SAM is implementing Machine Learning (ML) and Artificial Intelligence (AI) to enhance electric inspection. The inspection services are a set of processes, procedures, and software intended to organize, filter, and analyze large, remotely sensed data sets for the electric transmission, distribution, and generation markets. SAM's vision for the process is to utilize ML as a decision support tool for electric utilities to locate assets, identify anomalies and right-of-way encroachments, improve operations and maintenance, and manage vegetation. The LiDAR, ground control survey, and weather data obtained at the same time can be used for engineering design and rebuilds. The primary goal is to support the development of strategic maintenance plans, reduce costs, and improve reliability for electric utilities across multiple business units by utilizing SAM's "one flight, many solution" business strategy. SAM believes the innovations will dramatically change the way data sets are gathered and analyzed across the engineering, aerial patrol, inspection, inventory, and survey disciplines.
SAM ML will be utilized to identify anomalies that may be present in electric transmission and distribution grids utilizing images gathered from multiple sensors. SAM utilizes a variety of ground and aerial inspection techniques and sensors in the visible and invisible ranges of the electromagnetic spectrum. Currently, the automated inspection is supplemented with Subject Matter Experts (SMEs) reviewing and confirming the ML results, potential anomalies, and encroachments discovered on the image datasets. Through time, and improved ML training models, the reports can be generated faster, more uniform, and without human bias. The ultimate value proposition to the client is the ability to review the datasets and results year-over-year and detect changes for trend analysis to better predict when an electric component may fail. The suspect anomalies and identified encroachments would be prioritized and reported to the utility for corrective action to reduce the risk of interruption to service.
SiMON™ is a multi-phased process and software under development to streamline geospatial workflows that deal with the inspection, identification, classification, location and inventory of structures and components acquired via digital capture on a cloud platform.
Phase 1 of SiMON™ was developed by SAM to enable users to quickly and efficiently manage imagery data. This phase provides the ability to process structure photo workflows, from the receipt of the imagery to the finished deliverable, without being overwhelmed with Gigabytes of image data.
Phase 2 of SiMON™ is the automated analysis of image data utilizing Machine Learning models. Initially, the ML model locates, identifies, classifies, clips, and exports the structure into SiMON™ with a high rate of precision. The next generation of ML models will focus on the individual components of a structure. Each model takes only a few seconds per image to accomplish and approximately 1,000 images to effectively train a model.