A featured contribution from Leadership Perspectives: a curated forum reserved for leaders nominated by our subscribers and vetted by the Construction Tech Review Advisory Board.
UDMTEK
Machine Black-Box based on Machine Language Processing
Gi Nam Wang
MLP consists of static and dynamic machine language processing. The static machine language processing technology structurally grasps the relationship between the control logics, enabling various control logic standardization, verification, change management, and automatic inspection. The dynamic machine language processing technology analyzes the characteristics and relationships of various control sections in detail to create a relationship graph, creating a logical connection relationship between the control within and between sectional dynamics when the program is running.
To understand the program being executed, it is necessary to first express different low-level languages for various controllers in a common high-level language. The next step is grasping the relationship between static and dynamic information flows in the process of executing the control software. Dynamic information is the process of communicating with various sensors or control hardware related to the input and output of the control software.
Internal control characteristics are difficult to analyze and understand due to the use of different controllers and control software and the development of control engineers with other purposes. The existing monitoring and diagnosis systems cannot reflect internal control characteristic data using external observation data. For example, human diagnosis has limitations with only outward symptoms such as human fever, cough, and headache. Internal nervous data are necessary for diagnosis with external symptoms. Similarly, machine diagnosis has limitations with only external data of IoT and sensors in an automated process. Diagnosis is essential using data of control segment obtained by MLP and external data.
The BlackBox™ product of UDMTEK, based on machine language processing, enables us to reproduce, see, understand, interpret, analyze, learn, and predict machine control operations and health conditions. The BlackBox™ can build, evaluate, and verify Edge AI models specialized for each specific machine or process. We can develop robust predicting process health conditions through machine language processing-based edge analytics.
The Airplane's black box records only essential information when an airplane crashes, such as past flight trajectories and conditions and pilot operation records. The car's Black-Box stores video data about past car driving and enables analysis only with video data at the time of an accident.
However, The BlackBox™ of UDMTEK enables a machine or process to reproduce and analyze primary data while executing control logic for the problematic process in the past. It links all primary machine or process data with the control program, allowing engineers to understand the flow of detailed data generated while executing and reproducing the control program of the past process. Reproducing past abnormal processes provides an environment for establishing and evaluating analytic methods like prediction and machine condition diagnosis.
This process makes it possible to identify and verify the cause of line stoppage, process abnormality, and quality abnormality. It also understandably expresses the log data while executing the control program, synchronizes and represents the actual video control phenomenon, and allows you to see the explainable AI analysis result and analyze it to understand it. We can efficiently identify the cause of the failure by reproducing, tracking, and exploring the non-operational and line-stopped situations of the past facilities. The processing time could be shortened by improving the facility performance and loss and removing delay factors and irrationality through precise analysis.
The BlackBox™ clearly analyzes various control logics and explores the input-output relationship and the signal pattern of the control operation section to automatically trace the cause of the abnormal situation and quickly identify the cause of the silent stop. The necessary logs are collected by analyzing the control logic. The processing time is shortened by improving the facility performance and loss and removing delay factors and irrationality through precise analysis. We identify and verify the cause of abnormal situations by executing control programs and analyzing related data. This process establishes prediction and diagnosis methods by creating and analyzing control section patterns for each abnormal operation.
The BlackBox™, based on machine language processing, enables us to reproduce, analyze, and verify various equipment and processes, identify anomalies, and establish various health condition diagnosis and prediction methods for parts, modules, machines, processes, lines, and factories.It makes it possible to locate problematic machine or process behavior in the past, even when engineers cannot find it with mere recorded data.
