Goods near expiration date early warning method based on supply chain management system

Abstract

The application discloses a kind of goods near expiration date early warning methods based on supply chain management system, including, the quality evaluation of environmental data collected by sensor, processing and fusing goods information, obtain comprehensive feature vector;Based on the feature vector setting dynamic weight, and construct dynamic shelf life prediction model, the jump compensation mechanism introduced in model, when detecting sudden event, automatically adjust early warning range, optimize early warning precision;Through random extraction goods are detected, and the model prediction result is compared, and early warning range is dynamically updated, and early warning information is transmitted to business end using visual means, to realize accurate goods near expiration date early warning;The application not only significantly improves the accuracy and real-time of early warning, reduces the false positive rate, but also enhances the ability of supply chain management system to deal with sudden events, can effectively reduce the goods loss caused by environmental factors or sudden events, improve the efficiency and response speed of supply chain management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of goods expiration date warning, and particularly relates to a goods expiration date warning method based on a supply chain management system. BACKGROUND

[0002] With the rapid development of globalization and information technology, supply chain management systems have become a core component of modern logistics and production management. Especially in industries such as food, medicine and chemicals that have strict quality requirements, it is particularly important to have an accurate goods expiration date warning system. Current traditional supply chain management systems often rely on static goods life cycle prediction, usually based on historical data and experience rules for manual prediction. However, this method has significant limitations, especially when faced with complex and changing environmental conditions, it cannot provide accurate and real-time warnings. In recent years, with the development of sensor technology, Internet of Things technology and big data analysis methods, dynamic monitoring and analysis of environmental data of goods during transportation and storage has become an important means to improve the accuracy of goods shelf life prediction.

[0003] Currently, there are some shelf life prediction methods based on environmental monitoring data, mainly including the influence of temperature, humidity, gas concentration and other environmental factors on the quality of goods. However, most of the existing technologies focus on static monitoring and prediction models, lack of dynamic response capability to environmental variable changes, resulting in poor accuracy and timeliness of the prediction results. In addition, although sensors can collect temperature and humidity data in real time, due to the complexity of the interaction between environmental factors, most systems cannot fully capture the dynamic relationship between these variables and the quality of goods even if they meet the monitoring conditions. For example, small fluctuations in temperature or changes in humidity can have a significant impact on the quality of goods in a short period of time, but existing systems often have difficulty adjusting effectively in the event of an emergency or non-linear condition, thereby reducing the real-time and accuracy of the warning system. In addition, considering that goods may encounter unexpected events such as vibration during transportation, packaging damage or equipment failure, which can also have a sudden impact on the shelf life of goods, and the existing technology usually ignores these interference factors and lacks a recovery compensation mechanism, resulting in slow processing efficiency and large error range after warning. SUMMARY

[0004] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0005] In view of the above existing problems, the present application is proposed. Therefore, the present application provides a goods near-expiration early warning method based on a supply chain management system to solve the problems proposed in the background art.

[0006] To solve the above technical problems, the present application provides the following technical solutions: a goods near-expiration early warning method based on a supply chain management system, comprising:

[0007] Different types of sensors are arranged at the production, storage and transportation links of the supply chain to obtain environmental data in each sensor, the environmental data is quality evaluated, and goods information is obtained from the supply chain management system, the environmental data that passes the quality evaluation is merged with the goods information to obtain a comprehensive feature vector of the goods;

[0008] Considering the influence of environmental data on the shelf life of the goods, a dynamic weight is set for the comprehensive feature vector of the goods, a dynamic shelf life prediction model is constructed based on the comprehensive feature vector of the goods and the dynamic weight, a jump compensation mechanism triggered by an event is introduced into the dynamic shelf life prediction model, and an early warning range of the goods near-expiration is set according to the jump compensation mechanism triggered by the event;

[0009] Randomly selected goods are subjected to shelf life detection, the prediction result of the dynamic shelf life prediction model is compared with the detection result of the shelf life of the randomly selected goods, a model prediction error is obtained, the early warning range of the goods near-expiration is updated according to the model prediction error, the update result of each time is displayed in a visual manner, and is transmitted to the supply chain management system to realize accurate early warning of the goods near-expiration in the supply chain management system.

[0010] As a preferred scheme of the goods near-expiration early warning method based on the supply chain management system, wherein: the environmental data in each sensor is obtained, and the environmental data is quality evaluated, comprising:

[0011] The quality evaluation is expressed by a formula as follows:

[0012]

[0013] Wherein, v i is the current measurement value v of the i-th sensor; is the average value of the measurement values of the remaining sensors in the same region; δ i is the threshold value δ of the i-th sensor;

[0014] If the threshold value delta does not exceed the threshold value of the remaining sensors in the same region, it indicates that the environmental data collected by the current sensor is normal, otherwise it indicates that the environmental data collected by the current sensor is abnormal, the environmental data collected by the remaining sensors is synchronized through the time stamp, if there is a missing situation of the environmental data collected by the current sensor in the data synchronization process, then the data of the remaining sensors in the same region is interpolated and predicted.

[0015] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, the goods information is obtained from the supply chain management system at the same time, the environmental data evaluated by quality is combined with the goods information to obtain a comprehensive feature vector of the goods, including:

[0016] The environmental data collected by the sensor in the normal state is represented as the environmental data evaluated by quality, the goods information is obtained from the supply chain management system at the same time, and the environmental data evaluated by quality is combined with the goods information to obtain a comprehensive feature vector of the goods.

[0017] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, the influence of the environmental data on the shelf life of the goods is considered, and a dynamic weight is set for the comprehensive feature vector of the goods, including:

[0018] Considering that each environmental data in the environmental data has an influence on the shelf life of the goods and the influence degree is different, a weight value updated with time is allocated to each environmental data to obtain an allocated weight set.

[0019] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, the comprehensive feature vector of the goods and the dynamic weight are used to construct a dynamic shelf life prediction model, including:

[0020] The quality state of the goods at time t is obtained and initialized, and a discrete time step is set based on hours;

[0021] Based on the obtained allocated weight set and the comprehensive feature vector of the goods, the quality decay process of the goods is obtained through a difference equation;

[0022] The difference equation Q(t+Δτ) is represented as:

[0023] Q(t+Δτ)=Q(t)-Δτ×F(X(t),ω(t))

[0024] Wherein, Q(t) represents the quality state of the goods at time t, initialized as Q(0)=Q0; Δτ represents a discrete time step, in hours; F(X(t), ω(t)) represents a decay function; X(t) is a comprehensive feature vector of the goods, and ω(t) is a set of assigned weights;

[0025] The decay function F(X(t), ω(t)) is represented as:

[0026]

[0027] Wherein, f i represents the i-th factor function, and N represents a specific environment.

[0028] As a preferred scheme of the goods near-expiration early warning method based on the supply chain management system, the method further comprises:

[0029] According to the obtained quality state Q(t) at time t, a lower limit Q low of the quality safety of the goods is set.

[0030] When Q(t)≤Q low , it indicates that the goods have entered a near-expiration state, and transportation is not suitable, and the value of Q(t) is updated, and the difference equation Q(t+Δt) is recalculated.

[0031] The value of the difference equation updated each time is recorded, and the decay state of the quality of the goods corresponding to the difference equation is marked.

[0032] As a preferred scheme of the goods near-expiration early warning method based on the supply chain management system, a jump compensation mechanism triggered by an event is introduced into the dynamic shelf life prediction model, comprising:

[0033] If a sudden event is detected, jump correction is performed:

[0034] Q new (t)=Q(t)-ΔQ shock

[0035] Wherein, Q new (t) represents the value of Q(t) after the jump, and ΔQ shock represents the severity of the sudden event.

[0036] If a remedial measure is taken after the sudden event, a sudden event compensation term is defined:

[0037] Q new (t)=Q new (t)+ΔQ recover

[0038] Wherein, ΔQ recoverAn effectiveness evaluation of the remedial measures is represented;

[0039] If the emergency is repaired, the detection Q new (t) is updated to the value of the current Q new (t).

[0040] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, the jump compensation mechanism triggered by the event is used to set the early warning range of the goods expiration date, including:

[0041] According to the value of the updated Q(t), the first threshold value, the second threshold value and the third threshold value are set, each threshold value corresponds to a different early warning level, and the relationship between the threshold values is represented as: β1<β2<β3.

[0042] When Q(t) approaches the first threshold value β1, the prediction model issues a primary early warning to prompt the relevant personnel to pay attention to the current environment of the goods;

[0043] When Q(t) approaches the second threshold value β2, the prediction model issues a middle-level early warning to prompt the relevant personnel to quickly promote the current goods;

[0044] When Q(t) approaches the third threshold value β3, the prediction model issues a high-level early warning to prompt the relevant personnel to remove the goods.

[0045] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, random goods are extracted for quality state detection, the prediction result of the dynamic shelf life prediction model is compared with the detection result of the quality state of the random goods, and the model prediction error is obtained, including:

[0046] The model prediction error is achieved by minimizing the objective function, as follows:

[0047]

[0048] Wherein, Q rs (t) is the quality state of the random goods at time t, and Θ represents the internal model parameters of each factor function f i .

[0049] As a preferred scheme of the goods expiration date early warning method based on the supply chain management system, the early warning range of the goods expiration date is updated according to the model prediction error, the update result of each time is displayed in a visual way and transmitted to the supply chain management system, and the accurate early warning of the goods expiration date in the supply chain management system is realized, including:

[0050] The Q(t) update result of each model error and the hierarchical early warning information are respectively displayed in the form of a curve and a chart, and through a RESTful API or other enterprise bus interface, data intercommunication is carried out with a business end of a supply chain management system or an enterprise resource planning system (ERP), and meanwhile, in a cloud database, the Q(t) change record of each batch of goods and the jump information of all events are saved in real time.

[0051] Compared with the prior art, the application has the following beneficial effects:

[0052] 1. By introducing a dynamic weight adjustment mechanism and environmental data quality evaluation, the application can monitor and adjust the expiration date prediction of goods in real time, so that the early warning result is more accurate, and the limitation of the traditional static model that cannot cope with complex environmental changes is avoided, and the dynamic weight mechanism can adjust the weight in a timely manner according to the change of environmental conditions, so as to ensure that the model accurately reflects the shelf life state of the goods at all times.

[0053] 2. The event-triggered jump compensation mechanism is introduced in the dynamic shelf life prediction model, so that the supply chain management system can respond to unexpected events in the transportation process in real time, and adjust the early warning range according to the remedial measures, thereby improving the flexibility of the supply chain management system in response to unexpected environmental changes.

[0054] 3. The expiration date of the randomly selected goods is detected and compared with the model prediction result, the early warning range is dynamically updated, and information is transmitted to the supply chain management system through a visual means, so that the supply chain management personnel can make more refined management decisions according to the real-time updated early warning information, thereby improving the efficiency and response speed of goods management. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort. Among them:

[0056] Figure 1 The overall flowchart of the goods expiration date early warning method based on the supply chain management system according to an embodiment of the application;

[0057] Figure 2 The comparison chart of the quality decay process of goods of different models of the goods expiration date early warning method based on the supply chain management system according to an embodiment of the application;

[0058] Figure 3 The model error distribution comparison chart (box type) of the goods expiration date early warning method based on the supply chain management system according to an embodiment of the application. DETAILED DESCRIPTION

[0059] In order to make the above objectives, features and advantages of the present application more clear and comprehensible, specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present application.

[0060] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without the specific details presented herein. In other instances, well-known methods have not been described in detail in order to avoid obscuring aspects of the present application.

[0061] Secondly, the "one embodiment" or "embodiment" referred to herein can include specific features, structures or characteristics in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments.

[0062] The present application is described in detail with reference to the accompanying drawings. In the detailed description of the embodiments of the present application, the sectional view of the device structure is partially enlarged without the general proportion for the convenience of description, and the schematic view is only an example, which should not limit the scope of protection of the present application. In addition, three-dimensional spatial dimensions including length, width and depth should be included in actual manufacture.

[0063] Meanwhile, in the description of the present application, it should be noted that the terms "upper, lower, inner and outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first, second or third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0064] Unless otherwise specifically defined and limited in the present application, the terms "mounting, connecting, connection" should be interpreted broadly, for example: it can be fixed connection, detachable connection or integral connection; it can also be mechanical connection, electrical connection or direct connection, it can also be indirectly connected through intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0065] Example 1

[0066] Referring to Figure 1 For the first embodiment of the present application, the embodiment provides a goods expiration date early warning method based on a supply chain management system, comprising:

[0067] S1, different types of sensors are arranged in the production, storage and transportation links of the supply chain to obtain environmental data in each sensor, the environmental data is quality evaluated, goods information is obtained from the supply chain management system, the environmental data that passes the quality evaluation is merged with the goods information to obtain a comprehensive feature vector of the goods;

[0068] Specifically, sensors affected by environmental factors such as temperature, humidity, gas concentration (such as CO2, O2) and light intensity are arranged in the production and storage environment of the supply chain, and sensors affected by human factors such as vibration and collision can be integrated in the transportation link;

[0069] Specifically, each sensor has a unique number and is connected to the data platform of the supply chain management system through wireless or wired means;

[0070] It should be noted that, considering that the sensors come from different manufacturers or the models of the sensors are different, in order to avoid distortion of a single data source obtained by a single sensor, the environmental data collected by each sensor needs to be quality evaluated;

[0071] Specifically, the quality evaluation can be expressed by a mathematical formula as follows:

[0072]

[0073] Where, v i is the current measurement value v of the ith sensor; is the average value of the measurement values of the remaining sensors in the same region; δ i is the threshold value δ of the ith sensor, if the δ exceeds 15% to 20% of the threshold values of the remaining sensors in the same region (the setting of the sensor threshold value can be adjusted according to the actual situation), it means that the environmental data collected by the current sensor is abnormal, the environmental data collected by the remaining sensors is synchronized through the time stamp, if there is a missing situation of the environmental data collected by the current sensor in the data synchronization process, the data is interpolated and predicted based on the data of the remaining sensors in the same region;

[0074] It should be noted that both the time stamp and the interpolation and prediction are the most common processing methods in data preprocessing, so they will not be described in detail;

[0075] Further, the environmental data that passes the quality evaluation is recorded as:

[0076] E(t) = [T(t), H(t), G(t), V(t), C(t), N1]

[0077] wherein E(t) represents the environmental data at time t, T(t) represents the temperature at time t, H(t) represents the humidity at time t, G(t) represents the gas concentration at time t, V(t) represents the vibration cumulative amount at time t, C(t) represents the collision intensity at time t; N1 represents a set of other environmental data, excluding the aforementioned temperature, humidity, gas concentration, vibration cumulative amount and collision intensity;

[0078] Further, the packaging integrity, initial shelf life reference value and the like of the goods information obtained from the supply chain management system are combined with the environmental data evaluated by the quality to obtain the comprehensive feature vector of the goods;

[0079] Specifically, the comprehensive feature vector of the goods is represented as:

[0080] X(t) = [E(t), I, P0, M]

[0081] wherein X(t) represents the comprehensive feature vector of the goods at time t; I represents the integrity of the goods packaging, which is a percentage value; P0 represents the initial shelf life reference value; M represents a set of other goods information, excluding the aforementioned integrity of the goods packaging and initial shelf life reference value;

[0082] It should be noted that the obtained comprehensive feature vector of the goods can be regarded as the digital twin of each piece of goods in the supply chain, so as to realize the digital supply chain data management mode, and by switching the application scene, only the goods information needs to be exchanged, for example, the drug information or the precision machinery information, etc. can meet the needs of the environmental data E(t), and thus the migration operation of the supply chain data collection can be realized;

[0083] S2, considering the influence of environmental data on the shelf life of goods, setting a dynamic weight for the comprehensive feature vector of the goods, constructing a dynamic shelf life prediction model with the comprehensive feature vector of the goods and the dynamic weight, introducing an event-triggered jump compensation mechanism in the dynamic shelf life prediction model, and setting the early warning range of the goods expiration date according to the event-triggered jump compensation mechanism;

[0084] It should be noted that since each environmental data in the environmental data E(t) can affect the shelf life of the goods, and the influence degree is different, a weight value updated with time needs to be allocated to each environmental data;

[0085] Specifically, the allocated weight value is denoted as:

[0086] ω(t) = [ω T (t), ωH (t), ω G (t), ω V (t), N2]

[0087] wherein ω(t) represents the weight set assigned at time t, ω T (t), ω H (t), ω G (t), ω V (t) respectively correspond to the weight values of temperature T(t), humidity H(t), gas concentration G(t), vibration cumulative amount V(t) and collision intensity C(t); N2 corresponds to N1, representing the weight value of the other environmental data set N1;

[0088] Further, the quality state of the goods at time t is obtained, and is initialized, and a discrete time step is set based on hours, and based on the obtained assigned weight set ω(t) and the comprehensive feature vector X(t) of the goods, the quality decay process of the goods is obtained through a difference equation;

[0089] Specifically, the difference equation is represented as:

[0090] Q(t+Δτ) = Q(t) - ΔτxF(X(t), ω(t))

[0091] wherein Q(t) represents the quality state of the goods at time t, and is initialized as Q(0) = Q0; Δτ represents the discrete time step, which is hours; F(X(t), ω(t)) represents the decay function;

[0092] Further, the decay function F(X(t), ω(t)) is represented as:

[0093]

[0094] wherein fi represents the i-th factor function, and N represents a specific environment, such as the above-mentioned temperature T, humidity H, gas concentration G, etc. i

[0095] It should be noted that each factor function is nonlinear or nonlinear, for example, the temperature factor uses the Arrhenius model, the humidity factor uses the exponential decay model, and the collision or vibration factor uses the piecewise linear model, etc.

[0096] Further, according to the obtained quality state Q(t) at time t, the lower limit of the quality safety of the goods Q low is set, when Q(t) ≤ Q low , it indicates that the goods have entered the near-expiration state and are not suitable for transportation, and the value of Q(t) is updated and the difference equation Q(t+Δτ) is recalculated;

[0097] ​Further, the value of the differential equation of each update is recorded, and the differential equation is marked with the attenuation state of the quality of the goods (such as the loss rate of goods, the number of customer returns, etc.);

[0098] It should be noted that the record of each calculation value of the differential equation can trace the change trajectory of the sensor from the production to the transportation of the goods;

[0099] It should be noted that in the actual supply chain, in addition to the daily smooth environmental changes, sudden events (such as power failure of cold chain, packaging damage, violent loading and unloading, etc.) will also cause non-continuous jumps or shocks to the shelf life of goods; at the same time, in order to more finely warn the risk of expiration of goods, multi-level threshold is needed for disposal, rather than only dividing into two states of expiration and non-expiration;

[0100] Further, on the basis of the dynamic shelf life prediction model Q(t), a jump compensation mechanism triggered by events is introduced, and when a sudden event is detected (such as packaging damage, temporary power failure leading to sudden temperature rise, violent loading and unloading of logistics leading to excessive vibration, etc.), jump correction is performed;

[0101] Specifically, the jump correction is as follows:

[0102] Q new (t)=Q(t)-ΔQ shock

[0103] Wherein, Q new (t) represents the value of Q(t) after jump, and ΔQ shock represents the severity of the sudden event;

[0104] Further, if remedial measures are taken after the sudden event (such as rapid cooling, immediate replacement of damaged packaging), the sudden event compensation term is defined as:

[0105] Q new (t)=Q new (t)+ΔQ recover

[0106] Wherein, ΔQ recover represents the effectiveness evaluation of the remedial measures;

[0107] Further, if the value of Q new (t) changes after the sudden event is repaired, if it does not change, it means that the environment returns to normal, and the value of Q(t) is updated to the current value of Q new (t);

[0108] Further, according to the value of Q(t) after updating, a first threshold, a second threshold and a third threshold are set, each threshold corresponds to a different warning level;

[0109] Specifically, the first threshold value, the second threshold value and the third threshold value are respectively represented as: β1<β2<β3;

[0110] When Q(t) approaches β1, the prediction model issues a primary early warning, prompting the relevant personnel to pay attention to the current environment of the commodity;

[0111] When Q(t) approaches β2, the prediction model issues a middle-level early warning, prompting the relevant personnel to need to carry out rapid promotion of the current commodity;

[0112] When Q(t) approaches β3, the prediction model issues a high-level early warning, prompting the relevant personnel to need to take the commodity off the shelf;

[0113] It should be noted that the threshold processing is taken as the early warning module of the supply chain management system, which can perform different levels of disposal (such as commodity storage, changing the rapid logistics channel, increasing the promotion intensity, etc.);

[0114] S3, randomly extract goods for shelf life detection, compare the prediction results of the dynamic shelf life prediction model with the detection results of the randomly extracted goods shelf life, obtain the model prediction error, update the early warning range of the goods shelf life according to the model prediction error, display the update results of each time in a visual way, and pass to the supply chain management system, so as to realize the accurate early warning of the goods shelf life in the supply chain management system;

[0115] Further, the model prediction error is realized by minimizing the objective function, as follows:

[0116]

[0117] Wherein, Q rs (t) is the quality state of the randomly extracted goods shelf life at t, and Θ represents the internal model parameters of each factor function f i .

[0118] It should be noted that if the model error is 0, it means that the dynamic shelf life prediction model is consistent with the actual goods shelf life, otherwise it means that the result of the dynamic shelf life prediction model is not ideal, and the value of Q(t) in the jump compensation mechanism needs to be rechecked, and the early warning division of the threshold processing is correspondingly modified;

[0119] Specifically, the Q(t) update result of each model error and the hierarchical early warning information are respectively displayed in the form of curves and charts, and data intercommunication is performed with the business end of the supply chain management system or the enterprise resource planning system (ERP) through the external RESTful API or other enterprise bus interfaces, and in the cloud database, in addition to recording the value of each difference equation, the Q(t) change record of each batch of goods and the jump information of all events are also saved in real time, thereby providing a basis for future goods analysis or responsibility tracing of each link.

[0120] Embodiment 2

[0121] Reference Figure 2 and Figure 3 For the second embodiment of the present application, the embodiment provides a goods near-expiration early warning method based on a supply chain management system, comprising: comparing the present application scheme and the traditional scheme through experiments to verify the beneficial effects of the present application scheme, and the experiment takes a small supply chain management system as an example, wherein the configuration of the test environment is as follows:

[0122] The test objects are fresh milk (shelf life of 7 days, storage condition of 0-4°C) and chilled meat (shelf life of 5 days, storage condition of -2-2°C); considering the most common small supply chain management system, a corresponding sensor is deployed in each link, and the supply chain management system is based on the SAP ERP platform and integrates the MySQL database and the Tableau visualization tool; a DS18B20 temperature sensor (±0.5°C precision) and a SHT31 humidity sensor (±2%RH precision) are installed in the production link; an AMS CCS811 gas sensor (CO2, VOC monitoring) and a Bosch BMA400 vibration sensor (±8g range) are deployed in the storage link; an ST LIS3DH three-axis accelerometer (collision monitoring) and a LoRa wireless transmission module are integrated in the transportation link;

[0123] The comparison scheme is as follows:

[0124] Traditional static model: linear regression prediction based on historical data (only temperature and humidity as fixed weight factors); dynamic weight model: dynamically adjusting the weight of environmental factors and introducing a jump compensation mechanism;

[0125] The test process is as follows:

[0126] Data acquisition and preprocessing: simulate the whole process of production, storage (72 hours) and transportation (24 hours) of fresh milk and chilled meat respectively; collect environmental data every 5 minutes, and generate the comprehensive feature vector X(t) of the goods through the formula Remove abnormal sensor data (set threshold δ>15%); use cubic spline interpolation to fill in the missing data, and generate the comprehensive feature vector X(t) of the goods;

[0127] Model training and prediction, the traditional model uses fixed weights (temperature weight 0.6, humidity 0.4), and the prediction formula is: Q(t) = Q0-k x (T x 0.6 + H x 0.4), k is the environmental influence factor; the dynamic model initializes the dynamic weight ω(t), which is updated every hour; the decay function F(X(t), ω(t)) = ∑ i∈{T，H，G，…，N} ω i (t) x f i (X i (t)), wherein the temperature factor f T is an exponential decay; the humidity factor f H is an exponential decay; simulate the sudden event in transportation, i.e. the cold chain power failure for 2 hours (temperature rises to 8℃), trigger the jump compensation Q new (t) = Q(t) - ΔQ shock , after repair, the compensation ΔQ recover = 0.8 x ΔQ shock ;

[0128] Error calibration and visualization, randomly select 10% of the goods for quality inspection, compare the actual shelf life with the predicted value, and take the minimum error of the objective function as the optimization target; the early warning threshold is set to β1 = 20% (primary), β2 = 40% (intermediate), and β1 = 60% (high), and the final test results are shown in Table 1;

[0129] Table 1 Comparison of test data

[0130]

[0131] As can be seen from Table 1, the traditional model cannot adapt to the dynamic environment due to the fixed weights, for example, when the temperature suddenly rises during transportation, the prediction deviation reaches 18%, resulting in a warning accuracy of only 72%; the present application adjusts the dynamic weight (temperature weight from 0.6 to 0.8), the accuracy is improved to 94%, especially in the cold chain power failure scenario, the jump compensation mechanism improves the error correction efficiency by 3 times (the error rate of the present model is 4% and that of the traditional static model is 18%), reference Figure 2Under 100 hours of goods quality test, the fitting accuracy of goods quality attenuation trajectory can be significantly improved by dynamic weight adjustment, which reflects that the jump compensation mechanism can effectively deal with emergencies; and the traditional model needs manual intervention to judge the emergency, and the response time is as long as 4.2 hours; while the dynamic model of the application shortens the response time to 0.5 hours through real-time jump compensation, and automatically triggers a hierarchical early warning to avoid large-scale loss of goods; in addition, the traditional model has a high misjudgment rate of 25% due to the neglect of gas concentration (CO2 accelerates corruption) and vibration accumulation (risk of package damage); while the application reduces the misjudgment rate to 5%, and alone, the dynamic weight model combined with random quality inspection feedback (objective function optimization) further reduces the model error rate from 9% to 4%, which is significantly better than the static model, as shown in Figure 3 In the figure, the error distribution of the traditional model is extensive (18±5%), and there is a long tail phenomenon; while the error of the model of the application is concentrated in (4±1.5%), and there is no abnormal point, which proves that the jump compensation in the application scheme effectively suppresses the error divergence problem; the traditional model relies on batch data synchronization (delay of 120 seconds) and cannot support real-time decision-making, and the application compresses the delay to 30 seconds through interpolation prediction and LoRa wireless transmission, which meets the timeliness requirements of fresh logistics;

[0132] The above test verifies the significant advantages of the application in early warning accuracy and error rate, which meets the fine management and control requirements of current goods supply chain.

[0133] Those skilled in the art will appreciate that the embodiments of the application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the application can be implemented in various computer languages, such as object-oriented programming languages Java and interpreted scripting language JavaScript.

[0134] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices produce a machine that implements the flowcharts and / or block diagrams. Figure 1 one flow or multiple flows and / or blocksFigure 1 means for performing the function specified by the block or blocks.

[0135] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 flow or flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0136] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 flow or flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.

[0137] Although preferred embodiments of the application have been described herein, changes and modifications can be suggested to one skilled in the art, and it is intended that the scope of the application be limited only by the appended claims and equivalents thereof.

[0138] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described herein.

Claims

1. A method for goods expiration date early warning based on a supply chain management system, characterized in that, The application relates to a dynamic shelf life prediction method and system. The application comprises the following steps: Different types of sensors are arranged at the production, storage and transportation links of a supply chain to obtain environmental data in each sensor, the environmental data is quality-evaluated, meanwhile, goods information is obtained from a supply chain management system, the environmental data after quality evaluation is combined with the goods information to obtain a comprehensive feature vector of the goods; A dynamic weight is set for the comprehensive feature vector of the goods in consideration of the influence of the environmental data on the shelf life of the goods, and a dynamic shelf life prediction model is constructed by using the comprehensive feature vector and the dynamic weight of the goods, an event-triggered jump compensation mechanism is introduced into the dynamic shelf life prediction model, and a warning range of the shelf life of the goods is set according to the event-triggered jump compensation mechanism; The application comprises the following steps: The quality state of the goods at time t is obtained and initialized, and a discrete time step is set based on hours; the difference equation is represented as: wherein, denotes the mass state of the goods at time t, initialized as ; denotes the discrete time step, in hours; denotes the decay function; is the goods' comprehensive feature vector, is the set of assigned weights; The decay function is represented as: wherein, wherein T represents temperature, H represents humidity, G represents gas concentration, and N represents the remaining environment other than T, H, and G. A quality decay process of the goods is obtained by using a difference equation based on the obtained distributed weight set and the comprehensive feature vector of the goods; The event-triggered jump compensation mechanism is introduced into the dynamic shelf life prediction model, and comprises the following steps: wherein, is expressed as a jump in the value, is expressed as the severity of the incident; If a sudden event is detected, jump correction is performed: wherein, represents an assessment of the effectiveness of the remedial action; If the value of the detected changes after the emergency is repaired, it indicates that the environment is not back to normal, and the value of is updated to the current value. If a remedial measure is taken after the sudden event, a sudden event compensation term is defined:

2. The method of claim 1, wherein the supply chain management system-based goods near expiration warning method is characterized by, Randomly selected goods are detected for the shelf life, the prediction result of the dynamic shelf life prediction model is compared with the detection result of the shelf life of the randomly selected goods, a model prediction error is obtained, the warning range of the shelf life of the goods is updated according to the model prediction error, the updating result of each time is displayed in a visual way, and is transmitted to the supply chain management system, so that accurate warning of the shelf life of the goods in the supply chain management system is realized. The environmental data in each sensor is obtained, and the environmental data is quality-evaluated, comprising the following steps: wherein, is the current measurement value v of the i-th sensor; is the average of the measurement values of the remaining sensors within the same area; denotes the threshold value for the i-th sensor ; If the threshold value If the threshold value of the current sensor is not exceeded by the threshold values of the remaining sensors in the same region, it indicates that the environmental data collected by the current sensor is normal, otherwise it indicates that the environmental data collected by the current sensor is abnormal, the environmental data collected by the remaining sensors is synchronized through the time stamp, and if the environmental data collected by the current sensor is missing during the data synchronization process, the data of the remaining sensors in the same region is used for interpolation prediction.

3. The method of claim 2, wherein the supply chain management system-based goods near expiration warning method is characterized by, The quality evaluation is expressed by a formula as follows: Meanwhile, goods information is obtained from the supply chain management system, the environmental data after quality evaluation is combined with the goods information to obtain a comprehensive feature vector of the goods, comprising the following steps:

4. The method of claim 2 or 3, wherein the supply chain management system-based goods expiration date warning method is characterized by, The environmental data collected by the sensor under the normal state is expressed as the environmental data after quality evaluation, meanwhile, goods information is obtained from the supply chain management system, and the environmental data after quality evaluation is combined with the goods information to obtain a comprehensive feature vector of the goods. The dynamic weight is set for the comprehensive feature vector of the goods in consideration of the influence of the environmental data on the shelf life of the goods, comprising the following steps:

5. The method of claim 1, wherein the supply chain management system-based goods near expiration warning method is characterized by, Since each environmental data in the environmental data has influence on the shelf life of the goods and the influence degree is different, a weight value updated with time is allocated to each environmental data to obtain a distributed weight set. According to the acquisition of the quality state at time t , set the quality safety lower limit of the goods ; When , it indicates that the goods have entered the near-expiration state, and it is not suitable for transportation, and the value of is updated, and the difference equation is recalculated; The dynamic shelf life prediction model is constructed by using the comprehensive feature vector and the dynamic weight of the goods, and further comprises the following steps:

6. The method of claim 1, wherein the supply chain management system-based goods near expiration warning method is characterized by, The value of the difference equation after each updating is recorded, and the decay state of the quality of the goods corresponding to the difference equation is marked. The warning range of the shelf life of the goods is set according to the event-triggered jump compensation mechanism, comprising the following steps: According to the value of the updated , the first threshold value, the second threshold value and the third threshold value are set, each of which corresponds to a different early warning level, and the relationship between the threshold values is represented as: ; When belongs to the first threshold , the prediction model issues a preliminary warning, prompting relevant personnel to pay attention to the current environment of the goods; When belongs to the second threshold , the prediction model issues a medium-level early warning, prompting relevant personnel to need to carry out rapid promotion of the current goods; When belongs to the third threshold , the prediction model issues a high-level early warning, prompting relevant personnel to remove the goods from the shelves.

7. The method of claim 1 or 6, wherein the supply chain management system-based goods expiration date warning method is characterized by, Randomly extract goods for shelf life detection, compare the prediction results of the dynamic shelf life prediction model with the detection results of the randomly extracted goods shelf life, obtain the model prediction error, including: The model prediction error is achieved by minimizing the objective function, as follows: wherein, a quality state of the shelf life of the goods is randomly extracted at time t, denotes an internal model parameter of the factor function .

8. The method of claim 6, wherein the supply chain management system-based goods near expiration warning method is characterized by, Update the early warning range of the goods expiration date according to the model prediction error, display the update result of each time in a visual way, and transfer it to the supply chain management system to realize accurate early warning of goods expiration date in the supply chain management system, including: The model error of each time is updated The update result and the early warning information are respectively displayed in the form of curves and charts, and data intercommunication is performed with a business end of a supply chain management system or an enterprise resource planning system through a RESTful API or other enterprise bus interfaces, and the change record of each batch of goods and the jump information of all events are saved in real time in a cloud database. ​

Images