Data Recognition for Multi-Source Heterogeneous Experimental Detection in Cloud Edge Collaboratives

Introduction

State Grid Corporation of China proposed the “Three-year Action Plan for Lean Operation of Measurement Assets” in 2021 to strengthen research on measurement data. With measurement data management as the focus, the action plan aims to achieve the processing, storage, transmission, exchange, and management activities of measurement data, to excel in the entire process control and management supervision of measurement data, to accelerate the transformation from traditional measurement models to information models, to provide timely and accurate feedback and verification mechanisms for measurement data, and to utilize the advantages of measurement data in testing and inspection (Wang et al., 2021b). Currently, State Grid Corporation of China has in place a three-level measurement system consisting of the State Grid Measurement Center, Provincial Marketing Service Center (Measurement Center), and county-level measurement institutions. The company's existing professional laboratories at all levels carry out the company's inspection, testing, verification, and calibration services, providing quality supervision services for the company and society (He & Su, 2021).

However, the technology and management elements of the measurement system in State Grid Corporation of China are disconnected from the actual measurement work, and there is a lack of information and intelligent data fusion methods for system control. The company's power measurement on-site and laboratory testing and detection business is mainly faced with the problems such as the low automation level of experimental equipment, high degree of isomerization of sensor transmission signals, low level of resource management and sharing, and low level of intelligence (Wang & Feng, 2021; Candra et al., 2021). The detection business has not been effectively coordinated, with problems such as low collection efficiency, high error rate, and data silos in the detection data. The analysis and extraction of key information from most experimental detection data still largely rely on manual labor. The low efficiency in data acquisition, high data dispersion, and the difficulty in ensuring data accuracy directly result in insufficient support for efficient business development from experimental detection data. Patrol, live detection, equipment operation, and other data are mostly in a discrete state, with the problem of duplicate entry, increasing business workload, and other information. The level of laboratory resources and data sharing is low, and a large amount of testing data have not been effectively integrated and applied (Ma et al., 2021a; Khalil et al., 2021).

We propose a differentiable binarization network (DBNet)- and convolutional recurrent neural network (CRNN)-based data recognition method for multisource heterogeneous experimental detection in a cloud-edge collaborative environment to address the issue of inaccurate extraction and recognition of measurement experimental detection data using deep learning networks. Compared with the traditional methods, the innovation of the proposed method lies in:

• 1.

  To overcome the high error rate common to most existing text detection methods, the proposed method utilizes a DBNet network with high detection performance for experimental text detection and uses Dual Path Networks (DPN68) for optimization, which significantly enhance the detection accuracy.

• 2.

  Considering that text data recognition mostly relies on manual labor and has low accuracy, the proposed method designs an improved CRNN algorithm for text data recognition, which can further improve the reliability of recognition through a multihead attention mechanism.