Sanjay Srivastava

Modeling of manufacturing processes with ANNs for intelligent manufacturing

Modern manufacturing often caters to rapidly changing product specifications determined by the continuously increasing productivity, flexibility and quality demands. Metal forming and machining are two important manufacturing processes in present day manufacturing. Automatic selection of tools and accessories in these processes heavily relies on forming force/cutting force estimation. Complex relationships exist between process parameters and these forces. In the present work, the applicability and relative effectiveness of Artificial Neural Network based models has been investigated for rapid estimation of these, invoking the function approximation capabilities of the ANN models. The results obtained are found to correlate well with the finite element simulation data in cases of metal forming, and experimental data in cases of metal cutting. This work has considerable implications in selection of the tools and on-line monitoring of tool wear. The actual forming and cutting forces can be compared with predicted ones to signal the onset of tool wear, and thus prevent damage to the tool and work piece during the course of manufacturing.

Project Scheduling: Time-Cost Tradeoff Problems

We design and implement new methods to solve multiobjective time-cost tradeoff (TCT) problems in project scheduling using evolutionary algorithm and its hybrid variants with fuzzy logic, and artificial neural networks. We deal with a wide variety of TCT problems encountered in real world engineering projects. These include consideration of (i) nonlinear time-cost relationships of project activities, (ii) presence of a constrained resource apart from precedence constraints, and (iii) project uncertainties. We also present a hybrid meta heuristic (HMH) combining a genetic algorithm with simulated annealing to solve discrete version of multiobjective TCT problem. HMH is employed to solve two test cases of TCT.

Integrated ANN-HMH Approach for Nonlinear Time-Cost Tradeoff Problem

AbstractThis paper presents an integrated Artificial Neural Network - Hybrid Meta Heuristic(ANN-HMH) method to solve the nonlinear time-cost tradeoff(TCT) problem of real life engineering projects. ANN models help to capture the existing nonlinear time-cost relationship in project activities. ANN models are then integrated with HMH technique to search for optimal TCT profile. HMH is a proven evolutionary multiobjective optimization technique for solving TCT problems. The study has implication in real time monitoring and control of project scheduling processes.

MOGA-based time-cost tradeoffs: Responsiveness for project uncertainties

Existing methods for analyzing the responsiveness of time-cost tradeoff (TCT) profiles with regard to project uncertainties ignore the cost parameter of project activities. To comprise this problem a novel method is developed in this work - it examines the effects of project uncertainties on both, the duration as well as the cost of the activities. The method integrates the two important paradigms of soft computing - multiobjective genetic algorithm (MOGA), and fuzzy logic - the method is referred to as the integrated Fuzzy-GA. A rule based fuzzy logic framework is developed which brings up the changes in the duration and the cost of each activity for the inputted uncertainties. The framework is integrated with MOGA which would search for Pareto-optimal front (a TCT profile) for a given set of time-cost pair of each project activity. Two standard test problems from the literature are successfully attempted using MOGA. A test case of TCT problem is solved using the integrated Fuzzy-GA. The method provides an efficient tool to project manager to carry out a sensitivity analysis of time-cost tradeoff profiles under varying conditions of existing uncertainties normally encountered in realistic projects.

Integrated Fuzzy–HMH for project uncertainties in time–cost tradeoff problem

Abstract Time–cost tradeoff (TCT) problem in project scheduling studies how to schedule project activities to achieve a tradeoff between project cost and project completion time. It gives project planners both challenges and opportunities to work out the best plan that optimizes time and cost to complete a project. In this paper, we present a novel method which examines the effects of project uncertainties on both, the duration as well as the cost of the activities. This method integrates a fuzzy logic framework with Hybrid Meta-Heuristic. Hybrid Meta-Heuristic (HMH) is an innovative approach which hybridizes a multiobjective genetic algorithm and simulated annealing. Integration of HMH and fuzzy logic is referred to as ‘integrated Fuzzy–HMH’. A rule based fuzzy logic framework brings up changes in the duration and the cost of each activity for the input uncertainties and HMH searches for Pareto-optimal front (TCT profile) for a given set of time–cost pair of each project activity. Two standard test problems from the literature are attempted using HMH. A case study of TCT problem is solved using integrated Fuzzy–HMH. The method solves time–cost tradeoff problems within an uncertain environment and carries out its sensitivity analysis.

Multi-resource-constrained discrete time-cost tradeoff with MOGA based hybrid method

This work focuses on solving multi-resource-constrained discrete time-cost tradeoff (MRCDTCT) problems in project scheduling using a novel methodology which hybridizes a heuristic with a multi-objective genetic algorithm - a hybrid MOGA. The proposed method is pertinent for the real world project scheduling where the resources are constrained, and where generation of complete Pareto-optimal front is essential for a decision-maker. Accordingly, the entire time-cost tradeoff (TCT) profile is identified, wherein the decision-maker basically makes a sequence of decisions, which optimizes the overall performance of the project in terms of time and cost by satisfying not only the precedence constraints but also the resource constraints. A MOGA, is devised to search for the optimal profile, and a heuristic is developed and hybridized with MOGA to ensure the availability of the resource requirements for each instance of the project schedule of MOGA by adjusting the start time of non-critical activities exploiting their float values. To demonstrate the efficacy of MOGA employed in this work, two standard test problems from the literature are attempted. The hybrid MOGA is employed to solve two test problems of TCT, one with a single resource constraint, and the other with three resource constraints. The results are also compared with those of exhaustive enumeration technique - the close proximity between these results validates the suitability and accuracy of hybrid MOGA to solve MRCDTCT problems.

Optimizing the risk of occupational health hazard in a multiobjective decision environment using NSGA-II

We present a novel system to lessen the risk of occupational health hazards (OHH) of workers in the labor intensive industrieswith a job-combination approach. The work is carried out in a brick manufacturing (BM) unit at Hathras, India. The risk of OHH is assessed in terms of perceived discomfort level (PDL) of workers. PDL is computed with factor rating (FR) method. It is observed based on an initial survey in the BM unit that the workers, in general, aim to maximize their earnings by subjecting themselves to extreme work conditions due to economic reasons, and hence are exposed to greater risk of OHH resulting in higher values of PDL. We employ NSGA-II, an evolutionary multiobjective optimization (EMO) technique, to search for optimal PDL-earning tradeoff (PET) profile with two conflicting objectives, viz. minimization of PDL, and maximization of earnings.

Reducing the risk of heat stress using artificial neural networks based job-combination approach

We design and implement a system to reduce the risk of heat stress, a recognized occupational health hazard (OHH), in two labor intensive industries using a job-combination approach. A novel feature of the system is employing artificial neural networks (ANNs) as model free estimators to evaluate perceived discomforts (PDs) of workers for different job combinations proposed in the work.

An intelligent system to address occupational health of workers exposed to high risk jobs

Workers in labor-intensive manufacturing units, in general, maximize their earnings by subjecting themselves to greater risk of occupational health hazards (RoOHH) mainly due to economic reasons. To embark upon this issue, we introduce an intelligent system employing artificial neural network (ANN) and non-dominated sorting genetic algorithm (NSGA-II). Experimentations are carried out in a brick manufacturing unit in India. Observations spell out that firing is the most severe job among others. A job-combination approach is incorporated wherein firing workers do another job along with firing to reduce their exposure to high temperature zone while maintaining their earnings to a satisfactory level. RoOHH is measured in terms of risk assessment score (RAS). ANN models the psychological responses of workers in terms of RAS, and facilitates the evaluation of one of the fitness function of NSGA-II. NSGA-II searches for optimal work schedules in a job-combination to minimize RAS and maximize earnings simultaneously.

Discrete Time-Cost Tradeoff with a Novel Hybrid Meta-Heuristic

In this paper, we present a new hybrid meta-heuristic (HMH) technique for solving multiobjective discrete time-cost tradeoff (TCT) problem in project scheduling. The proposed technique hybridizes a multiobjective genetic algorithm and simulated annealing, and is apposite for problems where generation of complete Pareto front, a TCT curve in this case, is essential for a decision-maker. Discrete TCT problem is known to be NP-hard. We solved two test problems of discrete TCT using HMH – on comparing the Pareto front results of HMH with those of analytical method, HMH performs well in terms of efficiency and accuracy.