Dynamic cost forecasting model based on extreme learning machine - A case study in steel plant

- This research focuses on the dynamic cost forecasting.
- Steel material fluctuated and hard to earn profit without appropriate cost estimation.
- A hybrid-forecasting model with Grey Relation Analysis and extreme learning machine is proposed.
- Managers are able to reconsider the purchasing of raw material and adjust the pricing strategy to pursuit the profit target.

Without a doubt, the current commercial circumstance exists in a strong competition and is full of uncertainty. Managers focus on continuous effort for increasing profits and reducing cost in their organization. In the past two decades, the price of raw materials has fluctuated severely and steel plants found it hard to earn profits and keep the capacity normal. Whereas some steel plants reduced the overcapacity of production by revamping the blast furnace or cutting down the throughput, others tried to limit and eliminate unnecessary cost. However, these strategies have failed to keep pace with the frequent changes in the raw material prices and could not support the profit targets greatly. This study aims at addressing such concerns by proposing an extreme learning machine (ELM) to predict the major raw material price in steel plants. Typically, this paper focuses on integration of Grey Relation Analysis (GRA) with a hybrid forecasting model to forecast the cost of iron ore and coking coal that are majorly used in steel plants. Here we attempt to establish a dynamic cost system to forecast the manufacturing cost of end products and adjust the purchasing and production strategy. This forecasting model can offer an accurate and rapidly predicting result of raw material price. Managers can use this forecasting results to reconsider the purchasing of raw materials and adjust the pricing strategy to pursuit their company's profit targets.