Implementing continuous improvement using genetic algorithms Petter Øgland

Implementing continuous improvement using genetic algorithms Petter Øgland, Department of Informatics, University of Oslo QMOD/ICQSS Conference, Verona, Aug 28 th 2009

Structure of presentation 1. 2. 3. 4. 5. 6. Introduction Literature review of CQI methods The new CQI method Example of new method in practical use Discussion Conclusion

Classical QMOD: Deming & Lewin Juran (1986): Plan, Control, Improve Deming (1986): Plan, Do, Check, Act Lewin (1950): Unfreeze, change, freeze

Unpredictable organizations where project-by-project approaches fail

Genetic Algorithms: Cultivate the flock rather than the individuals

Research questions • RQ 1: Is it possible to use the GA approach for effective QMS design? • RQ 2: If it is possible, why is it not used?

Structure of presentation 1. 2. 3. 4. 5. 6. Introduction Literature review of CQI methods The new CQI method Example of new method in practical use Discussion Conclusion

GA for understanding OD • Genetic Algorithms (GA) has been suggested for QM as a part of a more general Complex Adaptive Systems (CAS) approach (Dooley et al. , 1995; Dooley, 2000) • GA on a metaphorical level (Goldstein, 1993; Nelson & Winter, 1982) • Simulation models based on GA (Bruderer & Singh, 1996) • GA as integrated part of decision support systems (Greer & Ruhe, 2003)

GA for implementing TQM Embracing control (OR) Embracing chaos (CAS) Organizational Lewin (1950) development (OD) Goldstein (1993) Dooley (2000) Quality management (TQM) Imai (1986)? ? ? Juran (1964) Deming (1986)

Structure of presentation 1. 2. 3. 4. 5. 6. Introduction Literature review of CQI methods The new CQI method Example of new method in practical use Discussion Conclusion

Genetic Algorithm (Wikipedia, 2009) • • • Choose initial population Evaluate the fitness of each individual in the population Repeat until termination: (time limit or sufficient fitness achieved) – – Select best-ranking individuals to reproduce Breed new generation through crossover and/or mutation (genetic operations) and give birth to offsping Evaluate the individual fitnesses of the offspring Replace worst ranked part of population with offspring

Structure of presentation 1. 2. 3. 4. 5. 6. Introduction Literature review of CQI methods The new CQI method Example of new method in use Discussion Conclusion

Example: The KLIBAS system • 1991 -95 – Formal development project – High prestige, management commitment – Project “completed”, but nothing worked • 1996 -99 – – Informal maintenance cycle Low prestige, little management commitment Problems, complaints requests fixed as reported A practical and useful system develop through many small iterations

Process maturity in KLIBAS due to managing knowledge/power Development project Maintenance process On paper Systematic (managed by people) Chaotic In reality Chaotic Systematic (managed by computer)

QMS as CAS with automated Pareto analysis at the nexus SYNOP AWS: Automatic weather stations PRECIP: Manual precipitation stations e-mail METAR: Airport weather stations e-mail Pareto analysis e-mail Monitoring of system outputs and users (customer satisfaction) e-mail System monitoring UASS: upper air sounding stations HIRLAM: quality control by use of forecast data

GA implementation of daily maintenance & development Enter office on the morning of day i. Evaluate population: Real-time and nightly automatic data collection for total system by use of e-mail. Select solutions for next population: Run a Pareto analysis for setting the agenda for the day. This defines the population of processes to be improved. i: = i + 1 Perform crossover and mutation: Read, write, discuss; design Exit office in the and implement afternoon of day i. etc. ; the daily practical work of process improvement.

Productivity indicator

Structure of presentation 1. 2. 3. 4. 5. 6. Introduction Literature review of CQI methods The new CQI method Example of new method in practical use Discussion Conclusion

Is GA the same as kaizen? Kaizen GA Similarities Technical kaizen Social GA sounds like GA like kaizen (Imai, 1986) (Goldberg, 2000) Differences Social implementation (skills and attitudes) Technical implementation (following an algorithm)

GA is a SPECIAL type of kaizen • It is strictly mathematical (an algorithm), not dependent on intuitive or cultural skills • It is ”stupid” in the sense that each ant in a colony has a lesser brain than an elephant • It is ”unfocused” as it aims for many improvements at the same time • It is ”inefficient” as it progresses by trial and error

But it works!

Why others do not use this approach 1. People are unwilling to be run by computer 2. The GA approach generates complexity 3. It is “common knowledge” that the unfreezechange-freeze approach is the “one best way” 4. TQM personnel lack technical skills for understanding GA 5. GA makes TQM invisible and thus a poor choice when wanting work acknowledgement

Structure of presentation 1. 2. 3. 4. 5. 6. Motivation Overview of current CQI methods The new CQI method Example of method in use Discussion Conclusion

Conclusion • There are sociological reasons why people might reject the GA approach to TQM, although it WORKS and it is SIMPLE to implement • The GA approach seems well-suited for designing QMS bottom-up in complex organizations or as a TQM method for people who enjoy living in chaos

Thank you