Meat product fresh-keeping state monitoring and dynamic evaluation system

Abstract

The invention discloses a meat product fresh-keeping state monitoring and dynamic evaluation system, and belongs to the field of food safety and quality monitoring. Comprising an acquisition module integrating multi-source sensing of gas, temperature, humidity, optics and the like, and multi-dimensional data related to meat product preservation is acquired in real time. The system is provided with a data preprocessing and fusion unit which is used for performing synchronous acquisition, de-noising and feature conversion on original signals and outputting high-quality feature data; and the intelligent monitoring and dynamic evaluation unit realizes high-precision discrimination and deterioration risk prediction of the fresh-keeping state based on a fusion algorithm. Monitoring and evaluation results are uploaded to a cloud analysis platform in a wired or wireless mode, archiving, trend analysis and decision optimization of historical and real-time data are achieved, and quality tracing of the whole process is supported. And the central control and display terminal manages each node in a centralized manner to realize intelligent early warning and visual display.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of food safety and quality monitoring, and more particularly relates to a meat product fresh-keeping state monitoring and dynamic evaluation system. BACKGROUND

[0002] With the continuous expansion of the meat product consumption market, the freshness and safety control of meat products in the whole process of processing, storage and transportation, and sales has become the focus of the food industry. At present, meat products are easily affected by microorganisms, temperature changes and environmental factors, and are prone to spoilage, resulting in decreased food quality and increased safety risks. Traditional meat product fresh-keeping state detection methods rely on manual sampling and laboratory physical and chemical analysis, which not only have long detection cycles and delayed responses, but also cannot achieve real-time monitoring of large quantities of products, and cannot meet the needs of modern meat product production and supply chain for refined and intelligent management.

[0003] In addition, some existing monitoring schemes based on a single sensor cannot comprehensively and accurately reflect the freshness changes of meat products due to the limitation of the monitoring index, and have poor adaptability to complex and variable environments, resulting in errors in the evaluation results. In the face of the urgent need for informationization and intelligentization of the meat product supply chain, it is of great practical significance and application prospect to develop a meat product fresh-keeping state monitoring system that integrates multi-source sensing, real-time data fusion and intelligent dynamic evaluation functions. SUMMARY

[0004] The present application aims to solve the technical problems of poor real-time performance, large evaluation errors and insufficient environmental adaptability in the existing meat product fresh-keeping state monitoring process. In view of the problems of delayed response and single information of traditional monitoring methods, a system is proposed that integrates multi-source sensing data acquisition, efficient data preprocessing and fusion, intelligent dynamic evaluation and remote communication functions, realizes real-time, comprehensive and accurate monitoring and early warning of the fresh-keeping state of meat products throughout the whole process, and improves the intelligentization of supply chain management and the level of food safety guarantee.

[0005] To achieve the above purpose, the present application adopts the following technical solutions: the system comprises: A multi-source sensing acquisition module, comprising a gas sensing array, a temperature and humidity sensor and an optical sensor, configured in a monitoring cabin or the inside of a meat product package, for acquiring multi-source sensing data related to the fresh-keeping state of meat products; A data preprocessing and fusion unit connected to the multi-source sensing acquisition module through wired or wireless means, for synchronously acquiring, denoising and feature converting the original signals from the multi-source sensing acquisition module, and outputting preprocessed multi-dimensional feature data; The intelligent monitoring and dynamic evaluation unit is connected with the data preprocessing and fusion unit through a high-speed bus, integrates feature extraction, variable point detection and fresh-keeping index prediction algorithms, is used for judging and evaluating the fresh-keeping state of the meat product, and generates evaluation results and early warning information. The remote data communication module is connected with the intelligent monitoring and dynamic evaluation unit, and uploads the monitoring and evaluation results to a cloud storage analysis platform; The cloud storage analysis platform is used for storing, archiving and deeply analyzing historical and real-time monitoring data, and feeding back big data analysis results or decision suggestions to the central control and display terminal; The central control and display terminal is connected with the intelligent monitoring and dynamic evaluation unit and the cloud storage analysis platform through a data interface, is used for displaying the fresh-keeping state of the meat product and early warning information in real time, and centrally managing the system running state, and remotely configuring parameters of the multi-source sensing acquisition module, the data preprocessing and fusion unit and the intelligent monitoring and dynamic evaluation unit.

[0006] In one scheme, the multi-source sensing acquisition module comprises a high-sensitivity gas sensing array, a digital temperature and humidity sensor and a visible light and near-infrared optical sensor, which are respectively integrated in suitable positions of an inner wall, a top cover or an observation window of a monitoring cabin or meat product packaging; The gas sensing array is used for detecting key gas components such as ammonia, hydrogen sulfide and carbon dioxide, the temperature and humidity sensor is used for monitoring the temperature and humidity of the storage environment, and the optical sensor is used for monitoring the color and related optical characteristics of the surface of the meat product; Each sensing unit is connected with the data preprocessing and fusion unit through a micro interface board or a flexible circuit in a wired (I2C, SPI) or wireless (BLE, ZigBee, LoRa) mode.

[0007] In one scheme, the data preprocessing and fusion unit is composed of an embedded processor and an analog / digital signal conditioning circuit thereof, and a wavelet transform noise reduction algorithm is embedded in the data preprocessing and fusion unit, so as to enhance the data stability; The collected data is normalized, standardized and feature extracted, and a multi-dimensional feature vector is output.

[0008] In one scheme, the intelligent monitoring and dynamic evaluation unit comprises a feature weight self-learning, a nonlinear variable point detection and a self-correcting fresh-keeping index prediction algorithm; Based on the input of the multi-source sensor, time sequence characteristics of the meat product under different processes and storage conditions are collected, a dynamically adjustable fusion feature expression model is constructed through an adaptive feature weight learning mechanism; The weight matrix is optimized through a target function, and a regularization constraint is combined to realize optimal representation of multi-dimensional sensing data, and the weight matrix is updated in real time with a sliding time window to adapt to batch, environment and storage and transportation conditions; For the state mutation recognition in the fresh-decay process of meat products, a nonlinear mutation point detection algorithm based on multi-source fusion is used, a reconstruction error measure is used, an autoregressive or deep prediction model is combined, and a change point probability of each time step is calculated; when the probability exceeds a preset threshold, the mutation time of quality is automatically determined, and the forward-looking perception ability of weak state changes is improved. The freshness index prediction model uses a multi-factor spatiotemporal regression network driven by historical big data, combines real-time collected fusion features, environmental parameters, batch and process information multi-input, and uses a recursive neural network with external factor correction structure to dynamically correct the prediction deviation caused by batch, season and climate change.

[0009] In one scheme, the remote data communication module uploads the monitoring and evaluation results and multi-dimensional time series data to a cloud storage analysis platform through 4G / 5G cellular, Ethernet, WiFi or LoRa communication mode in an encrypted manner.

[0010] In one scheme, the central control and display terminal is interconnected with each collection node, processor and cloud platform through a wired or wireless high-speed interface, realizes centralized management and visualization of system running state; The terminal supports remote configuration, dynamic upgrade and batch inspection of parameters of each collection node and processor, and is provided with a strategy rule engine and an intelligent alarm component, which automatically identifies abnormal trends, communication faults and hardware abnormalities, and implements multi-form early warning; All operation records and data flow are automatically archived, the terminal has a protective shell and a high-brightness display screen, is suitable for long-term deployment in cold chain and complex warehouse environment, and realizes full-process intelligent information interaction and management of meat product freshness monitoring.

[0011] In one scheme, the data preprocessing and fusion unit is arranged in a monitoring cabin external control box or a large integrated collection module, and is connected with the intelligent monitoring and dynamic evaluation unit through a high-speed bus (RS485, CAN, SPI or Ethernet), has an expansion interface, and the overall design considers anti-interference, low power consumption and compact structure.

[0012] In one scheme, the cloud storage analysis platform adopts a distributed database and big data processing architecture, archives, labels, dynamically analyzes features and models trends based on historical and real-time data, and outputs intelligent decisions and early warning suggestions based on AI decision and risk optimization algorithms. The cloud analysis result is fed back to the central control and display terminal through a secure data channel, supports visual management of historical records, quality trends and risk information by the user, forms a traceable data analysis and business closed loop.

[0013] Advantages of the present application: Through the collaborative deployment of high-sensitivity gas sensing array, digital temperature and humidity sensor, and visible light and near-infrared optical sensor, the system can comprehensively, multi-dimensionally and in real time collect and reflect the environmental parameters and state changes of meat products in the storage and flow process. After pre-processing and fusion unit of multi-source data, efficient synchronization, noise reduction and feature conversion, the data input into the evaluation model is highly stable and reliable.

[0014] Relying on the embedded intelligent algorithms such as multi-feature extraction, change point detection and preservation index prediction, the system can accurately distinguish and predict the preservation and deterioration dynamic process, sensitively capture quality mutations, and realize early automatic warning of risks. At the same time, the monitoring and evaluation results can be safely uploaded to the cloud platform through various wired or wireless communication methods. The cloud end uses distributed database and big data processing architecture to support deep archiving, dynamic feature analysis and trend modeling of historical and real-time data, and automatically outputs intelligent decision and risk control suggestions, forming a complete closed-loop quality traceability and digital archive management.

[0015] The central control and display terminal not only realizes remote configuration, upgrade and inspection of each node, but also supports centralized intelligent operation and maintenance and multi-form intelligent alarm of the system, provides intuitive quality trend and abnormal information display for users, and effectively improves the operation convenience and management intelligent level. The overall design takes into account the characteristics of modularity, scalability and anti-interference, and is suitable for long-term application in various complex environments such as meat product processing, cold chain storage and circulation.

[0016] Compared with the traditional manual detection or single sensing method, the present application greatly improves the accuracy, timeliness and automation level of meat preservation quality monitoring, helps to reduce deterioration loss, strengthens food safety control, and promotes the upgrading of the industry to high standards, intelligence and digitization. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The system block diagram of the present application; Figure 2 The flow chart of the intelligent monitoring and dynamic evaluation unit of the present application. DETAILED DESCRIPTION

[0018] In order to facilitate the understanding of the present application, the present application will be described more fully below with reference to the related drawings. The drawings show typical embodiments of the present application. However, the present application can be realized in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present application more thorough and comprehensive.

[0019] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0020] like Figure 1 As shown, the system of the present invention mainly consists of a multi-source sensor acquisition module, a data preprocessing and fusion unit, an intelligent monitoring and dynamic evaluation unit, a remote data communication module, and a central control and display terminal.

[0021] (1) The multi-source sensor acquisition module includes a gas sensor array, a temperature and humidity sensor and an optical sensor, which are integrated into the monitoring chamber or inside the meat product packaging and are connected to the data preprocessing and fusion unit via wired or wireless means.

[0022] Gas sensor arrays typically employ high-sensitivity metal-oxide-semiconductor sensors, electrochemical sensors, or infrared gas sensing components to detect key gaseous components such as ammonia, hydrogen sulfide, and carbon dioxide released during meat spoilage. These gas sensors are closely arranged on the inner wall or top of the monitoring chamber or meat product packaging, enabling rapid response to minute changes in gas concentration while maintaining close proximity to the surface of the meat sample, thus improving the representativeness and sensitivity of the data collection. Temperature and humidity sensors utilize high-precision digital modules to monitor the temperature and relative humidity of the meat storage environment in real time. Their sensing elements are generally fixed to the inner side wall of the chamber or packaging, 2-5 cm away from the meat product, to avoid direct contact contamination while ensuring the accuracy and speed of the measurement data.

[0023] The optical sensor section integrates visible light and near-infrared spectroscopy modules to characterize the optical features of surface color changes and harmful microbial metabolites in meat products. These sensors are typically mounted planarly on the top transparent area of ​​the packaging or at the observation window of the monitoring chamber, achieving non-destructive monitoring through timed scanning or reflection detection. All the aforementioned sensor units are connected together via a miniature interface board or flexible circuitry. Signals are synchronously transmitted to the data preprocessing and fusion unit via wired (I2C, SPI) or wireless (BLE, ZigBee, LoRa) methods, enabling real-time and efficient acquisition of multi-source information. The entire module is designed to balance sealing with the openness of the sensing window, preventing external interference while facilitating sensor maintenance and replacement, making it suitable for integrated deployment inside and outside packaging for different types of meat products.

[0024] (2) The data preprocessing and fusion unit synchronously collects, denoises and converts the original signals from various sensors, and the processed multi-dimensional data is connected to the intelligent monitoring and dynamic evaluation unit through a high-speed bus.

[0025] The unit is composed of an embedded processor and its supporting analog / digital signal processing circuit. All gas, temperature and humidity, optical sensors are synchronously accessed through multi-channel ADC or multi-channel digital interface to realize millisecond-level data synchronous acquisition. In view of the noise and interference that may exist in the sensor signals, the unit integrates a wavelet transform denoising algorithm to ensure the stability and reliability of the output signals. In order to support subsequent intelligent monitoring processing, the processor further converts the channel data into features, including normalization, standardization, feature extraction (such as gas composition peak, temperature and humidity change slope, spectral feature parameters), and converts high-dimensional original data into representative multi-dimensional feature vectors.

[0026] In terms of structural layout, the data preprocessing and fusion unit is generally located in the external control box of the monitoring cabin or the internal large integrated acquisition module of the meat product package, which facilitates physical connection with the sensor port and enables fast and stable transmission of processed data to the intelligent monitoring and dynamic evaluation unit through a high-speed bus (RS485, CAN, SPI, Ethernet). The unit also has an expansion interface to meet the needs of sensor type upgrade or quantity expansion in the future, and the overall design focuses on anti-interference, low power consumption and compact structure to adapt to the diversified installation and deployment in the meat product circulation packaging and cold chain transportation environment.

[0027] (3) The intelligent monitoring and dynamic evaluation unit integrates feature extraction, change point detection and preservation index prediction algorithms to realize dynamic discrimination and evaluation of the preservation state of meat products. The output results are connected to the central control and display terminal through a data interface, making it convenient for users to view the preservation state and warning information in real time.

[0028] As shown in Figure 2 , specifically, the system first automatically learns the feature expression weights of each gas and environmental parameter based on different processing techniques and storage conditions of meat products, dynamically adjusts the influence of different sensing signals in state discrimination, and forms a self-adaptive feature expression over time.

[0029] The intelligent monitoring and dynamic evaluation unit deeply integrates multi-sensor data intelligent processing and advanced algorithm architecture. First, in view of the differences in data distribution and evolution process of different processing techniques and storage conditions of meat products, the system uses an adaptive feature weight learning mechanism to build a multi-dimensional feature expression model. In the specific implementation process, the original feature matrix is assumed to be , where N is the number of samples, D is the number of feature channels (multi-class gas concentration, temperature and humidity, optical index), and the system performs weighted linear transformation Get fusion features where the weight matrix is dynamically adjusted by a self-learning mechanism to ensure that the contribution of key sensing signals under various processes and storage conditions is optimal. The weight matrix optimization uses the objective function where is the state label, is a regularization coefficient used to suppress irrelevant feature dimensions, and the model is iterated in real time with a time sliding window to form a time-adaptive feature expression.

[0030] Second, a nonlinear state change point detection algorithm based on multi-sensor fusion is introduced, which can identify the small but critical gas composition inflection point in the process of meat products from fresh to deterioration, and realize the forward-looking analysis of state mutation.

[0031] The system uses a nonlinear mutation point detection algorithm based on multi-sensor fusion. Specifically, for the feature sequence ( is the time step), a mutation measure based on reconstruction error is introduced, and the change point probability at a certain time is defined as where is the expected feature output by the autoregressive or deep prediction model using the data of the previous time, is a scaling parameter, is a sigmoid function. If exceeds the threshold value, it is determined that this time is the state inflection point of fresh-deterioration, and the forward warning of meat quality sudden change is realized. This algorithm couples with the asynchronous changes of gas main components, environmental temperature and humidity, and optical characteristics, improving the capture sensitivity of weak quality mutation signals.

[0032] In addition, the system develops a self-correcting freshness index prediction model, which uses a multi-factor time trace network driven by historical big data to dynamically correct the prediction bias caused by external factors such as batch differences and climate change, making the freshness period evaluation more personalized and scenario-adaptive.

[0033] The system designs a self-correcting multi-factor spatio-temporal regression network, which fully utilizes the historical big database and combines real-time input for dynamic modeling. Let the freshness index be its prediction model is where represents the multi-source fusion feature sequence at the past time, is the current external environment vector (such as climate, transportation temperature), is the batch identifier and process parameter, is the model parameter. The model uses LSTM with external factor correction structure, and the loss function is wherein to correct the weights, For suppressing batch-to-batch prediction bias. The system realizes continuous self-correction of the model with batches, seasons and transportation routes through online fine-tuning and sliding window historical error backtracking mechanism, making the preservation prediction results more personalized and scene-adaptive.

[0034] (4) The intelligent monitoring and dynamic evaluation unit also uploads the monitoring and evaluation results to the cloud storage analysis platform through the remote data communication module. The cloud storage analysis platform not only archives and analyzes historical and real-time data for a long time, but also feeds back big data analysis results or intelligent decision suggestions to the central control and display terminal, forming a feedback loop.

[0035] After completing local feature recognition, change point detection and preservation index prediction, the intelligent monitoring and dynamic evaluation unit uploads key monitoring and evaluation results and multi-dimensional time series data before or after preprocessing to the cloud storage analysis platform in real time or at regular intervals through the integrated remote data communication module.

[0036] The data communication module can select 4G / 5G cellular communication, Ethernet, WiFi or LoRa according to the actual scene, and is equipped with a secure encryption protocol (such as SSL / TLS) to ensure the confidentiality and integrity of the data during transmission. The cloud storage analysis platform is built on a distributed database and big data processing architecture. The platform automatically labels and archives uploaded data according to batches, processes and circulation links, and combines multi-dimensional dynamic feature modeling, time series clustering and quality evolution trend analysis to realize long-term data retention and multi-angle traceability review.

[0037] During data analysis, the cloud platform can also use aggregation statistics, anomaly detection, group mechanism modeling and AI decision algorithms to globally optimize the preservation risk of different batches, seasons or transportation routes, and generate personalized intelligent decision suggestions or dynamic early warning strategies based on model output. All cloud-based analysis results and decision suggestions are fed back to the central control and display terminal through a secure data channel to realize user-side visualization and real-time tracking of historical records, quality trends, risk distribution and intelligent decisions, and to build a data analysis and business loop. To ensure the maintainability and environmental adaptability of the system, the remote data communication module and antenna are usually integrated in the external communication interface area of the terminal main control box or the monitoring unit shell of the cold chain logistics node, with protection level and anti-interference design, while the cloud platform is deployed in a remote data center or enterprise's own private cloud, ensuring high availability and large-scale concurrent processing capability of data.

[0038] (5) The central control and display terminal can centrally manage the overall operation state of the system, and can remotely configure the parameters of each acquisition node or processor when necessary, so as to realize intelligent monitoring, dynamic evaluation and full-process information interaction of the meat preservation state.

[0039] The central control and display terminal as the core management and interaction hub of the system usually adopts an industrial integrated industrial control host or a touch panel, and is built-in with a high-reliability operating system and a customized management software to realize centralized management and control and information visualization of the whole intelligent monitoring and dynamic evaluation system. The terminal realizes stable interconnection through wired or wireless high-speed data interfaces with various data acquisition nodes, intelligent processors and cloud platforms, and collects real-time monitoring data, preservation indexes, state change point information and cloud big data analysis results of the whole process. At the management software level, the terminal integrates multi-thread communication and distributed task management mechanism, supports remote configuration and dynamic upgrade of the parameters of each acquisition node and processor in the system. Specifically, the administrator can flexibly adjust the key parameters such as acquisition frequency, detection sensitivity and data reporting strategy of each sensing channel through the graphical user interface, or one-key start deep inspection and temporary early warning measures for special batches and abnormal states. The terminal also has a strategy rule engine and an intelligent alarm component, which can automatically identify abnormal trends, communication failures or hardware abnormalities in system operation, and give instant warning through sound and light or push notification. At the same time, all operation records and data flow are automatically archived for subsequent traceability and system maintenance. In order to ensure the durability and operation convenience of the terminal device, it is usually installed in the central refrigeration warehouse dispatch control room, logistics sorting center or cold chain transportation vehicle cab, etc. core management area, equipped with dustproof and waterproof shell and high-brightness display screen, to ensure long-term stable operation in complex environment, and fully meet the information interaction and intelligent management needs of each link of the whole process of meat preservation.

[0040] The whole system realizes highly automated data acquisition, remote cloud analysis and visualization warning, provides an intelligent and dynamic full-process solution for meat supply chain preservation management, and effectively solves the problems of meat preservation state monitoring lag, large evaluation error and poor adaptability in the prior art.

[0041] Embodiment: In order to verify the effectiveness and reliability of the meat preservation state monitoring and dynamic evaluation system of the application, the actual deployment and operation of the system in the specific embodiment is given by taking the refrigerated storage of segmented pork of a cold chain storage enterprise as an application scene.

[0042] I. Experimental environment and system deployment A batch of split pork with batch number A20240601 in the enterprise cold storage is selected, with a net weight of 500 kg, a single weight of 1.5-2.0 kg, vacuum packaging, and a storage temperature of 2°C and a relative humidity of 85%. The multi-source sensing acquisition module of the application is arranged in the cold storage and the inner wall of the meat product package, a total of 10 monitoring nodes, each node integrating a gas sensing array (ammonia, hydrogen sulfide, carbon dioxide), a digital temperature and humidity sensor, and a visible-near infrared optical sensor. All nodes are interconnected with the external data preprocessing and fusion unit through the LoRa wireless network, and are uniformly connected to the central control and display terminal, and the system parameters are remotely configured. The acquisition cycle is set to once every 30 minutes, and the monitoring cycle covers 15 days.

[0043] II. Key monitoring data and process During the entire monitoring period, key data such as gas, temperature and humidity, and optical reflectance spectrum were collected and analyzed. Some of the raw data are shown in Table 1 (monitoring node 3 is taken as an example, and key moment data are selected): III. Data processing and dynamic evaluation process Each monitoring node synchronously collects the above-mentioned multiple groups of data every 30 minutes. The data are first denoised and normalized by the preprocessing and fusion unit, and the time series features (such as gas concentration gradient change, spectral feature downward trend, and environmental parameter fluctuation) are extracted based on the feature conversion model. The intelligent monitoring and dynamic evaluation unit uses a multi-source fusion recurrent neural network to dynamically analyze the feature data, generates a current freshness index prediction value (range 0-1, 1 represents optimal freshness, below 0.7 is considered as a mild spoilage warning, and below 0.5 is considered as serious spoilage), and outputs a freshness state evaluation conclusion.

[0044] As can be seen from Table 1, in the early stage of the experiment (June 1-June 6), the ammonia and hydrogen sulfide concentrations are extremely low, the carbon dioxide increase is slow, the spectral peak remains at a high level, the freshness index prediction value is higher than 0.9, and the meat product state is “fresh”. Starting from June 9, the gas sensing results show an accelerated increase, especially the ammonia and hydrogen sulfide increase significantly, the spectral peak rapidly decreases, the freshness index prediction value drops to 0.87, and the system enters the “caution” stage; by June 12, the key gas concentrations break the warning line (such as ammonia > 3.5 ppm), combined with the decrease in the spectral peak value, the dynamic evaluation algorithm comprehensively determines that it is “mildly spoiled”, and automatically pushes the critical warning to the central control terminal and the warehouse management personnel’s mobile phone. By June 15, the ammonia and hydrogen sulfide concentrations reach 7.8 ppm and 0.93 ppm respectively, the spectral peak value drops to 0.37, the freshness index drops to 0.43, and the system determines that it is “severely spoiled”, and forcibly triggers an alarm and recommends that the batch be removed.

[0045] IV. Cloud analysis and abnormal management closed loop All collected data and dynamic evaluation results are uploaded to the cloud through the LoRa+4G interface. The cloud analysis platform archives, labels, and manages the full data, clusters time series features, and performs trend prediction analysis, identifies the abnormal inflection point of gas concentration rise (June 11) within the storage period of this batch, and automatically compares with the historical data of nearly 500 batches of the enterprise, outputs the risk trend report and management suggestions. The platform also automatically inspects the operation of each node device, detects the abnormal signal fluctuation of node 1 on June 14, and automatically dispatches maintenance work orders to ensure device reliability.

[0046] V. Centralized management and display interaction The central control and display terminal refreshes the values and evaluation conclusions of each node in real time all day long. Management personnel can view the batch quality trend, early warning records, and device operation status at any time through the visual large screen and mobile terminal APP. The system supports hierarchical permission management and automatic archiving of the entire operation process, facilitating post-tracing and quality responsibility identification. According to the system's suggestions, management personnel began to check the actual sensory and microbial indicators of meat products on June 12. The detection found that the total number of microorganisms began to exceed the standard, which was highly consistent with the system's dynamic early warning, further verifying the accuracy and forward-looking nature of the monitoring and evaluation model of the invention.

[0047] This embodiment fully verifies that the system of the invention can realize high-frequency and stable collection and synchronous fusion of multi-source data throughout the whole process of cold-chain meat product storage, automatically realize dynamic discrimination, risk trend prediction, and multi-level intelligent early warning of the whole process from freshness to deterioration, greatly improving the identification and response efficiency of safety risks. Through cloud big data analysis, the batch management traceability loop is realized, and the system exhibits significant advantages of intelligence, automation, and high adaptability in practical application, providing technical support and innovative path for fine management of cold-chain meat products and food safety guarantee.

[0048] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiment methods. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0049] It should be understood that the technical solutions of the present application described above with the preferred embodiments are illustrative rather than limiting. Based on the description of the present application, those skilled in the art can modify the technical solutions recorded in each embodiment, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A system for monitoring and dynamically evaluating the preservation status of meat products, characterized in that: The system includes: The multi-source sensor acquisition module, including a gas sensor array, a temperature and humidity sensor, and an optical sensor, is configured inside the monitoring chamber or meat product packaging to collect multi-source sensor data related to the freshness status of meat products. The data preprocessing and fusion unit is connected to the multi-source sensor acquisition module via wired or wireless means. It synchronously acquires, denoises, and transforms the raw signals from the multi-source sensor acquisition module, and outputs preprocessed multi-dimensional feature data. The intelligent monitoring and dynamic evaluation unit is connected to the data preprocessing and fusion unit via a high-speed bus. It integrates feature extraction, change point detection, and freshness index prediction algorithms to identify and evaluate the freshness status of meat products and generate evaluation results and early warning information. The remote data communication module connects to the intelligent monitoring and dynamic evaluation unit and uploads the monitoring and evaluation results to the cloud storage and analysis platform. The cloud storage and analysis platform is used to store, archive, and perform in-depth analysis of historical and real-time monitoring data, and to feed back big data analysis results or decision-making suggestions to the central control and display terminal. The central control and display terminal is connected to the intelligent monitoring and dynamic evaluation unit and the cloud storage and analysis platform through a data interface. It is used to display the freshness status of meat products and early warning information in real time, and to centrally manage the system operation status. It can also remotely configure the parameters of the multi-source sensor acquisition module, the data preprocessing and fusion unit, and the intelligent monitoring and dynamic evaluation unit.

2. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The multi-source sensing acquisition module includes: a high-sensitivity gas sensing array, a digital temperature and humidity sensor, and visible light and near-infrared optical sensors, which are respectively integrated into the inner wall, top cover, or observation window of the monitoring chamber or meat product packaging. Gas sensor arrays are used to detect key gas components such as ammonia, hydrogen sulfide and carbon dioxide; temperature and humidity sensors monitor the temperature and humidity of the storage environment; and optical sensors monitor the surface color and related optical characteristics of meat products. Each sensing unit connects to the data preprocessing and fusion unit via a wired (I2C, SPI) or wireless (BLE, ZigBee, LoRa) connection through a miniature interface board or flexible circuit.

3. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The data preprocessing and fusion unit consists of an embedded processor and its analog / digital signal conditioning circuit, and incorporates a wavelet transform noise reduction algorithm to enhance data stability. The collected data is normalized, standardized, and its features are extracted to output a multidimensional feature vector.

4. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The intelligent monitoring and dynamic evaluation unit includes a feature weight self-learning algorithm, a nonlinear change point detection algorithm, and a self-correcting freshness index prediction algorithm. Based on multi-source sensor input, the temporal characteristics of meat products under different processing and storage conditions are collected. Through an adaptive feature weight learning mechanism, a dynamically adjustable fusion feature expression model is constructed. The weight matrix is ​​optimized through the objective function and combined with regularization constraints to achieve the optimal representation of multidimensional sensor data, and is updated in real time with the sliding time window to adapt to changes in batch, environment and storage and transportation conditions. For identifying state mutations in meat products during the fresh-to-spoilage process, a nonlinear mutation point detection algorithm based on multi-source fusion is used. By utilizing reconstruction error measurement and combining autoregressive or deep prediction models, the probability of a change point at each time step is calculated. When the probability exceeds a preset threshold, it is automatically determined as a moment of quality mutation, thereby improving the ability to anticipate subtle state changes. The freshness index prediction model uses a multi-factor spatiotemporal regression network driven by historical big data, combined with multi-factor inputs such as real-time collected fusion features, environmental parameters, batch and process information, and adopts a recurrent neural network with an external factor correction structure to dynamically correct prediction biases caused by batch, season and climate change.

5. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The remote data communication module encrypts and uploads the monitoring and evaluation results and multi-dimensional time-series data to the cloud storage and analysis platform via 4G / 5G cellular, Ethernet, WiFi or LoRa communication methods.

6. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The central control and display terminal is interconnected with each acquisition node, processor and cloud platform through wired or wireless high-speed interfaces to realize centralized management and visualization of the system's operating status; The terminal supports remote configuration, dynamic upgrade and batch inspection of parameters for each acquisition node and processor, and has a built-in policy rule engine and intelligent alarm component to automatically identify abnormal trends, communication failures and hardware anomalies and implement multi-form early warning. All operation records and data flow are automatically archived. The terminal has a protective shell and a high-brightness display screen, making it suitable for long-term deployment in complex cold chain and warehousing environments, and realizing intelligent information interaction and management of the entire process of meat product preservation monitoring.

7. The meat product preservation status monitoring and dynamic evaluation system according to claim 1, characterized in that: The data preprocessing and fusion unit is located in the external control box of the monitoring cabin or in a large integrated acquisition module, and is connected to the intelligent monitoring and dynamic evaluation unit via a high-speed bus (RS485, CAN, SPI or Ethernet). It has an expansion interface, and the overall design takes into account anti-interference, low power consumption and compact structure.

8. The meat product preservation status monitoring and dynamic evaluation system according to claim 5, characterized in that: The cloud storage and analysis platform adopts a distributed database and big data processing architecture to archive, tag, perform dynamic feature analysis and trend modeling on historical and real-time data, and output intelligent decision-making and early warning suggestions based on AI decision-making and risk optimization algorithms. The cloud-based analysis results are fed back to the central control and display terminal through a secure data channel, supporting users to visualize and manage historical records, quality trends, and risk information, forming a traceable data analysis and business closed loop.

Images