Accenture Global Solutions Limited

Shin Moriga

Makoto Murai

Takuya Kudo

Atsushi Suyama

Motoaki Hayashi

The present disclosure describes a method of controlling a manufacturing system using multivariate time series and includes storing recording data as a plurality of time series, each time series having a first recorded value and a final recorded value, interpolating, within a first time window, missing values in the plurality of time series using a Bayesian model, the missing values falling between a first and an end time of the respective time series, storing the interpolated values as prediction data, each interpolated value including an uncertainty, loading recorded data of a second time window, loading prediction data of the second time window, predicting, using the Bayesian model, values for each time series that is absent recorded data and prediction data, storing the predicted values, each prediction value including an uncertainty, and adjusting a device that generates the recorded data based on the prediction values within the second time window.

Prediction method and system for multivariate time series data in manufacturing systems

The present disclosure describes a computer-implemented method for detecting anomalies during lot production, wherein the products within a production lot are processed according to a sequence of steps that include manufacturing steps and one or more quality control steps interspersed among the manufacturing steps, the method comprising: obtaining process quality inspection data from each of the one or more quality control steps for a first production lot; obtaining product characteristics data for the products in the first production lot after the final step in the sequence; training a Gaussian process regression model using the process quality inspection data and the product characteristics data from the first production lot; generating a predictive distribution of the product characteristics data using the Gaussian process regression model that uses a bathtub kernel function; obtaining process quality inspection data from each of the quality control steps for a second production lot; identifying anomalies in the second production lot using the predictive distribution of the product characteristics data and the process quality inspection data from the second production lot; if no anomalies are detected in the second production lot, updating the Gaussian process regression model using the process quality inspection data from the second production lot; setting target values for one or more values in the process quality inspection data based on the predictive distribution of the product characteristic; and adjusting settings of one or more manufacturing steps based on the target values.

Anomaly detection method and system for manufacturing processes

The present disclosure describes a method of controlling a manufacturing system using multivariate time series, the method comprising: recording data from one or more devices in the manufacturing system; storing the recorded data in a data storage as a plurality of time series, wherein each time series has a first recorded value corresponding to a first time and a final recorded value corresponding to an end of the time series; interpolating, within a first time window, missing values in the plurality of time series using a Bayesian model, wherein the missing values fall between the first and end time of the respective time series; storing the interpolated values as prediction data in a prediction storage, wherein the interpolated values include the uncertainty of each interpolated value; loading the recorded data that fall within a second time window from the data storage; loading prediction data from the prediction storage that fall within the second time window and for which no recorded data are available; optimizing the parameters of the Bayesian model using the loaded recorded data and the prediction data; predicting, using the Bayesian model, values for each of the time series for which loaded recorded and prediction data are not available; storing the predicted values as prediction data in the prediction storage, wherein the prediction values include the uncertainty of each prediction value; and adjusting one or more of the devices that generate the recorded data based on the prediction data within the second time window.

Growing community of inventors

Tokyo, Japan

Shin Moriga