Machine thinking “Out of Box”

By: Reader

Following the success of industrialization and automation, artificial intelligence needed to be taught to make decision and provide operational commands to equipment and system. With the insights gathered out from the problem, Prescriptive analytics helps to derive solution and preventing issues before problem occurs. The results are shared with machine cognition for learning and also can be a methodology in deriving at “new” and “unconventional” solutions.

New issues on equipment, process and operation would provide unstructured data and input into the evaluation matrix. The unstructured data and information will have to be analyzed to understand the reason of such omission while revising the matrix and the control logic.

Logic control for equipment and processes.

By: Kock Chee Kiong, ARiMI-CERM, ARM.

General analysis can be performed on equipment and system using a matrix to assess and correlate frequency and severity of breakdown and failure in relation to the followings;

  • Probability
  • Correlation
  • Causation

The result obtained can be mapped for further assessment through Predictive Analytics.

Predictive analytics helps to determine the underlying issues and problems while developing actions plan during general analysis stage. Method such as Online analytical processing (Ref: data science creates teams that ask the right questions and deliver real values by Doug Rose, 2016, Apress) allows insight to be extracted from the matrix while processing the real time input and parameters.

General analysis and predictive analytics identify and evaluate the probability of issues leading to the problem, enabling issues to be articulated and helping to make a problem statement.

Prescriptive Analytics

Prescriptive analytics are paired to the equipment/machine and human user which allows autonomous control, creates an opportunity for collaborated learning and aids in problem solving. Operation control parameter and sensors will provide real time information and data into the control matrix, determining the peripheral of the system control. Prescriptive analytics further determines the overall system control, allowing decision to be made for intuitive control with real time information and parameters

Prescriptive analytics should be performed by human that operate the equipment and process with support from experts and computation power of technology in embracing mistakes (mission) and drive remedy afterthought and optimizing effectiveness of the solution.

Below are methods to be considered in applying Prescriptive Analytics.

  • Analyze the result using 80-20 rule (seeking for the list of possible issues to the problem).
  • Perform higher level analysis to narrow down on the numbers of leading issues to the problem.
  • Application for 90/10 rule – “Seeking the wisdom of the crowds” approach
  • Narrate the problem in a story mode by questioning the whole process and event. A correct story helps getting the right approach for problem solving. The Question and Answer seek alternate view and approach in delivering a same solution answer. This is similar when applying different algorithm and method in deriving for the same answer.

Matrix for Process Control and Equipment Maintenance.

By: Kock Chee Kiong, ARiMI-CERM, ARM.

Process automation helps and supports growth of the many manufacturing facilities.  With more manufacturing facilities that are equipped with cloud based operation and system control, physical control panels/controllers are no longer required on site. In many ways, technology compliments process automation with better operational control and fewer physical equipment.

Equipment and Process Control Matrix

Developing a matrix-based equipment maintenance and process control workflow provides operational logic control, allowing  issues to be mapped (using logic blocks) in a matrix. The matrix executes a control function after assessing the relationships between the logic blocks.

The equipment and process matrix helps to identify issues and to solve engineering/operational problem, providing intuitive control . Logic blocks would be a method to analyse issues and severity of the incident and problem, including issues that are previously not identified. This aids the development for equipment maintenance and process control operation strategy.

As we moving into Industry 4.0, emphasis will be placed on artificial intelligence (AI) to improve automated process control and reducing human intervention. Concerns on jobs security and loss of jobs for operators would arise. Before being fixated on the downside of the risk on Industry 4.0, human operator should be working together with AI and technology in creating a manufacturing environment with minimal downtime and allowing operation to be available in any part of the world.

Future of Engineering and Maintenance

By Kock Chee Kiong, ARM, ARiMI-CERM.

Operational and maintenance practices are aligned through policies and standard operating procedures developed by individual organization. Adhering to standard operating procedure helps facilities to operate normally and functioning optimally. Out of the normal issues such as breakdown and troubleshooting may not be explicitly identified and addressed in operating procedure. Problems and issues arise from equipment and process will then have to be tackled by experienced personnel or third-party vendor. Since there is a mixture of experiences existed among employees in the organization and manufacturing facilities, the more experienced and senior personnel will be relied upon to solve the problems when facing difficult issues in equipment and process operation.

Intuitive-based approach towards problem solving

Conventional method such as breakdown and failure mode analysis allow engineering team to identify root cause, but the resulting solution is usually retrospective. Generic engineering solutions maybe applied at times after issues and problems are surfaced out. The effectiveness to solve these problems will very much depend on experience of the personnel making such decision and the solution taken.

Engineering methodology has to be improved by moving towards an intuitive -based approach, rather than the conventional way of preventive, predictive maintenance and conditioning monitoring, troubleshooting. Algorithm derived from Big Data Analysis should be used to assess issues and problem with respective to the exiting situation/environment. The results from the analysis enable users to make better informed decision on formulating an effective solution to resolve issues faster.

The key advantage for such an apporach is the ability to resolve underlying issues and tackle the problem real time and in a timely and appropriate sequence. These are critical for engineering and maintenance personnel whose responsibility is to upkeep production and operation activities.

Intuition-based engineering for AI

The maturity of Artificial Intelligence (AI) and robot to have human intuition is still at an early stage, if human still have real concerns on safety when taking a driver-less plane, high speed rail or electric car.

Despite the advancement of neuro science and programming, intuition baes decision will require time to be developed and learnt by machine. AI would be analyzing bulks of data, making decision though computing and processing capabilities and doing the tasks with minimal or without human intervention. However, this process still lacks human intuition-based approach to solve issue.

What do I mean about human based intuition that AI and machine are lacking. A machine can extract and computes more data at a faster speed and would perform a similar task as the most experienced human operator. There would be no difference in executing this task for both and in fact, many would prefer technology and machineries over human. AI and machine can also pick up issues and alarms before linking these issues to incipient problems. If a problem without prior information and past encounters arises, could we trust AI to resolve the problem or you are betting on the human operator with vast experience but without knowledge on the new problem.

In solving technical issues and problems that are not predefined, human would make relation with experience he/she has, think about all possible root cause (qualifying the issues) and outcome (quantifying the approach/methodology in assessment) which is similar to a matrix-based analysis. For AI, it would be the same using the data and information that could be relevant for analysis.

Machines can learn the skillsets in human approach to solve problem in deriving with alternatives to solve the problems while also predicting list of issues to improvise or work on 1) should solutions failed or 2) the problems got out of control.

Working with vast information and skewed information with the input of intuitive assessment. This enables ignorant data to be analyzed for possibility.

Improve decision making by applying rules and factor application (weighted factor) – using fact-based application for decision making.

Developing the Intuitive-based approach

The formulation towards intuitive-based approach to solve engineering related problems would comprise AI and human user, working together in deriving cost-efficient action to be taken together with various stakeholder and third-party vendors. The workload for human operator work would also be reduced by using mobile devices which are programmed with routine maintenance procedure and equipment troubleshooting methodology.

Machine Intuition

By Kock Chee Kiong, ARM, ARiMI-CERM.

Routine work activities are being handled by machines with increased speed, precision and consistency. Problems and anomalies would show up along the way while work is being done. Resolving such problems promptly will depend on the skill set and experience of the personnel handling the machine to tackle these issues to the problem. It relies very much on experience and also collective agreement among co-worker/subject matter expert/supervisor for the decision to be made and the measure taken thereafter. There is still an “intuitive” gap between human and machine such as Artificial Intelligence (AI) on problem solving and application of solutions within short time span in engineering work.

Taught experience

The approach to solve problems lies in the action and decision of human, which is dependent on knowledge, past experience, gut feeling and intuition.  The development on big data analysis would help machine to make better informed decision. Machines learning should include solving issues that are not in databases and previously not analyzed. Algorithm involving predictive analytic and neuro-science would help machine and AI to develop “intuitive” solution. Continue reading “Machine Intuition”