Adaptive quantitative sub-packaging method for fried rice with multiple side dishes in central kitchen and apparatus therefor

Patent metadata

Field	Value
Publication number	US-11897648-B2
Application number	US-202117767446-A
Country	US
Kind code	B2
Filing date	Dec 3, 2021
Priority date	Dec 1, 2021
Publication date	Feb 13, 2024
Grant date	Feb 13, 2024

Abstract

The present invention belongs to the technical field of food processing, and in particular, relates to an adaptive quantitative sub-packaging method for fried rice with multiple side dishes in a central kitchen and an apparatus therefor. The present invention mainly includes self-construction of a key side dish recognition model, self-evaluation of the key side dish recognition model, adaptive quantification of batch fried rice, and the apparatus matched with the three-step operation.

First claim

What is claimed is:

1. An adaptive quantitative sub-packaging method for fried rice with multiple side dishes in a central kitchen, comprising the following steps:
step I, self-construction of a key side dish recognition model, comprising the following processes:
process I, for fried rice with a standard formula comprising rice, edible oil, m kinds of side dishes, and n kinds of seasoning liquids, obtaining finished fried rice to be sub-packaged by frying with a standardized process and fixed equipment of the central kitchen, and selecting an i-th side dish H_i of the finished fried rice to be sub-packaged as a key side dish required for quantification;
process II, taking m1 kg of the finished fried rice to be sub-packaged as a fried rice sample A for model construction, and separating the key side dish H_i in the fried rice sample A for model construction from other fried rice ingredients to obtain a fried rice sample A&H_i containing only the key side dish H_i, and a fried rice sample A-H_i without the key side dish H_i;
process III, starting a central control module matched with the method, wherein the central control module comprises a model self-construction module, a model self-evaluation module, and an adaptive quantitative module for the fried rice, and the model self-construction module, the model self-evaluation module, and the adaptive quantitative module for the fried rice are electrically connected, and
the model self-construction module comprises a material box self-resetting module and a startup model self-construction module, the model self-evaluation module comprises a material box self-resetting module and a startup model self-evaluation module, and the adaptive quantitative module for the fried rice comprises an adaptive quantitative operation module;
first, starting the material box self-resetting module in the model self-construction module in the central control module, transmitting, by the material box self-resetting module, a signal to the central control module, so as to control movement of a position sensor and a main conveyor belt, and resetting a material box on the main conveyor belt to completely coincide with a predefined station; and then laying the fried rice sample A&H_i in a single-layer and non-overlapping manner in w material boxes in other stations before an imaging station, and laying the fried rice sample A-H_i in a single-layer and non-overlapping manner in w material boxes in stations before the imaging station to obtain w material boxes containing only thin layers of the fried rice sample A&H_i and w material boxes containing only thin layers of the fried rice sample A-H_i, wherein a is w positive integer;
process IV, inputting, to the model self-construction module, a product name of the fried rice to be sub-packaged X1, a name of the key side dish X2, a station number corresponding to the material boxes containing the thin layers of the fried rice sample A&H_i and a number of the key side dishes H_i contained in each of the material boxes, and relevant information of a station number corresponding to the material boxes containing the thin layers of the fried rice sample A-H_i; and clicking the startup model self-construction module in a boundary of the model self-construction module, transmitting a signal to the central control module to control a camera to self-acquire the fried rice samples A&H_i and A-H_i and images I_A&H_i_h_j, I_A-H_i_h_j, and I_O_h_j of w empty material boxes illuminated by b light sources with different bands λ1, λ2, . . . , λ(b−1), and λb, self-optimizing, by the central control module, an optimal band λ_A and an optimal segmentation threshold C for key side dish recognition, and self-calculating and displaying an optimal recognition rate for modeling of D, wherein h∈[1, b], j∈[1, w], and h and b are positive integers; and
process V, when the model recognition rate D is greater than or equal to an expected recognition rate E, proceeding to step II of an operation of “self-evaluation of the key side dish recognition model”; and when the model recognition rate D is less than the expected recognition rate E, repeating the processes II, III, IV, and V of the step I;
step II, self-evaluation of the key side dish recognition model, comprising the following processes:
process i, taking m2 kg of the finished fried rice as a fried rice sample B for model construction, and separating the key side dish H_i in 2*(m2)/3 kg of the fried rice sample B from other fried rice ingredients to obtain a fried rice sample B&H_i containing only a key side dish H_i, a fried rice sample B-H_i without a key side dish H_i, and (m2)/3 kg of a fried rice sample B_B containing all ingredients;
process ii, starting the material box self-resetting module in the model self-evaluation module in the central control module, and transmitting, by the material box self-resetting module, a signal to the central control module to control the position sensor to reset the material box on the main conveyor belt to completely coincide with the predefined station after one cycle of operation; and then laying, by workers, the fried rice sample B_B in a single-layer and non-overlapping manner in w material boxes in the stations before the imaging station, laying the fried rice sample B-H_i in a single-layer and non-overlapping manner in w material boxes in the stations before the imaging station, and laying the fried rice sample B&H_i in a single-layer and non-overlapping manner in w material boxes in the stations before the imaging station to obtain w material boxes containing only thin layers of the fried rice sample B_B, w material boxes containing only thin layers of the fried rice sample B-H_i, and w material boxes containing only thin layers of the fried rice sample B&H_i;
process iii, inputting, to the model self-evaluation module in the central control module, a station number of the material boxes containing the thin layers of the fried rice sample B_B and a number of the key side dishes H_i contained in each of the material boxes, a station number of the material boxes containing the thin layers of the fried rice sample B-H_i, a station number of the material boxes containing the thin layers of the fried rice sample B&H_i, and relevant information of a number of the key side dishes H_i contained in each of the material boxes;
process iv, clicking the startup model self-evaluation module in the model self-evaluation module in the central control module, transmitting, by the startup model self-evaluation module, a signal to the central control module to open a light source corresponding to the optimal band λ_A through the central control module according to the optimal band λ_A and the optimal segmentation threshold C obtained in the step I, and acquiring, by the camera, an image I_B_B_j of w fried rice samples B_B, an image I_B-H_i_j of w fried rice samples B-H_i, and an image I_B&H_i_j of w fried rice samples B&H_i illuminated by the light source corresponding to λ_A; and sequentially recognizing, by the central control module, numbers of the key side dishes H_i in the images I_B_B_j, I_B-H_i_j, and I_B&H_i_j according to the optimal segmentation threshold C, and comparing recognition results with numbers of the key side dishes H_i input by the workers to calculate an evaluation recognition rate F; and
process v, when the evaluation recognition rate F is greater than or equal to an expected recognition rate G, automatically saving, by control software, the optimal band λ_A and the optimal segmentation threshold C and proceeding to step III of adaptive quantification of the fried rice; and when the model recognition rate F is less than the expected recognition rate G, repeating all the processes of the step I; and
step III, adaptive quantification of batch fried rice sub-packaging, comprising the following processes:
process 1, starting batch production to obtain the finishe

Assignees

• Univ Jiangsu

Inventors

• Zou Xiaobo
• Li Chuang
• Shi Jiyong
• Huang Xiaowei
• Li Zhihua
• Shen Tingting
• Hu Xuetao
• Zhang Jiukai
• Xiao Jianbo
• Guo Zhiming

Classifications

• B65B57/145 Primary

  for fluent material · CPC title

• B65B1/06

  by gravity flow · CPC title

• B65B1/32

  by weighing (check-weighing of filled containers or receptacles B65B1/46) · CPC title

• B65B25/001

  of foodstuffs, combined with their conservation (B65B25/041, B65B25/062, B65B25/067 take precedence) · CPC title

• B65B43/52

  using roller-ways or endless conveyors · CPC title