GE Proficy CSense

GE Proficy CSense is GE's monitoring and analysis software suite, providing business advantage through broad analytical capabilities and access to GE's cloud data storage. The release of the new product was announced by the press service of the company on July 8, 2013.

2023: U.S. authorities warn of dangerous holes in Siemens and GE industrial software

On January 17, 2023, the US Cybersecurity and Infrastructure Protection Agency (CISA) announced the discovery of dangerous vulnerabilities in industrial software (software) manufactured by Siemens, GE Digital and Contec. Read more here.

First for supermassives

• Proficy's software monitoring and analysis suite is the industry's first industrial data supermassive solution by combining Historic and Hadoop technology.
• The first integrated software solution to simultaneously maintain the health of core assets and optimize processes
• Companies can scale data collection from controller to enterprise, in the cloud, in a single solution

Total composition

Proficy's software package consists of six integrated products:

• Proficy Historic, GE's flagship data collection software product;

• Proficy Historic Analysis for advanced data analysis and visualization;

• Proficy SmartSignal for predictive analysis of monitoring systems status;

• Proficy CSense for process troubleshooting, process health monitoring, and closed loop process optimization. Two additional products are new in the company's offering and include Proficy Historic HD, which allows you to store especially large groups of data in the Hadoop cluster, as well as Proficy Knowledge Center, a console that provides process visualization with the ability to assess asset health and process optimization.

• Proficy Historic HD is the first solution to combine industrial data and analytics with the ability to create a supermassive data store based on Hadoop; this is the first step towards GE's goal of building an industrial internet. The software provides advanced analysis of industrial data supermasses to improve process efficiency, troubleshoot processes, and predict and prevent asset failures. Developed in collaboration with the GE Center for Advanced Software Solutions (Software Center of Excellence), Historic HD allows users to bring together the full range of archived time data, reducing storage costs by at least 85%.

• Proficy Knowledge Center is a model-driven, browser-based visualization application. By directly linking the fixed asset data model to archived time data, Proficy Knowledge Center supports unified modes for displaying current process and equipment states. The solution is based on Proficy Historic for efficient storage and display of a huge array of historical data. The results of the data analysis performed by Proficy CSense or Proficy SmartSignal are then rendered through the Knowledge Center in the context model, simplifying navigation and understanding.

Main purpose and tasks to be solved

Proficy CSense is an intelligent add-on to operational and historical production databases. Its key purpose is to use historical data to obtain information on the progress of technological processes, as well as to build functional links and generate statistical models of the dependence of performance targets (KPIs) of processes on the parameters of the technological mode of operation for a particular object. Based on the built models, this product allows you to simulate various modes of operation of objects without experimenting on real equipment. In addition, it is possible to solve a wide range of tasks for improving technological processes, among which we note the following:

• diagnosis and search for the reasons for deviations of process parameters from the desired values,
• equipment status monitoring tasks,
• building virtual sensors,
• objectives of stabilization and optimization of performance targets.

As a result, this product allows you to form solutions of various types, from off-line synthesis of control actions with the creation of active rules and recommendations for process adjustments to on-line solutions in advisor mode and the ability to create APC (Advanced Process Control). Thus, this product is a powerful device for building expert systems at production enterprises.

Methods and approaches used

Proficy CSense supports integration with a wide range of data sources:

• Microsoft Access, MySQL, SQL Server, Oracle databases;
• архивы OSI-Soft PI, Proficy Historian, Wonderware;
• APCS via OPC interface;
• various text data and user.NET formats

Fig.1 Proficy CSense Components

To process and prepare statistics for modeling, the package carries out correlation analysis of data, which is used both to select decorrelated input parameters of the model and determine key factors affecting KPIs, and to find time delays of KPIs from model inputs. It is also possible to conduct frequency analysis of data to determine the cyclicity of various indicators.

The construction of models here is implemented using modern machine learning methods. Moreover, the product allows you to build multidimensional models of varying degrees of complexity: both linear and highly non-linear. To build complex nonlinear dependencies, an apparatus of artificial neural networks is used. In addition, it is possible to program your models or add functions made in other applications. To solve optimization problems, a solver is integrated into the product, which allows solving problems with simple linear constraints of the interval type. Despite the complex mathematical apparatus used in the product, work in it is quite comfortable and does not seem laborious.

Рис.2 Импорт и объединение источников данных в TSh	Fig.3Interface of visual representation and processing of statistical data in TSh

Advantages

Proficy's monitoring and analysis software package can be used remotely, both to control an individual enterprise and to control all of the company's operations globally. As a result, it provides information to increase the availability of equipment and reduce unplanned downtime, increase production and reduce variability, and reduce maintenance and inventory costs. Proficy's monitoring and analysis suite has four key features:

• Improved data management: The platform is able to process and aggregate time and process data for authorized personnel anywhere with an Internet browser. Huge amounts of real-time data can be collected, archived and distributed at very high speeds and at very low storage costs: in the cloud. More data over longer periods of time can provide enhanced analysis and greater visibility into actual process efficiency versus optimal - providing a competitive advantage.

• Asset Improvement: Using advanced analysis and diagnostic tools, companies can monitor all critical fixed assets automatically and continuously, all 24 hours a day. The software "memorizes" the normal behavior of a fixed asset object and detects deviations by issuing warnings in advance about impending problems, diagnosing equipment and process failures, and prioritizing by importance. Early warning reduces emergency equipment failures and provides proactive maintenance instead of costly reactive.

• Process Optimization: Advanced analytical capabilities also help optimize processes. The product recognizes unknown correlations of process variables that may lead to reduced performance compared to optimal performance. This provides an alternative analysis for finding solutions that optimize processes - while improving the health of assets. Process optimization increases production while reducing deviations.

• Data visualization: The software package contains a network reporting tool to provide information on roles and assets, revealing the value of data in a specific context to the right person at the right time.

Composite Modules and How the Product Works

Software Product Proficy CSense consists of three main parts: Troubleshooter, Architect and Action Object Manager, together with Report Server (Fig. 1).

The Troubleshooter component is designed primarily for processing historical data and creating process plan models. The presence of a rich interface for visualizing and analyzing statistical information is one of the advantages of this application, which significantly helps in the preparation of historical data for modeling. Working in Troubleshooter consists of a number of intuitive steps:

Рис.4 Корреляционная матрица, спектральный анализ, статистика и определение задержек	Fig.5Model construction in TSh

• Import of data: in addition to integration with various sources, there is also an interface for combining databases (Fig. 2);
• Visualization and data preparation (Fig. 3) (trends, histograms, scatter plots, statistical characteristics, brushing, correlation and frequency analysis (Fig. 4), determination of time delays)
• Construction of models (Fig. 5) (nonlinear models, rule models, clear and fuzzy logic)
• Analysis of simulation results, determination of deviation causes, viewing of "what-if" scenarios and structure of "input-output" dependencies;
• Off-line optimization to assess the potential for possible process improvements; (Fig. 10)
• Building off-line solutions in the form of active rules and process control recommendations based on the previous steps.

	Model and Cause Analysis Window
Рис.6 Окно анализа моделей и причин отклонений в TSh	Figure 7 Example of a project in Architect

The Architect module provides simulations and on-line solutions (Fig. 7). In addition, it itself contains a fairly rich toolkit for multi-component modeling and optimization in real time. You can program your own models and objects, as well as integrate objects created in the external environment.

Action Object Manager & Report Server is a service for hosting and launching on-line solutions (Fig. 8) developed in Troubleshooter and Architect. Paired with the Report Server (Fig. 9), which provides a service for web reports and visual presentation of various analytical information, this module forms a Run-time shell of the original product.

Рис.8 Интерфейс размещения решений	Figure 9Report Server Sample Report

Results and Effects of Use

A significant effect from the use of Proficy CSense package can be obtained not only in those industries where a long archive of historical data is collected, but also where there is only an APCS level. If statistics show a large variability of processes and the task of stabilizing and optimizing objects is acute, as well as the possibility of active control of these objects, then the possible results in improving production efficiency can be very significant.

As an example, we will cite a pellet metallization plant at one of the Russian metallurgical plants, where a preliminary pilot project showed the possibility of optimizing the specific consumption of natural gas in the furnace by an average of 10-15% without deteriorating the quality of products (Fig. 10). Using the example of another industrial facility, it was possible not only to reduce the specific consumption of energy resources by an average of 7-8%, but also to stabilize the quality of the output product at the desired level (optimization of the specific consumption of water steam at the butadiene production plant, Fig. 11^[1]).

Рис.10 Оптимизация удельного потребления природного газа в печи металлизации окатышей	Fig. 11Minimization of steam consumption and stabilization of product quality in the butadiene production column

Notes