Agricultural product cold chain distribution intelligent monitoring system and distribution method

Abstract

The invention discloses an agricultural product cold chain distribution intelligent monitoring system and a distribution method, and relates to the technical field of cold chain distribution. The agricultural product cold chain distribution intelligent monitoring system comprises an agricultural product analysis module, a route selection module, a vehicle selection module, a driver selection module and a monitoring module; the agricultural product analysis module receives a delivery order, extracts agricultural product information, calculates a total value value Vtota l of the order according to the agricultural product information, and retrieves a related knowledge base to query storage and transportation requirements based on the agricultural product information; the route selection module inputs a departure place and a destination in the delivery order into a map engine to obtain an alternative route set; according to the invention, the route selection module is used for screening qualified routes according to the distribution cost upper limit Cmax, and a vehicle and driver scoring mechanism is combined to respectively select cold chain vehicles and drivers which are most matched with task demand scores for resource allocation, so that accurate scheduling based on task features is realized, the vehicle utilization rate and the distribution efficiency are improved, and the scheduling quality is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of cold chain distribution, and in particular to an intelligent monitoring system and distribution method for cold chain distribution of agricultural products. Background Art

[0002] Currently, the widely used cold chain distribution system for agricultural products is in use by numerous fresh produce e-commerce platforms, agricultural product wholesale markets, and third-party cold chain logistics companies. Relying on the Internet of Things, AI algorithms, and big data analytics, the system enables digital management of the entire transportation process, from order generation and route planning to real-time monitoring. By integrating temperature and humidity sensors, GPS positioning devices, and on-board cameras, the system provides real-time information on vehicle location, cabin environment, and driver behavior, ensuring that perishable agricultural products are maintained in optimal storage and transportation conditions throughout the entire transportation process. This type of system has been implemented by platforms such as JD Cold Chain, Miss Fresh, and SF Express Cold Chain, effectively improving the on-time delivery rate of fresh produce and the stability of agricultural product quality, becoming a key infrastructure for modern agricultural product supply chain management.

[0003] Most existing systems allocate vehicles and drivers based on manual experience or static rules. They lack a quantitative analysis and scoring mechanism for matching order value with transportation task difficulty. This can result in high-value agricultural products not being allocated to the optimal cold chain resources, impacting transportation safety and economic efficiency. Furthermore, driver fatigue is monitored only through passive recording by onboard cameras, lacking multi-dimensional behavioral modeling such as real-time facial expression analysis and gesture recognition, posing safety risks.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention provides an intelligent monitoring system and distribution method for cold chain distribution of agricultural products.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an intelligent monitoring system and distribution method for cold chain distribution of agricultural products, comprising: an agricultural product analysis module, a route selection module, a vehicle selection module, a driver selection module and a monitoring module;

[0007] The agricultural product analysis module receives delivery orders, extracts agricultural product information, calculates the total value Vtotal of the order based on the agricultural product information, and queries storage and transportation requirements based on the agricultural product information retrieval knowledge base;

[0008] The route selection module inputs the origin and destination of the delivery order into the map engine to obtain a set of candidate routes. The total value Vtotal is multiplied by the preset weight coefficient to obtain the cost cap Cmax. Routes with travel times exceeding the delivery time requirement and expenditure costs exceeding the cost cap Cmax are eliminated to obtain a set of qualified routes. The route with the shortest travel time in the qualified route set is selected as the delivery route.

[0009] The vehicle selection module retrieves vehicle information from the vehicle information table, combines this information with a weighted scoring model to derive a score for each vehicle, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Cold chain vehicles with scores close to the demand score are then assigned for delivery.

[0010] The driver selection module obtains driver information from the driver information table, combines this information with a weighted scoring model to derive a score for each driver, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Drivers with scores close to the demand score are assigned to deliveries.

[0011] The monitoring module uses a facial key point detection model to analyze the driver's facial expressions and head posture to obtain a status score, and provides classification prompts based on the status score.

[0012] The agricultural product analysis module receives delivery orders submitted by users through the order management interface. The delivery order contains at least the following fields: agricultural product category, weight and quantity, origin and destination, packaging coefficient, delivery time limit, and order time. The agricultural product price database is called to obtain the unit price of the category in the current delivery order. When multiple categories are involved, the sub-value of each category is estimated item by item, and the total value of the delivery order Vtotal is calculated. The final calculation method is as follows:

[0013]

[0014] Where: Pi is the unit price of the i-th agricultural product; Qi is its weight; Ci is the packaging coefficient; M is the calculation coefficient;

[0015] Access the agricultural product cold chain knowledge base to match the storage and transportation requirements for each category in the delivery order, including the recommended temperature range, recommended humidity range, and safe transportation time.

[0016] The route selection module encapsulates the departure and destination text information in the delivery order as map engine request parameters, calls the map service platform to obtain no less than n alternative routes, and extracts the following parameters for each alternative route: the total route distance Lroute, the estimated time and the toll. All alternative routes are traversed according to the cost upper limit Cmax, and the routes that meet the estimated time less than the delivery time limit and the toll less than the cost upper limit Cmax are retained as the qualified route set. The route with the shortest time in the qualified route set is selected as the delivery route.

[0017] The vehicle selection module retrieves the vehicle information table to obtain a dispatchable vehicle information set, including the vehicle number, age, number of vehicle repairs, cold chain level, purchase price, and vehicle mileage. A vehicle score Svehicle is calculated for each vehicle in the vehicle information set using the following formula:

[0018] Svehicle=w1×f1+w2×f2+w3×f3+w4×f4+w5×f5

[0019] Among them, f1 is the age of the delivery vehicle, f2 is the number of times the delivery vehicle has been repaired, f3 is the cold chain level, f4 is the purchase price of the delivery vehicle, and f5 is the mileage of the delivery vehicle; w1 is the weight coefficient of the age of the delivery vehicle, w2 is the weight coefficient of the number of times the delivery vehicle has been repaired, w3 is the weight coefficient of the cold chain performance, w4 is the weight coefficient of the purchase price of the delivery vehicle, and w5 is the weight coefficient of the mileage of the delivery vehicle;

[0020] According to the storage and transportation requirements, select the vehicle set D that meets the minimum cold chain level requirements;

[0021] The task requirement score Stata is calculated based on the total distance of the delivery route Lroute and the total order value Vtotal:

[0022] Task=α×normalize(Lroute)+β×normalize(Vtotal)

[0023] Among them, normalize() is the normalization function, α and β are adjustment coefficients, and Lroute is the total distance of the delivery route;

[0024] According to the difference Δj = |Svehicle-Stask| between the vehicle score Svehicle and the task requirement score Stask, the cold chain vehicle with the smallest difference is selected as the delivery vehicle.

[0025] The driver selection module queries the driver information table for a set of available drivers. The driver information includes age, driving experience, number of accidents, and historical on-time delivery rate. The minimum and maximum values ​​of age, driving experience, number of accidents, and historical on-time delivery rate are calculated, respectively. Then, the minimum-maximum normalization method is used to linearly map the original values ​​to the range of 0 to 1 to obtain statistical values ​​for age, driving experience, number of accidents, and historical on-time delivery rate. A comprehensive score Sdriver is calculated for each driver:

[0026] Sdriver=wA(1-A)+wE×E+wI(1-I)+wH×H

[0027] Where A is the age statistic, E is the driving experience statistic, I is the number of accidents statistic, and H is the historical on-time delivery rate statistic; wA, wE, wI, and wH are the weight coefficients of the age statistic A, the driving experience statistic E, the number of accidents statistic I, and the historical on-time delivery rate statistic H, respectively;

[0028] The driver demand score is calculated based on the total route distance Lroute and the total order value Vtotal: Sta = α2·Lroute + β2·Vtotal, where α2 and β2 are the weight coefficients of the total route distance Lroute and the total order value Vtotal respectively;

[0029] Calculate the matching difference Δm = |Sdriverj - Stask| for each driver. The smaller the difference, the more compatible the driver is with the delivery order. Sort all candidate drivers by difference from smallest to largest, and select the driver with the smallest difference as the delivery driver.

[0030] The final selected driver information, delivery vehicle information and order information are packaged, and the dispatch interface pushes reminder messages to the driver's mobile phone.

[0031] The monitoring module calls a facial key point detection model, uses the facial key point detection model to determine the positions of 68 facial key points, calculates the eye aspect ratio (EAR), and marks an eye closure event when the EAR is less than 0.2 for 15 consecutive frames; counts the total number of eye closures per minute; calculates the head pitch angle based on the 2D-3D PnP algorithm, combining head key points and camera internal parameters; when the pitch angle exceeds ±15°, it is marked as an attention shift, and the total number of attention shifts per minute is counted;

[0032] The driver status score of the current frame is calculated by weighting the number of eye closures and the number of attention shifts with a weight of 0.5 and 0.5 respectively.

[0033] Set warning threshold one and warning threshold two. When the driver status score is less than warning threshold one, no mark is made. When the driver status score is greater than warning threshold one, a first-level voice reminder is triggered through the in-vehicle radio. When the driver status score is greater than warning threshold two, a second-level reminder is triggered, the seat vibrates, and a first-level voice reminder is triggered through the in-vehicle radio, prompting the driver to stop and rest.

[0034] The specific method for intelligent monitoring and distribution of cold chain distribution of agricultural products is as follows:

[0035] S1. The agricultural product analysis module receives the delivery order and extracts the agricultural product information, calculates the total order value Vtotal, and searches the cold chain knowledge base to match the corresponding storage and transportation requirements;

[0036] S2. The route selection module submits the origin and destination to the map engine to generate alternative routes, calculates the cost cap based on the total order value, and selects the route with the best efficiency from the qualified routes;

[0037] S3. The vehicle selection module queries available vehicle information, evaluates vehicles that meet the cold chain requirements, and matches suitable vehicles based on route length and order value.

[0038] S4. The driver selection module queries available driver information, uses a scoring model to evaluate qualified drivers, and matches suitable drivers based on route length and order value;

[0039] S5. The monitoring module performs real-time analysis of the driver's facial and head posture, calculates the status score, and issues a classification prompt when the warning level is reached.

[0040] The present invention provides an intelligent monitoring system and distribution method for cold chain distribution of agricultural products. Compared with the existing technology, it has the following advantages:

[0041] The present invention uses a route selection module to screen qualified routes based on the delivery cost cap Cmax. Combined with the vehicle and driver scoring mechanism, it selects cold chain vehicles and drivers that best match the task requirement scores for resource allocation, thereby achieving precise scheduling based on task characteristics, improving vehicle utilization and delivery efficiency, and optimizing scheduling quality.

[0042] The present invention uses the agricultural product analysis module to calculate the total order value Vtotal and match it with cold chain storage and transportation requirements, thereby quantifying the importance of distribution tasks and identifying professional requirements, thereby improving the pertinence of distribution routes and resource allocation, and enhancing the overall cold chain logistics service quality.

[0043] The present invention calls the facial key point detection and head posture recognition algorithm through the monitoring module to obtain the driver's status score in real time, and issues graded warnings based on the score threshold, thereby realizing an intelligent reminder and control mechanism for fatigue driving and distraction, effectively improving safety during transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the system framework of the present invention;

[0045] Figure 2 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] See also Figure 1 ,This application provides an intelligent monitoring system and distribution method for agricultural products cold chain distribution, including: an agricultural products analysis module, a route selection module, a vehicle selection module, a driver selection module and a monitoring module;

[0048] The agricultural product analysis module receives delivery orders, extracts agricultural product information, calculates the total value Vtotal of the order based on the agricultural product information, and queries storage and transportation requirements based on the agricultural product information retrieval knowledge base;

[0049] The route selection module inputs the origin and destination of the delivery order into the map engine to obtain a set of candidate routes. The total value Vtotal is multiplied by the preset weight coefficient to obtain the cost cap Cmax. Routes with travel times exceeding the delivery time requirement and expenditure costs exceeding the cost cap Cmax are eliminated to obtain a set of qualified routes. The route with the shortest travel time in the qualified route set is selected as the delivery route.

[0050] The vehicle selection module retrieves vehicle information from the vehicle information table, combines this information with a weighted scoring model to derive a score for each vehicle, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Cold chain vehicles with scores close to the demand score are then assigned for delivery.

[0051] The driver selection module obtains driver information from the driver information table, combines this information with a weighted scoring model to derive a score for each driver, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Drivers with scores close to the demand score are assigned to deliveries.

[0052] The monitoring module uses a facial key point detection model to analyze the driver's facial expressions and head posture to obtain a status score, and provides classification prompts based on the status score.

[0053] The agricultural product analysis module receives delivery orders submitted by users through the order management interface. The delivery order contains at least the following fields: agricultural product category, weight and quantity, origin and destination, packaging coefficient, delivery time limit, and order time. The agricultural product price database is called to obtain the unit price of the category in the current delivery order. When multiple categories are involved, the sub-value of each category is estimated item by item, and the total value of the delivery order Vtotal is calculated. The final calculation method is as follows:

[0054]

[0055] Where: Pi is the unit price of the i-th agricultural product; Qi is its weight; Ci is the packaging coefficient; M is the calculation coefficient;

[0056] In the intelligent monitoring system for cold chain distribution of agricultural products, the agricultural product analysis module, as the system's front-end analysis unit, undertakes the key tasks of comprehensively interpreting user order information and modeling basic parameters. By analyzing the order content, this module not only provides a decision-making basis for subsequent route planning and resource matching, but also establishes a technical foundation for the intelligent identification of cold chain storage and transportation requirements. The agricultural product analysis module receives delivery orders submitted by users through the order management interface. The order structure adopts a unified data specification format, which at least includes fields such as agricultural product category, weight and quantity, departure place, destination, packaging coefficient, delivery time limit and order time.

[0057] Access the agricultural product cold chain knowledge base to match the storage and transportation requirements for each category in the delivery order, including the recommended temperature range, recommended humidity range, and safe transportation time.

[0058] The agricultural product analysis module uses each category in the order as an independent query condition, accesses the agricultural product cold chain knowledge base, and retrieves matching storage and transportation requirements, including:

[0059] Recommended temperature range (e.g. 0°C to 4°C, below -18°C);

[0060] Recommended humidity range (e.g. 60% to 80%);

[0061] Safe transportation time (the maximum transportation time allowed from departure to destination);

[0062] Cold chain grade requirements (such as Class A and Class B refrigeration standards).

[0063] The route selection module encapsulates the departure and destination text information in the delivery order as map engine request parameters, calls the map service platform to obtain no less than n alternative routes, and extracts the following parameters for each alternative route: the total route distance Lroute, the estimated time and the toll. All alternative routes are traversed according to the cost upper limit Cmax, and the routes that meet the estimated time less than the delivery time limit and the toll less than the cost upper limit Cmax are retained as the qualified route set. The route with the shortest time in the qualified route set is selected as the delivery route.

[0064] The vehicle selection module retrieves the vehicle information table to obtain a dispatchable vehicle information set, including the vehicle number, age, number of vehicle repairs, cold chain level, purchase price, and vehicle mileage. A vehicle score Svehicle is calculated for each vehicle in the vehicle information set using the following formula:

[0065] Svehicle=w1×f1+w2×f2+w3×f3+w4×f4+w5×f5

[0066] Among them, f1 is the age of the delivery vehicle, f2 is the number of times the delivery vehicle has been repaired, f3 is the cold chain level, f4 is the purchase price of the delivery vehicle, and f5 is the mileage of the delivery vehicle; w1 is the weight coefficient of the age of the delivery vehicle, w2 is the weight coefficient of the number of times the delivery vehicle has been repaired, w3 is the weight coefficient of the cold chain performance, w4 is the weight coefficient of the purchase price of the delivery vehicle, and w5 is the weight coefficient of the mileage of the delivery vehicle;

[0067] The vehicle selection module first retrieves the vehicle information table to obtain the information set of all vehicles currently in a dispatchable state (i.e., idle, operating normally, and without maintenance plans). The structured information of each vehicle should include at least the following parameter fields:

[0068] Vehicle number (for unique identification);

[0069] Years of use f1 (reflecting the degree of vehicle aging);

[0070] Cumulative number of repairs f2 (reflecting failure rate);

[0071] Cold chain level f3 (such as level I, level II, level III, reflecting refrigeration capacity and temperature control accuracy);

[0072] Purchase price f4 (used to indirectly reflect the vehicle's performance level);

[0073] Cumulative mileage f5 (reflects the wear and fatigue of the vehicle).

[0074] According to the storage and transportation requirements, select the vehicle set D that meets the minimum cold chain level requirements;

[0075] The task requirement score Stata is calculated based on the total distance of the delivery route Lroute and the total order value Vtotal:

[0076] Task=α×normalize(Lroute)+β×normalize(Vtotal)

[0077] Among them, normalize() is the normalization function, α and β are adjustment coefficients, and Lroute is the total distance of the delivery route;

[0078] The system calculates the task demand score based on the actual attributes of the delivery task to measure the intensity of the current task's demand for transportation resources. The score is based on the following two key dimensions: the total distance of the delivery route Lroute, which is used to measure the transportation stability and durability requirements; the total value of the agricultural product order Vtotal, which reflects the economic value of the task and the corresponding service level requirements.

[0079] According to the difference Δj = |Svehicle-Stask| between the vehicle score Svehicle and the task requirement score Stask, the cold chain vehicle with the smallest difference is selected as the delivery vehicle.

[0080] The present invention uses the agricultural product analysis module to calculate the total order value Vtotal and match it with cold chain storage and transportation requirements, thereby quantifying the importance of distribution tasks and identifying professional requirements, thereby improving the pertinence of distribution routes and resource allocation, and enhancing the overall cold chain logistics service quality.

[0081] The driver selection module queries the driver information table for a set of available drivers. The driver information includes age, driving experience, number of accidents, and historical on-time delivery rate. The minimum and maximum values ​​of age, driving experience, number of accidents, and historical on-time delivery rate are calculated, respectively. Then, the minimum-maximum normalization method is used to linearly map the original values ​​to the range of 0 to 1 to obtain statistical values ​​for age, driving experience, number of accidents, and historical on-time delivery rate. A comprehensive score Sdriver is calculated for each driver:

[0082] Sdriver=wA(1-A)+wE×E+wI(1-I)+wH×H

[0083] Where A is the age statistic, E is the driving experience statistic, I is the number of accidents statistic, and H is the historical on-time delivery rate statistic; wA, wE, wI, and wH are the weight coefficients of the age statistic A, the driving experience statistic E, the number of accidents statistic I, and the historical on-time delivery rate statistic H, respectively;

[0084] The system sets corresponding weight coefficients based on the impact of each indicator on delivery quality, and calculates a comprehensive score for each candidate driver to measure his or her overall transportation performance.

[0085] The system retrieves the set of drivers whose current status is "Available" from the driver information database, filtering out those who are currently in unavailable status such as "In Transit", "Vacation", "Pending Confirmation", etc. Each driver should include the following basic information fields:

[0086] Age (in years);

[0087] Driving experience (i.e., the number of years of holding a driver's license and engaging in actual transportation work);

[0088] The cumulative number of accidents (based on the number of accidents caused by liability during historical transportation);

[0089] Historical on-time delivery rate (calculate the percentage of on-time deliveries based on all delivery tasks in the past year).

[0090] The driver demand score is calculated based on the total route distance Lroute and the total order value Vtotal: Sta = α2·Lroute + β2·Vtotal, where α2 and β2 are the weight coefficients of the total route distance Lroute and the total order value Vtotal respectively;

[0091] Calculate the matching difference Δm = |Sdriverj - Stask| for each driver. The smaller the difference, the more compatible the driver is with the delivery order. Sort all candidate drivers by difference from smallest to largest, and select the driver with the smallest difference as the delivery driver.

[0092] The final selected driver information, delivery vehicle information and order information are packaged, and the dispatch interface pushes reminder messages to the driver's mobile phone.

[0093] When the system completes the final driver selection, it will package the driver's complete information (including number, contact information, rating record, etc.), the selected delivery vehicle information and delivery order details, and push them to the driver's mobile terminal through the scheduling interface, prompting him to accept the order and prepare for transportation.

[0094] The monitoring module calls a facial key point detection model, uses the facial key point detection model to determine the positions of 68 facial key points, calculates the eye aspect ratio (EAR), and marks an eye closure event when the EAR is less than 0.2 for 15 consecutive frames; counts the total number of eye closures per minute; calculates the head pitch angle based on the 2D-3D PnP algorithm, combining head key points and camera internal parameters; when the pitch angle exceeds ±15°, it is marked as an attention shift, and the total number of attention shifts per minute is counted;

[0095] The monitoring module first calls a facial key point detection model (such as a face annotation network based on Dlib or Mediapipe) to identify the locations of 68 key points in the driver's facial image in real time, focusing on the following areas: eye contour (12 key points);

[0096] Nose bridge and chin (for gesture recognition);

[0097] Facial boundaries and brow arches (to help enhance model stability) use the 2D-3D PnP algorithm (Perspective-n-Point) to accurately estimate head pose by combining the following elements:

[0098] 2D keypoint locations: image coordinates from the detection model;

[0099] The driver status score of the current frame is calculated by weighting the number of eye closures and the number of attention shifts with a weight of 0.5 and 0.5 respectively.

[0100] Set warning threshold one and warning threshold two. When the driver status score is less than warning threshold one, no mark is made. When the driver status score is greater than warning threshold one, a first-level voice reminder is triggered through the in-vehicle radio. When the driver status score is greater than warning threshold two, a second-level reminder is triggered, the seat vibrates, and a first-level voice reminder is triggered through the in-vehicle radio, prompting the driver to stop and rest.

[0101] The present invention calls the facial key point detection and head posture recognition algorithm through the monitoring module to obtain the driver's status score in real time, and issues graded warnings based on the score threshold, thereby realizing an intelligent reminder and control mechanism for fatigue driving and distraction, effectively improving safety during transportation.

[0102] The specific method for intelligent monitoring and distribution of cold chain distribution of agricultural products is as follows:

[0103] S1. The agricultural product analysis module receives the delivery order and extracts the agricultural product information, calculates the total order value Vtotal, and searches the cold chain knowledge base to match the corresponding storage and transportation requirements;

[0104] S2. The route selection module submits the origin and destination to the map engine to generate alternative routes, calculates the cost cap based on the total order value, and selects the route with the best efficiency from the qualified routes;

[0105] S3. The vehicle selection module queries available vehicle information, evaluates vehicles that meet the cold chain requirements, and matches suitable vehicles based on route length and order value.

[0106] S4. The driver selection module queries available driver information, uses a scoring model to evaluate qualified drivers, and matches suitable drivers based on route length and order value;

[0107] S5. The monitoring module performs real-time analysis of the driver's facial and head posture, calculates the status score, and issues a classification prompt when the warning level is reached.

[0108] The present invention uses a route selection module to screen qualified routes according to the delivery cost upper limit Cmax, and combines the vehicle and driver scoring mechanism to select cold chain vehicles and drivers that best match the task requirement scores for resource allocation, thereby achieving precise scheduling based on task characteristics, improving vehicle utilization and delivery efficiency, and optimizing scheduling quality.

[0109] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0110] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent monitoring system for cold chain distribution of agricultural products, characterized in that: include: Agricultural product analysis module, route selection module, vehicle selection module, driver selection module and monitoring module; The agricultural product analysis module receives delivery orders, extracts agricultural product information, calculates the total value Vtotal of the order based on the agricultural product information, and queries storage and transportation requirements based on the agricultural product information retrieval knowledge base; The route selection module inputs the origin and destination of the delivery order into the map engine to obtain a set of candidate routes. The total value Vtotal is multiplied by the preset weight coefficient to obtain the cost cap Cmax. Routes with travel times exceeding the delivery time requirement and expenditure costs exceeding the cost cap Cmax are eliminated to obtain a set of qualified routes. The route with the shortest travel time in the qualified route set is selected as the delivery route. The vehicle selection module retrieves vehicle information from the vehicle information table, combines this information with a weighted scoring model to derive a score for each vehicle, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Cold chain vehicles with scores close to the demand score are then assigned for delivery. The driver selection module obtains driver information from the driver information table, combines this information with a weighted scoring model to derive a score for each driver, and then calculates a demand score based on the length of the delivery route and the total value of the agricultural products. Drivers with scores close to the demand score are assigned to deliveries. The monitoring module uses a facial key point detection model to analyze the driver's facial expressions and head posture to obtain a status score, and provides classification prompts based on the status score.

2. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 1 is characterized in that: The agricultural product analysis module receives delivery orders submitted by users through the order management interface. The delivery order contains at least the following fields: agricultural product category, weight and quantity, origin and destination, packaging coefficient, delivery time limit, and order time. The agricultural product price database is called to obtain the unit price of the category in the current delivery order. When multiple categories are involved, the sub-value of each category is estimated item by item, and the total value of the delivery order Vtotal is calculated. The final calculation method is as follows: Where: Pi is the unit price of the i-th agricultural product; Qi is its weight; Ci is the packaging coefficient; M is the calculation coefficient; Access the agricultural product cold chain knowledge base to match the storage and transportation requirements for each category in the delivery order, including the recommended temperature range, recommended humidity range, and safe transportation time.

3. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 1 is characterized in that: The route selection module encapsulates the departure and destination text information in the delivery order as map engine request parameters, calls the map service platform to obtain no less than n alternative routes, and extracts the following parameters for each alternative route: the total route distance Lroute, the estimated time and the toll. All alternative routes are traversed according to the cost upper limit Cmax, and the routes that meet the estimated time less than the delivery time limit and the toll less than the cost upper limit Cmax are retained as the qualified route set. The route with the shortest time in the qualified route set is selected as the delivery route.

4. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 1, characterized in that: The vehicle selection module retrieves the vehicle information table to obtain a dispatchable vehicle information set, including the vehicle number, age, number of vehicle repairs, cold chain level, purchase price, and vehicle mileage. A vehicle score Svehicle is calculated for each vehicle in the vehicle information set using the following formula: Svehicle=w1×f1+w2×f2+w3×f3+w4×f4+w5×f5 Among them, f1 is the service life of the delivery vehicle, f2 is the number of maintenance times of the delivery vehicle, f3 is the cold chain level, f4 is the purchase price of the delivery vehicle, and f5 is the mileage of the delivery vehicle; w1 is the weight coefficient of the service life of the delivery vehicle, w2 is the weight coefficient of the number of maintenance times of the delivery vehicle, w3 is the weight coefficient of the cold chain performance, w4 is the weight coefficient of the purchase price of the delivery vehicle, and w5 is the weight coefficient of the mileage of the delivery vehicle.

5. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 4 is characterized in that: According to the storage and transportation requirements, select the vehicle set D that meets the minimum cold chain level requirements; The task requirement score Stata is calculated based on the total distance of the delivery route Lroute and the total order value Vtotal: Task=α×normalize(Lroute)+β×normalize(Vtotal) Among them, normalize() is the normalization function, α and β are adjustment coefficients, and Lroute is the total distance of the delivery route; According to the difference Δj = |Svehicle-Stask| between the vehicle score Svehicle and the task requirement score Stask, the cold chain vehicle with the smallest difference is selected as the delivery vehicle.

6. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 1 is characterized in that: The driver selection module queries the driver information table for a set of available drivers. The driver information includes age, driving experience, number of accidents, and historical on-time delivery rate. The minimum and maximum values ​​of age, driving experience, number of accidents, and historical on-time delivery rate are calculated, respectively. Then, the minimum-maximum normalization method is used to linearly map the original values ​​to the range of 0 to 1 to obtain statistical values ​​for age, driving experience, number of accidents, and historical on-time delivery rate. A comprehensive score Sdriver is calculated for each driver: Sdriver=wA(1-A)+wE×E+wI(1-I)+wH×H Among them, A is the age statistic, E is the driving experience statistic, I is the accident number statistic, and H is the historical on-time delivery rate statistic; wA, wE, wI, and wH are the weight coefficients of the age statistic A, driving experience statistic E, accident number statistic I, and historical on-time delivery rate statistic H, respectively.

7. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 6, characterized in that: The driver demand score is calculated based on the total route distance Lroute and the total order value Vtotal: Sta = α2·Lroute + β2·Vtotal, where α2 and β2 are the weight coefficients of the total route distance Lroute and the total order value Vtotal respectively; Calculate the matching difference Δm = |Sdriverj - Stask| for each driver. The smaller the difference, the more compatible the driver is with the delivery order. Sort all candidate drivers by difference from smallest to largest, and select the driver with the smallest difference as the delivery driver. The final selected driver information, delivery vehicle information and order information are packaged, and the dispatch interface pushes reminder messages to the driver's mobile phone.

8. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 1 is characterized in that: The monitoring module calls the facial key point detection model, uses the facial key point detection model to determine the positions of 68 facial key points, calculates the eye aspect ratio (EAR), and marks an eye closure event when the EAR is lower than 0.2 for 15 consecutive frames; The total number of eye closures per minute is counted. Based on the 2D-3D PnP algorithm, the head pitch angle is calculated by combining head key points and camera internal parameters. When the pitch angle exceeds ±15°, it is marked as attention shift, and the total number of attention shifts per minute is counted.

9. The intelligent monitoring system for cold chain distribution of agricultural products according to claim 8, characterized in that: The driver status score of the current frame is calculated by weighting the number of eye closures and the number of attention shifts with a weight of 0.5 and 0.5 respectively. Set warning threshold one and warning threshold two. When the driver status score is less than warning threshold one, no mark is made. When the driver status score is greater than warning threshold one, a first-level voice reminder is triggered through the in-vehicle radio. When the driver status score is greater than warning threshold two, a second-level reminder is triggered, the seat vibrates, and a first-level voice reminder is triggered through the in-vehicle radio, prompting the driver to stop and rest.

10. An intelligent monitoring and distribution method for cold chain distribution of agricultural products, characterized in that: The specific method for intelligent monitoring and distribution of cold chain distribution of agricultural products is as follows: S1. The agricultural product analysis module receives the delivery order and extracts the agricultural product information, calculates the total order value Vtotal, and searches the cold chain knowledge base to match the corresponding storage and transportation requirements; S2. The route selection module submits the origin and destination to the map engine to generate alternative routes, calculates the cost cap based on the total order value, and selects the route with the best efficiency from the qualified routes; S3. The vehicle selection module queries available vehicle information, evaluates vehicles that meet the cold chain requirements, and matches suitable vehicles based on route length and order value. S4. The driver selection module queries available driver information, uses a scoring model to evaluate qualified drivers, and matches suitable drivers based on route length and order value; S5. The monitoring module performs real-time analysis of the driver's facial and head posture, calculates the status score, and issues a classification prompt when the warning level is reached.

Images