US 12,405,603 B2
Industrial Internet of Things (IoT) for determining reparability of defective product, control method, and storage medium thereof
Zehua Shao, Chengdu (CN); Haitang Xiang, Chengdu (CN); Bin Liu, Chengdu (CN); Yuefei Wu, Chengdu (CN); and Lei Zhang, Chengdu (CN)
Assigned to CHENGDU QINCHUAN IOT TECHNOLOGY CO., LTD., Chengdu (CN)
Filed by CHENGDU QINCHUAN IOT TECHNOLOGY CO., LTD., Sichuan (CN)
Filed on Jan. 25, 2024, as Appl. No. 18/423,268.
Application 18/423,268 is a continuation of application No. 18/172,268, filed on Feb. 21, 2023, granted, now 11,921,498.
Claims priority of application No. 202211015348.9 (CN), filed on Aug. 24, 2022.
Prior Publication US 2024/0168466 A1, May 23, 2024
Int. Cl. G05B 19/418 (2006.01)
CPC G05B 19/41875 (2013.01) [G05B 19/41835 (2013.01)] 17 Claims
OG exemplary drawing
 
1. An industrial Internet of Things (IoT) for determining repairability of a defective product, including: an obtaining module, a total correction cost determination module, and a defective product processing module, wherein
the obtaining module is configured to obtain defective product data;
the total correction cost determination module is configured to:
in response to a determination that a number of the defective product is larger than 1, perform a filtering operation on a defective product vector of the defective product to obtain a first defective product vector and a second defective product vector, the defective product vector being constructed based on the defective product data, wherein the first defective product vector is a filtered-out defective product vector, and the second defective product vector is a defective product vector left after filtering;
determine, based on the first defective product vector, a first total correction cost;
cluster the second defective product vector to determine at least one cluster center set;
for each of the at least one cluster center set, determine, based on one or more center set features, a number of raw material and manpower required by the cluster center set through a prediction model, the prediction model being a machine learning model; wherein the prediction model is obtained through training based on a plurality of labeled training samples, and a training process of the prediction model includes:
inputting the plurality of labeled training samples to an initial prediction model, constructing a loss function by the labels and an output result of the initial prediction model, and updating a parameter of the initial prediction model iteratively based on the loss function; wherein when the loss function of the initial prediction model satisfies a preset condition, the training process is completed, and a trained prediction model is obtained; and the training samples including sample center set features of a plurality of sample cluster center sets, and the labels indicating a number of raw material and manpower required for the sample cluster center sets;
determine, based on the number of raw material and the manpower required by the each of the at least one cluster center set, at least one second total correction cost; and
determine, based on the first total correction cost and the at least one second total correction cost, a total correction cost; and
the defective product processing module is configured to determine, based on the total correction cost and a preset cost, whether the defective product is repairable, and perform repair of the defective product based on the determination that the defective product is repairable.