AI projects for manufacturing companies
Automation at the push of a button? Unfortunately, not (yet) possible. Experience shows that it is a continous process to realise the potential of digitization. Our guest author Verena Fink from the consulting company Woodpecker Finch where and how artificial intelligence (AI) can help.
Manufacturing companies with industrialised processes are in the best position to automate up to 80 percent of their processes. Artificial intelligence can help make a big leap forward along this path. We are used to analysing, standardising, consolidating, implementing and executing along the way. By using AI, we can instead allow machines to evaluate our own wealth of experience and thus shortcut problem solving. How can this work and where is it worth starting?
Business AI potential
The application areas of AI in production are both, diverse and promising. The areas range from product development to quality management to process optimisation and logistics. AI technologies can be used successfully in the industrial production environment, especially if a sufficient database has been created in advance.
Thus, text, speech, image and sound recognition can play just as important a role in manufacturing and production as, for example, action planning or multidimensional pattern recognition. In the area of “Business AI”, European companies have significant potential to leverage AI power based on our strong industrial and IoT background. This means leveraging industry experience and letting AI systems support and automate existing processes. Which areas are particularly relevant?
Product development and process optimisation with AI
AI can help manufacturing companies in various stages of product development. Processes can be simplified, scaled and accelerated. Pattern recognition can help evaluate test and simulation data more efficiently. At the same time, planning and optimisation algorithms can accelerate development processes.
AI-based quality management
Quality management is essential in manufacturing companies to ensure the highest product quality. But manual quality controls are costly and difficult to scale. That’s why AI-supported methods are increasingly being used in quality management. Computer vision and machine learning in particular help in the early detection of product damage, quality variations or quality problems. As a result, AI saves time and costs in production and ensures the required product quality.
Predictive maintenance with AI
AI in production is also gaining importance in maintenance through so-called predictive maintenance. For this purpose, algorithms for action planning monitor defined parameters and properties of the processes. For example, the optimal maintenance time can be determined based on the degree of wear of operating resources and maintenance activities can be optimised. Using AI-based methods, the system can detect critical condition information of various machine parts and automatically inform the user. In this way, unforeseen downtime can be avoided and maintenance processes and costs can be optimised.
Logistics and resource planning with AI
AI is used in manufacturing for supply chain optimisation, among other things. Semantic networks and machine learning integrate action planning and optimisation algorithms, for example. This optimises various aspects of the supply chain. These include goods receipt, storage, production and distribution.
First steps to algorithms
Every piece of data has a story. Organizations can start by asking themselves where each data point actually came from and then evaluate in which use case you could use it. It is always advisable to first create a detailed description of the data. Based on this, the missing values can then be determined. The idea is to get complete visibility into the data by focusing on mining, segmenting and finding patterns before you start thinking about the use case.
When you start such projects, make sure you give your people enough freedom to experiment with their treasure trove of data in the data warehouse or data lake.
No AI Pilot without added value
Once you have generated potential AI use cases for AI from your data, critically examine their added value. To avoid the AI project dying in beauty, the added value for users should be precisely described and validated.
These 8 tips will help you avoid stumbling blocks in your first AI projects:
- Choose technology wisely: The technology should not only fit your needs, but also be scalable if you later want to roll-out your AI solution company-wide.
- Not without IT: Approach your IT colleagues early. The first step is to find a common understanding of the technological requirements. If that doesn’t work, you can use external IT experts as translators or for knowledge transfer.
- Develop prototypes iteratively: In any case, work with prototypes to quickly test your AI-solution with test users. Let your decision-makers test as well to better understand and manage their expectations.
- Avoid redundancy: Define work steps and criteria in detail. This makes redundant work steps transparent if, for example, an identical data set is entered multiple times.
- Keep it simple: Avoid to use complex AI algorithms that are difficult to validate. Increase the amount of AI in your projects slowly and only when the need is clearly demonstrated.
- Invest in tool training: Plan and budget for training and education for AI users. Staff supported by AI also need to learn how to use the new solution.
- Make friends in the legal department: Strengthen the relationship with in-house lawyers to identify legal requirements related to liability, data protection and intellectual property.
- Clarify relationship: Consider early on which of your colleagues will later be affected by your AI solutions. Involve the user teams in the project from the beginning to help decide how the solution can accompany human work processes.
