System for real-time detection of supply chain disruptions and forecasting of financial impacts using AI-driven ERP environments

Abstract

A computer-implemented system (100) for real-time detection of supply chain disruptions and forecasting of the financial impact in a computer-driven ERP environment, wherein the system (100) comprises: a data acquisition interface (1) configured to continuously capture structured and unstructured operational data from one or more ERP modules, supply chain execution systems, and enterprise event streams; a data management and harmonization engine (2) configured to validate, normalize, deduplicate and semantically map the captured data into a unified, time-aligned canonical enterprise schema; a disturbance signal fusion and anomaly detection engine (3) configured to fuse signals from multiple sources and detect disturbance events using one or more machine learning models to output a disturbance value, event class and a list of affected nodes; a supply chain dependency graph and a digital twin builder (4) configured to create and update a dynamic dependency graph representing suppliers, facilities, transportation routes, SKUs, orders, contracts and lead times, and propagate disruption effects across the graph; a financial impact forecasting engine (5) configured to estimate the real-time and forward-looking effects of the disruption on one or more financial measures, including sales, margin, working capital, cash flow, service level penalties and inventory holding costs, and furthermore the effects on the cost of goods sold (COGS) and / or the landed costs using a Kl-based forecast with a range of uncertainty; a module for coordinating corrective actions and automating workflows (6) configured to generate ranked corrective actions and initiate ERP workflow steps, including reordering, reassignment, alternative sources of supply, production rescheduling and logistics diversions, wherein the ranked corrective actions include recommendations to resolve the disruption by using available stock, reassigning low-priority shipments or diverting shipments that can be delivered later within a lead time window; an explainability, audit trail and compliance logging module (7) configured to record model inputs, feature assignments, decision justifications, scenario assumptions and action results as an immutable audit trail; and a visualization, alerting and collaboration interface (8) configured to output real-time alerts, dashboards and scenario comparisons to authorized users and downstream systems via APIs and role-based access controls, wherein the interface (8) generates a quick report view that displays early disruption signals and impact on manufacturing costs, delivers notifications to mobile devices, triggers a special urgent notification if the disruption affects a high-priority trading partner, and accepts user responses on mobile devices that automatically and without delay trigger corrective actions or updates in the ERP system.

Description

INVENTION AREA

[0001] The present invention relates to enterprise software systems for monitoring supply chain risks. In particular, the invention relates to a computer-implemented system that operates in an AI-driven ERP environment to (i) detect disruptions in the supply chain in real time, (ii) transmit the effects of the disruptions on dependencies within the supply chain, and (iii) predict the financial impact on key business figures such as sales, margin, working capital, and cash flow with automated workflows for damage mitigation and auditable explainability. BACKGROUND OF THE INVENTION

[0002] The subject matter discussed in the "Background" section should not be considered prior art solely because it is mentioned in that section. Likewise, a problem mentioned in the "Background" section or related to the subject matter of the "Background" section should not be considered prior art. The subject matter in the "Background" section merely presents various approaches, which could themselves also be inventions.

[0003] IFRS Global and national supply chains have evolved from linear "supplier-factory-customer" models to complex, multi-tiered networks involving multiple suppliers, contract manufacturers, distribution centers, external logistics providers, and cross-border transportation routes. Even for a single finished product, a company may depend on multiple upstream suppliers and sub-suppliers, alternative bill of materials (BOM) options, capacity-constrained factories, and time-critical logistics routes. As a result, a disruption at one point in the network—such as late material deliveries, a quality complaint with a supplier, port congestion, a shortage of trucks, an unplanned factory closure, or a sudden surge in demand—can propagate, leading to cascading bottlenecks, production rescheduling, delivery delays, lost revenue, and increased operating costs.Complexity continues to increase due to frequent SKU proliferation, shorter product lifecycles, seasonal demand patterns, and global compliance requirements.

[0004] Companies typically manage procurement, inventory, production, sales, and finance through ERP environments that often span multiple instances across different subsidiaries and regions. These ERP ecosystems generate large volumes of operational signals, such as order confirmations, schedule changes, goods receipt postings, inventory transfers, backorder creation, work order completion, shipping milestone updates, invoice holds, and credit / debit adjustments. However, these signals are frequently scattered across heterogeneous data structures, inconsistent master data, different time zones, and varying coding standards. Vendor names, item codes, site identifiers, and contract references can differ by plant or business unit.As a result, early signs of disruption may remain hidden due to fragmentation and a lack of data harmonization, making it difficult to achieve real-time, company-wide risk transparency.

[0005] Traditional supply chain monitoring solutions in ERP environments typically rely on static exception rules, regular reporting, or manual escalations. Most dashboards operate with delayed snapshots (daily / weekly) and only detect problems once they have already significantly impacted operations—for example, when a shipment is late, a production line is down, or a customer order misses its promised delivery date. Rule-based exception handling also struggles with "weak signals," such as gradually increasing variability in lead times, correlated minor delays across multiple suppliers, unusual cancellation patterns, or subtle shifts in logistics reliability that precede a larger disruption.Furthermore, conventional systems often treat the detection of disturbances as a purely operational problem and financial forecasting as a purely financial task, so planners and finance teams have to manually consolidate data, assumptions and scenarios in spreadsheets or isolated tools.

[0006] A key limitation of current approaches is the lack of a mechanism that considers dependencies to model how disruptions propagate through the network. In practice, the impact of a delayed component is determined not only by the delay itself, but also by constraints such as safety stock policies, multi-level bill of materials dependencies, the availability of substitute materials, production capacity, transportation reliability, allocation priorities, contractual penalties for customers, and regional inventory levels. Without a continuously updated dependency model (e.g., a supply chain diagram or a digital twin), companies cannot accurately predict second-order impacts, such as which downstream SKUs will run out first, which customers will be affected, or how production schedules will shift.This often leads to reactive decisions, delayed corrective actions, unnecessary rush orders, and a suboptimal redistribution of inventory.

[0007] Another limitation is that existing forecasting methods often provide single-point estimates without quantified uncertainty. In real-world disruption scenarios, uncertainty is unavoidable: supplier recovery times are uncertain, transportation delays have distributions, demand may continue to fluctuate, and remedial actions may or may not be approved or implemented in a timely manner. If tools do not represent uncertainty ranges or scenario ranges, decision-makers either overreact (leading to excessive costs) or underreact (leading to service outages), and post-event debriefings fail to reveal the assumptions that influenced the decision.Furthermore, audit and compliance requirements—particularly in regulated industries—require companies to maintain traceable records of the data used, the reasons for risk classification, and the approvals and actions taken. Traditional ERP workflows often fail to capture model-driven "why" explanations and do not maintain an immutable chain of evidence that links operational triggers to financial outcomes and corrective actions.

[0008] Accordingly, there is a clear need for an improved system that (i) continuously captures ERP and supply chain events from various sources, (ii) harmonizes enterprise data into a consistent canonical representation, (iii) detects disruption events in real time using AI-driven multi-signal fusion, (iv) models the supply chain as a dynamically updated dependency graph / digital twin to propagate impacts across nodes, (v) forecasts financial impacts under baseline and counterfactual scenarios with uncertainty ranges, and (vi) coordinates ranked remediation actions through ERP workflows while maintaining explainable, auditable audit trails. Such a system would enable proactive disruption management, faster and more reliable decision-making, and a measurable reduction in lost revenue, service penalties, and operational inefficiencies.

[0009] The use of any examples or illustrative phrases (e.g., "as") relating to specific embodiments serves only to better illustrate the invention and does not constitute a limitation of the otherwise claimed scope of the invention. No wording in the description shall be construed as referring to an unclaimed element that is essential for carrying out the invention.

[0010] The information disclosed above in this "Background" section is provided solely for a better understanding of the background of the invention and may therefore contain information that is not part of the prior art already known to a person skilled in the art in this country. SUMMARY

[0011] Before describing the systems and methods presented here, it should be noted that this application is not limited to the specific systems and methods described, as there may be several possible embodiments not expressly presented in this disclosure. It should also be noted that the terminology used in the description serves only to describe the specific versions or embodiments and is not intended to limit the scope of this application.

[0012] In one embodiment, the present invention provides a computer-implemented system (100) configured for operation in an AI-driven ERP environment to enable real-time detection of supply chain disruptions and forecasting of the corresponding financial impact. The system (100) is designed to continuously monitor the company's operational signals, generate information about disruptions, assess business and financial risks, and support timely damage mitigation through automated workflows and auditable traceability.

[0013] In one embodiment, the system (100) comprises a data ingestion interface (1) configured to continuously capture and stream structured and unstructured data from one or more ERP modules, supply chain execution systems, supplier portals, and enterprise event streams. This data may include order events, shipping milestones, inventory movements, production status, supplier confirmations, and exception logs, thereby enabling near real-time visibility into operational changes.

[0014] In one embodiment, the system (100) further comprises a data management and harmonization engine (2) configured to validate, normalize, deduplicate, time-align, and semantically map the captured data into a unified canonical enterprise schema. The engine (2) can also perform entity resolution to unify supplier identities, item master data, location codes, and contract identifiers across multiple ERP instances and business units, thereby enabling consistent data interpretation for downstream analysis.

[0015] In one embodiment, the system (100) further comprises a disturbance signal fusion and anomaly detection engine (3) configured to fuse operational signals from multiple sources and detect disturbance events using one or more machine learning models. The engine (3) generates outputs that include at least one disturbance value, an event class, and a list of affected nodes, wherein the affected nodes include one or more suppliers, transportation routes, facilities, SKUs, storage locations, or customer orders that may be affected by the detected disturbance.

[0016] In one embodiment, the system (100) further comprises a supply chain dependency graph and a digital twin builder (4) configured to create and continuously update a dynamic dependency representation of supply chain entities and constraints. The generator (4) is configured to propagate disruption effects across dependencies such as supplier-to-plant flows, bill of materials relationships, capacity constraints, and lead time variability, thereby identifying downstream consequences such as predicted inventory shortages, bottlenecks, and service level risks.

[0017] In one embodiment, the system (100) further comprises a financial impact forecasting engine (5) configured to estimate, in real time and predictively, the impact of the disruption on one or more financial metrics, including revenue, margin, working capital, cash flow, penalty risks, rush costs, and inventory holding costs. The forecasting engine (5) generates forecasts with uncertainty ranges and supports baseline and alternative scenarios, enabling decision-makers to assess risk areas and expected outcomes under different risk mitigation strategies.

[0018] In one embodiment, the system (100) further comprises a module (6) for coordinating risk mitigation measures and automating workflows, configured to generate ranked risk mitigation measures and initiate ERP workflow steps for execution. The module (6) can recommend and / or trigger measures such as alternative procurement, reordering, production rescheduling, inventory redistribution, customer prioritization, and logistics rerouting, while simultaneously supporting approval routing and the selection of multiple targets based on cost, service level, margin impact, and risk exposure.

[0019] In one embodiment, the system (100) further comprises an explainability, audit trail, and compliance logger (7) configured to store model inputs, derived features, feature assignments, decision justifications, scenario assumptions, action selections, and realized results. The logger (7) maintains an immutable or tamper-proof monitoring audit trail to support internal controls, debriefings, and compliance requirements, including the ability to answer questions such as "Why was this flagged?" and "What has changed?"

[0020] In one embodiment, the system (100) further comprises a visualization, alerting, and collaboration interface (8) configured to provide authorized users with real-time alerts, dashboards, and scenario comparison views, and to integrate the outputs into downstream enterprise systems via secure APIs. The interface (8) provides role-based access for users from procurement, planning, operations, finance, and management, and supports collaboration features such as acknowledgment tracking, escalation workflows, and the assignment of responsibilities for risk mitigation.

[0021] In further embodiments, the system (100) supports extended functions, including scenario simulation (baseline vs. counterfactual), multi-target ranking of risk mitigation measures, continuous monitoring of model deviations, and feedback-based retraining using realized results, thereby maintaining accuracy and reliability over time in evolving ERP-driven supply chain environments. BRIEF DESCRIPTION OF THE DRAWING

[0022] To clarify various aspects of some embodiments of the present invention, a more detailed description of the invention is given with reference to specific embodiments shown in the accompanying drawing. It is understood that this drawing represents only illustrative embodiments of the invention and is therefore not to be considered a limitation of its scope. The invention is described and explained with additional specificity and detail using the accompanying drawing.

[0023] To make the advantages of the present invention easily understandable, a detailed description of the invention is discussed below in conjunction with the accompanying drawing, which, however, should not be regarded as limiting the scope of the invention to the accompanying drawing, in which: Fig. a block diagram representation of the system (100) for real-time detection of supply chain disruptions and forecasting of financial impacts using AI-driven ERP environments. DETAILED DESCRIPTION

[0024] The present invention relates to the system (100) for real-time detection of disruptions in the supply chain and for forecasting financial impacts using AI-driven ERP environments.

[0025] Fig. shows a detailed block diagram representation of the system (100) for real-time detection of supply chain disruptions and forecasting of financial impacts using AI-driven ERP environments.

[0026] For the purposes of this specification, an ERP environment comprises enterprise resource planning systems and modules used for procurement, warehousing, production, sales, logistics, finance, treasury, and enterprise risk functions, and may additionally include related execution systems such as manufacturing execution systems, warehouse management systems, transportation systems, supplier portals, and enterprise data platforms. A disruption refers to any event, condition, or developing pattern that causes or is likely to cause a deviation from the expected performance of the supply chain, including but not limited to delivery delays, bottlenecks, capacity reductions, demand shocks, quality defects, compliance blockages, and supply route failures.A canonical business schema refers to a unified representation of business entities and transactions, mapping suppliers, items / SKUs, routes, locations, orders, contracts, and time-series measurements into consistent structures. A digital twin refers to a computational representation of nodes, constraints, and flows in the supply chain, used to simulate, estimate, and propagate the impact of disruptions. Financial metrics can include impacts on revenue, gross margin, EBITDA, changes in working capital, impacts on the cash conversion cycle, penalties, rush costs, inventory holding costs, and impacts on service levels.

[0027] In one embodiment, the invention provides a computer-implemented system (100) that is deployed on one or more computing devices, including servers, cloud instances, virtual machines, containers, or hybrid enterprise deployments. The system (100) is connected to enterprise data sources via one or more secure connectors, including API gateways, message queues, event buses, database connectors, file-based batch exports, and controlled middleware adapters. The system (100) is configured to operate continuously and to be informed about the conditions in the enterprise's supply chain in near real time.

[0028] In one embodiment, the system (100) is logically divided into a series of functional modules (1) to (8) that together perform the continuous acquisition of operational signals, the control and harmonization of enterprise data, the detection of faults through AI-based analysis, the dependency-aware propagation of impacts, the forecasting of financial impacts with uncertainty, the coordination of corrective actions through ERP workflows, the logging of supervisory audits with explainability, and role-based visualization and alerting. The modules can be implemented as microservices or components that communicate via authenticated inter-service communication and can be horizontally scaled depending on the event volume, enterprise size, and required response latency.

[0029] In one embodiment, the system (100) comprises a data acquisition interface (1) configured to continuously acquire data from one or more sources, including ERP transaction logs, procurement and inventory modules, production planning modules, order management modules, supplier collaboration portals, transportation milestone systems, and enterprise event streams. The acquisition interface (1) can acquire structured data such as tables, transaction records, and time series measurements, and can also acquire unstructured data such as exception comments, supplier communications, incident texts, or operational notes.

[0030] In one embodiment, the data acquisition interface (1) supports streaming acquisition to capture changes from event to event with low latency and also supports batch acquisition for periodic synchronization when streaming is unavailable. The interface (1) can add timestamps, source identifiers, and event correlation identifiers to each captured event to maintain event sequence and enable traceability. Additionally, the interface (1) can apply secure authentication and encryption during data acquisition to protect corporate data both in transit and at rest.

[0031] In one embodiment, non-limiting examples of captured signals include order creation and confirmation, schedule changes, advance delivery notifications, goods receipt postings, inventory transfers, backorder creation, invoice holdups, production outages, machine capacity changes, delivery milestone delays, forecast changes, returns / cancellations, and quality rejections. These signals form a continuous operational footprint used for fault derivation and downstream analysis.

[0032] In one embodiment, the system (100) comprises a data management and harmonization engine (2) configured to validate, standardize, normalize, and integrate the acquired signals before they are used by AI / ML models or simulation components. The engine (2) performs data validations, such as completeness checks, range checks, referential integrity checks, and logical consistency checks between related ERP transactions. The engine (2) can generate a governance status indicating whether data records are valid, partially valid, or suspicious ( ) and can forward exceptions for review or automated correction.

[0033] In one embodiment, the engine (2) performs normalization operations, such as the harmonization of units of measurement, currency conversion, time zone adjustment, naming conventions, and date and time formats. The engine (2) also performs deduplication and event consolidation to prevent repeated or resent messages from inflating the noise signals. Furthermore, the engine (2) semantically maps the processed data to a canonical enterprise schema, thus enabling consistent downstream calculations across multiple ERP instances and business units.

[0034] In one embodiment, the engine (2) performs entity resolution to unify supplier identities, item / SKU identifiers, location codes, lane identifiers, and contract references, particularly when different subsidiaries maintain different master data representations. The engine (2) can assign data quality scores and generate governance outputs such as missing supplier confirmations, inconsistent lead time entries, abnormal unit costs, or mismatched item mappings, which can be displayed to authorized users via the visualization interface (8) to support corrective action.

[0035] In one embodiment, the system (100) comprises a disturbance signal fusion and anomaly detection engine (3) configured to detect disturbances by analyzing the managed and harmonized data generated by the engine (2). The engine (3) fuses multiple operational signals and creates feature sets that represent the behavior of the enterprise's supply chain over time. Such feature sets may include patterns of lead time variances, changes in supplier confirmations, trends in the frequency of delays, missed delivery dates, inventory withdrawal rates, production yield anomalies, capacity bottlenecks, abnormal cancellation or return rates, demand spikes, forecast error patterns, and correlated anomalies across suppliers, transportation routes, and facilities.

[0036] In one embodiment, the engine (3) detects disturbances using one or more machine learning approaches, including sequential event models that learn temporal patterns of ERP events, anomaly detection models trained on historical normal behavior, supervised models trained on labeled disturbance histories, and graph-based models that leverage relationships between suppliers, plants, and logistics routes. The engine (3) can combine these methods through ensemble fusion to improve robustness and reduce false alarms, particularly in complex ERP environments where disruptive operational events occur.

[0037] In one embodiment, the engine (3) outputs a disruption assessment that includes at least one disruption value indicating the severity and / or probability, an event class indicating the type of disruption, such as supplier delay, logistics delay, quality defect, demand surge, or capacity reduction, and a list of affected nodes identifying the suppliers, SKUs, routes, plants, warehouses, and customer orders that are affected or at risk. The output may further include temporal attributes such as the detected start time, the expected duration range, and the escalation priority.

[0038] In one embodiment, the system (100) comprises a supply chain dependency graph and a digital twin builder (4) configured to dynamically generate and continuously update a representation of the supply chain dependencies. The builder (4) represents nodes, including suppliers, facilities, warehouses, logistics routes, SKUs, orders, contracts, and customers. The builder (4) also represents edges that capture supply relationships, transportation flows, substitution links, bill of materials dependencies, capacity dependencies, and contractual obligations. The builder (4) updates the dependency graph based on newly captured events, the results of master data harmonization, and changes in supply or demand conditions.

[0039] In one embodiment, when a disruption is detected by the Engine (3), the Builder (4) propagates the disruption's impact across the dependency graph to estimate downstream consequences. During propagation, constraints such as lead time distributions, safety stock policies, reorder points, production capacity and cycle time limitations, transportation reliability, allocation rules, and substitution possibilities can be evaluated. The Builder (4) can simulate inventory flows and estimate the expected downtime for affected SKUs at specific locations, as well as identify bottlenecks that amplify chain reaction effects on dependent finished goods or customer orders.

[0040] In one embodiment, the digital twin part of the Builder (4) performs a constraint-aware simulation to generate a time-staggered view of supply bottlenecks, production constraints, and fulfillment risks across the entire network. This enables an early assessment of second- and third-order impacts beyond the initially affected node, allowing mitigation measures to be planned before service outages occur.

[0041] In one embodiment, the system (100) comprises a financial impact forecasting engine (5) configured to estimate both real-time and forward-looking financial impacts resulting from detected disruptions and propagating effects. The engine (5) processes disruption expenditures from the engine (3) and dependency impacts from the builder (4) and converts operational failures into financial risks. The engine (5) can estimate impacts on at-risk revenues due to lost sales, margin erosion due to substitution or acceleration, changes in working capital due to increased transit inventory or safety stock withdrawals, cash flow changes due to payment term delays, penalty risks due to service level violations, and cost increases due to overtime production or premium freight.

[0042] In one embodiment, the engine (5) uses AI forecasting models that take into account historical time series patterns, seasonality, demand elasticity, price and cost structures, contract rules and penalty clauses, allocation priorities, and finance-specific assumptions such as payment terms and cash cycle parameters. The engine (5) generates a baseline forecast representing the expected outcome without disturbances or maintaining the status quo, and additionally generates one or more counterfactual forecasts representing alternative outcomes under different combinations of actions.

[0043] In one embodiment, the engine (5) provides uncertainty ranges by generating confidence intervals or risk areas using probabilistic methods such as Bayesian inference, conformal forecasting, or Monte Carlo simulation. This enables decision-makers to compare not only point estimates but also risk distributions, thereby supporting risk-aware decision-making and preventing over- or under-reactions under uncertain recovery conditions.

[0044] In one embodiment, the system (100) comprises a module (6) for coordinating remedial actions and automating workflows, configured to generate feasible remedial actions and initiate ERP workflows to execute selected actions. The module (6) identifies risk mitigation strategies such as sourcing from alternative suppliers, accelerating shipments or changing the mode of transport, redistributing inventory across regions, rescheduling production and rebalancing capacities, substituting materials or components, adjusting customer priorities, and triggering contract renegotiations.

[0045] In one embodiment, the module (6) evaluates risk mitigation measures using multi-criteria optimization, which balances service level restoration, costs, margin impact, risk, feasibility constraints, and compliance limitations. The module (6) can generate a ranked action list for each package of measures, including the expected financial and operational outcomes. Selected measures can be executed via ERP APIs or workflow functions, including the creation of purchase requisitions or purchase orders, the creation of transportation orders, the adjustment of production schedules, the updating of allocation rules, the submission of approval requests, and the initiation of exception handling workflows.

[0046] In one embodiment, module (6) supports governance by enforcing approval hierarchies and logging approvals, rejections, or changes to recommended actions. Module (6) can also incorporate restrictions relating to supplier qualifications, regulatory limitations, and contractual constraints to ensure that risk mitigation measures remain compliant and practically feasible.

[0047] In one embodiment, the system (100) comprises an explainability, audit trail, and compliance logger (7) configured to record the entire lifecycle of fault detection, prediction, decision-making, workflow execution, and realized results. The logger (7) stores the data sources used, model inputs, derived features, output features, decision thresholds, scenario assumptions, and model versions, including configuration hashes, thereby enabling the reproducibility of system outputs during testing or post-event verification.

[0048] In one embodiment, the logger (7) stores recommended actions, approval timestamps, user acknowledgments, execution confirmations, and subsequent outcome measurements such as actual delivery performance, actual express costs, actual revenue losses, and service-level results. The logger (7) can maintain a tamper-proof or append-only storage structure so that the recorded path cannot be altered without detection, thereby supporting regulatory compliance, internal controls, and accountability.

[0049] In one embodiment, the logger (7) enables audit queries such as “Why was this disturbance reported?”, “What has changed compared to a previous assessment?”, “Which data and model version generated this prediction?”, and “Which corrective action was selected and what was its observed effect?”. This capability improves transparency and supports governance teams in validating decision rationales under complex disturbance conditions.

[0050] In one embodiment, the system (100) includes a visualization, alerting, and collaboration interface (8) configured to provide real-time insights and alerts to authorized enterprise users and systems. The interface (8) can provide dashboards displaying disruption hotspots, risk assessments, affected nodes, predicted timelines for inventory shortages, and recommended remediation actions. The interface (8) can also provide financially oriented views summarizing risky revenues, margin impacts, working capital changes, and scenario comparisons for management.

[0051] In one embodiment, the interface (8) supports the transmission of alerts via one or more channels, including email, messaging, ERP notifications, incident management tools, and API callbacks to downstream applications. The interface (8) also supports collaboration features such as responsibility assignment, acknowledgment tracking, escalation workflows, comment threads, and action status monitoring to ensure that risk mitigation measures progress within the required response times.

[0052] In one embodiment, the interface (8) enforces role-based access control and policy-based authorization, ensuring that sensitive information about suppliers, prices, and contracts is accessible only to authorized personnel. The interface (8) can implement audit-friendly user activity logs that are aligned with the compliance logger (7) to ensure the traceability of user actions and approvals.

[0053] In one embodiment, the system (100) includes a continuous learning loop in which realized results from disturbances, predictions, and corrective actions are captured and used to improve model performance. The system can monitor data deviations (changes in the input distributions), concept deviations (changes in the relationship between signals and disturbances), and performance deviations (changes in accuracy over time). Such monitoring can be performed via dashboards that track precision / recall for disturbance detection and prediction error for estimating financial impact.

[0054] In one embodiment, retraining and recalibration can be triggered based on detected deviations, periodic schedules, or exceedances of performance thresholds. The system can also recalibrate predicted probabilities and confidence intervals to ensure the uncertainty range remains reliable. In this way, the system (100) maintains its adaptability and accuracy in dynamic, ERP-driven supply chain environments where patterns and risks evolve over time.

[0055] In another embodiment, the visualization, alerting, and collaboration interface (8) generates a “quick report view” that is automatically populated when the disturbance signal fusion and anomaly detection engine (3) issues an early warning signal or disturbance event. The quick report view summarizes on a compact screen (i) a disturbance value, (ii) an event class, (iii) a list of affected nodes, (iv) predicted time windows for supply bottlenecks, (v) bottlenecks derived from the dependency graph and the digital twin builder (4), and (vi) a financial overview from the financial impact forecasting engine (5), wherein the financial overview includes at least the impact on the cost of goods sold (COGS) and / or the cost of goods sold, as well as uncertainty margins.

[0056] In another embodiment, the quick report view is presented as a role-based "at-a-glance" report for users from the areas of supply planning, procurement, finance and management, and can include an at-a-glance difference between the base scenario and the counterfactual scenario to show the incremental COGS and the margin effect of the disruption.

[0057] Notifications with Guidance and Ranked Solution Recommendations In one embodiment, the module for coordinating remediation and automating workflows (6) generates actionable guidance upon detection of a disruption or early warning. This guidance is embedded in the notifications generated by the visualization, alerting, and collaboration interface (8). The guidance includes ranked recommendations for resolving or mitigating the disruption by: (i) consuming available inventory to protect high-value demand, (ii) reallocating low-priority shipments to higher-priority demand or commitments, and / or (iii) rerouting shipments that can be delivered later within a lead time window, thus reducing the short-term impact on service levels while maintaining availability.

[0058] In another embodiment, each recommended remedy is displayed with the estimated cost / benefit (including impact on manufacturing costs, change in service level and risk assessment) together with a summary generated by the explainability, audit trail and compliance logger (7) so that the user can understand “why this remedy is recommended”.

[0059] Urgent notification of impacts on trading partners with high priority.

[0060] In one embodiment, the system manages a priority attribute for one or more trading partners within the unified canonical enterprise schema created by the data management and harmonization engine (2) and / or within the dependency graph managed by the dependency graph and digital twin builder (4). When the list of affected nodes indicates that a disturbance affects a “high-priority trading partner,” the visualization, alerting, and collaboration interface (8) triggers a special urgent notification, escalates the severity of the alert, and initiates an accelerated acknowledgment workflow with response time tracking.

[0061] In another embodiment, the urgent notification can trigger an automated escalation chain (e.g., planner → manager → executive) if no confirmation is received within a threshold defined in the policy.

[0062] The figure and the preceding description provide examples of embodiments. Those skilled in the art will recognize that one or more of the described elements can be combined to form a single functional element. Alternatively, certain elements can be divided into several functional elements. Elements from one embodiment can be added to another embodiment. For example, the sequence of the processes described here can be changed and is not limited to the manner described here. Furthermore, the actions of a block diagram need not be implemented in the sequence shown, nor does it necessarily have to be executed all actions. In addition, those actions that are not dependent on other actions can be executed in parallel with the other actions. The scope of embodiments is by no means limited by these specific examples.

[0063] Although the embodiments of the invention have been described in language relating to structural features and / or methods, it should be noted that the appended claims are not necessarily limited to the specific features or methods described. Rather, the specific features and methods are disclosed as examples of embodiments of the invention.

Claims

[1] A computer-implemented system (100) for real-time detection of supply chain disruptions and forecasting of the financial impact in an AI-driven ERP environment, wherein the system (100) comprises: a data acquisition interface (1) configured to continuously capture structured and unstructured operational data from one or more ERP modules, supply chain execution systems, and enterprise event streams; a data management and harmonization engine (2) configured to validate, normalize, deduplicate and semantically map the captured data into a unified, time-aligned canonical enterprise schema; a disturbance signal fusion and anomaly detection engine (3) configured to fuse signals from multiple sources and detect disturbance events using one or more machine learning models to output a disturbance value, event class and a list of affected nodes; a supply chain dependency graph and a digital twin builder (4) configured to create and update a dynamic dependency graph representing suppliers, facilities, transportation routes, SKUs, orders, contracts and lead times, and propagate disruption effects across the graph; a financial impact forecasting engine (5) configured to estimate the real-time and forward-looking effects of the disruption on one or more financial measures, including sales, margin, working capital, cash flow, service level penalties and inventory holding costs, and furthermore the effects on the cost of goods sold (COGS) and / or the landed costs using a Kl-based forecast with a range of uncertainty; a module for coordinating remedial actions and automating workflows (6) configured to generate ranked remedial actions and initiate ERP workflow steps, including reordering, reassignment, alternative sources of supply, production rescheduling, and logistics diversions, wherein the ranked remedial actions include recommendations to resolve the disruption by using available stock, reassigning low-priority shipments, or diverting shipments that can be delivered later within a lead time window; an explainability, audit trail and compliance logging module (7) configured to record model inputs, feature assignments, decision justifications, scenario assumptions and action results as an immutable audit trail; and a visualization, alerting and collaboration interface (8) configured to output real-time alerts, dashboards and scenario comparisons to authorized users and downstream systems via APIs and role-based access controls, wherein the interface (8) generates a quick report view that displays early disruption signals and impact on manufacturing costs, delivers notifications to mobile devices, triggers a special urgent notification if the disruption affects a high-priority trading partner, and accepts user responses on mobile devices that automatically and without delay trigger corrective actions or updates in the ERP system. [2] System (100) according to claim 1, wherein the disturbance signal fusion and anomaly detection engine (3) is configured to generate early warning signals prior to the occurrence of a significant disturbance by analyzing lead time deviations, inventory velocity deviations, correlated delivery delays and supplier confirmation anomalies, and wherein the visualization, alerting and collaboration interface (8) displays the early warning signals in the quick report view together with a COGS impact indicator. [3] System (100) according to claim 1, wherein the data acquisition interface (1) is configured to receive at least one of the following elements as time-stamped event streams: order events, invoice events, inventory movements, shipping milestones, supplier confirmations, demand signals and plant performance telemetry data. [4] System (100) according to claim 1, wherein the data management and harmonization engine (2) performs entity resolution to unify supplier identities, item masters, location codes and contract identifiers across heterogeneous ERP instances and subsidiaries, and further links a trading partner priority attribute to one or more of the unified supplier identities or customer / trading partner identities. [5] System (100) according to claim 1, wherein the disturbance signal fusion and anomaly detection engine (3) uses a multimodal model comprising at least one of the following elements: an attention-based transformer for sequential ERP events, a graphical neural network about supplier relationships, and an anomaly detector trained on historical disturbance labels and near miss patterns. [6] System (100) according to claim 1, wherein the dependency graph and digital twin builder (4) calculates the propagation of disturbances using constraint-aware simulation based on lead time distributions, capacity constraints, safety stock policies and route reliability assessments to output predicted downtimes and bottleneck locations and to determine whether an affected node corresponds to a high-priority trading partner for urgent escalation. [7] System (100) according to claim 1, wherein the financial impact forecasting engine (5) generates scenario-based forecasts for a base scenario and one or more counterfactual scenarios and creates confidence intervals using conformal predictions, Bayesian inference or Monte Carlo simulations, wherein at least one counterfactual scenario includes a destination change scenario for a shipment during transport to compare the effects of taxes / duties and landing costs between alternative destinations. [8] System (100) according to claim 1, wherein the module for coordinating remedial actions and automating workflows (6) evaluates remedial actions using a multi-criteria optimization that balances service levels, costs, margin impacts and risk exposure, and writes selected actions back to the ERP via authenticated APIs, wherein a user response received via a mobile notification triggers the write-back to the ERP without delay using transaction update operations. [9] System (100) according to claim 1, wherein the explainability, audit trail and compliance logger (7) stores feature attribution outputs, summaries of causal factors and model version hashes and supports audit queries for "why" and "what-has-changed" analyses over time windows and furthermore stores mobile responses from users, triggered corrective actions and corresponding ERP update confirmations as linked audit events. [10] System (100) according to claim 1, wherein the visualization, alerting and collaboration interface (8) provides role-specific views for delivery planners, procurement, finance and management and offers guided escalation workflows with confirmation tracking and response time metrics, wherein the interface (8) issues a policy for urgent notifications in the event of disruptions affecting trading partners with high priority and embeds instructions for action in the notification to enable a direct mobile response and the automated execution of corrective actions.

Images