A method of detecting and classifying anomalies during semiconductor processing includes executing a wafer recipe a semiconductor processing system to process a semiconductor wafer; monitoring sensor outputs from a sensors that monitor conditions associated with the semiconductor processing system; providing the sensor outputs to models trained to identify when the conditions associated with the semiconductor processing system indicate a fault in the semiconductor wafer; receiving an indication of a fault from at least one of the models; and generating a fault output in response to receiving the indication of the fault.

Aspects of the disclosed technology encompass the use of a deep-learning controller for monitoring and improving a manufacturing process. In some aspects, a method of the disclosed technology includes steps for: receiving control values associated with a process station in a manufacturing process, predicting an expected value for an article of manufacture output from the process station, and determining if the deep-learning controller can control the manufacturing process based on the expected value. Systems and computer-readable media are also provided.

A method includes receiving trace sensor data associated with a first manufacturing process of a processing chamber. The method further includes processing the trace sensor data using one or more trained machine learning models that generate a representation of the trace sensor data, and then generate reconstructed sensor data based on the representation of the trace sensor data. The method further includes comparing the trace sensor data to the reconstructed sensor data. The method further includes determining one or more differences between the reconstructed sensor data and the trace sensor data. The method further includes determining whether to recommend a corrective action associated with the processing chamber based on the one or more differences between the trace sensor data and the reconstructed sensor data.

A dimension tolerance determining method and a system are disclosed. The method includes: determining initial tolerances of parts in assembled product; generating a plurality of tolerance data sets according to initial tolerances of parts, each tolerance data set containing a calculated part tolerance of each part; inputting the plurality of tolerance data sets into a key parameter generation module to generate a plurality of key parameters corresponding to the assembled product, wherein each key parameter corresponds to one tolerance data set; when the key parameters are located within a design range, calculating a plurality of assembly yield rates based on the tolerance data sets corresponding to the key parameters located within the design range; and selecting the tolerance data set corresponding to one of the assembly yield rates as the tolerances of the parts in the assembled product; wherein the key parameter generation module includes a neural network model.

An inspection system may receive inspection datasets from a defect inspection system associated with inspection of one or more samples, where an inspection dataset of the plurality of inspection datasets associated with a defect includes values of two or more signal attributes and values of one or more context attributes. An inspection system may further label each of the inspection datasets with a class label based on respective positions of each of the inspection datasets in a signal space defined by the two or more signal attributes, where each class label corresponds to a region of the signal space. An inspection system may further segment the inspection datasets into two or more defect groups by training a classifier with the values of the context attributes and corresponding class labels for the inspection datasets, where the two or more defect groups are identified based on the trained classifier.

A set of multi-mode elastic wave generating and detecting devices and field sensors are utilized in a real-time in-situ monitoring system based on the quality assessment of a specially designed article made by an additive manufacturing machine. The original invention disclosed in U.S. patent application Ser. No. 15/731,366 involves the transmission and reception of waves into a periodic test artifact while it is being built. The current invention involves the transmission and reception of multi-mode waves into a test artifact, the processing of data from narrow and wide field-of-view sensors, and correlating and relating the waveforms and sensor data while it is being built using physics-based and machine learning models. The disclosed system may initiate control and real-time corrective actions based on the properties and characteristics of the obtained waveforms and sensor data and their correlations and functional relationships.

Methods and systems for reducing substrate particle scratching using machine learning are provided. A machine learning model is trained to predict process recipe settings for a substrate temperature control process to be performed for a current substrate at a manufacturing system. First training data and second training data are generated for the machine learning model. The first training data includes historical data associated with prior process recipe settings for a prior substrate temperature control process performed for a prior substrate at a prior process chamber. The second training data is associated with a historical scratch profile of one or more surfaces of the prior substrate after performance of the prior substrate temperature control process according to the prior process recipe settings. The first training data and the second training data are provided to train the machine learning model to predict which process recipe settings for the substrate temperature control process to be performed for the current substrate correspond to a target scratch profile for one or more surfaces of the current substrate.

First training data including a set value related to a molding machine, a measured value obtained by measuring a physical quantity related to molding, and a degree of quality of a molded product generated by the molding machine is collected, a first learning model for outputting a degree of quality of a molded product when a set value and a measured value are input is generated by machine learning based on collected first training data, second training data including a defect degree for each defect type of a molded product, a measured value, and a set value capable of reducing the defect degree is collected, and a second learning model for outputting a set value capable of reducing a defect degree when a defect degree and a measured value are input is generated by machine learning based on collected second training data and a degree of quality output from the first learning model.

A prediction system configured to predict a defect of a target product includes a first pre-trained model trained based on a defect characteristic value indicating a defect associated with a location in an existing product, a feature of a three-dimensional shape of the existing product, and conditional information indicating a manufacturing condition of the existing product. The first pre-trained model is configured to, when a feature of a three-dimensional shape of the target product is input, output a defect characteristic value indicating a defect associated with a location in the target product.

A method includes receiving, from one or more sensors associated with manufacturing equipment, current trace data associated with producing, by the manufacturing equipment, a plurality of products. The method further includes performing signal processing to break down the current trace data into a plurality of sets of current component data mapped to corresponding component identifiers. The method further includes providing the plurality of sets of current component data and the corresponding component identifiers as input to a trained machine learning model. The method further includes obtaining, from the trained machine learning model, one or more outputs indicative of predictive data and causing, based on the predictive data, performance of one or more corrective actions associated with the manufacturing equipment.