International Business Machines Corporation

Chandrasekhara K Reddy

Jayant R Kalagnanam

Horst Cornelius Samulowitz

Dhavalkumar C Patel

Ashish Sabharwal

Arun Kwangil Iyengar

Young Min Lee

Jeffrey Owen Kephart

Deepak Srinivas Turaga

Robert J Baseman

Dung Tien Phan

Elham Khabiri

Gerald James Tesauro

Tamir Klinger

Kiran A Kate

Srideepika Jayaraman

Raya Horesh

Anuradha Bhamidipaty

Toshiyuki Hama

Zhiguo Li

Stuart A Siegel

Lianjun An

Shrey Shrivastava

Andrew J Davenport

Sanjeeb Dash

Young T Chae

Techniques for model evaluation and selection are provided. A plurality of models trained to generate predictions at each of a plurality of intervals is received, and a plurality of model ensembles, each specifying one or more of the plurality of models for each of the plurality of intervals, is generated. A test data set is received, where the test data set includes values for at least a first interval of the plurality of intervals and does not include values for at least a second interval of the plurality of intervals. A first model ensemble, of the plurality of model ensembles, is selected based on processing the test data set using each of the plurality of model ensembles.

Determining and selecting prediction models over multiple points in time using test data

A computer implemented method of preparing process data for use in an artificial intelligence (AI) model includes collecting and storing raw data as episodic data for each episode of a process. An episode data generator assigns an episode identifier each set of episodic data. The raw data per episode is transformed into a standardized episodic data format that is usable by the AI model. Metrics are assigned to the episodic data and the episodic data is aggregated in an episode store. The data in the episode store is used by a feature extraction and learning module to extract and rank features.

Dynamic and continuous composition of features extraction and learning operation tool for episodic industrial process

Techniques for generating model ensembles are provided. A plurality of models trained to generate predictions at each of a plurality of intervals is received. A respective prediction accuracy of each respective model of the plurality of models is determined for a first interval of the plurality of intervals by processing labeled evaluation data using the respective model. Additionally, a model ensemble specifying one or more of the plurality of models for each of the plurality of intervals is generated, comprising selecting, for the first interval, a first model of the plurality of models based on (i) the respective prediction accuracies and (ii) at least one non-error metric.

Determining and selecting prediction models over multiple points in time

A computing device includes a processor and a storage device. A vehicle asset modeling module is stored in the storage device and is executed by the processor to configure the computing device to perform acts of identifying and clustering a plurality of assets based on static properties of a vehicle asset using a first module of the vehicle asset modeling module. The clustered plurality of assets is determined based on dynamic properties of the vehicle asset using a second module. Event prediction is performed by converting a numeric data of the clustered plurality of assets to a natural language processing (NLP) domain by a third module. One or more sequence-to-sequence methods are performed to predict a malfunction of a component of the vehicle asset and/or an event based on past patterns. Prediction information is stored in the storage device.

Vehicle asset modeling using language processing methods

A computing device includes a processor and a storage device. A wafer asset modeling module is stored in the storage device and is executed by the processor to configure the computing device to perform acts identifying and clustering a plurality of assets based on static properties of a wafer asset using a first module of the wafer asset modeling module. The clustered plurality of assets is determined based on dynamic properties of the wafer asset using a second module of the wafer asset modeling module. Event prediction is performed by converting a numeric data of the clustered plurality of assets to a natural language processing (NLP) domain by a third module of the wafer asset modeling module. One or more sequence-to-sequence methods are performed to predict a malfunction of a component of the wafer asset and/or an event based on past patterns. Prediction information is stored in the storage device.

Wafer asset modeling using language processing methods

A sub-process sequence is identified from a temporal dataset. Based on time information, predictors are categorized as being available or not available during time periods. The predictors are used to make predictions of quantities that will occur in a future time period. The predictors are grouped into groups of a sequence of sub-processes, each including a grouping of one or more of the predictors. Information is output that allows a human being to modify the groups. The groups are finalized, responsive to any modifications. Prediction models are extracted based on dependencies between groups and sub-processes. A final predication model is determined based on a prediction model from the prediction models that best meets criteria. A dependency graph is generated based on the final prediction model. Information is output to display the final dependency graph for use by a user to adjust or not adjust elements of the sequential process.

System for automatic deduction and use of prediction model structure for a sequential process dataset

A method for automatically constructing and organizing a navigation graph includes receiving input data including a plurality of reports, at least two of the reports including plots, extracting a plurality of variables from the plots, building a knowledge graph from the input, wherein each node of the knowledge graph is associated with an individual one of the plots and an edge is added between two of the nodes sharing at least one of the variables in common, adding an edge weight to each of the edges of the knowledge graph, and organizing the nodes of the knowledge graph for navigation, wherein the knowledge graph is displayed in a user interface.

Automatic construction and organization of knowledge graphs for problem diagnoses

A system, method, and computer program product for automatically selecting from a plurality of analytic algorithms a best performing analytic algorithm to apply to a dataset is provided. The automatically selecting from the plurality of analytic algorithms the best performing analytic algorithm to apply to the dataset enables a training a plurality of analytic algorithms on a plurality of subsets of the dataset. Then, a corresponding prediction accuracy trend is estimated across the subsets for each of the plurality of analytic algorithms to produce a plurality of accuracy trends. Next, the best performing analytic algorithm is selected and outputted from the plurality of analytic algorithms based on the corresponding prediction accuracy trend with a highest value from the plurality of accuracy trends.

Automated intelligent data navigation and prediction tool

A method for configuring a manufacturing plant includes generating a set of initial production schedules for manufacturing an entity, determining a set of consistent production schedules given each of the initial production schedules, selecting an instance of the consistent production schedules, generating a schedule for manufacturing instances of the entity, and operating the manufacturing plant using the schedule.

Production system having constraint evaluator generating constrained sequences

A method and system for optimizing modules of a steel manufacturing process includes a plurality of manufacturing modules for a manufacturing process. Each of the modules have a plurality of steps. The plurality of modules include at least an upstream module, a casting module, and a downstream module. Each of the plurality of modules have parameters, and include at least one variable event. The variable event is adjustable for optimization of the manufacturing process while the parameters are being maintained for each of the plurality of modules. A communication system is used for exchanging information between the modules while the manufacturing process is occurring to adjust the at least one variable event for optimizing the manufacturing process.

Algorithmic framework for the integrated optimization of hot strip mill and upstream molten metal operations for a large-scale steel manufacturing plant

A static heat transfer model is derived from a system of dynamic equations by integrating the dynamic equations over different time periods. That static heat transfer model links periodic (e.g., monthly) energy usage with cooling and heating degree hours, humidifying and dehumidifying hours. Its coefficients of measuring correlations correspond to the thermal parameters of buildings. Temporal data from a building may be used to estimate the overall heat transfer parameters. A clustering scheme may be developed to decompose all the buildings into different clusters based on one or more similarity criteria. The overall heat transfer parameters are separated into values for the wall, roof and window using multiple buildings' data in the same cluster or group.

Estimating building thermal properties by integrating heat transfer inversion model with clustering and regression techniques for a portfolio of existing buildings

Generating an optimal planning of building retrofit for a portfolio of buildings may include providing a plurality of objective functions that may be selected for maximizing cost reduction, maximizing green house gas emission reduction, or maximizing energy reduction, or combinations thereof. The objective function may be solved based on information including at least a retrofit cost for retrofitting a building, payback period specifying the length of time needed to recover the retrofit cost, a budget available for retrofitting the building, expected price of energy, estimated energy savings from retrofitting and estimated green house gas emission from retrofitting. The planning of building retrofit may be generated based on the solutions of one or more of the objective functions, which may provide for an optimal plan of building retrofit.

Optimal planning of building retrofit for a portfolio of buildings

An automated method optimally designs plates to satisfy an order book at a steel plant so as to maximize the yield of the plates designed while using capacity fully to reduce the production of surplus slabs or plates, and satisfy order deadlines. Our method consists of four main components: (1) mother plate design, (2) slab design, (3) cast design, and (4) material allocation. A column generation framework for mother plate design is used where the problem is decomposed into a master problem and a subproblem. The master problem is used to evaluate packing patterns that should be used to fulfill the order book and the subproblem generates potential one-dimensional and two-dimensional feasible packing patterns as candidates to be evaluated by the master problem. The solution to the master problem produces a list of mother plates that need to be produced. These mother plates are transformed into candidate slabs, which are represented via an interval graph. The maximal cliques in the interval graph&#x2014;maximal subsets of slabs which can be grouped together to form casts&#x2014;are enumerated to generate candidate casts, from which redesign information is sent to the plate/slab design processes. The cast design process is integrated with the plate design/slab design processes in the sense that information is passed from the plate/slab design processes to the cast design process and vice versa. Finally, our method has a material allocation component to assign order plates to inventory mother-plates and slabs.

Growing community of inventors

Kinnelon, NJ, United States of America

Chandrasekhara K Reddy