| Artificial Intelligence and it's Practical Application in | | | | Vendors typically don't like to refer to their AI |
| the Manufacturing Environment | | | | based scheduling applications as AI due to the fact |
| As the manufacturing industry becomes | | | | that the phrase has some stigma associated with |
| increasingly competitive, manufacturers need to | | | | it. Buyers are perhaps reluctant to spend money |
| implement sophisticated technology to improve | | | | on something as ethereal sounding as AI but are |
| productivity. Artificial intelligence, or AI, can be | | | | more comfortable with the term "constraint |
| applied to a variety of systems in manufacturing. | | | | based scheduling". |
| It can recognize patterns, plus perform time | | | | |
| consuming and mentally challenging or humanly | | | | Constraint-based scheduling needs accurate data |
| impossible tasks. In manufacturing, it is often | | | | A good constraint-based scheduling system |
| applied in the area of constraint based production | | | | requires correct routings that reflect steps in the |
| scheduling and closed loop processing. | | | | right order, and good data on whether steps can |
| AI software uses genetic algorithms to | | | | be parallel or whether they need to be sequential. |
| programatically arrange production schedules for | | | | The amount of thorough planning that is required |
| the best possible outcome based on a number of | | | | for a successful system to be launched is one of |
| constraints, which are pre-defined by the user. | | | | the largest drawbacks. |
| These rule-based programs cycle through | | | | If a management team has not defined and |
| thousands of possibilities, until the most optimal | | | | locked in accurate routings in terms of operation |
| schedule is arrived at which best meets all criteria. | | | | sequence and operation overlap, and if it has not |
| Another emerging application for AI in a | | | | correctly identified resource constraints with |
| manufacturing environment is process control, or | | | | accurate run and set-up times with a correct |
| closed loop processing. In this setting, the | | | | set-up matrix, what it winds up with is just a |
| software uses algorithms which analyze which | | | | very bad finite schedule that the shop cannot |
| past production runs came closest to meeting a | | | | produce. Tools like AI should not be thought of as |
| manufacturer's goals for the current pending | | | | a black box solution, but rather as a tool that |
| production run. The software then calculates the | | | | needs accurate inputs in order to produce a |
| best process settings for the current job, and | | | | feasible schedule that can be understood by the |
| either automatically adjusts production settings or | | | | users. |
| presents a machine setting recipe to staff which | | | | |
| they can use to create the best possible run. | | | | Constraint-based scheduling within an ERP |
| This allows for the execution of progressively | | | | (enterprise resource planning) system |
| more efficient runs by leveraging information | | | | In selecting a solution, there are a number of |
| collected from past production runs. These recent | | | | system prerequisites that you need to look for. |
| advances in constraint modeling, scheduling logic, | | | | The better an enterprise application integrates |
| and usability have allowed manufacturers to reap | | | | various business disciplines, the more powerful it |
| cost savings, reduce inventory and increase | | | | will be in terms of delivering constraint based |
| bottom line profits. | | | | scheduling. This means that if an application suite |
| | | | offers functionality cobbled together from |
| AI - A brief history | | | | different products the manufacturer has |
| The concept of artificial intelligence has been | | | | purchased, it may be harder to use that suite to |
| around since the 1970s. Originally, the primary goal | | | | deliver good scheduling functionality. This is |
| was for computers to make decisions without | | | | because a number of business variables that |
| any input from humans. But it never caught on, | | | | reside in non-manufacturing functionality can |
| partly because system administrators couldn't | | | | affect capacity. |
| figure out how to make use of all the data. Even | | | | When an ERP package has been configured for |
| if some could comprehend the value in the data, it | | | | constraint based or finite scheduling, it is generally |
| was very hard to use, even for engineers. | | | | routed to a scheduling server which calculates |
| On top of that, the challenge of extracting data | | | | start and finish times for the operations with |
| from the rudimentary databases of three | | | | consideration to existing orders and capacity. |
| decades ago was significant. Early AI | | | | When the shop order is executed, the scheduling |
| implementations would spit out reams of data, | | | | system updates the information regarding |
| most of which wasn't sharable or adaptive to | | | | operations and sends the results back to the |
| different business needs. | | | | enterprise server. |
| | | | Scheduling functionality within an ERP solution |
| The resurgence | | | | ought to work in a multiple-site environment. Let's |
| AI is having resurgence, courtesy of a ten-year | | | | say you need to calculate a delivery date based |
| approach called neural networks. Neural networks | | | | on a multi-site, multilevel analysis of material as |
| are modeled on the logical associations made by | | | | well as capacity throughout your whole supply |
| the human brain. In computer-speak, they're | | | | chain. The system should allow you to plan given |
| based on mathematical models that accumulate | | | | all the sites in your supply chain and the actual |
| data based on parameters set by administrators. | | | | work scheduled for each of those work centers. |
| Once the network is trained to recognize these | | | | Manually or automatically, you should be able to |
| parameters, it can make an evaluation, reach a | | | | schedule work and immediately give your |
| conclusion and take action. A neural network can | | | | customer a realistic idea of when the order will be |
| recognize relationships and spot trends in huge | | | | completed. |
| amounts of data that wouldn't be apparent to | | | | |
| humans. This technology is now being used in | | | | More benefits of AI, constraint based applications |
| expert systems for manufacturing technology. | | | | Apart from the immediately apparent capacity |
| | | | management benefits of constraint based |
| Practical application in the real world | | | | scheduling, there are a number of less obvious |
| Some automotive companies are using these | | | | analytical capabilities. Scheduling functionality |
| expert systems for work process management | | | | typically allows you to conduct predictive analyses |
| such as work order routing and production | | | | of what would happen if certain changes are |
| sequencing. Nissan and Toyota, for example, are | | | | made to an optimized schedule. So if a plant |
| modeling material flow throughout the production | | | | manager is pressured by a particular account |
| floor that a manufacturing execution system | | | | executive to prioritize an order on behalf of a |
| applies rules to in sequencing and coordinating | | | | customer, that plant manager can produce |
| manufacturing operations. Many automotive plants | | | | excellent data on how many other orders would |
| use rules-based technologies to optimize the flow | | | | be late as a result. Furthermore, this functionality |
| of parts through a paint cell based on colors and | | | | can provide predictive analyses on the effect of |
| sequencing, thus minimizing spray-paint | | | | added capacity in the plant. This enables |
| changeovers. These rules-based systems are able | | | | manufacturers to see if equipment purchases will |
| to generate realistic production schedules which | | | | truly deliver an increase in capacity, or if it will |
| account for the vagaries in manufacturing, | | | | simply result in a bottleneck further downstream |
| customer orders, raw materials, logistics and | | | | in the manufacturing process. |
| business strategies. | | | | |