Abstract: Herein are techniques to generate candidate rulesets for machine learning (ML) explainability (MLX) for black-box ML models. In an embodiment, an ML model generates classifications that each associates a distinct example with a label. A decision tree that, based on the classifications, contains tree nodes is received or generated. Each node contains label(s), a condition that identifies a feature of examples, and a split value for the feature. When a node has child nodes, the feature and the split value that are identified by the condition of the node are set to maximize information gain of the child nodes. Candidate rules are generated by traversing the tree. Each rule is built from a combination of nodes in a tree traversal path. Each rule contains a condition of at least one node and is assigned to a rule level. Candidate rules are subsequently optimized into an optimal ruleset for actual use.

Abstract: The present invention relates to a fuel dispenser control system which controls the fuel dispensing process and accepts payment for the fuel dispensed through a card reader. The dispenser controller has a microprocessor with read-only-memory (ROM) and read-and-write-memory (RAM). A series of commands are stored in the ROM for controlling the dispensing process and accepting payment for the fuel dispensed. Configuration circuits' translate the communication language of the dispenser controller into a communication protocol readable by the dispensers such that the dispensing process of various dispenser brands can be controlled and payment for the fuel accepted. Response data from the dispenser is stored in the RAM, and then passed up a register system.

Abstract: A method is disclosed for developing the contour of tools employed for forming members exhibiting complex shapes. The members may be precipitation, heat treatable, metals or metal alloys which are age formed, although they be of any material which exhibits a relationship between a strain applied by a forming tool, or otherwise, and a resulting strain after release of the applied strain. The resulting member may be formed to the desired contour as a result of exposure to an elevated temperature but the member may also be cold formed. The invention is particularly concerned with a methodology for simplifying the analysis of integrally stiffened structures of complex shape. The method of the invention assures proper results on the first occasion the tool is used, thereby resulting in considerable savings of labor and material.

Abstract: A communication module has a backplane interface circuit that enables the module to communicate with other components of the programmable controller, and has a network interface circuit which sends and receives messages via a serial communication link. A memory within the module stores a designation of a starting time, a designation of a length of a message repeat interval, and a file of message information. A clock circuit determines when the starting time occurs and thereafter determines the expiration of each successive occurrence of the message repeat period. A control mechanism responds to the clock circuit determining an occurrence of the starting time or determining an expiration of a message repeat interval, by operating network interface circuit to transmit a message, that contains the message information, over the serial communication link.

Abstract: A 1-bit error correction circuit based on CRC calculation is provided with a syndrome generation circuit which determines input parallel data of m bits and which have been converted from n number of m-bit serial data. A 1-bit error detection circuit cyclically supplies a syndrome to a remainder calculation circuit and decodes remainder data obtained from this cyclic supply and detects 1-bit errors. A actual data reproduction circuit calculates the exclusive OR of output data of a predetermined register of a 1'st.about.n'th register of a syndrome generation circuit and data supplied to a predetermined register and obtains parallel data which is the actual data. A correction circuit which calculates a exclusive OR of parallel data obtained from a actual data reproduction circuit and 1-bit error data detected by the 1-bit error detection circuit and outputs corrected data.

Abstract: Disclosed is a method and apparatus which provide control over industrial operations utilizing sequential processes. The invention focusses particularly upon machine transfer line operations in which milling, drilling, boring, and other machining processes are performed upon a work piece on a transfer line according to predetermined sequences of operations. The invention utilizes a sequential, integrated motion and process logic educator (SIMPLE) controller based on an IBM personal computer platform located at each workstation and at the transfer driver workstation. Intelligent monitoring of the operations of processes defined as a sequence of actuations is achieved with networked SIMPLE controllers wherein distributed computing is provided at the various workstations which comprise the transfer line. Distributed computing and a system of interlocks for use with a sequential program sufficiently narrows the scope of operations at each workstation to a manageable level.

Abstract: A process control apparatus varies a setpoint weighting factor .beta. according to both an error signal and a measurable dynamic characteristic of the process. The dynamic characteristic is either in terms of the normalized gain or normalized dead time of the process. For a PID controller, an initial value of the initial setpoint weighting factor .beta., a time at which the initial value should be varied, and the new value of the factor are all automatically set by a self-tuning procedure, wherein a processor is programmed to execute a set of correlation formulae which is applicable to a wide range of industrial processes.

Abstract: The present invention is a system and method for improving stream composition and/or product property estimates from process models, which better reflect the true conditions of the process. One or more process simulation models are run on a computer in parallel with the actual process to provide estimates of stream composition and/or product properties to be used to control the process. Adjustments are made to the models to maintain them in alignment with continuously measured key process variables that are closely related to stream composition and/or product properties where such a relationship exists. This greatly improves the ability of the model to track the actual process. Additional adjustments are made to both the models and to the model estimates based on differences between measured and calculated stream composition/product properties.

Abstract: A method for optimizing the efficiency of a set of machines comprising a double-controlled turbine and a generator. A correlation is determined between the efficiency .beta., flow Q, variable height of fall H, gate opening .DELTA..gamma. and runner opening .phi. on a model set of machines by test series. The determination yields a model optimum curve for the respective height of fall, the model optimum curve having the optimum gate opening as a function of flow and runner opening. An optimum efficiency for an operating point of a large-scale system is determined by, based on the model optimum curve, making variations of the gate opening .DELTA..gamma. and of the runner opening .phi. on the large-scale system, the variations being such that the flow of the model set of machines would remain constant for the same variations, and wherein the flow of the large-scale system is also presumed to be constant. The power outputs before and after the variations are compared to determine stepwise the .DELTA..gamma./.phi.

Abstract: A method of rapid prototyping through controlled layerwise deposition and extraction involving the precision control of material subtractive or removal processes in conjunction with material additive processes wherein each layer is formed by selectively depositing and removing portions of the layers in accordance with corresponding software slices of a computer model of the prototype part, and with complementary support material included in the layers and ultimately removed when the complete part has been fabricated.

Abstract: A retrofit auto-indexing system for a turret punch press is described in which a predetermined location is provided to which both the upper and lower turrets of a non-indexing turret punch press are rotated. Upper and lower punching tools are mounted respectively in the upper and lower turrets in such a manner so as to be rotated simultaneously, using a single servomotor, to a desired angle. The auto-indexing system is controlled using a microcomputer which is connected to receive the numerical control programmed data, determine whether the code it is receiving relates to the auto-indexing function and, when it does, institute the instructed auto-indexing function. Otherwise, the microcomputer passes the numerical control programmed data directly to a minicomputer which operates the turret punch press in a known manner.

Abstract: The detection and correction of errors in digital data transmitted by or stored in a media channel is provided by processing the data through a triple orthogonally interleaved error correction system. On the transmit/store side of the system, the data is encoded three times prior to placement in the media channel with two different interleaving steps performed between the encoding steps. The first interleave is an orthogonal row shuffling interleave that provides enhanced protection against burst errors. On the receive/play back side, the data is decoded and deinterleaved, with included errors detected and corrected to enable recovery of the original data. To enhance the error correction, a circuit is used for generating a symbol accurate error flag identifying symbols containing errors thereby allowing the error correcting decoders to focus on and correct the data.

Abstract: The present invention stores the image data of an original image pattern read by an image reader in processed data storage means, and stores the calculation formulas for producing designs such as checkers and strip patterns in calculation formula storage means. And, the needle location data of a selected design is generated by substituting the line and column numbers of the image data in the calculation formulas stored in the calculation formula storage means. By this construction, within the image data, embroidering data of the design which is not affected by the visible outline thereof can be formed. In addition, by carrying out the logical AND between the data of the needle locations corresponding to a design and the image data of an original data, embroidering data of the design which is not affected by the visible outline of the image data is formed within the image data.

Abstract: An error control decoder (44) for use in decoding signals encoded with an M-ary convolutional code and a method for decoding such a code. The error control decoder includes a branch metrics module (46) that determines differences between the correlation values of each of eight possible symbols used to represent data encoded and the largest of these correlation values. An add-compare-select module (50) determines path metric values for each of 64 states by adding selected branch metric values to prior state metrics for the two possible paths that lead to a current state metric. The minimum path metric of the two is then assigned as a new state metric, and a logic level identifying the selected path metric is stored in a path history module (62). This procedure is repeated for each of the 64 states. A minimum state metric from the prior symbol period is determined and used to normalize the current state metrics to avoid overflow.

Abstract: A peripheral data acquisition, monitor, and adaptive control system is disclosed in which a personal computer (PC) and one or more input/output (I/O) Bridge devices interface signals from electronically-controlled devices to the PC via the keyboard port, thus permitting data, such as measurement data, to be automatically and directly entered into application programs such that the PC can take action and control outputs based upon the measurement data. A software control program is also disclosed which allows the user to configure the system for orienting the user as to which devices are being controlled, reading digital and analog inputs, making decisions based upon the information using specific user-defined conditions, and setting digital outputs. The program will also keep track of all I/O events by storing them to a disk file.

Abstract: First and second devices, such as a laser (11) and an optical fiber (12), are aligned by first positioning the laser on an x-y-z table (13) (such a table is capable of responding to electrical signals to make precise movements in mutually orthogonal x,y and z directions). The laser beam is imaged onto a machine vision camera (19) which develops signals representing the image of the laser beam and directs them to a computer (16). The computer analyzes the signal, calculates the center of the image, and determines from such calculation any deviations in the x and y directions of the position of the laser from its desired alignment position. Next, the optical fiber (12) is imaged on a machine vision camera (17). Signals from the camera representative of the image of the optical fiber end are directed to the computer (16) which calculates the center of the image and determines any deviation from its desired position.

Abstract: In an error correction system, which incorporates Cyclic Redundancy Check (CRC) and a Reed-Solomon (RS) code, original data are encoded simultaneously by CRC and RS encoders. The CRC encoder processes the words 0-9 of the original data to form CRC check bytes, which become the last bytes in words 6, 7, 8 and 9. The RS encoder processes 104-byte words 0-9 of original data to form 16 redundant RS bytes per word. The words 0-5 are simultaneously supplied from a codeword memory to CRC and RS encoders. The redundant RS bytes generated by the RS encoder are sent back to the codeword memory, while the CRC encoder saves the CRC check bytes in local registers. For words 6-9 the RS encoder processes the bytes of original data and stops processing before the CRC bytes. It then temporarily saves the generated data in the codeword memory and waits for the CRC encoder to finish encoding. Once the CRC encoder completes, the CRC bytes corresponding to the words 6-9 are processed by the RS.

Abstract: Method and system for evaluating sheet metal forming tooling design, for use with a draw die including a punch and binder designed to form the sheet metal into a part, utilizing improved implicit time integration methods that reduce numerical instability, thereby enhancing convergence of numerical solutions. The sheet metal and tool surface of the punch are each represented as a mesh having a plurality of nodes. Contact nodes between the sheet metal mesh nodes and the tool surface mesh can be identified. A first embodiment includes minimizing discontinuities generated by unloading by determining a stress increment of a sampling point in the sheet metal mesh according to an incremental deformation theory of plasticity. A second embodiment includes modelling a draw-bead as a plurality of nonlinear elastic springs to minimize discontinuities in the spring force during unloading.

Abstract: A personal computer (PC) and a programmable controller are used to automatically reposition the turn bars and compensation rollers used in a folder for a web printing press. By providing linear transducers for each of the turn bars and compensation rollers disposed at each level of the folder, analog information as to the actual position of each is provided to the programmable controller. The PC provides digital information to the programmable controller as to the desired position and an error signal is computed for controlling the operation of the drive motors so as to reduce the error signal to zero. When setting up the folder to run a new job, the turn bar settings and compensation roller positions are manually accomplished, but upon command, the settings for that job may be stored in the memory of the PC for use the next time that particular job is to be run.

Abstract: A numerical control apparatus having a three-dimensional graphic display function is provided. A three-dimensional space is divided into lattice points in storing the form data of a workpiece and a tool. When the same lattice data of a plane including an X-axis and Y-axis of three-dimensional coordinates is continuous in a Z-direction, i.e., the same form continues, those lattice point data are stored using a piece of lattice data and a continuity number, and the lattice data are stored in the form of a list of links. This allows the form data of the workpiece and tool to be stored using a small memory capacity and the accuracy of display to be improved where needed without increasing the memory capacity.