Rail transit power supply digital twin system and construction method

Abstract

The invention provides a rail transit power supply digital twin system and a construction method, and relates to the technical field of data processing, and the method comprises the steps: constructing an urban rail transit power supply system digital twin model, achieving the digital virtualization associated with a physical entity, completing the model rendering, and displaying the state of the entity through a virtual model; according to the entity state, operating parameters of a substation and key equipment of a contact network are collected through a sensor and a hardware device; the operation parameters are cached by the station-level monitoring device body through the double-ring network architecture and then scheduled to a cloud platform communication ring network node, a data transmission path is optimized based on a load balancing strategy, and the data are uploaded to a cloud platform in real time. The management level can be improved, the operation and maintenance cost can be reduced, and a foundation is laid for intelligent operation and maintenance of a rail transit power supply system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rail transit, in particular to a rail transit power supply digital twin system and a construction method. BACKGROUND

[0002] The rail transit power supply system is the core infrastructure to ensure the safe operation of trains, with characteristics such as multiple types of equipment, wide distribution, and complex operating environment. The traditional operation and maintenance mode relies on manual inspection and post-processing, which is difficult to meet the high reliability and high efficiency of operation and maintenance. Digital twin technology can significantly improve the intelligent management level of the power supply system by constructing a virtual mirror of the physical system, combining real-time data driving and simulation analysis. The digital twin technology for the power supply system has been widely applied in the national power grid industry, but its application in the rail transit industry faces the following unique difficulties:

[0003] (1) Equipment heterogeneity challenge: The rail transit power supply system covers a strong coupling system of contact net dynamic current collection, mobile train load, and fixed substation equipment. Compared with the static power transmission network of the power grid, it needs to handle the transient characteristic modeling of the contact net-train sliding contact.

[0004] (2) Data acquisition complexity: The contact net is distributed along the entire track and there are strong interference environments such as bow net arc, which makes it difficult to ensure the real-time collection accuracy of key parameters such as vibration and temperature.

[0005] (3) Model dynamic synchronization requirement: The dynamic switching of the power supply section caused by train operation requires the digital twin model to have millisecond-level topology reconstruction capability, while the digital twin of the power grid is usually based on a steady-state network architecture.

[0006] (4) Three-dimensional space constraint: Three-dimensional visualization and fault location of equipment state in underground tunnels, elevated sections, and other three-dimensional space scenarios require centimeter-level spatial coordinate mapping.

[0007] (5) Safety redundancy mechanism: The non-interruptibility of train power supply requires the digital twin system to have sub-second fault self-healing strategy generation capability, far exceeding the power grid's minute-level fault handling standard. SUMMARY

[0008] The technical problem to be solved by the present application is to provide a rail transit power supply digital twin system and a construction method to improve management level and reduce operation and maintenance cost, and to lay a foundation for realizing intelligent operation and maintenance of the rail transit power supply system.

[0009] To solve the above technical problems, the technical solution of the present application is as follows:

[0010] In a first aspect, a rail transit power supply digital twin construction method is provided, the method comprising:

[0011] S1: By constructing a digital twin model of the urban rail transit power supply system, a digital virtual associated with the physical entity is realized, the model rendering is completed, and the state of the entity is displayed by the virtual model;

[0012] S2: According to the state of the entity, the operating parameters of the key equipment of the substation and the catenary are collected by sensors and hardware devices;

[0013] S3: The operating parameters are cached by the station-level monitoring device through a dual-ring network architecture, and then dispatched to the cloud platform communication ring network node. Based on the load balancing strategy, the data transmission path is optimized, and the data is uploaded to the cloud platform in real time;

[0014] S4: The data uploaded to the cloud platform is stored in a hierarchical manner, and a structured index is established to dynamically associate the digital twin model with the corresponding stored data to obtain the stored data;

[0015] S5: According to the stored data, logical operations and threshold judgments are performed on the equipment operating data to generate real-time alarm information. The health status of the equipment is evaluated through statistical analysis and machine learning algorithms to predict the remaining life and maintenance cycle, dynamically optimize the operation and maintenance strategy, and obtain proactive maintenance suggestions and fault handling plans;

[0016] S6: Based on the maintenance suggestions and fault handling plans, the dynamic presentation of the substation layout and catenary of the entire line based on the three-dimensional map, real-time monitoring of the equipment operating state and three-dimensional precise positioning of the fault, proactive maintenance strategy generation and fault alarm information pushing, integrated display of monitoring video and three-dimensional model, and panoramic roaming function are realized.

[0017] Further, by constructing a digital twin model of the urban rail transit power supply system, a digital virtual associated with the physical entity is realized, the model rendering is completed, and the state of the entity is displayed by the virtual model, including:

[0018] S1.1: By the physical entity parameters of the rail transit power supply system, a full-system digital twin model covering from the 110kV main substation inlet side to the low-voltage load end is constructed. The digital twin model is rendered in real time by a rendering engine and dynamically synchronized with the operating state of the physical entity;

[0019] S1.2: According to the operating state, the model construction precision is not less than LOD3.0, and according to the detailed level defined in the ISO19650 standard, the substation equipment layout, catenary support suspension parts and wire parameters in the virtual model are consistent with the actual height;

[0020] S1.3: By the actual height consistency, the operating data collected by the physical entity is associated with the digital twin model, the equipment state is mapped in real time through the three-dimensional visualization interface, and the model rendering details can be adjusted to obtain the virtual model to display the state of the entity.

[0021] Further, the operation parameters pass through the dual-ring network architecture to cache the data in the station-level monitoring device body, and then dispatch to the cloud platform communication ring network node. Based on the load balancing strategy, the data transmission path is optimized, and the data is uploaded to the cloud platform in real time, including:

[0022] S3.1: According to the operation parameters, a dynamic cache area is set in the station-level monitoring device, and time-sensitive data priority caching and non-critical data hierarchical caching are performed;

[0023] S3.2: According to the dynamic cache area, a transmission path set is dynamically constructed based on the topology perception module of the dual-ring network architecture, and the path set includes the main ring network path and the standby ring network path;

[0024] S3.3: Through the transmission path set, real-time collection of each link transmission index is realized, and a dynamic weight path is generated based on the improved ant colony optimization algorithm;

[0025] S3.4: According to the dynamic weight path, data aggregation is performed through the cloud platform communication ring network node to realize time series alignment, data packet integrity verification and transmission path backtracking marking.

[0026] Further, through the uploaded data and hierarchical storage, an index is established to associate the digital twin model with the stored data, including:

[0027] S4.1: Through the aggregated data stream, hierarchical storage decision is made according to data characteristics, and stored in the in-memory database cluster. The digital twin model dynamically renders the interface, writes to the distributed time series database, binds the device historical state analysis service, and stores to the erasure code storage pool, with the addition of the device full life cycle digital archive label;

[0028] S4.2: According to the hierarchical storage decision, a spatio-temporal four-dimensional index system is constructed to realize spatial grid indexing according to the GPS coordinates of the substation plus the catenary mileage pile number, sliding time window indexing based on high-precision timestamp, generating inverted index associated with device digital identity card, and establishing hot-temperature-cold data cross-layer index mapping table;

[0029] S4.3: Through the spatio-temporal four-dimensional index system, an index dynamic update engine is deployed to realize automatic triggering of associated data re-labeling when the digital twin model parameters change, reconstruction of spatial grid index when device topology relationship changes are detected, and synchronous updating of inverted index confidence weight when receiving the device health assessment results generated by S5;

[0030] S4.4: According to the index dynamic update engine, set the data life cycle management strategy, including automatically converting to warm data when the associated device has no alarm for 3 consecutive periods, meeting the complete operation cycle of the device, and automatically loading associated historical data when S5 predicts similar failure patterns.

[0031] Further, according to the stored data, logical operation and threshold judgment are performed on the equipment operation data to generate real-time alarm information, the health state of the equipment is evaluated through statistical analysis and machine learning algorithm, the remaining life and maintenance cycle are predicted, the operation and maintenance strategy is dynamically optimized, and active maintenance suggestions and fault handling plans are generated, including:

[0032] S5.1 According to the stored data, a three-level threshold joint decision mechanism is implemented, the threshold interval is dynamically adjusted, short-term degradation prediction is performed, a LSTM-Transformer hybrid model is deployed for long-term life evaluation, a Weibull-PHM-GARCH combined model is constructed, a transfer learning framework is adopted, laboratory breakdown data and field operation data are fused, and a three-dimensional insulation defect thermal map is obtained;

[0033] S5.2 According to the three-dimensional insulation defect thermal map, a digital twin driven virtual operation and maintenance sand table is created, real-time equipment state parameters are injected, the historical fault mode library stored in S4 is loaded, the reliability / economic indicators of different maintenance strategies are evaluated through Monte Carlo-reinforcement learning hybrid simulation, a Pareto optimal solution set is generated, and a decision support package is output;

[0034] S5.3 Through the output decision support package, a diagnostic model iteration closed loop is constructed, the prediction accuracy is automatically verified every week, model parameter correction is started after major maintenance, a fault knowledge graph is established, and maintenance suggestions and fault handling plans are obtained.

[0035] Further, according to the maintenance suggestions and fault handling plans, the following functions are realized based on the three-dimensional map: dynamic presentation of the whole line substation layout and catenary, real-time monitoring of equipment operation state and three-dimensional accurate positioning of faults, active maintenance strategy generation and fault alarm information push, integrated display of monitoring video and three-dimensional model, and panoramic roaming function, including:

[0036] S6.1: According to the maintenance suggestions and fault handling plans, BIM+GIS+point cloud fusion modeling technology is adopted to realize substation three-dimensional model precision ≤3mm / m, catenary dynamic deformation mapping, underground pipeline perspective visualization, deploy real-time rendering engine of ray tracing, support dynamic light simulation, weather effect superposition;

[0037] S6.2: Through the rendering engine, multi-dimensional monitoring views are constructed, including spatial view, time view and logical view, to realize three-dimensional positioning of faults;

[0038] S6.3: According to the three-dimensional positioning, dynamically integrate the operation and maintenance strategies generated in S5, and deploy intelligent push gateway;

[0039] S6.4: Through the push gateway, a monitoring video-three-dimensional model linkage mechanism is established, and a panoramic roaming engine is developed.

[0040] In a second aspect, a digital twin system for power supply of rail transit includes:

[0041] An acquisition module is configured to realize digital virtualization associated with physical entities by constructing a digital twin model of the power supply system of urban rail transit, complete model rendering, and display the state of the entities with the virtual model; and acquire operating parameters of key equipment of a substation and a catenary according to the state of the entities through sensors and hardware devices.

[0042] A scheduling module is configured to schedule data to a cloud platform communication ring network node after the data is cached by a station-level monitoring device ontology through a double-ring network architecture, optimize a data transmission path based on a load balancing strategy, and upload the data to the cloud platform in real time; perform hierarchical storage on the data uploaded to the cloud platform, establish a structured index, dynamically associate a digital twin model with corresponding stored data, and obtain the stored data.

[0043] A processing module is configured to perform logical operation and threshold judgment on equipment operating data according to the stored data, generate real-time alarm information, evaluate the health status of the equipment by statistical analysis and machine learning algorithm, predict the remaining life and maintenance cycle, dynamically optimize the operation and maintenance strategy, obtain active maintenance suggestions and fault handling plans, realize dynamic presentation of substation layout and catenary based on a three-dimensional map, real-time monitoring of equipment operating state and three-dimensional accurate positioning of faults, generation of active maintenance strategy and push of fault alarm information, integrated display of monitoring video and three-dimensional model, and panoramic roaming function.

[0044] In a third aspect, a computing device includes:

[0045] One or more processors;

[0046] A storage device configured to store one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method.

[0047] In a fourth aspect, a computer readable storage medium stores a program, when the program is executed by a processor, the method is implemented.

[0048] The above-mentioned scheme of the present application at least has the following beneficial effects:

[0049] By adapting LOD3.0 level precision modeling and ISO19650 standard, transient characteristic simulation of catenary-train sliding contact is realized, and by combining millisecond level topology reconstruction technology, the problem of strong coupling modeling of mobile load and fixed equipment is effectively solved, and the model dynamic error rate is reduced to below 0.5%.

[0050] Adopt sliding window filtering and working condition feature matching technology, in the strong interference environment such as pantograph-catenary arc, the signal-to-noise ratio of catenary vibration data is improved by 12dB, the temperature collection error is controlled within ±1.5℃, and the key parameter collection accuracy is guaranteed.

[0051] Through the centimeter-level coordinate mapping engine and the dynamic association of the structured index, the three-dimensional positioning error of the underground tunnel equipment is less than or equal to 5cm, and the accuracy is improved by 8 times compared with the traditional two-dimensional positioning mode.

[0052] The health assessment system based on the machine learning algorithm can predict the local overheating fault of the catenary in advance by 72 hours, the response time of the active maintenance strategy is less than 800ms, the efficiency is improved by 90% compared with artificial inspection, and the operation and maintenance cost is reduced by 45%.

[0053] The double-ring network architecture and the load balancing strategy make the data transmission packet loss rate less than 0.01%, and in combination with the sub-second fault self-healing strategy, the uninterrupted power supply requirement of rail transit can be met.

[0054] Through the integrated display of monitoring video and three-dimensional model, the space-time alignment of electrical parameters and visual data is realized, the pantograph-catenary arc detection accuracy can be improved, and the false alarm rate can be reduced. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 It is a flowchart of a rail transit power supply digital twin construction method provided by an embodiment of the present application.

[0056] Figure 2 It is a rail transit power supply digital twin system schematic diagram provided by an embodiment of the present application.

[0057] Figure 3 It is a flowchart of a construction method of the present application.

[0058] Figure 4 It is a hardware flowchart of an embodiment of the present application.

[0059] Figure 5 It is a software flowchart of an embodiment of the present application. DETAILED DESCRIPTION

[0060] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0061] As Figure 1 shown, an embodiment of the present application proposes a rail transit power supply digital twin construction method, which comprises the following steps:

[0062] Step S1: by constructing a digital twin model of the urban rail transit power supply system, realizing the digital virtualization associated with the physical entity, completing the model rendering, and displaying the state of the entity with the virtual model;

[0063] Step S2: according to the state of the entity, collecting the operation parameters of the substation and the key equipment of the catenary through sensors and hardware devices;

[0064] Step S3: the operation parameters are cached by the station-level monitoring device through the dual-ring network architecture, and then dispatched to the cloud platform communication ring network node, and the data transmission path is optimized based on the load balancing strategy, and the data is uploaded to the cloud platform in real time;

[0065] Step S4: the data uploaded to the cloud platform is stored in a hierarchical manner, and a structured index is established, and the digital twin model is dynamically associated with the corresponding stored data to obtain the stored data;

[0066] Step S5: according to the stored data, logical operation and threshold judgment are performed on the equipment operation data, real-time alarm information is generated, the health status of the equipment is evaluated through statistical analysis and machine learning algorithm, the remaining life and maintenance period are predicted, the operation and maintenance strategy is dynamically optimized, and active maintenance suggestions and fault handling plans are obtained;

[0067] Step S6: according to the maintenance suggestions and fault handling plans, realize the dynamic presentation of the whole line substation layout and the catenary based on the three-dimensional map, the real-time monitoring of the equipment operation state and the three-dimensional accurate positioning of the fault, the generation of the active maintenance strategy and the push of the fault alarm information, the integrated display of the monitoring video and the three-dimensional model and the panoramic roaming function.

[0068] In the embodiment of the application, by constructing the digital twin system of the rail transit power supply system, complete mapping and dynamic interaction from physical entity to virtual space are realized, based on BIM+GIS and multi-physical field coupling modeling technology, millimeter-level precision equipment state visualization and dynamic simulation are realized, through the dual-ring network transmission architecture and intelligent load balancing strategy, the reliability and timeliness of real-time transmission of massive data are guaranteed, combined with multi-modal machine learning algorithm and digital twin simulation engine, the prediction-diagnosis-decision closed loop of equipment health state is constructed, the fault early warning accuracy is improved, the operation and maintenance response efficiency is improved, the three-dimensional visualization platform realizes the fault accurate positioning and maintenance strategy virtual pre-rehearsal, significantly reduces the operation and maintenance cost, the system solves the industry pain points such as the state perception lag of traditional power supply system, the difficulty of fault positioning and passive maintenance decision through the intelligent operation and maintenance mode driven by data, and provides whole life cycle guarantee for rail transit power supply safety.

[0069] In a preferred embodiment of the application, the above step S1 can include:

[0070] Step S1.1: Construct a full-system digital twin model covering from the 110kV main station input line side to the low-voltage load end through the physical entity parameters of the rail transit power supply system, and the digital twin model is rendered in real time through a rendering engine and dynamically synchronized with the running state of the physical entity.

[0071] Step S1.2: According to the running state, the model construction accuracy is not less than LOD3.0, and according to the detailed level defined in the ISO19650 standard, the equipment layout of the substation, the contact net support suspension parts and the wire parameters in the virtual model are consistent with the actual engineering height.

[0072] Step S1.3: By being consistent with the actual height, the running data collected by the physical entity is associated with the digital twin model, the device state is mapped in real time through a three-dimensional visualization interface, and the model rendering details are adjusted to obtain a virtual model to display the state of the entity.

[0073] In the embodiment of the present application, by constructing a high-precision full-system digital twin model, the accuracy and reliability of the state mapping of the power supply system are improved, the modeling standard based on LOD3.0 realizes the millimeter-level restoration of all factors from the substation to the contact net, ensures that the deviation rate of the virtual model parameters and the engineering entity is less than 0.1%, adopts the structured data association mechanism of the ISO19650 standard to realize the real-time bidirectional mapping of the physical parameters and electrical characteristics of the equipment, and the dynamically adjustable three-dimensional rendering engine supports multi-scale visualization from the macro system level to the micro component level.

[0074] In the embodiment of the present application, the specific steps include:

[0075] Step S1.1: Extract geometric parameters such as device size, spatial layout, installation coordinates, electrical parameters such as rated voltage, current, wiring method and component list such as switch cabinet, disconnector, contact net support model from the rail transit power supply system engineering design drawings, equipment specification book, and field survey data, perform three-dimensional laser scanning on the substation equipment and contact net key components, obtain millimeter-level precision point cloud data as the basic data source for model construction, based on BIM building information model technology, construct digital twin model layer by layer according to voltage level 110kV main station→33kV / 25kV substation→contact net→low-voltage load, covering main station input line side equipment, substation internal layout, contact net anchor section arrangement and low-voltage distribution network, realize geometric modeling of each subsystem through modeling tool CATIA, ensure that the model space relationship is 1:1 matched with the actual engineering, form a full-link virtual mirror from high-voltage input to low-voltage output, use Unity rendering engine to perform material mapping and lighting rendering on the model, establish the state mapping interface of the model and the physical entity, and obtain device running data in real time through OPCUA or MQTT protocol to drive the visual state update of the corresponding device in the model.

[0076] Step S1.2: Define the model detail hierarchy, the model needs to include the main structure of the equipment, key components and installation details, the geometric dimension error is ≤3mm / m, support the state monitoring and parameter query of the equipment level, the equipment of the substation is modeled in layers, the first layer: the overall layout of the main station / substation, the second layer: the spatial distribution of the equipment group, the third layer: the three-dimensional structure of the single equipment and the interface parameters such as the cable connection position, the sensor installation point, follow ISO19650 to classify and encode the model data, assign a unique ID to each device / component, associate technical documents, operation records and real-time monitoring data, ensure that the catenary model parameters are consistent with the actual project: support column coordinates (X / Y / Z), height, inclination angle, suspension components (such as positioners, droppers) specifications and connection relationship, contact line / force cable material, cross section, tension parameters.

[0077] Step S1.3: Configure a data mapping table for each virtual device, associate the real-time data interface of the physical sensor (such as the switch cabinet protection and control device, the catenary pantograph monitoring system) through the device ID, realize one-to-one accurate correspondence of one device to one tag, develop an interactive three-dimensional interface, support the following functions: multi-view browsing: overhead view, 360° rotation, equipment level zoom, support positioning the geographical location of each substation / catenary anchor section through the GIS map, real-time state mapping: intuitively display the device running state through color coding (green-normal, yellow-warning, red-fault), superimpose real-time data tags; dynamic adjustment of details: users can customize rendering details, support exporting model view or generating patrol path animation, periodically correct model deviation through field measurement data, ensure that the virtual and actual states are consistent, when the physical equipment is replaced or the layout is adjusted, automatically trigger the model update process, synchronize the associated sensor data interface and visualization configuration.

[0078] In a preferred embodiment of the present application, according to the entity state mapped by the three-dimensional visualization interface of the digital twin model, different working conditions can be adapted, and the local discharge of the equipment can be more accurately captured, providing a strong basis for equipment fault warning, the station-level communication node of the double-ring network architecture sends acquisition instructions to the specified sensor cluster and synchronously starts the monitoring unit, ensuring the efficiency and timeliness of data acquisition, the double-ring network architecture provides reliable communication guarantee, avoids data transmission interruption, enables the operation data of the catenary and substation to be collected quickly and stably, implements sliding window filtering on the catenary data and executes working condition feature matching on the substation data, effectively removes noise and interference in the data, improves the accuracy and reliability of the operation parameters, provides a high-quality data basis for subsequent data analysis, equipment state evaluation and fault diagnosis, etc., and helps to improve the intelligent management level and operation safety of the rail transit power supply system.

[0079] In the embodiment of the present application, the specific steps include:

[0080] Step S2.1: Extracting data related to the physical entity of the rail transit power supply system from the digital twin model, including real-time operating parameters of the equipment (such as voltage, current, temperature, etc.), historical operating data, and simulated equipment state information in the model, determining an index system for evaluating the entity state according to the characteristics and needs of the power supply system, such as the load rate of the equipment, the temperature change rate, the degree of partial discharge, etc., combining various factors affecting the evaluation of the entity state to form a factor set, setting the evaluation levels of the entity state to form a comment set, each factor has different importance in evaluating the entity state, and needs to be assigned a weight to form a factor weight vector, for each factor in the factor set, determine the membership degree of each evaluation level in the comment set, the membership degree reflects the possibility of the factor belonging to a certain evaluation level, the value range is between 0 and 1, combine the membership degrees of all factors to each evaluation level to form a fuzzy relation matrix, through where b j the comprehensive membership degree of the entity state to the a-th evaluation level, w s is the s-th element in the factor weight vector W, r sa is the s-th row and a-th column element in the fuzzy relation matrix R, i is the serial number of the evaluation factor in the factor set U, a is the serial number of the evaluation level in the comment set V, R is the fuzzy relation matrix, V is the comment set, B' is the comprehensive evaluation result, the evaluation result of the entity state is obtained, the evaluation result can be divided into different levels such as normal, warning, and failure, according to the geographical layout and structural characteristics of the rail transit power supply system, the equipment such as the catenary and the substation is segmented in space, for example, it is divided according to different station intervals, different voltage level regions, etc., the trigger conditions of distributed optical fiber temperature measurement and lifting amount detection are set for each segment, such as when the equipment load of a certain segment exceeds a certain threshold or the temperature changes abnormally, the distributed optical fiber temperature measurement and lifting amount detection of the segment is triggered, according to the set trigger conditions and regional division, the corresponding collection instructions are generated, which clearly specify the segment position to be detected, the detection parameters (temperature, lifting amount) and the detection frequency, the operating mode of the equipment is analyzed, such as normal operation mode, overload operation mode, maintenance mode, etc., the operating mode of the equipment can be identified by monitoring the current, voltage, switch state and other parameters, for different operating modes, determine the corresponding partial discharge detection frequency band, for example, in normal operation mode, select a lower detection frequency band, in overload operation mode, appropriately increase the detection frequency band, to accurately detect the partial discharge situation, according to the identified operating mode and the determined detection frequency band, generate the collection instruction of the partial discharge detection frequency band switching.

[0081] Step S2.2: The device parameter acquisition instruction set generated in step S2.1 is encapsulated, necessary header information (such as source address, target address, instruction type, etc.) and verification information are added according to the communication protocol requirements of the dual-ring network architecture, the integrity and accuracy of the instruction are ensured, the station-level communication node in the dual-ring network architecture is configured, including setting the communication parameters (such as baud rate, communication protocol, IP address, etc.), establishing communication connection with the sensor cluster and the monitoring unit, sending the encapsulated acquisition instruction to the designated sensor cluster through the station-level communication node, in the sending process, a reliable communication mechanism such as acknowledgement mechanism is adopted to ensure that the instruction can accurately and correctly reach the target sensor, when the sensor cluster receives the acquisition instruction, the overhead line monitoring unit and the substation monitoring unit are started synchronously, the overhead line monitoring unit is responsible for acquiring the related parameters of the overhead line, such as temperature, lifting amount, partial discharge, etc., and the substation monitoring unit is responsible for acquiring the operating parameters of the equipment in the substation, such as voltage, current, power, etc.

[0082] Step S2.3: The size and step length of the sliding window are determined, the window size determines the number of data points participating in filtering, and the step length determines the distance of the window moving each time, for example, the window size is set to 10 data points, and the step length is 1 data point, the original data collected by the overhead line monitoring unit is filtered by sliding window according to the defined window size and step length, in each window, the average value, median or other statistical quantity of the data is calculated as the filtering result of the window, the filtering result is taken as the processed overhead line data, according to different operating conditions of the substation equipment, a corresponding working condition characteristic library is established, the working condition characteristics include the change law and characteristic mode of the current, voltage, power and other parameters of the equipment under different working conditions, the key features such as amplitude, frequency and phase of the current are extracted from the original data collected by the substation monitoring unit, the extracted features are matched with the feature patterns in the working condition characteristic library to judge the working condition of the equipment, according to the matching result, the data is corrected and processed to remove the interference data caused by the change of working condition, and the accurate operating parameters of the substation are obtained.

[0083] In a preferred embodiment of the present application, the above step S3 can include:

[0084] S3.1: According to the operating parameters, a dynamic cache area is set in the station-level monitoring device, and time-sensitive data priority caching and non-critical data hierarchical caching are performed;

[0085] S3.2: According to the dynamic cache area, a transmission path set is dynamically constructed based on the topology perception module of the dual-ring network architecture, the path set includes a main ring network path and a standby ring network path;

[0086] S3.3: Real-time acquisition of each link transmission index is realized through the transmission path set, and a dynamic weight path is generated based on the improved ant colony optimization algorithm.

[0087] S3.4: According to the dynamic weight path, the cloud platform communication ring network node is used to perform data aggregation, realize time sequence alignment, data packet integrity verification and transmission path backtracking marking.

[0088] In the embodiment of the application, through the technologies of dynamic caching, double ring network path construction, intelligent path optimization and data aggregation verification, the efficiency, reliability and accuracy of data transmission are realized, the dynamic caching mechanism preferentially guarantees the fast storage and transmission of time-sensitive data, avoids the loss of key information, and at the same time, the non-key data is processed in stages, the cache resource allocation is optimized, the double ring network architecture combines topology perception to construct the main and backup transmission paths, ensures automatic switching when the single ring network fails, improves the system fault tolerance, analyzes the link load, delay and other indicators in real time by improving the ant colony algorithm, dynamically generates the optimal transmission path, reduces data congestion and delay, improves network utilization, and data aggregation processing realizes time sequence alignment, integrity verification and path backtracking marking. Through the whole link optimization of caching-path-verification, the real-time and reliable cloud uploading of the power supply system operation data is guaranteed, the dynamic synchronization of the digital twin model and the efficient execution of intelligent operation and maintenance decision are supported, and the monitoring real-time and operation safety of the rail transit power supply system are significantly improved.

[0089] In the embodiment of the application, the specific steps include:

[0090] Step S3.1: According to the running parameters, the time-sensitive data priority P1: fault alarm data (such as overvoltage, short circuit signal), real-time running parameters (such as real-time voltage / current of the overhead line, device temperature transient value), device control instruction feedback data, the delay is required to be less than or equal to 10ms, the packet loss rate is required to be less than or equal to 0.1%, the P2 level device state statistical data (such as daily / weekly load rate average) is required to be less than or equal to 500ms, the P3 level historical waveform file (such as partial discharge spectrum) is required to be less than or equal to 10s, the elastic cache space is divided in the memory of the station level monitoring device, the total capacity is dynamically adjusted according to the number of devices, the cache capacity is equal to the product of the number of devices, the real-time data rate of a single device and 10s, the P1 data occupies a fixed cache partition (≥40% capacity), the P2 / P3 data shares the remaining space, when the P1 data bursts, the P2 / P3 cache space can be preempted, when new data arrives, the data type is analyzed and the priority label is matched, the P1 data is directly written into the fixed cache area, the first-in-first-out plus emergency insertion strategy is adopted: if the cache area is full, the P1 data that enters the earliest is deleted, if it is emergency alarm data, it is directly inserted into the head and triggers immediate transmission, the P2 data is cached according to the time window plus data heat: the device statistical data with high frequency access is retained for 24 hours, the low frequency data is retained for 7 days, the P3 data adopts the on-demand loading plus elimination mechanism, the historical files that are not accessed for more than 48 hours are automatically compressed and stored to the local hard disk, and the cache space is released.

[0091] Step S3.2: Collect all node information in the dual ring network (station-level monitoring device, switch, cloud platform access point), build a node list, record the IP address, MAC address, bandwidth capacity (such as 1 Gbps / 10 Gbps) of each node, scan the ring network connection in real time through LLDP, generate a link matrix L, define the main transmission link in the clockwise direction (such as node 1→node 3→node 5→cloud platform), carry P1 / P2 data, the standby ring network path, define the redundant link in the counterclockwise direction (node 1→node 7→node 5→cloud platform), default to carry P3 data, automatically switch to P1 / P2 data channel when the main ring network fails, generate an initial path set, dynamically adjust the path in combination with the data priority of the buffer area, if the P1 data buffer area capacity exceeds 80%, add bandwidth quota to the main ring network path, send a heartbeat packet every 50 ms to detect the link state, if the delay of a link in the main ring network exceeds the threshold or the packet loss rate >5%, mark it as a fault link, trigger the standby path switching.

[0092] Step S3.3: Real-time collection of bandwidth utilization, transmission delay, error rate, and remaining capacity of each link, define the link state vector S i =[bandwidth utilization, delay, error rate, remaining capacity] as the input parameter of path optimization, through correct the pheromone evaporation, combine heuristic function and pheromone concentration to calculate the transition probability: Where τ ij (t) represents the pheromone concentration on link (i-j) at time t, the pheromone concentration reflects the tendency of ants to choose this link, the higher the concentration, the more likely ants will choose this link, τ ij (t+1) represents the updated pheromone concentration on link (i-j) at time (t+1), ρ is the evaporation coefficient, taking a value of 0.3, α p is the priority weight factor, which is valued according to the priority of the data, m is the total number of ants, i.e. the number of ants participating in path search, L k is the path length of the kth ant, is the probability of the kth ant choosing to transfer from node i to node j at the current node, β is the pheromone importance parameter, taking a value of 1.5, η ij is the heuristic function, γ is the heuristic information importance parameter, taking a value of 1, l∈public path represents all allowed paths from the current node i, i and j represent the node numbers in the network.

[0093] Step S3.4: Time stamp calibration, all data is added with high-precision time stamp in the station-level monitoring device, synchronized with the cloud platform clock based on NTP (Network Time Protocol), multi-source data (substation / catenary) is sorted by time stamp in the cloud platform communication ring network node, linear interpolation method is used to fill in the missing time points, CRC-32 check is used for P1 / P2 data, check code is generated and attached to the end of the data frame, the receiving end recalculates the check code and compares it, the error data packet is marked as invalid and triggers retransmission (retransmission times ≤ 3 times), lightweight check (such as MD5 digest) is used for P3 data, check is only performed when storage space is insufficient, reducing the computational overhead, a path identification field is added in the data frame header, recording the node sequence the data has passed through, such as n1→n3→n5, variable-length encoding (node ID is compressed to 4 bytes) is used, the cloud platform associates the path label with the data storage record after receiving the data, supporting subsequent fault location (such as when a certain link fails, quickly filtering the affected data packets).

[0094] In a preferred embodiment of the present application, the above step S4 can include:

[0095] S4.1: Through the aggregated data stream, hierarchical storage decisions are made according to data characteristics, stored in the in-memory database cluster, associated with the digital twin model dynamic rendering interface, written into the distributed time series database, bound to the device historical state analysis service, and transferred to the erasure code storage pool, with the device full life cycle digital archive label attached.

[0096] S4.2: According to the hierarchical storage decision, a spatio-temporal four-dimensional index system is constructed to realize spatial grid indexing according to substation GPS coordinates plus catenary mileage pile number, sliding time window indexing based on high-precision time stamp, generating inverted index associated with device digital identity card, and establishing hot- warm- cold data cross-layer index mapping table.

[0097] S4.3: Through the spatio-temporal four-dimensional index system, an index dynamic update engine is deployed to automatically trigger associated data re-labeling when the digital twin model parameters change, reconstruct the spatial grid index when the device topology relationship changes, and synchronize the inverted index confidence weight when receiving the device health assessment results generated by S5.

[0098] S4.4: According to the index dynamic update engine, set the data life cycle management strategy, including automatically converting to warm data when the associated device has no alarm for 3 consecutive periods, meeting the complete operation period of the device, and automatically loading associated historical data when S5 predicts similar failure patterns.

[0099] In the embodiments of the present application, through hierarchical storage, spatio-temporal index construction, dynamic index updating and life cycle management, the efficiency of data storage, the accuracy of retrieval and the intelligent scheduling of resources are realized. The hierarchical storage decision matches different storage media (in-memory database, time series database, erasure code storage pool) according to data characteristics (real-time, importance, frequency of use), ensures that real-time data quickly responds to digital twin model rendering, historical data supports long-term state analysis, and archived data guarantees full life cycle tracing, avoids efficiency loss caused by data mixing, and the spatio-temporal four-dimensional index system (spatial grid, time window, device reverse, cross-layer mapping) realizes multi-dimensional fast retrieval of data. The index dynamic updating engine automatically adapts to model parameter changes, device topology changes and health evaluation results, ensures that the index is synchronized with the physical system state in real time, avoids the lag and error rate of manual maintenance of the index, and the life cycle management strategy classifies data according to its hotness (hot data is accessed frequently, warm data is archived regularly, and cold data is stored for a long time). While ensuring data availability, it reduces storage costs (such as automatically reducing non-alarming data to warm storage), and intelligently loads associated historical data when predicting faults, improving the accuracy of analysis models. Overall, this step builds a whole-chain data governance system of storage-index-updating-management, provides bottom-layer storage architecture support for precise data-driven digital twin models and intelligent operation and maintenance of power supply systems, and significantly improves data utilization efficiency and system reliability.

[0100] In the embodiments of the present application, the specific steps include:

[0101] Step S4.1: Differentiate data with high real-time requirements (such as real-time running parameters of devices, alarm information, etc.) and data with low real-time requirements (such as monthly statistical reports of devices, annual inspection records, etc.), identify key data (such as data related to device safety and system stable operation) and non-key data (such as general log information), count the access frequency of data, separate data with high access frequency and data with low access frequency, and store data with high real-time requirements and high access frequency in the in-memory database cluster. For example, store real-time current and voltage data of devices in the in-memory database to quickly respond to the dynamic rendering needs of the digital twin model. For data with time series characteristics, such as historical running state data of devices, write to the distributed time series database. At the same time, bind these data with device historical state analysis services to facilitate subsequent data analysis and mining, and transfer data with lower importance and lower usage frequency to the erasure coding storage pool. When transferring, add a device full life cycle digital archive label to the data to trace and manage the entire life cycle data of the device, associate the data in the in-memory database cluster with the digital twin model dynamic rendering interface to ensure that the digital twin model can obtain the latest device running data for dynamic rendering, and bind the data in the distributed time series database with the device historical state analysis service to provide data support for the historical state analysis of the device.

[0102] Step S4.2: Build a spatio-temporal four-dimensional indexing system, collect GPS coordinates and contact network mileage stake information of substations, and divide the entire power supply network into several spatial grids according to the information, for example, divide the network into different grid areas in units of certain latitude and longitude ranges and mileage stake intervals, associate each data record with its spatial grid number to realize fast retrieval of data by spatial position, use high-precision timestamps in the data to generate sliding time window indexes, for example, divide the data into different time windows according to time sequence in units of hours, days, months, etc., when querying data in a certain time period, the relevant data can be quickly located through the time window index, assign a unique digital identity card to each device, establish an inverted index, and associate the digital identity card of the device with related data records, for example, when querying all related data of a device, the corresponding record can be quickly found through the digital identity card of the device, according to the hotness of the data (hot data, warm data, cold data), establish a hot-warm-cold data cross-layer indexing mapping table, which records the mapping relationship between data of different hotness in different storage layers, facilitating data migration and retrieval.

[0103] Step S4.3: Deploy the dynamic index update engine. When the parameters of the digital twin model change, the dynamic index update engine automatically triggers the relabeling of associated data. For example, if the parameters of a device in the digital twin model are adjusted, the engine will relabel the data records related to that device to ensure the consistency between the data and the model. It monitors changes in the topology of devices in real time. When a change in the topology of a device is detected, such as the addition of a device, the removal of a device, or a change in the connection relationship of a device, the dynamic index update engine will reconstruct the spatial grid index to reflect the latest network topology. It receives the device health assessment results generated in S5 and updates the confidence weight of the inverted index synchronously according to the assessment results. For example, if the health assessment result of a device shows that its reliability has decreased, the engine will correspondingly reduce the confidence weight of the inverted index of the data records related to that device.

[0104] Step S4.4: Set up a data lifecycle management strategy to monitor the alarm status of associated devices in real time. When an associated device has no alarms for three consecutive cycles, the relevant data is automatically changed from hot data status to warm data status. For example, for the real-time operating data of a certain device, if there are no alarm messages for three consecutive hours, its data status is marked as warm data. Track the device's operating cycle, and when the device completes a full operating cycle, the relevant data is processed accordingly. For example, the data within an annual operating cycle of the device is archived and its storage optimized. When a similar failure mode is predicted in S5, the data lifecycle management strategy automatically loads the associated historical data. For example, if a certain device is predicted to experience a certain failure, the system will automatically load relevant data from similar failures in the past, providing a reference for fault diagnosis and handling.

[0105] In a preferred embodiment of the present invention, step S5 may include:

[0106] Based on the stored data, S5.1 implements a three-level threshold joint judgment mechanism, dynamically adjusts the threshold range, predicts short-term degradation, deploys an LSTM-Transformer hybrid model for long-term lifetime assessment, constructs a Weibull-PHM-GARCH combined model, adopts a transfer learning framework, integrates laboratory breakdown data and field operation data, and obtains a three-dimensional insulation defect heat map.

[0107] Based on the 3D insulation defect heat map, S5.2 creates a digital twin-driven virtual operation and maintenance sandbox, injects real-time equipment status parameters, loads the historical fault mode library stored in S4, and evaluates the reliability / economic indicators of different maintenance strategies through Monte Carlo-reinforcement learning hybrid simulation to generate Pareto optimal solution sets and output decision support packages.

[0108] S5.3, by outputting the decision support package, constructing the diagnostic model iterative closed loop, automatically verifying the prediction accuracy every week, starting the model parameter correction after major maintenance, establishing the fault knowledge graph, obtaining the maintenance suggestion and the fault handling plan.

[0109] In the embodiment of the present application, through three-level threshold joint judgment, mixed model prediction and intelligent simulation decision, the accuracy of equipment state evaluation, the scientificity of maintenance strategy and the self-adaptability of model iteration are realized, the three-level threshold joint judgment and dynamic adjustment mechanism can effectively filter data noise, the combination of LSTM-Transformer mixed model captures the short-term degradation trend of equipment, Weibull-PHM-GARCH combined model analyzes long-term life, at the same time, the laboratory and field data are fused by using transfer learning to form a three-dimensional insulation defect thermal map, accurately locate the equipment hidden danger, inject real-time parameters and historical fault modes into the virtual operation and maintenance sand table, through Monte Carlo-reinforcement learning hybrid simulation, the reliability and economy of different maintenance strategies are quantified, the Pareto optimal solution is generated, data-driven decision support is provided for condition-based maintenance, over-maintenance or insufficient maintenance is avoided, the diagnostic model iterative closed loop verifies the prediction accuracy automatically, corrects the parameters after maintenance and constructs the fault knowledge graph, realizes the continuous optimization of the model, ensures that the maintenance suggestion and the fault handling plan are dynamically updated with the equipment state, improves the pertinence and timeliness of the operation and maintenance strategy, as a whole, this step constructs the intelligent operation and maintenance closed loop of state evaluation-strategy generation-model optimization, significantly improves the precision of equipment health management, reduces the operation and maintenance cost, and provides core technical support for the transformation of rail transit power supply system.

[0110] In the embodiment of the present application, the specific steps include:

[0111] Step S5.1: Threshold level definition, first level threshold (warning threshold): based on equipment rated parameters and industry standards such as overhead line conductor temperature rating 70°C, the warning threshold is set to 65°C, triggering yellow warning when real-time data exceeds this threshold, second level threshold (failure threshold): based on equipment safety limit parameters such as insulator withstand voltage critical value 25kV, the failure threshold is set to 22kV, triggering red fault alarm and linkage trip protection, third level threshold (dynamic adjustment threshold): through historical data statistics such as equipment operating parameter fluctuation range in the past 12 months, when the same equipment continuously exceeds the first level threshold for 3 consecutive periods and the trend continues to deteriorate, or exceeds the second level threshold once, triggering a deep diagnosis process, short-term degradation prediction LSTM-Transformer hybrid model, normalizing real-time collected equipment parameters (such as substation switchgear partial discharge quantity, overhead line vibration amplitude), generating time series data matrix (time step = 5 minutes, window length = 240 points, corresponding to 20 hours), LSTM layer captures short-term dependence, sets 128 memory units, dropout rate 0.2, Transformer encoder extracts long-term dependence features (such as load cycle impact on equipment wear), attention head number 8, feedforward network dimension 256, output layer predicts future 12-hour parameter degradation trend through full connection layer, long-term life assessment Weibull-PHM-GARCH combined model, based on equipment historical failure data, fitting failure rate, integrating equipment operating stress data (voltage fluctuation amplitude, current overload times), establishing health index, HI = 0.3 triggers life warning, modeling equipment parameter fluctuation sequence (such as 25kV bus voltage standard deviation), predicting future 3-month operating fluctuation rate, correcting Weibull model parameters, high voltage breakdown test data, labeling defect types, collecting equipment appearance defect data through ultraviolet imager and infrared thermal imager, combining GIS positioning to generate spatial coordinates, freezing the first four convolution layers of pre-trained ResNet50, fine-tuning the last three fully connected layers, migrating laboratory defect features to on-site data classification (accuracy ≥ 92%), rendering defect density heat map according to equipment spatial position (such as substation 3D model coordinates, overhead line support mileage marker), red area represents high insulator defect area, supports drilling to view specific equipment defect types and confidence (such as "2nd substation 10kV switchgear insulation crack, confidence 95%").

[0112] Step S5.2: Real-time acquisition of device state parameters through OPCUA protocol, refresh frequency 100ms, ensure sand table and physical system state synchronization, load near 5 years fault data from S4 stored distributed time series database, classify by device type (such as isolating switch, lightning arrester), each fault mode contains trigger condition, fault impact range, historical maintenance record, support dragging virtual maintenance personnel / device on sand table, simulate implementation process of different maintenance strategies, Monte Carlo simulation, current health status of device (HI value), maintenance resource constraints such as available insulator inventory 50, maintenance personnel 10 groups, operation and maintenance cost parameters, generate 10000 kinds of random fault evolution paths such as insulator crack→partial discharge→phase-to-phase short circuit probability distribution, calculate power outage time, maintenance cost, power supply reliability index under each strategy, device health index HI (0-1), maintenance resource balance, fault impact range (small / medium / large), maintenance strategy set immediate replacement, state monitoring, degraded operation, each action is associated with cost and benefit function, through R=0.6×reliability improvement rate-0.4×maintenance cost / budget, through Q-learning algorithm iteration 100000 times, generate Pareto optimal solution set, optimal maintenance strategy and implementation steps, fault probability prediction under different strategies, required spare parts list, maintenance team scheduling plan.

[0113] Step S5.3: Weekly verification: trigger at 0 o'clock every Monday, compare the device degradation trend predicted by the model last week with the actual data, calculate the root mean square error and the mean absolute percentage error, if RMSE>15% or MAPE>20%, mark the model for correction, after a major maintenance event, automatically import the post-maintenance parameters and trigger model retraining, use Bayesian optimization algorithm to adjust the learning rate (range 1e-5~1e-3) and batch size (16~128) of the LSTM-Transformer model to minimize the validation set MSE, iterate 50 times to complete parameter optimization, device entities: substations, contact net support, switch cabinets, etc., attributes include model, commissioning time, historical fault times, defect entities: insulation cracks, conductor wear, joint overheating, etc., attributes include defect level, inducing factors, maintenance entities: replacing insulators, tightening bolts, insulation testing, etc., attributes include maintenance duration, required tools, historical effectiveness evaluation, build a graph through triplets (device, association, defect), (defect, cause, fault), (maintenance, handle, defect), support complex queries, generate structured maintenance recommendations based on the current state of the device, historical cases in the knowledge graph, and the optimal strategy output by the simulation, after each major maintenance, record the new handling process, spare parts usage record, and effectiveness evaluation result into the contingency plan library, and automatically update the maintenance entity attributes in the knowledge graph.

[0114] In a preferred embodiment of the application, the above step S6 can include:

[0115] S6.1: According to the maintenance suggestion and fault handling plan, the BIM+GIS+point cloud fusion modeling technology is adopted to realize the substation three-dimensional model precision ≤3mm / m, the contact network dynamic deformation mapping, the underground pipeline perspective visualization, the light ray tracing real-time rendering engine is deployed to support dynamic light simulation and weather effect superposition;

[0116] S6.2: Through the rendering engine, multi-dimensional monitoring views are constructed, including spatial view, time view and logical view, to realize three-dimensional positioning of faults;

[0117] S6.3: According to the three-dimensional positioning, the operation and maintenance strategy generated in S5 is dynamically integrated, and an intelligent push gateway is deployed;

[0118] S6.4: Through the push gateway, a monitoring video-three-dimensional model linkage mechanism is established, and a panoramic roaming engine is developed.

[0119] In the embodiment of the application, through high-precision modeling, multi-dimensional visualization and intelligent interaction technology, the intuitiveness, accuracy and efficiency of device operation and maintenance are realized, the BIM+GIS+point cloud fusion modeling technology constructs a substation three-dimensional model with millimeter-level precision and contact network dynamic deformation mapping, combined with light ray tracing rendering to realize complex scene simulation such as weather and lighting, to provide immersive and high-accuracy device space layout and operation status view for operation and maintenance personnel, underground pipeline perspective visualization solves the problem of hidden engineering difficult to monitor in traditional operation and maintenance, multi-dimensional monitoring views (space, time, logic) support fast positioning of fault points from multiple perspectives such as geographic positioning, time series and device logical relationship, shorten the fault troubleshooting time, dynamically integrate operation and maintenance strategies and distribute them in real time through the intelligent push gateway, ensure that the maintenance suggestion and handling plan accurately reach the execution end, improve the decision response speed, the monitoring video-three-dimensional model linkage and the panoramic roaming engine realize real-time mapping of the scene and the virtual model, support remote operation and maintenance personnel to intuitively master the device installation environment through panoramic roaming, combined with video monitoring to dynamically verify the execution effect of the maintenance strategy, this step constructs an operation and maintenance visualization system of high-precision modeling-multi-dimensional display-intelligent interaction, significantly improves the fault positioning efficiency and reduces the spatial cognitive cost of operation and maintenance personnel, provides immersive technical support for unmanned inspection and remote operation and maintenance of the rail transit power supply system, and promotes the transformation of the operation and maintenance mode to visualization, intelligentization and high efficiency.

[0120] In the embodiment of the application, the specific steps include:

[0121] Step S6.1: Establish a substation building information model using tools such as Revit, accurately label equipment geometry, installation location (X / Y / Z coordinate accuracy ±3mm), and component relationships (such as switch cabinet and cable connection ports), establish a parametric model for the catenary system, define the physical properties of the support, dropper, and wire, use vehicle-mounted laser radar to scan the substation and catenary along the line, obtain point cloud data (density ≥100 points / m2), remove noise points using CloudCompare software, generate high-precision three-dimensional point cloud grids, perform ground penetrating radar scanning on underground pipelines to obtain burial depth and strike data, construct an underground space point cloud model, import high-precision geographic information data (scale 1:500) including latitude and longitude coordinates, elevation data, and topography, align the point cloud data with the BIM model coordinates, ensure the overall accuracy of the three-dimensional model of the substation ≤3mm / m, deploy strain sensors on catenary wires, insulators, and other components to collect real-time deformation data, adjust the attitude of the corresponding components in the BIM model through API interface driving, set transparency layers in the three-dimensional model, with an underground pipeline layer transparency of 30% and a surface layer transparency of 80%, support perspective viewing of underground structures through mouse wheel zooming, use Unity / UE engine integrated with ray tracing plugins to perform real-time rendering of indoor and outdoor scenes in the substation, support dynamic lighting simulation and weather effect superposition, and perform LOD grading on complex models: display complete details such as bolts and wiring terminals at close range (<10m), simplify non-critical components at medium distance (10-50m), and use proxy models at long distance (>50m) to ensure a rendering frame rate ≥60FPS.

[0122] Step S6.2: Based on GIS geographic coordinates and BIM three-dimensional models, construct a full-scene spatial view, support mouse dragging and roaming, view angle locking (such as automatically inspecting along the catenary milepost), click on any device to pop up a real-time running parameter floating window, interface with the time series database stored in S4, display device parameter curves such as catenary voltage fluctuation trends and substation power factor changes on the time axis (minute / hour / day / month), support time-space linkage queries, construct device logic connection diagrams such as power supply network topology and protection device action logic chains, mark device states with different colors: green = normal, yellow = warning, red = fault, support fault propagation path simulation such as a certain feeder circuit breaker tripping → visualizing the impact of the coverage of 3 stations, when S5 generates a fault alarm, the system automatically marks the fault point with red flashing in the spatial view, simultaneously highlights the parameter mutation curves 1 hour before and after the fault in the time view, and labels the affected downstream devices in the logic view, catenary fault positioning accuracy ≤0.5 meters (through milepost and BIM model coordinate mapping), and substation equipment positioning accuracy to specific terminal row.

[0123] Step S6.3: Develop a RESTful API interface to parse the maintenance strategy generated in S5 into three-dimensional coordinates, match the fault location results in S6.2, and mark the equipment / position involved in the maintenance strategy with a blue semi-transparent cover in the three-dimensional model. Click on the cover to display the specific operation steps, push the text and image work order to the maintenance personnel's mobile APP through the MQTT protocol, push the strategy execution progress board to the dispatch center large screen, and push the AR navigation path to the on-site intelligent safety helmet. Set the push priority according to the fault level: emergency fault (red) real-time pop-up reminder with vibration, general warning (yellow) displayed at the top of the APP message center, ensuring that the maintenance strategy reaches the relevant personnel within 10 seconds.

[0124] Step S6.4: Deploy 4K monitoring cameras at key locations of the substation and catenary, obtain the camera internal and external parameters, establish the mapping relationship between the video picture pixel coordinates and the BIM model world coordinates, realize the function of automatically switching to the corresponding camera picture by clicking on the three-dimensional model equipment, automatically retrieve the video stream of the three cameras near the location when the three-dimensional model detects abnormal equipment state, identify whether there are abnormal phenomena such as discharge sparks and equipment deformation, and feed back the results to the three-dimensional model label. Automatically roam along the preset route (such as substation equipment inspection path and catenary along the walking board route) at a speed of 0.5m / s, pause automatically at fault points and pop up the detail page, support keyboard / handle control of view angle movement, support high-altitude view angle and ground view angle switching, integrate environmental sound effects, automatically play the corresponding environmental sound when roaming to the equipment nearby, enhance the sense of presence, support VR device access, and realize immersive three-dimensional maintenance scene experience.

[0125] As shown in Figure 2 Embodiments of the present application also provide a rail transit power supply digital twin system, comprising:

[0126] The acquisition module is configured to realize digital virtualization associated with physical entities by constructing a digital twin model of the urban rail transit power supply system, complete model rendering, and display the state of the entity with a virtual model; and according to the entity state, issue instructions to the station level layer, and collect the operating parameters of the key equipment of the substation and the catenary through sensors and hardware devices.

[0127] The dispatch module is configured to cache the data through the station level monitoring device ontology after the operating parameters pass through the dual-ring network architecture, dispatch the data to the cloud platform communication ring network node, optimize the data transmission path based on the load balancing strategy, upload the data to the cloud platform in real time, perform hierarchical storage on the data uploaded to the cloud platform, establish a structured index, dynamically associate the digital twin model with the corresponding stored data, and obtain the stored data.

[0128] The processing module is used for performing logical operation and threshold judgment on the equipment operation data according to the stored data, generating real-time alarm information, evaluating the equipment health state through statistical analysis and machine learning algorithm, predicting the remaining life and maintenance cycle, dynamically optimizing the operation and maintenance strategy, obtaining the active maintenance suggestion and fault processing plan, realizing the dynamic presentation of the whole line substation layout and the overhead contact system based on the three-dimensional map, the real-time monitoring of the equipment operation state and the three-dimensional accurate positioning of the fault, the generation of the active maintenance strategy and the push of the fault alarm information, the integrated display of the monitoring video and the three-dimensional model and the panoramic roaming function.

[0129] Embodiments of the present application also provide a computing device, comprising a processor and a memory storing a computer program, the computer program being executed by the processor to perform the method described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0130] Embodiments of the present application also provide a computer readable storage medium storing instructions, when the instructions are executed on a computer, the computer executes the method described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0131] The above is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the art, without departing from the principles of the present application, can make a number of improvements and refinements, these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. A method for constructing a digital twin of rail transit power supply, characterized in that, The method comprises: S1: by constructing a digital twin model of the urban rail transit power supply system, realizing a digital virtualization associated with the physical entity, completing model rendering, and displaying the state of the entity with a virtual model; S2: according to the state of the entity, collecting the operating parameters of the key equipment of the substation and the catenary through sensors and hardware devices; S3: after the operating parameters are cached by the station-level monitoring device through a dual-ring network architecture, the data is dispatched to the cloud platform communication ring network node, and the data transmission path is optimized based on a load balancing strategy, and the data is uploaded to the cloud platform in real time; S4: the data uploaded to the cloud platform is stored in stages, and a structured index is established, the digital twin model is dynamically associated with the corresponding stored data to obtain the stored data; S5: according to the stored data, logical operation and threshold judgment are performed on the equipment operating data to generate real-time alarm information, the equipment health status is evaluated through statistical analysis and machine learning algorithm, the remaining life and maintenance cycle are predicted, the operation and maintenance strategy is dynamically optimized, and active maintenance suggestions and fault handling plans are obtained; S6: according to the maintenance suggestions and fault handling plans, realize the dynamic presentation of the substation layout and the catenary based on the three-dimensional map, the real-time monitoring of the equipment operating state and the three-dimensional accurate positioning of the fault, the generation of active maintenance strategy and the push of fault alarm information, the integrated display of monitoring video and three-dimensional model and the panoramic roaming function.

2. The rail transit power supply digital twin construction method according to claim 1, characterized in that, By constructing a digital twin model of the urban rail transit power supply system, realizing a digital virtualization associated with the physical entity, completing model rendering, and displaying the state of the entity with a virtual model, including: S1.1: by the physical entity parameters of the rail transit power supply system, a full-system digital twin model covering from the 110kV main substation inlet side to the low-voltage load end is constructed, the digital twin model is rendered in real time by a rendering engine, and is dynamically synchronized with the operating state of the physical entity; S1.2: according to the operating state, the model construction precision is not less than LOD3.0, according to the detailed level defined in the ISO19650 standard, the substation equipment layout, catenary support suspension parts and wire parameters in the virtual model are consistent with the actual height of the project; S1.3: based on the consistency of the model and the actual height of the project, the operating data collected by the physical entity is associated with the digital twin model, the equipment state is mapped in real time through a three-dimensional visualization interface, and the model rendering details can be adjusted to obtain the state of the virtual model.

3. The rail transit power supply digital twin construction method according to claim 2, characterized in that, After the operating parameters are cached by the station-level monitoring device through a dual-ring network architecture, the data is dispatched to the cloud platform communication ring network node, and the data transmission path is optimized based on a load balancing strategy, and the data is uploaded to the cloud platform in real time, including: S3.1: according to the operating parameters, a dynamic cache area is set in the station-level monitoring device, and time-sensitive data is preferentially cached and non-critical data is cached in stages; S3.2: according to the dynamic cache area, a topology perception module based on the dual-ring network architecture dynamically constructs a transmission path set, which includes a main ring network path and a backup ring network path; S3.3: through the transmission path set, real-time collection of each link transmission index is realized, and a dynamic weight path is generated based on an improved ant colony optimization algorithm; S3.4: According to the dynamic weight path, the cloud platform communication ring network node performs data aggregation, realizes time series alignment, data packet integrity verification and transmission path backtracking marking.

4. The rail transit power supply digital twin construction method according to claim 3, characterized in that, Through the uploaded data and hierarchical storage, the index is established to associate the digital twin model with the stored data, including: S4.1: Through the aggregated data stream, hierarchical storage decision is made according to data characteristics, and the data is stored in the memory database cluster, associated with the digital twin model dynamic rendering interface, written into the distributed time series database, bound to the device historical state analysis service, stored in the erasure code storage pool, and attached with the device full life cycle digital archive label; S4.2: According to the hierarchical storage decision, a space-time four-dimensional index system is constructed to realize the establishment of a space grid index according to the GPS coordinates of the substation and the contact net mileage pile number, the generation of a sliding time window index based on high-precision timestamp, the generation of an inverted index associated with the device digital identity card, and the establishment of a hot-temperature-cold data cross-layer index mapping table; S4.3: Through the space-time four-dimensional index system, the index dynamic update engine is deployed to realize automatic triggering of associated data re-labeling when the digital twin model parameters change, reconstruction of the space grid index when the device topology relationship changes, and synchronous updating of the inverted index confidence weight when the device health assessment result generated by S5 is received; S4.4: According to the index dynamic update engine, set the data life cycle management strategy, including automatically converting to warm data when the associated device has no alarm for 3 consecutive periods, meeting the complete operation period of the device, and automatically loading associated historical data when S5 predicts similar failure modes.

5. The rail transit power supply digital twin construction method according to claim 4, characterized in that, According to the stored data, logical operation and threshold judgment are performed on the device operation data to generate real-time alarm information, and statistical analysis and machine learning algorithm are used to evaluate the device health status, predict the remaining life and maintenance period, dynamically optimize the operation and maintenance strategy, and generate proactive maintenance suggestions and fault handling plans, including: S5.1 According to the stored data, a three-level threshold joint decision mechanism is implemented, the threshold interval is dynamically adjusted, short-term degradation prediction is performed, a LSTM-Transformer hybrid model long-term life evaluation is deployed, a Weibull-PHM-GARCH combined model is constructed, a transfer learning framework is used, laboratory breakdown data and field operation data are fused, and a three-dimensional insulation defect thermal map is obtained; S5.2 According to the three-dimensional insulation defect thermal map, a digital twin driven virtual operation sand table is created, real-time device state parameters are injected, a historical fault mode library stored by S4 is loaded, a Monte Carlo-reinforcement learning hybrid simulation is performed, reliability / economic indicators of different maintenance strategies are evaluated to generate a Pareto optimal solution set, and a decision support package is output; S5.3 Through the output decision support package, a diagnostic model iteration closed loop is constructed, the prediction accuracy is automatically verified every week, the model parameter correction is started after major maintenance, a fault knowledge graph is established, and maintenance suggestions and fault handling plans are obtained.

6. The rail transit power supply digital twin construction method according to claim 5, characterized in that, According to the maintenance suggestion and the fault handling plan, the full-line substation layout and the overhead line dynamic presentation based on the three-dimensional map are realized, the equipment operation state real-time monitoring and the fault three-dimensional accurate positioning are realized, the active maintenance strategy generation and the fault alarm information pushing are realized, the monitoring video and the three-dimensional model integrated display and the panorama roaming function are realized, including: S6.1: According to the maintenance suggestion and the fault handling plan, the BIM+GIS+point cloud fusion modeling technology is used to realize that the three-dimensional model precision of the substation is less than or equal to 3mm / m, the overhead line dynamic deformation mapping and the underground pipeline perspective visualization are realized, the real-time rendering engine of the light tracing is deployed, and the dynamic light simulation and the weather effect superposition are supported; S6.2: The multi-dimensional monitoring view is constructed through the rendering engine, the spatial view, the time view and the logic view are realized, and the fault three-dimensional positioning is realized; S6.3: According to the three-dimensional positioning, the operation and maintenance strategy generated in S5 is dynamically integrated, and the intelligent pushing gateway is deployed; S6.4: Through the pushing gateway, the monitoring video-three-dimensional model linkage mechanism is established, and the panorama roaming engine is developed.

7. A digital twin system for power supply of rail transit, the system implements the method of any one of claims 1 to 6, characterized in that, The acquisition module is used to realize the digital virtualization associated with the physical entity by constructing the digital twin model of the urban rail transit power supply system, complete model rendering, and display the state of the entity with the virtual model; according to the entity state, the operation parameters of the key equipment of the substation and the overhead line are collected through sensors and hardware devices; The scheduling module is used to store the data in the cloud platform communication ring network node through the double ring network architecture after the data is cached by the station-level monitoring device ontology, and optimize the data transmission path based on the load balancing strategy, and upload the data to the cloud platform in real time; the data uploaded to the cloud platform is stored in stages, and a structured index is established, the digital twin model is dynamically associated with the corresponding stored data to obtain the stored data; The processing module is used to generate real-time alarm information by logically operating and threshold judging the equipment operation data according to the stored data, evaluate the equipment health state by statistical analysis and machine learning algorithm, predict the remaining life and maintenance period, dynamically optimize the operation and maintenance strategy, and obtain the active maintenance suggestion and the fault handling plan; according to the maintenance suggestion and the fault handling plan, the full-line substation layout and the overhead line dynamic presentation based on the three-dimensional map are realized, the equipment operation state real-time monitoring and the fault three-dimensional accurate positioning are realized, the active maintenance strategy generation and the fault alarm information pushing are realized, the monitoring video and the three-dimensional model integrated display and the panorama roaming function are realized. One or more processors; 8. A computing device, comprising: The storage device is used to store one or more programs, when the one or more programs are executed by the one or more processors, so that the one or more processors realize the method in any one of claims 1 to 6. The computer readable storage medium stores a program, which is executed by the processor to realize the method in any one of claims 1 to 6. ​ 9. A computer-readable storage medium, characterized in that, ​

Images