KR100418620B1 - System for managementing supply-chain and method for operating the same - Google Patents

Abstract

The present invention measures the degree of uncertainty according to the marketing in multiple stages through collaborative demand analysis and prediction, realizes the optimal volume planning in multiple stages considering the difference in procurement price conditions and the customization of the stages, and utilizes real-time sales information. In order to provide a supply chain management system and an operation method for implementing an optimal volume distribution for minimizing opportunity loss through sales analysis and distribution plan for each store, the supply chain management system of the present invention is collected from a store operator. A database where a plurality of store information is registered is established, connected to the operation server consisting of various sub-systems for product demand analysis and prediction, quantity planning, and sales analysis and distribution before and during the product sales season, and connected to the Internet network. Demand analysis and forecast, quantity planning and sales analysis by the operation server And a plurality of store operator side computer terminals selectively providing store information so as to be distributed, and communicate with the store operator side computer terminal through an internet network, and linked with the operation server to analyze and forecast demand by the operation server. It is characterized in that it comprises a manager computer terminal for displaying the results of the volume planning and sales analysis and distribution to the manager.

Description

System for managementing supply-chain and method for operating the same}

The present invention relates to a supply chain management system of products, in particular, analyzes and forecasts demand in a collaborative manner for products with extremely short product life cycles and long procurement periods, and sales as well as volume planning and ordering. The present invention relates to a supply chain management system and a method of operation of a product that can maximize profits and minimize risk by systematically combining analysis and distribution optimization.

In other words, through the system using the Internet-based information server, the demand forecasting information between each retailer and the supplier and the information are exchanged so that the characteristics of each store and the attributes of the sales products are taken into consideration, and thus the collaborative information between the supplier and each store operator. Demand forecasting combining exchange and optimization analysis method, multi-level volume planning model considering mass customization, and optimal supply chain management system through proper allocation of stores for minimizing opportunity loss and its operation method.

Recently, the production method of fashion products is based on small quantity production of many kinds, and the change of fashion is fast, the life cycle is short, and it requires customized products for each consumer's taste.

In particular, in the case of fashion products, the demand itself is extremely uncertain, and at the same time, the wider selection of products makes the competition between products more intense and the emergence of competitors and global sourcing based on the global environment. In light of the growing competition with weapons of cost advantage, it is difficult to cope with existing production methods.

Recently, supply chain management (SCM) and enterprise resource planning (ERP) models have been in the spotlight as a solution model suitable for this environment, and applied to many industries as a tool to improve the structure of the entire industry. This is on the trend.

For example, an enterprise resource planning (ERP) model is a comprehensive system that integrates and plans and manages overall information necessary for management of materials, personnel, and information in all areas of a company such as production, sales, materials, accounting, and personnel. There is great significance as a system.

However, the enterprise resource planning model assumes definite demands, such as MRP (Material Requirement Planning) and MRP II in the past, and sets demand and timing based on the Bill of Material (BOM) according to the main production schedule. Inverted decision, the unplanned situation caused by the development of demand fluctuation, supply delay, etc., which has been pointed out as a fundamental MRP problem, and the neurality caused by excessive change of the plan are raised. have.

Although to fix the above problems, although a fixed policy of the production planning period is used, the fundamental problems cannot be solved.

On the other hand, in the case of the supply chain management model, the problems of the aforementioned ERP model remain the same, and the supply chain management model is required according to the characteristics of the industry, so that the order lead time is long and the product life cycle In the case of fashion products with a very short cycle and various types, the existing general supply chain management model is not suitable in view of the characteristics that the order must be made long before the order is placed.

In particular, in recent years, the proliferation of global outsourcing has resulted in a prolonged lead time, and the above-mentioned problem is remarkable, and the effects of loss due to lack of inventory on profits and market share are increasing with competition. Will be further weighted.

In addition, current ERP and supply chain management models are based on MRP based on definite demand and static BOM without considering risk, so they do not reflect the characteristics of these industries based on risk. I can't.

For reference, the Merrill Lynch report reported that, for advanced home appliances, these malicious inventory growth rates reached 26% annually, the highest level of 17 years, despite the current supply chain management and enterprise resource planning system. Rather, the problem is getting worse.

In the case of high-tech products, clothing, precious metals, accessories and sundries, furniture, etc., the product life cycle is very short, but the order lead time is extremely long, so the uncertainty of the demand forecast is large and supply and demand imbalance is the most serious problem.

Therefore, there is an urgent need for a new model that complements the limitations of existing systems based on uncertainties in supply chain management.

The present invention has been made to solve the above problems, through the cooperative demand analysis and forecasting, quantity planning, etc. to measure the instability of the demand for each stage of the new product with a short life cycle, the difference in the procurement price conditions of each stage and Its purpose is to provide a supply chain management system and operation method that consists of a multi-level volume plan considering mass customization and a distribution system that minimizes the loss of opportunity in consideration of the characteristics of meeting demand through alternative materials.

The supply chain management system of the present invention for achieving the above object is the product information analysis and forecasting, quantity planning and before and during the sales season of the product through the database of store information registered, and cooperative information with the store operator and Operation server consisting of various sub-systems for sales analysis and distribution, and a plurality of stores that selectively provide store information through the Internet network for demand analysis and prediction, quantity planning and sales analysis and distribution by the operation server. It communicates with the computer terminal on the store side via a computer terminal and the Internet network, and interlocks with the operation server to visually provide the manager with the results of demand analysis and forecasting, quantity planning and sales analysis and distribution by the operation server. Characterized in that it comprises a computer terminal on the administrator side.

More specifically, before the start of the sales season for a particular product, the characteristics of the demand distribution according to the change of point of time of each product can be analyzed by analyzing the characteristics of stores, detailed categories, and demand information using data from past seasons. A collaborative demand analysis subsystem that analyzes dynamic changes; By providing information on new products for each product to the merchandisers in the store by collaborative information exchange with the merchandisers in each store, the demand forecasting information of the merchandisers is applied to the cooperative demand analysis module to calculate the total demand for each new product, the demand for each store, and the SKU. Initial demand forecasting including unit demand is performed, and the error occurred in the initial demand forecasting process is corrected based on the demand information obtained by sending the new product to the test market and selling the product immediately before the sale season of the product. A collaborative prediction subsystem that performs; A collaborative volume planning sub-system that executes a volume purchasing plan for raw materials and finished products based on POS data in consideration of mutual substitution and complementarity between products so that multi-level decision making reflects uncertainty of information through optimization techniques considering uncertainty and; A sales analysis and distribution sub-system, which performs sales analysis and demand forecasting information correction and optimal volume distribution according to the demand characteristics of each store during the sales season of the product based on continuously updated POS data; A graphic user interface unit for providing information calculated from the respective sub-systems to an administrator-side computer terminal; It is characterized by including the Internet network for data transmission between store operators and system administrators.

Meanwhile, the supply chain management operating method of the present invention receives store information from a merchandiser of each store through the Internet network before the sale season for a specific product begins, and analyzes the characteristics of stores, analyzes by detailed categories, and accuracy of demand information. Analyzing the demand of each product by analyzing the demand of the product, and providing the merchandise of the merchandiser with the information on the new product for each product, the total demand forecasting, the total demand forecasting and the SKU Accurate demand forecasting for products by performing initial demand forecasting through unit demand forecasting, and correcting errors generated in the initial demand forecasting process based on demand information obtained by sending the new product to the test market and marketing the product immediately before the sale season of the product. Step by step, and optimization techniques that take into account uncertainty. Implementing a purchase plan for raw materials and finished products based on POS data in consideration of interchangeability and complementarity between products to make multi-level decision making reflecting uncertainty of information, and continuously updating POS data. Based on the analysis of sales analysis and demand forecasting information during the sales season of the product and the step of performing the optimal volume distribution according to the demand characteristics of each store is characterized in that it is made.

1 is a block diagram of a supply chain management system of the present invention

2 is a block diagram of the cooperative demand analysis sub-system of the present invention

3 is a flowchart for explaining a method for analyzing store characteristics according to the present invention.

4a to 4b is a flowchart for explaining a detailed category analysis method according to the present invention

5 is a flowchart illustrating a method for analyzing accuracy of demand information according to the present invention.

6 is a GUI data visually showing the accuracy of the sales volume and the demand information by sales period of the knitwear using the cooperative demand analysis and prediction sub-system of the present invention

7 is a flowchart illustrating a total demand prediction method according to the present invention.

8 is a flowchart illustrating a method for predicting total demand for each store according to the present invention.

9 is a flowchart for explaining a SKU unit demand forecasting method according to the present invention

10 is a block diagram of a cooperative quantity planning sub-system according to the present invention

11a to 11b is a flow chart for explaining the cooperative quantity planning method according to the present invention

12 is a flowchart illustrating a sales analysis method using POS data according to the present invention.

Figure 13 is a flow chart for explaining the optimal volume distribution method according to the present invention

* Description of the symbols for the main parts of the drawings *

10: operation server 12: collaborative demand analysis and prediction subsystem

14: Cooperative Volume Planning Subsystem 16: Sales Analysis and Distribution Subsystem

18: graphical user interface 20: administrator computer terminal

30_1 ~ 30_n: Computer terminal of the store side 40: Internet network

100: database 200: enterprise resource management sub-system

Hereinafter, the supply chain management system and its operation method of the present invention will be described with reference to the accompanying drawings.

First, in the case of high-tech products, clothing, precious metals, accessories, miscellaneous goods, and furniture, the life cycle of products is extremely short, but the lead time of orders is extremely long, so the uncertainty of demand forecast is large and supply and demand imbalance can be severe.

Accordingly, the present invention minimizes the uncertainty of the demand forecast for the industry having the above characteristics in consideration of the following three aspects.

First, accurate demand forecasting:

By organizing views on the demands of distributors and merchandisers in contact with consumers by season, it enables optimal demand forecasting for each product category and automates the optimal assortment of each store.

Second, supply chain reaction speed is improved:

As far as possible, implement a category-specific automated product ordering module that delays the point at which product characteristics are determined to maximize the effect of risk-pooling. At the same time, it implements a system that makes the most of additional market information by enabling additional orders by immediately responding to each SKU product unit as needed under various price conditions and uncertainty conditions in multiple stages.

Third, inventory management minimizes loss of opportunity and effective management of parts and shortages:

In the case of expensive products with short product life cycles and high demands, the cost of opportunity loss due to shortages is high, so the approach of simply minimizing the cost or avoiding the worst case generates considerable opportunity cost. Therefore, when companies are aggressively developing markets, they evaluate and manage them scientifically.

As such, the supply chain management system of the present invention presents various modules that can be specifically realized for the above three aspects and forms a management support system that supports decision-making of the top management and the middle management by systematically organizing knowledge. It provides a high level solution to revolutionize the industrial competitiveness in this field.

First, the present invention has the following characteristics.

One). Cooperative Demand Forecasting and Customer Relationship Management (CRM):

Statistical access to consumers based on the Internet, and the use of optimization tools to systematically connect demand information from scientifically selected distributors and professional merchandisers and process the information in an interactive manner. Present a model.

In addition, by managing the demand patterns of the past season by category, it is possible to grasp the characteristics of the store and the characteristics of customers. This is the most important part of CRM. Furthermore, it implements a system that analyzes sales by store and period by improving demand information through POS data during the season.

2). Efficient order management and mass customization:

Recently, a small number of leading companies in Korea are increasing reaction production to switch to the QR method, but it is difficult to grasp the appropriate amount, and has undergone a lot of trial and error.

In the case of fashion products, the opportunity cost is very large, and the demand fluctuations are so large that it has a big impact on profit. Therefore, this study comprehensively considers the accuracy of demand information, the cost incurred by additional orders, the differences in demand patterns by category such as various sizes and color patterns, and the benefits of co-ordering.

In addition, the optimization module, which allows orders to be ordered in product categories or SKU units in two or multiple stages, enables management to manage individual units, and has advanced functions to realize mass customization by delaying the point at which product characteristics are determined. .

3). Inventory management to minimize opportunity loss through proper distribution of stores:

Fashion products with short product lifecycles and demanding preferences have a high cost of opportunity loss due to shortages, so a hedging approach that circumvents simple or worst-case scenarios, such as allocation and volume disposal by time-based sales ratios, Incurs significant opportunity costs.

Therefore, a system for scientifically evaluating and managing the market is urgently needed for the aggressive market development of a company.

Thus, the present invention proposes a distribution model that can maximize the profits according to the price difference for each product category, time, and implements the optimal assortment and automation of services for each store.

4). Enhance the operational capabilities of store operators by organizing knowledge by managing patterns and managing patterns of consumer responses to new product shipments each year, and at the same time, enabling continuous exchange of information and operation with operators in each store. And enable customer relationship management.

It also revolutionizes the industry's competitiveness in this field by supporting vendor decision-making in the form of intelligent management support systems.

Referring to the supply chain management system of the present invention having such characteristics in more detail as follows.

1 is a configuration diagram of the supply chain management system of the present invention, the operation server 10 consisting of various sub-systems for product demand analysis and prediction, quantity planning and distribution analysis and distribution, and the operation server 10 and A plurality of store-side computer terminals for inputting various data through the Internet network so that the system operator-side computer terminal 20 and the operation server 10 that are linked to each other can perform demand analysis and prediction, quantity planning, and sales analysis and distribution. (30_1,30_2,30_3, ..., 30_n) and the internet network 40 for communication between the operation server 10 and the store-side computer terminals 30_1,30_2,30_3, ..., 30_n. It is composed.

Here, the operation server 10 is a cooperative demand analysis and prediction sub-system 12, a cooperative volume planning sub-system 14, sales analysis and distribution sub-system 16 and information calculated from each sub-system It consists of a graphical user interface 18 for visually presenting to the administrator, which is interoperable.

Such a supply chain management system first analyzes demand through the collaborative exchange of information with merchandisers in each store, long before the system operator begins the sales season for the product. In other words, before the start of the sales season for a particular product, each store receives sales information on past seasons through the Internet network 16 and registers it in the database 100, and then analyzes and analyzes the characteristics of the store from the registered information. Demand analysis by product is analyzed as a whole by analyzing the accuracy of information by category and demand information.

Then, the information on the new product (catalog of the product, etc.) is provided to the merchandiser of the store through the Internet network 40, and the merchandiser of each store is based on the empirical knowledge from the information on the new product provided by the operation server 10. After predicting the demand for a new product, the predicted information is again provided to the operation server 10 through the Internet network 40.

Accordingly, the operation server 10 registers the demand forecast information provided from selected few representative stores or merchandisers of all stores in a database, and then, based on the registered information, forecasts total demand for new products, forecasts total demand for each store, and SKUs. Initial demand forecasting such as demand forecasting of units is carried out, and then, new products are sent to the test market immediately before the product is sold (1 ~ 3 weeks before) to be marketed, thereby correcting errors occurring in the initial demand forecasting process.

Afterwards, the company plans to establish a quantity purchase plan for raw materials and products based on POS data.The optimization method considering uncertainty allows multi-level decision making to reflect the uncertainty of information, and considering the possibility of mutual substitution and complementation between products. Implement a volume purchasing plan to minimize risks from risk-pooling.

Lastly, based on the POS data, which is continuously updated to maximize profits according to the price difference by product category and time period, the sales analysis and demand forecasting information is modified while the product is being sold, Allow for volume distribution. For reference, the notation code "200" indicates an enterprise resource planning department system (ERP), which is an integrated information system that effectively manages all human and material resources in a company and ultimately enhances the company's competitiveness. It is a representative enterprise reengineering technique that maximizes productivity by rebuilding an integrated system of management resources covering all sectors such as production, sales, human resources, accounting, funding, cost, and fixed assets.

The software supplied to realize this enterprise resource management system is called an ERP package, and this package is a system that can be integrated and operated without having to process a separate interface by automatically connecting data when inputting from one system. .

Hereinafter, the supply chain management system of the present invention will be described in detail.

2 is a block diagram of a Collaborative Demand Analysis (CDA) and a Collaborative Forecasting Engine (CFE) according to the present invention.

As mentioned above, system operators can use Store Characteristics Analysis (SCA), a detailed category, to analyze product demand through collaborative information exchange with merchandisers in each store, well before the start of the sales season for the product. Detailed Category Analysis (DCA) and Information Accuracy Analysis ( IAA) are conducted.

That is, as shown in the figure, the sales information for the past season provided from each store through the Internet network 40 is registered in the database 100, and then based on this store characteristics analysis, detailed category analysis and demand information Analyze the accuracy.

Here, the store characteristic analysis (SCA) module 13a is based on the composition ratio of sales at each store in the past season before the initial quantity order after the concept and detailed characteristics of the product are determined in the initial planning stage. This is to select stores that can best reflect the overall sales.

The reason for analyzing the store characteristics is the foundation for estimating the distribution of demand without loss of information by efficiently combining the demand forecasting information of merchandisers in each store participating in the demand forecasting process to estimate the sales volume of each product in the actual season. To provide data.

Here, in collecting the demand forecast information of the merchandisers of each store, it is expensive to actually participate in the process of forecasting the merchandisers of all the stores, so it is necessary to extract a representative representative store according to the demand forecast and to bear the cost. Efficiently forecast demand without

At this time, the sales ratio by store is divided by period. The reason for this is that there is a significant difference in the cost of overstock or lack of inventory depending on the normal price, discount price, and additional discount sales. This is to select the best representative store by reflecting the cost of overstock or lack of inventory due to the difference. In this case, the system administrator wants a category of stores with similar sales rates. Sorting by number allows cost-effectiveness analysis.

In other words, through the data conversion using the sales data of the previous season registered in the database 100, the sales data for each SKU product unit or category of each store by the sales period, and at the same time sales price, inventory management costs, etc. Calculate the various overstock and undercosts associated with the discount.

Based on these data, stores are classified into various categories through optimization techniques as needed, and the characteristics of the classified stores are reported to the managers to help managers understand the characteristics of the stores and reflect them in the decision-making process. In doing so, it should be a guide for management on big uncertain issues.

The above analysis of store characteristics helps managers to manage each store differently according to the characteristics of the store.

First, it is possible to infer demand characteristics and influence factors by stores by classifying stores with similar characteristics.

Second, some representative stores can be selected before the season to support the forecasting of demand for seasonal products based on the prospects of each store's sales.

Third, during the season, we will improve the accuracy of the demand for seasonal products based on initial sales trends by some representative stores.

Fourth, analyzing the past season data for each store to identify the characteristics of the sales composition ratio of each product line, and through the estimated overpriced period, discounted overstock cost, and understock cost By identifying the level of lost sales, it should be used as a supplementary material for differentiated inventory policies based on the characteristics of the store.

In fact, in the collaborative demand analysis and prediction sub-system according to the present invention, in order to analyze the characteristics of the store, variable costs including the sales volume of each product, the sales price of each product period, and the inventory management cost of each product period in the data of the same season in the past Opt overstock for each product period (Product p is variable cost at time t + (price at time t-price at (t + 1))), inventory shortage cost of each product Upt (variation cost at time t price of product p ), The number of stores (I), the sales volume of product p at time point t of each store i, the total sales volume of each store i (Si), and the sales volume of product p at time point t of store i in the total sales volume Enter data from the database on the quantity of the rip.

At this time, the specific gravity (Ript) of the sales quantity of the sales quantity of the product p at the time t of the store i occupies the total sales quantity is calculated by the following formula.

Ript = Sipt / SUM_p {Sipt}

Here, Sipt represents the sales quantity of the product p at the time point t of the store i.

After inputting the data in this way, assuming that the normal price sales and discounted price sales are performed by period after one time of volume distribution, when the store named i distributes the products according to the sales ratio of the representative store named j, the normal price sales and the discounted price sales are calculated. The cost Dij at this time is obtained.

After deciding the number of representative stores, it decides which stores are classified based on which representative stores are classified. In this case, a decision is made to minimize the total cost of the oversold and undersized costs according to the normal and discounted selling prices for each period. do.

By performing these steps, the result is presented to the manager by plotting the effect of changing the number of representative stores to k = 1,2,3, ..., I and the cost of increasing the number of representative stores. do. In this case, the information may be additionally provided by enumerating a representative store named j belonging to a store named i at each stage.

In addition, in comparing the characteristics of each store, additional information may include location (metropolitan city center, metropolitan outskirts, small and medium city centre, etc.), population (population near store), weather (seasonal average temperature, rainfall, humidity, clear blue). Etc.), along with the classification of stores, so that they can be compared with the classification of representative stores, so that managers can easily understand the characteristics of the stores.

3 is a flowchart illustrating a method for analyzing store characteristics according to an operation method of a supply chain management system of the present invention.

As shown in FIG. 3, Sipt representing the stock shortage cost Upt information at time t of product p, excess inventory cost Opt information at time t of product p, and the sales quantity of product p at time t of store i The information and Ript information indicating the share of the sales quantity of the sales quantity of the product p in the total sales quantity at the time point t of the store i are input from the database (S301).

Subsequently, the retail price i and the discounted price sale amount are allocated when the store named i distributes the product according to the sales ratio of the representative store j based on the premise that the regular price sale and the discounted price sale are performed for each period after the volume distribution is performed once using the information input from the database. Calculate the cost Dij at this time (S302).

Subsequently, after making a decision to minimize the total cost including the excess inventory cost and the lack of inventory cost according to the normal selling price and the discount selling price for each period (S303), the representative store is selected (S304).

Subsequently, after analyzing the cost effect of the selected representative store (S305), it is determined whether the cost effect is satisfied (S306), and if not satisfied, increase the number of representative stores (S307), and then from step S303 to S305. Repeat the steps, and if the cost effectiveness is satisfactory, the output through the graphical user interface so that the administrator can easily visually recognize (S308).

Subsequently, the detailed category analysis (DCA) module 13b will be described.

Detailed Category Analysis (DCA) refers to managing products in units of SKUs. To break down the classification system.

In general, products can be categorized by style, color, material, size, etc., based on the ratio of sales in the same category. It is an analytical method that determines the number of types to match supply and demand by unit of SKU.

For example, if the method of analyzing the subcategory based on the size,

First, given the major classification The sales volume of each product in the category is classified by size to determine a product group having the same characteristics. In this case, in order to subdivide the product category by reflecting the characteristics of the composition ratio of sales by size, data on the difference in sales composition ratio by size of the same category product in the past season is used.

In the process of testing the difference in sales composition ratio by size according to the previously provided subdivision criteria within a given major category, the group that is not significant uses the same product composition ratio by size of each product based on the current category. However, for the significant groups, subcategories were classified by the optimal classification technique while increasing the number of subcategories (k), and the subcategorization criteria could be identified by graphing the characteristics of the factors for each subgroup. do.

If increasing the number of subcategory groups fails to find meaningful subclassification criteria, the classification criteria can be registered and used at the current category level.However, if a meaningful criteria is found, the subcategory groups (k) can be increased. Further refine.

The characteristics of this method are to automatically present subcategories of different characteristics by the optimization technique, so that managers can learn, identify, and adopt the characteristics of categories according to statistical criteria and optimal classification techniques.

Through the subdivision category classification, consumers' preferences can be distinguished more clearly. For example, the size-specific preferences may vary greatly depending on the pattern of the color and the pattern of the stripes. In other words, the vertical stripes will sell a lot of sizes to fit the body of the shorter person, the horizontal stripes will sell a lot of sizes to fit the body of the taller person. In addition, the dark color may be sold a lot of body size of a fat person, and reflects these characteristics to have different SKU units according to the characteristics within the same category to utilize the characteristics of the demand precisely.

Such a detailed category analysis (DCA) is not easy for merchandisers in each store participating in the demand forecasting process to estimate the sales of each product in the actual season to consider all the parts required for accurate prediction. Therefore, they have the ability to induce them to calculate demand forecasts in a wide range of categories and to complement the ability of demand forecasters (merchants in each store) by adding precise scientific analysis. For reference, this detailed category analysis is performed at the same time as the analysis of the store characteristics.

Here, the data inputted from the database to the cooperative demand analysis and prediction sub-system for the detailed category analysis include the following.

Sales by period of each product by category, sales price by period of each product, variable costs including inventory management costs by period of each product, excess inventory cost opt by period of each product, inventory shortage cost upt of each product, time point for size i of product p Data on the sales volume at t (Spit), the proportion of product p for size i of product p (Rpi) from the total size of product p (sales), and the total sales volume of product p (w) The transformation uses the following formula:

Rpi = Api / Wp

Dpq = SUM_i {Up1 * (Spil-Rqi * Wp) ⁺ } + SUM_i {Op1 * (Rqi * Wp-Spi1) ⁺ } +

SUM_i {Up2 * (Spi2- (Rqi * Wp-Spi1) ⁺ ) ⁺ } + SUM_i {Op2 * ((Rqi * Wp-Spi1) ⁺ -Spi2) ⁺ }

Here, Api represents the total sales ratio of the size i to the product p in each store.

After that, the number of categories is classified by changing the number of categories in consideration of the excess inventory cost and the lack of inventory cost according to the normal and discounted selling price by period, and at the same time, statistical analysis is performed on the homogeneity of the classified products. For example, if the product of each product group is classified according to the size, the test of the homogeneity using the difference in sales composition for each group is performed and judged at the 5% significance level to further classify the detailed category. Determine whether to proceed.

This sorting process is not until the difference in sales between product and size by product group is not significant or when no significant but special characteristic elements are found in the same category or the effect of classification is insignificant. Abort the classification process.

This detailed category analysis method will be described with reference to the drawings.

4A to 4B are flowcharts illustrating a detailed category analysis method according to an operation method of a supply chain management system of the present invention.

As shown in Figs. 4A to 4B, first, after determining the year and season of the analysis target (S401), one category is selected from the major categories of the product (S402).

Subsequently, a corresponding subclassification criterion is provided for a given large classification category selection value (S403), and the significance of the sales amount by size of each subclassification criterion provided in step S403 is tested (S404).

Subsequently, the characteristics of the corresponding factors for each subclassification criteria are presented in a graph (S405), and then there is a significant subclassification standard (S406). One (S407), if there is a significant classification criteria, set the number of subcategory groups (k) to 2 (S408) The product is classified into subcategories (S409), and then the homogeneity of all subcategories is tested (S410).

In this case, if there is no homogeneity, the subcategory group is increased by one more step (S411), and then, step S409 is again performed. If it is determined that the homogeneity is determined, the classification characteristics of the current subcategory are provided as a graph (S412).

Therefore, when the manager determines the detailed classification criteria by looking at the detailed category classification characteristics provided in the graph (S413), the detailed category analysis according to the present invention is completed.

Next, the third step in the cooperative demand analysis step, which is an accuracy analysis of demand information (IAA) module 13c, will be described.

Information Accuracy Analysis (IAA) tracks changes in the correlation between estimated sales and final demand by period for each product category, and graphically displays the pattern of accuracy of demand information through a graphical user interface. It provides a visual understanding of changes in accuracy, and provides visual information to classify the type of change and the relationship between the incidence of sales and the accuracy of demand information.

To do this, first select the product category. In other words, the classification standard that managers use conventionally Can be used as is, but Product categories can be subdivided according to the management's judgment. The correlation coefficient between the final sales and demand forecasting information for each time point is calculated to visually show the management the change in accuracy of the information.

Here, in determining the time point of the pattern analysis, in consideration of the initial prediction time, additional information time, and the closing time of the test market in the previous season, when the sales are in each store, for example, the daily and weekly units are determined. However, according to the additional order, the last point that can be distributed within the season is adopted as the range of the pattern analysis point.

After that, the correlation coefficient for each point of time between the final sales information and demand forecasting information is calculated to create separate data for each category, and the management can compare the patterns by showing the year and category data through the GUI. .

Such a demand information accuracy analysis method can be briefly described as the flow chart shown in FIG.

As shown in FIG. 5, after determining a large classification category of the product (S501), it is determined a time point of analyzing the demand information accuracy pattern for the corresponding products in the same season (S502). At this time, in determining the analysis time of the demand information accuracy pattern, the period is determined on a daily and weekly basis, including an initial prediction time point, additional information time point, a closing time of the test market in the previous season, and a period in which sales occur in each store. .

Later, during the same season, time-based demand forecasts for the products in the category were collected from the database. After inputting (S503), after inputting the final demand data of the corresponding products in the same season (S504), the correlation coefficient between steps S503 and S504 is estimated (S505).

Finally, the accuracy pattern of the demand information by year and category is presented to the manager in a chart (S506).

For reference, FIG. 6 illustrates GUI data that visually shows the accuracy of sales volume and demand information for each sales period of knitwear.

Next, the Collaborative Forecasting Engine (DFE) will be described.

The cooperative prediction sub-system 12b organizes the knowledge of store managers to efficiently accumulate and manage information. Furthermore, the cooperative forecasting sub-system 12b enables the demand forecasting in which the supplier and each store's information are integrated in a cooperative manner.

In particular, the fashion industry requires both the logical capacity of merchandisers to plan volume and the emotional ability to represent the sense of new fashion by design managers. Therefore, it is very difficult to have a competent merchandiser with both. In addition, even in the case of competent merchandisers, it is difficult to continuously develop skills and organize knowledge.

Accordingly, in view of the above difficulties due to the dependency of a small number of individuals, the above-mentioned difficulties are solved by enabling more systematic information management and accumulation of knowledge.

In reality, the information about each store's demand will be best understood by the store's operator, and this is accumulated with personal knowledge, and each of these store operators is virtually unchanged, so the function of the system due to human fluctuations is large. It is not affected.

Therefore, as the knowledge is accumulated based on the invariant of these store operators and the stability of the system, the highly intelligent It can be developed into a marketing system.

This collaborative forecasting engine (CFE) is composed of an initial demand forecast (IDF) module 15a and a test market demand forecast (TMF) module 15b, as shown in FIG. The prediction module is subdivided into total demand forecast (IDFT) 15c, store specific demand forecast (IDFR) 15d, and SKU demand forecast (IDFS) 15e.

Initially, Initial Demand Forecasting (IDF) module 15a refers to a cooperative demand forecasting method between store operators and suppliers for initial demand forecasting.

This initial demand forecast is conducted after the concept of the product is established and before the initial initial order of the product is made. At this point, there is almost no information on demand. It is almost impossible to forecast the demand of new products by item from historical data.

Therefore, information is obtained from the operators of each store who are sensitive to the trends and trends of professional merchandisers and consumers. In other words, the characteristics of the demand function of each store are assumed to be best understood by store managers, and the knowledge of these store managers is systematically organized to efficiently accumulate and manage information. Support for forecasting demand in a collaborative manner that integrates forecasts and information from each store.

In this regard, the IDF module 15a is based on information about representative stores that best reflect the overall demand selected by store characteristics analysis from last season's data. It performs three functions: store-specific aggregate demand forecast (IDFR) and SKU demand forecast (IDFS).

First, the total demand forecast (IDFT) will be described in more detail as follows.

The aggregate demand forecast (15c) is made up of style (for single size) and product (for complex size), and the initial aggregate demand forecast for each general category used by general companies before being classified into subcategories. Obtain For example, in the case of the same-size product such as a tie, this is the total demand forecast of the product line by category, which includes the same feature in the product line-up classified by the property. In addition, in the case of knitwear, the classification is made according to the organization, color, etc. of each product within each product, and there is a variety of complex sizes in the products belonging to one category of these classifications, so the single product is one category The demand forecast for each product unit corresponds to the total demand forecast.

In the procedure, assuming that the product category is determined to be C1, C2, C3, ..., Ck through the store characteristic analysis (13a), the analysis of the total sales volume by category for each store will be carried out in future stores and categories. Estimate the change in total sales for each star. This section provides a selection of appropriate forecasts from several model alternatives based on managerial experience. Therefore, the total sales estimate (RGic) by category for each store is determined.

Next, after determining the estimated total sales (RGic), the total sales of the group c to which the product p belongs to last year's data are inputted from the database of the store's expected sales amount (representative store or all stores) for each product last year. The ratio of sales to stores i gives the total demand (Fip) for all stores of product p in the last season, predicted by store i.

Next, we need to estimate the overall demand through the appropriate weighted average of the demand forecasts obtained from each store.To obtain the weight, the optimal weight is obtained from the demand and actual sales of the past products predicted based on store i through the optimization module. Obtain

Min SUM_p {Op * (SUM_i (Wi * Fip) -Dp) ⁺ + Up * (Dp-SUN_i (Wi * Fip))}

Wi = Decision Variable

Find Optimal Wi.

Subsequently, the store's expected sales of each product (Sip (new)) for each product this year are entered into the database, and then the total demand for all products of this year's product p (new), which is estimated by store i, is SFip (new). Obtain

Next, find the total demand forecast (Fp (new)) for all stores of product p (new) this year.

Fp (new) = SUM_i (SFip (new) * Wi)

Finally, the average and variance of the demand distribution for the new product is determined. First, the demand average of the new product P (new) is Ep (new) = Fp (new).

However, variance is obtained by the following procedure.

Demand Dispersion of New Product p (new) = Vp (new) = Ac * V'p (new)

Where V'p (new) is the sample dispersion value of SFip (new)

For reference, Ac is an adjustment coefficient of the product category c, which is a value obtained by adjusting the difference between the variance of actual demand and the estimated demand variance of the product category c last season.

This aggregate demand prediction method will be described with reference to the flowchart shown in FIG.

As shown in FIG. 7, when a category of a product is determined (S701), a regression analysis is performed on the total sales of each store and each category (S702).

Since then, there are a few After presenting the alternative (S703), the appropriate alternative is determined (S704), and then the total sales estimated amount (RGic) for each store and category is determined (S705).

Subsequently, after inputting the demand forecast value of the i-store operator for the product p of the new season (S706), the total demand expected amount of the product p of the new season predicted based on the i-store is determined (S707).

While the process proceeds to step S707, on the one hand, the i-store operator's sales amount prediction value (Sip) of the last season's product p and the actual sales amount (Gip) of the i-store for the product p's last season are input (S708). In step S709, the total demand of the product p of the previous season is determined based on the i store.

Thereafter, the weight Wi for the i-store prediction demand through the optimization module is calculated (S710).

Lastly, based on the total demand expected amount determined in step S707 and the weight (Wi) calculated in step S710, the total demand expected amount for the product p is determined in the new season, and the variance value of the demand is determined and entered into the database (S711). The total demand prediction according to the present invention is completed.

Next, the second store-specific forecasting module (IDFR) will be described.

The Initial Demand Forecast per Retailer (IDFR) (15d) finds the first-order forecasted demand Fip (new) at store i inferred for the new product p (new) in category c as follows:

Sales and total sales ratio of store i in category c to which new product p (new) obtained through regression analysis or last year's data p is Fip (new) is calculated as follows from the sales and the total sales ratio of the store i in the category c.

In other words, A formula of Fip = Fp * (RGic) / {SUM_i (RGic)} is used.

Such a store-specific aggregate demand prediction method can be simply described as a flowchart shown in FIG.

As shown in FIG. 8, the average and variance values of the total demand estimates for the product p are input from the database in the new season obtained through the aforementioned aggregate demand forecasting method (S801), and at the same time, obtained through the aggregate demand forecasting method. Regression analysis of the total sales amount of each store and category is received from the database (RGic) (S802).

Subsequently, based on a regression analysis result of the average, variance value, and total sales amount of each store and category of the expected total demand, the estimated sales amount (Fip) at the i store of the product p is calculated in a new season (S803). If presented in a graph (S804), the total demand forecast for each store according to the present invention is completed.

Finally, the SKU unit demand forecast (IDFS) module 15e of the initial demand forecast module is described as follows.

In the subcategory analysis module 13b, each of the products may be divided into a product style large category category into detailed product categories having homogeneity in size composition ratio. The coefficient of variation (Ei) of products in a particular item i (eg, size i) in each subcategory is used to calculate the mean and standard deviation of the aggregate demand for each SKU unit.

In the subcategory analysis, Ri and Ei (for example, Ei = coefficient of variation of size i and Ri = ratio of size i) are calculated for the products in each category.

Here, the average and standard deviation of the demand for the SKU of the new product p size i are obtained by the following equation.

Demand average = Mu_pi = Ri * Ep

Standard deviation = Std_pi = sqrt (Vp) * Ei * Mu_pi / sqrt {SUM_i (Ei * Mu_pi) ^ 2}

This SKU unit demand forecasting method is a flow chart shown in FIG. 9 and can be briefly described.

As shown in FIG. 9, the average and variance values of the total demand estimates for the product p in the new season obtained through the aforementioned aggregate demand prediction method are input from the database (S901), and at the same time, the detailed classification items obtained from the detailed category analysis. The variation coefficient Ei for the size i and the ratio Ri of the subcategory item i are input from the database (S902).

Subsequently, based on the average, variance value and coefficient of variation (Ei), and the ratio (Ri) of the detailed category item i, the SKU unit of the detailed classification item (size) i of the product p in the new season is inputted in step S901. By calculating the average demand (Mu_pi), the standard deviation (Std_pi) (S903), and presenting it graphically (S904), the SKU unit demand prediction according to the present invention is completed.

The above has described the initial demand forecasting module 15a among the two modules constituting the cooperative forecasting sub-system, and the following describes the second module, Test Market Forecasting (TMF) module 15b. Shall be.

The test market demand prediction module 15b is used just before the start of the season and predicts the sales volume of the previous season from the sales of the representative store.

In other words, in the period of one to three weeks just before the start of the season, several stores are selectively displayed in actual stores and the reactions of consumers are measured to estimate the future sales direction of the products.

In the case of a large number of stores, it is too expensive to actually display the products in every store to see the reactions of the consumers, and at the same time, it is very difficult to accurately identify the reactions of the consumers by distributing the products on time. Therefore, it selectively examines the response of consumers to a few stores (representative stores) and uses it as data for further orders as information on the overall sales.

The initial demand forecasting module 15a mentioned above reflects the store operator's response, but the test market demand forecasting module 15b reflects the direct consumer's response without filtering, rather than the store operator's response.

However, in terms of cost and management, not all stores can be tested, so it is important to select a representative test store that best reflects the overall sales trend. If a test store is unilaterally selected only for stores with large sales, the test itself will contain significant errors. Therefore, it is necessary to select an appropriate store according to various composition ratios of all stores.

In addition, high-priced fashion products have a relatively high opportunity cost due to lack of inventory and a relatively small management cost due to overstocking, but the obsolescence cost of shortening the product life cycle takes a considerable cost. Therefore, by selecting a test shop that appropriately reflects the two costs of overstocking and obsolescence of shortening the lifecycle, it is possible to have a reliable response survey.

Furthermore, the new optimal gross sales forecasts and sales forecasts at each store are derived by properly reflecting and combining the sales status at representative stores and initial sales forecasts of each store manager.

The test market demand prediction module 15b estimates a change pattern of sales according to a normal price, a discount price, and an additional discount based on sales data and initial estimated sales information of a test store. In addition, it is possible to analyze the cost effect considering the effect of improving the explanatory power of sales that occurs along with the change in the number of representative stores and the cost of the increase in the number of test stores.

Properly combining the initial sales response of some representative stores and the initial forecast sales data before the season improves the accuracy in predicting the normal and discounted sales volume for the entire period of sales. In particular, it maximizes demand forecasting power by combining consumer response and store operator forecast data in an optimal state.

At this time, the database input data for the test market demand prediction,

Determination of the test store candidates through the store characteristic analysis data, the test period, the number of test stores (k), and the unit to classify the product p (the part reflecting consumer's preference such as the style of the product, the color, the clothing, and the organization of the textile The difference in size may be excluded from the classification if it does not fall within the consumer's propensity to buy the product.) And the period of the regular price sale, the first discount sale period, and the additional discount sale period in the same season. Period data, etc.

For reference, the data input from the same season's data includes the number of stores (I), the sales quantity of product p (Sipt) at time point t of store i, the variable cost including period sales price, inventory management cost by period, and excess inventory cost by period. , The shortage cost, the quantity of sales of product p at point in time t of store i (Ript), the quantity sold (p jp) during the test period of product p in store j, test in store j This includes data on sales volume (Pj) during the period.

At this time, optimization for each time point is performed as follows.

Min SUM_ {p, t} {Opt * AApt + Upt * BBpt}

subject to

SUM_j {Wjt * Pjp} + IFp * Ht -AApt + BBpt = Spt for all p, t

Wjt> = 0 Decision Variable,

Lack of product p at time AApt = t.

Excess amount of product p at time BBpt = t.

IFp = initial demand forecast for product p.

Ht = weight for IFp.

Next, the estimated sales quantity TFpt at the time point t of the new product p using the data of the sales quantity (Np) at store j of new product p and the representative representative store to which each store i belongs to the optimization result is expressed by the following equation. You can get it,

TFpt = SUM_j {Npj * Wjt} + IFp * Ht

The estimated sales volume for the entire period at store i of new product p can be obtained by the following equation.

TFip = SUM_t TFpt * (Si / SUM_i {Si})

Si = SUM_ {p, t} {Sipt}

Hereinafter, the cooperative quantity planning sub-system among the cooperative demand analysis and prediction sub-system, the cooperative volume planning sub-system, and the sales analysis and distribution sub-system constituting the supply chain management system of the present invention will be described in detail.

FIG. 10 is a schematic diagram of a collaborative quantity planning sub-system according to the present invention. When knowledge of demand is systematized, a distribution of future total demand at each volume purchasing decision point may be determined, and numerical information about uncertainty may be determined. Have. Information on the distribution of demand for each store is also established.

In this regard, the collaborative quantity planning sub-system establishes a quantity purchase plan for raw materials and finished products based on data obtained by the supplier of the product through cooperative relationships with the operators of each store.

In particular, it is difficult to make rational decisions in the volume planning of fashion products when merchandisers face the problem of uncertainty. Generally, decisions are made that do not reflect the concept of uncertainty, but in the present invention, information about demand distribution based on emotional ability that represents the sense of new fashions of design managers or store managers is combined with logical decision-making tools. This allows you to have a system merchandiser with two abilities – emotional and logical.

In addition, even a competent merchandiser is very difficult to continuously demonstrate his or her ability to systematically develop and develop knowledge so that decision makers' cognitive abilities generally overcome the realistic limitation that it is difficult to consider three or four factors at the same time. Can develop into an intelligent system.

As shown in the drawing, the collaborative quantity planning sub-system includes an OPM2 (Optimal Procurement m-Stage) module (14a) and an OPMP (Optimal Procurement m-stage per-SKU) module (14b). Module 14a makes initial volume purchase decisions with quantified information on demand uncertainty, at which point the decision is made to purchase goods that can be used for a variety of purposes, such as raw materials (fabrics and subsidiary materials). Do it. In addition, decisions are made to secure volumes for each category without detailed product characteristics being determined.

In addition, based on information on the distribution of demand determined in cooperation with the operators of each store, the quantity planning sub-system establishes a quantity purchase plan for raw materials and finished products in two stages.

Risks that reduce risks through decision making in two stages (steps 1 and 2) that reflect information uncertainty through optimization techniques that consider uncertainties, and reduce risks by taking into account the interchangeability and complementarity of products under the same classification. Implement risk-pooling. In particular, in consideration of the point in time when the characteristics of the product is determined to have a structure that delays this point within the limit as possible to achieve an optimal decision.

The OP2m module 14b is composed of a first-order module and a second-order module, which have the precondition that the product group having the same characteristics is determined by classifying the sales volume of each product by size.

In order to calculate the initial quantity, the average value m_pi, standard deviation sig_pi data for the size i of each product or one product p in the product family, and the second order point prediction value calculated from the historical data and the final demand are calculated in the database. The correlation coefficient (rho) is entered, and the total order quantity (Qp) of the initial product p is obtained by solving a Stochastic programming problem that maximizes expected profit.

On the other hand, the second volume module means that all stores, representative store operators, or professional merchandisers re-establish the estimated sales volume by dividing the sales volume by size for each product.

Therefore, a new sales forecast (nm_pi) for the demand of the size i of the product p, and additional input data for the demand correlation coefficient (rho) and the average price (price) is needed.

In order to determine the order quantity Qpi of the size i of the secondary product p, a solution can be found through the traditional newsvendor problem.

Subsequently, an additional order of size SKU unit for product p may be repeatedly applied to the following OPmp module.

Next, the second module of the collaborative quantity planning sub-system will be described.

In making a quantity purchase decision with quantified information on demand uncertainty, no substitution between products is allowed, but a product that can be purchased under various conditions over multiple stages (steps 1, 2, ..., n). For example, the purchase decision on the tie and the like) is focused on.

Based on the information on the uncertainty of demand at each decision point, a quantity planning sub-system is used to plan the quantity purchase so that optimal decisions can be made in consideration of the accuracy and trend of the information.

First, the initial input data for determining the order quantity for each j = 1,2, ... m is the category information to which the product p in the SKU unit belongs, except that p belongs to the category c. There is a demand correlation coefficient (rho_cj), an average value (m_p), and a standard deviation (sig_p) between time-point predictions.

And the input data of j = 2,3, ... includes data that all stores, representative store managers or professional merchandisers have decided the expected sales volume for each store by SKU unit, and j-1 stage for SKU unit p by store. There are new sales forecasts at, the demand correlation coefficient (rho_cj), mean estimates, and standard deviation estimates between j-order point-in-time estimates from historical data.

Here, the estimated average value of step j is m_pj = m_p + rho_cj * sig_p * (Dpj-Mpj) / Tau_pj, and the estimated standard deviation value is determined by a formula such as sig_pj = sig_p * sqrt {1- (rho_cj) ^ 2}. Become.

Where Dpj is the realized sales for product p and Tau_pj is the standard deviation of the realized sales for product p.

Such a quantity planning method using the collaborative quantity planning sub-system of the present invention will be described below with reference to a flowchart.

11 is a flow chart for explaining the cooperative quantity planning method according to the present invention. First, when the detailed category is determined in the cooperative demand analysis sub-system (S1101), the demand information accuracy analysis for each time point is performed after the determined category. Obtain a correlation parameter for (S1102).

Thereafter, for each product, it is determined whether the product can be delayed at the point of time when the characteristics of the product are determined (S1103), and if it is determined that the delay is necessary, the initial quantity purchase plan and the secondary quantity purchase plan are established using the OPm2 module. If it is determined that there is no need to delay, the OPmp module is used to establish a multi-stage purchase plan.

In other words, if it is possible to determine whether to delay the point in time when the characteristics of the product are determined for each product, and if it is possible to delay the point in time when the characteristics of the product are determined, uncertainty for the initial volume planning, which is the first stage of the OPm2 module, is determined. After the optimization operation is performed through the reflected optimization technique (S1104), the initial volume order for the product group having the same characteristics is determined by classifying the sales volume of each product in the category by size (S1105).

After that, it is determined whether the initial quantity (step 1) is the order point (S1106), and if the order point is reached, the initial order quantity is sent to the supplier (S1107). Then, it is determined whether it is time to order the second quantity. That is, it is determined whether it is time to determine the quantity in SKU units of the product (S1108), and if it is time to determine the quantity, optimization is performed through an optimization technique reflecting uncertainty for each product in the category (S1109). The order quantity is determined (S1110).

Finally, it is determined whether the additional order is necessary (S1111), and if it is determined that there is no need for the additional order, the quantity plan by the Opm2 module is completed.

On the other hand, if it is determined in step S1103 that it is not possible to delay the time when the characteristics of the product for each product is determined, establish a quantity purchase plan in multiple stages using the OPmp module.

In other words, after determining the first order quantity, which is the first stage of the OPmp module, optimization is performed through an optimization technique that reflects uncertainty (S1112). Determine the first order quantity for the product line (S1113).

Thereafter, it is determined whether it is time to order the first quantity (S1114), and if it is time to order, the first order quantity is sent to the supplier (S1115). After that, it is determined whether the additional quantity of the order time (S1116), When it is determined that it is time to determine the quantity, the optimization operation is performed through an optimization technique reflecting uncertainty for each product (S1117), and the additional order quantity of each product is determined (S1118). In this case, the optimization is performed based on the latest sales information and updated demand information.

Subsequently, it is determined whether an additional order is necessary (S1119). If it is determined that there is a need for an additional order, the step S1117 is performed, and if it is determined that there is no need for an additional order, the quantity planning by the OPmp module is completed.

In the above, the cooperative demand analysis and prediction sub-system and the cooperative volume planning sub-system constituting the supply chain management system of the present invention have been described. Finally, the sales analysis and distribution sub-system will be described.

The sales analysis and distribution system consists of POS data sales analysis and demand information modification POS sales analysis and demand forecasting module (POS-SAF) and RTRS module for optimal distribution.

On the other hand, POS-based sales analysis and demand forecasting module of the sales analysis and distribution sub-system predicts the sales volume of each season from the sales of all stores (POS), and uses the test results of the last market before the season, and the sales products of each store at the beginning of the season. We estimate the normal price and the discounted price sales, and further classify and allocate detailed periods to obtain sales forecasts reflecting weekly and daily characteristics.

To do this, the data to be entered into the database includes:

First, the detailed category inputs C1, C2, ..., Ck, and the product p may be simply expressed as Ck = c, and the sales period is assumed to be 1,2, 3, ... T.

Sales quantity of product p at time point t of store i in the same season in the past (Sipt) and average value of sales volume at time point t of product category c of store i in the total sales volume REict, its standard Deviation values RSict, c product category nck.

Here, if the test market demand forecast value of the new product p of store i just before the season is Fip, the formula for the demand forecast by period of the new product performed at the beginning of the season is as follows.

SE_ipt = REict * Fip: Estimated Sales Value by Period

SS_ipt = RS_ict * Fip: Standard deviation value of sales ratio by period

And assuming that the demand is realized at a, the demand forecast formula for the remaining period of the new product is

SE_ipt = SUM_ {t '= 1, ..., a} Sipt' * RE_ict / SUM_ {t '= 1, ..., a} RE_ict': Estimated value over time

SS_ipt = SUM_ {t '= 1, ..., a} Sipt' * RS_ict / SUM_ {t '= 1, ..., a} RE_ict': standard deviation value

Here, Sipt 'represents the realized sales quantity of product p at time point t' of store i.

This sales analysis can use the test market results conducted just before the season to predict the normal price and the discount price sales volume of each product at the beginning of the season at the beginning of the season. The sales estimate reflecting this is available.

The sales analysis method using such POS data will be described in more detail with reference to the flowchart shown in FIG. 12.

First, the detailed categories (C1, C2, ..., Ck) of each product are read from the database at the values determined in S407 and S413, and the sales period (1, 2, 3, ... T) of each product is determined. (S1201).

Then, after inputting the sales quantity (Sipt) for the product p at the time point t of the store i of the last season (S1202), the share of the sales volume at the time point t of the product category c of the i store in the total sales volume The mean value REict and the standard deviation RSict are obtained (S1203).

Then, to the new product p in store i just before the season After inputting the test market demand prediction value (Fip) for (S1204), and based on this, the estimated average value and the standard deviation of the sales ratio for each period (S1205).

Subsequently, based on the time point when the demand is realized, the graph calculates and calculates the average sales forecasted average value and the standardized standard deviation of the forecasted sales for each period in which the demand is not realized (S1206), and judges whether or not it is currently in season (S1207). ), If not in season, end sales analysis; if in season, repeat S1206.

Next, the distribution (RTRS) module will be described.

Through the distribution module, managers can appropriately distribute the necessary quantity according to the demand characteristics of the store so that inventory replenishment can be achieved by store. The cycle time for replenishment for the distribution of the quantity is set in the unit period. And since the selling price, the lack of inventory cost, the purchase price of the product, and the excess inventory cost are different according to the time points, this information can be used to calculate the optimal service rate, and at the same time, considering the substitution effect between similar products of the same category. Let's determine the rate.

In addition, the stock replenishment cycle time is further classified into detailed periods to establish a distribution plan based on sales forecasts that reflect daily characteristics.

The distribution of demand at each time point t is determined based on the above-mentioned SEipt and SSipt, and the following parameter values and functions are derived.

Fij (Xij) = i Service rate when product is distributed Z quantity to j store

The minimum volume distribution that satisfies the target service rate when Uij = i

Minimal volume allocation to meet target service rates through Dj category j stores

Inventory of Si = i Product

The optimal resource allocation decision problem that maximizes the service function considering the substitution effect is made into a non-linear optimization problem and the solution is solved by the optimization algorithm.

The result is an optimal distribution that satisfies the target service rate for each product category at each time point t, and at the same time, distributes each i-product to j-store, which maximizes the overall expected profit considering the substitution effect of the product, even under the limited inventory of products. You can get the value.

The distribution method using the distribution module (RTRS) will be described with reference to the flowchart of FIG. 13 as follows.

First, after determining the inventory replenishment cycle time (S1301), the product shortage at the time of sale, the product shortage and excess inventory cost at that time are calculated from the product purchase price (S1302).

Subsequently, the demand distribution of the cycle time period reflecting the sales information realized up to the point in time is estimated by using the sales amount realized for each time point as information (S1303). In this case, the average and standard deviation of the demand is a value obtained through sales analysis using the above-mentioned POS data.

Thereafter, the target service level of each product in each store is determined, and the target service level of the replaceable product family in each store is determined (S1304). That is, the service level (Uij) for each product is determined, and the service level (Dj) for the product family that can be replaced in the j store is determined.

Subsequently, the optimal resource distribution decision problem according to the objective function considering the alternative demand is made into a non-linear optimization problem, and the solution is found by the optimization algorithm (S1305). Distribution to satisfy the target service rate per product category is performed (S1306).

After that, check whether the season has ended (S1307), and if the season has not ended, return to step S1303 to repeat the process.

As described above, the supply chain management system and its operation method of the present invention utilizes store information provided by merchandisers through the internet network to support the step-by-step demand of new product with short life cycle through cooperative demand analysis and forecasting and quantity planning. Implement indispensability, multistage volume planning considering differences in procurement price conditions, mass customization, and allocation to minimize opportunity losses in consideration of the characteristics of demand demand through substitutes.

In this case, the store information may be registered as a database of the operation server through the Internet network, but if necessary, such information may be registered in the database offline.

As described above, the supply chain management system and its operation method of the present invention has the following effects.

For products with extremely short product life cycles and long procurement periods, product demand analysis and forecasting, volume planning are established in a collaborative manner between system operators and store operators. By systematically combining sales analysis and distribution optimization, you can minimize opportunity losses in your stores and maximize profits.

Therefore, the demand itself is extremely uncertain and at the same time wider selection of products, the competition between products becomes more intense and competition is intensified with the emergence of competitors against the background of global environment and the cost advantage of global sourcing. In response to existing production methods, it is possible to provide an optimal supply chain management system that thoroughly eliminates risks in all processes, from forecasting demand for products with uncertainty to purchasing and distributing volumes.

Claims (31)

Operational server consisting of sub-systems for product demand analysis and forecasting, quantity planning, and sales analysis and distribution before and during the sales season of the product through the establishment of a database where store information is registered and cooperative information with the store manager. ; A plurality of store-side computer terminals selectively providing store information through an internet network so that demand analysis and prediction, quantity planning, and sales analysis and distribution by the operation server are performed; The manager side computer terminal communicates with the computer side of the store through the internet network and visually provides the manager with the results of demand analysis and forecast, quantity planning and sales analysis and distribution by the operation server. Supply chain management system, characterized in that comprises a. The supply chain management system according to claim 1, wherein the store information is registered in the database through an internet network, and the store information is used as data for operating each sub-system. A collaborative demand analysis sub-system which analyzes the demand of each product by analyzing the characteristics of the store, the analysis by the detail category, and the accuracy of the demand information by the cooperative information combination with the merchandiser of each store before the sale season for the specific product; Based on the demand forecasting information of the merchandisers obtained by providing the merchandise of the store with information on new products for each product Initial demand forecasting, including total demand for new products, total demand for each store, and demand in units of SKUs A cooperative forecasting sub-system for correcting forecasts for products by modifying the A collaborative volume planning sub-system that executes a volume purchasing plan for raw materials and finished products based on POS data in consideration of mutual substitution and complementarity between products so that multi-level decision making reflects uncertainty of information through optimization techniques considering uncertainty ; A sales analysis and distribution sub-system, which performs sales analysis and demand forecasting information correction during the sales season of products based on continuously updated POS data, and performs optimal distribution according to the demand characteristics of each store; Graphical user interface unit for visually providing the information calculated from the respective sub-system through the computer terminal of the administrator; Supply chain management system comprising an internet network for data transmission between a plurality of store operators and system administrators. According to claim 3, The collaborative prediction sub-system based on information on representative stores that best reflects the overall demand selected through the analysis of store characteristics in the data of last season (IDFT), total demand by store Initial demand forecasting (IDF) modules that perform forecasting (IDFR) and SKU demand forecasting (IDFS), and test market demand forecasting (TMF) that predicts sales for all seasons from representative store sales immediately before the product's sales season begins. Supply chain management system, characterized in that consisting of modules. 4. The OP2m module according to claim 3, wherein the collaborative quantity planning sub-system makes an initial quantity purchase decision for the goods through numerical information on demand uncertainty based on the information on the distribution of demand determined in cooperation with the store operator. And, supply chain management system, characterized in that it is composed of Opmp module that is not allowed to replace between products, but makes a purchase decision for products that can be purchased under various conditions in multiple stages. According to claim 3, The sales analysis and distribution sub-system using the test market results performed just before the season to predict the normal price and the discount price sales of the sales products for each period at the beginning of the season, and further classify the detailed period Analyze sales volume reflecting weekly and monthly characteristics, and set cycle time for replenishment of inventory, and calculate the optimal service rate using information on sales price, stock shortage cost, product purchase price and excess stock cost by time Supply chain management system, characterized in that to distribute the required amount according to the demand characteristics. Analyzing the demand for each product by receiving store information from a merchandiser of each store through an internet network before the sale season for a specific product begins, analyzing the characteristics of the store, analyzing the detailed categories, and analyzing the accuracy of the demand information; Based on the demand forecasting information of the merchandisers obtained by providing information on new products for each product to the store's merchandisers, the initial demand forecast is made through the total demand forecast assuming sales of the new product, the total demand forecast for each store, and the demand forecast in SKU units Correcting errors generated in the initial demand forecasting process based on demand information obtained by sending the new product to a test market and selling the product immediately before a sale season of the product to perform accurate demand forecasting of the product; Performing a quantity purchase plan for raw materials and finished products based on POS data in consideration of mutual substitution and complementarity between products in order to make a multi-level decision reflecting uncertainty of information through an optimization technique considering uncertainty; Supply chain management system, comprising the step of performing the optimal distribution according to the characteristics of the sales analysis and demand forecasting information and store during the sales season of the product based on the updated POS data continuously How to operate. The method of claim 7, wherein the store characteristic analysis, From the sales information for the previous season, the sales volume of each product by period, the sales price of each product period, the variable cost including the inventory management fee of each product period, the excess inventory cost of each product period, the inventory shortage cost of each product, the number of stores (I) , The sales quantity of product p at each time point t of each store i, the total sales quantity of each store i (Si), and the ratio of the sales volume of the sales quantity of product p to the total sales volume (Ript) at the time point t of the store i. Inputting information about the database from the database, Calculating the cost Dij at this time by calculating a normal price sale and a discounted sale price when distributing the product according to the sales ratio of the representative store j in the store i; Making a decision to minimize the total cost of the overstock and understock costs in accordance with the normal and discounted selling prices over time; Selecting a representative store, Analyzing the cost effectiveness of selected representative stores; Determining satisfaction of the analyzed cost effect; According to the result of the determination, the number of representative stores is increased to make a decision to minimize the total cost of adding up the overstock cost and the lack of inventory cost according to the normal selling price and the discount selling price for each period, or graphic the satisfaction of the cost effect. A method of operating a supply chain management system comprising the step of outputting so that the manager can visually easily recognize through a user interface. The method of claim 7, wherein the result of the store characteristic analysis is to estimate the distribution of demand from the demand forecasting information of the merchandisers in each store participating in the demand forecasting process in order to estimate the sales volume of each product in the actual sales season of the product. A method for operating a supply chain management system, characterized in that it is used as basic data. 10. The method of claim 9, wherein the demand forecasting report is provided by selected plurality of representative store merchandisers capable of providing reliability to the demand forecasting. The method according to claim 7, wherein the detailed category analysis is classified into styles, colors, materials, and sizes for each product, and is used as data for reasonably determining the number of representative stores based on the composition ratio of sales in the same category. Operation of the supply chain management system. The method of claim 7, wherein the detailed category analysis comprises: sales amount by period of each product by category, sales price by period of each product, variable cost including inventory management cost by period of each product, excess inventory cost by period of each product, and Out of stock cost (Upt), sales volume at time t for size i of product p (Spit), sales volume for size i of product p (rpi) of total size sales of product p (rpi), product p Supply chain management system characterized in that the analysis using the information on the total sales volume of the (Wp). The method according to claim 7, wherein the detailed category analysis is to test the significance of the homogeneity using the difference in the sales composition of each group through the detailed classification according to the size of the product of each product group to perform additional subclassification if it is outside the error range. , If the significance is within the error range operating method of the supply chain management system characterized in that no further classification. The method of claim 7, wherein the subcategory analysis, Determining the year and season of the analysis, Selecting one category from the major categories of products, Providing the corresponding classification criteria for the major category selection values; Testing the significance of the sales amount by size of the provided subdivision criteria; Graphically presenting the characteristics of the relevant factors by subclassification criteria for each major category; Checking for the presence of significant classification criteria, Selecting a current category or increasing the number of subcategory groups according to the verification result; Classifying the product into detailed categories by the selected category group through an optimization module; Determining homogeneity in all subcategories; Supply chain management, characterized in that the step of increasing the number of sub-category group according to the determination result of homogeneity to classify the product into the sub-category of the selected category group, or output the classification characteristics of the current sub-category How the system works. The method of claim 7, wherein the accuracy analysis of the demand information tracks the correlation change between the (estimated) sales amount and the final demand amount by period for each product category, and displays the demand information accuracy pattern for each time point and visualizes the accuracy change of the demand information. Supply chain management system operating method characterized in that the analysis of the occurrence of sales, the accuracy of the demand information and the type of change in the demand information by providing a form that can be recognized. According to claim 7, Analysis of the accuracy of the demand information, Determining the major categories of products, Determining when to analyze the demand information accuracy pattern for the corresponding category products in the same season; Leading data from the database for point-in-time demand forecasts and final demand data for these category products in the same season; Estimating a correlation between a demand forecast value and a final capacity for each of the corresponding category products; A method of operating a supply chain management system comprising the step of outputting a demand information accuracy pattern for each year, category. According to claim 7, wherein the total demand forecast to estimate the change in the total sales of each store and category, to obtain the total sales forecast (RGic) by store and category, and the expected sales amount of the operator by store for each product last year (representative store Or all stores), and then predict the total demand in all stores for a specific product of the previous season and the weighted average obtained from the expected sales and actual sales from each store operator. How to operate. The method of claim 7, wherein the aggregate demand forecast, Once the product category is determined, regression analysis of total sales by store and category is performed. Presenting alternatives to trends in total sales and determining appropriate alternatives; Determining total sales estimates for each store and category; Determining the total demand forecast for the new season's products based on the store (i) based on the demand forecasts predicted by the operator of the store (i) for the new season's product (p); During the above steps, the store operator's sales forecast for the last season's product (p) and the store's actual sales for the last season's product were read from the database to predict the store's sales in the last season. Determining total demand, Calculating weights for the demand forecast of the store through an optimization module; And determining a total demand forecast amount and a variance value for the product in the new season based on the total demand forecast amount and the weight. According to claim 7, wherein the total demand forecast for each store is determined by regression analysis of the sales and sales of the category c in the category c belonging to the product p and the sales of the store i in the category c belonging to the new product p (new) as the last year data; A method of operating a supply chain management system, characterized in that it predicts through the total sales ratio. The method of claim 7, wherein the total demand forecast for each store, Regressing the average value and the variance value of the total demand expected amount and the total sales amount of each store and category of the last season in the database; A method of operating a supply chain management system, characterized in that the product (p) in the new season comprises the step of calculating the expected sales in the store (i). The method of claim 7, wherein the SKU unit demand forecast classifies each product into a detailed category in a detailed category analysis, and uses the coefficient of variation Ei of the products belonging to a specific item i (size) of each subcategory. A method of operating a supply chain management system, characterized in that it is estimated by obtaining the average and standard deviation of the demand for the SKU unit. The method of claim 7, wherein the SKU unit demand forecast, Reading the average value and the variance of the total demand expected amount and the coefficient of variation and detail category for the subcategory items obtained in the detailed category analysis step in a database; A method of operating a supply chain management system, characterized in that it comprises the step of calculating the demand average, standard deviation for the SKU unit of the sub-category of the product (p) in the new season. The method of claim 7, wherein the demand forecasting of the test market selects a test store by reflecting two costs, a management cost according to excess inventory and an obsolescence cost due to a shortened life cycle. A method of operating a supply chain management system comprising combining initial sales estimates of store operators to calculate an optimal total sales forecast and a sales forecast at each store. The method of claim 7, wherein the test market demand prediction includes a test store candidate through the store characteristic analysis data, a test period, a number of test stores (k), a unit for classifying a product p, and a normal price selling period in the same season in the past, A method of operating a supply chain management system, characterized in that it is predicted from the information of the first discount sale period and the additional discount sale period. 25. The operation of the supply chain management system according to claim 24, wherein the determination of the unit for classifying the product p is performed based on the part reflecting consumer's preference such as the product's style, color, obedience, and fabric structure. Way. 25. The method of claim 24, wherein the information of the same season in the past includes: the number of stores (I), the sales quantity (Sipt) of the product p at the time t of the store i, the variable cost including the period sales price, the inventory management fee by period, and the excess inventory cost by period , The shortage cost, the share of the quantity sold of product p at the point in time t of store i (Ript), the quantity sold during the test period of product p in store j (Pjp), at store j Method of operation of the supply chain management system characterized in that it further comprises information about sales quantity (Pj), sales quantity (Np) in store j of new product p, j representative stores to which each store i belongs during the test period. The method of claim 7, wherein the multi-step decision making reflecting the uncertainty of the information, Information about the mean value (m_pi), standard deviation (sig_pi), and second order point estimate from the historical data for each product or one product p in the product family, and the correlation coefficient (rho) between final demand Determining the initial volume through Based on the new sales forecast (nm_pi), demand correlation coefficient (rho), and average price for the demand of size i of product p, a representative store operator or a professional merchandiser breaks down the sales volume by size for each product A method of operating a supply chain management system comprising the step of determining a secondary. The method of claim 7, wherein the quantity purchase plan is a category information (Suppose that p belongs to category c) belonging to the SKU unit product p at the time of purchasing decision of each product (assuming that p belongs to category c), j order from the historical data Using data on the demand correlation coefficient (rho_cj), mean value (m_p), and standard deviation (sig_p) between order-point predictions, the representative store manager or professional merchandiser decides the estimated sales volume for each store by SKU unit, Supply chain characterized by using the new sales forecast in the j-1 step for the SKU unit p for each store, the demand correlation coefficient (rho_cj), average value estimate and standard deviation estimate information between the j order order point estimate from the past data How to use the management system. The method of claim 7, wherein the sales analysis, Estimating the average and variance of the sales ratio by period and category; Forecasting demand by category, store and unit period during the period from A method of operating a supply chain management system comprising the steps of estimating demand for each category, store and unit period for the remaining period. 8. The method of claim 7, wherein the optimal volume allocation, Calculating service functions considering the substitution effect, out-of-stock cost, overstock cost, integrated service level, available capacity and sales volume by store and product; A method of operating a supply chain management system comprising the step of determining the distribution of each product for each store through optimization by category. 8. The operation of the supply chain management system according to claim 7, wherein the step of executing the quantity purchase plan is performed after the initial demand forecasting, and is executed again after performing the accurate demand forecasting using the test market. Way.