Abstract: In example embodiments, techniques are provided that use two different ML models (a symbol association ML model and a link association ML model), one to extract associations between text labels and one to extract associations between symbols and links, in a schematic diagram (e.g., P&ID) in an image-only format. The two models may use different ML architectures. For example, the symbol association ML model may use a deep learning neural network architecture that receives for each possible text label and symbol pair both a context and a request, and produces a score indicating confidence the pair is associated. The link association ML model may use a gradient boosting tree architecture that receives for each possible text label and link pair a set of multiple features describing at least the geometric relationship between the possible text label and link pair and produces a score indicating confidence the pair is associated.

Abstract: In example embodiments, a multi-stage PFS diagram generation technique is used to iteratively define the layout of a PFS diagram from a subset of engineering data in a 3D model of an industrial process. The multi-stage PFS diagram generation technique may repeatedly call an automatic layout generator, which each time solves for one unknown quality of the PFS diagram (e.g., relative positions of components in the PFS diagram, positions on components in the PFS diagram, sizes of the components in the PFS diagram). The PFS diagram may be adapted based on user preferences, for example to define the subset of engineering data, or to constrain aspects of its layout. Updated PFS diagrams may be generated by selecting different user preferences.

Abstract: In one embodiment, techniques are provided for aligning source infrastructure data to be compatible with a conceptual schema (e.g., BIS) implemented through an underlying database schema (e.g., DgnDb). Data aligned according to the conceptual schema may serve as a “digital twin” of real-world infrastructure usable throughout various phases of an infrastructure project, with physical information serving as a “backbone”, and non-physical information maintained relative thereto, forming a cohesive whole, while avoiding unwanted data redundancies. Source-format-specific bridge software processes may be provided that that know how to read and interpret source data of a respective source format, and express it in terms of the conceptual schema.

Abstract: In various example embodiments, techniques are provided for updating content of a markup language document. A software process receives a markup language document having one or more sections and a corresponding enhancement document that includes a plurality of commands describing updates to the markup language document. The software process converts the markup language document into a first hierarchical graph in which each section of the markup language document is arranged as a parent of any subsections of the respective section. The software process also converts the enhancement document into a second hierarchical graph including one or more of the commands. The software process merges the first hierarchical graph and the second hierarchical graph, the merging to apply commands of the second hierarchical graph to the first hierarchical graph to produce an updated hierarchical graph. The software process then outputs an updated markup language document based on the updated hierarchical graph.

Abstract: In example embodiments, techniques are provided for decoupling user access to infrastructure models from proprietary software that maintains and updates the infrastructure models. A backend application may include an infrastructure modeling backend module that, among other functions, handles communication with an infrastructure modeling frontend module of a frontend application that provides user access to the infrastructure model, infrastructure modeling hub services that maintain repositories for the infrastructure model, and an infrastructure modeling native module that creates, performs operations upon, and updates local instances of a database that stores the infrastructure model.

Abstract: In example embodiments, a superelevation tool extracts and displays superelevation data from a 3D roadway model by accessing roadway meshes and a horizonal alignment from the 3D roadway model, extracting a plurality of template drops from the one or more roadway meshes at locations along the horizonal alignment to produce an ordered list of template drops and processing the template drops of the ordered list to identify top-facing roadway edges in each template drop that represent top pavement surface of the roadway at the location of the template drop, iteratively searching for a superelevation candidate and detecting superelevation data from the superelevation candidate at least in part by comparing cross-slopes of the top-facing roadway edges of consecutive template drops in the ordered list, wherein a superelevation candidate includes at least two or more template drops having cross-slopes that are locked, and providing a visualization of the detected superelevation data.

Abstract: In various example embodiments, techniques are provided for training and/or using a semantic deep learning model, such as a segmentation-enabled CNN model, to detect corrosion and enable its quantitative evaluation. An application may include a training dataset generation tool capable of semi-automatic generation of a training dataset that includes images with labeled corrosion segments. The application may use the labeled training dataset to train a semantic deep learning model to detect and segment corrosion in images of an input dataset at the pixel-level. The application may apply an input dataset to the trained semantic deep learning model to produce a semantically segmented output dataset that includes labeled corrosion segments. The application may include an evaluation tool that quantitatively evaluates corrosion in the semantically segmented output dataset, to allow severity of the corrosion to be classified.

Abstract: In example embodiments, techniques are described for implementing a code service for managing codes for elements in a digital twin of infrastructure according to an edge computing paradigm. The techniques may use an “edge base,” wherein each edge computing device (e.g., a client computing device or VM) executes a code service that maintains a local, periodically-synchronized copy of a portion of a code database for the digital twin. A cloud container of a blob storage service of a cloud datacenter may maintain a master copy of the code database. Read operations by an application on a client computing device may be performed against its as-of-last-synchronization local code database. Write operations by the application may likewise be performed against the local code database, serialized via a write lock maintained in the cloud container that permits only a single client computing device to modify its local code database at a time.

Abstract: In one embodiment, a technique for numerical integration in material point method (MPM)-based geotechnical analysis and simulation is provided. Input terms for an element of a background mesh are received. The input terms including material points in the element that describe a continuum of soil, rock and/or groundwater. A set of constraints is created that defines an optimization problem. The set of constraints provide that numerical integration of the material points weighted by unknown integration weights equal exact integration for finite element shape functions. The optimization problem defined by the constraints is solved by an optimization algorithm to minimize numerical integration error for polynomials up to a given order to produce a set of integration weights. The set of integration weights is scaled to conserve the mass of the material points to produce optimized integration weights.

Abstract: In example embodiments, techniques are provided for visually comparing digital content for an infrastructure project according to time using 4-D construction modeling software. The 4-D construction modeling software includes a cloud-based 4-D comparison service and a local 4-D modeling client. The 4-D comparison service includes a digital content alignment service and a 4-D difference engine. The digital content alignment service aligns different pieces of digital content and produces views that provide visual comparison between different pieces of digital content. The 4-D difference engine automatically determines differences between different pieces of digital content. The 4-D modeling client includes a 4-D comparison user interface (UI) process that receives user input used to generate, and then displays a generated visual comparison between different pieces of digital content.

Abstract: In example embodiments, a new model for modeling monopiles in is provided that, in addition to distributed lateral load along the monopile, considers distributed moment along the length of the pile, base moment at the pile tip, and base shear force at the pile tip. The new model may avoid the overly conservative designs for large diameter piles (e.g., 10 m+) with small length-to-diameter ratios (e.g., <6), while using standardized reaction curves (i.e., p-y curves and t-z curves) and considering axial and combined loading.

Abstract: In example embodiments, techniques are provided for using machine learning to extract machine-readable labels for text boxes and symbols in P&IDs in image-only formats. A P&ID data extraction application uses an optical character recognition (OCR) algorithm to predict labels for text boxes in a P&ID. The P&ID data extraction application uses a first machine learning algorithm to detect symbols in the P&ID and return a predicted bounding box and predicted class of equipment for each symbol. One or more of the predicted bounding boxes may be decimate by non-maximum suppression to avoid overlapping detections. The P&ID data extraction application uses a second machine learning algorithm to infer properties for each detected symbol having a remaining predicted bounding box. The P&ID data extraction application stores the predicted bounding box and a label including the predicted class of equipment and inferred properties in a machine-readable format.

Abstract: Techniques are provided for near real-time anomaly event detection and classification with trend change detection for smart water grid operation management. In the first phase, a trend change is detected in each of one or more sensors by comparing new sensor data of a sensor with a historical trend pattern of the same sensor. In the second phase, and after the trend changes are detected, a valid event evaluation time window can be determined based on combining and analyzing the detected trend changes for flow and pressure sensors, e.g., at least one flow sensor and at least one pressure sensor from the same supply zone of the smart water grid. The valid event evaluation time window can be used with anomaly events that are detected in near-real time to classify the anomaly events in near-real time as valid, e.g., true anomaly events, or invalid, e.g., false alarms.

Abstract: In one embodiment, material points are received that cover at least a portion of an element of a background mesh that describes a continuum of soil, rock and/or groundwater. MPM-based geotechnical analysis and simulation is conducted at least in part by performing a numerical integration over the material points to produce a system matrix and right-hand side vector. The numerical integration applies hierarchical shape functions to the material points. The MPM-based geotechnical analysis and simulation also may subtract out contributions of any lower-order polynomials from higher-order polynomials of the hierarchical shape functions when interpolating one or more state variables for the material points to the background mesh. The MPM-based geotechnical analysis and simulation also may subtract out contributions any lower-order polynomials from higher-order polynomials of the hierarchical shape functions when calculating one or more boundary conditions for the material points.

Abstract: In example embodiments, a user interface of a software application is provided for visualizing high-dimensional datasets, which simultaneously displays marginal distributions and joint distributions of variables that represent different attributes (e.g., properties) of entities (e.g., elements of infrastructure). The marginal distributions and joint distributions are combined into a single visualization that may be shown in a single window of the application. The visualization may include a graph (e.g., a bar chart) for each of the variables showing the marginal distribution of the variable, wherein each graph is displayed along a different portion of a perimeter of a closed shape (e.g., a circle). The visualization may also include graphics (e.g., lines) connecting portions of the bar charts showing the joint distribution for possible pairs of variables, wherein each graphic (e.g., line) is displayed with visual properties (e.g., a thickness) that indicates co-occurrence frequency of values of the variables.

Abstract: Techniques are provided for accessing derivative properties from a database utilizing a logical schema instruction that includes a base object identifier. A logical schema instruction may be received and analyzed to identify a predefined keyword. A portion of the logical schema instruction, that is located in relation to the predefined keyword, may be identified. The portion may include a base object identifier and a property identifier, e.g., a derivative property identifier. In an embodiment, the identified portion of the logical instruction is not translated to a database schema instruction at prepare time. Instead, an extract function is executed at runtime access and analyze a database table that corresponds to a base object associated with the base class identifier. A property in the database table that corresponds to the property identifier may be identified. Advantageously, a base object identifier may be utilized to access a derivative property.

Abstract: In one embodiment, a technique for providing absolute orientation of a mobile device involves aligning the orientation sensor of the mobile device with the Earth's coordinate system. An initial orientation from the orientation sensor of the mobile device is accessed, a user is prompted to move the mobile device in a direction of the initial orientation, and using a position sensor, a set of position information that describes positions of the mobile device is captured while the mobile device is being moved in the direction. Based on the set of position information, a vector of movement is determined. Using the vector of movement, an orientation difference is calculated between the Earth's coordinate system and a coordinate system of the mobile device. Upon demand, an absolute orientation of the mobile device may be produced by accessing a live orientation from the orientation sensor and adding the orientation difference.

Abstract: In various example embodiments, machine vision-based techniques for non- contact SHM are provided that may integrate both phase-based algorithms and template matching algorithms to enable selection of one or more machine vision algorithms that are effective at measuring responses (e.g., displacement, strain, acceleration, velocity, etc.) under present conditions. Results of a single algorithm or a combination of results from multiple algorithms may be returned. In such techniques, improved template matching algorithms may be employed that provide sub-pixel accuracy. Responses may be adjusted to cancel out camera vibration and video noise. Defects or damage may be determined by tracking changes in displacement within an area of interest.

Abstract: A method of modifying or controlling a highway traffic system may include training a machine learning model using historical traffic data corresponding to a roadway traffic system in which the historical traffic data is indicative of traffic patterns over a historical time interval. The method may include obtaining, by the machine learning model, traffic data corresponding to the roadway traffic system and determining a probability of traffic congestion occurrence based on the obtained traffic data corresponding to the roadway traffic system. The method may include comparing the probability of traffic congestion occurrence to a traffic control probability threshold, and responsive to the probability of traffic congestion exceeding the traffic control probability threshold, adjusting operations associated with one or more traffic controls that correspond to the roadway traffic system.

Abstract: Techniques are provided for modifying coloring of images utilizing machine learning. A trained model is generated utilizing machine learning with training data that includes images of a plurality of different scenes with different illumination characteristics. New original images of a scene may each be downsampled and transformed to a corresponding output image utilizing the trained model. A color transformation from each original image to its corresponding output image may be determined. In an embodiment, the color transformation is determined utilizing a spline fitting approach. The determined color transformations may be applied to each of the original images to generate corrected images. Specifically, the color transformation that is applied to a particular original image is the color transformation determined for the input image that corresponds to the particular original image. The corrected images are utilized to generate a digital model of the scene, and the digital model has accurate model texture.