Neural logic controllers
AssigneeSiemens Aktiengesellschaft
InventorsArquimedes Martinez Canedo, Hasan Sinan Bank, Hartmut Ludwig
Life of this patent
Estimating the term…
Abstract
A method for executing a machine learning model with a controller includes a processor within the controller writing input values to a process image within the controller. The term process image refers to a predefined address space within volatile memory of the controller. A co-processor connected to the controller reads the input values from the process image and applies a machine learning model to the input values to generate output values. The co-processor writes output values to the process image and the processor reads those output values from the process image. The process can then execute an application program that utilizes the one or more output values.
Claims13
A method for executing a machine learning model with a controller, the method comprising:
- writing, by a processor within the controller, one or more input values to a process image within the controller, wherein the process image is a predefined address space within volatile memory of the controller, wherein the input values include parameters of a neural code corresponding to a machine learning model;
- reading, by a co-processor connected to the controller, the input values during runtime of the controller from the process image;
- executing, by the co-processor, the neural code using the input values to generate one or more output values;
- writing, by the co-processor, the one or more output values to the process image;
- reading, by the processor, the one or more output values from the process image;
- and
- executing, by the processor, a control program that utilizes the one or more output values.
A system for executing machine learning code within a controller device, the system comprising:
- a programmable logic controller (PLC) comprising:
- a non-transitory storage medium storing a process image;
- a processor executing a plurality of instructions, wherein a subset of the instructions depend on execution of machine learning code not executable by the processor;
- a co-processor connected locally to the programmable logic controller and configured to:
- (i) read input values during runtime of the programable logic controller from the process image, wherein the input values are written by the processor to the process image, the input values including parameters of the machine learning code,
- (ii) use the input values to execute the machine learning code, thereby yielding one or more output values, and
- (iii) write the one or more output values to the process image, wherein the processor is configured to read the one or more output values from the process image and execute a control program that utilizes the one or more output values.
Dependencies are inferred from the claim text.