System and method for scenario-driven optimization of sourcing cost

Abstract

To solve the problem that sourcing is a practice of searching for and selecting entities based on set criteria, and determining best supplier entities at the lowest costs can develop a competitive advantage, however, existing sourcing techniques lack effective ways to target best suppliers for right part bundling and to efficiently optimize sourcing costs for sustained savings.SOLUTION: The present disclosure leverages optimization methods using machine learning and technical data to provide computationally efficient solutions. Relationships between parts and corresponding attributes are obtained for parts bundling, and best supplier locations are mapped for selected part bundles. Sourcing cost minimization means is used in conjunction with an iterative process to grow or limit supplier capabilities by constraining order quantity, bundle combination, or number of supplier entities based on multiple alternative scenarios being defined to systemically minimize costs and manage risks driven by changes in customer demand or manufacturing location.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] (Related applications) This application claims priority to U.S. Application No. 17 / 524,220 filed November 11, 2021, the entire disclosure of which is incorporated herein by reference.

[0002] (Technical field) This disclosure generally relates to the analysis of supply chain costs using machine learning and optimization, and more specifically to a system and method for scenario-driven optimization of procurement costs. [Background technology]

[0003] Supply chain management is the management of the flow of goods and services, encompassing all processes that transform raw materials into final products. To maximize customer value and gain a competitive advantage in the market, it is necessary to proactively streamline the supply-side activities of a business. Parts procurement, also known as procurement, is a crucial component of supply chain management. Businesses that can find the most suitable suppliers at the lowest cost can develop a competitive advantage. To determine the lowest procurement cost, a scenario-driven approach that considers various input combinations is effective. However, existing procurement techniques lack an objective and effective method for predicting target costs combined with a systematic scenario-driven approach to select the optimal supplier for appropriate parts bundling and optimize costs. The large number of parts and supplier combinations results in very long computation times for optimization. This limitation restricts the number of scenarios that can be considered, reducing process efficiency. [Overview of the project] [Means for solving the problem]

[0004] Embodiments of this disclosure present technical improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, one embodiment provides a processor implementation method for bundling parts, selecting corresponding suppliers, and optimizing overall procurement costs. This method can improve the functionality of the current system by reducing computational execution time for optimization, saving time and computational resources, enabling the consideration of more scenarios, and ultimately providing a more efficient solution. The method comprises the steps of: one or more hardware processors receiving input requests from one or more businesses; one or more hardware processors querying a parts repository based on the input requests to identify multiple parts; one or more hardware processors applying Jacquard similarity to each set of parts from multiple parts by calculating similarity scores between each set of parts from the multiple parts to obtain one or more combinations of similar parts sets; and, based on the similarity scores, identifying at least one corresponding supplier from a supplier repository and tagging at least one corresponding supplier to each of the one or more combinations of similar parts sets. The process includes: assigning; one or more hardware processors performing a comparison between a similarity score and a threshold; one or more hardware processors identifying one or more target suppliers from one or more combinations of sets of similar parts and at least one supplier based on the comparison; one or more hardware processors determining the procurement cost of each set of similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios; and one or more hardware processors optimizing the total procurement cost derived from the procurement costs of each set of similar parts based on one or more constraints.

[0005] In one embodiment, one or more predefined scenarios include (i) an overall cost reduction for a specific combination set of parts and one or more corresponding target suppliers, (ii) at least one potential supplier identified to fulfill parts for a particular business under a predefined quantity boundary, (iii) at least one potential supplier identified to fulfill parts for a particular business under a given expenditure opportunity, (iv) a maximum number of potential suppliers identified for delivery across the entire business, or (v) at least one combination thereof.

[0006] In one embodiment, one or more constraints include at least one of the following: requirements, maximum quantity of parts, minimum quantity of parts, maximum expenditure, and maximum number of suppliers.

[0007] In one embodiment, the method further includes the step of predicting the target cost of subsequent time instances using a neural network autoregressive input (NNARX) model, wherein the NNARX model comprises at least three layers, the first input layer comprising neurons representing historical values ​​of costs, as well as current and historical values ​​of external driving variables, the last output layer comprising neurons representing the target cost, and at least one hidden layer comprising neurons configured to compute the node weights of the NNARX model with a normalized linear activation function, each node of the hidden layer being connected to one or more nodes of the input layer and nodes of the output layer; further, the method comprises the step of splitting the input data into a training set and a test set; and further, training... The process includes steps of: fitting an NNARX model to a set; calculating predicted costs and prediction errors (measured by mean absolute percentage error or MAPE) using the resulting model for a test set; predicting target costs for subsequent time instances using a model with a prediction error of less than 5%; comparing (i) the predicted target costs for subsequent time instances with (ii) optimized procurement costs for each of one or more predefined scenarios; and identifying a focus scenario from one or more predefined scenarios based on the comparison, which serves as an output.

[0008] In one embodiment, the method includes the steps of: iterating until the optimized total procurement cost of a focus scenario is less than or equal to a target cost; identifying one or more potential objective improvements; modifying (i) the costs associated with each component and (ii) at least one of one or more constraints; and obtaining a new optimized cost for the focus scenario based on the modified costs and one or more constraints.

[0009] In another embodiment, a system is provided for bundling parts, selecting corresponding suppliers, and optimizing overall procurement costs. The system comprises: a memory for storing instructions; one or more communication interfaces; and one or more hardware processors connected to the memory via the one or more communication interfaces, wherein the one or more hardware processors receive input requests from one or more businesses by instruction; query a parts repository based on the input requests to identify multiple parts; apply Jacquard similarity to each set of parts from multiple parts by calculating similarity scores between each set of parts from the multiple parts to obtain one or more combinations of similar parts sets; identify at least one corresponding supplier from a supplier repository based on the similarity scores and tag at least one corresponding supplier to each of the one or more combinations of similar parts sets; perform a comparison between the similarity scores and thresholds; identify one or more target suppliers from the one or more combinations of similar parts sets and at least one supplier based on the comparison; determine the procurement cost for each set of similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios; and optimize the total procurement cost derived from the procurement costs of each set of similar parts based on one or more constraints.

[0010] In one embodiment, one or more predefined scenarios include (i) an overall cost reduction for a specific combination set of parts and one or more corresponding target suppliers, (ii) at least one potential supplier identified to fulfill parts for a particular business under a predefined quantity boundary, (iii) at least one potential supplier identified to fulfill parts for a particular business under a given expenditure opportunity, (iv) a maximum number of potential suppliers identified for delivery across the entire business, or (v) at least one combination thereof.

[0011] In one embodiment, one or more constraints include at least one of the following: requirements, maximum quantity of parts, minimum quantity of parts, maximum expenditure, and number of suppliers.

[0012] In one embodiment, one or more hardware processors, by instruction, predict the target cost of a subsequent time instance using a neural network autoregressive input (NNARX) model, the NNARX model comprising at least three layers, the first input layer comprising neurons representing historical values ​​of cost, as well as current and historical values ​​of extrinsic or external driving variables, the last output layer comprising neurons representing the target cost, and at least one hidden layer comprising neurons configured to compute the node weights of the NNARX model with a normalized linear activation function, each node of the hidden layer being connected to one or more nodes of the input layer and nodes of the output layer; further, the system is configured to perform a comparison between (i) the predicted target cost for a subsequent time instance and (ii) the optimized procurement cost for each of one or more predefined scenarios; and further, based on the comparison, identify a focus scenario from one or more predefined scenarios, the focus scenario serving as an output.

[0013] In one embodiment, one or more hardware processors are further configured to iterate through instructions until the optimized total procurement cost of a focus scenario is less than or equal to a target cost; identify one or more potential objective improvements; modify (i) the cost associated with each component and (ii) at least one of one or more constraints; and obtain a new optimized cost for the focus scenario based on the modified cost and one or more constraints.

[0014] In yet another embodiment, one or more non-temporary machine-readable information storage media are provided, which, when executed by one or more hardware processors, contain one or more instructions that trigger a method for bundling parts, selecting corresponding suppliers, and optimizing overall procurement costs. The method comprises the steps of: one or more hardware processors receiving input requests from one or more businesses; one or more hardware processors querying a parts repository based on the input requests to identify a plurality of parts; one or more hardware processors applying Jacquard similarity to each set of parts from the plurality of parts by calculating similarity scores between each set of parts from the plurality of parts to obtain one or more combinations of similar parts sets; and, based on the similarity scores, identifying at least one corresponding supplier from a supplier repository and tagging at least one corresponding supplier to each of the one or more combinations of similar parts sets. The process includes: assigning; one or more hardware processors performing a comparison between a similarity score and a threshold; one or more hardware processors identifying one or more target suppliers from one or more combinations of sets of similar parts and at least one supplier based on the comparison; one or more hardware processors determining the procurement cost of each set of similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios; and one or more hardware processors optimizing the total procurement cost derived from the procurement costs of each set of similar parts based on one or more constraints.

[0015] In one embodiment, one or more predefined scenarios include (i) a particular set of component combinations and an overall cost reduction of one or more corresponding target suppliers, (ii) at least one potential supplier identified to fulfill components for a particular operator below a predefined quantity threshold, (iii) at least one potential supplier identified to fulfill components for a particular operator with a given spending opportunity, and (iv) the maximum number of potential suppliers identified for delivery across all operators, or (v) a combination thereof.

[0016] In one embodiment, one or more constraints include at least one of requirements, maximum quantity of components, minimum quantity of components, maximum spending, and number of suppliers.

[0017] In one embodiment, the method further includes predicting a target cost for a subsequent time instance using a neural network autoregressive input (NNARX) model, the NNARX model including at least three layers, a first input layer including neurons representing historical values of cost and current and historical values of external driving variables, a last output layer including neurons representing the target cost, and at least one intermediate layer including neurons configured to calculate the node weights of the NNARX model with a normalized linear activation function, each node of the intermediate layer being connected to one or more nodes of the input layer and a node of the output layer; further including executing a comparison between (i) the predicted target cost for a subsequent time instance and (ii) the procurement cost optimized for each of one or more predefined scenarios; further including identifying a focus scenario from one or more predefined scenarios based on the comparison, the focus scenario functioning as an output.

[0018] In one embodiment, the method further includes repeatedly performing until the optimized total procurement cost of the focus scenario is less than or equal to the target cost; identifying one or more potential target improvement points; modifying (i) the cost associated with each component and (ii) at least one of the one or more constraints; and obtaining a new optimized cost for the focus scenario based on the modified cost and the one or more constraints.

[0019] It should be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention as claimed. The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and together with the description serve to explain the principles disclosed.

Brief Description of the Drawings

[0020] [Figure 1] An exemplary system for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs according to one embodiment of the present disclosure is shown. [Figure 2] An exemplary high-level block diagram of the system of FIG. 1 for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs according to one embodiment of the present disclosure is shown. [Figure 3] An exemplary flowchart showing a method for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs using the system of FIGS. 1-2 according to one embodiment of the present disclosure is shown. [Figure 4] An exemplary diagram of a component ontology according to one embodiment of the present disclosure is shown. [Figure 5] A dendrogram showing the results of component bundling according to one embodiment of the present disclosure is shown. [Figure 6] A flowchart showing a method for obtaining one or more combinations of sets of similar components according to an embodiment of the present disclosure is shown. [Figure 7]An exemplary flowchart illustrating a method for selecting one or more target suppliers according to embodiments of this disclosure is shown. [Figure 8] Figure 1-2 illustrates an exemplary neural network autoregressive input (NNARX) model having various layers, as implemented by the system shown in one embodiment of this disclosure. [Figure 9] This block diagram shows a process for obtaining a new optimized cost for a focus scenario using the iterative target cost seeker shown in Figure 2, according to one embodiment of the present disclosure. [Modes for carrying out the invention]

[0021] Exemplary embodiments are described with reference to the accompanying drawings. In each drawing, the leftmost digit(s) of the reference numeral identifies the drawing in which the reference numeral first appears. Where convenient, the same reference numeral is used throughout the drawings to refer to the same or similar parts. While examples and features of the disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the scope of the disclosed embodiments.

[0022] As mentioned above, supply chain management refers to the flow of goods and services and includes all processes that transform raw materials into final products. This requires proactively streamlining supply-side activities in order to maximize customer value and gain a competitive advantage in the market. Parts procurement, also known as parts procurement, is the process of finding and selecting individuals / businesses based on established criteria. Parts procurement is carried out in a variety of fields and for a variety of reasons. Businesses that can find the most suitable suppliers at the lowest cost can develop a competitive advantage. To determine the lowest procurement cost, it is effective to consider scenarios that examine various input combinations. Due to the large number of combinations of parts and suppliers, the computational execution time for optimization becomes very long. This limitation restricts the number of scenarios that can be considered, reducing the efficiency of the process. However, existing parts procurement technologies lack an effective method for targeting the best suppliers for proper bundling of parts and systematically negotiating optimal contract terms for sustainable savings.

[0023] Embodiments of this disclosure provide a system and method for addressing the above technical issues by bundling parts, selecting corresponding suppliers, and optimizing overall parts procurement, leveraging a clear combination of needs for production volume and capacity, along with detailed data on the technology and future volume forecasts of key parts for a future product portfolio, and cost structures across supplier locations. A parts bundler creates an optimal parts family with common technical attributes, and a target supplier selector maps the optimal supplier locations for the selected parts bundle. An iterative target cost seeker is further implemented by the system of this disclosure, which uses a procurement cost optimizer that includes multiple alternatives for expanding or limiting supplier capacity, for example, by constraining order quantities, bundle combinations, or the number of suppliers. This process can systematically manage risks caused by changes in customer requirements or manufacturing locations.

[0024] Similar reference letters always represent corresponding features throughout the drawings. Preferred embodiments are shown in each drawing, more specifically with reference to Figures 1 to 9, and these embodiments are described in relation to the following exemplary systems and / or methods.

[0025] Figure 1 shows an exemplary system 100 according to an embodiment of the present disclosure for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs. In one embodiment, the system 100 includes one or more hardware processors 104, a communication interface device(s) or input / output (I / O) interface(s) 106 (also referred to as an interface(s)), and one or more data storage devices or memories 102 operationally connected to the one or more hardware processors 104. The one or more processors 104 may be one or more software processing components and / or hardware processors. In one embodiment, the hardware processor may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuits, and / or any device that processes signals based on operation instructions. In other functions, the processor(s) may be configured to take in and execute computer-readable instructions stored in memory. In one embodiment, system 100 can be implemented on various computing systems such as laptop computers, notebooks, handheld devices (e.g., smartphones, tablet phones, mobile communication devices, etc.), workstations, mainframe computers, servers, and network clouds.

[0026] The I / O interface device(s) 106 may include various software and hardware interfaces, such as a web interface and a graphical user interface, and can facilitate multipoint communication over various network and protocol types, such as wired networks like LANs and cables, and wireless networks like WLANs, cellular, or satellite. In one embodiment, the I / O interface device(s) may include one or more ports for connecting multiple devices to each other or to another server.

[0027] Memory 102 may include any computer-readable media known in the art, such as volatile memory, including static random access memory (SRAM) and dynamic random access memory (DRAM), and / or non-volatile memory, including read-only memory (ROM), erasable programmable ROM, flash memory, hard disk, optical disk, and magnetic tape. In one embodiment, memory 102 stores a database 108, which includes information input request types, parts repositories, Jaccard similarity scores for parts, and so on. Database 108 further includes information such as (i) sets of similar parts to be combined, (ii) a supplier repository, and (iii) a selection of corresponding suppliers, target suppliers, procurement costs, and optimized procurement costs for input requests. Database 108 further includes data for one or more constraints and one or more predefined scenarios used to optimize procurement costs, each scenario being a digital representation of alternative yet unique combinations of input types and values. Furthermore, database 108 includes information corresponding to focus scenarios identified from one or more predefined scenarios, one or more potential objective improvements, and the cost of each component to be modified in order to obtain a new modification cost for each component. Furthermore, memory 102 includes one or more modules, such as a component bundler, a procurement cost optimizer, a target supplier selector, a target cost estimator, a neural network autoregressive input (NNARX) model, and an iterative target cost seeker. The aforementioned modules are implemented as at least one of a logically self-sufficient portion of a software program, a self-sufficient hardware component, and / or a self-sufficient hardware component having a logically self-sufficient portion of a software program embedded in each of the hardware components that, when executed, perform the methods described herein. Memory 102 may further include information relating to the inputs / outputs of each step performed by the systems and methods of this disclosure.In other words, the input(s) supplied at each step and the output(s) generated at each step are stored in memory 102 and can be used for further processing and analysis.

[0028] Figure 2, with reference to Figure 1, shows an exemplary high-level block diagram of the system 100 of Figure 1 for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs according to embodiments of the present disclosure. As referred to in Figure 1, the components / modules shown in Figure 2 are contained in memory 102 and invoked to perform the methods described herein.

[0029] Figure 3, with reference to Figures 1 and 2, shows an exemplary flowchart illustrating a method for bundling components, selecting corresponding suppliers, and optimizing overall procurement costs using the system of Figures 1 and 2, according to one embodiment of the present disclosure. In one embodiment, system(s) 100 comprises one or more data storage devices or memory 102 operationally connected to one or more hardware processors 104, and is configured to store instructions for performing each step of the method by one or more processors 104. Each step of the method of the present disclosure is described with reference to the components of system 100 in Figure 1, the block diagram of system 100 shown in Figure 2, the flowchart shown in Figure 3, and Figures 4 through 9. In one embodiment, in step 202 of the present disclosure, one or more hardware processors 104 receive input requests from one or more businesses (e.g., individual customers, corporate organizations (or B2B businesses), etc.). The input requests correspond to the purchase of one or more components. One or more components may correspond to any type of object. For example, parts can include machines, consumer products, vehicle components, manufacturing units, and assembly units. Those skilled in the art will understand that the above examples of parts should not be construed as limiting the scope of this disclosure.

[0030] Upon receiving an input request, in step 204 of this disclosure, one or more hardware processors 104 query a parts repository based on the input request to identify several parts. The parts repository includes parts lists, corresponding materials, processes, geometric features, tolerances, surface finish specifications, etc. The system queries the repository to obtain a complete list of parts that need to be procured. Table 1 below shows several exemplary parts identified from the parts repository. [Table 1] TIFF2023071638000002.tif177143

[0031] In one embodiment, in step 206 of the present disclosure, one or more hardware processors 104 apply Jacquard similarity to each pair of parts from multiple parts by calculating similarity scores between each set of parts from the multiple parts in order to obtain one or more combinations of similar sets of parts. In other words, the similarity scores between each set (or each pair of parts) of parts from multiple parts are calculated to obtain one or more combinations of similar sets of parts. Step 206 described above will be better understood through the following description.

[0032] Part bundling can be achieved by using a part ontology that includes various attributes from part data. Parts can be bundled using unsupervised machine learning methods for clustering known in the art, based on high similarity coefficients calculated, for example, using the Jaccard method. Figure 4 shows a part ontology used for bundling. More specifically, Figure 4 shows an exemplary diagram of a part ontology according to one embodiment of the present disclosure, with reference to Figures 1 to 3. The Jaccard similarity can be calculated as a ratio. TIFF2023071638000003.tif6170

[0033] In equation (1) above, the numerator is the number of common attributes between two parts A and B (intersection), and the denominator is the total number of unique attributes between the two parts A and B (union). For example, considering parts ID0 and ID1, the common attributes are shaft, metal, cylindrical, medium tolerance, machining, and turning, i.e., 6 common attributes. The unique attributes between parts ID0 and ID1 are shaft, solid shaft, metal, steel, cylindrical, medium tolerance, precision finish, machining, turning, grinding, hollow shaft, aluminum, hole, semi-finishing, and hole machining, totaling 15 attributes. The Jaccard similarity score is 6 / 15 = 0.4. Similarly, the similarity between parts ID1 and ID2 is (7 / 17) = 0.412, and the similarity between parts ID0 and ID2 is (5 / 18) = 0.278. Next, the system constructs a pairwise distance matrix where distance = (1 - Jaccard similarity score). In the example above, the distance matrix is ​​as shown in Table 2 below. [Table 2] TIFF2023071638000004.tif26139

[0034] Interpretation: The minimum distance is between part 1 and part 2, indicating that these parts are the most similar. They form the first bundle. Similarly, bundling of parts continues until all parts are included in one large bundle. This result is stored in a linkage matrix as shown in Table 3 below. [Table 3] TIFF2023071638000005.tif36143

[0035] System 100 forms part bundles by grouping the parts that are closest to each other. Once a group is formed, that group forms a new (virtual) part (see, for example, the first column above). This process continues until all parts are grouped into one large bundle. This result is best visualized by a tree diagram as shown in Figure 5. More specifically, Figure 5 shows a tree diagram illustrating the result of part bundling according to one embodiment of the present disclosure, with reference to Figures 1 to 4. In an example embodiment of the present disclosure, if the user wants to form two bundles, based on the analysis, P1 and P2 form bundle 1 and P0 forms bundle 2. Summary information following the part bundling step is shown in Table 4 below. [Table 4] TIFF2023071638000006.tif41164

[0036] Step 206 described above can be better understood from the flowchart in Figure 6 and the description of the embodiment above. More specifically, Figure 6 shows a flowchart illustrating a method for obtaining one or more combinations of sets of similar parts according to one embodiment of the present disclosure, with reference to Figures 1 to 5. As can be seen from Figure 6, first data corresponding to parts and ontologities is obtained, and the part data is converted into an ontology format having multiple attributes associated therewith. Next, pairs of parts are selected, and the intersection and union of their attributes are determined accordingly. Jaccard similarity is calculated based on the intersection and union of the part attributes. These steps are repeated until the last pair of parts. Once the Jaccard similarity scores have been calculated up to the last pair of parts, a distance matrix of distance = (1 - Jaccard similarity score / coefficient) is created. In other words, once parts are retrieved from the repository and represented by their ontology attributes, the parts are compared with each other in a pairwise manner / method, as shown in the flowchart in Figure 6, and intersection attributes (IA) and union attributes (UA) are determined. The Jaccard similarity coefficient is calculated as |IA| / |UA|. Once all pairwise similarity calculations are complete, a distance matrix is ​​calculated where the distance for each pair is (1 - similarity coefficient). Crossing this matrix, the parts with the smallest distances are bundled, effectively forming new parts. This process continues until all parts are identified in a bundled state.

[0037] Upon obtaining one or more combinations of sets of similar parts, in step 208 of this disclosure, one or more hardware processors 104 identify at least one corresponding supplier from the supplier repository based on the similarity score and tag at least one corresponding supplier to each of the one or more combinations of sets of similar parts. The supplier repository includes information such as a list of suppliers, along with location, material, and processing specifications. The information contained in the supplier repository is not intended to limit the scope of this disclosure. Table 5 below shows the supplier repository. [Table 5] TIFF2023071638000007.tif245150

[0038] The target supplier selector can use the unsupervised machine learning clustering method Jaccard similarity, but it can be applied to find similarity between each part bundle and different suppliers, as opposed to finding similarity between different parts. This similarity can be calculated and displayed as a "similarity score" for each bundle-supplier pair. In this process, it is possible to flexibly select the attributes to include in the similarity score calculation by selecting only the necessary columns from the supplier data file. For a given bundle, suppliers whose similarity score is at least or equal to a user-specified cutoff score can be identified as "target suppliers" for that bundle. The above explanation will be better understood by the following example. Calculation when supplier selection = Gyro and bundle = B1(P1, P2) Common attributes = {metal, machining, turning, drilling} All attributes = {aluminum, shaft, turning, gear, hole machining, steel, hollow shaft, metal, machining, gear cutting, spur gear, brass, milling, thread cutting} Similarity score = 4 / 14 = 0.286

[0039] Referring to the steps in Figure 3, in step 210 of this disclosure, one or more hardware processors 104 perform a comparison between the similarity score and a threshold. In step 212 of this disclosure, one or more hardware processors 104 identify one or more target suppliers from one or more combinations of a set of similar parts and at least one supplier based on this comparison. After completing similar calculations for all other bundle-supplier combinations in this manner, the results from target supplier selection are shown in Table 6 below. [Table 6] TIFF2023071638000008.tif93162

[0040] The bolded attributes in some cells of Table 6 indicate the target suppliers for each bundle that have a user-specified cutoff value of 0.28 (e.g., a threshold). The steps described above are better understood from the steps in the flowchart of Figure 7. More specifically, Figure 7 shows an exemplary flowchart illustrating a method for selecting one or more target suppliers according to embodiments of the present disclosure, with reference to Figures 1 through 6. As can be seen from Figure 7, following parts bundling, supplier data is retrieved from the supplier repository. For each bundle-supplier pair, the ratio of the intersection of attributes (IA) to the union of attributes (UA) is calculated to obtain a similarity score (S=IA / UA) (similar to that described for bundles). If the similarity score is greater than the specified cutoff value, the supplier is selected as a target supplier for that bundle. This process is repeated until all bundle-supplier comparisons are completed.

[0041] Once one or more target suppliers are identified / selected, in step 214 of this disclosure, one or more hardware processors 104 determine the procurement cost for each set of similar parts corresponding to one or more identified target suppliers based on one or more predefined scenarios. Once the procurement cost for each set of similar parts is determined, the total procurement cost is derived accordingly, and then the total procurement cost is optimized based on one or more constraints.

[0042] One or more predefined scenarios, each being a digital representation of a unique combination of input type and value choices, include: (i) an overall cost reduction for a specific set of parts and one or more corresponding target suppliers; (ii) at least one potential supplier identified to fulfill parts for a specific operator under a predefined quantity range; (iii) at least one potential supplier identified to fulfill parts for a specific operator under a given expenditure opportunity; (v) the maximum number of candidate suppliers identified for delivery dates across each operator; or (vi) at least one combination of these. One or more predefined scenarios and one or more constraints are illustrated in Table 7 below. [Table 7] TIFF2023071638000009.tif166143

[0043] Based on the selection of one or more of the above scenarios, the focus of negotiations is expected to narrow as follows: 1. Scenario 1: What are the overall cost reduction opportunities with a given parts bundle and target suppliers? 2. Scenario 2: Which supplier will deliver a cost-competitive product within the delivery range? 3. Scenario 3: Under the constraints of minimum delivery and maximum expenditure, which supplier can meet the supply requirements for the dedicated factory? 4. Scenario 4: Given the limitations on the maximum number of suppliers and minimum delivery quantities, which suppliers can meet the supply requirements for the dedicated factory? 5. Scenario 5: What is the maximum number of suppliers required across all factories to meet the delivery requirements? 6. Scenario 6: Given a given expenditure opportunity, which supplier can meet the supply requirements for the dedicated factory?

[0044] The attempt to obtain the appropriate scenario for selecting a "focus scenario" may depend on what the business is trying to achieve through parts procurement negotiations. For example, 1. Scenarios 1, 2, and 3: Target suppliers that meet the requirements and choose which suppliers to grow, control, or withdraw from. 2. Scenarios 1, 4, and 5: Consider single sourcing to "minimize the impact on the factory" and understand the potential impact on the overall cost increase. 3. Scenarios 1, 2, 5, and 6: Consider using a minimum number of suppliers to ensure delivery while understanding the impact on the overall cost increase.

[0045] The above predefined scenarios are better understood from the following illustrative explanation. 1. Unconstrained Scenario: In this scenario, the only constraint is the requirement for a specific part at the factory. This is the base scenario. X ijk - The quantity of bundle k shipped from supplier i to factory j (decision variable) C ijk - Unit cost for shipping bundle k from supplier i to factory D jk - Request for bundle k at factory j The objective is to minimize the total cost. The following conditions apply: TIFF2023071638000010.tif6170 or earlier. TIFF2023071638000011.tif121702.Minimum and Maximum Delivery Scenario: In this scenario, in addition to requirements, there are constraints on minimum and maximum deliveries. X ijk - The quantity of bundle k shipped from supplier i to factory j (decision variable) C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Request for bundle k at factory j MINQ im - Minimum quantity of category m parts to purchase from supplier i MAXQ im - Maximum quantity of category m parts to purchase from supplier i The objective is to minimize the total cost. TIFF2023071638000012.tif is subject to the following conditions. TIFF2023071638000013.tif 241703. Minimum delivery and maximum expenditure scenario: In this scenario, in addition to the requirements, there are constraints on minimum delivery and maximum expenditure. X ijk - Quantity of bundle k shipped from supplier i to factory j (decision variable) C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Requirement for bundle k at factory j MINQ im - Minimum quantity of parts of category m purchased from supplier i MAXS im - Maximum expenditure limit for parts of category m purchased from supplier i The objective is to minimize the total cost. TIFF2023071638000014.tif is subject to the following conditions. TIFF2023071638000015.tif 241704. Minimum delivery and maximum number of suppliers scenario: In this scenario, in addition to the requirements, there are constraints on minimum delivery and the maximum number of suppliers. X ijk - Quantity of bundle k shipped from supplier i to factory j (decision variable) C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Requirement for bundle k at factory j MINQ im - Minimum quantity of parts of category m purchased from supplier i MAXSUPP jk - Maximum number of suppliers of bundle k to factory j y ijk - Binary variable = 1 (when supplier i ships bundle k to factory j), otherwise = 0 A large number such as M - 1,000,000 The objective is to minimize the total cost. The following conditions apply: TIFF2023071638000016.tif6150 or earlier. TIFF2023071638000017.tif361705.Single Procurement and Maximum Delivery Scenario: In this scenario, in addition to the requirements, there is a constraint that each delivery must have a single supplier, along with a maximum delivery quantity. C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Request for bundle k at factory j MINQ im - Minimum quantity of category m parts to purchase from supplier i y ijk -Binary variable = 1 (if supplier i ships bundle k to factory j), otherwise = 0 Large numbers like M-1,000,000 The objective is to minimize the total cost. The following conditions apply: TIFF2023071638000018.tif6170 or below. TIFF2023071638000019.tif361706.Single Procurement and Maximum Spending: In this scenario, in addition to the requirements, there is a constraint that each delivery must have a single supplier, along with a maximum spending limit. C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Request for bundle k at factory j MAXS im - Maximum spending limit for category m parts purchased from supplier i y ijk -Binary variable = 1 (if supplier i ships bundle k to factory j), otherwise = 0 Large numbers like M-1,000,000 The objective is to minimize the total cost. The following conditions apply: TIFF2023071638000020.tif6170 or earlier. TIFF2023071638000021.tif36170

[0046] The following describes how at least some of the above scenarios are formed / predefined by System 100 of the present disclosure. For the sake of brevity and for a better understanding of the embodiments of the present disclosure, only a few scenario formations are shown by System 100. Those skilled in the art will understand that embodiments that define / form scenarios should not be construed as limiting the scope of the present disclosure.

[0047] A scenario explores different combinations of inputs and corresponding outputs. In a procurement cost minimization problem, various inputs include suppliers, destinations (factories, production units, or manufacturing units, etc.), parts or bundles with corresponding requirements, the cost of supplying one unit of a given part or bundle to a given supplier and destination, minimum / maximum limits on delivery quantities, minimum / maximum limits on expenditures, and minimum / maximum number of suppliers. In one exemplary embodiment, a scenario is formed by selecting and combining a specific subset of the above inputs. Some inputs are required in all scenarios (e.g., suppliers, destinations, requirements, etc.), while others are arbitrary combinations (e.g., minimum delivery quantity and maximum expenditure, etc.). Generally, the total procurement cost should tend to increase as the number or combination of inputs increases.

[0048] The following outline is an example of an approach to selecting the structure of several predefined scenarios. Step 1: The unconstrained scenario evaluates opportunities based on supplier quotes regarding procurement. Step 2: Minimum and maximum delivery scenarios analyze the competitiveness of individual suppliers in terms of the minimum and maximum capacity they offer. The volume allocation and associated expenditures in the scenarios further clarify whether the available capacity can cover the required combination, and how much it will cost to cover it across the entire component set. Negotiations among (or across) the businesses (suppliers, manufacturers, consumers, etc.) should also take into account the suppliers' track record and cost structure to identify suppliers who can grow with additional volume or expenditure, suppliers who need to be fixed in terms of expenditure or capacity control, or suppliers who may be able to withdraw with further volume and expenditure reductions. The minimum and maximum delivery scenarios can be further utilized by changing the boundaries, for example, by lowering the maximum delivery for a supplier that is withdrawing from the market and raising the minimum delivery for a growing supplier. This allows for an analysis of the impact on total costs from the perspective of the competitive costs that suppliers can offer. Step 3: The minimum delivery / maximum expenditure scenario works by changing the delivery or expenditure boundary, allowing for further analysis of the balance of volume and cost across the entire allocation to improve opportunities across the entire “growth / fixed / exit plan.” Once the optimal solution (for example, the optimal solution refers to expenditure and allocation combinations across suppliers) is compared, this can drive negotiations between businesses (or entities such as suppliers, manufacturers, and consumers) with a better understanding of the opportunities to reduce total costs and manage the “growth / fixed / exit plan.” In other words, each of these solutions, such as expenditure across suppliers and volume allocation combinations across suppliers, is compared to determine which of them provides the best results with respect to one or more strategies such as growth, fixed, and exit plans.

[0049] Using machine learning techniques for selecting component bundles and target suppliers reduces the number of decision variables in optimization, leading to a reduction in computational execution time. As described above, one or more constraints include at least one of the following: requirements, maximum quantity of components, minimum quantity of components, maximum expenditure, and number of suppliers. Furthermore, those skilled in the art will understand that the above-described predefined scenarios should not be construed as limiting the scope of this disclosure. In other words, other scenarios can be evaluated and defined in real-time / near-real-time or well in advance, depending on the requirements (e.g., based on input requests from customers / users).

[0050] Steps 214 and 216 above are better understood from the following explanation. The procurement cost optimizer implemented by system 100 (contained in memory 102, see block diagram of system 100 in Figure 2) uses the parts bundle and the unit costs from one or more specific / selected target suppliers to determine the minimum procurement cost for various constraints and combinations thereof defined by the scenario. Procurement costs can be minimized using optimization algorithms, as is known in the art. The optimization problem can be set based on selected business objectives and constraints. Simulation algorithms can be an alternative option for optimization. Below is an example of using optimization to minimize procurement costs, using minimum and maximum quantity limits from selected suppliers.

[0051] In this example, the constraints are the minimum and maximum quantities of a specific category that can be purchased from the supplier, with default values ​​of "zero" and "unlimited," respectively. X ijk - The quantity of bundle k shipped from supplier i to factory j (decision variable) C ijk - Unit cost for shipping bundle k from supplier i to factory j D jk - Request for bundle k at factory j MINQ im- Minimum quantity of category m parts to purchase from supplier i MAXQ im - Maximum quantity of category m parts to purchase from supplier i The objective is to minimize the total cost. The following conditions apply: TIFF2023071638000022.tif6170 or earlier. TIFF2023071638000023.tif24170

[0052] Output decision variable value X ijk The objective function TIFF2023071638000024.tif6170 can be used to calculate total procurement costs. This is the sumproduct of the unit cost and quantity for each supplier-parts bundle-factory combination.

[0053] For the sake of brevity and for a better understanding of the embodiments of this disclosure, the relevant systems and methods described for the next steps use the special case in which each component is in a bundle by itself and all suppliers are subject suppliers for all components. The procurement cost optimizer uses two input files. In the following files, only the first few rows and columns of data from the data files are shown to illustrate the input / output structure, and the use of such data is not to be construed as limiting the scope of this disclosure. 1) Consider LaneCosts.csv - Table 8 below includes the following columns, which have mostly self-evident meaning. a. Variables: x_1, x_2, x_3, ... => Number of decision variables. In the example use case scenario, there are 6 parts / bundles, 4 suppliers, 4 factories, and 96 decision variables. b. Supplier: c. Supply destination: Factory d. Parts: Parts or bundles e. Cost: The unit cost (in US dollars) for procuring a part or bundle from the supplier to the customer. f. Category: User-defined or calculated grouping of parts [Table 8] TIFF2023071638000025.tif156152 2) Consider constraints.csv - Table 9 below describes the constraints for the optimization problem. The last column (N) of each row is the RHS value of the constraint, and the (N-1)th column is the mathematical symbol (<=, >=, =) representing the relationship between the constraints. Columns 1 through (N-2) contain the constraint coefficients of the decision variables. [Table 9] TIFF2023071638000026.tif76168

[0054] For each scenario, there are two main output files from the procurement cost optimizer (among several). The first output file is the value of the objective function (Equation 2). This is the minimum parts procurement cost obtained under the constraints described in the scenario (Equations 3 to 6). The output of this scenario 2 (also called MinMaxQty) is shown below, and is stored in MinMaxQty_optobj.csv in memory (not shown in the figure). Optimal total cost of parts procurement - MinMaxQty = 1,962,100,000 The second output from the procurement cost optimizer is the value of each decision variable, i.e., the quantity of each supplier / part / factory combination required to achieve the minimum procurement cost mentioned above. This is stored in MinMaxQty_optout.csv, an example of which is shown in Table 10 below. The sum of the "total_cost" column is equal to the optimal cost of 1,962,100,000. [Table 10] TIFF2023071638000027.tif252170 TIFF2023071638000028.tif61170

[0055] Those skilled in the art will understand that the predefined scenarios and one or more constraints described above are examples and should not be interpreted as limiting the scope of this disclosure. In other words, depending on the input requirements and the level of involvement and negotiation across the entire business, the scenarios and one or more constraints can be configured and modified to meet the requirements.

[0056] Once the total procurement cost is optimized, one or more hardware processors 104 predict the target cost for subsequent time instances using a neural network autoregressive input (NNARX) model (also known as a neural network or neural network model, used interchangeably herein). The NNARX model comprises at least three layers: a first input layer containing neurons representing the historical values ​​of the cost and the current and historical values ​​of one or more external driving variables; a final output layer containing neurons representing the target cost; and at least one hidden layer containing neurons configured to compute the node weights of the NNARX model using a normalized linear activation function. Each node in the hidden layer is connected to one or more nodes in the input layer (depending on the number of nodes present in the input layer) and to the nodes in the output layer. The above steps and the neural network model can be better understood by the following description.

[0057] Setting target (total) costs for scenario improvement is a critical technical issue addressed using supervised machine learning approaches, and this disclosure uses a neural network autoregressive input (NNARX) model to predict target total costs for current or future time periods. The NNARX implemented by the system and method of this disclosure is shown in Figure 8. More specifically, Figure 8 shows an exemplary representation of a neural network autoregressive input (NNARX) model having various layers, as implemented by the system in Figures 1 to 2, according to embodiments of this disclosure, with reference to Figures 1 to 7. More specifically, the neural network autoregressive input (NNARX) model has at least three layers, including an input layer, one or more hidden layers, and a final output layer. Here, in addition to the total (procurement) cost on a historical basis, the price of raw materials on a historical basis (e.g., steel price) is used as an exogenous input (or external driving variable). In a neural network model, the neurons in the first input layer represent the historical values ​​of costs, as well as the current and historical values ​​of external driving variables; the neurons in the final output layer represent the target cost; and the neurons in the hidden layers are used to calculate weights using a normalized linear activation function or ReLU. As shown in Figure 8, nodes in any two consecutive layers are interconnected. After defining the model as described above, the input data becomes a training set and a test set, and the parameters of the NNARX model are generated by fitting them to the training data. More specifically, the input data is divided into a training set and a test set, the NNARX model is fitted to the training set, and the resulting / fitted model is used on the test set to calculate the predicted cost and prediction error (measured through mean absolute percent error or MAPE), and the model then predicts the target cost for subsequent time instances with a prediction error of less than 5%. In other words, the generated model is used to predict the predicted cost on the test data. The prediction error is then calculated when evaluated using mean absolute percent error (MAPE). TIFF2023071638000029.tif18105

[0058] In this case, the model yields a prediction error of 4.56%, but such an error should not be interpreted as limiting the scope of this disclosure. Because the prediction error is less than 5%, this model is used to predict the target cost for subsequent time instances. Examples of input and output records are shown in Table 11. The total cost prediction in bold is used in the iterative target seeker method (described later). [Table 11] TIFF2023071638000030.tif244128 TIFF2023071638000031.tif203137

[0059] Those skilled in the art will understand that the training and testing setups described by the systems and methods of this disclosure should not be construed as limiting the scope of this disclosure. In other words, in an example embodiment of this disclosure, the training and testing of the NNARX model, and its suitability, can vary depending on the observed requirements and scenarios.

[0060] Next, one or more hardware processors 104 perform a comparison between (i) the predicted target cost for subsequent time instances and (ii) the optimized procurement cost for each of the one or more predefined scenarios, and based on the comparison, identify a focus scenario from the one or more predefined scenarios, which serves as the output (e.g., the target cost).

[0061] The focus scenario is selected based on the smallest difference from the target value. The base (unconstrained) scenario is an ideal scenario used to provide a baseline, so its total cost may generally be slightly lower than the target cost. In this case, the MinMaxQty scenario is selected, as shown in Table 12 (bold indicates reference values). [Table 12] TIFF2023071638000032.tif57145

[0062] Next, one or more hardware processors 104 repeatedly perform the following steps until the optimized total procurement cost of the focus scenario is less than or equal to the target cost. These steps include identifying one or more potential objective improvements, modifying (i) the costs associated with each component and (ii) at least one of one or more constraints, and obtaining a new optimized cost for the focus scenario based on the modified costs and one or more constraints. The steps described above are performed by an iterative cost seeker, as shown in Figures 2 and 9. The steps described above can be better understood from the following explanation and examples. Once a focus scenario is identified, the system 100 attempts to further reduce the total cost (towards the target) by modifying the inputs. In this case, a high cost in the focus scenario may be the result of (i) using a supplier with a high unit cost, or (ii) being forced to select a different supplier due to constraints such as minimum quantity (MinQty). To illustrate these differences precisely, the output of the focus scenario is compared to the base scenario, as shown in Table 13 below. [Table 13] TIFF2023071638000033.tif245168 TIFF2023071638000034.tif92163

[0063] Given the outputs of various cost optimizer scenarios and the target cost, the iterative target cost seeker selects the scenario with the cost closest to the target and identifies potential for further cost reductions. To this end, three scenarios were examined with respect to their total cost and the projected target procurement cost of US$19,559,400.

[0064] The iterative target cost seeker compares the output of each decision variable (supply_qty) in the focus scenario with the corresponding output in the base scenario, and highlights the difference as potential for improvement, as follows: For example, let's assume the following focus scenario: Target cost = 1957500000 Tolerance range for the value (%) = 0.2 Focus Scenario = MinMaxQty Total cost of the focus scenario = 1962100000 Difference rate between focus cost and target cost = 0.23499361430395915 The one or more identified potential objective improvements include: Part, Destination, BaseSource, BaseSupplyQty, BaseUnitCost, FocalSource, FocalSupplyQty, FocalUnitCost, UnitCostDiff, TotalCostDiff P4, Indianapolis, Marina, 100000.0, 1160, Marina, 50000.0, 1160, 0, 0.0 P4, Indianapolis, Marina, 100000.0, 1160, Lakes, 50000.0, 1180, 20, 1000000.0 P4, St. Louis, Marina, 25000.0, 1220, Marina, 15000.0, 1220, 0, 0.0 P4, St. Louis, Marina, 25000.0, 1220, Steelers, 10000.0, 1300, 80, 800000.0 P5, St. Louis, Marina, 30000.0, 1240, Steelers, 30000.0, 1280, 40, 1200000.0 P8, Cleveland, Marina, 140000.0, 1400, Marina, 90000.0, 1400, 0, 0.0 P8, Cleveland, Marina, 140000.0, 1400, Empire, 50000.0, 1470, 70, 3500000.0 P11, St. Louis, Marina, 40000.0, 1160, Marina, 10000.0, 1160, 0, 0.0 P11, St. Louis, Marina, 40000.0, 1160, Steelers, 30000.0, 1200, 40, 1200000.0

[0065] The improvements mentioned above can be illustrated in Table 14 with the following example regarding the requirements for P4 in Indianapolis. [Table 14] TIFF2023071638000035.tif141167

[0066] The first row shows that in the base scenario, the entire 100K requirement is sourced from Marina, the lowest-cost supplier. However, in the focus scenario, only 50K is sourced from Marina, with the remaining 50K sourced from another supplier, Lakes, at a unit cost $20 higher. This is because there is a minimum Qty requirement of 50K from Lakes. If the amount sourced from Marina can be increased, it may be possible to get closer to the goal with a cost reduction of (20 * 50,000) = $1 million.

[0067] The above information is available here, and you can negotiate with Lakes to bring costs closer to the base cost of $1160 or reduce the MinQty requirements. Either option or a suitable combination will help reduce costs. In other words, (i) the costs associated with each component and (ii) at least one of one or more constraints will be modified based on the above available information (e.g., negotiations between businesses such as the business organization / vendor / supplier).

[0068] Once the new inputs are determined, a new scenario is constructed and can be executed through the source cost optimizer to further reduce the total procurement cost toward the target. In other words, the new optimization cost for the focus scenario can be obtained (or acquired) based on the modified cost and one or more constraints as described above. The above steps for obtaining the new optimization cost for the focus scenario can be better understood in the following explanation.

[0069] Example: Suppose the user reduces the unit cost of P4 from Lakes to Indianapolis, and the new cost is $1170 instead of $1180. Re-optimization yields the following result, and the new optimal total cost is $1961600000. This new optimal total cost is closer to the target value of $1957500000 than the focus scenario's $1962100000. However, since the target value has not yet been reached, the target seeker recommends further iterations. The above explanation can be realized with a block diagram of an iterative target cost seeker as shown in Figure 9. More specifically, Figure 9 is a block diagram of the process of a method for obtaining a new optimized cost for a focus scenario using the iterative target cost seeker of Figure 2, according to an embodiment of the present disclosure, with reference to Figures 1 through 8. As can be seen from Figure 9, following the selection of part bundles and target suppliers, an initial set of scenarios can be processed via the analysis engine, and total cost metrics can be calculated for each scenario to obtain a set of metrics. These collected sets of metrics can be compared to the target total cost. The total cost of the base scenario is generally slightly lower than the target cost, as it constitutes an ideal scenario used to provide a baseline. One of the initial scenarios (other than the base scenario) can be selected for recommendation or further refinement. This selection is rule-driven, and the rule may be "select the scenario with the smallest total cost difference from the target cost." The selected scenario is usually called the focus scenario. If the total cost of the focus scenario is lower than the target cost, further refinement may not be necessary. System 100 can recommend this to the decision-maker as the preferred scenario. If the total cost of the focus scenario exceeds the target cost and a predefined tolerance, further refinement may be necessary. Refinement can be done by modifying inputs and / or constraints. The recommended quantities for a scenario can be compared to the recommended quantities for the base scenario for each category. Since the base is an ideal scenario where each requirement is sourced from the lowest-cost supplier, any deviation from this case may warrant a more thorough examination.These differences are listed as potential target improvement opportunities, along with the calculated improvement costs. If a re-optimization decision is made (e.g., a user decision), the data may overlap with the focus scenario. After obtaining new cost or constraint data, the optimization can be rerun and the results compared to the target cost. This process can be repeated until the target cost is reached or the maximum number of iterations has been performed.

[0070] Existing methods lack an effective way to systematically negotiate optimal contract terms for sustainable savings by targeting the best suppliers that can provide the appropriate parts bundles. To proactively leverage rapidly changing technologies or the available capabilities of critical goods with constraints and choices, embodiments of the disclosure can enable targeting capable suppliers and potentially negotiating multi-year contracts with potentially larger obligations. Targeting the best suppliers and creating appropriate parts procurement bundles can be further enhanced by the iterative target cost seeker described herein, with a clear structure of contract elements relating to cost, capability, and, in some cases, opportunities to address complex business requirements. Driving such parts procurement contracts through a systematic negotiation process requires the use of core parts and supplier-related information, along with agreements on capability combinations and capability-building elements, and is further required to be carried out in an uncertain market environment that presents the technical challenges that can be overcome by the systems and methods of the disclosure.

[0071] The typical parts procurement cost for a manufactured product accounts for approximately 65% ​​of the cost of goods sold. Cost benefits obtained through supplier quotation analysis or negotiation are difficult to maintain due to constantly changing customer requirements, product offerings, and supplier business risks. Manufacturers are increasing their suppliers globally to find the optimal capabilities to meet the emerging technical requirements for their products, but they face the risk of managing cost fluctuations that directly impact profits or productivity, as there is no effective method for systematically negotiating optimal contract terms for sustainable savings by targeting the best suppliers for the right parts bundles. To address the technical problems described above, embodiments of this disclosure provide a system and method that leverages detailed data on the technology and future volume forecasts of key parts for a future product portfolio, as well as a clear combination of needs for production volume and capacity, along with cost structures across supplier locations. Using this data, the system of this disclosure provides relationships between parts and corresponding capacity opportunities. The parts bundles described herein create an optimal family of parts with shared technology and estimated total demand, while the target supplier selector implemented by the system maps the optimal supplier locations for the selected parts bundles. The iterative target cost seeker uses a procurement cost optimizer that includes multiple alternatives to increase or limit supplier capabilities, for example, by constraining order quantities, bundle combinations, or the number of suppliers. Furthermore, by performing the above method of the present disclosure, the system 100 enables the integration of data formats for parts (e.g., representing the attributes of each part and the quantity of each part), thereby enabling links across corresponding processing and supplier information.

[0072] This specification describes the subject matter so that those skilled in the art can implement and utilize these embodiments. The scope of the embodiments of the subject matter is defined by the claims and may include other modifications that may be made available to those skilled in the art. Such other modifications are intended to be included in the claims if they have similar elements that do not differ from the language of the claims, or if they have equivalent elements that differ only slightly from the language of the claims.

[0073] It is understood that the scope of protection extends not only to such programs but also to computer-readable means containing messages, and such computer-readable storage means includes program-code means for carrying out one or more steps of the method when the program is executed on a server or mobile device or some suitable programmable device. The hardware device can be any kind of programmable device, including, for example, any kind of computer such as a server or personal computer or any combination thereof. The device may also include means that can be, for example, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, for example, an ASIC and an FPGA, or at least one microprocessor and at least one memory on which software processing components are arranged. Thus, the means may include both hardware and software means. Embodiments of the method described herein can be implemented in hardware and software. The device may also include software means. Alternatively, embodiments may be implemented on different hardware devices, for example, using multiple CPUs.

[0074] Embodiments of this specification may comprise hardware and software elements. Software-implemented embodiments include, but are not limited to, firmware, resident software, and microcode. Functions performed by the various components described herein may be implemented by other components or combinations thereof. For the purposes of this specification, a computer-enabled or computer-readable medium may be any device capable of comprising, storing, transmitting, propagating, or transferring programs for use by or in connection with an instruction execution system, apparatus, or device.

[0075] The illustrated steps are provided to illustrate the exemplary embodiments shown, and it is expected that ongoing technological developments will alter how certain functions are performed. These examples are presented herein for illustrative purposes only and are not limiting. Furthermore, boundaries of functional components are arbitrarily defined herein for the convenience of explanation. Alternative boundaries can be defined as long as certain functions and their relationships are adequately performed. Alternatives (including equivalents, extensions, modifications, deviations, etc., of those described herein) will be apparent to those skilled in the art based on the teachings contained herein. Such alternatives are included within the scope of the disclosed embodiments. Furthermore, the words “comprising,” “having,” “containing,” and “including,” as well as other similar forms, are intended to be semantically equivalent and open-ended, in that the item(s) following any one of these words does not mean that the item(s) described are exhaustive or limited to the item(s) described. Also, note that, as used herein and in the appended claims, the singular forms “a,” “an,” and “the” include plural references unless clearly indicated by the context.

[0076] Furthermore, one or more computer-readable storage media may be used when carrying out embodiments consistent with the Disclosure. Computer-readable storage media refers to any type of physical memory capable of storing information or data that is readable by a processor. Thus, computer-readable storage media can store instructions for execution by one or more processors, including instructions for causing a processor(s) to perform steps or stages consistent with embodiments described herein. The term “computer-readable media” should be understood to include tangible media, excluding carrier and transient signals; i.e., non-transient. Examples include random-access memory (RAM), read-only memory (ROM), volatile memory, non-volatile memory, hard drives, CD-ROMs, DVDs, flash drives, disks, and any other known physical storage media.

[0077] The true scope of the disclosed embodiments is indicated by the following claims, and this disclosure and the embodiments are intended to be merely illustrative. [Explanation of symbols]

[0078] 100 Systems 102 memory 108 Databases 104 Hardware Processors 106 Interfaces

Claims

1. 1. A processor implementation method, comprising: one or more hardware processors receiving (202) input requests from one or more businesses; querying a parts repository based on the input request by the one or more hardware processors to identify a plurality of parts (204); applying Jaccard similarity to each set of parts from the plurality of parts by calculating a similarity score between each set of parts from the plurality of parts (206), to obtain one or more combinations of sets of similar parts; the one or more hardware processors identifying at least one corresponding supplier from a supplier repository based on the similarity scores and tagging the at least one corresponding supplier with each of the one or more combinations of sets of similar parts (208); the one or more hardware processors performing (210) a comparison between the similarity score and a threshold; the one or more hardware processors identifying (212) one or more target suppliers from the one or more combinations of sets of similar parts and the at least one supplier based on the comparison; determining (214) a procurement cost for each set of the similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios by the one or more hardware processors; optimizing (216) a total procurement cost resulting from the procurement costs of each set of similar parts based on one or more constraints; predicting, via the one or more hardware processors, a target cost for a subsequent time instance using a neural network autoregressive input (NNARX) model, the NNARX model including three layers, a first input layer including neurons representing historical values ​​of cost and current and historical values ​​of external driving variables, a final output layer including neurons representing the target cost, a hidden layer including neurons configured to calculate node weights of the NNARX model with a normalized linear activation function, each node of the hidden layer being connected to one or more nodes of the input layer and a node of the output layer, the input requests being divided into a training set and a test set, the NNARX model being used on the test set to calculate a prediction error and a predicted cost for the test data, and the NNARX model being used with the prediction error to predict the target cost for the subsequent time instance; said performing, via one or more hardware processors, a comparison of (i) the predicted target costs for subsequent time instances with (ii) the optimized procurement costs for each of one or more predefined scenarios; identifying, via one or more hardware processors, a focus scenario from the one or more predefined scenarios based on the comparison, the focus scenario serving as an output; until the optimized total procurement cost of the focus scenario is less than the target cost with a tolerance value. identifying one or more target improvements; modifying at least one of (i) the cost associated with each part and (ii) one or more constraints based on the identified objective improvements; obtaining a new optimized cost for the focal scenario based on the revised cost and the one or more constraints; iteratively executing, via one or more hardware processors, 2. A processor implementation method comprising:

2. 2. The processor-implemented method of claim 1, wherein the one or more predefined scenarios include at least one of: (i) an overall cost reduction for a particular combination set of parts and one or more corresponding target suppliers; (ii) at least one potential supplier identified to fulfill parts for a particular business under a predefined quantity boundary; (iii) at least one potential supplier identified to fulfill parts for a particular business at a given spending opportunity; (iv) a maximum number of potential suppliers identified for distribution across businesses; or (v) a combination thereof.

3. The processor-implemented method of claim 1 , wherein the one or more constraints include at least one of a requirement, a maximum quantity of parts, a minimum quantity of parts, a maximum spend, and a number of suppliers.

4. a memory (102) for storing instructions; one or more communication interfaces (106); one or more hardware processors (104) connected to the memory (102) via the one or more communication interfaces (106); A system (100) comprising: The one or more hardware processors (104) are caused by the instructions to: receiving input requests from one or more businesses; querying a parts repository based on the input request to identify a plurality of parts; applying Jaccard similarity to each set of parts from the plurality of parts by calculating a similarity score between each set of parts from the plurality of parts to obtain one or more combinations of sets of similar parts; identifying at least one corresponding supplier from a supplier repository based on the similarity score, and tagging the at least one corresponding supplier with each of the one or more combinations of the set of similar parts; performing a comparison between the similarity score and a threshold; identifying one or more target suppliers from the one or more combinations of the set of similar parts and the at least one supplier based on the comparison; determining a procurement cost for each set of the similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios; optimizing a total procurement cost resulting from the procurement costs of each set of similar parts based on one or more constraints; a neural network autoregressive input (NNARX) model is used to predict a target cost for a subsequent time instance, the NNARX model including three layers, a first input layer including neurons representing historical values ​​of cost and current and historical values ​​of external driving variables, a final output layer including neurons representing a target cost, a hidden layer including neurons configured to calculate node weights of the NNARX model with a normalized linear activation function, each node of the hidden layer being connected to one or more nodes of the input layer and to nodes of the output layer, the input requests are divided into a training set and a test set, the NNARX model is used on the test set to calculate a prediction error and a predicted cost for the test data, and the NNARX model is used with the prediction error to predict the target cost for the subsequent time instance; Performing a comparison of (i) the projected target costs for subsequent time instances and (ii) the optimized procurement costs for each of one or more predefined scenarios; identifying a focus scenario from the one or more predefined scenarios based on the comparison, the focus scenario serving as an output; until the optimized total procurement cost of the focus scenario is less than the target cost with a tolerance value. modifying at least one of (i) the cost associated with each part and (ii) one or more constraints based on the identified objective improvements; obtaining a new optimized cost for the focal scenario based on the revised cost and the one or more constraints; Identifying one or more targeted improvements; By doing so, it is possible to execute it repeatedly. The system is configured as follows:

5. 5. The system of claim 4, wherein the one or more predefined scenarios include at least one of: (i) an overall cost reduction for a particular combination set of parts and one or more corresponding target suppliers; (ii) at least one potential supplier identified to fulfill parts for a particular business under a predefined quantity boundary; (iii) at least one potential supplier identified to fulfill parts for a particular business at a given spending opportunity; (iv) a maximum number of potential suppliers identified for distribution across businesses; or (v) a combination thereof.

6. The system of claim 4 , wherein the one or more constraints include at least one of a requirement, a maximum quantity of parts, a minimum quantity of parts, a maximum spend, and a number of suppliers.

7. One or more non-transitory machine-readable information storage media containing one or more instructions that, when executed by one or more hardware processors, cause a method for component bundling, corresponding supplier selection, and overall procurement cost optimization, the method comprising: receiving, by one or more hardware processors, input requests from one or more entities; the one or more hardware processors querying a parts repository based on the input request to identify a plurality of parts; the one or more hardware processors applying Jaccard similarity to each set of parts from the plurality of parts by calculating a similarity score between each set of parts from the plurality of parts to obtain one or more combinations of sets of similar parts; identifying at least one corresponding supplier from a supplier repository based on the similarity score and tagging the at least one corresponding supplier with each of the one or more combinations of sets of similar parts; the one or more hardware processors performing a comparison between the similarity score and a threshold; the one or more hardware processors identifying, based on the comparison, one or more target suppliers from the one or more combinations of sets of similar parts and the at least one supplier; the one or more hardware processors determining a procurement cost for each set of the similar parts corresponding to the one or more identified target suppliers based on one or more predefined scenarios; optimizing a total procurement cost resulting from the procurement costs of each set of similar parts based on one or more constraints; predicting, via the one or more hardware processors, a target cost for a subsequent time instance using a neural network autoregressive input (NNARX) model, the NNARX model including three layers, a first input layer including neurons representing historical values ​​of cost and current and historical values ​​of external driving variables, a final output layer including neurons representing the target cost, a hidden layer including neurons configured to calculate node weights of the NNARX model with a normalized linear activation function, each node of the hidden layer being connected to one or more nodes of the input layer and a node of the output layer, the input requests being divided into a training set and a test set, the NNARX model being used on the test set to calculate a prediction error and a predicted cost for the test data, and the NNARX model being used with the prediction error to predict the target cost for the subsequent time instance; said performing, via one or more hardware processors, a comparison of (i) the predicted target costs for subsequent time instances with (ii) the optimized procurement costs for each of one or more predefined scenarios; identifying, via one or more hardware processors, a focus scenario from the one or more predefined scenarios based on the comparison, the focus scenario serving as an output; until the optimized total procurement cost of the focus scenario is less than the target cost with a tolerance value. identifying one or more target improvements; modifying at least one of (i) the cost associated with each part and (ii) one or more constraints based on the identified objective improvements; obtaining a new optimized cost for the focal scenario based on the revised cost and the one or more constraints; iteratively executing, via one or more hardware processors, One or more non-transitory machine-readable information storage media, including:

8. 8. The one or more non-transitory machine-readable information storage media of claim 7, wherein the one or more predefined scenarios include (i) an overall cost reduction for a particular combination set of parts and one or more corresponding target suppliers, (ii) at least one potential supplier identified to fulfill parts for a particular business under a predefined quantity boundary, (iii) at least one potential supplier identified to fulfill parts for a particular business at a predetermined spending opportunity, and (iv) a maximum number of potential suppliers identified for distribution across the business, or (v) a combination thereof.

9. 8. The one or more non-transitory machine-readable information storage media of claim 7, wherein the one or more constraints include at least one of a requirement, a maximum quantity of parts, a minimum quantity of parts, a maximum spend, and a number of suppliers.

Images