Stored grain inventory management neural network

A system, comprising:

• one or more processors;
• and a memory comprising instructions, wherein the one or more processors are configured by the instructions to:
• (a) from an acquisition system comprising an electromagnetic transceiver and a radio frequency (RF) antenna array coupled thereto that comprises a plurality of antenna probes that are distributed around an interior holding space of a container and are operable by said electromagnetic transceiver in transmitter/receiver pairs to impart RF energy to held contents of the container from transmitting antennae of said transmitter/receiver pairs and collect resulting scattered RF signals from receiving antennae of said transmitter/receiver pairs, receive first uncalibrated electromagnetic data at a plurality of radio frequencies, which first uncalibrated electromagnetic data comprises S-parameter measurements composed of both magnitude and phase;
• (b) process the first uncalibrated electromagnetic data to derive therefrom a phaseless and magnitudinally normalized electromagnetic dataset that preserves magnitudinal information from the S-parameter measurements, normalized transmitter-wise, and disregards phase;
• and
• (c) generate prediction parameters for said container and said contents thereof by passing the phaseless and magnitudinally normalized electromagnetic dataset through a neural network trained on likewise phaseless and magnitudinally normalized data corresponding to a synthetic training set previously derived using a computer model of the container and simulated antenna probes.