Research On Rolling Forecast Of Shipbuilding Enterprises Material Cost Based On LSTM

Ship products are typically complex products with complex structure and technology,long production cycle,and many uncertain factors.Some equipment depends on imports,is greatly affected by exchange rate,and the comprehensive cost of labor,raw materials,financing and so on is rising."Difficult profit" has always been a practical problem plaguing ship enterprises.According to the data released by China Shipbuilding Industry Association,although the shipping market has been steadily improving in recent years,enterprises are still facing serious risks of profit decline and cost rise.How to strengthen cost forecast and control is very important to improve the profit space of shipping enterprises.By analyzing ship manufacturing process and identifying the characteristics and influencing factors of material cost,this paper constructs a rolling forecast model of ship material cost based on LSTM.On this basis,considering that cost prediction requires a certain amount of historical data,material cost data and corresponding influencing factors of53 64,000 tons bulk carriers in a shipyard were collected for 63 months to verify the model.The results show that: 1)most of the predicted data are within an acceptable range of error from the actual values,and the individual cases with large prediction deviations are traced back to factors such as design changes,proving the effectiveness of the selected method and model construction.2)For the ship material cost forecast such a complex problem,LSTM can automatically capture the relationship between complex time series data and mine the hidden information in the data,thus having strong robustness and generalization ability.3)Rolling prediction has good reference value for real-time cost progress,providing data support for companies to identify cost risks and strengthen cost process control.With the promotion of intelligent transformation,companies have efficiently aggregated various data such as design,production,procurement,and operation.Shipbuilding manufacturing will shift from "entity-led" to "digital-led" and from "entity manufacturing" to "predictive manufacturing".Through the integration of finance,business,and information technology,cost prediction accuracy and the scientificity of prediction models can be further studied from the perspective of big data,and a data-supported cost prediction system and analysis model can be established to achieve cost prediction and control of the shipbuilding manufacturing process that is knowable,visible,and controllable.