Electronics factory

Factory operational data analytics

Shandong City Commercial Banks Alliance

Bank IT log search and analytics

Guilin Bank

Bank transaction link analytics

Agricultural Bank

IT Network packet traffic analysis

Huadian Corporation

Energy distribution system analytics

Shandong City Commercial Banks Alliance

Bank IT network security portrait

China National Petroleum Corporation

Email activities and risks analytics

MLP used in electronics manufacturing assembly lines at electronics factory - a case brief

MLP monitors laser welding machines, robots, several types of quality checking machines, and PLCs. MLP collects streaming operational data and logs in real time. Dozens of types of semi-structured data streams are collected continuously. Hundreds of data fields are extracted and processed from the semi-structured data.

Value created by MLP:

  • Production yield optimization
  • Production efficiency optimization
  • Timely fault detection for the machines
  • Prediction of laser lamp failure using machine learning
  • Descriptive analytics of the production
  • Real time dashboards of the operation

MLP used in analyzing and optimizing the operation of city-wide energy grid at Huadian - a case brief

Huadian corporation is a large energy conglomerate with 7 publicly traded subsidiaries, and a revenue of $30B.

At Huadian Corporation's city-wide home heating power grid, MLP monitors power plant, heat exchangers and associated equipments' data. Values such as pressures, temperatures, flow rates, electricity power consumption, etc. are captured in real time, along with weather information. MLP analyzes the data, and applies machine learning algorithms to determine optimal operating conditions. It also forecasts energy needs for the next several days based on weather forecast data

Value created by MLP:

  • Determine optimum operating conditions to minimize energy cost for the required heat
  • Predict fuel requirement for future so as not to waste burning more fuel than needed
  • Detect faulty equipment, and predict failure for preventive maintenance
  • Detect inefficiency progressing in heat exchangers and suggest repair
  • Provide complete data analytics for operators and management

case 1:

MLP used in the consolidation and analysis of all of IT logs at SCCBA - a case brief

SCCBA consolidates IT operations of 16 commercial banks in Shandong province

At SCCBA, MLP collects operational logs from servers, applications, network devices, and business transaction systems (order of TB/day). The logs are parsed using MLP's parsing engine to extract hundreds of key operating parameters for further data analysis and machine learning. The logs are also indexed into MLP's search engine

Value created by MLP:

  • Parse, store and archive IT logs for full text search capability to troubleshoot operational problems and to fulfil government requirements
  • Provide insights into bank's transaction operations, information security, network operations, server operations, application performance, capacity planning, etc.
  • Provide a one-stop application for all of IT operational information
  • A dashboard to view in the operations center

case 2:

MLP used in internal security posture monitoring at SCCBA - a case brief

At SCCBA, MLP uses machine learning to create a portrait profile of internal TCP connection pairs. It uses firewall logs to get the connection information. It looks for new connections, long and short duration connections, and lost connections. Then, any abnormalities are flagged and made into an alert. The connections are also color-coded based on their behavior changes.

Value created by MLP:

  • Internal security of accesses from various member banks into the central IT services and application is monitored
  • By the application of machine learning, changes and abnormalities are detected for security threats and policy violations
  • Now IT security professionals have visibility and control over the security of internal banking applications and their client accesses

MLP used to analyze banking transaction applications at Guilin Bank - a case brief

Guilin bank is a large regional bank in China, located in Guilin in the southwestern region.

At Guilin Bank, MLP is deployed to collect network, security, systems, applications and transaction events and logs. The transactions originate from teller terminals, mobile phone banking, web banking, ATM, and POS systems. MLP uses its parsing and in-memory stream analytics capability to extract key IDs and other parameters, stitches the transaction steps together, and reconstructs the transaction flows. Then conducts further statistical analyses, and machine learning based classification, and anomaly detection.

Value created by MLP:

  • Provides transaction behaviors, trends and statistics in numerous descriptive analytics
  • Using machine learning, detects anomalies of transaction delays and faults, and helps find their root causes in the infrastructure
  • Provides large screen display for the operations center

MLP used to analyze worldwide email messaging at CNPC - a case brief

CNPC is the fourth largest company in the Fortune Global 500 list, with 1.6 million employees and $263B revenue

MLP monitors all email messages passing through all email servers of CNPC at various locations worldwide. There are dozens of internal email domains and millions of external domains. It extracts over 20 features from each email message, applies multidimensional unsupervised machine learning technology, and determines abnormal email activities and risks to the company.

Value created by MLP:

  • Complete behavior analysis of internal as well as external email users, email domains and email servers
  • Find high-risk email users that may pose a threat to the company
  • Find security threats such as data leak, malware, etc.
  • Keep records of all email activities and provide executive reports
  • Provide full text search capability on historic data for forensic analysis

MLP used for network traffic analysis at ABC - a case brief

Agricultural Bank of China is a large national bank with assets of 13.24 Trillion Yuan and revenue of 421.96 Billion Yuan.

At Agricultural Bank of China, network traffic connections were analyzed using MLP's packet decoder. The connections were reconstructed and protocols were decoded. It includes TCP, UDP, HTTP, DNS, FTP, etc.

Value created by MLP:

  • Analyzed statistics including GeoIP location, DNS resolution, etc., for network department's visibility and planning
  • Detected anomalies in connection data by classification as well as multidimensional unsupervised machine learning in order to determine security and performance issues
  • Behaviors of users were analyzed in regard to the web sites they were visiting, queries they were performing on search engines, etc. for the benefit of employee monitoring
  • Anomalies were detected in DNS traffic for security purposes
Copyright © 2018 Yosemei. All rights reserved