Mtell Reservoir a high performance repository for time-series data, maintenance and operational events, and other relationship data.

Transcription

1 Mtell Summit comprises a suite of foundation Mtell applications including Mtell Reservoir, Mtell CloudSync, Mtell Previse, and Mtell View. Summit allows open access to third party applications, including its open repository, to combine time synchronized data from any source. While Mtell tools provide analysis and learning, open API s allow delivery of any raw data and computational results into diverse client applications for display, reporting, or extended analysis. Mtell Summit is the premier remote monitoring center application for gathering and combining time-series and descriptive, relationship data for complete analysis and benchmarking. Mtell Previse the state-of-the-art toolset for machine learning-based analysis of related time-indexed data for preparing Mtell s pattern recognition monitoring agents. Also supports general purpose learning of complex behavioral data patterns for many optimization and decision-making initiatives in any organization. Mtell View Mtell View fuels situational awareness for personnel remotely overseeing equipment at diverse locations. Early alerts combined with aggregation and correlation effectively steer investiga tions and root cause analysis by equipment type, usage, and across sites. Mtell Reservoir a high performance repository for time-series data, maintenance and operational events, and other relationship data. Mtell CloudSync and other third party data ingestion tools provide controlled, easy connectivity and input from disparate sources. Third Party Extensions through open API s Mtell Summit permits other applications to enter data into the repository. API s also allow the extraction and processing of data in contemporary client applications such as R, Mathematica, Matlab, Apache Spark, etc.

2 Mtell Reservoir a new caliber historian fully meeting the demands of the enterprise: BIG data, massive data retention, rapid data delivery to innumerable users, and diverse applications. Reservoir s fresh design meets those requirements by fully leveraging processing capabilities of new software and hardware with extraordinarily inexpensive, distributed storage. Scalability Plus Performance Mtell Summit enables BIG data scalable to thousands of sites, millions of assets, with billions of sensors, and trillions of sensor readings. Multiple hardware nodes assure outstanding disk input/output and CPU processing capability. Federated views of equipment across multiple sites deliver enhanced asset health monitoring, and facilitate remote maintenance workprocess across many locations. The diagnostic capability, accuracy, and the range of condition monitoring using simultaneous machine learning across many machines at many locations are all increased using Mtell Summit. Extensive analysis and data delivery into other applications extends the use cases for Mtell Summit. Storage for all sensor time-series data Federated views across multiple manufacturing sites Local and remote data center synchronization Power to process large datasets Foundation for predictive analytics: Mtell Analytics plus third party reporting and analysis tools including R, Mathematica, Matlab, Apache Spark, etc. Scalability to multi-cpu clusters for: - Increased data processing requirements - Faster disk I/O operations

3 Mtell Previse The plant floor model of Mtell Previse is extended into Mtell Summit for extra duties on much larger (federated) data sets from multiple sites. Summit also supports additional analysis techniques for other optimization and decision-making services that can extend across diverse manufacturing processes equipment at many locations. Transfer Learning is a key capability, where Mtell Summit learns on one machine and transfers that learning in the form of pattern signatures to monitoring Agents on similar machines at other locations. Mtell Summit unlocks a further advance in retaining and sharing knowledge across fleets or pools of equipment. Mtell calls this process Population-based Learning where Mtell Summit combines group analysis and learning of behavioral patterns from similar processes and equipment, regardless of where they are located. Summit aggregates all the sensor information for groupings of similar equipment to massive sets in Mtell Reservoir. Internally, deep learning extracts the patterns of operations and failures, learning the shared behavioral characteristics of the entire set at the same time. Mtell Agents produced this way provide a new level of accuracy of pattern recognition, with only limited labeled data requirements. Such Agents are readily shared across the set members even if they are located at different sites and with different customers. Starting from day one, newly installed equipment of the same type, sensors, and usage can be equipped with Agents for monitoring normal and failure behavior that were prepared from older working equipment. By gathering all that time-series data into a great big storage, many other things are possible and desirable. A BIG data reservoir serves as the storage and source of all related data that is connected by timestamps. Additional data such as notes, work orders, photographs, videos, etc., can be inserted into the archives to be readily accessible whenever a user calls up a relevant historical trend. The Mtell Reservoir BIG Data sets allow analysts to perform ad hoc discovery, organization, and enrichment to prepare data for other analytical tools, reports, and dashboards. Remotely connecting operations and maintenance systems facilitates the highest performing assets at the lowest risk, and best financial performance.

4 Mtell View bundled visualization application Mtell View delivers contextually developed data about the performance and failure characteristics of assets and process equipment. At the enterprise level, Mtell View provides an intuitive navigation scheme that quickly alerts users, guiding them rapidly and effectively to important and prioritized information. Federated views allow subject matter experts (SME s) in a remote monitoring center to oversee equipment at diverse locations simultaneously. All information about any assets including condition-based alerts, maintenance work orders, and Agent properties, is aggregated and correlated in views highlighting situational awareness. Heat maps give extremely visual ways to show concentrations of specific degradation and failure in many dimensions. Analysts can quickly perform investigations and root cause analysis by location, across sites, across asset groups/fleets, by failure mode, equipment type, customer, usage. Consequently, the failure profiles and associated risk are immediately evident. Such clarifying views are essential to owner-operators, remote service providers, and original equipment manufacturers who wish to monitor and manage distributed assets from a central location. Mtell Reservoir is the full function enterprise storage for all time synchronized data. Mtell Reservoir At the enterprise datacenter, a new caliber of repository for historical data retention and delivery must meet the needs of more users, diverse applications, and emerging BIG data applications. Mtell Reservoir replaces and extends contemporary time-series historians to leverage enormous advances in computer hardware and software. For example, Mtell Reservoir recorded total data ingestion rates at 100 million points per second on a modest 4 node Hadoop cluster and scales almost linearly with additional hardware. Sites including fleets of equipment CloudSync sophisticated transfer Mtell Reservoir large volume data complex processing Mtell Reservoir leverages the Hadoop and OpenTSDB (time-series database) software technology. The Apache Hadoop software library allows for load-sharing by distributing the processing of large data sets across clusters of computers. Hadoop scales from a single server to thousands, each offering local computation and input/ output storage. Additionally, the OpenTSBD is a data management framework designed specifically for handling time-synchronized and indexed data. Implementing the Mtell Reservoir on Hadoop with OpenTSBD provides large improvements over traditional plant historians, especially for retrieval and display of very large data sets. Additionally, Mtell Reservoir facilitates specific maintenance process library functions, general purpose archiving, and information comparison functions, including report generation and third-party analysis.

5 Mtell CloudSync Mtell CloudSync provides extremely high data ingestion rates. CloudSync connectivity streams data from plant historians, but also accepts specific batch uploads of comma-separated value (CSV) files and custom developed data input services. Its elegant and sophisticated bi-directional architecture ensures CloudSync performs stream-based processing across challenging and bandwidth limited network connections such as satellite links. Transmitted streams include sensor data values, alerts, events, and maintenance activities. Automatic, lossless data compression means more efficient data transfers, and dynamic throttling keeps transfer within configured bandwidth limits. Signal prioritization assures the most pertinent data are received first, and the system will recover older data as bandwidth becomes available. CloudSync also delivers machine learning signatures from Mtell Summit into Autonomous Agents at remote sites. Focus on the future, not the past. A powerhouse tool set assures Mtell Summit predicts what can happen and advises the action to avoid it. Asset Health Monitoring Mtell Summit provides comprehensive asset health monitoring and analysis for myriad machines at multiple locations. Facilitating the 3 P s of Maintenance Analytics 1. Performance where information is readily available in views, charts, and trends to examine current and past asset performance. 2. Predictive where the solution can detect and warn of impending equipment failures well before serious damage occurs, including root cause analysis and process defect profiling. 3. Prescriptive offering key advice on mitigation, ordering inspection or repair, along with advice on avoiding impending issues.

6 Remote & Predictive Asset Health Monitoring Comprehensive remote maintenance monitoring has been elusive; promised by many but never really delivered. Other solutions provide simple graphics indicating present and past asset behavior and require remote personnel to poke around in an attempt to discern problems. Such limited remote center functionality offer little more than post incident phone support. Mtell Summit changes all that! Real, accurate real predictive failures warnings generated at the manufacturing location and/or at the Data Reservoir. Mtell provides precise, early warnings of impending issues alerting WHEN and WHY a failure would occur, and advising the appropriate prescriptive action to take, well before damage sets in. Now, the Mtell agent-based software assures subject matter experts at a remote operating center can oversee equipment located at many sites. They are forewarned, can investigate and take immediate action to alleviate the issue by recommending adjusted production, arranging minor servicing, and may avoid the problem altogether. The remote asset health worker can use the Mtell Reservoir for many horizontal analytical tasks that improve both manufacturing process and equipment efficiency, including: Reporting and Analysis Investigation of trends Profiling failure signatures across pools of similar equipment Root cause analysis Sub-component analysis Process analysis that permits review of multiple sensor data streams across time Equipment benchmarking; enabled by managing nameplate and brand information across equipment to compare variances in performance affected by location, usage, etc. Batch/discrete process analysis, including comparing behavioral patterns across multiple batches Signature Analysis Investigation comparison of signatures across time and across similar equipment Anomaly detection and conversion of anomaly failures into more precise signatures that detect far earlier than anomalies Event signatures including investigation of hidden failures Failure signatures the most precise and earliest way to detect degradation Efficiency signatures the inverse of degradation, inquiring why some equipment operates more efficiently Process defect signatures similar to failure detection on equipment, where the focus is on discrete events such as a batch processing where variations can occur across batch runs

7 A Time-synchronized Data Repository for any Use The content of the Mtell Reservoir is not limited to cross-site federated views and machine learning for asset health management. Time-series sensor data are lightly governed before ingestion; a cleansing procedure assures all data points are valid and within range before ingestion and machine learning. Consequently, the rich content is available for many business users to explore, combine with other data, build reports. Data can be extracted and processed by alternative client applications, or structured data warehouses, to answer questions that have not been possible or practical in the past. An Information Library The Mtell Reservoir is more than an elevated plant historian with a bump in CPU speed. First it is the scalable high performance industrial big data reservoir for time-series sensor data streams and event records. Second, it is a library containing a whole host of information about assets and asset performance; mappings to equipment models, sensor templates for fleets of similar equipment, and a failure library based on the ISO standard. Reporting and Analysis Process performance Equipment performance Model types Site performance benchmarks Equipment benchmarks Breakdown by location, operator, etc. Industrial Machine Learning Library Deep learning Signature library Sensor knowledge base Industrial Library Industrial signal processing Feature extraction Hierarchical sensor model Virtual sensing Sensor upset detection Shift change / Maintenance Performance logs Universal plant model Equipment taxonomy Nameplate parameters Linked to sensors Sensor ranking/importance Mtell Summit has been implemented successfully across several industrial sites. Reliability engineers can use Mtell Summit for maintenance workflow management, and can also generate and share unlimited machine learning agents at sites across the globe. Mtell Summit is the essential solution that merges process and maintenance data from multiple locations and presents them in rich, intuitive graphical displays in web browsers, tablets, and smartphones.

8 Cutting Edge Features CloudSync Remote Site Connect Multi-Site Tag Namespace Analysis & Reporting On-ramp to big data, sends data from remote facilities Configurable bandwidth throttling State-of-the-art compression and security Supports live streams and batch-oriented backfill Resilient synchronization with store-and-forward and system restarts Equipment Model Extensible equipment model library & taxonomy Failure Code Library based on ISO Automated import of equipment hierarchy and taxonomy from EAM/CMMS systems Mapping of sensor values into equipment templates For unique names from multiple remote plant historians Equipment types, associations, and sensor groups Big Data Scalability Leverages Hadoop and multiple editions including MapR, Cloudera, and Hortonworks Scalability add nodes for extra CPU and I/O capacity OpenTSDB for flexible management of time series data World Class Performance Low-latency response times for analytical applications Mixed workloads: quick trends to machine learning Drag-and-drop report generation Key filtering and management for all reports based on equipment hierarchy (site, equipment, sensors) and sensor templates (models, sub-assemblies) Correlates sensor streams with maintenance work orders, operator actions, and production activities/events Mashup trends overlay events on sensor signals Industrial Strength Processing Detects issues in sensor reliability & calibration issues Intelligent interpolation using uni/multi-variate techniques Incorporates virtual sensors, rules, and calculations Mtell Summit provides high performance, industrial strength, predictive analytics for enterprise-wide decision making Hotel Circle North. Suite 120. San Diego, CA (619) Mtelligence Corporation (dba. Mtell). All Rights Reserved. MTL-134