Technical Field

The present invention relates to a system for gathering data about a product in spare parts services, determining product attributes by using at least one mathematical model, and estimating which spare part will be used with reference to a failure call, in after sales services.

Background of the Invention

Today, each customer call made to a customer call center in a current status process of an after-sales customer support service is triggered upon a new failure log is created in a failure tracking system and/or CRM (customer relationship management) system. During this call, details of failure occurrence (for example, failure details from most general to most specific) are learned from customers. Then, customer details (for example, customer profile and location) are enriched and product details (for example, product code, material type, material group and product hierarchy) are removed from the previous product assembly history in a CRM system. Thereafter, the related failure log is assigned to a nearby technical service with regard to the customer’s location. Lastly, the technical service pays a feasibility visit to control the failure cause and the defective component. For the related failure incident, each customer visit is managed as a unique maintenance line item and the spare part consumption or maintenance workmanship for the related customer visit is set off to this line item.

Considering the studies included in the state of the art, it is understood that there is need for a system which enables to predict the class of categorical spare part to be used by means of classification-type algorithms. The Chinese patent document no. CN110909997A, an application in the state of the art, discloses a spare part demand prediction method and a spare part demand prediction system. The said invention provides a spare part demand prediction method: firstly a graph structure is built; wherein any node of the graph structure is used for representing a spare part or a machine; the relationship between spare parts, the relationship between spare parts and machines or the relationship between machines and spare parts are created by the data gathered from at least one new spare part and at least one historical spare part; based on the designed graph structure and the historical damage data of one historical spare part of any new spare part in the at least one new spare part, it is enabled to determine predicted damage data of any new spare part in the at least one new spare part and then to determine the demand quantity of any new spare part based on the predicted damage data of any new spare part. The invention further provides a spare part demand prediction device and electronic equipment.

Summary of the Invention

An objective of the present invention is to realize a system which enables to predict the optimum spare part in accordance with a new failure log and a customer profile reaching a call center and then to transmit it to a technical service.

Detailed Description of the Invention

“A Real-Time Spare Part Prediction System” realized to fulfil the objective of the present invention is shown in the figure attached, in which:

Figure l is a schematic view of the inventive system.

The components illustrated in the figure are individually numbered, where the numbers refer to the following: 1. System

2. Failure tracking server

3. Data warehouse

4. Enterprise resource planning server

5. Data preparation server

6. Modelling server

The inventive system (1) for gathering data about a product in spare parts services, determining product attributes by using at least one mathematical model, and estimating which spare part will be used with reference to a failure call, in after sales services comprises: at least one failure tracking server (2) which is configured to manage attributes about the failure logs created and to track customer visits; at least one data warehouse (3) which is configured to store the data gathered periodically by means of data transfer procedures; at least one enterprise resource planning server (4) which is configured to enable use of resources such as workforce, machinery and material required for production of goods and services; at least one data preparation server (5) which is configured to acquire data about products, problems and customers from data sources; and at least one modelling server (6) which is configured to obtain the related data from the data preparation server (5) and to make inferences related to the main underlying reasons for the spare part consumption.

The failure tracking server (2) included in the inventive system (1) is configured to manage main attributes about the failure logs, that are created at the operational level, at the header level and to track customer visits, that are paid for active or inactive failure calls, within the framework of maintenance and repair history. The failure tracking server (2) is configured to set off the spare part consumed in a customer visit or maintenance and repair workmanship to a related maintenance confirmation item in terms of quantity and time. The failure tracking server (2) is configured to carry out management of master data apart from operational data. The failure tracking server (2) is configured to manage data such as attributes of a customer who is provided with after-sales service, attributes of a spare part that is consumed in a customer visit and information of technical service who paid the customer visit or performed the first assembly, within the framework of a B2C (business-to-customer e-commerce) business model.

The data warehouse (3) included in the inventive system (1) is configured to store data such as product code, material type, material group, product hierarchy about a product. The data warehouse (3) is configured to extract and store the data created at the operational level in enterprise resource planning systems (ETL-extract, transform and load), periodically by means of designated data transfer procedures. The data warehouse (3) is configured to periodically acquire the data about the operation occurring on the failure tracking server (2) and the enterprise resource planning server (4) and the temporal master data and to store these data within itself.

The enterprise resource planning server (4) included in the inventive system (1) is configured to provide arrangement of business processes and data flow at the operational level. The enterprise resource planning server (4) is configured to operate by means of integrated management algorithms enabling efficient use of resources such as workforce, machinery and material required for production of goods and services in enterprises, and to comprise attributes of end products.

The data preparation server (5) included in the inventive system (1) is configured to extract the first raw data set from various data sources, in accordance with a designated SQL query procedure. The data preparation server (5) is configured to gather data from a data source of failure log keeping data such as failure log id, date and time of event, complaint or symptom codes, failure status, the related customer number and the number of defective product at the operational level of failure log. The data preparation server (5) is configured to gather data from the data source of maintenance item keeping the consumption of spare parts consumed in a customer visit and the confirmation information of maintenance-repair workmanship. The data preparation server (5) is configured to receive data from a product data source that keeps features being managed in the enterprise resource planning server (4) such as material code, material type, material group, product hierarchy, brand and product costing group. The data preparation server (5) is configured to receive data from a data source of product details that keeps information such as date of production, start of warranty date. The data preparation server (5) is configured to receive data from a customer data source that keeps basic information about a customer having a defective product such as customer profile, customer location information with district/province/region details. The data preparation server (5) is configured to associate each observation with a unique failure call and to characterize it at a higher level of abstraction. The data preparation server (5) is configured to gather consumption of spare part in each consumption item related to the same failure event, to transfer values of spare part in the related maintenance items, and to singularize repeating values by combining these values in a single derived target class. The data preparation server (5) is configured to extract only combinations of positive-associated spare part by using a predetermined filter algorithm. The data preparation server (5) is configured to convert target class values of combined spare parts into more refined forms according to rules of association being revealed as a result of the second step. The data preparation server (5) is configured to determine relatively important rules according to a predetermined minimum confidence parameter. The data preparation server (5) is configured to sort filtered association rules in a descending order according to criterion values of lift and confidence. The data preparation server (5) is configured to control whether the target class value of a combined spare part in a created data set involves both antecendent and consequent values mentioned in the association rules. The data preparation server (5) is configured to extract the resulting value from the target class value of a combined spare part and to process this new value obtained to a new target class, in case of involving. The data preparation server (5) is configured to use a combined target class value directly, in the event that the target class value of a combined spare part in the set thereof does not involve both antecendent and consequent values mentioned in the association rules. The data preparation server (5) is configured to ensure that a refined transaction set obtained is converted into an item set according to complaint codes, product hierarchies and attribute values of a spare part product group. The data preparation server (5) is configured to extract usage rules of spare parts at different levels of complaint code and product hierarchy according to criterion values of lift, confidence and support for the created item sets. The data preparation server (5) is configured to determine the rules that are relatively weak among usage rules of spare parts, in accordance with a predetermined algorithm.

The modelling server (6) included in the inventive system (1) is configured to classify characteristics of historical failure logs (particularly, complaint details), characteristics of defective product, information about the customer having the defective product and the most important (and dominant) factors and patterns that explain the main underlying fact (or phenomenon) for the spare part consumption for the spare part consumption according to the accountable technical service performing the assembly, within the framework of specific concrete rules or closed forms. The modelling server (6) is configured to determine attributes to be used in a spare part prediction operation and effect (weight) of these attributes on prediction. The modelling server (6) is configured to scan a neighbourhood for a training record of k that is closest to an unknown sample by means of a kNN (K- nearest neighbour) classifier and to identify the proximity as a distance function in the form of Euclidean distance, in the event that an unknown record is given in a spare part prediction step. The modelling server (6) is configured to digitize values of each attribute before calculation of distance. The modelling server (6) is configured to normalize categorical attributes by a digitization weight value assigned to each significant attribute by adapting the weight, interdimensional similarity obtained in the operation of attribute selection. The modelling server (6) is configured to measure the similarity between a new failure event and neighbour objects in the last dataset, by means of a predetermined algorithm. The modelling server (6) is configured to abstract the number per a value of spare part target class, the values of total and average similarity for the nearest-neighbour objects. The modelling server (6) is configured to determine the spare part values in the result list of prediction according to the number of repetition (frequency). The modelling server (6) is configured to determine the spare parts included in the result list of spare part prediction according to an average similarity value. The modelling server (6) is configured to conclude the result list of spare part prediction by means of a predetermined k-limit algorithm.

In the inventive system (1), information about a product are stored in the data warehouse (3) at first. The enterprise resource planning server (4) manages use of resources such as workforce, machinery and material required for production of goods and services. The information received from the failure tracking server (2) and the data warehouse (3) are made ready to be processed by the data preparation server (5) and the modelling server (6) analyses the related data by using various predetermined algorithms and creates a prediction list of spare parts.

With the intention, it is enabled to resolve a failure transmitted to a company, by means of a first feasibility visit; to reduce the average number of visits per failure; to reduce expenses of sales operation and consumption of unnecessary spare parts; and to improve the quality level of after-sales services.

Within these basic concepts; it is possible to develop various embodiments of the inventive system (1); the invention cannot be limited to examples disclosed herein and it is essentially according to claims.