Development of grinding intelligent monitoring and big data-driven decision making expert system towards high efficiency and low energy consumption: experimental approach

Grinding has been extensively applied to meet the urgent need for tight tolerance and high productivity in manufacturing industries. However, grinding parameter settings and process control still depend on skilled workers’ engineering experience. The process stability in complicated non-uniform wear can't be guaranteed. Moreover, it is impossible to obtain energy-saved grinding strategies. Intelligent monitoring methods are well-recognized to help conquer present trial–error processing deficiencies. However, discrete manufacturing companies have to face increasing difficulties to identify the monitored big data and make credible decisions directly. A decision-making expert system driven by monitored power data (EconG^©) is thus developed. EconG^© provides a 4-level database structure to efficiently manage multi-source heterogeneous data. Signal conditioning, peaks-valleys feature exaction, and compression approaches are proposed for reducing the storage volume of real-time monitored data. The data size has been reduced to 6.5% of the source. A mathematical comparison model based on the power feature is embedded to diagnose burns, which has been validated by the 16th and 55th surface grinding results. Mapping relation model from inputs, signals to outputs has been built by the power feature-extended artificial neural network algorithm. Prediction accuracy is improved by introducing adaptive control and dynamic changes in material removal. EconG^© breaks a single analysis based on grinding parameters. Energy-saved grinding strategies could be intelligently acquired through the presented Pareto optimization method. In the future, a broader and deeper implementation of EconG^© will guild manufacturers to respond quickly to explosive demands on intellectualization, sustainability, and flexibility in the arrived 4th industrial revolution.