Power grid project multi-dimensional digital supervision method based on big data

Abstract

The invention relates to a power grid project multi-dimensional digital supervision method based on big data, and belongs to the technical field of power grid project digital supervision, and the method comprises the following steps: S1, carrying out the real-time collection and fusion of multi-dimensional data, and constructing a supervision system through four supervision layers and a digital twinborn body; s2, multi-dimensional supervision indexes are intelligently calculated, and a supervision dimension index system is constructed; s3, risk dynamic early warning and root cause tracing; and S4, decision optimization and closed-loop control. According to the method, multi-source heterogeneous data integration is carried out by fusing multi-dimensional data sources of BIM model, GIS positioning, Internet of Things sensing data, environment monitoring and business system data, so that single data source deviation is avoided, information islands are broken, multi-target collaborative optimization is realized, a'supervision-decision-execution 'closed loop is realized, an optimal regulation and control strategy is convenient to calculate, and the system performance is improved. An optimal regulation and control strategy can be generated based on the reinforcement learning model, and an early warning instruction is directly connected to a field terminal, so that the power grid project monitoring and early warning efficiency can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of digital supervision of power grid projects, and in particular to a multi-dimensional digital supervision method for power grid projects based on big data. Background Art

[0002] Grid project supervision refers to a series of activities carried out by government departments and relevant institutions to supervise, manage and control the entire process of grid project planning, design, construction, acceptance, operation, maintenance, transformation and decommissioning in accordance with laws, regulations and technical standards. Its core goal is to ensure the safety, reliability, economy and environmental protection of grid projects, guarantee the stable operation of the power system and reliable power supply to power users, and promote fair competition and sustainable development in the power market.

[0003] Effective power grid project supervision can effectively prevent casualties and property losses caused by power grid accidents, ensure continuous and stable electricity supply to thousands of households and important users, promote the optimal allocation of power resources on a larger scale, and reduce the impact of power grid construction and operation on the environment. However, traditional power grid project supervision relies on manual inspections and decentralized systems, and has problems such as data islands, delayed responses, and insufficient risk warnings. The multi-source heterogeneous data of construction progress, equipment status, environmental parameters, and safety behavior lack deep fusion and analysis capabilities, and the supervision dimension is single, making it difficult to achieve multi-objective collaborative optimization of "quality-safety-progress-cost-environment". For this reason, a multi-dimensional digital supervision method for power grid projects based on big data is proposed. It integrates multi-source heterogeneous data by combining BIM models (building information), GIS positioning (spatial information), IoT sensor data (equipment status), environmental monitoring (meteorology / geology), and business system data (progress / cost), thereby avoiding the bias of a single data source and breaking down information silos. It also combines computer vision (YOLO target detection), time series prediction (LSTM), and anomaly detection (isolation forest) algorithms to capture quality defects, safety hazards, and environmental risks in real time, thereby shifting risk prevention and control from "passive response" to "active prediction," improving regulatory effectiveness, and implementing multi-objective collaborative optimization to achieve a closed loop of "regulation-decision-making-execution," making it easier to calculate the optimal control strategy. Summary of the Invention

[0004] The present invention provides a multi-dimensional digital supervision method for power grid projects based on big data, which solves the problems raised by the above-mentioned existing background technologies and can realize the integration of multi-source heterogeneous data, thereby avoiding the deviation of a single data source, breaking down information silos, capturing quality defects, safety hazards and environmental risks in real time, and improving the supervision effect.

[0005] The present invention solves the above technical problems with the following solution: a multi-dimensional digital supervision method for power grid projects based on big data, comprising the following steps:

[0006] S1, real-time collection and fusion of multi-dimensional data, building a supervision system through four supervision layers and digital twins. The four supervision layers include perception layer, data layer, analysis layer, and application layer. The structured and unstructured data of the supervision system are accessed through a unified data interface specification. The spatiotemporal alignment algorithm is used to map BIM coordinates, GIS positioning, and sensor data to the digital twin. The specific fusion formula is as follows:

[0007] D_fused=α·D_sensor+β·D_BIM+γ·D_environment

[0008] (α, β, γ are dynamic weight coefficients, which are adaptively calculated by the entropy weight method);

[0009] S2: Intelligent calculation of multi-dimensional supervision indicators to build a supervision dimension indicator system. The supervision dimension indicator system includes quality supervision, safety supervision, progress supervision, cost supervision, and environmental supervision. The system chart is as follows:

[0010]

[0011]

[0012] S3: Dynamic risk warning and root cause tracing, detect abnormal indicators based on the isolation forest algorithm, build a knowledge graph to trace risk-related factors. The specific code is as follows:

[0013] graph LR

[0014] A[Abnormal equipment temperature]-->B(Construction environment humidity>80%)

[0015] A-->C (cooling system failure)

[0016] C-->D[Maintenance records missing];

[0017] S4: Decision optimization and closed-loop control, using reinforcement learning models to generate control strategies and send instructions to on-site terminals.

[0018] On the basis of the above technical solution, the present invention can also be improved as follows.

[0019] Furthermore, in step S1, the perception layer includes IoT devices deployed in the power grid project, and the IoT devices include UWB positioning tags, stress sensors, high-definition cameras, drone inspection systems, and environmental monitoring terminals;

[0020] The data layer establishes a multi-source data lake, integrating BIM models, GIS geographic information, construction logs, equipment operation data, and meteorological environment data;

[0021] The analysis layer is based on the Spark distributed computing framework and runs a multi-dimensional supervision algorithm engine;

[0022] The application layer generates dynamic supervision dashboards and decision instructions.

[0023] Furthermore, in step S3, the specific steps of detecting abnormal indicators based on the isolation forest algorithm are as follows:

[0024] a. Preprocessing operation, Min-Max scaling to the [0,1] interval,

[0025] from sklearn.preprocessing import MinMaxScaler

[0026] scaler = MinMaxScaler()

[0027] X_scaled=scaler.fit_transform(X);

[0028] b. Build an isolation forest model, and set the key parameters as follows:

[0029]

[0030] The parameter selection is based on the icons as follows:

[0031]

[0032] c. Abnormal score calculation and threshold determination. The abnormal score formula is as follows:

[0033] s(x,n)=2^{-\frac{E(h(x))}{c(n)}},

[0034] E(h(x)): average path length of sample x in the forest

[0035] c(n): path length normalization factor (n is the number of samples),

[0036] The judgment logic is as follows:

[0037] #Get the anomaly score (-1 to 1, the closer to 1, the more abnormal)

[0038] anomaly_scores=model.decision_function(X_scaled)

[0039] #Dynamic threshold setting (based on the 3σ principle)

[0040] threshold = np.mean(anomaly_scores) + 3*np.std(anomaly_scores) #Mark abnormal points

[0041] is_anomaly=anomaly_scores>threshold.

[0042] Furthermore, in step S4, the reinforcement learning model construction framework is as follows, and the core element definition chart is as follows:

[0043]

[0044] The weight coefficient is adjusted dynamically.

[0045] ω3=\begin{cases}0.3&\text{Risk\_score<5}\\0.8&\text{Risk\_score≥5}\end{cases};

[0046] Algorithm selection: PPO;

[0047] The network structure is as follows,

[0048]

[0049]

[0050] The training process diagram is as follows:

[0051]

[0052] The beneficial effects of the present invention are as follows: the present invention provides a multi-dimensional digital supervision method for power grid projects based on big data, which has the following advantages:

[0053] 1. Integrate multi-source heterogeneous data by integrating BIM models (building information), GIS positioning (spatial information), IoT sensor data (equipment status), environmental monitoring (meteorological / geological), and business system data (progress / cost), thereby avoiding the bias of a single data source and breaking down information silos;

[0054] 2. Combining computer vision (YOLO object detection), time series prediction (LSTM), and anomaly detection (isolation forest) algorithms, it can capture quality defects, safety hazards, and environmental risks in real time, thereby shifting risk prevention and control from "passive response" to "active prediction" and improving regulatory effectiveness;

[0055] 3. Multi-objective collaborative optimization realizes the closed loop of "supervision-decision-execution", facilitates the calculation of the optimal control strategy, and generates the optimal control strategy based on the reinforcement learning model. The early warning instructions are directly transmitted to the on-site terminal, which can effectively improve the efficiency of power grid project monitoring and early warning.

[0056] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and to implement it according to the contents of the description, the following preferred embodiments of the present invention are described in detail with reference to the accompanying drawings. The specific implementation methods of the present invention are given in detail by the following embodiments and the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0058] Figure 1 A flowchart of a method for multi-dimensional digital supervision of power grid projects based on big data provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following is combined with Figure 1 The principles and features of the present invention are described, and the examples given are only for the purpose of explaining the present invention and are not intended to limit the scope of the present invention. The following paragraphs describe the present invention in more detail by way of example with reference to the accompanying drawings. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are in a very simplified form and are not in exact proportions, and are only used for the purpose of conveniently and clearly assisting in illustrating the embodiments of the present invention.

[0060] Taking the multi-dimensional digital supervision of UHV converter station construction as an example,

[0061] Project background and project scale: Voltage level ±800kV, construction content: valve hall, GIS room, main transformer area, cooling system, construction period: 18 months;

[0062] The monitoring system architecture is as follows

[0063]

[0064] Example 1:

[0065] Valve hall steel structure installation quality supervision

[0066] S1. Data acquisition: The laser scanner generates steel frame point cloud data (accuracy 0.5mm), and the welding robot feeds back current / temperature parameters;

[0067] S2. Quality analysis: compare the point cloud data with the BIM model, use the CNN defect detection model to identify weld bubbles, and use the dynamic weight formula to calculate the quality risk index. The calculation formula is as follows:

[0068] Q_{risk}=0.6\times\Delta D_{deform}+0.3\times T_{weld}+0.1\times H_{ambient}

[0069] (Delta D_{deform} is the deformation, T_{weld} is the welding temperature, and H_{ambient} is the ambient humidity);

[0070] S3, decision execution. When \Delta D_{deform}>1.8mm, the system automatically pauses welding and sends process adjustment instructions to the robot.

[0071] Example 2:

[0072] Cross-operation safety warning, risk of collision between crane and high-altitude workers

[0073] S1, data fusion, UWB positioning personnel coordinates (update frequency 10Hz), crane sensor feedback boom angle and load;

[0074] S2, risk prediction, the LSTM trajectory prediction model calculates the position of personnel / crane in the next 5 seconds. The collision risk formula is as follows:

[0075] Risk_{collision}=\frac{1}{1+e^{-(0.5\times d^{-1}+0.3\times v_{rel})}}

[0076] (d is the real-time distance, v_{rel} is the relative speed);

[0077] S3, decision execution, when Risk_{collision}>0.8, on-site broadcast alarm + crane automatic speed limit.

[0078] Example 3:

[0079] Optimized the pouring schedule for extremely cold concrete, and delayed the pouring schedule in -30℃ environment

[0080] S1. Multi-source monitoring: temperature sensors are embedded in the concrete, and drones use thermal infrared to monitor the temperature distribution in the curing shed.

[0081] S2, progress control, reinforcement learning model dynamically adjusts resource input,

[0082] #Action Space Example (Resource Allocation Strategy)

[0083]

[0084] S3. Cost balance: The system calculates the heating cost C_{heat} and the delay penalty P_{delay} and selects the optimal solution.

[0085] Example 4:

[0086] Early warning and disposal closed loop: At 3 a.m. one day, the system detected an abnormal SF6 gas concentration (>1200ppm) in the GIS room.

[0087] S1. Multi-dimensional correlation analysis: Environmental sensors → SF6 concentration suddenly increased; Video analysis → Valve seal installation offset (YOLOv7 recognition accuracy 98%); Maintenance records → No recent maintenance records for this equipment;

[0088] S2. Risk transmission chain tracing,

[0089] graph TB

[0090] A[SF6 leakage]-->B(seal ring offset 2.3mm)

[0091] B-->C{root cause}

[0092] C-->D[Insufficient installation torque]

[0093] C-->E[material shrinkage at low temperature];

[0094] S3, automatic closed-loop processing,

[0095] a. Trigger ventilation system + evacuation broadcast;

[0096] b. Dispatch maintenance work orders to the nearest personnel, based on UWB positioning distance;

[0097] c. Modify the installation process library of similar equipment, and the torque value is changed from 80Nm to 85Nm.

[0098] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any ordinary technician in this industry can smoothly implement the present invention as shown in the drawings and described above. However, any equivalent changes, modifications and evolutions made by technicians familiar with this profession without departing from the scope of the technical solution of the present invention using the technical content disclosed above are all equivalent embodiments of the present invention. At the same time, any equivalent changes, modifications and evolutions made to the above embodiments based on the essential technology of the present invention are still within the scope of protection of the technical solution of the present invention.

Claims

1. A multi-dimensional digital supervision method for power grid projects based on big data, characterized by: The steps include: S1, real-time collection and integration of multi-dimensional data, building a supervision system through four supervision layers and digital twins. The four supervision layers include the perception layer, data layer, analysis layer, and application layer. The structured and unstructured data of the supervision system are accessed through unified data interface specifications, and the spatiotemporal alignment algorithm is used to map BIM coordinates, GIS positioning and sensor data to the digital twin. S2, intelligent calculation of multi-dimensional supervision indicators, building a supervision dimension indicator system, which includes quality supervision, safety supervision, progress supervision, cost supervision, and environmental supervision; S3, dynamic risk warning and root cause tracing, detects abnormal indicators based on the isolation forest algorithm and constructs a knowledge graph to trace risk-related factors; S4, decision optimization and closed-loop control, uses reinforcement learning models to generate control strategies and send instructions to on-site terminals.

2. The multi-dimensional digital supervision method for power grid projects based on big data according to claim 1 is characterized in that: In step S1, the perception layer includes IoT devices deployed in the power grid project, and the IoT devices include UWB positioning tags, stress sensors, high-definition cameras, drone inspection systems, and environmental monitoring terminals; the data layer establishes a multi-source data lake, integrating BIM models, GIS geographic information, construction logs, equipment operation data, and meteorological and environmental data; the analysis layer runs a multi-dimensional supervision algorithm engine based on the Spark distributed computing framework; and the application layer generates dynamic supervision dashboards and decision-making instructions.

3. The multi-dimensional digital supervision method for power grid projects based on big data according to claim 1 is characterized in that: In step S3, the specific steps of detecting abnormal indicators based on the isolation forest algorithm are as follows: a. Preprocessing operation, Min-Max scaling to the interval [0,1]; b. Build an isolation forest model and set key parameters; c. Anomaly score calculation and threshold determination.

4. The multi-dimensional digital supervision method for power grid projects based on big data according to claim 1 is characterized in that: In step S4, the reinforcement learning model construction framework includes a state space (S), an action space (A), a reward function (R), and a state transition (P).

Images