration instructions via 5G core sub-link sub-module 313); Continuous logic: after triggering, it lasts for 1 minute and automatically falls back to normal acquisition via a hardware timer.

[0253] Triggering condition: Equipment health assessment submodule 322 outputs a fault signal ( (or the fire alarm submodule detects a smoke concentration >500ppm).

[0254] The modular prefabricated cabin unit 400 integrates power supply and distribution equipment, fire protection system and edge computing components. It adopts a separate steel structure design for high-pressure cabin and low-pressure cabin, combined with fireproof and heat insulation technology to achieve physical isolation, rapid installation and environmental protection.

[0255] In this embodiment, the modular prefabricated cabin unit 400 includes a high-pressure cabin module 410, a low-pressure cabin module 420, and a fire-fighting integrated module 430, wherein:

[0256] The high-voltage compartment module 410 includes a switchgear submodule 411 and a transformer submodule 412, wherein:

[0257] Switchgear module 411 houses a 10kV metal-clad withdrawable switchgear with a Q355B steel plate frame. An arc flash protection sensor is installed inside the cabinet; when the arc flash energy reaches a preset threshold, the circuit breaker trips within 10ms, and a fault signal is uploaded via the Modbus protocol. Switchgear module 411 is horizontally positioned on the left side of the high-voltage compartment (800mm from the front panel). Its bottom surface is welded to the compartment floor plate using L-shaped angle steel (75mm×75mm×8mm), with a 5mm rubber shock-absorbing pad installed between the angle steel and the cabinet. The top of the cabinet is fixed to the compartment top with M10 bolts to ensure mechanical stability during operation.

[0258] Transformer submodule 412 is a dry-type transformer with an outer casing having a protection rating of at least IP54. The cooling fan and temperature controller are linked for operation. The device is internally equipped with a temperature controller and cooling fan; when the winding temperature reaches a preset threshold, the fan automatically starts to dissipate heat and uploads the temperature status via a communication interface. Transformer submodule 412 is vertically arranged on the right side of the high-voltage compartment (1.5m from the switchgear). The bottom is constructed with a 1.2m long base welded from #10 channel steel, and the transformer is fixed to the base with M12 bolts. A 300mm heat dissipation channel is provided between the outer casing and the compartment side panels, and louvers are installed on the top to ensure efficient heat dissipation.

[0259] The low-pressure chamber module 420 includes an intelligent power distribution module 421 and an edge computing server submodule 422, wherein:

[0260] The intelligent power distribution module 421 integrates a molded case circuit breaker and an intelligent power meter, supporting Modbus protocol communication. It enables power parameter acquisition and circuit breaker status monitoring, and responds to opening and closing control commands from the edge computing unit. Located at the front of the low-pressure compartment (500mm from the front panel), the intelligent power distribution module 421 is supported by a 200mm×100mm×5mm rectangular steel pipe welded bracket. The terminal box is fixed to the bracket with four M8 bolts at the corners. The bracket is connected to the compartment floor using M10 expansion bolts (300mm spacing) to ensure a secure installation.

[0261] The edge computing server submodule 422 is an industrial-grade server, installed with a shock-absorbing bracket, and adapted to a wide temperature operating environment. The edge computing server submodule 422 is vertically installed on a 4U standard guide rail at the rear of the low-pressure compartment (1.2m from the bottom plate), and is adapted to industrial standard installation through guide rail sliders. A shock-absorbing bracket is added to the bottom, and the bracket is fixed to the guide rail with M6 bolts. A pressure plate is set on the top of the server to limit the top of the compartment, which is adapted to the vibration environment.

[0262] The fire protection integrated module 430 includes a fire extinguishing device submodule 431 and a fire detection submodule 432, wherein:

[0263] The fire extinguishing device submodule 431 integrates a heptafluoropropane gas fire extinguishing system. The nozzles are arranged at preset intervals and have a linkage triggering mechanism. Upon receiving a fire detection signal, it delays the release of the agent and triggers an audible and visual alarm, providing status feedback via dry contacts. The nozzles are evenly spaced at 2.5m intervals on the top of the chamber (50mm from the top), and are suspended via DN25 galvanized steel pipes—the pipe supports are spaced 1.5m apart (fixed with expansion bolts). The steel pipes and nozzles are connected by threads and sealed to ensure uniform agent coverage.

[0264] The fire detection submodule 432 employs a smoke and heat composite detector, with detection sensitivity meeting relevant national standards. It collects smoke concentration and temperature data in real time, triggering an audible and visual alarm when the alarm threshold is reached, and activating the fire extinguishing system. The alarm signal is uploaded to the cloud platform collaborative management unit 200 via the 5G core sub-link submodule 313, triggering a remote early warning. The smoke and heat detectors are ceiling-mounted in the center of the cabin ceiling (100mm from the top), with a spacing of ≤10m. The base is secured with M6 expansion bolts, and the detector body is snap-fitted to the base. The wiring ports are protected with waterproof connectors.

[0265] As a further explanation of this embodiment, three sets of Φ100mm epoxy resin through-wall sleeves are installed at the bottom of the high-pressure compartment and low-pressure compartment partition (200mm from the bottom plate), filled with fireproof mud. After the cables pass through, they are sealed with sealant to meet fire prevention and protection requirements. In addition, heat dissipation holes (600mm×400mm, 1.5m from the ground) are opened on the rear side plate of the transformer compartment, and metal protective nets and axial flow fans are installed. The fans are fixed by welding with 50mm×50mm×5mm angle steel brackets to ensure reliable operation of the heat dissipation system.

[0266] The automated adaptive control unit 500 is used to execute the final control instructions of the edge computing and AI decision-making unit 100. Through the multi-system linkage logic of power supply and distribution circuit switching, micro-environment control and decision and control module 530, it realizes real-time power flow monitoring and emergency response to sudden events.

[0267] In this embodiment, the automated adaptive control unit 500 includes an intelligent power distribution execution module 510 and a microenvironment control module 520, wherein:

[0268] The intelligent power distribution execution module 510 uses solid-state circuit breakers to construct an intelligent power distribution network. The intelligent power distribution execution module 510 includes a fault isolation submodule 511 and a dynamic load distribution submodule 512, wherein:

[0269] The fault isolation submodule 511 is based on a high-speed current sensor and a solid-state circuit breaker to achieve rapid isolation of short-circuit faults.

[0270] As a further explanation of this embodiment, the high-speed current sensor in this embodiment collects the three-phase current in real time and calculates the waveform distortion rate through the FPGA; when the current exceeds 5 times the rated value or the distortion rate exceeds the threshold, the FPGA sends a trip command to the circuit breaker, and uploads the fault type to the edge computing and AI decision unit 100 through the Modbus / TCP protocol; at the same time, the fault signal triggers the circuit breaker to trip through hard wiring, and feeds back the action status to the intelligent power distribution terminal through the RS485 interface to ensure that the fault isolation is completed within 2ms.

[0271] The dynamic load distribution submodule 512 dynamically adjusts the transformer tap position through a fuzzy PID controller based on the load prediction results of the edge computing and AI decision unit 100.

[0272] As a further explanation of this embodiment, the specific process of dynamically adjusting the transformer tap position using a fuzzy PID controller is as follows:

[0273] First, load forecast data (updated every 1 minute) output by the edge computing and AI decision-making unit 100 is collected as the basis for load regulation decisions.

[0274] Subsequently, based on the collected load data, the following two core inputs are calculated:

[0275] Load forecast deviation: The difference between the actual load and the forecast load (describes "how large the forecast error is");

[0276] Deviation change rate: The rate at which the load forecast deviation changes (characterizing whether the error is increasing or decreasing);

[0277] Then, based on the preset fuzzy rules (example: if the deviation is "positive large" (actual load is much higher than the predicted value) and the rate of change is "positive small" (deviation growth slows down), the "+2 level" adjustment command is output), the number of tap adjustment steps is derived (the minimum adjustment unit is ±1 level, which directly determines the tap level change).

[0278] Meanwhile, the basic PID parameters (proportional coefficient) Integral coefficient Differential coefficients It is not fixed, but rather corrected in real time through fuzzy reasoning, allowing the controller to adapt to the dynamic changes in the load.

[0279] Finally, the final adjustment amount is calculated every 10 seconds to drive the tap changer motor: the motor step angle is 1.8° to ensure adjustment accuracy; the single adjustment step size is ≤1.25% of the rated voltage, achieving "low impact and fine granularity" voltage optimization.

[0280] The microenvironment control module 520 achieves adaptive adjustment of the tunnel microenvironment through an IoT edge controller, including a multi-parameter collaborative control submodule 521 and a zoned temperature equalization submodule 522, wherein:

[0281] The multi-parameter collaborative control submodule 521 dynamically adjusts the ventilation fan speed based on CO concentration, visibility, and traffic flow using a model predictive control algorithm.

[0282] The zone temperature balancing submodule 522 adjusts the power of the tunnel lighting system through a PID controller to achieve cabin temperature difference control. This design uses the temperature difference between tunnel zones as the controlled object. Distributed temperature sensors collect data in real time, and the PID controller dynamically adjusts the LED lighting power (adjustment range 20%-100%) based on the temperature difference deviation and its rate of change. The thermal effect of the lighting system is used to balance the cabin temperature, forming a closed-loop control of "temperature acquisition - deviation calculation - power adjustment" to ensure that the zone temperature difference is maintained within a reasonable range.

[0283] In this embodiment, the decision-making and control module 530 includes a collaborative control submodule 531 and an emergency response decision tree submodule 532, wherein:

[0284] The collaborative control submodule 531 constructs an inter-system communication mechanism based on an event bus architecture, specifically including:

[0285] Assign dynamic priorities to systems such as fire protection, ventilation, and lighting (e.g., fire protection system priority is the highest level 1) to ensure that the response time of critical systems meets the real-time requirements of fire emergency response;

[0286] The Raft consensus algorithm is used to achieve state synchronization between control units, with synchronization accuracy meeting industrial-grade distributed system collaborative control standards. The Raft consensus algorithm categorizes control nodes into leader, follower, and candidate states, and ensures the consistency and reliability of control commands across multiple nodes through log replication and election mechanisms. In this embodiment, command synchronization for systems such as fire protection, ventilation, and lighting is combined with event bus priority management (with fire protection systems having the highest priority) to ensure that the response time of critical systems meets emergency requirements, achieving multi-system collaborative control.

[0287] The emergency response decision tree submodule 532 constructs an emergency response model for sudden events based on a Bayesian network inference engine, specifically including:

[0288] A set of typical fault modes for the entire life cycle of a predefined tunnel power supply and distribution system (covering basic faults such as single-phase grounding and three-phase short circuits, as well as derivative composite faults).

[0289] By updating the fault probability distribution in real time through Bayesian inference, the optimal emergency response strategy is generated, and the timeliness of strategy generation meets the rapid response requirements for power system fault handling.

[0290] As a further explanation of this embodiment, this embodiment uses a Bayesian network inference engine to construct an emergency response model for sudden events: This model is based on typical fault modes (including single-phase grounding, three-phase short circuit, etc.) throughout the entire life cycle of the tunnel power supply and distribution system. It describes the correlation between fault characteristics and fault types through a preset conditional probability table, collects protection device signals and electrical characteristic quantities in real time, updates the fault probability distribution using Bayesian inference, and quickly generates the optimal emergency response strategy (such as the circuit breaker disconnection sequence). The timeliness of strategy generation meets the requirements for power system fault handling and ensures the safe operation of the system.

[0291] It should be noted that the intelligent power distribution execution module 510 in this embodiment adopts Modbus / TCP and RS485 interfaces, the microenvironment control module 520 supports CANopen and DALI protocols, and the decision and control module 530 implements event-driven communication based on message queues. At the hardware level, an industrial-grade controller and fieldbus architecture are used to ensure the real-time performance and reliability of the control logic. The overall design complies with relevant standards for power systems and industrial automation, and supports rapid on-site deployment and commissioning.

[0292] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0293] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive control system for intelligent substation power supply and distribution based on edge computing modular prefabricated tunnel modules, characterized in that, include: The edge computing and AI decision-making unit (100) is used to realize the rapid acquisition, processing and real-time decision-making of multi-dimensional data of tunnel power supply and distribution. By deploying computing resources and machine learning algorithms at edge nodes close to the data source, it generates power supply and distribution adaptive control strategies and converts the strategies into executable control instructions. Based on the collaborative control instructions issued by the cloud platform collaborative management unit (200), it generates the final control instructions through a priority determination mechanism. The edge computing and AI decision-making unit (100) generates the final control instruction through a priority determination mechanism, including the following steps: S100.

1. The local real-time instructions and collaborative control instructions received by the edge computing and AI decision-making unit (100) are parsed, and key features for priority calculation are extracted. The key features include instruction type, target object, parameter value range and timeliness attribute. S100.

2. Based on device status and command attributes, calculate the local command priority value using a two-factor weighting model. Priority value of coordinated control instructions The calculation formula is: ； in, Indicates the urgency factor; Indicates the current temperature rise of the equipment; This represents the average temperature rise under historical load conditions, used to reflect the degree of equipment abnormality. ； in, Indicates the optimization factor; This represents the difference between the real-time load rate and the rated load rate. This indicates the rated load rate, reflecting the urgency of system regulation. S100.3, Based on local instruction priority value Priority value of coordinated control instructions The calculation results are used to execute hierarchical decision-making logic; The cloud platform collaborative management unit (200) constructs a global data storage and analysis center to receive data from edge computing nodes and provide a global perspective on power grid operation. Through big data analysis and resource optimization algorithms, it issues collaborative control instructions to the edge computing and AI decision-making unit (100) to assist in optimizing local decision-making. The intelligent sensing and IoT unit (300) is used to monitor the operating parameters of the tunnel power supply and distribution equipment, the cabin environment and fire status in real time, and uses the multi-protocol communication link module (310) to transmit the data to the edge computing and AI decision-making unit (100) and the cloud platform collaborative management unit (200) to build a full-scene monitoring system through the full-scene data fusion analysis module (320); The modular prefabricated cabin unit (400) integrates power supply and distribution equipment, fire protection system and edge computing components. It adopts a separate steel structure design for high-pressure cabin and low-pressure cabin, combined with fireproof and heat insulation technology to achieve physical isolation, rapid installation and environmental protection. The modular prefabricated cabin unit (400) includes a high-pressure cabin module (410), a low-pressure cabin module (420), and a fire protection integrated module (430), wherein: The high-voltage compartment module (410) includes a switch cabinet sub-module (411) and a transformer sub-module (412). The low-pressure chamber module (420) includes an intelligent power distribution module (421) and an edge computing server submodule (422). The fire protection integrated module (430) includes a fire extinguishing device submodule (431) and a fire detection submodule (432). An automated adaptive control unit (500) is used to execute the final control instructions of the edge computing and AI decision unit (100). Through the multi-system linkage logic of power supply and distribution circuit switching, micro-environment control and decision and control module (530), it realizes real-time power flow monitoring and emergency response to sudden events.

2. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 1, characterized in that, The edge computing and AI decision-making unit (100) includes a data processing module (110), a policy generation module (120), and an instruction conversion module (130), wherein: The data processing module (110) is used to clean and normalize the raw data transmitted by the intelligent sensing and IoT unit (300), and to remove noise using a sliding window filtering algorithm, with the window width set to a preset value. The strategy generation module (120) generates an adaptive control strategy for power supply and distribution based on a multi-layer neural network. The input of the neural network is historical load data, and the output is the power prediction value for future periods. The instruction conversion module (130) is used to convert the control strategy into executable control instructions, adopting an industrial control standard protocol, and the instruction generation delay meets the real-time control requirements.

3. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel module according to claim 1, characterized in that, In step S100.3, the hierarchical decision logic specifically includes: First priority strategy: when When the instruction type is emergency, the edge computing and AI decision unit (100) generates the highest priority control instruction, and executes the device protection action through the automated adaptive control unit (500), with a threshold value of [missing information]. This is a threshold for equipment safety urgency, used to distinguish between normal operation and emergency failure scenarios, and the threshold... Equipment safety margin determination based on modular prefabricated cabin unit (400); Parameter fusion strategy: when Alternatively, if the instruction type is optimization, it performs a weighted fusion of the control parameters of local real-time instructions and collaborative control instructions: ； in, This indicates the final control parameters that are fused and sent to the automated adaptive control unit (500), and Meets the electrical safety boundaries and mechanical life constraints of the modular prefabricated cabin unit (400); This represents the local control parameters generated by the edge computing and AI decision-making unit (100); This refers to the cloud-based control parameters generated by the cloud platform collaborative management unit (200); Second priority strategy: when At that time, the edge computing and AI decision-making unit (100) prioritizes executing cloud-based collaborative instructions and adjusts system operating parameters and thresholds through the automated adaptive control unit (500). A global optimization priority threshold is used to balance local security and cloud energy efficiency goals, and the threshold... Determination of the global energy efficiency model based on the cloud platform collaborative management unit (200).

4. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 2, characterized in that, The edge computing and AI decision-making unit (100) further includes a policy verification module (140), a parameter optimization module (150), and a version management module (160), wherein: The strategy verification module (140) verifies the effectiveness of the control strategy through the Monte Carlo simulation algorithm. The number of simulations is preset to the industry standard value. Based on the device operation parameters collected by the intelligent sensing and IoT unit (300), an evaluation report containing the risk probability distribution is generated. The evaluation report is used to correct the priority judgment threshold parameter built into the edge computing and AI decision unit (100). The parameter optimization module (150) dynamically adjusts the strategy parameters based on the deep reinforcement learning framework, with the goal of minimizing grid loss. The optimization cycle is dynamically issued by the cloud platform collaborative management unit (200) according to the load fluctuation characteristics, and directly outputs the dual-factor weight parameters used for priority calculation in the edge computing and AI decision unit (100). The version management module (160) uses blockchain hash storage technology to record the historical version of the strategy, and supports one-click rollback to the previous valid version in case of an anomaly. The rollback time meets the equipment fault tolerance requirements of the modular prefabricated cabin unit (400). The rollback trigger condition is linked with the security verification result of the edge computing and AI decision unit (100).

5. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 1, characterized in that, The global data storage and analysis center of the cloud platform collaborative management unit (200) adopts a distributed time-series database cluster architecture to perform millisecond-level sampling and storage of equipment operation data uploaded by the edge computing and AI decision-making unit (100) and realize global visualization of the power grid operation status. The distributed time-series database cluster architecture specifically includes: A digital twin is constructed based on a 3D tunnel topology model and a power flow algorithm to render a load distribution heat map of the power supply and distribution network in real time. By using the Long Short-Term Memory (LSTM) network to predict trends in historical data, a risk warning map of power grid operation for the next 24 hours is generated.

6. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel module according to claim 5, characterized in that, The cloud platform collaborative management unit (200) uses big data analysis and resource optimization algorithms to generate collaborative control instructions, including the following steps: S200.

1. A consistent hashing algorithm is used to manage the tunnel group power distribution network by region, and the load rate of each region is monitored in real time. When the load rate of a certain region exceeds the preset overload threshold of the rated value, the load transfer strategy of the adjacent region is triggered, and the transfer power threshold is calculated by the following formula. : ； in, For the rated capacity of the area, As a dynamic adjustment factor, the preset overload threshold is the industry's conventional load rate critical value; the regional interconnection coefficient is determined by the topological connectivity of the power distribution network and is a parameter used to quantify the degree of electrical connection between different regions in the tunnel group power distribution network. S200.

2. Taking the minimization of power grid loss, the maximization of new energy consumption, and the maximization of power supply reliability as objective functions, a non-dominated sorting genetic algorithm is used for iterative optimization to generate a coordinated control instruction that includes the transformer tap adjustment level, the number of capacitor switching groups, and the execution sequence of the load transfer strategy. The coordinated control instruction is an approximate Pareto front solution set, which is selected and executed by the edge computing and AI decision unit (100) based on the local real-time state.

7. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 1, characterized in that, The multi-protocol communication link module (310) includes an industrial bus sub-link sub-module (311), a wireless sensing sub-link sub-module (312), and a 5G core sub-link sub-module (313), wherein: The industrial bus sub-link sub-module (311) uses the Modbus protocol to build a dedicated communication channel for power equipment, and performs millisecond-level periodic acquisition of voltage, current and temperature parameters of transformers and switchgear in the tunnel, adapting to the strong real-time monitoring requirements of power equipment. The wireless sensing sub-link sub-module (312) builds an environmental monitoring network based on the LoRaWAN protocol, collects temperature, humidity and smoke concentration parameters, optimizes node energy consumption through a sleep wake-up low-power communication mechanism, and adapts to tunnel cabling restricted scenarios. The 5G core sub-link sub-module (313) adopts a 5G URLLC slicing channel to perform real-time priority transmission of fire alarm and gas concentration data; The full-scene data fusion and analysis module (320) includes a multimodal data fusion submodule (321), an equipment health assessment submodule (322), and an adaptive sampling control submodule (323), wherein: The multimodal data fusion submodule (321) is used to align electrical parameters, environmental parameters, and fire protection parameters in a spatiotemporal dimension, and to use a Kalman filter algorithm to eliminate data noise. The equipment health assessment submodule (322) constructs a fault diagnosis model based on DS evidence theory, extracts the circuit breaker opening and closing coil current waveform and transformer oil temperature change rate as core feature quantities, and realizes early fault warning of equipment. After receiving the abnormal operating condition signal sent by the edge computing and AI decision-making unit (100), the adaptive sampling control submodule (323) automatically adjusts the sampling frequency of key equipment and constructs a three-level sampling mechanism of normal cycle sampling, abnormal trigger sampling, and high-frequency tracking sampling.

8. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 1, characterized in that, In the high-pressure chamber module (410): The switch cabinet sub-module (411) has a built-in 10kV metal armored withdrawable switch cabinet with a Q355B steel plate frame and an arc protection sensor installed inside the cabinet. The transformer submodule (412) adopts a dry-type transformer, and the outer shell adopts a structure with a protection level of not less than IP54. The cooling fan and the temperature controller are linked for control. In the low-pressure chamber module (420): The intelligent power distribution module (421) integrates a molded case circuit breaker and an intelligent power meter, and supports Modbus protocol communication; The edge computing server submodule (422) adopts an industrial-grade server, is installed with a shock-absorbing bracket, and is adapted to a wide temperature operating environment; In the fire protection integrated module (430): The fire extinguishing device submodule (431) integrates a heptafluoropropane gas fire extinguishing system, with nozzles arranged at preset intervals and equipped with a linkage triggering mechanism; The fire detection submodule (432) adopts a smoke and heat sensing composite detector.

9. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 2, characterized in that: The automated adaptive control unit (500) includes an intelligent power distribution execution module (510) and a micro-environment control module (520), wherein: The intelligent power distribution execution module (510) uses solid-state circuit breakers to construct an intelligent power distribution network. The intelligent power distribution execution module (510) includes a fault isolation submodule (511) and a dynamic load distribution submodule (512), wherein: The fault isolation submodule (511) is based on a high-speed current sensor and a solid-state circuit breaker to achieve rapid isolation of short-circuit faults. The dynamic load distribution submodule (512) dynamically adjusts the transformer tap position through a fuzzy PID controller based on the load prediction results of the edge computing and AI decision unit (100). The microenvironment control module (520) achieves adaptive adjustment of the tunnel microenvironment through an IoT edge controller, including a multi-parameter collaborative control submodule (521) and a zoned temperature equalization submodule (522), wherein: The multi-parameter collaborative control submodule (521) dynamically adjusts the speed of the ventilation fan based on CO concentration, visibility and traffic flow using a model predictive control algorithm. The partition temperature equalization submodule (522) adjusts the power of the tunnel lighting system through a PID controller to achieve cabin temperature difference control.

10. The adaptive control system for intelligent substation power supply and distribution based on edge computing in a modular prefabricated tunnel as described in claim 1, characterized in that: The decision and control module (530) includes a collaborative control submodule (531) and an emergency response decision tree submodule (532), wherein: The collaborative control submodule (531) is based on an event bus architecture to build an inter-system communication mechanism, specifically including: Assign dynamic priorities to fire protection, ventilation, and lighting systems to ensure that the response time of critical systems meets the real-time requirements of fire emergency response; The Raft consensus algorithm is used to achieve state synchronization between control units, and the synchronization accuracy meets the standards for collaborative control of industrial-grade distributed systems. The emergency response decision tree submodule (532) constructs an emergency response model for sudden events based on a Bayesian network inference engine, specifically including: A predefined set of typical failure modes throughout the entire lifecycle of a tunnel power supply and distribution system; By updating the fault probability distribution in real time through Bayesian inference, the optimal emergency response strategy is generated, and the timeliness of strategy generation meets the rapid response requirements for power system fault handling.

Images