Intelligent canteen health management system and management method based on multi-source data analysis

Abstract

The invention relates to an intelligent canteen health management system and management method based on multi-source data analysis, and the system comprises an identification perception system, an acquisition system, a data processing system, and a nutrition analysis system. The identification perception system is arranged above a dining table of a canteen and is used for identifying specific dishes and surrounding environment; the identified and perceived dishes are collected, the collected data are uploaded to the data processing system to be processed and analyzed, and finally a report is generated through the analyzed conclusion. According to the invention, global data connection and dynamic decision engine can be realized, a closed-loop service architecture is adopted as a whole, an end-edge-cloud three-level response system is constructed, and the data acquisition rate is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent catering management and health data analysis, and relates to an intelligent canteen health management system and method based on multi-source data analysis. BACKGROUND

[0002] Traditional canteens are accelerating the transformation towards intelligence and health. However, the existing intelligent canteen system generally faces three major bottlenecks: multi-source data fragmentation, health intervention lag, and service loop missing.

[0003] Specifically, the following technical solutions exist in the field of intelligent canteens at present:

[0004] 1. RFID (Radio Frequency Identification) system: Through the implantation of RFID chips (working frequency 13.56 MHz) in the tray, automatic settlement is realized. The typical disclosed "intelligent tableware recognition system" uses a radio frequency reader to complete dish recognition at a speed of 0.3 seconds per piece, and generates an account together with the background order system

[0005] 2. Multi-modal payment technology: The "face-two-dimensional code dual-mode payment terminal" proposed by some patents integrates a living body detection algorithm (false recognition rate ≤0.001%) and a dynamic two-dimensional code generation module, which supports identity verification and payment operation within 300ms

[0006] 3. AI nutrition analysis engine: The "diet assessment method based on image recognition" disclosed by some patents uses an improved ResNet-50 model (accuracy 98.7%) for dish recognition, and calculates the calorie value (error ±5%) in combination with weight sensor data

[0007] These technologies perform well in automatic settlement, payment efficiency, and basic nutrition analysis, but have the following key defects:

[0008] 1. Data dimension missing: Existing systems only collect single-dimensional data such as dish weight and type, and cannot obtain deep behavioral characteristics such as user eating speed (e.g. number of times of picking up dishes per unit time) and food temperature (affecting nutrient absorption)

[0009] 2. Model update lag: The nutrition recommendation model relies on offline training (update cycle ≥24h), and cannot respond in real time to physiological index changes (e.g. no dynamic adjustment mechanism when blood glucose fluctuation during meal exceeds 2mmol / L)

[0010] 3. Multi-source data fragmentation: Although some patents propose to integrate payment and health data, they do not establish a cross-device data synchronization mechanism (e.g. timestamp deviation of weighing sensor and body temperature monitoring device >500ms), resulting in inaccurate analysis results

[0011] At the same time, there is currently no technical solution that can meet the following requirements:

[0012] 1. A closed-loop system architecture that integrates multimodal data (images, weight, temperature, user vital signs) in real time;

[0013] 2. A dynamic health intervention model based on time series analysis (response delay < 1s);

[0014] 3. Cross-hardware protocol data synchronization mechanism (time deviation < 50ms);

[0015] Therefore, a smart canteen health management system and management method based on multi-source data analysis are designed. Summary of the Invention

[0016] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a smart canteen health management system and method based on multi-source data analysis. This invention enables full-domain data connectivity and a dynamic decision engine, adopts a closed-loop service architecture, and constructs a three-level "end-edge-cloud" response system to improve data collection rates.

[0017] This invention is achieved through the following technical solution: a smart canteen health management system based on multi-source data analysis. The system includes an identification and sensing system, a data acquisition system, a data processing system, and a nutrition analysis system. The identification and sensing system is installed above the canteen's food counters to identify specific dishes and the surrounding environment. The system collects the identified and sensed dishes, uploads the collected data to the data processing system for processing and analysis, and finally generates a report based on the analysis conclusions.

[0018] Preferably, the identification and sensing system includes a camera and a temperature and humidity sensor, both installed on the top of the canteen. The camera is installed at a 90° downward angle on the top of the dining table, 1m ± 0.1m away from the dining table, to ensure coverage of a 1m diameter food pick-up area.

[0019] A smart canteen health management method based on multi-source data analysis, the method comprising the following steps:

[0020] 1) Equipment installation;

[0021] 2) Customer meal check-in;

[0022] 3) Dish identification;

[0023] 4) Data acquisition;

[0024] 5) Health data analysis and summary;

[0025] 6) Comprehensive data analysis: meals, reports, anomalies;

[0026] 7) Push out analysis summaries and health advice;

[0027] 8) Health data analysis;

[0028] 9) Food and drink check-in.

[0029] Preferably, the acquisition system is a multimodal data synchronous acquisition system, including...

[0030] 1) Visual data stream: Acquire 1920×1080 resolution images at a frame rate of 30fps;

[0031] 2) Biometric data: Food temperature (±0.5℃) is monitored via an infrared array (8×8 pixels) built into the plate;

[0032] 3) Spatiotemporal alignment mechanism: The NTP protocol is used to achieve time synchronization of multiple devices (error <10ms);

[0033] 4) Payment data flow: Integrate third-party payment SDKs (WeChat v3.0.9 / Alipay v4.18.0 / UnionPay v6.2.1).

[0034] Preferably, the health data analysis summary includes:

[0035] 1. Heterogeneous data fusion processing

[0036] 1) Establish a unified data format and model;

[0037] 2) Develop a data cleaning pipeline to remove outliers (such as sudden weight changes > 500g / second) and compensate for environmental interference (initiate infrared temperature measurement calibration when humidity > 85%).

[0038] 3) Privacy protection processing: Biometric data is transmitted with SM4 encryption, and user profile data undergoes differential privacy processing (ε = 0.45);

[0039] 2. Nutrition Knowledge Map

[0040] 1) Construct a 9-dimensional nutrition database containing 1200 kinds of ingredients;

[0041] 2) Food exchange portion parameter table: 6 food exchange matrices (grains and tubers 25g / serving, fruits and vegetables 500g / serving, etc.), and a calorie equivalence comparison table for 120 kinds of foods;

[0042] 3. Personalized Solution Engine

[0043] 1) A food exchange algorithm based on dynamic programming;

[0044] 2) Nutrition plan generation protocol: The fields include 16 nutritional indicators + 8 types of allergen alerts.

[0045] Preferably, the comprehensive data analysis specifically includes:

[0046] 1. Real-time nutrition calculation engine

[0047] 1) An improved YOLO algorithm model is used to achieve real-time food identification with a recognition latency of <200ms and an accuracy of ≥99.5%;

[0048] 2) Real-time calculation of nutrient intake:

[0049]

[0050] Where w is the current weight of the dish, and e is the baseline nutritional value;

[0051] 2. Tiered response mechanism

[0052] 1) Establish a three-level early warning system:

[0053] Warning level Trigger condition Response mode Level 1 Single sodium intake > 30% of daily recommended value LED screen flashing warning Level 2 Consecutive 3 meals of protein exceeding the standard Mobile terminal push alternative recipes Level 3 Diabetes risk index > 0.85 Smart tableware restricts meal quantity

[0054] 2) Payment linkage rules: The payment function is frozen when the deviation rate is >20%, and the order is automatically released after 30 seconds of payment timeout.

[0055] 7. The management method of the smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: the health data analysis method is:

[0056] 1. Dynamic updates to user profiles:

[0057] P new =0.85×P old +0.15×Scurrent0.15×S current

[0058] P represents the health profile, and S represents the current dietary characteristics;

[0059] 2. Long-term trend report generation:

[0060] Generation cycle: Daily / Weekly;

[0061] Indicators include: trend charts of 12 metabolic parameters and radar chart of nutritional balance.

[0062] As a preferred method, the food check-in record will be fed back to the dish recognition system for closed-loop execution. The specific execution method is as follows:

[0063] 1. Verify the deviation rate between actual intake and recommendations (threshold <20%);

[0064] 2. Update user compliance rating:

[0065]

[0066] 3. Hardware status maintenance: Camera lens contamination alarm (triggered when light transmittance <85%).

[0067] The beneficial effects of this invention are as follows:

[0068] The smart canteen health management system and method based on multi-source data analysis designed in this invention can achieve...

[0069] 1. Through a heterogeneous device protocol conversion module (supporting 6 types of protocols such as Modbus / TCP and MQTT) and a spatiotemporal alignment algorithm (positioning error ≤5cm), it integrates 9-dimensional data streams in real time (including food temperature, user food picking behavior, etc.).

[0070] 2. Dynamic Decision Engine: Develop a health prediction system based on an LSTM-Bayesian hybrid model, updating the user's health profile every 30 minutes. When the connected continuous glucose monitor (CGM) data fluctuates above 3.9 mmol / L (referring to the "Guidelines for the Prevention and Treatment of Type 2 Diabetes in China"), it triggers the recommendation of alternative recipes.

[0071] 3. Closed-loop service architecture: Construct a three-tiered "device-edge-cloud" response system to achieve:

[0072] 1) Complete anomaly detection and LED screen warning push within 800ms

[0073] 2) Generate a personalized health report (including analysis of 8 metabolic indicators) within hours.

[0074] 3) Implement RFID electronic tag locking for high-risk dishes.

[0075] This system breaks through the traditional one-way service model and achieves the following in complex dining scenarios:

[0076] 1. The effective data collection rate reached 88.7% ± 2.1%;

[0077] 2. Critical operation delays reduced to 1.6 seconds;

[0078] 3. The compliance rate with medical orders was 79.4%;

[0079] Compared with existing smart canteen systems, the health management system and method based on multi-source data analysis provided by this invention have the following advantages:

[0080] 1. Achieve precision nutrition management

[0081] By integrating multimodal data and improving algorithm models, the error rate of traditional nutritional analysis was reduced from 15% to 3.2%, and the accuracy rate of dish identification reached ≥99.5%. In pilot projects in senior living communities, the matching rate between user dietary plans and health indicators reached 92%, an improvement of 56 percentage points compared to the original system.

[0082] 2. Improve user health compliance

[0083] The tiered early warning mechanism and closed-loop feedback system reduced the dietary deviation rate to 18% (compared to 42% for traditional systems). Practical application data shows that the rate of nutritional intake compliance among patients with chronic diseases increased by 79%, and the average decision-making time for a single meal was reduced to 45 seconds.

[0084] 3. Expand the smart catering service ecosystem

[0085] It supports seamless integration with medical institution HIS systems and wearable devices, improving health data flow efficiency by 40%. The system is compatible with 5 types of catering scenarios (elderly care communities / hospitals / corporate canteens, etc.) and is particularly suitable for targeted management of 8 types of chronic disease groups, including diabetes. It has been successfully integrated into the central kitchen systems of 32 elderly care institutions.

[0086] 4. Create business value-added space

[0087] Payment system integration improves transaction efficiency by 3 times (800ms / transaction), and brings operators an additional 25% revenue through health data value-added services. The corporate annuity linkage model has partnered with two Fortune 500 companies, reducing employee health management costs by 37%. Attached Figure Description

[0088] Figure 1 This is the overall flowchart of the present invention. Detailed Implementation

[0089] To enable those skilled in the art to more clearly understand the purpose, technical solution, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0090] In the description of this invention, it should be understood that the orientation or positional relationship indicated by terms such as "upper", "lower", "left", "right", "inner", "outer", "lateral", and "vertical" is based on the orientation or positional relationship shown in the accompanying drawings and is only for the convenience of describing this invention, and is not intended to indicate or imply that the device or component referred to must have a specific orientation, and therefore should not be construed as a limitation of this invention.

[0091] The invention will now be described in detail with reference to the accompanying drawings: Figure 1 As shown, a smart canteen health management system based on multi-source data analysis is disclosed. The system includes an identification and sensing system, a data acquisition system, a data processing system, and a nutrition analysis system. The identification and sensing system is installed above the canteen's food counters to identify specific dishes and the surrounding environment. The system collects the identified and sensed dishes, uploads the collected data to the data processing system for processing and analysis, and finally generates a report based on the analysis conclusions.

[0092] The identification and sensing system includes a camera and a temperature and humidity sensor, both installed on the top of the canteen. The camera is installed at a 90° angle above the dining table, 1m ± 0.1m away from the table, to ensure coverage of a 1m diameter food pick-up area.

[0093] A smart canteen health management method based on multi-source data analysis, the method comprising the following steps:

[0094] 1) Equipment installation;

[0095] 2) Customer meal check-in;

[0096] 3) Dish identification;

[0097] 4) Data acquisition;

[0098] 5) Health data analysis and summary;

[0099] 6) Comprehensive data analysis: meals, reports, anomalies;

[0100] 7) Push out analysis summaries and health advice;

[0101] 8) Health data analysis;

[0102] 9) Food and drink check-in.

[0103] The acquisition system is a multimodal data synchronous acquisition system, including...

[0104] 1) Visual data stream: Acquire 1920×1080 resolution images at a frame rate of 30fps;

[0105] 2) Biometric data: Food temperature (±0.5℃) is monitored via an infrared array (8×8 pixels) built into the plate;

[0106] 3) Spatiotemporal alignment mechanism: The NTP protocol is used to achieve time synchronization of multiple devices (error <10ms);

[0107] 4) Payment data flow: Integrate third-party payment SDKs (WeChat v3.0.9 / Alipay v4.18.0 / UnionPay v6.2.1).

[0108] The health data analysis and summary includes:

[0109] 1. Heterogeneous data fusion processing

[0110] 1) Establish a unified data format and model;

[0111] 2) Develop a data cleaning pipeline to remove outliers (such as sudden weight changes > 500g / second) and compensate for environmental interference (initiate infrared temperature measurement calibration when humidity > 85%).

[0112] 3) Privacy protection processing: Biometric data is transmitted with SM4 encryption, and user profile data undergoes differential privacy processing (ε = 0.45);

[0113] 2. Nutrition Knowledge Map

[0114] 1) Construct a 9-dimensional nutrition database containing 1200 kinds of ingredients;

[0115] 2) Food exchange portion parameter table: 6 food exchange matrices (grains and tubers 25g / serving, fruits and vegetables 500g / serving, etc.), and a calorie equivalence comparison table for 120 kinds of foods;

[0116] 3. Personalized Solution Engine

[0117] 1) A food exchange algorithm based on dynamic programming;

[0118] 2) Nutrition plan generation protocol: The fields include 16 nutritional indicators + 8 types of allergen alerts.

[0119] The comprehensive data analysis specifically includes:

[0120] 1. Real-time nutrition calculation engine

[0121] 1) An improved YOLO algorithm model is used to achieve real-time food identification with a recognition latency of <200ms and an accuracy of ≥99.5%;

[0122] 2) Real-time calculation of nutrient intake:

[0123]

[0124] Where w is the current weight of the dish, and e is the baseline nutritional value;

[0125] 2. Tiered response mechanism

[0126] 1) Establish a three-level early warning system:

[0127] Warning level Trigger condition Response mode Level 1 Single sodium intake > 30% of daily recommended value LED screen flashing warning Level 2 Consecutive 3 meals of protein exceeding the standard Mobile terminal push alternative recipes Level 3 Diabetes risk index > 0.85 Smart tableware restricts meal quantity

[0128] 2) Payment linkage rules: The payment function is frozen when the deviation rate is >20%, and the order is automatically released after 30 seconds of payment timeout.

[0129] The health data analysis method is as follows:

[0130] 1. Dynamic updates to user profiles:

[0131] P new =0.85×P old +0.15×S current P represents the health profile, and S represents the current dietary characteristics;

[0132] 2. Long-term trend report generation:

[0133] Generation cycle: Daily / Weekly;

[0134] Indicators include: trend charts of 12 metabolic parameters and radar chart of nutritional balance.

[0135] After the food check-in is completed, the data will be fed back to the dish recognition system for closed-loop processing. The specific execution method is as follows:

[0136] 1. Verify the deviation rate between actual intake and recommendations (threshold <20%);

[0137] 2. Update user compliance rating:

[0138]

[0139] 3. Hardware status maintenance: Camera lens contamination alarm (triggered when light transmittance <85%).

[0140] The smart canteen health management system and method designed in this invention, based on multi-source data analysis, reduces the error rate of traditional nutrition analysis from 15% to 3.2% through multimodal data fusion and improved algorithm models, achieving a dish recognition accuracy rate of ≥99.5%. In pilot projects in elderly care communities, the matching degree between user diet plans and health indicators reached 92%, an improvement of 56 percentage points compared to the original system. Through a hierarchical early warning mechanism and closed-loop feedback system, the dietary deviation rate was reduced to 18% (42% in the traditional system). Actual application data shows that the nutritional intake compliance rate of patients with chronic diseases increased by 79%, and the decision-making time for a single meal was shortened to an average of 45 seconds. This invention supports seamless integration with medical institution HIS systems and wearable devices, improving health data flow efficiency by 40%. The system is compatible with 5 types of catering scenarios (elderly care communities / hospitals / corporate canteens, etc.) and is particularly suitable for targeted management of 8 types of chronic disease groups, including diabetes. It has been successfully integrated into the central kitchen systems of 32 elderly care institutions. Its payment system integration has improved transaction efficiency by 3 times (800ms / order). Through health data value-added services, it brings an additional 25% revenue to operators. The corporate annuity linkage model has partnered with 2 Fortune 500 companies, reducing employee health management costs by 37%, thereby creating commercial value-added space.

[0141] The specific embodiments described herein are merely illustrative of the principles and effects of the invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this invention should still be covered by the claims of this invention.

Claims

1. A smart canteen health management system based on multi-source data analysis, characterized in that: The system includes an identification and sensing system, a data acquisition system, a data processing system, and a nutrition analysis system. The identification and sensing system is installed above the dining table in the cafeteria to identify specific dishes and the surrounding environment. The system collects the identified and sensed dishes, and the collected data is uploaded to the data processing system for processing and analysis. Finally, a report is generated based on the analysis results.

2. The smart canteen health management system based on multi-source data analysis according to claim 1, characterized in that: The identification and sensing system includes a camera and a temperature and humidity sensor, both installed on the top of the canteen. The camera is installed at a 90° angle above the dining table, 1m ± 0.1m away from the table, to ensure coverage of a 1m diameter food pick-up area.

3. A smart canteen health management method based on multi-source data analysis, characterized in that: The method includes the following steps: 1) Equipment installation; 2) Customer meal check-in; 3) Dish identification; 4) Data acquisition; 5) Health data analysis and summary; 6) Comprehensive data analysis: meals, reports, anomalies; 7) Push out analysis summaries and health advice; 8) Health data analysis; 9) Food and drink check-in.

4. The smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: The acquisition system is a multimodal data synchronous acquisition system, including: 1) Visual data stream: Acquire 1920×1080 resolution images at a frame rate of 30fps; 2) Biometric data: Food temperature (±0.5℃) is monitored via an infrared array (8×8 pixels) built into the plate; 3) Spatiotemporal alignment mechanism: The NTP protocol is used to achieve time synchronization of multiple devices (error <10ms); 4) Payment data flow: Integrate third-party payment SDKs (WeChat v3.0.9 / Alipay v4.18.0 / UnionPay v6.2.1).

5. The management method of the smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: The health data analysis and summary includes:

1. Heterogeneous data fusion processing 1) Establish a unified data format and model; 2) Develop a data cleaning pipeline to remove outliers (such as sudden weight changes > 500g / second) and compensate for environmental interference (initiate infrared temperature measurement calibration when humidity > 85%). 3) Privacy protection processing: Biometric data is transmitted with SM4 encryption, and user profile data undergoes differential privacy processing (ε = 0.45); 2. Nutrition Knowledge Map 1) Construct a 9-dimensional nutrition database containing 1200 kinds of ingredients; 2) Food exchange portion parameter table: 6 food exchange matrices (grains and tubers 25g / serving, fruits and vegetables 500g / serving, etc.), and a calorie equivalence comparison table for 120 kinds of foods; 3. Personalized Solution Engine 1) A food exchange algorithm based on dynamic programming; 2) Nutrition plan generation protocol: The fields include 16 nutritional indicators + 8 types of allergen alerts.

6. The management method of the smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: The comprehensive data analysis specifically includes:

1. Real-time nutrition calculation engine 1) An improved YOLO algorithm model is used to achieve real-time food identification with a recognition latency of <200ms and an accuracy of ≥99.5%; 2) Real-time calculation of nutrient intake: Where w is the current weight of the dish, and e is the baseline nutritional value; 2. Tiered response mechanism 1) Establish a three-level early warning system: 2) Payment linkage rules: The payment function is frozen when the deviation rate is >20%, and the order is automatically released after 30 seconds of payment timeout.

7. The management method of the smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: The health data analysis method is as follows:

1. Dynamic updates to user profiles: P new =0.85×P old +0.15×S current P represents the health profile, and S represents the current dietary characteristics; 2. Long-term trend report generation: Generation cycle: Daily / Weekly; Indicators include: trend charts of 12 metabolic parameters and radar chart of nutritional balance.

8. The management method of the smart canteen health management system based on multi-source data analysis according to claim 3, characterized in that: After the food check-in is completed, the records will be fed back to the dish recognition for closed-loop execution. The specific execution method is as follows:

1. Verify the deviation rate between the actual intake and the recommendation (threshold <20%).

2. Update user compliance rating:

3. Hardware status maintenance: Camera lens contamination alarm (triggered when light transmittance <85%).

Images