FsTech List of Products and Services

"DTEmpower" is a data analysis and modeling software that can extract valuable information from various data it holds and build high-quality data models. It provides a graphical operating environment and powerful algorithms, allowing users without data modeling experience to easily perform data analysis and model construction. With its unique algorithm AIOD for outlier detection, it can identify samples that exhibit abnormal characteristics in local subsets. 【Features】 ■ Data Cleaning ■ Data Clustering ■ Sensitivity Analysis ■ Regression Models ■ Zero Coding *For more details, please refer to the PDF materials or feel free to contact us.

We will introduce case studies on data cleaning and AIAgent training using our intelligent data modeling software "DTEmpower." This software provides users with a variety of algorithms related to different aspects of data modeling, such as sensitivity analysis and model training. Using a simulation dataset based on the Styblinski-Tang function, we verified the effectiveness of the model training algorithm "AIAgent." The response surface is essentially aligned with theoretical values, confirming high accuracy. [Case Overview] - The goal of data modeling is to obtain a predictive model from x (x1 to x5) to y. - To confirm the superiority of AIAgent, a new project was simultaneously created, and validation using an integrated learning algorithm was also conducted. - The response surface of the model trained by AIAgent is essentially aligned with theoretical values, confirming high accuracy. *For more details, please download the PDF or feel free to contact us.*

We would like to introduce the application examples of data preprocessing and visualization using our intelligent data modeling software "DTEmpower." This software enables effective utilization of existing data even in situations where there is no experience in data modeling. Most of the sample points in this case are distributed along the same curve corresponding to normal operations, but in datasets with anomalies, some sample points are significantly deviated from the curve, allowing for a visual confirmation of potential anomalies. [Case Overview] - Preparing an appropriate dataset is an important task in modeling, so we use the built-in data preprocessing and visualization tools. - Samples from two types of data sources used here are plotted on the same scatter plot and displayed in two colors. *For more details, please download the PDF or feel free to contact us.

We would like to introduce an application case of timing prediction for parameters of our intelligent data modeling software "DTEmpower." This product is equipped with graphical process construction features, allowing all data and model operations to be performed simply by manipulating the modules within the toolbox. In the time series forecasting problem, our software addressed issues related to data characteristics and model training, ultimately improving the R2 index of the prediction model from 0.68 to 0.94. [Case Overview] - Faced with a complex dynamic biochemical process model characterized by strong external interference, time-varying factors, coupling, and non-linearity, we built a data-driven system based on a large amount of measurement data. - By rationally selecting external features and introducing feature engineering techniques such as MDI/PCA, we not only enhanced the richness of input information but also improved model prediction accuracy and mitigated issues arising from excessive input. *For more details, please download the PDF or feel free to contact us.

We would like to introduce the high-speed design of glass molds using the data analysis and modeling software "DTEmpower." The current mold design process involves continuously adjusting the design mold B to create a mold that meets the requirements. After obtaining the necessary glass model, there is a desire to establish a flow that allows for the direct and rapid design of the appropriate mold. Ultimately, we obtained deviation data between the glass model A that meets the design requirements and mold B, and learned the mapping relationship. We provided design methods such as constructing a data model. [Background and Issues] - The current mold design process creates a mold that meets the requirements by continuously adjusting the design mold B. - There is a desire to establish a flow that allows for the direct and rapid design of the appropriate mold after obtaining the necessary glass model. *For more details, please download the PDF or feel free to contact us.

We would like to introduce a case study of applying the data modeling and analysis software "DTEmpower" to transformer winding temperature warnings. The machine learning method allows for a more sensitive detection of abnormal data points in winding temperature simply by setting the difference between the temperature measured by sensors and the temperature estimated by the model. Additionally, early warnings based on machine learning only require setting the degree of deviation from normal values, which essentially establishes a dynamic early warning zone. This approach offers greater flexibility and improved reliability compared to traditional static warning bands. 【Problems and Challenges】 - Ensuring the stability and reliability of transformers is a critical issue, and responses to failures need to be swift and effective. - The main cause of transformer failures is the decrease in insulation capacity. - To mitigate the risk of transformer failure due to decreased insulation capacity, early warnings for winding temperature are necessary. *For more details, please download the PDF or feel free to contact us.

We will introduce a case study that applies the data modeling platform "DTEmpower" to accurately predict electricity usage and support smart power supply. This product features a time series forecasting function, which allows for the construction of data-driven electricity usage prediction models. Using the necessary data for forecasting, we assist in making predictions about future electricity usage. The benefits of the forecasting model include the ability to predict changes in electricity usage in advance, as well as improvements in the economic efficiency of the power system and social benefits. 【Flow for Building the Forecasting Model】 ■ Setting Timing Variables ■ Preprocessing Time Series Data *For more details, please download the PDF or feel free to contact us.

We will introduce the technical analysis work related to bearing parameter alarms and gearbox fault diagnosis based on the data modeling software "DTEmpower." This product can provide deep data analysis necessary for industrial data processing as a concise and rigorous one-stop solution. Additionally, it enables data analysis, modeling, and design based on machine learning, significantly improving product development efficiency. 【Condition Monitoring & Parameter Alarms】 ■ Extraction of Sensory Feature Characteristics ■ Quantitative Optimization of Sensitive Feature Alarms *For more details, please download the PDF or feel free to contact us.

Based on the issues in the de-NOx system of power plants, we will introduce solutions using the data analysis and modeling software "DTEmpower," as well as optimal control of the system. The de-NOx system has significant thermal hysteresis effects, making it difficult to construct physical and chemical models. Therefore, a control method that combines "predictive control," "optimal control," and "feedback correction" utilizing machine learning based on datasets is effective. As a result of applying data analysis, we can accurately and quickly predict changes in NOx concentration within the de-NOx system through machine learning, providing a reduction in the impact of the boiler's thermal hysteresis effect and optimizing rational system control. [Issues with the de-NOx System] - Unable to respond quickly to fluctuations in nitrogen oxide (NOx) concentration. - It is necessary to excessively spray ammonia at the flue gas outlet to reduce NOx concentration. - Excessive spraying of ammonia can lead to dust accumulation on the catalyst and clogging of the air preheater, resulting in decreased boiler operating efficiency and increased operational costs. *For more details, please download the PDF or feel free to contact us.

Wind turbines are mainly composed of parts such as blades, pitch control systems, gearboxes, generators, yaw control systems, and hubs. The hub connects the base of the blades to the main shaft of the wind turbine, and the blades experience complex alternating loads such as thrust, torque, and bending moments. Speed is transmitted from the hub to the main drive system through pitch bearings. Therefore, it is necessary to strictly manage the strength and lifespan requirements of the hub throughout the wind turbine. *For more details, you can view the related links. For further information, please download the PDF or feel free to contact us.*

The in-house developed data modeling software DTEmpower is a data modeling platform designed for industrial users, developed by delving into the data modeling needs of industrial companies. DTEmpower offers a wide range of algorithms for data modeling, including data cleaning, feature extraction, feature selection, and model training. It enables the improvement of model quality while reducing the need for user experience through algorithm development for specific cases, an intelligent scheduling engine, and super reference optimization. DTEmpower is equipped with a complete set of diagnostic solutions for failure diagnosis of turbomachinery. *For more detailed information, please refer to the related links. For further inquiries, feel free to download the PDF or contact us.*

Fuel cells are devices that directly convert the chemical energy of fuel into electrical energy and are expected to be the fourth generation of power generation technology, following hydro, thermal, and nuclear power. This electrochemical energy can supply the direct current power needed for external loads, and fuel cells do not emit nitrogen oxides or sulfur oxides. They have high power generation efficiency, a wide range of fuel options, low environmental pollution, and high reliability. In this case study, we will introduce rapid temperature field prediction using data analysis based on issues in the development and design of fuel cells. *For more detailed information, please refer to the related links. For further inquiries, feel free to download the PDF or contact us.*

Using the general-purpose data analysis and modeling software DTEmpower, we will introduce data modeling for hull resistance. When utilizing simulations in hull design, the time cost per case tends to be high, while the computational resources available are limited. Therefore, designers need to consider alternative performance evaluation methods that are not simulation-based in order to accelerate the optimization iterative process. *For more detailed information, please refer to the related links. For further details, you can download the PDF or feel free to contact us.*

Using the general-purpose data analysis and modeling software DTEmpower, we will implement data-driven design for hull lines. The design of hull lines is a crucial aspect of ship design, significantly impacting the technical performance and economic efficiency of vessels. Traditional ship shape design methods involve manual design, followed by a development process that includes CAD-assisted design and CFD-based technical evaluation. However, it has been believed that a more efficient and intelligent workflow can be established. By leveraging a data-driven intelligent design optimization platform, we can efficiently and quickly find the "appropriate" design targeting performance indicators. *For more detailed information, please refer to the related links. For further inquiries, feel free to download the PDF or contact us.*

Using the general-purpose data analysis and modeling software DTEmpower, we will conduct a data-driven strength evaluation of the reusable spacecraft wall structure. Reusable spacecraft are a crucial tool for enabling transportation between space and the ground, and the evaluation of the load-bearing capacity of wall structures, as typical load-bearing components, is significantly related to the overall safety performance of the spacecraft. However, traditional finite element methods often require over 100 minutes of computation time for a single buckling analysis of an enhanced cylindrical exterior, leading to high costs. *For more detailed information, please refer to the related links. For further inquiries, feel free to download the PDF or contact us.*

The descent rate of an aircraft during landing significantly exceeds the normal landing descent rate, resulting in excessive landing impact loads and a reduction in the strength margin of the landing gear and fuselage structure. Landing loads are related not only to mass but also to the pull from the rotor during the flight state at landing and the structure of the landing gear itself. Conventional methods for predicting aircraft landing loads require spending hours on dozens of simulations under operational conditions, making it impossible to quickly and accurately obtain design solutions that meet the requirements. *For more detailed information, please refer to the related link. For further details, you can download the PDF or feel free to contact us.*

In March 2024, China drilled a 10,000-meter oil well in the center of the Taklamakan Desert, making it the first of its kind in the country and the second in the world. It is difficult to ascend into the sky, and even more challenging to descend to the ground. For aerospace, an altitude of 10,000 meters is significant, but when it comes to going underground, 10,000 meters can be said to represent the limits of human technology. For every 100 meters dug into the ground, the temperature increases by about 2°C, and pressure also rises. At a depth of 10,000 meters, one would be exposed to temperatures exceeding 200°C and pressures exceeding 130 MPa. *For more detailed information, please refer to the related links. For further details, you can download the PDF or feel free to contact us.*