The Internet of Things (IoT) accelerates the pace of data creation, in both type and volume. Some 90% of all digital data has reportedly been generated in the last two years. That is 2.5 quintillion bytes, or 2.5 billion GB, of data per day. It’s the equivalent of streaming 2.5 billion hours of standard-definition video on Netflix. Imagine the possibilities if a company used this enormous amount of data to drive insightful decisions.
Enterprises today, however, aren’t prepared to handle the amount of data they generate across different business segments. According to Forbes, nearly 80% of enterprises have little to no visibility into what is happening across their pools of unstructured data. This is text-heavy information, with a mixture of dates, numbers, and facts. Think of it as information thrown into an Excel spreadsheet without any organization of its columns, making it difficult to understand. Hackers can easily target this potentially sensitive information, especially since it lacks protection.
Industrial IoT (IIoT) data is growing faster than any other segment. Companies use only 3% of that data in a meaningful way. Many companies, such as those in electronics manufacturing services (EMS), still aren’t taking an insight-driven approach. When companies fail to implement data-analytics tools, they face risks, including the following:
- Inefficiency: Machines sit idle and labor goes unused because of improper tracking of shop-floor status
- Increased operational costs: Inconsistent supply-chain management and unnecessary scheduled downtime cause losses in production output
- Missed business insights: Insufficient diagnostics results in unplanned downtime, costly repairs, and poor customer satisfaction
Cultivating Big Data
The EMS industry is growing rapidly, and research indicates that it will exceed $650 billion by 2024. Driving that growth is increased demand across the globe for consumer and smart electronics devices to support wearables, automotive applications, and mission-critical IoT applications.
EMS is a competitive industry. It manufactures electronics components such as printed circuit boards, microelectronics, optoelectronics, radio-frequency devices, and wireless devices. The difference between a successful EMS company and one that’s struggling to sustain its business is the ability to provide turnkey services such as data-analytics capabilities.
If properly harnessed, big data can drive dramatic efficiency gains for enterprises, especially for EMS in the IIoT segment. With the introduction of intelligent sensors, mission-critical communications, and automation, EMS companies can harness big data collected from production, material, storage, and other management systems. However, big data is useless without some level of analytics to convert the data into something meaningful and actionable. Managed correctly and systematically, big data provides real-time machine status and enables better insights and decision-making.
Four types of analytics can help companies make sense out of data formatted for reporting: descriptive, diagnostic, predictive, and prescriptive.
1. Descriptive analytics: What is happening?
Descriptive analysis is the most common type of analytics. It helps users understand what's happening in an area of focus. One example is understanding the lifecycle of a product through an IoT chipset manufacturing floor, such as where and when the product was manufactured and by which machine. Another example is understanding the performance of a manufacturing line, such as yield, beat rate, and cost to run. A test engineer can use a visualization tool to present the data in a way that enhances its understanding.
2. Diagnostic analytics: Why did it happen?
Diagnostic analysis is the next step in data analytics. It looks at the available data to determine why something has happened. In problem-solving, diagnostic analysis identifies the cause of the problem. It’s common for analysts to review time-series data to create an analytics dashboard for the entire business. If an IoT chipset manufacturing line isn’t performing well, diagnostic analysis of data from multiple machines can help decipher why production dropped. For example, the root cause could be an incorrect machine parameter, a change in the factory environment, or an operator mistake. These causes become evident in a diagnostic analysis.
3. Predictive analytics: What is likely to happen?
Predictive analysis uses algorithms to study past data trends to predict what might happen in the future. For the IoT chipset factory, it’s about predicting when a machine might fail and scheduling preventive maintenance before that occurs. Another example is determining the optimal production-floor temperature that will allow machines to have the greatest amount of uptime.
4. Prescriptive analytics: What should I do about it?
Prescriptive analysis offers the highest value, but it’s also the most complex to administer. This model utilizes the understanding of what happened, why it happened, and other outcomes from predictive analysis to help determine the next set of actions. In the IoT chipset factory example, prescriptive analytics can work across machine performance data, operator skill-set data, and raw material data. The results enable the analyst to recommend actions that improve end-to-end manufacturing operations.
Depending on the stage of its development, a company may need one or another type of analytics. As it evolves, the company will progress from using simple descriptive analytics to employing more complex, predictive, and prescriptive analytics.
Reaping Big-Data and Analytics Benefits
Big data and analytics can help a company sort through an incredible amount of information and make decisions more confidently. EMS companies benefit from the use of industrial data and analytics in three key ways:
- Better product quality: An EMS company is able to analyze the vast number of intelligent sensor data from every test during each step of production. It can easily identify, isolate, and correct problematic outliers by viewing real-time trends. Analysis of historical data enables the EMS company to quickly identify the root cause of quality issues.
- Higher productivity and throughput: Yield and test time of various operations across global networks often experiences significant variability. With data analytics, an EMS company is able to isolate any variations in the operation. It can make changes to optimize the processes almost instantaneously. This helps ensure higher productivity and throughput across global operations.
- Improved asset utilization: Data analytics collects and analyzes environmental data via IoT sensors installed throughout the production process. An EMS company should leverage machine health and measurement data to predict equipment and fixture failures in real time. This eliminates the need for periodic maintenance and extends the mean time between failures, which ultimately leads to greater asset utilization.
A Better, Data-Driven Approach
The goal of most businesses or operations is to meet sales demands while increasing shareholder profit. Operations teams constantly seek solutions to maximize throughput, minimize downtime, and reduce expenses to manage the factory floor efficiently and increase profit margins.
Conventionally, EMS companies regularly perform preventive maintenance on manufacturing equipment, regardless of whether the tester has been running at full capacity. The downtime during this maintenance translates into millions of dollars in lost productivity. Insufficient regular diagnostics can overlook critical problems, ultimately resulting in unplanned downtime and costly repairs.
Instead, EMS companies should proactively analyze sensor and operational data from the factory and production line to mitigate and resolve production issues quickly and avoid interruptions. They can accomplish this using an Industry 4.0-ready data-analytics solution with powerful features to help enterprises realize the true power of industrial analytics.
The right data-analytics software can enable enterprises to find, visualize, and understand big data for business knowledge improvement. It should include features such as visualization tools, real-time asset monitoring, and advanced algorithms that predict and anticipate anomalies.
Janet Ooi is a lead in Keysight IoT Industry Solutions Marketing.