Event-driven Method Research Articles

Hadoop in Banking: Event-Driven Performance Evaluation.

In today's data-intensive atmosphere, performance evaluation in the banking industry depends on timely and accurate insights, leading to better decision making and operational efficiency. Traditional methods for assessing bank performance often need to be improved to handle the volume, velocity, and variety of data generated in real time. This study proposes an event-driven approach for performance evaluation in banking alongside a Hadoop-based architecture. Infused with real-time event analytics, this scalable framework can process and analyze fast-moving transactional data. Hence, the framework allows banks to monitor key performance indicators and detect real-time operational anomalies. This is supported by the Hadoop ecosystem, which provides distribution of the processing and storage, making it fit for handling large datasets with high fault tolerance and parallel computation levels. This study analyzes transaction and user engagement data using Hive queries, focusing on credit card transactions via MasterCard. Two cases are examined: a detailed snapshot of individual transactions and a five-day trend analysis. Metrics like active users, card registrations, and retention are visualized through dashboards. Findings reveal user activity patterns and areas for improvement, emphasizing scalable, data-driven approaches for transaction analytics. This framework conceives a functional approach for banks to exploit extensive data-analytic capabilities to strive for competitive advantage and survivability of a business by adding any required metrics. The findings signify that the Hadoop-integrated event-driven analytics method could act as a game changer for performance evaluation in the banking sector.

Energy-conserving event-triggered control for multiagent systems with limited resources

An energy-conserving event-driven control method is developed for multiagent systems with unknown disturbances to achieve low-energy formation control. Firstly, an energy-conserving event-driven scheme is developed to save energy by introducing a second trigger function, which is utilized to switch the control input to zero when the consensus error is small. Then, a dynamic energy-conserving event-driven scheme is designed to further decrease the frequency of communication and control by establishing a positive dynamic variable to increase the trigger threshold. Under the designed energy-conserving event-based schemes, the energy consumption during the system stability is significantly reduced, the communication burden is decreased, and the system is Zeno-free and bounded stable. Finally, a four-agent formation example is illustrated to show the effectiveness of the proposed energy-conserving schemes.

Event-driven weakly supervised video anomaly detection

Inspired by the observations of human working manners, this work proposes an event-driven method for weakly supervised video anomaly detection. Complementary to the conventional snippet-level anomaly detection, this work designs an event analysis module to predict the event-level anomaly scores as well. It first generates event proposals simply via a temporal sliding window and then constructs a cascaded causal transformer to capture temporal dependencies for potential events of varying durations. Moreover, a dual-memory augmented self-attention scheme is also designed to capture global semantic dependencies for event feature enhancement. The network is learned with a standard multiple instance learning (MIL) loss, together with normal-abnormal contrastive learning losses. During inference, the snippet- and event-level anomaly scores are fused for anomaly detection. Experiments show that the event-level analysis helps to detect anomalous events more continuously and precisely. The performance of the proposed method on three public datasets demonstrates that the proposed approach is competitive with state-of-the-art methods.

KGroot: A knowledge graph-enhanced method for root cause analysis

Fault localization in online microservices is a challenging task due to the vast amount of monitoring data, diversity of types and events, and complex interdependencies among services and components. Fault events in services are propagative and can trigger a cascade of faults in a short period of time. In the industry, fault localization is typically conducted manually by experienced personnel. This reliance on experience is unreliable and lacks automation. Different modules present information barriers during manual localization, making it difficult to quickly align during urgent faults. This inefficiency lags stability assurance to minimize fault detection and repair time. Although actionable methods aimed to automate the process, the accuracy and efficiency are less than satisfactory. The precision of fault localization results is of paramount importance as it underpins engineers’ trust in the diagnostic conclusions, which are derived from multiple perspectives and offer comprehensive insights. Therefore, a more reliable method is required to automatically identify the associative relationships among fault events and propagation paths. To achieve this, a knowledge graph-enhanced root cause analysis (KGroot) method is designed for efficient and effective diagnosis of recurring failures in complex microservices environments. As the first event-driven knowledge graph method, KGroot uses event knowledge and the correlation between events to perform root cause reasoning for Root Cause Analysis (RCA). A Fault Event Knowledge Graph (FEKG) is built based on historical data, an online graph is constructed in real-time when a failure event occurs, and the similarity between each event knowledge graph and online graph is compared using GCNs to pinpoint the fault type through a ranking strategy. Comprehensive experiments demonstrate that KGroot can locate the root cause with an accuracy of 93.5% top 3 potential causes in second-level. This performance matches the level of real-time fault diagnosis in the industrial environment and significantly surpasses state-of-the-art baselines in RCA in terms of effectiveness and efficiency. (KGroot is available at https://github.com/daixixiwang/KGroot).

Event-driven piezoelectric energy harvesting for railway field applications

Piezoelectric energy harvesters (PEHs) hold promises for revolutionizing railway structural health monitoring (SHM). However, the challenges of designing high-robust PEHs for handling complex and ever-shifting vibrations in practical railway scenarios remain formidable. Unlike conventional methods that align the fundamental frequency of the PEH with the dominant frequency of the vibration source, this study employs an event-driven enhancement method. This study commences by employing a high-fidelity dynamic model of the PEH and an advanced vehicle-track coupled dynamic model to evaluate energy harvesting efficiency across various railway scenarios. The vehicle-induced track vibrations are predicted across various train speeds, track structures, and track irregularities, considering 12 distinct vehicle-track conditions and totaling 3600 simulation datasets, each representing a specific event. We employ the particle swarm optimization (PSO) algorithm to identify optimal PEH designs for different events. To streamline this process, instead of offering one optimal solution for each event, we employ the K-means algorithm to cluster similar events. Subsequently, we choose the centroid design for each cluster as the representative solution for events within that cluster. This approach allows us to use a limited number of representative PEHs to effectively address most events across different clusters. Finally, we thoroughly examine and evaluate the dynamic responses of these representative designs to demonstrate their robust performance. This study explores PEH design optimization from a fresh perspective, bridging the gap between theoretical design and practical implementation in rail transportation systems.

Model predictive control for non-holonomic robots with disturbances: A quasi-differential type event-driven method

For non-holonomic wheeled mobile robots with coupled constraints and bounded disturbances, a novel event-driven model predictive control (EMPC) strategy is designed in this paper. Firstly, to alleviate the computational and data transmission burden of the system, an EMPC algorithm considering the quasi-differential information of continuous state discrepancies is proposed, rather than only considering the system state at a single moment. Secondly, to accelerate the convergence speed of the system state, an exponential robust constraint is introduced when designing the optimization problem. Furthermore, the parameter conditions that ensure the feasibility and system stability are theoretically provided. In the end, simulation and comparison are provided to demonstrate the effectiveness of this algorithm.

Advantages of the Event Method for the Simulation of Water Quality in Pressurised Water Systems

In this paper, several methods for the calculation of water quality evolution in drinking water distribution networks are analysed. The Lagrangian Time-Driven method has been implemented in the Epanet simulation software since version 2.0. In version 2.2, some improvements were implemented to deal with mass imbalances (Lagrangian Time-Driven improved method). However, it sometimes presents inaccuracies in calculations, especially when there are short-length pipes. To solve this problem, the implementation of the Lagrangian Event-Driven method is proposed, which provides more accurate quality results. In order to detail the differences and similarities of the results of the different methods and to determine under what conditions the results provided by Epanet are sufficiently adjusted, two practical examples have been carried out, one of them on a hydraulic model of a real network.

EPOC: A 28-nm 5.3 pJ/SOP Event-driven Parallel Neuromorphic Hardware with Neuromodulation-based Online Learning.

Bio-inspired neuromorphic hardware with learning ability is highly promising to achieve human-like intelligence, particularly in terms of high energy efficiency and strong environmental adaptability. Though many customized prototypes have demonstrated learning ability, learning on neuromorphic hardware still lacks a bio-plausible and unified learning framework, and inherent spike-based sparsity and parallelism have not been fully exploited, which fundamentally limits their computational efficiency and scale. Therefore, we develop a unified, event-driven, and massively parallel multi-core neuromorphic online learning processor, namely EPOC. We present a neuromodulation-based neuromorphic online learning framework to unify various learning algorithms, and EPOC supports high-accuracy local/global supervised Spike Neural Network (SNN) learning with a low-memory-demand streaming single-sample learning strategy through different neuromodulator formulations. EPOC leverages a novel event-driven computation method that fully exploits spike-based sparsity throughout the forward-backward learning phases, and parallel multi-channel and multi-core computing architecture, bringing 9.9× time efficiency improvement compared with the baseline architecture. We synthesize EPOC in a 28-nm CMOS process and perform extensive benchmarking. EPOC achieves state-of-the-art learning accuracy of 99.2%, 98.2%, and 94.3% on the MNIST, NMNIST, and DVS-Gesture benchmarks, respectively. Local-learning EPOC achieves 2.9× time efficiency improvement compared with the global learning counterpart. EPOC operates at a typical clock frequency of 100 MHz, providing a peak 328 GOPS/51 GSOPS throughput and a 5.3 pJ/SOP energy efficiency.

Event-driven method of measuring effectiveness of information security controls in software development environments

Event-driven method of measuring effectiveness of information security controls in software development environments

Toward model-free safety-critical control with humans in the loop

This vision article shows how to build on the framework of event-triggered Control Barrier Functions (CBFs) to design model-free controllers for safety-critical multi-agent systems with unknown dynamics, including humans in the loop. This event-triggered framework has been shown to be computationally efficient and robust while guaranteeing safety for systems with unknown dynamics. We show how to extend it to model-free safety critical control where a controllable ego agent does not need to model the dynamics of other agents and updates its control based only on events dependent on the error states of agents obtained by real-time sensor measurements. To facilitate the process of real-time sensor measurements critical in this approach, we also present CBF relative degree reduction methods, which can reduce the number of such measurements. We illustrate the effectiveness of the proposed framework on a multi-agent traffic merging decentralized control problem and on highway lane changing control with humans in the loop and relative degree reduction. We also compare the proposed event-driven method to the classical time-driven approach.

Parallel dynamic NSGA-II with multi-population search for rescheduling of Seru production considering schedule changes under different dynamic events

Seru production is usually accompanied by various dynamic events in the real world, such as worker tool breakdowns, new batch arrivals, batch cancellations, and batch size changes. Therefore, dynamic Seru production is proposed. To cope with the dynamic seru production, we adopt an event-driven predictive-reactive scheduling method, i.e., when a dynamic event occurs, rescheduling is triggered to produce a new schedule with good efficiency. However, the new schedule may deviate significantly from the previous schedule. Therefore, it is critical to minimize schedule changes while maximizing efficiency (e.g. makespan) simultaneously. First, we formulate a multi-objective dynamic seru production model that applies to various dynamic events to minimize both schedule changes and makespan. Then, we develop dynamic NSGA-II to produce a new schedule. To effectively respond to dynamic events, a population re-initialization strategy is proposed to fast-track non-dominated solutions by making use of the characteristics of dynamic events, and a communication mechanism is designed to accelerate the convergence speed. However, these two mechanisms may lead to reducing its diversity. To maintain population diversity, a multi-population search strategy based on hyper-heuristic is adopted. The hyper-heuristic algorithm is incorporated to evolve an additional seru rescheduling population, which employs NSGA-ΙΙ as the high-level strategy to manipulate several low-level heuristics to generate efficient local search heuristic sequences. However, this approach also results in a longer computation time. Therefore, a parallel mechanism is designed to save computational time. Finally, we have conducted extensive experiments to demonstrate that our algorithm outperforms existing algorithms and provides several management insights.

ED-DQN: An event-driven deep reinforcement learning control method for multi-zone residential buildings

Residential Heating, Ventilation, and Air conditioning (HVAC) systems are responsible for a significant amount of energy consumption, but their management is challenging due to the complexities of building thermodynamics and human activities. Reinforcement learning (RL) has been adopted to tackle this issue, but traditional RL methods require massive training data, long learning periods, and frequent equipment adjustments. To address these issues, we construct a new event-driven Markov decision process (ED-MDP) framework, which enables adjustments of control policies triggered by events, reducing unnecessary operations. Moreover, we propose an event-driven deep Q network (ED-DQN) method, which optimizes the action selection based on the triggered events. In the HVAC control problem, the proposed ED-DQN can effectively capture dynamic non-linear features of thermal comfort, and reduce the equipment damage caused by frequent adjustments. Our experimental results show that compared to three benchmark methods and three RL methods, our ED-DQN achieved state-of-the-art performance in both energy saving and thermal comfort violations. Moreover, our method demonstrates promising performance when applied to new test thermal environments, indicating its robustness and adaptability for optimizing residential HVAC controls.

Event-Driven Non-Invasive Multi-Core Cable Current Monitoring Based on Sensor Array

Current monitoring of multi-core cables plays an important role in various applications since cables are widely utilized in distribution and transmission systems. However, existing current sensing devices which are mostly designed for the single conductor break the cable jacket to measure conductor currents in multi-core cables. Recently various non-invasive sensing devices with reconstruction algorithms are designed for rapid access to currents in multi-core cables but don't make full use of abundant information during current monitoring. Thus, this paper proposes an event-driven non-invasive multi-core cable current monitoring method using an array of magnetic field sensors. Real-time currents of the multi-core cable are determined by solving conductor localization and current measurement problems. Since the observed changes of the magnetic field around the cable are caused by current or position changes, conductor localization is driven by the events which are classified as electrical events and mechanical events respectively. The key idea is to improve the accuracy of conductor localization through effective electrical events. Then current monitoring is achieved by solving current measurement problems using more accurate conductor positions and the measured magnetic field. This method is utilized in the real cable current monitoring, a sustaining decrease of relative current errors is observed, which is below 1% in long-term monitoring.

Event-driven sorting algorithm-based Monte Carlo method with neighbour merging method for solving aerosol dynamics

A new event-driven sorting algorithm-based merging Monte Carlo (SAMMC) method is proposed and developed for solving the general dynamic equation in aerosol dynamics. A neighbour merging method is proposed to maintain a constant-volume and constant-number scheme with minimal interference to the numerical particle population, where absolute volume difference (AVD) and relative volume difference (RVD) are used as the crucial merging criteria. The SAMMC method can be used for simulating all aerosol dynamic processes with very high computational accuracy, especially effective in those aerosol dynamic processes generating additional numerical particles. In the present study, comprehensive computational conditions are used to study their impacts on computational accuracy and efficiency by comparing the SAMMC method to previous MC methods and analytical solutions. Numerical results show that the SAMMC method has excellent agreement with analytical solutions for all specified cases of different aerosol dynamic processes and shows higher computational accuracy than equal-weight-based MC methods. In addition, the computational accuracy of the SAMMC method in the total particle number concentration is much higher than those of the weighted fraction Monte Carlo (WFMC) method and sorting algorithm-based merging weighted fraction Monte Carlo (SAMWFMC) method in non-homogeneous coagulation. The SAMMC method can also achieve the same computational precision as the multi-Monte Carlo (MMC) method at only slightly higher computational cost in homogeneous coagulation. More importantly, the SAMMC method can deal with breakage-related processes and simultaneous coagulation and nucleation with very high computational accuracy and efficiency, while the numerical results of the MMC method may significantly deviate from analytical solutions due to the introduction of systematic errors. Furthermore, the RVD can achieve higher computational accuracy in multi-breakage modelling than AVD, but AVD and RVD have almost the same computational efficiency and accuracy in other aerosol dynamic processes.

Simulation-Based Method for the Calculation of Passenger Flow Distribution in an Urban Rail Transit Network Under Interruption

In the extensive urban rail transit network, interruptions will lead to service delays on the current line and spread to other lines, forcing many passengers to wait, detour, or even give up their trips. This paper proposes an event-driven simulation method to evaluate the impact of interruptions on passenger flow distribution. With this method, passengers are regarded as individual agents who can obtain complete information about the current traffic situation, and the impact of the occurrence, duration, and recovery of interruption events on passengers’ travel decisions is analyzed in detail. Then, two modes are used to assign passenger paths: experience-based pre-trip mode and response-based entrap mode. In the simulation process, the train is regarded as an individual agent with a fixed capacity. With the advance of the simulation clock, the network loading is completed through the interaction of the three agents of passengers, platforms, and trains. Interruption events are considered triggers, affecting other agents by affecting network topology and train schedules. Finally, taking Chongqing Metro as an example, the accuracy and effectiveness of the model are analyzed and verified. And the impact of interruption on passenger flow distribution indicators such as inbound volume, outbound volume, and transfer volume is studied from both the individual and overall dimensions. The results show that this study provides an effective method for calculating the passenger flow distribution of an extensive urban rail transit network in the case of interruption.

A hierarchical event-driven emergency control method utilizing battery energy storage system

Event-driven emergency control (EEC) executes control actions immediately following a risky disturbance, which serves as an effective and efficient scheme to restore power system stability and prevent cascading events. With the increasing variational energy sources in the power grid, battery energy storage systems (BESS) have been widely deployed in the system for response-driven control services such as primary frequency control, but the role of BESS in EEC is yet to be recognized. Viewing BESS as a type of switching power source, this paper aims to investigate the capability of BESSs to participate in power system EEC through immediate tripping. A hierarchical control method considering the deviated control costs among different types of control actions, including BESS shedding and load shedding, is proposed, where trajectory sensitivity analysis is used in a hierarchical way to efficiently decide the optimal control strategy. The proposed method has been tested on an IEEE 39-bus system integrated with widespread renewable energy sources (RES) and BESSs for event-driven emergency control. It is demonstrated that the proposed method can well utilize the available BESS resources to reduce the control burden and costs of load shedding while meeting the required stability criteria, which validates the efficacy of BESSs in event-driven emergency control of renewable power systems.

Improved event-driven simulation method for fuel transport in a mesh-like pipeline network

One of the tasks of an oil company is to fulfil site demands. Since not all sites have production, it is essential to transport products between sites. The order of pumping operations is determined by the scheduling plan. The scheduling plan includes the type and quantity of products to be delivered, the start of deliveries, and routes. The type and quantity of products to be delivered, the start of deliveries and routes are included in the scheduling plan. A scheduling plan is feasible if it does not contain inconsistencies. Checking a one-month schedule is a complex task in real multi-source mesh-like pipeline networks, thus it is advisable to use a discrete event simulation system. This work presents improvements on our previous simulation model. In the improved simulation model, it is possible to reroute, split, and reverse the flow direction of the products in the pipe, amongst other things.

Simulation of energy-efficient operation for metro trains: A discrete event-driven method based on multi-agent theory

Along with the rapid development of urban rail transit systems, the increase in operating mileage is the direct result of the enormous energy consumption. Among them, the energy consumption of train traction operation is close to half of the total energy consumption, leading to a gradual increase in operation cost. Based on the discrete event-driven method of multi-agent simulation, this paper simulates the process of the train operation with four function modules. Ingenuity, trains with onboard equipment would store braking regenerative energy and then transmit regenerative energy to other traction trains in the same power supply section. Moreover, the evaluation method of renewable energy utilization efficiency also is proposed with different operation schemes to calculate train operation energy consumption. Further, an adaptive large neighborhood search algorithm is designed to improve computational efficiency for real-time applications. Lastly, the operation of the Beijing Metro Yizhuang line is utilized as a case study to evaluate the simulation model and algorithm. The results show that the simulation method can reasonably simulate the process of train operation according to the actual situation. By optimizing the train operation time and dwell time of different onboard power storage equipment scenarios, the traction energy consumption can be reduced by 3.95% and up to 17.71%, which has a remarkable effect on energy saving.

An Intelligent Train Operation Method Based on Event-Driven Deep Reinforcement Learning

Train operation control in urban railways is challenging due to its high dynamics, complex environment, and level of comfort and safety. To address these challenges, in this article, the authors propose a new deep reinforcement-based train operation (DRTO) method which includes: 1) A deterministic deep reinforcement learning algorithm, 2) a dynamic incentive system, which is used to ensure safe operation in a multitrain environment, and 3) an event-driven method, which is used to improve the DRTO performance based on an event-driven strategy. To evaluate the performance, we thoroughly compare the proposed method with other operation control solutions on both synthetic and real datasets. Our results demonstrate that DRTO is effective in: 1) Decreasing the energy consumption of train operation, 2) increasing passenger comfort, and 3) achieving a good tradeoff between efficiency and safety. In addition, the effectiveness of the event-driven strategy and the dynamic incentive system is demonstrated in the experiments.

A Fine-Grained Modeling Approach for Systolic Array-Based Accelerator

The systolic array provides extremely high efficiency for running matrix multiplication and is one of the mainstream architectures of today’s deep learning accelerators. In order to develop efficient accelerators, people usually employ simulators to make design trade-offs. However, current simulators suffer from coarse-grained modeling methods and ideal assumptions, which limits their ability to describe structural characteristics of systolic arrays. In addition, they do not support the exploration of microarchitecture. This paper presents FG-SIM, a fine-grained modeling approach for evaluating systolic array accelerators by using an event-driven method. FG-SIM can obtain accurate results and provide the best mapping scheme for different workloads due to its fine-grained modeling technique and deny of ideal assumption. Experimental results show that FG-SIM plays a significant role in design trade-offs and outperforms state-of-the-art simulators, with an accuracy of more than 95%.