CN117391599A

CN117391599A - Commodity sorting method based on big data

Info

Publication number: CN117391599A
Application number: CN202311344599.6A
Authority: CN
Inventors: 杨裕
Original assignee: Zhuhai City Polytechnic
Current assignee: Zhuhai City Polytechnic
Priority date: 2023-10-18
Filing date: 2023-10-18
Publication date: 2024-01-12

Abstract

The application provides a commodity sorting method based on big data, which comprises the following steps: according to the sorting requirements of the fresh combined commodities, the fresh commodities are transferred from an inventory area to a sorting area through automatic equipment, sorted according to the characteristics of the fresh commodities, and arranged according to the requirements of varieties, specifications, weights and numbers; sorting sub-commodities according to sorting requirements of all sub-commodities in the fresh combined commodities, and carrying out combined sorting on the sub-commodities; for the fresh combined commodity, if a certain fresh commodity is out of stock, updating the information of the fresh combined commodity, and judging whether to re-sort or replace the out-of-stock fresh commodity in the fresh combined commodity; when new fresh commodities are marketed, sub commodities in the combined commodity to be adjusted are judged so as to introduce new commodity combinations and create more sales opportunities; and determining the final sorting result of the fresh combined commodity through detection of the scanning equipment.

Description

Commodity sorting method based on big data

Technical Field

The invention relates to the technical field of information, in particular to a commodity sorting method based on big data.

Background

With the explosive growth of electronic commerce in recent years, especially the rise of fresh e-commerce platforms, consumers increasingly seek a high quality, diversified and fast-responding shopping experience. The characteristics of fresh goods, such as short shelf life, susceptibility to external environmental influences, etc., determine their high dependence on time and accuracy during sorting, storage and distribution. In addition, the variety and demand of fresh goods varies with seasons and market trends, which increases the complexity of sorting operations. In the past, the sorting of fresh goods relies on the manual work to go on, and not only inefficiency is makeed mistakes easily. For example, in large electronic commerce warehouses, both sorting and sorting of combined goods, quick response is required to ensure freshness of the goods. However, it is difficult to achieve this goal in a limited time with conventional methods. Modern consumers seek fresh and distinctive fresh goods. With the continuous marketing of new fresh goods, the warehouse needs to quickly adjust its storage and sorting strategies to meet the new needs of the market. This quick response is extremely difficult with conventional methods, but is critical to improving sales and meeting consumer needs. Unlike the sorting of single articles, the sorting of combined articles involves the combining and deployment of multiple articles. This means that when new fresh goods are presented on the market, it is not only necessary to consider how to properly store and sort these new goods, but also how to combine them with existing goods to create new sales opportunities. For example, when a new fruit and vegetable is presented on the market, how to combine it with other goods to form a new, attractive combined commodity becomes an important issue for the fresh electronic commerce. With advances in technology, large data has been widely used in a variety of industries. However, in the field of sorting fresh commodities, how to fully utilize big data technology to achieve higher sorting efficiency and accuracy is still a problem to be studied urgently. Through analysis of historical sales data, seasonal trends, and consumer preferences, big data may provide powerful support for the storage, sorting, and combining of fresh goods. Therefore, in order to meet the demands of modern consumers, improve the sorting efficiency of fresh commodities, and timely adjust and innovate when new fresh commodities are marketed, a new fresh commodity sorting method based on big data is needed.

Disclosure of Invention

The invention provides a commodity sorting method based on big data, which mainly comprises the following steps:

according to the sorting requirements of the fresh combined commodities, the fresh commodities are transferred from an inventory area to a sorting area through automatic equipment, sorted according to the characteristics of the fresh commodities, and arranged according to the requirements of varieties, specifications, weights and numbers; sorting sub-commodities according to sorting requirements of all sub-commodities in the fresh combined commodities, and carrying out combined sorting on the sub-commodities; for the fresh combined commodity, if a certain fresh commodity is out of stock, updating the information of the fresh combined commodity, and judging whether to re-sort or replace the out-of-stock fresh commodity in the fresh combined commodity; when new fresh commodities are marketed, sub commodities in the combined commodity to be adjusted are judged so as to introduce new commodity combinations and create more sales opportunities; acquiring the demand and the stock of the commodity through a data analysis and prediction technology according to seasonal variation and stock fluctuation factors of the fresh combined commodity, and adjusting production and purchasing plans according to analysis results so as to meet the demands of different seasons and stock; and determining the final sorting result of the fresh combined commodity through detection of the scanning equipment.

Further, according to the sorting requirement of the fresh combined commodity, the fresh commodity is transferred from the stock area to the sorting area through the automatic equipment, and is sorted according to the characteristics of the fresh commodity, and is arranged according to the requirements of varieties, specifications, weights and numbers, and the method comprises the following steps:

acquiring sorting requirements of fresh combined commodities, including fresh varieties, specifications, weight and quantity; acquiring data of a logistics management system, wherein the data comprise storage positions, inventory states, supplier information and warehouse-in dates of each sub commodity; using automatic equipment, including an unmanned carrier and an automatic sorting system, and moving fresh goods from an inventory area to a sorting area according to sorting demand labels of sub-goods and the positions of the inventory area; sorting by using automatic equipment according to the characteristics and sorting requirements of fresh commodities; the method comprises the steps of using sensor equipment, including a weight sensor and a size sensor, to perform data detection on the sorted fresh goods and obtain specification, weight and quantity information; and judging whether the sorting requirement is met or not according to the detection data, and determining whether the sorting requirement is re-performed or adjusted.

Further, the sorting of the sub-commodities according to the sorting requirement of each sub-commodity in the required fresh combined commodity, and the combined sorting of the sub-commodities comprises:

Using a freshness testing instrument to judge the freshness of fresh goods by measuring the conductivity of the fresh goods; acquiring an image of a fresh commodity through a camera, training a convolutional neural network by using a fresh commodity picture marked with the degree of ripeness, and inputting a new commodity into a model to judge the degree of ripeness of the commodity; judging the quality of the fresh goods according to the freshness and the maturity of the fresh goods, and marking the quality of the fresh goods by using an automatic labeling machine; the method comprises the steps of using a size sensor to obtain the size of fresh commodities, and performing size grouping on the commodities according to a size grouping standard; and carrying out combined sorting according to the types, the numbers, the qualities and the sizes of the sub-commodities.

Further, if a certain fresh article is out of stock, the information of the fresh combined article is updated, and whether to re-sort or replace the out-of-stock fresh article in the fresh combined article is judged, including:

acquiring inventory data of all sub-commodities according to information of fresh combined commodities, wherein the data comprises names of the commodities, inventory quantity and positions in a warehouse; comparing the inventory data of each sub-commodity with the required quantity of the sub-commodity in the combined commodity by adopting a numerical comparison mode to obtain a sub-commodity list with sufficient inventory and insufficient inventory; acquiring a sub-commodity list with insufficient inventory, searching for alternative commodities with similar attributes in a commodity database through data matching, and obtaining an alternative commodity list of each commodity with insufficient inventory; through analysis of the candidate commodity list, the optimal candidate commodity is selected according to the attribute similarity, the price and the stock quantity of the commodity, and a sub commodity replacement scheme is obtained; according to the sub commodity replacement scheme, the composition of the fresh combined commodity is updated, meanwhile, the inventory is deducted, and updated fresh combined commodity information and inventory information are obtained; according to the updated information of the fresh combined commodity, using automatic equipment to re-sort the fresh combined commodity; acquiring sorted fresh combined commodities, and then performing quality inspection and quantity verification; further comprises: determining whether to replace according to the similarity between the replaceable fresh commodity attribute and the fresh commodity attribute in the absence of the commodity; and determining other sub-commodities needing to be replaced according to the replaced fresh combined commodity information.

The method for determining whether to replace according to the similarity between the replaceable fresh commodity attribute and the fresh commodity attribute in the absence of the commodity specifically comprises the following steps:

and acquiring attribute information of the out-of-stock commodity according to the out-of-stock fresh commodity list. Euclidean distance between the replaceable fresh and out-of-stock items is calculated as attribute similarity using the euclidean_distances function in the Scikit-learn library. And judging whether the attribute of the replaceable commodity is similar to the attribute of the absent commodity by setting a similarity comparison rule, and determining whether the absent commodity is replaced according to a judging result to acquire a replaceable fresh commodity list. The attributes of the replaceable commodity are matched with the attributes of the backorder commodity using the data processing tool Pandas, with its data matching and screening functions. And obtaining a replaceable commodity list similar to the attribute of the backorder commodity through the matching result, and determining the final replaceable fresh commodity. And replacing the stock-out commodity with a replaceable commodity to obtain the information of the replaced fresh combined commodity, including the specification, the weight and the quantity.

The method for determining other sub-commodities to be replaced according to the replaced fresh combined commodity information specifically comprises the following steps:

And acquiring attribute information of other sub-commodities in the fresh combined commodity. And matching the attribute of the other sub-commodities with the updated commodity attribute according to the replaced fresh combined commodity information, judging whether the other sub-commodities need to be replaced, and determining the other sub-commodities needing to be replaced. And acquiring a list of alternative other sub-commodities, and matching the attribute of the alternative commodity with the attribute of the sub-commodity to be replaced by adopting an attribute matching algorithm to determine the final alternative other sub-commodities. And updating other sub-commodities needing to be replaced into replaceable commodities, and determining the attribute of the updated fresh combined commodity, including the specification, the weight and the quantity, to obtain updated fresh combined commodity information. And judging whether the updated fresh combined commodity meets the expected requirement by adopting a data statistics method. And determining whether to further adjust according to the judging result. And determining the final fresh combined commodity according to the updated fresh combined commodity information and the verification result.

Further, when a new fresh commodity is marketed, determining sub-commodities in the combined commodity to be adjusted to introduce a new commodity combination and create more sales opportunities, including:

Acquiring fresh commodity information on new market according to a commodity inventory database, wherein the fresh commodity information comprises commodity names, categories, selling prices, supplier information and warehouse-in dates; analyzing all the attributes of the obtained fresh goods by using a collaborative filtering algorithm, comparing the attributes with the attributes of the existing fresh combined goods, and judging whether the fresh goods are compatible with the existing fresh combined goods; if the new commodity is compatible with the existing combined commodity, updating the information of the combined commodity; analyzing the updated estimated sales and profit margins of the combined commodity by using a linear regression algorithm, and evaluating the estimated sales effect of the new combined commodity; sorting the new combined commodity according to the analysis result if the expected sales effect of the new combined commodity reaches the expected value; if the expectation is not reached, adjusting the sub-commodity of the combined commodity; according to the determined new combined commodity, carrying out actual sorting on the new combined commodity, and carrying out quality inspection and quantity verification; the sales condition of the new combined commodity is monitored and counted in real time, wherein the sales condition comprises sales quantity and sales amount; comparing the expected sales amount with the actual sales amount according to the sales condition, and evaluating the sales effect; if the sales effect reaches the expected value, continuing sales; if the sales effect does not reach the expected value, reselecting the new commodity; according to the new commodity list with sales potential, the allocation relation between the new commodity and the existing commodity is determined by using a combination optimization algorithm, and a new fresh combination commodity list is output; comparing the new fresh combined commodity list with the existing fresh combined commodity list, utilizing difference analysis to find out the combined commodity to be regulated, and outputting the combined commodity list to be regulated; further comprises: based on a collaborative filtering algorithm, evaluating the compatibility of the fresh goods and the existing fresh combined goods; determining whether the new commodity can be combined with certain sub-commodities by evaluating the association and competitiveness between the new commodity and the sub-commodity in the existing combined commodity; and determining whether the sub-commodity can be effectively paired with the sub-commodity of the existing combined commodity according to the supply chain and inventory management condition of the new commodity, and ensuring effective management of supply and inventory.

Based on collaborative filtering algorithm, evaluate the compatibility of fresh goods of new generation and current fresh combination commodity, specifically include:

and obtaining scoring data of the user on the existing fresh combined commodity according to the historical purchasing record of the user. And calculating the similarity between the users by a cosine similarity calculation method. And selecting a certain number of users with similar purchasing behavior with the target user as neighbors according to the similarity calculation result. And determining attributes related to the newly-born fresh commodity according to the purchase records of the neighbor users, wherein the attributes comprise commodity category, price interval, production place, freshness and quality. And evaluating the compatibility of the fresh goods and the existing fresh combined goods by analyzing the scores of the neighbor users on the fresh goods and the scores of the existing fresh combined goods.

The method for determining whether the new commodity can be combined with some sub-commodities by evaluating the relevance and the competitiveness between the new commodity and the sub-commodity in the existing combined commodity comprises the following steps:

and acquiring detailed information of the new fresh combined commodity through an enterprise database, wherein the detailed information comprises commodity characteristics, expected selling prices and target consumer groups. Sub-commodity information in the new fresh combined commodity is obtained, wherein the sub-commodity information comprises sales data, cost and customer feedback of each sub-commodity. And obtaining cosine similarity of the new commodity and each sub-commodity in commodity characteristics, expected selling price and target consumer groups by utilizing a Scikit-learn library of a data analysis tool Python, and obtaining a relevance index. And (3) factor analysis of the SPSS is performed by using a statistical analysis tool to obtain the competitive power index of the new commodity and each sub-commodity in sales data, cost and customer feedback. And according to the obtained competitive power index, setting a threshold value, judging which sub-commodities and the new commodity have the competitive power index lower than the set threshold value, and ensuring that the competition between the new commodity and the sub-commodities is low. And screening sub-commodities with the relevance index higher than a first preset threshold and the competitiveness lower than a second preset threshold according to the obtained sub-commodity list with low competitiveness, and taking the sub-commodities with the relevance index higher than the first preset threshold as candidate sub-commodities possibly combined with the new commodity. And predicting sales and customer satisfaction of the new commodity combined with the candidate sub-commodity by using a data analysis tool Tableau and combining a Monte Carlo simulation method, wherein parameters of simulation experiments comprise commodity price, market demand and competitiveness. If the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list.

The method for determining whether the sub-commodity can be effectively paired with the sub-commodity of the existing combined commodity according to the supply chain and the inventory management condition of the new commodity, and ensuring the effective management of supply and inventory comprises the following steps:

and acquiring supply chain information of the new commodity through a commodity data acquisition system, wherein the supply chain information comprises production date, quality guarantee period, transportation time and manufacturer. And acquiring the supply chain information of the sub-commodity of the existing combined commodity according to the supply chain information of the new commodity. And judging the similarity of the new commodity and the sub commodities in the supply chain management by adopting a K-means based clustering algorithm according to the production date, the quality guarantee period, the transportation time and the manufacturer data. And acquiring inventory management information of the new commodity, including inventory, sales period and sales volume, through an inventory data acquisition system. And acquiring inventory management information of the sub-commodity of the existing combined commodity according to the inventory management information of the new commodity. And judging the similarity of the new commodity and the sub commodities in inventory management according to the inventory, the sales period and the sales data by adopting a linear regression algorithm. And setting a threshold according to the obtained similarity result, and judging which sub-commodities have high supply chain and inventory similarity with the new commodity if the similarity is larger than a preset threshold, wherein the sub-commodities are candidate commodities possibly combined with the new commodity, so as to obtain a candidate commodity list. And (3) adopting a Monte Carlo simulation experiment to predict the sales and customer satisfaction degree which can occur by simulating the sales condition of the new commodity and the candidate commodity combination. Parameters of the simulation experiment include commodity price, market demand and competitiveness. If the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list.

Further, according to the seasonal variation and stock quantity fluctuation factors of the fresh combined commodity, the demand quantity and stock quantity of the commodity are obtained through a data analysis and prediction technology, and the production and purchase plans are adjusted according to the analysis result so as to meet the demands of different seasons and stock quantities, including:

extracting sales volume and sales date of the commodity according to the history sales record of the fresh combined commodity; obtaining seasonal sales trends of the commodities according to the extracted sales data by a time sequence analysis method; acquiring the current stock quantity of the fresh combined commodity according to the commodity number or commodity name by adopting the inquiry function of the stock management system; setting an inventory safety threshold by comparing seasonal sales trends with current inventory data, and if the inventory quantity is lower than or higher than the threshold, determining that the conditions of insufficient inventory or excessive inventory exist; according to inventory adjustment requirements, a time sequence prediction model is applied to predict commodity demand in a period of time in the future, and demand prediction data are obtained; using a production and purchase module in the ERP system to obtain the adjustment direction of the production and purchase plan and generating an adjusted production and purchase plan; according to the adjusted production and purchasing plans, producing and purchasing corresponding fresh commodities; according to the new fresh commodity inventory, inputting new inventory data through an updating function of an inventory management system, and updating the inventory information of the fresh combined commodity; and comparing the updated inventory data with the demand forecast data, if the inventory data meets or exceeds the demand forecast data, determining that the commodity demand in a future period is met, outputting a verification result meeting the demand, and otherwise, readjusting the production and purchase plans.

Further, the detecting by the scanning device, determining the final sorting result of the fresh combined commodity comprises:

placing fresh combined commodities in a detection area of scanning equipment according to a preset placing mode by an operator, and starting a preset scanning program through a scanning equipment interface; scanning fresh combined commodities to be detected through a sensor and a camera in a detection area, determining the colors and textures of the commodities, and judging whether the commodities have spoilage conditions or not; adopting a convolutional neural network to classify scanning images and identifying the types and the quantity of commodities in the images; comparing the recognized commodity types and quantity results according to preset combined commodity standards to verify; weight sensing equipment is adopted to detect the weight of the fresh combined commodity; judging whether the types and the amounts of the fresh combined commodities are matched with the weights according to a reference manual of the preset weights and the types and the amounts of the commodities; comparing the detection result of the weight sensing device with the output of the image recognition algorithm, and if the weight detection result is consistent with the type and the number of the commodities obtained by the image recognition, confirming the information of the fresh combined commodities, including the type, the number and the weight of the commodities; if the combination commodities are inconsistent, the fresh combination commodities are sorted again, and scanning and detection are carried out again; and generating a sorting report of the fresh combined commodity according to the confirmed result.

The technical scheme provided by the embodiment of the invention can have the following beneficial effects:

the invention discloses a commodity sorting method based on big data. The method transfers the commodities from the stock area to the sorting area according to the characteristics of fresh commodities, and arranges the commodities according to the requirements of varieties, specifications, weights and quantity. And sorting the sub-commodities according to the sorting requirements of the required fresh combined commodities, and carrying out combined sorting on the sub-commodities. If a certain fresh commodity is out of stock, updating the information of the fresh combined commodity, and judging whether to re-sort or replace the out-of-stock commodity. When new fresh commodities are marketed, sub-commodities in the combined commodity are adjusted, new commodity combinations are introduced, and more sales opportunities are created. And acquiring the demand and the stock of the commodity by a data analysis and prediction technology, adjusting the production and purchase plans, and meeting the demands of different seasons and stock. Finally, the final sorting result of the fresh combined commodity is determined through detection of the scanning equipment. The method comprehensively utilizes a plurality of technologies, realizes the efficient sorting of fresh combined commodities, and improves the operation efficiency and accuracy.

Drawings

Fig. 1 is a flow chart of a commodity sorting method based on big data according to the present invention.

Fig. 2 is a schematic diagram of a commodity sorting method based on big data according to the present invention.

Fig. 3 is a further schematic diagram of a method for sorting goods based on big data according to the present invention.

Detailed Description

The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The commodity sorting method based on big data in the embodiment specifically comprises the following steps:

s101, transferring fresh commodities from an inventory area to a sorting area through automatic equipment according to sorting requirements of fresh combined commodities, sorting according to characteristics of the fresh commodities, and arranging according to requirements of varieties, specifications, weights and numbers.

The sorting requirements of the fresh combined commodity are obtained, wherein the sorting requirements comprise fresh varieties, specifications, weight and quantity. And acquiring data of the logistics management system, wherein the data comprise storage positions, stock states, supplier information and warehouse-in dates of each sub commodity. And the automatic equipment comprises an unmanned carrier and an automatic sorting system, and fresh commodities are moved from the stock area to the sorting area according to the sorting demand labels of the sub-commodities and the positions of the stock area. And sorting by using automatic equipment according to the characteristics and sorting requirements of fresh commodities. The sensor equipment comprises a weight sensor and a size sensor, and the data detection is carried out on the fresh goods after the sorting is finished to obtain the specification, weight and quantity information. And judging whether the sorting requirement is met or not according to the detection data, and determining whether the sorting requirement is re-performed or adjusted. For example, according to the sorting requirements of the fresh combined commodity, it is assumed that one fresh combined commodity includes apples and oranges, and the sorting requirements are labeled as apples of variety, large size, 500 g in weight and 10 in number. According to the data of the logistics management system, if the apples are stored in the A stock position, the oranges are stored in the B stock position, the stock state is sufficient, the supplier information is supplier A, and the stock date is 2021 year, 1 month and 1 day. According to the equipment movement scheme, the automated guided vehicle will take 10 apples and 10 oranges from the bin a of the stock area and move them to the sorting area, respectively. And an automatic sorting system is used for sorting according to the sorting demand labels of fresh commodities, and 10 large apples and 10 oranges are respectively placed in corresponding sorting areas. The data detection is performed on the sorted fresh goods using a weight sensor and a size sensor. The average weight of apples was measured to be 510 g and the average weight of oranges was measured to be 520 g. The apples are re-sorted as they weigh more than the prescribed 500 grams. The weight of the orange meets the requirements. And (5) sorting the apples again according to the situation, and putting the apples with the weight meeting the requirements into a sorting area.

S102, sorting the sub-commodities according to sorting requirements of all the sub-commodities in the required fresh combined commodities, and carrying out combined sorting on the sub-commodities.

And judging the freshness of the fresh commodity by measuring the conductivity of the fresh commodity by using a freshness testing instrument. And acquiring an image of the fresh commodity by a camera, training a convolutional neural network by using the fresh commodity picture marked with the maturity, and inputting the new commodity into a model to judge the maturity of the commodity. And judging the quality of the fresh goods according to the freshness and the maturity of the fresh goods, and marking the quality of the fresh goods by using an automatic labeling machine. And obtaining the size of the fresh commodity by using a size sensor, and grouping the commodity by using a size grouping standard of the set size. And carrying out combined sorting according to the types, the numbers, the qualities and the sizes of the sub-commodities. For example, the conductivity of a batch of apples was measured using a freshness tester to give apple 1 with a conductivity of 80 μS/cm, apple 2 with a conductivity of 85 μS/cm, and apple 3 with a conductivity of 90 μS/cm. And (3) inputting images of apples into a model for judging the maturity according to the previously trained convolutional neural network model, so that apple 1 is 80% ripe, apple 2 is 70% ripe, and apple 3 is 90% ripe. According to the freshness and the ripeness of the fresh commodity, quality judgment can be performed, the freshness and the ripeness are respectively evaluated according to the range of 0-100, and the quality of apple 1 is calculated to be 80×80% =64, the quality of apple 2 is calculated to be 85×70% =59.5, and the quality of apple 3 is calculated to be 90×90% =81. The quality of fresh goods is marked by using an automatic labeling machine, and apples 1 and 2 are marked as 'general' and 'poor', and apples 3 are marked as 'good'. If the size of the batch of apples is measured by using a size sensor, apple 1 with a size of 5cm, apple 2 with a size of 6cm and apple 3 with a size of 7cm are obtained. According to the set size grouping standard, the apples in the group can be divided into small groups smaller than 5cm and divided into 5-7cm and divided into three large size groups larger than 7cm. Apple 1 belongs to the small-size group, apple 2 belongs to the medium-size group, and apple 3 belongs to the large-size group. According to the types, the numbers, the qualities and the sizes of apples, combination sorting can be performed, apples with good quality and large size can be placed in a high-end gift basket, apples with general quality and medium size can be placed in a general pin, and apples with poor quality and small size can be placed in a low-end gift basket.

And S103, if a certain fresh commodity is out of stock, updating information of the fresh combined commodity, and judging whether to re-sort or replace the fresh commodity out of stock in the fresh combined commodity.

And acquiring inventory data of all sub-commodities according to the information of the fresh combined commodity, wherein the data comprises the names of the commodities, the inventory and the positions in a warehouse. And comparing the inventory data of each sub-commodity with the required quantity of the sub-commodity in the combined commodity by adopting a numerical comparison mode to obtain a sub-commodity list with sufficient inventory and insufficient inventory. And acquiring a sub-commodity list with insufficient inventory, and searching for alternative commodities with similar attributes in a commodity database through data matching to obtain an alternative commodity list of each commodity with insufficient inventory. And through analysis of the candidate commodity list, selecting the optimal candidate commodity according to the attribute similarity, the price and the inventory of the commodity, and obtaining a sub commodity replacement scheme. And according to the sub-commodity replacement scheme, updating the composition of the fresh combined commodity, and simultaneously deducting the inventory to obtain updated fresh combined commodity information and inventory information. And (5) according to the updated information of the fresh combined commodity, using an automatic device to re-sort the fresh combined commodity. And acquiring the sorted fresh combined commodity, and then performing quality inspection and quantity verification. For example, there is a fruit splice consisting of apples, oranges and bananas, and it is now necessary to acquire inventory data for all sub-commodities and compare them to the required quantity. If the stock quantity of apples is 100, the stock quantity of oranges is 80, and the stock quantity of bananas is 60. Whereas fruit pieces require 1 apple, 2 oranges and 3 bananas. The results obtained by numerical comparison were adequate apples, 100 stock, 100 required amount, 80 stock, 200 required amount, 60 stock and 300 required amount. Then, according to the sub commodity list with insufficient stock, searching for an alternative commodity in a commodity database, wherein the alternative commodity is found in the database, namely, the alternative commodity 1 of orange is orange A, the price is 2 yuan, and the stock quantity is 100. The orange alternative commodity 2 is orange B, the price is 3 yuan, and the stock quantity is 150. Through analysis of the candidate commodity list, the optimal candidate commodity can be selected according to the attribute similarity, the price and the stock quantity of the commodity. An orange alternative commodity 1 was selected as an alternative. According to the alternative of the commodity, the fruit splice trays were updated to be composed of 1 apple, 2 oranges a and 3 bananas. Meanwhile, the inventory is deducted, the inventory of apples is reduced by 1, the inventory of oranges A is reduced by 2, and the inventory of bananas is reduced by 3. The updated fruit splice tray information and the inventory information are finally obtained, wherein the fruit splice tray comprises 1 apple, 2 oranges A and 3 bananas. The apples were 99 in stock, the oranges a were 98 in stock and the bananas were 57 in stock. And (5) according to the updated fruit splice tray information, using automatic equipment to re-sort fresh combined commodities, and if sorting is completed, obtaining 10 parts of fruit splice trays. Then, quality inspection and quantity check are carried out to ensure that the fruit splice tray meets the sales requirement. After inspection, 1 fruit piece was discarded if it had a problem in quality. Thus, 9 parts of fruit splice trays with qualified quality are finally obtained, and the sales requirement can be met.

And determining whether to replace according to the similarity between the replaceable fresh commodity attribute and the fresh commodity attribute in the absence of the commodity.

And acquiring attribute information of the out-of-stock commodity according to the out-of-stock fresh commodity list. Euclidean distance between the replaceable fresh and out-of-stock items is calculated as attribute similarity using the euclidean_distances function in the Scikit-learn library. And judging whether the attribute of the replaceable commodity is similar to the attribute of the absent commodity by setting a similarity comparison rule, and determining whether the absent commodity is replaced according to a judging result to acquire a replaceable fresh commodity list. The attributes of the replaceable commodity are matched with the attributes of the backorder commodity using the data processing tool Pandas, with its data matching and screening functions. And obtaining a replaceable commodity list similar to the attribute of the backorder commodity through the matching result, and determining the final replaceable fresh commodity. Replacing the stock-out commodity with a replaceable commodity to obtain replaced fresh combined commodity information, wherein the information comprises specification, weight and quantity; for example, the product that is now out of stock is red Fuji apple, and a product with similar properties needs to be found as a substitute. And obtaining attribute information of the red Fuji apples, wherein the red Fuji apples are red, sweet in taste and medium in size. According to the attribute information of the green apples, the golden apples and the purple apples, the green apples are green in color, sour in taste and medium in size, the golden apples are yellow in color, sweet in taste and large in size, and the purple apples are purple in color, sweet in taste and small in size. The attribute similarity of red Fuji apples to each apple was calculated using the euclidean_distances function in the Scikit-learn library. If the green apples are not similar in color and taste to the red Fuji apples, the similarity is low. A rule is set that requires at least two products of similar properties to be considered as replaceable products. The properties of red Fuji apples were matched to those of other apples using the Pandas library. Based on the previous similarity calculations, gold apples and red Fuji apples were similar in taste and size, so gold apples were selected as alternative commodities. Based on the results of the matching and screening, golden apple is the final alternative. The red Fuji apples in the original fresh combination are replaced by golden apples at present, and replaced fresh combination commodity information comprising specification, weight and quantity is obtained.

And determining other sub-commodities needing to be replaced according to the replaced fresh combined commodity information.

And acquiring attribute information of other sub-commodities in the fresh combined commodity. And matching the attribute of the other sub-commodities with the updated commodity attribute according to the replaced fresh combined commodity information, judging whether the other sub-commodities need to be replaced, and determining the other sub-commodities needing to be replaced. And acquiring a list of alternative other sub-commodities, and matching the attribute of the alternative commodity with the attribute of the sub-commodity to be replaced by adopting an attribute matching algorithm to determine the final alternative other sub-commodities. And updating other sub-commodities needing to be replaced into replaceable commodities, and determining the attribute of the updated fresh combined commodity, including the specification, the weight and the quantity, to obtain updated fresh combined commodity information. And judging whether the updated fresh combined commodity meets the expected requirement by adopting a data statistics method. And determining whether to further adjust according to the judging result. And determining the final fresh combined commodity according to the updated fresh combined commodity information and the verification result. For example, there is a fresh combination commodity including orange, apple and banana. Each sub-commodity has attribute information such as weight, place of origin, and color. Now a part of sub-commodity is replaced, and the original orange is changed into the grapefruit. It is desirable to determine whether to replace other sub-commodities and alternatively other sub-commodity lists by means of attribute matching according to the replaced fresh combined commodity information. If the replaced fresh combined commodity information is grapefruit, the weight is 200g, and the color is yellow in China at the production place; apple, 150g in weight, producing area in the united states, red in color; banana, weight 120g, brazil of origin, yellow color; an attribute matching algorithm is used to compare the attribute information of each sub-commodity with the updated commodity attribute information. Matching is performed according to three attributes of weight, place of production and color. Then the grapefruit is 200g in weight, the color yellow is matched with the original orange in China at the production place, and if the orange has the following properties: orange, weight 180g, place of origin china, orange color, can calculate the score of attribute matching, if weight difference 20g, place of origin matching score 100, color matching score 0. And according to the set matching rule and weight, the matching score of the grapefruit and the orange can be obtained. And then, comparing the attributes of the apples and the bananas one by one with the original commodity to match, and calculating to obtain a matching score. According to the matching result, whether other sub-commodities need to be replaced or not can be judged. If the match score threshold is set to 80, if the match score is below the threshold, then a replacement is deemed necessary. If the apple match score is 70, the banana match score is 90. And according to the set threshold value, determining that the apple needs to be replaced. Next, alternative other sub-merchandise listings need to be obtained. And matching the attribute of the replaceable commodity with the attribute of the sub commodity to be replaced through an attribute matching algorithm. If the alternative commodity is pear, the attribute is that the pear, 160g weight, the place of origin china, the colour yellow, calculate the matching score of pear and apple and be 90. And determining that the pear is an alternative other commodity according to the matching score. And finally, updating other sub-commodities needing to be replaced into replaceable commodities, and determining the attribute of the updated fresh combined commodity. After replacing apples with pears, obtaining updated fresh combined commodity information, namely grapefruit, weight of 200g, china in the producing area and yellow color; 160g pear, china in origin, yellow color; banana, weight 120g, brazil of origin, yellow color; and then, counting indexes such as total weight, average weight and the like of each sub-commodity, or evaluating the quality of the commodity by using an analysis algorithm to judge whether the updated fresh combined commodity meets the expected requirement. And determining whether further adjustment or optimization is needed according to the judgment result. If the updated fresh combined commodity does not meet the expected requirement, the updated fresh combined commodity can be adjusted or optimized according to the verification result, and other alternative sub-commodities are selected. And finally, determining the final fresh combined commodity according to the updated fresh combined commodity information and the verification result.

S104, judging sub-commodities in the combined commodity to be adjusted when new fresh commodities are marketed, so as to introduce new commodity combinations and create more sales opportunities.

And acquiring fresh commodity information on new markets according to the commodity inventory database, wherein the fresh commodity information comprises commodity names, categories, selling prices, supplier information and warehouse-in dates. And analyzing various attributes of the obtained fresh goods by using a collaborative filtering algorithm, comparing the attributes with the attributes of the existing fresh combined goods, and judging whether the fresh goods are compatible with the existing fresh combined goods. If the new merchandise is compatible with the existing combined merchandise, the information of the combined merchandise is updated. And analyzing the updated estimated sales and profit margins of the combined commodity by using a linear regression algorithm, and evaluating the estimated sales effect of the new combined commodity. And sorting the new combined commodity according to the analysis result if the expected sales effect of the new combined commodity reaches the expected value. If the expectation is not reached, the sub-commodity of the combined commodity is adjusted. And according to the determined new combined commodity, carrying out actual sorting on the new combined commodity, and carrying out quality inspection and quantity verification. The sales of the new combined commodity is monitored and counted in real time, including sales amount and sales amount. And comparing the expected sales and the actual sales according to sales conditions, and evaluating sales effects. If the sales effect reaches the expectations, sales continue. If the sales effect does not reach the expectations, the new merchandise is reselected. And according to the new commodity list with sales potential, the allocation relation between the new commodity and the existing commodity, a possible new combined commodity is determined by using a combination optimization algorithm, and a new fresh combined commodity list is output. And comparing the new fresh combined commodity list with the existing fresh combined commodity list, finding out the combined commodity to be adjusted by utilizing difference analysis, and outputting the combined commodity list to be adjusted. For example, a fresh supermarket obtains fresh commodity information of new market according to a commodity inventory database of the fresh supermarket, wherein the first commodity is named as apple, the class is fruit, the selling price is 5 yuan/jin, the supplier is supplier A, and the warehouse-in date is 2021-01-01; the trade name of the second commodity is fresh milk, the class is dairy products, the selling price is 10 yuan/bottle, the supplier is supplier B, and the warehouse-in date is 2021-01-05; the properties of these fresh goods are analyzed using collaborative filtering algorithms, such as evaluating their sales, popularity, and market competitiveness. These attributes are then compared to the attributes of the existing fresh combination merchandise to evaluate whether the fresh combination merchandise is compatible with the existing fresh combination merchandise. If the existing fresh combination commodity is a fruit basket, the fruit basket comprises apples, oranges and bananas. The apples newly marketed can be compared with apples in the existing combined commodity to see if their selling prices, supplier information, and date of warehousing match. If so, the information of the combined commodity, such as the provider information and the warehouse entry date of the apples, is updated. The updated estimated sales and profit margins of the combined commodity are analyzed by using a linear regression algorithm, the estimated sales of the new combined commodity is evaluated by calculating historical sales data and market trends, and if the historical data shows that the average sales of the fruit basket is 1000 yuan, the profit margin is 20%. The sales and profit margins of the updated combined commodity can be predicted by a linear regression algorithm, the sales are predicted to be 1200 yuan, and the profit margin is predicted to be 25%. And sorting the new combined commodity according to the analysis result if the expected sales effect of the new combined commodity reaches the expected value. If the expectation is not reached, the sub-commodity of the combined commodity is adjusted, the unpopular commodity is replaced or the selling price is adjusted. After the new combined commodity is determined, the new combined commodity is actually sorted, quality inspection and quantity verification are carried out, all commodities are ensured to meet the quality standard, and sorting is carried out according to the quantity of orders. And carrying out real-time monitoring and statistics on sales of the new combined commodity, including sales and sales. If the sales of the new combined commodity in the first month is 200 parts, the sales amount is 2400 yuan. And comparing the expected sales and the actual sales according to sales conditions, and evaluating sales effects. If the sales effect reaches the expectations, sales continue. If the sales effect does not reach the expectations, new goods are reselected and replaced with more popular fruits. And according to the new commodity list with sales potential, the allocation relation between the new commodity and the existing commodity, a possible new combined commodity is determined by using a combination optimization algorithm, and a new fresh combined commodity list is output. A new fresh combined commodity list comprising apples, fresh milk and oranges is determined through a combined optimization algorithm. And comparing the new fresh combined commodity list with the existing fresh combined commodity list, finding out the combined commodity to be adjusted by utilizing difference analysis, and outputting the combined commodity list to be adjusted. The sales of the oranges are found to be low by the variance analysis, and it is considered to adjust suppliers of the oranges or to increase sales promotion.

And evaluating the compatibility of the fresh goods and the existing fresh combined goods based on a collaborative filtering algorithm.

And obtaining scoring data of the user on the existing fresh combined commodity according to the historical purchasing record of the user. And calculating the similarity between the users by a cosine similarity calculation method. And selecting a certain number of users with similar purchasing behavior with the target user as neighbors according to the similarity calculation result. And determining attributes related to the newly-born fresh commodity according to the purchase records of the neighbor users, wherein the attributes comprise commodity category, price interval, production place, freshness and quality. And evaluating the compatibility of the fresh goods and the existing fresh combined goods by analyzing the scores of the neighbor users on the fresh goods and the scores of the existing fresh combined goods. For example, there are 5 users A, B, C, D and E and 3 fresh portfolios X, Y and Z. Firstly, scoring data is obtained according to historical purchasing records of users, wherein the score of a user A to a commodity X is 5, the score of the user A to a commodity Y is 3, and the score of the user A to a commodity Z is 4; the score of the user B on the commodity X is 4 points, and the score of the user B on the commodity Y is 2 points; the score of the user C on the commodity X is 5 points, and the score of the user C on the commodity Z is 3 points; the score of the user D on the commodity Y is 4 points, and the score of the user D on the commodity Z is 5 points; the user E scores 3 points for commodity X, 2 points for commodity Y and 1 point for commodity Z; next, the similarity between users is calculated using a cosine similarity calculation method. A certain number of users with similar purchase behavior are selected as neighbors, here two users B and C with similar purchase behavior as target user a are selected as neighbors. Then, determining the attribute related to the newly-born fresh commodity according to the purchase record of the neighbor user, wherein the category of the commodity X is vegetables, the price interval is 10-20 yuan, the production place is local, the freshness is fresh, and the quality is high; the category of the commodity Z is fruits, the price interval is 20-30 yuan, the place of origin is import, the freshness is fresh, and the quality is general; and finally, evaluating the compatibility of the fresh goods A with the existing fresh combined goods by analyzing the scores of the neighbor users on the fresh goods and the scores of the existing fresh combined goods, wherein the classification of the fresh goods A is vegetables, the price interval is 20-30 yuan, the production place is local, the freshness is fresh, the quality is high, and the evaluation result is that the compatibility of the fresh goods A with the existing fresh combined goods X is higher according to the scores of the neighbor users B and C on the goods X. And according to the scoring of the commodity Z by the neighbor user C, the evaluation result can be obtained that the compatibility of the newly-born fresh commodity A and the existing newly-born fresh combined commodity Z is lower.

By evaluating the relevance and competitiveness of new merchandise to sub-merchandise in an existing combined merchandise, it is determined whether new merchandise can be combined with certain sub-merchandise.

And acquiring detailed information of the new fresh combined commodity through an enterprise database, wherein the detailed information comprises commodity characteristics, expected selling prices and target consumer groups. Sub-commodity information in the new fresh combined commodity is obtained, wherein the sub-commodity information comprises sales data, cost and customer feedback of each sub-commodity. And obtaining cosine similarity of the new commodity and each sub-commodity in commodity characteristics, expected selling price and target consumer groups by utilizing a Scikit-learn library of a data analysis tool Python, and obtaining a relevance index. And (3) factor analysis of the SPSS is performed by using a statistical analysis tool to obtain the competitive power index of the new commodity and each sub-commodity in sales data, cost and customer feedback. And according to the obtained competitive power index, setting a threshold value, judging which sub-commodities and the new commodity have the competitive power index lower than the set threshold value, and ensuring that the competition between the new commodity and the sub-commodities is low. And screening sub-commodities with the relevance index higher than a first preset threshold and the competitiveness lower than a second preset threshold according to the obtained sub-commodity list with low competitiveness, and taking the sub-commodities with the relevance index higher than the first preset threshold as candidate sub-commodities possibly combined with the new commodity. And predicting sales and customer satisfaction of the new commodity combined with the candidate sub-commodity by using a data analysis tool Tableau and combining a Monte Carlo simulation method, wherein parameters of simulation experiments comprise commodity price, market demand and competitiveness. If the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list. For example, an enterprise is currently selling three fresh fruits: apple, pear and peach enterprises wish to push out a new combination package containing a new fresh commodity of citrus. In order to determine how to combine citrus with existing fruit, enterprises obtain detailed information of citrus from a database, which is characterized by high nutritional value, expected selling price of 2 yuan/jin, and target consumer group of young. Sales data, cost and customer feedback of apples, pears and peaches are obtained. And (3) calculating cosine similarity of citrus to apples, pears and peaches in commodity characteristics, expected selling prices and target consumer groups by using a Scikit-learn library of Python, wherein if the similarity to apples is 0.8, the similarity to pears is 0.7 and the similarity to peaches is 0.6. The SPSS was used to calculate the competitiveness of citrus and other three commodities in terms of sales data, cost, customer feedback, if 0.5 with apples, 0.4 with pears, and 0.6 with peaches. The threshold was set to 0.55 based on the competitiveness index, with only apples and pears having a competitiveness index below 0.55. In combination with the relevance index, both apples and pears are highly relevant to citrus, and therefore they are candidate combination products. Predicting sales and customer satisfaction of citrus and apple or pear combination by using Tableau combined with Monte Carlo simulation, wherein if simulation results show that the sales of citrus and apple combination is predicted to be 10000 yuan, the customer satisfaction is 80%; the sales of the citrus and pear combination was expected to be 9000 yuan with a customer satisfaction of 78%. Based on the simulation results, it was decided to sell a combination of citrus and apple, because the sales and customer satisfaction of such a combination were relatively high.

And determining whether the sub-commodity can be effectively paired with the sub-commodity of the existing combined commodity according to the supply chain and inventory management condition of the new commodity, and ensuring effective management of supply and inventory.

And acquiring supply chain information of the new commodity through a commodity data acquisition system, wherein the supply chain information comprises production date, quality guarantee period, transportation time and manufacturer. And acquiring the supply chain information of the sub-commodity of the existing combined commodity according to the supply chain information of the new commodity. And judging the similarity of the new commodity and the sub commodities in the supply chain management by adopting a K-means based clustering algorithm according to the production date, the quality guarantee period, the transportation time and the manufacturer data. And acquiring inventory management information of the new commodity, including inventory, sales period and sales volume, through an inventory data acquisition system. And acquiring inventory management information of the sub-commodity of the existing combined commodity according to the inventory management information of the new commodity. And judging the similarity of the new commodity and the sub commodities in inventory management according to the inventory, the sales period and the sales data by adopting a linear regression algorithm. And setting a threshold according to the obtained similarity result, and judging which sub-commodities have high supply chain and inventory similarity with the new commodity if the similarity is larger than a preset threshold, wherein the sub-commodities are candidate commodities possibly combined with the new commodity, so as to obtain a candidate commodity list. And (3) adopting a Monte Carlo simulation experiment to predict the sales and customer satisfaction degree which can occur by simulating the sales condition of the new commodity and the candidate commodity combination. Parameters of the simulation experiment include commodity price, market demand and competitiveness. If the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list. For example, the new commodity is milk, and supply chain information of the commodity can be obtained, wherein the supply chain information comprises that the production date is 2022, 1 month and 1 day, the quality guarantee period is 30 days, the transportation time is 3 days, and the manufacturer is manufacturer A. Then, according to the supply chain information of the new commodity, the supply chain information of the sub commodity of the yoghurt can be obtained by obtaining that the existing combined commodity is the yoghurt. The production date of the yoghurt is 2022, 1 month and 2 days, the shelf life is 20 days, the transportation time is 2 days, and the manufacturer is manufacturer B. The similarity in supply chain management of the new commodity and the sub-commodity of the yoghurt can be compared by a supply chain matching algorithm based on clustering. Next, by the stock data acquisition system, the stock quantity of the new commodity was 1000, the sales cycle was 30 days, and the sales quantity was 100. Inventory management information of the existing combined commodity yoghurt can be obtained. If the stock quantity of the yoghourt is 500, the sales period is 20 days, and the sales quantity is 50. And comparing the similarity of the new commodity and the sub commodity of the yoghurt in inventory management through an inventory matching algorithm based on linear regression. And according to the obtained matching result, obtaining a set similarity threshold value of 8, and if the supply chain similarity and the inventory similarity of the new commodity and the sub commodity of the yoghurt are both more than 8, the yoghurt is one of candidate commodities of the new commodity. Then, a Monte Carlo simulation experiment is adopted to obtain the new commodity with the price of 10 yuan, the market demand of 1000 pieces and the competitiveness of 8. The combined sales situation of the new commodity and the yoghourt can be simulated, and sales and customer satisfaction which can occur can be predicted. And finally, if the simulated sales and the customer satisfaction reach the preset standards, the sales exceeds 10000 yuan, and the customer satisfaction score exceeds 4 points, then the new commodity can be determined to be combined with the yoghurt, and a final combined commodity list is obtained.

S105, acquiring the demand and the inventory of the commodity through a data analysis and prediction technology according to seasonal variation and inventory fluctuation factors of the fresh combined commodity, and adjusting production and purchasing plans according to analysis results so as to meet the demands of different seasons and inventory.

And extracting the sales quantity and the sales date of the commodity according to the historical sales record of the fresh combined commodity. And obtaining seasonal sales trends of the commodities according to the extracted sales data by a time sequence analysis method. And acquiring the current stock quantity of the fresh combined commodity according to the commodity number or commodity name by adopting the inquiry function of the stock management system. By comparing seasonal sales trends with current inventory data, an inventory safety threshold is set and if the inventory is below or above the threshold, then an under-inventory or over-inventory condition is deemed to exist. And according to the inventory adjustment demand, a time sequence prediction model is applied to predict the commodity demand in a period of time in the future, so as to obtain demand prediction data. And using a production and purchase module in the ERP system to obtain the adjustment direction of the production and purchase plan and generating an adjusted production and purchase plan. And (5) producing and purchasing corresponding fresh commodities according to the adjusted production and purchasing plans. According to the new fresh commodity inventory, new inventory data is input through the updating function of the inventory management system, and the inventory information of the fresh combined commodity is updated. And comparing the updated inventory data with the demand forecast data, if the inventory data meets or exceeds the demand forecast data, determining that the commodity demand in a future period is met, outputting a verification result meeting the demand, and otherwise, readjusting the production and purchase plans. For example, sales data of a certain fresh combined commodity in the past year (sales date, sales volume), (2019-01-01, 50), (2019-01-02, 45), (2019-01-03, 60), (2019-12-29, 70), (2019-12-30, 65), (2019-12-31, 80) are extracted from the historical sales record of the fresh combined commodity, and seasonal sales trend analysis is performed on the sales data by a time series analysis method. If the analysis result shows that the sales amount of the fresh combined commodity is higher in 6-8 months in summer, the sales amount is lower in 12-2 months in winter, and the sales amounts in 3-5 months in spring and 9-11 months in autumn are at intermediate levels. Then, the stock management system inquires the current stock quantity of the fresh combined commodity, and the current stock quantity is 100 kg. According to seasonal sales trend and current inventory data, setting an inventory safety threshold, wherein the inventory safety threshold is 50 kg, namely, when the inventory quantity is lower than 50 kg, the inventory is determined to be insufficient, and when the inventory quantity is higher than 150 kg, the inventory is determined to be excessive. And then, predicting the commodity demand in a future period by using a time sequence prediction model. If the predicted result shows that the demand of the fresh combined commodity in the future week is (date, demand), (2020-01-01, 60), (2020-01-02, 55), (-)

2020-01-03, 70), (2020-01-07, 75), according to the demand forecast data and the inventory adjustment demand, combining the production and purchasing modules in the ERP system to obtain the adjustment direction of the production and purchasing plans, and the adjustment result is to increase the production and purchasing quantity. And (5) producing and purchasing corresponding fresh commodities according to the adjusted production and purchasing plans. If the adjusted production plan is 100 kg, the purchasing plan is 50 kg. According to the new fresh commodity inventory, the new inventory data is input by using the updating function of the inventory management system, the inventory information of the new fresh combined commodity is updated, and the updated inventory data is 200 kg. By comparing the updated inventory data with the demand forecast data, the inventory data is found to exceed the demand forecast data and is therefore deemed to meet the demand for the commodity for a period of time in the future. And outputting a verification result meeting the requirement. If the inventory data does not meet or exceed the demand forecast data, then the step of readjusting the production and procurement plans is performed.

S106, determining the final sorting result of the fresh combined commodity through detection of the scanning equipment.

The method comprises the steps of placing fresh combined commodities in a detection area of a scanning device according to a preset placing mode by an operator, and starting a preset scanning program through a scanning device interface. And scanning the fresh combined commodity to be detected through a sensor and a camera in the detection area, determining the color and the texture of the commodity, and judging whether the commodity has a spoilage condition or not. And adopting a convolutional neural network to classify the scanned images and identifying the types and the quantity of commodities in the images. And comparing the recognized commodity types and quantity results according to preset combined commodity standards to verify. Weight sensing equipment is adopted to detect the weight of the fresh combined commodity. And judging whether the types and the amounts of the fresh combined commodities are matched with the weights according to a reference manual of the preset weights and the types and the amounts of the commodities. Comparing the detection result of the weight sensing device with the output of the image recognition algorithm, and if the weight detection result is consistent with the type and the number of the commodities obtained through image recognition, confirming the information of the fresh combined commodities, including the type, the number and the weight of the commodities. If the combined commodities are inconsistent, the sorting of the fresh combined commodities is carried out again, and the scanning and the detection are carried out again. And generating a sorting report of the fresh combined commodity according to the confirmed result. For example, a fresh combination commodity contains 3 commodities, apples, bananas and tomatoes. And placing apples, bananas and tomatoes in a detection area of the scanning equipment according to a preset mode, and starting scanning through a touch screen interface. The sensor detects the red and smooth texture of tomatoes, the yellow and specific shape of bananas, and the green and circular shape of apples. The cameras then shoot the goods and the system also checks for signs of spoilage, black spots of tomatoes. The convolutional neural network analyzes the photographed images to identify 2 apples, 5 bananas and 3 tomatoes. The system confirms that the number of identified items is correct by comparison with the pre-set combined item criteria. The total weight detected by the weight sensing device matches the expected weights of 2 apples, 5 bananas and 3 tomatoes. The system confirms that the detection result of the weight sensing device is consistent with the output result of the image recognition algorithm. Since the detected weight matches the image-identified commodity category and number, the system confirms that this combination contains 2 apples, 5 bananas and 3 tomatoes. Finally, the system generates a sort report listing the types, amounts, and total weight of the fresh combination products.

The foregoing describes specific embodiments of the present invention. It is to be understood that the invention is not limited to the particular embodiments described above, and that various changes or modifications may be made by those skilled in the art within the scope of the appended claims without affecting the spirit of the invention. The embodiments of the present application and features in the embodiments may be combined with each other arbitrarily without conflict.

Claims

1. A method for sorting goods based on big data, the method comprising:

according to the sorting requirements of the fresh combined commodities, the fresh commodities are transferred from an inventory area to a sorting area through automatic equipment, sorted according to the characteristics of the fresh commodities, and arranged according to the requirements of varieties, specifications, weights and numbers; sorting sub-commodities according to sorting requirements of all sub-commodities in the fresh combined commodities, and carrying out combined sorting on the sub-commodities; for the fresh combined commodity, if a certain fresh commodity is out of stock, updating the information of the fresh combined commodity, and judging whether to re-sort or replace the out-of-stock fresh commodity in the fresh combined commodity; when new fresh commodities are marketed, sub commodities in the combined commodity to be adjusted are judged so as to introduce new commodity combinations and create more sales opportunities; acquiring the demand and the inventory of the commodity according to the seasonal variation and inventory fluctuation factors of the fresh combined commodity, and adjusting the production and purchasing plans according to the analysis result so as to meet the demands of different seasons and inventory; and determining the final sorting result of the fresh combined commodity through detection of the scanning equipment.

2. The method of claim 1, wherein the transferring fresh goods from the inventory area to the sorting area by the automated equipment according to the sorting requirements of the fresh combined goods, sorting according to characteristics of the fresh goods, and arranging according to variety, specification, weight, and quantity requirements, comprises:

acquiring sorting requirements of fresh combined commodities and data of a logistics management system; moving the goods from the inventory area to the sorting area using automated equipment, including an automated guided vehicle; and (3) detecting data of the sorted fresh commodities by using a sensor device, and determining whether sorting is performed again or sorting requirements are adjusted according to the detected data.

3. The method of claim 1, wherein the sorting sub-commodities according to the sorting requirement of each sub-commodity in the required fresh combined commodity, and the sub-commodities are sorted in combination, respectively, comprises:

judging the freshness of the fresh goods by using a freshness testing instrument; acquiring images of fresh commodities and judging the maturity of the commodities by using a convolutional neural network; according to the freshness and the maturity of the commodity, an automatic labeling machine is used for quality labeling; obtaining the size of fresh goods, and setting grouping standards to group the sizes; and carrying out combined sorting according to the types, the numbers, the qualities and the sizes of the sub-commodities.

4. The method of claim 1, wherein the updating information of the fresh combination commodity if a certain fresh commodity is out of stock for the fresh combination commodity, and determining whether to re-sort or replace the out-of-stock fresh commodity in the fresh combination commodity comprises:

acquiring inventory data of all sub-commodities, wherein the inventory data comprises commodity names, inventory amounts and warehouse positions; comparing the stock data with the required quantity in the combined commodity to identify sub-commodity with insufficient stock; matching sub-commodities with insufficient inventory in a commodity database, and finding out alternative commodities with similar attributes; determining the optimal alternative commodity through comprehensive analysis of attribute similarity, price and inventory; based on the candidate commodities, updating the constitution and inventory information of fresh combined commodities; sorting the updated fresh combined commodities by using automatic equipment, and checking quality and quantity after finishing; further comprises: determining whether to replace according to the similarity between the replaceable fresh commodity attribute and the fresh commodity attribute in the absence of the commodity; according to the replaced fresh combined commodity information, other sub-commodities needing to be replaced are determined;

the method for determining whether to replace according to the similarity between the replaceable fresh commodity attribute and the fresh commodity attribute in the absence of the commodity specifically comprises the following steps: acquiring attribute information of the out-of-stock commodity according to the out-of-stock fresh commodity list; calculating Euclidean distance between the replaceable fresh commodity and the out-of-stock commodity as the attribute similarity by using an euclidean_distances function in the Scikit-learn library; judging whether the attribute of the replaceable commodity is similar to the attribute of the absent commodity by setting a similarity comparison rule, and determining whether the absent commodity is replaced according to a judging result to acquire a replaceable fresh commodity list; using a data processing tool Pandas, and matching the attribute of the replaceable commodity with the attribute of the backorder commodity by utilizing the data matching and screening functions of the Pandas; general purpose medicine

After the matching result, a replaceable commodity list similar to the attribute of the backorder commodity is obtained, and the final replaceable fresh commodity is determined; replacing the stock-out commodity with a replaceable commodity to obtain replaced fresh combined commodity information, wherein the information comprises specification, weight and quantity;

the method for determining other sub-commodities to be replaced according to the replaced fresh combined commodity information specifically comprises the following steps: acquiring attribute information of other sub-commodities in the fresh combined commodity; according to the replaced fresh combined commodity information, matching the attribute of other sub-commodities with the updated commodity attribute, judging whether the other sub-commodities need to be replaced or not, and determining the other sub-commodities needing to be replaced; acquiring a list of alternative other sub-commodities, and matching the attribute of the alternative commodity with the attribute of the sub-commodity to be replaced by adopting an attribute matching algorithm to determine the final alternative other sub-commodities; updating other sub-commodities needing to be replaced into replaceable commodities, and determining the attribute of the updated fresh combined commodity, including the specification, the weight and the quantity, so as to obtain updated fresh combined commodity information; judging whether the updated fresh combined commodity meets the expected requirement by adopting a data statistics method; determining whether to further adjust according to the judging result; and determining the final fresh combined commodity according to the updated fresh combined commodity information and the verification result.

5. The method of claim 1, wherein the determining sub-items in the combined commodity that need to be adjusted to introduce new commodity combinations and create more sales opportunities when new fresh commodity is marketed comprises:

acquiring information of fresh goods newly marketed from a goods inventory database, wherein the information comprises names, categories, selling prices, supplier information and warehousing dates; analyzing the new commodity attribute by utilizing a collaborative filtering algorithm, comparing the new commodity attribute with the existing combined commodity attribute, and determining whether the new commodity attribute is compatible with the existing combined commodity attribute; under the condition of compatibility, updating the combined commodity information, and analyzing the estimated sales and profit margin by using a linear regression algorithm; according to the analysis, carrying out actual sorting, quality inspection and quantity checking of the new combined commodity; monitoring sales and sales of the new combined commodity in real time, and evaluating differences from expected sales; determining combined commodity needing to be adjusted, and outputting a combined commodity list needing to be adjusted; further comprises: based on a collaborative filtering algorithm, evaluating the compatibility of the fresh goods and the existing fresh combined goods; determining whether the new commodity can be combined with certain sub-commodities by evaluating the association and competitiveness between the new commodity and the sub-commodity in the existing combined commodity; according to the supply chain and inventory management conditions of the new commodity, determining whether effective pairing can be carried out with sub-commodities of the existing combined commodity, and ensuring effective management of supply and inventory;

Based on collaborative filtering algorithm, evaluate the compatibility of fresh goods of new generation and current fresh combination commodity, specifically include: according to the historical purchasing record of the user, scoring data of the user on the existing fresh combined commodity is obtained; calculating the similarity between users by a cosine similarity calculation method; selecting a certain number of users with similar purchasing behavior with the target user as neighbors according to the similarity calculation result; determining attributes related to fresh goods, including goods category, price interval, production place, freshness and quality, according to purchase records of neighbor users; evaluating compatibility of the fresh goods and the existing fresh combined goods by analyzing the scores of the neighbor users on the fresh goods and the scores of the existing fresh combined goods;

the method for determining whether the new commodity can be combined with some sub-commodities by evaluating the relevance and the competitiveness between the new commodity and the sub-commodity in the existing combined commodity comprises the following steps: acquiring detailed information of new fresh combined commodities, including commodity characteristics, expected selling prices and target consumer groups, through an enterprise database; acquiring sub-commodity information in the new fresh combined commodity, wherein the sub-commodity information comprises sales data, cost and customer feedback of each sub-commodity; obtaining cosine similarity of the new commodity and each sub-commodity in commodity characteristics, expected selling price and target consumer groups by utilizing a Scikit-learn library of a data analysis tool Python, and obtaining a relevance index; factor analysis of SPSS is carried out by using a statistical analysis tool, so that competitive power indexes of the new commodity and each sub commodity in sales data, cost and customer feedback are obtained; according to the obtained competitive power index, a threshold value is set, and the fact that the competitive power index of the sub-commodity and the new commodity is lower than the set threshold value is judged, so that the competition between the new commodity and the sub-commodity is low; screening sub-commodities with the relevance index higher than a first preset threshold and the competitiveness lower than a second preset threshold according to the obtained sub-commodity list with low competitiveness, and taking the sub-commodities with the relevance index higher than the first preset threshold as candidate sub-commodities possibly combined with the new commodity; using data analysis tool Tableau in combination with Monte Carlo simulation

The method comprises the steps of predicting sales and customer satisfaction of new commodities combined with candidate sub-commodities, wherein parameters of a simulation experiment comprise commodity price, market demand and competitiveness; if the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list;

the method for determining whether the sub-commodity can be effectively paired with the sub-commodity of the existing combined commodity according to the supply chain and the inventory management condition of the new commodity, and ensuring the effective management of supply and inventory comprises the following steps: acquiring supply chain information of a new commodity through a commodity data acquisition system, wherein the supply chain information comprises a production date, a quality guarantee period, a transportation time and a manufacturer; acquiring the supply chain information of sub-commodities of the existing combined commodity according to the supply chain information of the new commodity; judging similarity of the new commodity and sub commodities in supply chain management according to production date, quality guarantee period, transportation time and manufacturer data by adopting a K-means based clustering algorithm; acquiring inventory management information of new commodities, including inventory quantity, sales period and sales quantity, through an inventory data acquisition system; acquiring inventory management information of sub-commodities of the existing combined commodity according to the inventory management information of the new commodity; adopting a linear regression algorithm, and judging whether the new commodity has similarity with sub commodities in inventory management according to the inventory, the sales period and sales data; according to the obtained similarity result, setting a threshold value, if the similarity is larger than a preset threshold value, judging which sub-commodities have high similarity with the supply chain and the inventory of the new commodity, wherein the sub-commodities are candidate commodities possibly combined with the new commodity, and obtaining a candidate commodity list; a Monte Carlo simulation experiment is adopted, and sales and customer satisfaction which can occur are predicted by simulating sales conditions of new commodity and candidate commodity combinations; parameters of the simulation experiment comprise commodity price, market demand and competitiveness; if the simulated sales and the customer satisfaction reach the preset standards, determining that the new commodity can be combined with the sub-commodities to obtain a final combined commodity list.

6. The method of claim 1, wherein the obtaining the demand and the inventory of the commodity according to the seasonal variation and the inventory fluctuation factors of the fresh combined commodity, and adjusting the production and the purchasing plans according to the analysis result to meet the demands of different seasons and inventory comprises:

extracting sales volume and sales date of the commodity; a time sequence analysis method is applied to obtain seasonal sales trends; acquiring the current inventory of the commodity and setting an inventory safety threshold; applying a time sequence prediction model to obtain future commodity demand prediction data; generating an adjusted production and purchase plan, and carrying out corresponding fresh commodity production and purchase; and updating the inventory information of the fresh combined commodity and verifying whether the commodity requirement in a future period of time is met.

7. The method of claim 1, wherein the determining, by detection by the scanning device, a final sorting result of the fresh combined commodity comprises:

placing the commodity in a detection area of the scanning equipment by an operator, and starting a preset scanning program; using a sensor and a camera to scan the commodity to determine the color and texture of the commodity; carrying out image classification by adopting a convolutional neural network, and identifying the types and the quantity of commodities; verifying the types and the quantity of the commodities, and detecting the weight by using a weight sensing device; comparing the weight detection result with the image identification output, and confirming or re-sorting the commodities; and generating a sorting report of the fresh combined commodity.