A manufacturing defect factor searching method includes: classifying manufacturing monitoring data into a set of non-defective products having an inspection result indicating a non-defective product and a set of defective products having the inspection result indicating a defective product, in accordance with a correspondence relationship between the manufacturing monitoring data and product inspection data indicating the inspection result of the product manufactured in the manufacturing line, the manufacturing monitoring data being collected from a manufacturing line of a product and being multivariate; estimating, for each item of the manufacturing monitoring data, a mixture distribution function approximating to a statistical distribution of each of the set of non-defective products and the set of defective products; resolving the mixture distribution function into components; and generating a list of items including a resolved component having a correlation with a manufacturing quality defect from among items of the manufacturing monitoring data.

A prediction model creation apparatus includes a feature amount acquisition unit that acquires values of types of feature amounts that are calculated from operating state data indicating an operating state of a production facility that produces a product, for both a normal time at which the production facility produces the product normally and a defective time at which a defect occurs in the product that is produced, a feature amount selection unit that selects a feature amount effective in predicting the defect from among the acquired types of feature amounts, based on a predetermined algorithm that specifies a degree of association between the defect and the types of feature amounts, from the values of the types of feature amounts acquired at the normal time and the defective time, and a prediction model construction unit that constructs a prediction model for predicting occurrence of the defect, using the selected feature amount.

A technique to generate predicted data for control or monitoring of a production process to improve a parameter of interest. Context data associated with operation of the production process is obtained. Metrology/testing is performed on the product of the production process, thereby obtaining performance data. A context-to-performance model is provided to generate predicted performance data based on labeling of the context data with performance data. This is an instance of semi-supervised learning. The context-to-performance model may include the learner that performs semi-supervised labeling. The context-to-performance model is modified using prediction information related to quality of the context data and/or performance data. Prediction information may include relevance information relating to relevance of the obtained context data and/or obtained performance data to the parameter of interest. The prediction information may include model uncertainty information relating to uncertainty of the predicted performance data.

A product state estimation device includes: an examination result acquisition device that acquires an examination result related to a state of a product obtained through each shot by a die-casting machine; a time series data acquisition device that acquires time series data based on an output from a sensor that detects an operation state of the die-casting machine at each shot; a time series data manipulation device that performs manipulation that clips data of a predetermined time interval out of the time series data; an estimation model generation device that generates an estimation model by using a neural network with the examination result of the product and the manipulated time series data as learning data; and an estimation device that estimates information related to quality of the product based on the manipulated time series data obtained from a plurality of detection signals at each shot by using the estimation model.

A machine tool machining dimensions prediction device (100) includes: a data collector (10) to acquire driving state information of a machine tool; a feature amount extractor (211) to extract a feature amount from the driving state information; a data analyzer (311) to analyze the extracted feature amount; and a machining quality prediction model generator (312) to generate, from the analyzed information, a prediction model of a machining dimension of a workpiece. The machine tool machining dimensions prediction device (100) applies the feature amount and the driving state information to the prediction model during machining of the workpiece to predict a machining quality and refers to a machining dimension quality regulation to determine whether the machining quality satisfies a standard.

Embodiments of systems and methods for supporting predictive and preventative maintenance are disclosed. One embodiment includes manufacturing cells within a manufacturing environment, where each manufacturing cell includes a cell controller and welding equipment, cutting equipment, and/or additive manufacturing equipment. A communication network supports data communications between a central controller and the cell controller of each of the manufacturing cells. The central controller collects cell data from the cell controller of each of the manufacturing cells, via the communication network. The cell data is related to the operation, performance, and/or servicing of a same component type of each of the manufacturing cells to form a set of aggregated cell data for the component type. The central controller also analyzes the set of aggregated cell data to generate a predictive model related to future maintenance of the component type.

Various systems and methods for modeling a manufacturing assembly line are disclosed herein. Some embodiments relate to operating a processor to receive cell data, extract feature data from the cell data, determine a plurality of faults, determine a priority level for each fault by applying the extracted feature data to a predictive model, determine at least one high priority fault, and generate at least one operator alert based on the at least one high priority fault.

Various systems and methods for modeling a manufacturing assembly line are disclosed herein. Some embodiments relate to operating a processor to receive cell data, extract feature data from the cell data, determine a plurality of cell configurations, determine an efficiency score by applying the feature data to a predictive model generated for predicting a production level of the manufacturing assembly line, determine at least one target cell configuration from the cell configurations based on the efficiency score, and apply the at least one target cell configuration to at least one cell by implementing each target cell configuration to a corresponding cell.

Various systems and methods for modeling a manufacturing assembly line are disclosed herein. Some embodiments relate to operating a processor to receive cell data and line production data, determine one or more production associations between the cell data and the line production data; evaluate the one or more production associations to identify one or more critical production associations; retrieve the cell data and the line production data associated with the one or more critical production associations; and train a predictive model with the retrieved cell data and the retrieved line production data to predict the production level of the manufacturing assembly line.

One variation of a method for predicting manufacturing defects includes: accessing a first set of inspection images of a first set of assembly units recorded by an optical inspection station over a first period of time; generating a first set of vectors representing features extracted from the first set of inspection images; grouping neighboring vectors in a multi-dimensional feature space into a set of vector groups; accessing a second inspection image of a second assembly recorded by the optical inspection station at a second time succeeding the first period of time; detecting a second set of features in the second inspection image; generating a second vector representing the second set of features in the multi-dimensional feature space; and, in response to the second vector deviating from the set of vector groups by more than a threshold difference, flagging the second assembly unit.