High-efficiency And Alternative Closed-mode Analysis To Optimize Production Planning

With the development of intelligent production,extracting useful knowledge user from use data and production data to guide and optimize Manufacturing planning becomes a research hotspot,The UP-Growth algorithm reduces transaction-weighted utility(TWU)values accumulated in its tree structure by pruning strategy,but still requires a lot of time to calculate the actual utility of the candidate patterns.HUI-Miner is One-stage mining algorithm and does not need to scan the database multiple times,does not generate candidate patterns,calculating the true utility value of the pattern by utility list directly,but this algorithm is more suitable for static data mining.The A-Close algorithm generates candidate patterns step by step based on Apriori and the algorithm execution time is too long,The MERIT algorithm adopts a strategy that storing the weights of each element separately.This strategy generates a large number of candidate modes when creating a new node in each mode combination,which takes up a lot of memory and time.In order to optimize the deficiencies of traditional algorithms,this paper proposes an high utility and erasable closed pattern mining algorithm HECPM,which includes two sub-algorithms,SHUPM and ECPM.mining pattern from demand side and the production side respectively.The target product output on the demand side is used as the input of the production side,then through the production side supply chain mechanismand,find a set of products what creating high profits.the enterprise decision layer combines the analysis result of demand side and the production side to establishing profitable product manufacturing plans,The sub-algorithm,SHUPM uses the sliding window technology and the utility list SHUP-List to analyze the data stream efficiently based on the utility parameters,and understanding the preference of customers,The sub-algorithm ECPM definite erasable rule based on the profit parameter to pre-processes the product data,The dNC-Sets structure of each pattern and the relationship between the pattern are determined according to the standard data structure dNC-Sets,and the most common excavation based on the proven alternative closed-mode theorem Excellent solution,get the best combination of product parts.In this Python environment,the sub-algorithm SHUPM and the conventional algorithms UP-Growth and HUI-Miner perform simulation experiments in IBM's user datasets Connect and T10I4D100 K.The sub-algorithm ECPM and the traditional algorithms A-Close and MERIT are compared and simulated in two production data from IBM Pumsb and Accidents.Three performance indicators of algorithm execution time,memory consumption and algorithm scalability are analyzed in the experiment.The simulation results show that in the standard user dataset and production data,the traditional algorithm HECPM algorithm has improved performance in terms of algorithm execution time,memory consumption and algorithm scalability.