A method for realizing dish individualization based on a big data environment

Abstract

The application discloses a kind of based on big data environment and realizes the identification control method of dish individualization, including the following steps: according to acquisition information, database is established;According to touch detection technology, the food species of identification and judgment is selected;Characteristic curve comparative analysis determines whether to update or correct database.The application has the advantages of avoiding food damage and fully utilizing the idle computing resources of networked equipment, improving the centralized computing power of server.

Description

Technical Field

[0001] This invention relates to the field of classification and recognition technology, and in particular to a method for personalized identification and control of dishes based on a big data environment. Background Technology

[0002] Due to limitations in computing power, existing smart restaurants lack an effective real-time correction system to achieve real-time monitoring and alarm mechanisms for abnormal behavior. Whether the problem of insufficient centralized computing power of servers can be solved has become a pressing technical challenge in this field in recent years.

[0003] Currently, the image information captured by image recognition cannot provide necessary real-time feedback, which can damage food when grasping it. For example, a "method for grasping with a robotic bionic hand" disclosed in Chinese patent literature, with announcement number CN113942009A, only obtains the soft and hard attribute data of the target object based on the tactile and visual information of the target object, and adjusts the tactile grasping parameters of the robotic bionic hand based on the soft and hard attribute data of the target object and continues to grasp. It lacks a data iteration system and cannot correct and update when the data is incorrect. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of monotonous restaurant menus and the inability of databases to update alarms in real time in the existing technology. It provides a personalized identification and control system and method for dishes based on a big data environment, which has the advantages of avoiding food damage, making full use of the idle computing resources of networked devices, and improving the centralized computing power of servers.

[0005] The technical solution adopted by this invention to solve the above-mentioned technical problems is a method for personalized identification and control of dishes based on a big data environment, comprising the following steps: N1: establishing a database based on collected information; N2: selecting and identifying food types based on tactile detection technology; N3: performing feature curve comparison and analysis to determine whether the database needs to be updated or corrected. This invention solves the problems of existing technologies, such as the inability to update and alarm the database in real time and the insufficient centralized computing power of the server. It has the advantages of avoiding food damage, making full use of the idle computing resources of networked devices, and improving the centralized computing power of the server.

[0006] Preferably, step N1 specifically comprises the following steps: N1.1: Collecting culinary information on dishes from various countries and regions, manually recording, correcting data, and compiling a database through regional analysis; N1.2: Evaluating information, including basic information about the dishes, their popularity, and their target audience; N1.3: Based on the culinary information and market assessment, filtering data to obtain the specific dishes required by the restaurant and the required quantities of raw materials after market assessment; N1.4: Adjusting the quantities of raw materials according to actual dining conditions to build a multi-layered food structure. Through comprehensive collection of culinary information, the constructed food database becomes more complete.

[0007] Preferably, step N2 specifically comprises the following steps: N2.1: Detecting the distance to the food to be grasped, the folding module automatically starts to stretch and approach the food; N2.2: Analyzing the grasping force based on the food's hardness obtained from the sensing system; N2.3: Selecting an appropriate force level for grasping based on the established multi-layered food structure grasping force determination system database; N2.4: If the hardness level is greater than or equal to 4, the conveying module starts, conveying the food to be grasped to the hard food detection end. The characteristic curve is analyzed through PLC contact coding. If the hardness level is less than 4, it is directly recognized by 3D vision at the soft food detection end. Switching between recognition methods improves recognition efficiency.

[0008] Preferably, step N3 specifically comprises the following steps: N3.1: The feature curve after conversion and recognition by the structured feature data acquisition module is compared and confirmed with the target knowledge base; N3.2: Database confirmation, re-identification of the test object, and judgment analysis to determine whether the database needs to be updated or corrected; N3.3: Using the collected real-time data to train the neural network algorithm model. After training, the real-time data of a certain point is input into the algorithm model through the cloud to determine whether the point is a normal point or an abnormal point. When it is an abnormal point, a predictive alarm message is generated; N3.4: Mechanical fault confirmation. If there is a discrepancy in the database identification, it is determined whether the food is damaged. If damaged, the food is discarded; if not damaged, the system automatically applies for warranty fault. The database can be updated and corrected in real time through the cloud, greatly avoiding recognition errors.

[0009] Preferably, in step N1.4, the data selected for the multi-layered structure of the food includes food texture, food smoothness, and food surface feel. The structural model data constructed using these three food features is more accurate.

[0010] Preferably, in step N2.3, the grasping force determination system and corresponding characteristic curves are classified into levels. Level classification improves the efficiency of data analysis.

[0011] Preferably, in step N3.3, edge computing is used for data identification training and correction updates. Edge computing makes full use of the idle computing resources of networked devices, solving the problem of insufficient centralized computing power on servers. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments or prior art, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0014] Figure 1 This is an overall block diagram of a method for personalized identification and control of dishes based on a big data environment, according to the present invention. Detailed Implementation

[0015] The specific implementation of the technical solution of the present invention will be further described below through examples and in conjunction with the accompanying drawings.

[0016] The present invention describes a method for personalized identification and control of dishes based on a big data environment, referencing... Figure 1As shown, the method includes the following steps: N1: Establish a food database based on the collected data; N2: Select and identify food types based on tactile detection technology; N3: Perform feature curve comparison analysis to determine whether the database needs to be updated or corrected. Step N1 specifically includes the following steps: N1.1 Collect cooking information on dishes from various countries and regions, and manually record, correct, and compile a database through regional analysis; N1.2 Evaluate information, including basic information about the dishes, their popularity, and their target audience; N1.3 Based on the cooking information and market assessment, perform data filtering to obtain the dishes required by specific restaurants and the required raw material quantities after market assessment; N1.4 Adjust the raw material quantities according to actual dining conditions, and construct a multi-layered food structure, food texture, food smoothness, surface tactile feel, a level judgment system for the grasping force, and corresponding feature curves. Step N2 specifically includes the following steps: N2.1 Detecting the distance to the food to be grasped; the folding module automatically starts stretching to approach the food when the distance is too long; N2.2 Analyzing the food's hardness based on the sensor system's analysis of the food's texture; N2.3 Selecting an appropriate force level for grasping based on the established database of food multi-layer structure grasping force levels; N2.4 If the hardness level is greater than or equal to 4, the conveying module starts, conveying the food to be grasped to the hard food detection end. The characteristic curve is analyzed through PLC contact coding. If the hardness level is less than 4, it is directly identified by 3D vision at the soft food detection end. Step N3 specifically includes the following steps: N3.1 The structured feature data acquisition module converts and identifies the feature curves and compares them with the target knowledge base for confirmation; N3.2 Database confirmation, re-identifying the test object, performing judgment analysis, and determining whether the database needs to be updated or corrected; N3.3 Using the collected real-time data to train the neural network algorithm model, after training, the real-time data of a certain point is input into the algorithm model through an SSH client and cloud connection to determine whether the point is a normal point or an abnormal point, and generating a predictive alarm information when it is an abnormal point; N3.4 Mechanical fault confirmation, if there is a discrepancy in the database identification, determining whether the food is damaged. If damaged, the food is discarded; if not damaged, the system automatically applies for warranty fault. The multi-layer structure selection data of the food includes food texture, food smoothness, and food surface feel; the grasping force judgment system and corresponding feature curves adopt a hierarchical classification; edge computing is used in step N3.3 for data recognition training and correction updates.

Claims

1.A method for identifying and controlling dish personalization based on a big data environment, the method comprising the steps of: The method comprises the following steps: N1: according to the collection information, establish a database; N2: according to the touch detection technology to select the identification and judgment of food category; If the hardness grade is greater than or equal to 4, through the PLC contact coding analysis characteristic curve, the hardness grade is less than 4, directly in the soft food detection end through 3D vision recognition; N3: characteristic curve comparative analysis, determine whether to update or correct the database. 2.The method of claim 1, wherein, The step N1 specifically comprises the following steps: N1.1: collect the cuisine cooking information of each country and region, and record, data correction and database preparation through regional analysis; N1.2: evaluation information, including basic information of dishes, preference degree, and audience degree; N1.3: according to the cuisine cooking information and market evaluation, data screening is carried out to obtain the specific restaurant required dishes and the required raw material quantity after market evaluation of dishes; N1.4: adjust the raw material quantity according to the actual dining situation, and build food multi-level structure. 3.The method of claim 1, wherein, The step N2 specifically comprises the following steps: N2.1: detect the distance of the food to be grabbed, and the folding module automatically starts to stretch and approach the food to be detected; N2.2: according to the food softness and hardness obtained by the perception system analysis, the grabbing force is analyzed; N2.3: according to the food multi-level structure grabbing force determination system database, select the appropriate force level for grabbing; N2.4: if the hardness grade is greater than or equal to 4, the conveying module starts to convey the food to be detected to the hard food detection end. 4.The method of claim 1, wherein, The step N3 specifically comprises the following steps: N3.1: the structured feature data acquisition module converts the identified feature curve and the target knowledge base for comparison and confirmation; N3.2: database confirmation, re-identify the test object, and perform determination analysis to determine whether the database needs to be updated or corrected; N3.3: use the collected real-time data to train the neural network algorithm model; N3.4: mechanical fault confirmation, database recognition has difference, determine whether the food has been damaged, if damaged, discard the food, if not damaged, the system automatically applies for warranty fault. 5.The method of claim 4, wherein, The step N3.3 specifically comprises: after training, input the collected real-time data of a certain point to the algorithm model through the cloud, and judge whether the point is normal or abnormal point in turn. 6.The method of claim 2, wherein, In the step N1.4, the food multi-level structure selects data including food texture, food smoothness and food surface touch. 7.The method of claim 3, wherein, In the step N2.3, the grabbing force determination system and the corresponding characteristic curve are divided into grades. 8.The method of claim 4 or 5, wherein, In the step N3.3, edge computing is used for data recognition training and correction update.

Images