Convolution modeling and learning system for predicting geometric shape accuracy of 3D printed products

A method for predicting geometric shape accuracy of 3D printed products, the method comprising:

• determining an input representation of an object in response to first input data;
• generating an electronic model of the object in response to the input representation;
• predicting a shape deviation of the electronic model relative to the input representation of the object;
• determining at least one layer function to define at least one layer of the electronic model of the object;
• calculating a convolution transfer function that when convolved with the at least one layer function generates the at least one shape deviation of the electronic model, the at least one shape deviation comprising a whole shape deviation of all layers of the at least one layer of the electronic model of the object and comprising a feature error comprising a distortion of a specific geometric feature of the at least one layer of the electronic model of the object, wherein convolving of the convolution transfer function includes data representative of a state vector of the at least one layer and a system response combined in the convolution transfer function;
• convolving the convolution transfer function with the at least one layer function to generate the at least one shape deviation;
• and
• generating machine control instructions for an additive manufacturing machine corresponding to the first input data offset according to the convolution transfer function to cancel the at least one shape deviation from the electronic model.