KR102158970B1 - Sales and inventory managment system for computer products - Google Patents

Abstract

The present invention relates to a computer product sale and inventory management system capable of increasing sale efficiency and inventory management of computer products. The computer product sale and inventory management system comprises: a recommendation agent which selects a recommended product by using a collaborative filtering technique; a payment agent which provides information of the recommended computer product to a user terminal and executes a payment process when receiving a purchase request; and an inventory management agent which executes inventory management of computer products.

Description

Computer product sales and inventory management system {SALES AND INVENTORY MANAGMENT SYSTEM FOR COMPUTER PRODUCTS}

The present invention relates to a computer product sales and inventory management system, and more particularly, to a computer product sales and inventory management system that can increase the sales efficiency and inventory management of computer products.

Along with the development of communication networks and terminals with various functions that support them, shopping mall services that provide various computer products online are being provided, and users who use such shopping mall services use their terminals to access various computer products on the Internet. It can be conveniently purchased through.

In such a shopping mall service, in order to promote sales, a discount rate for computer products is set or a method of inducing sales by exposing stock products to a higher level is taken, but there is a limit to sales promotion by such a uniform method.

In addition, since computer products come in various types and features, it is difficult to select them.

In addition, in the shopping mall service, in order to accurately manage the inventory quantity of computer products to be sold, the type of inventory product and its entry and exit quantity (including incoming and outgoing quantity) are accurately checked when inventory products are received or released from the warehouse. Should be.

However, even if the stock quantity of the stock product is checked correctly, if the stock quantity of the stock product is insufficient, the supplier who calculates the shortfall for each stock product to supplement the stock quantity, and supplies the stock product to compensate for the shortfall You have to place orders individually (this is a traditional inventory management technique).

If there are not many kinds and quantities of products in stock, it is not difficult to grasp the stock quantity and make up for shortfalls, but if the kinds and quantity of stock products increase, it is very difficult to understand the stock quantity and make up for the shortfall.

KR 10-1679381 B1

Accordingly, one object of the present invention is to provide a computer product sales and inventory management system that can increase sales efficiency and inventory management of computer products as conceived to solve the above problem.

Other objects and advantages of the present invention will be described below, and will be understood by examples of the present invention. Further, the objects and advantages of the present invention can be realized by means and combinations indicated in the claims.

In order to achieve the above object, the present invention provides a computer product sales and inventory management system, comprising: a recommendation agent for selecting recommended computer products using a collaborative filtering technique; A payment agent that provides information on the recommended computer product to a user's terminal and executes a payment process when a purchase is requested; And an inventory management agent that executes inventory management of computer products, wherein the inventory management agent predicts a demand amount of computer products in consideration of a quantitative factor of periodic sales quantity and a qualitative factor due to discounts and holidays. Demand forecast module; An order quantity calculation module for determining an optimal economical order quantity Q ₁ based on the following equations (1) to (7) in order to minimize the total cost related to inventory; A decision-making module that performs optimal decision-making on order quantity and order date based on demand forecasting using Sugeno's fuzzy integration; And it characterized in that it comprises a stock grasping module for determining the actual inventory amount of the computer product.

(1)

(One)

(2)

(2)

(3)

(3)

(4)

(4)

If predicted demand (T), average daily use (M), number of days (K), safety stock (X), order quantity (N),

(5)

(5)

(6)

(6)

Therefore, the optimal economical order quantity (Q1) is,

(7)

(7)

)는 사용자 A의 아이템 q에 대한 평가치로, 하기 수식 (9)에 의해 사용자 A의 최근접 이웃의 평가치들을 가중 평균하여 예측하는 것을 특징으로 한다.In addition, the recommendation agent assigns each weight based on the user's disposition, age, gender, region, income standard, and foreign/domestic preferences, and stores them in the database, and is different from the user in order to select other users similar to the user. The similarity between users is calculated by the following equation (8), and based on the calculated similarity, the N neighbors that are most similar to the recommended users are selected, and the evaluation values of the recommended users are predicted based on the evaluation values of the selected neighbors, and then in descending order. Generates a recommendation list of, and recommends the computer product of the TOP3 user to the user among the generated recommendation list, and the evaluation value (

) Is an evaluation value of user A's item q, characterized by a weighted average of the evaluation values of user A's nearest neighbors according to Equation (9) below.

(8)

(8)

는 사용자 A와 사용자 B의 유사도이고, q는 사용자 A와 사용자 B가 공통으로 평가한 컴퓨터 제품의 인덱스이고,

는 컴퓨터 제품 i에 대한 사용자 A의 평가점수이고,

는 컴퓨터 제품 i에 대한 사용자 B의 평가점수이고,

는 사용자 A의 평가점수 평균값이고,

는 사용자 B의 평가점수 평균값임.here,

Is the similarity between user A and user B, q is the index of computer products commonly evaluated by user A and user B,

Is the score of user A for computer product i,

Is the score of user B for computer product i,

Is the average value of user A's evaluation score,

Is the average value of user B's evaluation score.

(9)

(9)

는 사용자 A의 최근접 이웃인 사용자 j의 이용 가능한 평가치들의 평균값을 의미함.here,

Denotes the average of the available evaluation values of user j, user A's nearest neighbor.

In addition, the inventory management agent applies a fuzzy neural network, and the input layer includes three variables of demand, inventory, and order, as well as importance and ordering of products as input vectors, and product quantity and product name as output layers.

In addition, the inventory identification module includes a 3D camera mounted and moved on a drone to photograph inventory of computer products, a data preprocessing unit that preprocesses the data collected by the 3D camera, and the type of inventory by analyzing the preprocessed data. And it characterized in that it comprises a data analysis unit to determine the number.

In addition, the data analysis unit extracts an inventory area using an edge detector method, uses a support vector machine (SVM) and a convolutional neural network (CNN) to determine the type of inventory, and corresponds to the number of inventory. It is characterized by using a single volume of stock.

As described above, according to the present invention, there is an effect of providing a computer product sales and inventory management system that can increase sales efficiency and inventory management of computer products.

1 is a block diagram of a computer product sales and inventory management system according to the present invention.
2 is a diagram schematically showing the configuration of the management server according to FIG. 1.
3 is a diagram schematically showing the configuration of the inventory identification module according to FIG. 2.
4 is a block diagram of a fuzzy neural network according to the present invention.

Details of other embodiments are included in the detailed description and drawings.

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, and may be implemented in various different forms. In the following description, when a certain part is connected to another part, this is not only a case where it is directly connected, but also It includes the case of being connected with another medium in the middle. In addition, parts not related to the present invention in the drawings are omitted in order to clarify the description of the present invention, and like reference numerals are attached to similar parts throughout the specification.

Hereinafter, the present invention will be described with reference to the accompanying drawings.

1 is a block diagram of a computer product sales and inventory management system according to the present invention.

Referring to FIG. 1, a computer product sales and inventory management system includes a user terminal 100, a manager terminal 200, and a management server 300.

The user terminal 100, the manager terminal 200, and the management server 300 are a processor, controller, integrated circuit, programmable logic circuit, or other database of any number or combination in order to perform the functions of the present invention. And a signal processing device, and in addition, one or more memories, a transmitter and a receiver, a display, and a communication module capable of communicating with various devices may be included. In addition, it includes a computer program that performs data operation through a specified algorithm and displays the result so that it can be recognized externally through a display, and a user interface (UI; User Interface) implemented to allow the user to manipulate it. I can.

The user terminal 100, the manager terminal 200, and the management server 300 may include any device or equipment including a memory associated with a processor. The processor can implement an operating system and can execute computer programs known as instructions, software codes, executable files, applications, and apps.

The user terminal 100 and the manager terminal 200 may include voice or data communication devices such as a mobile phone, a mobile phone, a smart phone, and a notebook.

2 is a diagram schematically showing the configuration of the management server according to FIG. 1.

Referring to FIG. 2, the management server 300 includes a fuzzy-neural network 310, a payment agent 320, a recommendation agent 330, and an inventory management agent 340.

The fuzzy-neural network 310 is an intelligent complex system that combines a neural network and a fuzzy system technology.

Neural networks are effective when applying the pattern classification problem, but it is not easy to explain the output results. Fuzzy systems are suitable for judging and explaining unstable or ambiguous results, but it is not easy to automatically create rules for judgment. Therefore, in order to compensate for these shortcomings, a fuzzy-neural network that combines a neural network and a fuzzy system technology is adopted.

The payment agent 320 enables a user to perform a transaction using the user terminal 100 and basically make a mobile payment for a purchased product. Since the computer product sales and inventory management system in the present invention can target users in various countries, it is desirable to implement it so as to be able to interwork with the mobile payment system that is already used in each country.

The recommendation agent 330 is responsible for selecting (recommending) recommended computer products using a collaborative filtering technique.

)는 사용자 A의 아이템 q에 대한 평가치로, 하기 수식 (9)에 의해 사용자 A의 최근접 이웃의 평가치들을 가중 평균하여 예측하는 것을 특징으로 한다.In addition, the recommendation agent 340 assigns each weight based on the user's disposition such as age, gender, region, income standard, and foreign/domestic preferences and stores them in the database, and selects other users similar to the user. The similarity between the user and other users is calculated by the following equation (8), and based on the calculated similarity, the N neighbors most similar to the recommended user are selected, and the evaluation value of the recommended user is predicted based on the evaluation value of the selected neighbor. After that, it creates a recommendation list in descending order, recommends the computer product of the TOP 3 user among the generated recommendation list to the user, and the evaluation value (

) Is an evaluation value of user A's item q, characterized by a weighted average of the evaluation values of user A's nearest neighbors according to Equation (9) below.

(8)

(8)

는 사용자 A와 사용자 B의 유사도이고, q는 사용자 A와 사용자 B가 공통으로 평가한 컴퓨터 제품의 인덱스이고,

는 컴퓨터 제품 i에 대한 사용자 A의 평가점수이고,

는 컴퓨터 제품 i에 대한 사용자 B의 평가점수이고,

는 사용자 A의 평가점수 평균값이고,

는 사용자 B의 평가점수 평균값이다.here,

Is the similarity between user A and user B, q is the index of computer products commonly evaluated by user A and user B,

Is the score of user A for computer product i,

Is the score of user B for computer product i,

Is the average value of user A's evaluation score,

Is the average value of user B's evaluation score.

(9)

(9)

는 사용자 A의 최근접 이웃인 사용자 j의 이용 가능한 평가치들의 평균값을 의미한다.here,

Denotes the average value of available evaluation values of user j, user A's nearest neighbor.

In other words, in the present invention, for recommendation of computer products, collaborative filtering is used according to the user's tendencies such as age, gender, region, income standard, and foreign/domestic preferences.

When a user signs up as a member, he/she writes age, gender, region, income standard, foreign/domestic preferences, etc., and weights them accordingly and finds other users with similar preferences and recommends Computo products recommended by that person. will be.

In the method of recommending collaborative filtering in the present invention, each weight is assigned to age (30%), gender (25%), region (10) based on the registered age, gender, region, income standard, and foreign/domestic preference when registering as a member. %), income standard (15%), foreign/domestic preference (20%). Among the weights, ages were classified by predicting that the computer products preferred by users in their 10s to 90s would be different, and the regions were divided into 8 provinces and 7 cities in Korea to select. In addition, foreign/domestic preference allows you to choose which of the foreign products and domestic products are preferred, and the income standard is divided into more than or less than the middle class based on an annual salary of 55 million won, which increases the tax burden.

In this way, users with similar tendencies are grouped and computer products are recommended with rating information of the group group.

The inventory management agent 340 is responsible for performing inventory management of computer products.

Referring back to FIG. 2, the inventory management agent 340 includes a demand quantity prediction module 342, an order quantity calculation module 344, a decision making module 346, and an inventory identification module 348.

The demand quantity prediction module 342 is responsible for predicting the quantity of demand of computer products by considering both the quantitative element of the periodic sales quantity and the qualitative element due to discounts and holidays.

The order quantity calculation module 344 is responsible for determining the optimal economical order quantity Q ₁ based on the following equations (1) to (7) in order to minimize the total cost related to inventory.

(1)

(One)

(2)

(2)

(3)

(3)

(4)

(4)

If predicted demand (T), average daily use (M), number of days (K), safety stock (X), order quantity (N),

(5)

(5)

(6)

(6)

Therefore, the optimal economical order quantity (Q1) is,

(7)

(7)

The decision-making module 346 is in charge of performing optimal decision-making on the order quantity and order date based on demand prediction by using Sugeno's fuzzy integration.

The inventory determination module 348 is responsible for the function of determining the actual inventory amount of computer products.

Referring to FIG. 3, the inventory identification module 348 includes a 3D camera 348a, a data preprocessor 348b, and a data analysis unit 348c.

The 3D camera 348a is for photographing inventory of computer products.

In order for the 3D camera 348a to acquire a picture, it needs to be physically moved. If an AS/RS (Automated Storage and Retrieval System) is established in the inventory warehouse, the camera can be installed on the device and moved. In the case of the AS/RS system, if you use a recording device that shoots at a speed of 30 fps or higher because it moves at 10 to 25 km/h, the AS/RS mobile device moves and you can shoot at the same time, so you can quickly acquire photos. If the facility is not established, a camera can be mounted on the drone to move the camera to the data collection location.

The data preprocessor 348b is responsible for preprocessing the data collected by the 3D camera 348a.

The data preprocessor 348b performs camera calibration and depth data conversion.

There are various types of 3D cameras, but most are composed of RGB cameras and IR cameras. Since the two cameras are physically located at different locations, an image using the two cameras becomes an image with different focus. Also, the resolution of RGB cameras in most cameras is much higher than that of IR cameras. In this way, there is a difference in resolution and a difference in the position of the camera, so camera calibration is required to correct it so that an image of the same resolution and the same point are captured. Chess-Board Patterns were used to obtain parameter values for camera calibration. Cameras with two different capabilities and functions can be calibrated to obtain the same type of information. This makes it easier to proceed with data analysis in the future.

When data is acquired by the 3D camera 348a, the acquired data includes an RGB image and an image representing the distance to the object (Depth).

In the case of storing RGB data and depth data into data, there is a disadvantage that it is difficult to intuitively check in stages such as final result verification. For this reason, in the case of RGB data, the photo itself was saved, and the depth data was also saved in the form of a photo.

The data analysis unit 348c is responsible for a function of analyzing the data preprocessed by the data preprocessing unit 348b to determine the type and number of inventory.

Using the data obtained using the 3D camera 348a, it will be necessary to detect what kind of stock is on the stock shelf. Once you've identified the type of inventory, you need to figure out how many inventory is on that shelf.

The data analysis unit 348c extracts an inventory area using an edge detection method, uses a support vector machine (SVM) and a convolutional neural network (CNN) to determine the type of inventory, and It is characterized by using a single volume of the inventory to determine the number.

-Inventory area extraction using edge detection

When a 3D camera shoots an inventory area as shown in [Reference Figure 1], the area where inventory is located and the area where inventory is not located must be divided. In addition, it is necessary to extract the edge of the boundary part in the area where the stock is located.

[Reference Figure 1]

Inventories have a step edge edge shape. For this reason, in the process of extracting the corresponding area, it can be determined that an edge is between a pixel representing the distance from the innermost side of the stock wall and a pixel that is considered closer than that. As can be seen in [Reference Fig. 2], in this simple manner, it is possible to detect an area of the entire inventory, such as a red line.

[Reference Figure 2]

-Check inventory type using CNN and SVM

[Reference Figure 3] shows a methodology for grasping the type of image data using CNN and SVM. The same algorithm is also applied in the present invention. First, when an inventory image is acquired, the inventory type must be learned in the SVM to determine what inventory the image is in. CNN is used to extract image features to be trained in SVM. Inventory features are extracted using CNN, and the extracted features are used as training data for SVM. Afterwards, the previously learned data is used to check what kind of stock the image captured and entered in the stock area is.

The reason why the method combining CNN and SVM is adopted in the present invention is to accurately identify the type of inventory. If CNN is used, the characteristics of inventory can be identified most accurately. However, if only CNN is used, an error may occur when an image that has not been learned is entered. In order to prevent such an error, the characteristics extracted from the CNN are put into the SVM to identify the type. If the SVM is used to classify the stock type, it is easy to recognize when an unlearned image comes in.

-Stock count calculation

In order to determine the number of inventory, some assumptions are first required.

First, inventory must be stacked from bottom to top, and inventory must be stacked first near the wall. In other words, when viewed from the front, there should not be a phenomenon in which the empty space in the rear part is covered by stock items.

Second, as shown in [Reference Fig. 4], the stocks should not be arranged obliquely to the wall, but should be arranged parallel to the wall.

Third, if there is a distance between the wall and the nearest inventory, the distance should be constant even if the type of inventory changes.

[Reference Figure 4]

If the above assumptions are met, the distance between the 3D camera and the stock product will be known. Knowing the distance to the stock item will allow you to count how many stock items are behind the stock item closest to the camera.

Referring to FIG. 4, the inventory management agent 340 applies a fuzzy-neural network to obtain an accurate inventory of materials, but the input layer uses three variables of demand, inventory, and order, and the importance and ordering of products as input vectors. And the quantity and product name of the product as the output layer

The order quantity may be data including purchase and maintenance costs. The hidden layer can converge by learning while extending from 2 nodes to 13 nodes.

The output vector represents the optimal economic quantity and product name. The learning rate of the algorithm was 0.6 and the average error was 0.05.

Those of ordinary skill in the art to which the present invention pertains will appreciate that the present invention can be implemented in other specific forms without changing the technical spirit or essential features thereof. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. The scope of the present invention is indicated by the scope of the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts are included in the scope of the present invention. It must be interpreted.

100: user terminal
200: manager terminal
300: management server
310: fuzzy-neural network
320: payment agent
330: referral agent
340: Inventory management agent

Claims (5)

As a computer product sales and inventory management system,
A recommendation agent that selects recommended computer products using a collaborative filtering technique;
A payment agent that provides information on the recommended computer product to a user's terminal and executes a payment process when a purchase is requested; And
Including an inventory management agent for executing inventory management of computer products,
The inventory management agent,
A demand quantity prediction module for predicting a demand quantity of computer products by considering both a quantitative element of the periodic sales quantity and a qualitative element due to discounts and holidays;
An order quantity calculation module for determining an optimal economical order quantity Q ₁ based on the following equations (1) to (7) in order to minimize the total cost related to inventory;
A decision-making module that performs optimal decision-making on order quantity and order date based on demand forecasting using Sugeno's fuzzy integration; And
Computer product sales and inventory management system, characterized in that it comprises an inventory identification module to determine the actual inventory of the computer product.

(One)

(2)

(3)

(4)
If predicted demand (T), average daily use (M), number of days (K), safety stock (X), order quantity (N),

(5)

(6)
Therefore, the optimal economical order quantity (Q1) is,

(7) The method of claim 1,
The recommendation agent,
Each weight is assigned and stored in the database based on the user's disposition, age, gender, region, income standard, and foreign/domestic preference,
In order to select another user similar to the user, the degree of similarity between the user and other users is calculated by the following equation (8),
Based on the calculated similarity, select the N neighbors most similar to the recommended users, predict the evaluation values of the recommended users based on the evaluation values of the selected neighbors, and create a recommendation list in descending order,
Among the generated recommendation list, computer products of TOP3 users are recommended to users,
The evaluation value (

) Is an evaluation value of user A's item q, and predicts a weighted average of evaluation values of user A's nearest neighbors according to Equation (9) below.

(8)
here,

Is the similarity between user A and user B, q is the index of computer products commonly evaluated by user A and user B,

Is the score of user A for computer product i,

Is the score of user B for computer product i,

Is the average value of user A's evaluation score,

Is the average value of user B's evaluation score.

(9)
here,

Denotes the average of the available evaluation values of user j, user A's nearest neighbor. The method of claim 1,
The inventory management agent applies a fuzzy neural network, but the input layer includes three variables of demand, inventory, and order, as well as the importance and ordering of products as an input vector, and the quantity and product name as an output layer. Sales and inventory management system. The method of claim 1,
The inventory identification module includes a 3D camera mounted and moved on a drone to photograph the inventory of computer products, a data preprocessing unit that preprocesses the data collected by the 3D camera, and the type and number of inventory by analyzing the preprocessed data. Computer product sales and inventory management system comprising a data analysis unit to grasp. The method of claim 4,
The data analysis unit extracts an inventory area using an edge detector method, uses a support vector machine (SVM) and a convolutional neural network (CNN) to determine the type of inventory, and uses the corresponding inventory to determine the number of inventory. Computer product sales and inventory management system, characterized in that using a single volume.

Images