Knowledge graph construction and Internet of Things optimisation for power grid data knowledge extraction

Xiangju Sun; Ting Hao; Xing Li

Journal & Issue Details

Applied Mathematics and Nonlinear Sciences

Journal Details

Format: Journal
eISSN: 2444-8656
First Published: 01 Jan 2016
Publication timeframe: 2 times per year
Languages: English

Grid data is an energy network that fully integrates the information communication and power system, and information interaction runs through all aspects of the smart grid, especially under the trend of electricity marketisation [1, 2, 3, 4]. The power grid has a closer relationship with stakeholders such as operators, service providers and users, which makes the security of information a critical issue [5, 6]. The data attributes of information systems used in the grid include three properties: confidentiality, in addition to the confidentiality of the information content and also the confidentiality of the information status; integrity, which is the characteristic that indicates authorised impossibility of alteration or destruction, including data integrity and system integrity; and availability, which is a feature of information that can be accessed by authorised entities and used on demand. The most important of these is the confidentiality and integrity of data [7, 8].

In the power grid, data are generated in every link of power generation, transmission, substation, distribution, electricity consumption and power regulation, and there are many pieces of data generation equipment, the generation rate is extremely fast, and the transmission path is two-way, resulting in a huge data scale, a variety of types and formats, and a non-uniform structure [9, 10]. The operation of the power system requires cross-departmental, cross-system and cross-regional real-time interaction of information and requires rapid data transmission and interaction and strong data analysis and processing capabilities as the support [11,12]. Big data will improve the management and operation level of the power grid and may even spawn new value-added services like cloud-based data acquisition, data and information management, and big data analytics combined [13, 14]. Big data technology can play an important role in the construction and operation of smart grids, and smart grids can be regarded as the application of big data technologies in power systems [15].

The multi-source heterogeneity problem of grid data improves the security and utilisation of grid data [16, 17]. The smart grid data management platform eliminates the heterogeneity of multiple sources of grid data through data support and service optimisation, thereby improving the data application and operation efficiency of the smart grid [18, 19].

The knowledge graph has since spread and developed in academia and industry [20, 21, 22]. This technology provides a way to extract structured knowledge from a large number of texts and images and has a unique semantic understanding advantage, so it has broad application prospects. The knowledge graph is a symbolic representation of the objective world and an important method for intelligent semantic retrieval [23]. Through the integration and specification of data, the knowledge graph has become an important basic technology to promote data value mining and support intelligent information services. As society and businesses demand knowledge graph construction, the demand for knowledge graph construction is increasing. Currently, researchers are conducting research by combining artificial intelligence techniques, deep neural network algorithms and natural language processing with database technology. Some results have been achieved in things such as the construction of knowledge graphs [24].

IoT is also seen as a network of objects and objects interconnected with each other. IoT uses various information-sensing devices and systems, such as various types of sensors, barcode identification devices and global positioning systems (GPS) [25]. Through the combination of various access networks and the Internet, a large-scale intelligent network is formed. The Internet is used to achieve communication between people, and the Internet of Things is to make people communicate with objects. One big difference between IoT and the internet is that IoT is based on objects, that is, devices. What the user is most concerned about is what ability the object has and what services it can provide to the user. However, at the same time, the current research of the Internet of Things in semantic description, intelligent search, etc. is still far behind the Internet field.

Based on the aforementioned problems, in order to explore the value of data resources of power grid enterprises, what we propose in this article is a semantic search KGSS algorithm which is based on a knowledge graph. Smart field word segmentation technology is used to extract knowledge from structured, semi-structured and unstructured data in power grid enterprises, and the corresponding knowledge graph is established. Semantic search based on similarity strategies is then used. Based on knowledge graph technology, the KGSS algorithm realises intelligent analysis that supports semantic search.

Knowledge graph of power grid data knowledge extraction

Knowledge extraction mainly consists of a schema layer at the upper level and a data layer at the lower level, where a pattern layer is at the heart of the knowledge graph. What is stored is refined knowledge, which is generally managed using ontology. Therefore, the knowledge graph can be seen as an extension of ontology technology. The knowledge graph data layer is the instantiation of the pattern layer, and a wealth of instance information is added on the basis of the knowledge summarised by the pattern layer. Instance information is usually stored in the form of graphs, and traditional relational databases can also implement storage functions, but as the knowledge graph becomes larger and more complex, the storage of inter-entity association relationships becomes very difficult and redundant, and the query efficiency of association relationships will be greatly reduced. In addition, once the modification of some data is involved in the knowledge graph, the modification of relational data will be much more difficult, so the instance data in the knowledge graph data layer generally chooses to use the graph database for storage.

As the size of the knowledge graph grows, its data management issues become increasingly important. The semantic Web field has now developed triplet libraries dedicated to storing RDF data. A graph database for managing property graphs is developed. But at present, there is no recognised reasonable and effective knowledge graph database.

From the perspective of knowledge extraction, the general knowledge graph focuses on the breadth of knowledge and covers coarse-grained knowledge.

From a knowledge representation perspective, a generic knowledge graph represents knowledge as multiple interrelated triples.

From the perspective of knowledge integration, the general knowledge graph has a certain tolerance for the quality of knowledge extraction.

From the perspective of knowledge reasoning, due to the wide knowledge coverage and shallow depth of the general knowledge graph, the reasoning path on the map is relatively short.

From the perspective of atlas application, the general knowledge graph is mainly used in the information search and automatic question answering. In addition to the aforementioned aspects, the main applications of domain knowledge graphs also include decision analysis and business management.

2.1

KGSS algorithm

The KGSS algorithm is mainly composed of three stages: knowledge extraction, knowledge fusion and semantic search. Data are divided into structured, semi-structured and unstructured data based on different data formats. For these different formats of data, the KGSS algorithm uses different strategies to extract knowledge of these data, as shown in Figure 1.

Knowledge extraction framework based on power grid data. MPP, massively parallel processing

Data such as the materials of the power grid company and the operation and inspection of various departments are structured data. These data are dominated by relational data tables. Data are collected through massively parallel processing (MPP). Substantively, MPP is the parallel spread of tasks across multiple servers and nodes. After each node has completed the calculation, each node summarises the calculation results together.

The OWL is used to transform the collected data into a unified physical form. Specifically, OWL is used to represent the classes and properties of the data.

Definition of a class: Classes express the relationship of belonging and inheritance between entities. Each base class is defined first, the hierarchical relationship between classes through parent classes and subclasses is described, and then each subclass is classified.

Genus definition: Property forms include object property and data type property. The former specifies entity categories as well as functionality, while data type attributes associate data with object types.

Knowledge extraction of semi-structured data: For semi-structured data, Hadoop technology is used for data processing. These semi-structured data are generally regular data for timing, such as some metric information, such as voltage, frequency and maximum.

Knowledge extraction of unstructured data: In addition to structured and semi-structured data, there are also some unstructured data such as text images in the distribution network. To this end, the continuous bag of words model with part of speech (POS-CBOW) is used to extract unstructured data. It is mainly composed of an input layer, filter layer, projection layer, annotation layer and output layer.

Firstly, the M word sequence is read sequentially from the window: (1) $C (t - M), C (t - M + 1), \dots, C (t + M)$ C(t - M),C(t - M + 1), \ldots,C(t + M)

The hash table is then used to calculate the corresponding position of these word sequences: (2) $W (t - M), W (t - M + 1), \dots, W (t + M)$ W(t - M),W(t - M + 1), \ldots,W(t + M)

The upper and lower M word contents of the W (t) word are read: (3) $Context (W (t))$ Context(W(t))

Then, the top and bottom of the W (t) word are added: (4) $V (t) = \sum_{t = N}^{t + N} Context [W (t)]$ V(t) = \sum\limits_{t = N}^{t + N} Context[W(t)]

Finally, the vector value of the calculated word W (t) is the word vector regression function: (5) $P [W (t)] Context [W (t)] = \prod_{t - n}^{t + n} f [V (t) θ]$ P[W(t)]Context[W(t)] = \prod\limits_{t - n}^{t + n} f[V(t)\theta ] (6) $f [V (t) θ] = \frac{1}{{1 + \exp [- V (t) θ x]}}$ f[V(t)\theta ] = {1 \over {\left\{ {1 + \exp [ - V(t)\theta x]} \right\}}}

2.2

Knowledge fusion

After obtaining the knowledge entities, it is necessary to calculate the relationship between the entities and build a 3-tuple of knowledge for the distribution network, and reference the 3rd entity in the prediction tuple of the projection embedding model (PEM).

Firstly, {E, R, D} a pending 3-tuple is constructed. E, R are known elements. D is an unknown element. E to R perform the correlation operation: (7) $E \oplus R = D_{e} E + D_{r} R + b_{c}$ E \oplus R = {D_e}E + {D_r}R + {b_c} where D_e and D_r are diagonal matrices of the k × k dimension representing the entity and relationship weights, respectively, and b_c represents the correlation bias. Using this association operation, the projective function is represent as follows: (8) $h (e, r) = g (W^{c} f (e \oplus r) + b_{p})$ h(e,r) = g\left({{W^c}f(e \oplus r) + {b_p}} \right)

The entities are then selected for distance calculations, and the prediction task is converted into a sorting problem for the candidate entities. Activation functions are represented according to f (*) and g(*) using sigmoid and tanh functions as activation functions. The entity score is as follows: (9) $h (e, r) = sigmoid (W_{{[i, :]}^{c}} tanh (e \oplus r) + b_{p})$ h(e,r) = sigmoid\left({{W_{{{[i,:]}^c}}}\tanh (e \oplus r) + {b_p}} \right)

Finally, a top-down knowledge graph construction method is adopted. Firstly, ontology is extracted from the data source, and ontology learning is performed, including terminology extraction, ontology concept learning and ontological relationship learning, followed by entity learning, where entity links and entity populations are populated, and finally, a knowledge graph is built.

2.3

Semantic search

The KGSS algorithm uses the similarity strategy to calculate search requests to entities in the knowledge graph by mapping. The semantic search steps based on similarity are as follows: the recognition model is used to identify the distribution network entities being searched. Then, the similarity between the entities being searched and the entities in the knowledge graph is calculated. (10) $sim (a_{i}, a_{j}) = \frac{\sum_{i = 1}^{n} (a_{i} \times a_{j})}{\sqrt{\sum_{i = 1}^{n} a_{i}^{2}} + \sqrt{\sum_{j = 1}^{n} a_{j}^{2}}}$ sim({a_i},{a_j}) = {{\sum\limits_{i = 1}^n ({a_i} \times {a_j})} \over {\sqrt {\sum\limits_{i = 1}^n a_i^2} + \sqrt {\sum\limits_{j = 1}^n a_j^2} }}

In the knowledge graph, the more an entity is associated with other entities, the more important that entity is and the higher the weight. The weights for each entity are calculated as follows: (11) ${weight}_{a_{i}} = \sum_{a_{j}}^{A} \frac{1}{d {(a_{i}, a_{j})}^{2}}$ weight_{{a_i}} = \sum\limits_{{a_j}}^A {1 \over {d{{({a_i},{a_j})}^2}}} where A = {a₁, a₂,…, a_n} is the extracted collection of n entities and d(a_i, a_j) is the distance between the entity a_i and entity a_j in the knowledge graph.

Knowledge graph construction and Internet of Things optimisation framework for grid data knowledge extraction

This article first constructs a definition based on the knowledge graph of power grid data knowledge extraction and briefly discusses its algorithm architecture, composition and the relationship and difference with power grid data. Finally, from the aspects of Internet of Things optimisation, including the coordination and control of distributed devices in a wide area, the integration of the power system and transportation system, the integration of the power system and natural gas network, the integration of the power system and natural gas network, and the research of knowledge graph construction and Internet of Things optimisation of power grid data knowledge extraction are discussed.

The research object of smart grid is the power system, while the research object of energy Internet is different from that of smart grid. It includes not only the power system and transportation system but also the natural gas network.

In smart grid, the only form of energy transmission and use is electric energy. In the energy Internet, the forms of energy conversion are diverse, including but not limited to electric energy, chemical energy, heat energy and other forms, and can be converted into each other.

At present, in power grid research, the research strategy adopted in distributed equipment such as distributed generation, energy storage and controllable load is mainly local absorption and control. In energy Internet, due to the large number of distributed devices, the research focus will change from local absorption to wide area coordination.

The main research object of the information system of electric power network is the traditional industrial control system, but in energy Internet, the open information network such as Internet becomes a new research form and content (Figure 2).

Knowledge graph construction and Internet of Things optimisation framework for knowledge extraction from power grid data

Experimental results and analysis

4.1

Experimental data preprocessing

Taking a power grid company as a test point, information such as citizen marketing, operation and inspection, electricity consumption, fault records of power grid stations and power grid ledgers was collected, and the amount of data reached 5.4 TB. For structured data, data are indexed with fields. For semi-structured data, data metrics are extracted as time fields, and time metrics are indexed as data. For unstructured data, data are indexed by word breakers.

Text preprocessing refers to the process of preprocessing the corpus obtained from the grid company, including deduplication, denoising, clause segmentation, word segmentation and part-of-speech annotation.

Deduplication: It refers to the removal of duplicate information, the deletion of recurring corpus and the avoidance of unnecessary processing.

Denoising: In this process, disturbing information such as pictures and tables that appear during data collection is deleted.

Clauses: A cluster refers to the division of the corpus according to a defined delimiter, which is a period.

Participles: A particle refers to the division of words in a sentence because Chinese is different from English, and there is no space in the sentence of Chinese to separate words. Therefore, we need to use the word segmentation tool to divide words. Commonly used word segmentation tools are HIT LIP, Chinese Academy of Sciences word segmentation tool and Stanford Chinese word segmentation tool. Word segmentation is a basic work in the experiment in this study because the accuracy of word segmentation is very important, so if the word segmentation is not accurate, it will affect the subsequent experiment. The study selects the word segmentation tool of the Chinese Academy of Sciences, which has added a user dictionary, and the user can first manually define some words as recognition standards to solve the problem of splitting complete words when using the tool, resulting in incorrect recognition.

4.2

Experimental evaluation criteria

Accuracy and recall are important performances that measure semantic search algorithms. The accuracy rate is the proportion of the correct results of the search to the total search results. The recall rate is the ratio of the correct result of the search to the correct result that actually exists. The larger the values of the accuracy and recall rates, the better the performance of the search algorithm. (12) $R = \frac{TP}{TP + FN}$ R = {{TP} \over {TP + FN}} (13) $P = \frac{TP}{TP + FP}$ P = {{TP} \over {TP + FP}} (14) $F_{1} = \frac{2 PR}{P + R}$ {F_1} = {{2PR} \over {P + R}}

The traditional keyword search (referred to as keyword search) is selected as a reference, and the accuracy and recall rate of the KGSS algorithm and keyword search are compare and analysed. The keyword search uses SQL statement queries. The KGSS algorithm uses SPARQL statements to query based on the knowledge graph.

4.3

Experimental environment

The development environment is a Linux system; the GPU uses NVIDIA GeForce RTX 2080Ti (11 GB) and Python version 3.6.5; the framework uses PyTorch 1.7 and TensorFlow 1.15 versions; and CUDA uses version 10.1. In order to prevent the overfitting phenomenon of the model, the experimental model improves the generalisation ability of the model, after several sessions of training to obtain the optimal parameters.

4.4

Analysis of experimental results

From Figure 3, the KGSS algorithm and keyword search have the largest difference in the query rate in search condition 5. Mainly because of search condition 5 (lightning strike malfunctioning equipment), it fully reflects the difference in the search strategy of the two algorithms. In the keyword search system, “lightning strike” and “failure” have not established a relationship. The KGSS algorithm establishes the relationship structure between “lightning strike” and “failure” through the knowledge graph (Table 1).

Table 1

Experimental results

Search criteria/item	P/%	R/%	F₁/%

1	87.45	89.32	88.62
2	79.32	80.32	80.71
3	81.25	82.91	82.18
4	85.32	84.78	85.52
5	84.21	83.12	83.98
6	83.92	82.41	83.69
7	100	100	0
8	81.25	80.18	80.32

From Figures 3 and 4, the accuracy and recall rate of the KGSS algorithm are generally better than those of the keyword search algorithm. Under different search conditions, the accuracy and recall rates of the KGSS algorithm are relatively stable. As shown in Figure 3, the accuracy rate of the KGSS algorithm varies around 90%, and the accuracy rate is above 80%. This is superior to that of the KGSS algorithm by searching through entity attributes and relationships with other entities, improving the accuracy of search results.

In addition, in Figure 4, it can be noted that the recall rate of the keyword search in search condition 7 (engineers with extensive fault recovery experience) is 0, while the corresponding accuracy rate reaches 100%. The reason is that keyword searches do not get “experienced” semantics, and the search results are empty. Therefore, the accuracy rate is 100%, and the recall rate is 0%.

Conclusion

In order to explore the value of data resources of power grid enterprises, this article proposes a semantic search KGSS algorithm based on the knowledge graph. The intelligent field word segmentation technology is used to extract knowledge from structured, semi-structured and unstructured data in power grid enterprises, and the corresponding knowledge graph is established. The semantic search based on similarity strategies is then used. Based on knowledge graph technology, the KGSS algorithm realises intelligent analysis that supports semantic search. There are massive amounts of data in power grid enterprises, and using these massive amounts of data to build a knowledge graph is a complex project. The semantic search for the application of knowledge graphs to power grids is still in its infancy. In future works, the algorithm will be further optimised to improve the timeliness and accuracy of search.

Figures & Tables

Experimental results

Search criteria/item	P/%	R/%	F₁/%

1	87.45	89.32	88.62
2	79.32	80.32	80.71
3	81.25	82.91	82.18
4	85.32	84.78	85.52
5	84.21	83.12	83.98
6	83.92	82.41	83.69
7	100	100	0
8	81.25	80.18	80.32

References

[1] Tatebe O, Morita Y, Matsuoka S, et al. Grid Datafarm Architecture for Petascale Data Intensive Computing[C]//IEEE/ACM International Symposium on Cluster Computing & the Grid. ACM, 2001. TatebeO MoritaY MatsuokaS Grid Datafarm Architecture for Petascale Data Intensive Computing[C] IEEE/ACM International Symposium on Cluster Computing & the Grid. ACM 2001 Search in Google Scholar

[2] Vazhkudai S, Schopf J M. Predicting sporadic grid data transfers[C]//Proceedings 11th IEEE International Symposium on High Performance Distributed Computing. IEEE, 2002. VazhkudaiS SchopfJ M Predicting sporadic grid data transfers[C] Proceedings 11th IEEE International Symposium on High Performance Distributed Computing. IEEE 2002 Search in Google Scholar

[3] Efthymiou C, Kalogridis G. Smart Grid Privacy via Anonymization of Smart Metering Data[C]//Smart Grid Communications (SmartGridComm), 2010 First IEEE International Conference on. IEEE, 2010. EfthymiouC KalogridisG Smart Grid Privacy via Anonymization of Smart Metering Data[C] Smart Grid Communications (SmartGridComm), 2010 First IEEE International Conference on. IEEE 2010 10.1109/SMARTGRID.2010.5622050 Search in Google Scholar

[4] Kim Y J, Thottan M, Kolesnikov V, et al. A secure decentralized data-centric information infrastructure for smart grid[J]. Communications Magazine IEEE, 2010, 48(11):58–65. KimY J ThottanM KolesnikovV A secure decentralized data-centric information infrastructure for smart grid[J] Communications Magazine IEEE 2010 48 11 58 65 10.1109/MCOM.2010.5621968 Search in Google Scholar

[5] Meng X, Zhixiang J I, Yang Y. Research on Big Data Platform and Its Key Technologies in Smart Grid[J]. Distribution & Utilization, 2015. MengX ZhixiangJ I YangY Research on Big Data Platform and Its Key Technologies in Smart Grid[J] Distribution & Utilization 2015 Search in Google Scholar

[6] Ding J, Xi H, Han H, et al. A Smart Grid-oriented Data Placement Strategy for Data-intensive Cloud Environment[J]. Dianli Xitong Zidonghua/Automation of Electric Power Systems, 2012, 36(12):66–70+100. DingJ XiH HanH A Smart Grid-oriented Data Placement Strategy for Data-intensive Cloud Environment[J] Dianli Xitong Zidonghua/Automation of Electric Power Systems 2012 36 12 66 70+100 Search in Google Scholar

[7] Zhang Y, Qian T, Tang W. Buildings-to-distribution-network integration considering power transformer loading capability and distribution network reconfiguration[J]. Energy, 2022, 244. ZhangY QianT TangW Buildings-to-distribution-network integration considering power transformer loading capability and distribution network reconfiguration[J] Energy 2022 244 10.1016/j.energy.2022.123104 Search in Google Scholar

[8] T. Qian, Xingyu Chen, Yanli Xin, W. H. Tang, Lixiao Wang. Resilient Decentralized Optimization of Chance Constrained Electricity-gas Systems over Lossy Communication Networks [J]. Energy, 2022, 239, 122158. QianT. ChenXingyu XinYanli TangW. H. WangLixiao Resilient Decentralized Optimization of Chance Constrained Electricity-gas Systems over Lossy Communication Networks [J] Energy 2022 239 122158 10.1016/j.energy.2021.122158 Search in Google Scholar

[9] T. Qian, Y. Liu, W. H Zhang, W. H. Tang, M. Shahidehpour. Event-Triggered Updating Method in Centralized and Distributed Secondary Controls for Islanded Microgrid Restoration[J]. IEEE Transactions on Smart Gird, 2020, 11(2): 1387–1395. QianT. LiuY. ZhangW. H TangW. H. ShahidehpourM. Event-Triggered Updating Method in Centralized and Distributed Secondary Controls for Islanded Microgrid Restoration[J] IEEE Transactions on Smart Gird 2020 11 2 1387 1395 10.1109/TSG.2019.2937366 Search in Google Scholar

[10] CH FANG, YN TAO, JG EANG, et al. Mapping Relation of Leakage Currents of Polluted Insulators and Discharge Arc Area[J]. Frontiers in Energy Research, 2021. FANGCH TAOYN EANGJG Mapping Relation of Leakage Currents of Polluted Insulators and Discharge Arc Area[J] Frontiers in Energy Research 2021 10.3389/fenrg.2021.777230 Search in Google Scholar

[11] Tehrani S O, Shahrestani A, Yaghmaee M H. Online electricity theft detection framework for large-scale smart grid data[J]. Electric Power Systems Research, 2022, 208:107895- TehraniS O ShahrestaniA YaghmaeeM H Online electricity theft detection framework for large-scale smart grid data[J] Electric Power Systems Research 2022 208 107895 10.1016/j.epsr.2022.107895 Search in Google Scholar

[12] Wang J, Wu L, Zeadally S, et al. Privacy-preserving Data Aggregation against Malicious Data Mining Attack for IoT-enabled Smart Grid[J]. ACM Transactions on Sensor Networks (TOSN), 2021. WangJ WuL ZeadallyS Privacy-preserving Data Aggregation against Malicious Data Mining Attack for IoT-enabled Smart Grid[J] ACM Transactions on Sensor Networks (TOSN) 2021 10.1145/3440249 Search in Google Scholar

[13] Zhou S, Li H, Zhang J, et al. Application Research of Big Data Real-time Processing Technology in Smart Grid[J]. IOP Conference Series: Earth and Environmental Science, 2021, 696(1):012044 (8pp). ZhouS LiH ZhangJ Application Research of Big Data Real-time Processing Technology in Smart Grid[J] IOP Conference Series: Earth and Environmental Science 2021 696 1 012044 (8pp) 10.1088/1755-1315/696/1/012044 Search in Google Scholar

[14] Lu W, Ren Z, Xu J, et al. Edge Blockchain Assisted Lightweight Privacy-Preserving Data Aggregation for Smart Grid[J]. IEEE Transactions on Network and Service Management, 2021, 18(2):1246–1259. LuW RenZ XuJ Edge Blockchain Assisted Lightweight Privacy-Preserving Data Aggregation for Smart Grid[J] IEEE Transactions on Network and Service Management 2021 18 2 1246 1259 10.1109/TNSM.2020.3048822 Search in Google Scholar

[15] Chen B, Yu X. Research on Smart Grid Data Layout Strategy Based on Cloud Computing[J]. Journal of Physics: Conference Series, 2020, 1453(1):012065 (4pp). ChenB YuX Research on Smart Grid Data Layout Strategy Based on Cloud Computing[J] Journal of Physics: Conference Series 2020 1453 1 012065 (4pp) 10.1088/1742-6596/1453/1/012065 Search in Google Scholar

[16] Ding Y, Wang B, Wang Y, et al. Secure Metering Data Aggregation With Batch Verification in Industrial Smart Grid[J]. IEEE Transactions on Industrial Informatics, 2020, 16(10):6607–6616. DingY WangB WangY Secure Metering Data Aggregation With Batch Verification in Industrial Smart Grid[J] IEEE Transactions on Industrial Informatics 2020 16 10 6607 6616 10.1109/TII.2020.2965578 Search in Google Scholar

[17] Kohli J K, Sharma R. Short Term Solar Radiation Forecasting for Photovoltaic Installations in Smart Grid: A Data Driven Approach[J]. SSRN Electronic Journal, 2020. KohliJ K SharmaR Short Term Solar Radiation Forecasting for Photovoltaic Installations in Smart Grid: A Data Driven Approach[J] SSRN Electronic Journal 2020 10.2139/ssrn.3575423 Search in Google Scholar

[18] Croce D, Giuliano F, Tinnirello I, et al. Privacy-Preserving Overgrid: Secure Data Collection for the Smart Grid[J]. Sensors (Basel, Switzerland), 2020, 20(8). CroceD GiulianoF TinnirelloI Privacy-Preserving Overgrid: Secure Data Collection for the Smart Grid[J] Sensors (Basel, Switzerland) 2020 20 8 10.3390/s20082249721886032316139 Search in Google Scholar

[19] Sehanobish A, Sandora M C, Abraham N, et al. Explaining the Effectiveness of Multi-Task Learning for Efficient Knowledge Extraction from Spine MRI Reports[J]. 2022. SehanobishA SandoraM C AbrahamN Explaining the Effectiveness of Multi-Task Learning for Efficient Knowledge Extraction from Spine MRI Reports[J] 2022 10.18653/v1/2022.naacl-industry.16 Search in Google Scholar

[20] Weichselbraun A. Inscriptis – A Python-based HTML to text conversion library optimized for knowledge extraction from the Web[J]. 2021. WeichselbraunA Inscriptis – A Python-based HTML to text conversion library optimized for knowledge extraction from the Web[J] 2021 10.21105/joss.03557 Search in Google Scholar

[21] Chen T, Zhu J, Zeng Z, et al. Compressor Fault Diagnosis Knowledge: a Benchmark Dataset for Knowledge Extraction from Maintenance Log Sheets based on Sequence Labeling[J]. IEEE Access, 2021, PP (99):1–1. ChenT ZhuJ ZengZ Compressor Fault Diagnosis Knowledge: a Benchmark Dataset for Knowledge Extraction from Maintenance Log Sheets based on Sequence Labeling[J] IEEE Access 2021 PP (99):1–1. 10.1109/ACCESS.2021.3072927 Search in Google Scholar

[22] Yang M, Chen K, Sun S, et al. A Pattern Driven Graph Ranking Approach to Attribute Extraction for Knowledge Graph[J]. IEEE Transactions on Industrial Informatics, 2021, PP (99):1–1. YangM ChenK SunS A Pattern Driven Graph Ranking Approach to Attribute Extraction for Knowledge Graph[J] IEEE Transactions on Industrial Informatics 2021 PP (99):1–1. 10.1109/TII.2021.3073726 Search in Google Scholar

[23] Nordsieck R, Heider M, Winschel A, et al. Knowledge Extraction via Decentralized Knowledge Graph Aggregation[C]//2021 IEEE 15th International Conference on Semantic Computing (ICSC). IEEE, 2021. NordsieckR HeiderM WinschelA Knowledge Extraction via Decentralized Knowledge Graph Aggregation[C] 2021 IEEE 15th International Conference on Semantic Computing (ICSC). IEEE 2021 10.1109/ICSC50631.2021.00024 Search in Google Scholar

[24] Chen N, Qiu T, Zhao L, et al. Edge Intelligent Networking Optimization for Internet of Things in Smart City[J]. IEEE Wireless Communications, 2021, 28(2):26–31. ChenN QiuT ZhaoL Edge Intelligent Networking Optimization for Internet of Things in Smart City[J] IEEE Wireless Communications 2021 28 2 26 31 10.1109/MWC.001.2000243 Search in Google Scholar

[25] Choi J, Park J. MIMO Design for Internet-of-Things: Joint Optimization of Spectral Efficiency and Error Probability in Finite Blocklength Regime[J]. IEEE Internet of Things Journal, 2021, PP (99):1–1. ChoiJ ParkJ MIMO Design for Internet-of-Things: Joint Optimization of Spectral Efficiency and Error Probability in Finite Blocklength Regime[J] IEEE Internet of Things Journal 2021 PP (99):1–1. 10.1109/JIOT.2021.3073239 Search in Google Scholar

Applied Mathematics and Nonlinear Sciences

Knowledge graph construction and Internet of Things optimisation for power grid data knowledge extraction

Xiangju Sun,
Xiangju Sun,
Ting Hao and
Ting Hao and
Xing Li
Xing Li

Published Online: 30 Nov 2022

Volume & Issue: AHEAD OF PRINT

Page range: -

Received: 03 Jul 2022

Accepted: 16 Sep 2022

Journal & Issue Details

Applied Mathematics and Nonlinear Sciences

PDF Preview

Article

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Figures & Tables

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Experimental results

References

Recent Articles

Plan your remote conference with Sciendo

Applied Mathematics and Nonlinear Sciences

Knowledge graph construction and Internet of Things optimisation for power grid data knowledge extraction

Xiangju Sun,Xiangju Sun,Ting Hao andTing Hao andXing LiXing Li

Published Online: 30 Nov 2022

Volume & Issue: AHEAD OF PRINT

Page range: -

Received: 03 Jul 2022

Accepted: 16 Sep 2022

Journal & Issue Details

Applied Mathematics and Nonlinear Sciences

PDF Preview

Article

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Figures & Tables

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Experimental results

References

Recent Articles

Plan your remote conference with Sciendo

Xiangju Sun,
Xiangju Sun,
Ting Hao and
Ting Hao and
Xing Li
Xing Li