Intelligent analysis and distribution method for vegetables directly supplied to vegetable base

Abstract

The invention relates to the technical field of agricultural product distribution, and relates to an intelligent analysis and distribution method for vegetables directly supplied to a vegetable base. According to the method, the maturity associated characteristics of the picked vegetables are obtained, and the maturity is obtained and classified through a maturity evaluation algorithm; constructing a maturity-delivery time-delivery environment-predicted received vegetable quality score prediction model based on historical delivery data, combining predicted delivery time and predicted delivery process environment parameters, and according to the predicted received vegetable quality score output by the prediction model, carrying out delivery distribution on vegetables of different maturity types; meanwhile, the accuracy rate of the prediction model is analyzed based on user feedback information, and the prediction model is iteratively optimized according to the accuracy rate of the prediction model. Through a closed-loop feedback mechanism, model parameters can be continuously corrected to match an actual distribution scene, so that the prediction precision of the model and the adaptability of the system are improved, and the stability of vegetable quality in a long-term distribution process is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural product distribution, and relates to a vegetable intelligent analysis and distribution method for direct supply of a vegetable base. BACKGROUND

[0002] In the direct supply mode of a vegetable base, the quality maintenance of vegetables from picking to reaching the hands of users is a core link for improving user experience. With the increasing demand of consumers for the freshness of fresh vegetables, an intelligent analysis method that can accurately evaluate the maturity of vegetables and predict the quality changes in the distribution process is urgently needed to realize efficient and high-quality distribution of vegetables from the base to users.

[0003] Chinese Patent Publication No. CN119761936A discloses a farm fresh vegetable precise distribution system and method based on intelligent algorithm. The system integrates multi-source data through a data collection module, analyzes market demand changes through an intelligent prediction module, adjusts planting plans and supply chain networks through a resource optimization configuration module, executes distribution plans through a precise distribution module, and provides information query functions through a user interaction module. This scheme improves the flexibility of farm production and the efficiency of the supply chain, and minimizes costs and time.

[0004] However, the existing technology has the following problems: 1. The existing technology focuses on market demand prediction and resource allocation, and does not quantitatively evaluate the maturity of picked vegetables. The maturity of vegetables directly affects the quality changes in the distribution process, and the lack of evaluation will lead to a mismatch between the distribution scheme and the actual state of the vegetables, which may cause premature vegetables to rot or late-maturing vegetables to be in a suboptimal state for consumption.

[0005] 2. The resource optimization configuration of the existing technology focuses on planting plans and inventory levels, and does not dynamically allocate based on the predicted quality scores of vegetables of different maturity under specific distribution conditions, affecting the scientificity of vegetable distribution strategies. When there is a difference between order demand and vegetable inventory, low-maturity vegetables may be preferentially allocated to long-distance areas, resulting in insufficient quality when received by users. SUMMARY

[0006] To overcome the defects of the prior art, the present application provides a vegetable intelligent analysis and distribution method for direct supply of a vegetable base, which realizes precise distribution with high quality and low loss through maturity grading, quality prediction, priority allocation, and feedback iteration.

[0007] The technical solution adopted by the present application to solve its technical problems is: a vegetable intelligent analysis and distribution method for direct supply of a vegetable base, comprising: obtaining maturity-related features of picked vegetables, and analyzing the maturity-related features using a maturity evaluation algorithm to obtain maturity.

[0008] The picked vegetables are classified based on the preset maturity range, and vegetables with the same maturity range are classified into the same maturity category.

[0009] A prediction model of maturity-delivery time-delivery environment-estimated received vegetable quality score is constructed based on historical delivery data, which takes vegetable maturity, delivery time and delivery process environment parameters as inputs and outputs estimated received vegetable quality score.

[0010] Different maturity category vegetables are allocated for delivery according to the predicted delivery time of the delivery area and the predicted delivery process environment parameters, combined with the predicted received vegetable quality score of different maturity category vegetables output by the prediction model.

[0011] Feedback information of users after receiving the vegetables is collected through online feedback questionnaires, the accuracy of the prediction model is analyzed based on the feedback information, and the prediction model is iteratively optimized according to the accuracy of the prediction model.

[0012] Compared with the prior art, the present application has the following beneficial effects: (1) The present application obtains the appearance feature data and internal ingredient data of the picked vegetables, analyzes the maturity by using a maturity evaluation algorithm, and classifies the picked vegetables based on the preset maturity range, thereby realizing accurate quantification and grading of vegetable maturity, providing a scientific basis for subsequent delivery allocation, and improving the quality consistency of the vegetables received by users.

[0013] (2) The present application constructs a prediction model of maturity-delivery time-delivery environment-estimated received vegetable quality score based on historical delivery data, accurately predicts the quality of the vegetables delivered according to the time and environmental parameters of the delivery area, thereby optimizing the delivery allocation strategy and improving the user's satisfaction with the freshness of the vegetables.

[0014] (3) The present application outputs the predicted received vegetable quality score of different maturity category vegetables according to the prediction model, and performs priority delivery allocation combined with the order demand quantity of the delivery area and the inventory quantity of each maturity category vegetable, thereby realizing optimal utilization of vegetable resources by dynamically matching different maturity vegetables and delivery conditions, ensuring that high-score quality vegetables meet the order demand first, and reducing vegetable loss caused by improper delivery.

[0015] (4) The present application collects feedback information of users after receiving the vegetables, analyzes the accuracy of the prediction model based on the feedback information, and iteratively optimizes the prediction model according to the accuracy of the prediction model, thereby continuously correcting the model parameters to match the actual delivery scenario through a closed-loop feedback mechanism, improving the prediction accuracy of the model and the adaptability of the system, and ensuring the stability of the quality of the vegetables in the long-term delivery process. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the following embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0017] Figure 1 The method flow step schematic diagram of the present application.

[0018] Figure 2 The setting step schematic diagram of the maturity evaluation algorithm in the present application.

[0019] Figure 3 The prediction model iterative optimization judgment flow schematic diagram in the present application. DETAILED DESCRIPTION

[0020] Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. Note that the relative arrangement, numerical expressions, and numerical values of the components and steps set forth in these embodiments are not limiting to the scope of the present application unless otherwise specifically stated. It should also be understood that the sizes of the various portions shown in the drawings are not drawn to scale for the sake of convenience in description.

[0021] The following description of at least one example embodiment is merely illustrative in nature and is in no way limiting to the scope of the application and its applications or uses. Techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail, but should be considered part of the specification, where appropriate.

[0022] In all the examples shown and discussed herein, any specific values should be interpreted as merely illustrative, and not as a limitation. Other examples of the example embodiments can therefore have different values.

[0023] Reference will now be made to Figure 1 As shown, the present application provides a vegetable intelligent analysis and distribution method for direct supply of vegetable bases, comprising: S1, obtaining the maturity correlation characteristics of the picked vegetables, and using a maturity evaluation algorithm to analyze the maturity correlation characteristics to obtain the maturity.

[0024] It should be noted that the maturity correlation characteristics of the picked vegetables are obtained, specifically: the appearance feature data and internal component data of the picked vegetables are obtained based on an image acquisition device and a spectrum detection device, the appearance feature data includes color, shape index and hardness, and the internal component data includes sugar content and acidity.

[0025] The obtained appearance feature data and internal component data are preprocessed to eliminate abnormal data and perform standardization conversion to obtain the preprocessed maturity correlation features. The standardization conversion refers to Z-score standardization processing.

[0026] Based on the preprocessed maturity correlation features, the maturity of the picked vegetables is calculated through a set maturity evaluation algorithm.

[0027] In a specific embodiment, the color degree refers to multi-angle shooting of each picked vegetable under natural environment lighting conditions through a high-resolution image acquisition device, extraction of the effective area of the vegetable through an image segmentation algorithm, and analysis of the color of the effective area using an HSV color model. The image segmentation algorithm and the HSV color model are both prior art, and will not be described in detail.

[0028] The shape index refers to determination of an edge contour based on the effective area of the vegetable, determination of the longest axis length and the shortest axis length based on the edge contour of the vegetable, and taking the ratio of the shortest axis length to the longest axis length as the shape index, which is used to reflect the regularity of the vegetable.

[0029] The hardness refers to testing on the surface of each vegetable through a fruit hardness tester, recording the maximum pressure value per unit area of each vegetable at different surface positions, and taking the average value as the hardness. The sugar content refers to adhesion of a corresponding detection probe of a near-infrared spectrum detection device to the surface of the vegetable, multiple spectrum scans of the surface of the vegetable to obtain average spectrum feature data, and reading of the soluble solid content of the vegetable as the sugar content based on a pre-labeled sucrose concentration and spectrum feature data corresponding relationship.

[0031] The acidity refers to matching of the average spectrum feature data of the surface of the vegetable with a pre-selected acidity and spectrum feature data corresponding relationship, and output of the acidity value of the vegetable.

[0032] The pre-labeled sucrose concentration and spectrum feature data corresponding relationship and the acidity and spectrum feature data corresponding relationship are both calibrated through a large amount of experimental data tests, for example, accurate measurement of the actual sugar content and acidity value of a batch of vegetable samples covering different sugar content and acidity in the laboratory using a standard sugar meter, simultaneous acquisition of the spectrum data of the vegetables using a near-infrared device, and establishment of a quantitative relationship model between the spectrum data and the measured sucrose concentration / acidity value using a chemometrics method. The chemometrics method is, for example, a PLS regression method.

[0033] The elimination of abnormal data refers to elimination of data that is obviously beyond a reasonable range, such as negative sugar content and hardness value far higher than the maximum possible value of the same type of vegetable. Such abnormal data is usually caused by equipment failure or operation error.

[0034] As Figure 2 shown in the figure, the maturity evaluation algorithm is set in the following way: a large number of experimental vegetable samples are collected, each vegetable sample contains pretreated color, shape index, hardness, sugar content and acidity, and the corresponding real maturity.

[0035] A multiple linear regression equation between the maturity-related characteristics and the real maturity is established by linear regression method. The multiple linear regression equation is , are the regression coefficients of color, shape index, hardness, sugar content and acidity, respectively, are the pretreated color, shape index, hardness, sugar content and acidity, respectively, is the constant term.

[0036] The constant term and the regression coefficients of each feature of the multiple linear regression equation are estimated by the least squares method, so that the residual sum of squares between the predicted maturity of the multiple linear regression equation and the real maturity is minimized.

[0037] The pretreated maturity-related characteristics of the picked vegetables are input into the established multiple linear regression equation, and the maturity of the picked vegetables is output.

[0038] S2, based on the preset maturity range, the picked vegetables are classified, and the vegetables with the same maturity in the same preset maturity range are divided into the same maturity category.

[0039] It should be noted that the classification of the picked vegetables based on the preset maturity range is as follows: according to the vegetable species and the preset optimal edible maturity interval, a plurality of maturity grading ranges are determined.

[0040] The maturity of the picked vegetables is matched with each maturity grading range of the corresponding species of vegetables, and the vegetables with the same maturity in the same maturity grading range are divided into the same maturity category.

[0041] The number of vegetables in each maturity category is counted, and if the number of vegetables in a certain maturity category exceeds the preset proportion threshold of the total number of vegetables in this batch, the grading range boundary is adjusted and then classified again.

[0042] In one specific embodiment, the optimal edible maturity interval of different species of vegetables can be determined by referring to the fresh vegetable industry standard or market consumption research data.

[0043] The grading rules of the plurality of maturity grading ranges are based on the optimal edible maturity interval as the benchmark, and are expanded in the direction of lower and higher maturity, forming 3-5 grading ranges. For example, tomatoes can be classified into unripe level, sub-mature level, optimal mature level and over-mature level.

[0044] The preset proportion threshold refers to setting the proportion of the number of different kinds of vegetables in different maturity categories in the total number of the batch of vegetables according to production needs, for example, the preset proportion threshold of the optimal maturity level of tomatoes is not more than 60%, thereby effectively avoiding excessive sorting pressure in the subsequent process.

[0045] When the number of vegetables in the optimal maturity level exceeds 60% of the total number of the batch of vegetables, the boundaries are expanded by 5% to the submature level and the overmature level respectively, until the proportions of all categories meet the preset proportion threshold requirement.

[0046] The present application realizes accurate quantification and grading of vegetable maturity by obtaining the appearance feature data and internal component data of the picked vegetables, using a maturity evaluation algorithm to analyze the maturity, and classifying the picked vegetables based on a preset maturity range, thereby providing a scientific basis for subsequent distribution allocation and improving the quality consistency of the vegetables received by users.

[0047] S3, a prediction model of maturity-distribution time-distribution environment-expected received vegetable quality score is constructed based on historical distribution data, wherein the prediction model takes vegetable maturity, distribution time and distribution process environment parameters as inputs, and takes expected received vegetable quality score as output.

[0048] It should be noted that the construction of the prediction model of maturity-distribution time-distribution environment-expected received vegetable quality score is as follows: collecting historical distribution data of the same kind of picked vegetables in the vegetable base, the historical distribution data including the initial maturity of the vegetables in each historical distribution record, the actual distribution time, each environment parameter in the distribution process environment parameter sequence, and the quality score of the received vegetables by users, to form a historical distribution data set.

[0049] The historical distribution data set is divided into a training set and a test set, and the feature parameters of the users receiving vegetables with a quality score higher than a preset quality score threshold are retained from the training set as model input variables.

[0050] Based on the screened model input variables and the corresponding quality score of the received vegetables by users, an initial prediction model is constructed using a machine learning algorithm, the initial prediction model is tested by the test set to optimize the model parameters, and the prediction model of maturity-distribution time-distribution environment-expected received vegetable quality score is obtained.

[0051] In a specific embodiment, the actual distribution time refers to the total time from when the vegetables are picked and packaged to when the user signs for the vegetables.

[0052] Each environment parameter in the distribution process environment parameter sequence includes average temperature, maximum temperature, average humidity, minimum humidity, and average vibration frequency and maximum vibration frequency.

[0053] Among them, temperature is a key factor affecting the respiration of vegetables and microbial activity, which can be measured by temperature sensors. The average temperature reflects the overall thermal environment throughout the delivery process. Long-term exposure to high average temperature can enhance the respiration of vegetables, accelerating the consumption of sugar and water, and leading to a decrease in freshness. The maximum temperature reflects the extreme high-temperature situation during the delivery process. Short-term high temperature may trigger a surge in local enzyme activity, leading to browning or rotting of vegetables.

[0054] Humidity affects the transpiration and water balance of vegetables, which determines their freshness, and can be measured by humidity sensors. Low average humidity can cause continuous water loss in vegetables, resulting in wilting leaves and shriveled fruits. The minimum humidity reflects the extreme dryness during the delivery process. Short-term low humidity may cause the vegetable skin to crack, directly affecting users' evaluation of the appearance and freshness of the vegetables.

[0055] Vibration can cause mechanical damage to the physical structure of vegetables, which can be measured by vibration sensors. The average vibration frequency reflects the intensity of continuous jolting during the delivery process. Long-term high-frequency vibration may cause vegetable cells to rupture and juice to flow out. The maximum vibration frequency corresponds to extreme jolting conditions that may cause scratches on the vegetable skin or internal tissue damage, accelerating oxidation and spoilage of the vegetables, leading to damage and deterioration when received by users, and thus reducing the quality score.

[0056] The machine learning algorithm takes into account that the prediction model involves multiple input variables, such as maturity, delivery time, and delivery process environmental parameters, and that there may be complex nonlinear relationships between each input variable and the predicted quality score of the received vegetables. Neural network algorithms have strong advantages in handling nonlinear relationships and multivariate problems, so this algorithm is chosen. For example, the neural network algorithm uses a backpropagation neural network algorithm that adjusts network weights through backpropagation, effectively improving prediction accuracy.

[0057] The way to test and optimize the initial prediction model parameters through the test set is to use prediction accuracy and mean absolute error as test indicators. The prediction accuracy is the proportion of samples with consistent predicted and received vegetable quality scores to the total number of samples. The mean absolute error is the average absolute difference between the predicted and received vegetable quality scores.

[0058] When the prediction accuracy reaches the set prediction accuracy and the mean absolute error is less than the set absolute difference, the model parameter optimization stops. Otherwise, adjust the number of hidden layer nodes of the neural network, retrain the model and test until the above criteria are met.

[0059] The application constructs a maturity-delivery time-delivery environment-expected received vegetable quality score prediction model based on historical delivery data, realizes accurate prediction of vegetable delivery quality according to delivery area time and environment parameters, so as to optimize delivery allocation strategy and improve user satisfaction with vegetable freshness.

[0060] S4, according to the expected delivery time of the to-be-delivered area and the predicted delivery process environment parameters, combining the predicted model output of the expected received vegetable quality score of different maturity categories of vegetables, the delivery allocation of different maturity categories of vegetables is carried out.

[0061] It should be noted that the acquisition method of the expected delivery time of the to-be-delivered area and the predicted delivery process environment parameters is: obtaining the geographic position of the to-be-delivered area, planning the delivery path of the vegetable base and the to-be-delivered area based on high-precision map, determining the delivery transportation basic time based on the effective time consumption of the delivery tool in the historical same delivery path.

[0062] The delivery transportation basic time is corrected based on the traffic condition data on the delivery path to obtain the expected delivery time of the to-be-delivered area.

[0063] According to the geographic position of the area along the delivery path between the vegetable base and the to-be-delivered area, the environment parameters matched with the arrival time and the geographic position of the area are obtained from the regional meteorological station, and the obtained environment parameters are sorted by time to generate a predicted delivery process environment parameter sequence, and each environment parameter in the predicted delivery process environment parameter sequence is obtained.

[0064] In a specific embodiment, the delivery transportation basic time is the average value of the effective time consumption of the delivery tool in each delivery transportation corresponding to the historical same delivery path. Wherein the effective time consumption is the abnormal time consumption excluding traffic accidents.

[0065] The acquisition method of the expected delivery time of the to-be-delivered area is: importing the delivery path of the vegetable base and the to-be-delivered area into the existing map software to obtain the expected unblocking time length of each traffic congestion section on the delivery path, superimposing the sum of the expected unblocking time length of each traffic congestion section and the delivery transportation basic time to obtain the expected delivery time of the to-be-delivered area.

[0066] It should be noted that the delivery allocation of different maturity categories of vegetables is specifically: determining the reference maturity of different maturity categories of vegetables, substituting it into the predicted model with the expected delivery time of the to-be-delivered area and each parameter in the predicted delivery process environment parameter sequence, and outputting the expected received vegetable quality score of different maturity categories of vegetables.

[0067] The maturity categories with the expected received vegetable quality score higher than the set quality score threshold are screened out.

[0068] According to the order demand quantity of the to-be-delivered region and the inventory quantity of the screened vegetables of different maturity categories, priority allocation rules are used for delivery allocation.

[0069] In a specific embodiment, the way of determining the reference maturity of the vegetables of different maturity categories is to screen the minimum maturity from the maturity of all the vegetables of different maturity categories as the reference maturity.

[0070] It should be noted that the priority allocation rules are set in such a way that if the inventory quantity of a certain maturity category of vegetables screened is greater than the order demand quantity, the vegetables of this maturity category are set as the first priority.

[0071] If the inventory quantity of a single maturity category of vegetables does not meet the order demand quantity, it is determined whether the total inventory quantity of the combined multiple categories meets the order demand quantity, and the combined categories that meet the order demand quantity are sorted according to the average expected vegetable quality score, and the combined category with the highest average expected vegetable quality score is selected as the second priority.

[0072] If the expected received vegetable quality score of all maturity categories or combined categories is lower than the set quality score threshold, the early warning mechanism is triggered.

[0073] In a specific embodiment, the multiple combined categories are randomly combined with the screened different maturity categories, it is determined whether the total inventory quantity of the randomly combined categories meets the order demand quantity, the expected received vegetable quality scores of all maturity categories of all combined categories that meet the order demand quantity are screened, and the average expected received vegetable quality score is calculated by mean value.

[0074] According to the prediction model, the expected received vegetable quality scores of different maturity categories of vegetables are output, and the order demand quantity of the to-be-delivered region and the inventory quantity of the vegetables of different maturity categories are combined for priority delivery allocation. By dynamically matching different maturity vegetables with delivery conditions, optimal utilization of vegetable resources is achieved, thereby ensuring that high-score quality vegetables meet the order demand first and reducing vegetable loss caused by improper delivery.

[0075] S5, feedback information of users receiving vegetables is collected through an online feedback questionnaire, the accuracy of the prediction model is analyzed based on the feedback information, and the prediction model is iteratively optimized according to the accuracy of the prediction model.

[0076] As shown in Figure 3 the prediction model is iteratively optimized according to the accuracy of the prediction model, specifically, after the users receive the vegetables, a feedback questionnaire is pushed through an online order platform, and the actual vegetable quality score of the users receiving the vegetables is extracted from the feedback questionnaire.

[0077] The deviation value of the actual vegetable quality score of the user after receiving the vegetable from the corresponding predicted vegetable quality score is calculated, and the accuracy of the prediction model is determined based on the deviation value and a set deviation range threshold.

[0078] If the accuracy is lower than a preset accuracy standard, the deviation source is determined according to the actual delivery time of the to-be-delivered area and the actual delivery process environment parameter sequence, the output feature of the prediction model is adjusted based on the deviation source, and the prediction model is retrained to complete iteration.

[0079] In a specific embodiment, the accuracy determination method of the prediction model is that if the deviation value of the actual vegetable quality score of the user after receiving the vegetable from the corresponding predicted vegetable quality score is lower than a set deviation range threshold, it indicates that the quality score prediction of the vegetable is accurate, the number of vegetables with accurate quality score prediction among the vegetables received by the user is counted, and the ratio of the number to the total number of vegetables received by the user is taken as the accuracy of the prediction model.

[0080] It should be noted that the deviation source determination content is specifically that the actual delivery time of the to-be-delivered area is analyzed for deviation from the predicted delivery time thereof to obtain a delivery time deviation degree, and if the delivery time deviation degree is greater than a set time deviation degree threshold, the deviation source is a delivery time estimation deviation.

[0081] The actual delivery process environment parameter sequence of the to-be-delivered area is analyzed for similarity from the predicted delivery process environment parameter sequence thereof to obtain a delivery process environment parameter sequence similarity degree, and if the delivery process environment parameter sequence similarity degree is less than a set similarity threshold, the deviation source is a delivery environment estimation deviation. The similarity analysis can be a cosine similarity analysis method.

[0082] When the deviation source is not a delivery time estimation deviation or a delivery environment estimation deviation, it is determined that the deviation source is a vegetable category data deviation.

[0083] In a specific embodiment, the delivery time deviation degree is obtained by obtaining the difference between the actual delivery time and the predicted delivery time thereof, and taking the ratio of the difference to the predicted delivery time as the delivery time deviation degree.

[0084] The output feature of the prediction model is adjusted based on the deviation source, and the specific method is that if the deviation source is a vegetable category data deviation, the historical data sample amount of the category is increased, and the prediction model of the category is retrained.

[0085] If the deviation source is a delivery environment estimation deviation, a dynamic delivery environment correction is added to the model to strengthen the prediction accuracy of the delivery environment.

[0086] If the deviation source is the delivery time estimation deviation, the historical time period traffic data of the delivery path is accessed to optimize the delivery time estimation, and the input accuracy of the time parameter is improved.

[0087] The application collects feedback information of users receiving vegetables, analyzes the accuracy of the prediction model based on the feedback information, iteratively optimizes the prediction model according to the accuracy of the prediction model, and continuously corrects the model parameters to match the actual delivery scene through the closed-loop feedback mechanism, thereby improving the prediction accuracy of the model and the adaptability of the system, and ensuring the stability of the quality of vegetables in the long-term delivery process.

[0088] The above formulas are dimensionless values, and the formulas are obtained by collecting a large amount of data to simulate the latest real situation, and the preset parameters in the formulas are set by a person skilled in the art according to the actual situation.

[0089] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0090] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solutions. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0091] In addition, the functional modules in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0092] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0093] Finally, the above is only a preferred embodiment of the present application and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for intelligent analysis and distribution of vegetables directly supplied from vegetable bases, characterized in that, include: The maturity correlation features of harvested vegetables were obtained, and the maturity was obtained by analyzing the maturity correlation features using a maturity assessment algorithm. Vegetables harvested are classified based on a preset maturity range, and vegetables with the same maturity range are grouped into the same maturity category. A predictive model is constructed based on historical delivery data, which includes maturity, delivery time, delivery environment, and expected quality score of the vegetables received. The predictive model takes vegetable maturity, delivery time, and delivery process environment parameters as inputs and the expected quality score of the vegetables received as output. Based on the estimated delivery time and predicted environmental parameters of the delivery process in the area to be delivered, and combined with the prediction model, the expected quality score of vegetables received for different maturity categories is output, and vegetables of different maturity categories are distributed accordingly. Feedback from users after receiving vegetables was collected through online questionnaires. The accuracy of the prediction model was analyzed based on the feedback information, and the prediction model was iteratively optimized based on the accuracy of the prediction model.

2. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 1, characterized in that: The acquisition of maturity-related features of harvested vegetables specifically includes: The appearance and internal composition data of harvested vegetables are obtained using image acquisition and spectral detection equipment. The appearance data includes color, shape index and hardness, and the internal composition data includes sugar content and acidity. The acquired appearance feature data and internal component data are preprocessed to remove outlier data and undergo standardized transformation to obtain the preprocessed maturity correlation features. Based on the preprocessed maturity correlation features, the maturity of vegetables after harvesting is calculated using a set maturity assessment algorithm.

3. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 2, characterized in that, The maturity assessment algorithm is configured as follows: A large number of experimental vegetable samples were collected. Each vegetable sample included its color, shape index, firmness, sugar content, acidity, and corresponding true ripeness after pretreatment. A multiple linear regression equation between maturity-related characteristics and actual maturity was established using linear regression methods. The least squares method is used to estimate the constant term and regression coefficients of each feature in the multiple linear regression equation, so as to minimize the sum of squared residuals between the predicted maturity and the actual maturity of the multiple linear regression equation. The maturity correlation features of the harvested vegetables after pretreatment are input into the established multiple linear regression equation, and the maturity of the harvested vegetables is output.

4. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 1, characterized in that, The classification of harvested vegetables based on a preset maturity range is specifically as follows: Based on the type of vegetable and the preset optimal ripeness range, multiple ripeness grading ranges are determined; The maturity of harvested vegetables is matched with the maturity grading range of corresponding vegetable types, and vegetables with the same maturity grading range are classified into the same maturity category. The number of vegetables in each maturity category is counted. If the number of vegetables in a certain maturity category exceeds the preset proportion threshold of the total number of vegetables in that batch, the grading range boundary is readjusted and the vegetables are reclassified.

5. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 4, characterized in that, The prediction model for constructing maturity-delivery time-delivery environment-expected vegetable quality score is as follows: Collect historical delivery data from vegetable bases that are the same type of vegetables as those harvested. The historical delivery data includes the initial maturity of vegetables, actual delivery time, environmental parameters in the environmental parameter sequence of the delivery process, and the quality rating of the vegetables received by users in each historical delivery record, forming a historical delivery dataset. The historical delivery dataset is divided into a training set and a test set. The feature parameters of the vegetables received by users that are higher than the preset quality score threshold are retained from the training set as model input variables. Based on the selected model input variables and the corresponding quality ratings of the vegetables received by users, an initial prediction model is constructed using machine learning algorithms. The model parameters are then optimized by testing the initial prediction model on a test set, resulting in a prediction model for maturity, delivery time, delivery environment, and expected quality ratings of the vegetables received.

6. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 1, characterized in that, The estimated delivery time and predicted environmental parameters for the delivery process in the area to be delivered are obtained as follows: Obtain the geographical location of the area to be delivered, plan the delivery route between the vegetable base and the area to be delivered based on a high-precision map, and determine the basic delivery time based on the effective time of delivery tools on the same delivery route in history. Based on traffic data along the delivery route, the base delivery time is adjusted to obtain the estimated delivery time for the area to be delivered. Based on the geographical locations of the areas along the delivery route between the vegetable base and the delivery area, environmental parameters matching the geographical locations and arrival times of the areas are obtained from the local meteorological station. The obtained environmental parameters are sorted by time to generate a sequence of environmental parameters for the predicted delivery process, thus obtaining each environmental parameter in the sequence of environmental parameters for the predicted delivery process.

7. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 6, characterized in that, The distribution and allocation of vegetables at different maturity levels specifically involves: Determine the reference maturity level for different maturity categories of vegetables, and substitute it with the estimated delivery time of the delivery area and each environmental parameter in the predicted delivery process environmental parameter sequence into the prediction model to output the expected quality score of the vegetables received for different maturity categories. Select maturity categories of vegetables that are expected to receive quality scores higher than the set quality score threshold; Based on the order demand in the areas to be delivered and the inventory of vegetables of each maturity level selected, a priority allocation rule is used for delivery allocation.

8. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 7, characterized in that, The priority allocation rule is set as follows: If the inventory of vegetables of a certain maturity level exceeds the order demand, then vegetables of that maturity level are set as the first priority. If the inventory of vegetables of a single maturity category does not meet the order demand, determine whether the total inventory of multiple category combinations meets the order demand. For category combinations that meet the order demand, sort them by the average expected vegetable quality score and select the category combination with the highest average expected vegetable quality score as the second priority. If the expected vegetable quality scores for all maturity categories or combinations are lower than the set quality score threshold, an early warning mechanism will be triggered.

9. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 1, characterized in that, The iterative optimization of the prediction model based on its accuracy specifically involves: After the user receives the vegetables, a feedback questionnaire is sent through the online order platform, and the actual quality rating of the vegetables after the user receives the vegetables is extracted from the feedback questionnaire. Calculate the deviation between the actual vegetable quality score given by the user after receiving the vegetables and the corresponding expected vegetable quality score, and determine the accuracy of the prediction model based on the deviation value and a set deviation range threshold. If the accuracy rate is lower than the preset accuracy rate standard, the source of deviation is determined based on the actual delivery time and environmental parameter sequence of the actual delivery process in the area to be delivered. The output features of the prediction model are adjusted based on the source of deviation, and the prediction model is retrained to complete the iteration.

10. The intelligent analysis and distribution method for vegetables directly supplied from vegetable bases according to claim 9, characterized in that, The specific details of determining the source of the deviation are as follows: The actual delivery time of the area to be delivered is analyzed to determine the deviation from the estimated delivery time. If the deviation exceeds the set time deviation threshold, the source of the deviation is the estimated delivery time deviation. The actual delivery process environmental parameter sequence of the area to be delivered is compared with its predicted delivery process environmental parameter sequence to obtain the similarity of the delivery process environmental parameter sequence. If the similarity of the delivery process environmental parameter sequence is less than the set similarity threshold, the source of the deviation is the delivery environment prediction deviation. If the source of the deviation is not due to deviation in delivery time estimation or deviation in delivery environment estimation, then the source of the deviation is determined to be deviation in vegetable type data.

Images