Product Quality Prediction Research Articles

Instrumental Texture Profile Analysis (TPA) of Shelled Sunflower Seed Caramel Snack Using Response Surface Methodology

Models capable of predicting product quality of shelled sunflower seed caramel snack have been developed using response surface methodology. The textural profile analysis was conducted on the snacks using a texture analyzer. The quality attributes measured were hardness, cohesiveness, springiness, chewiness, and resilience as a function of sugar and sunflower kernels content. The sugar and shelled seed proportions affect the textural characteristics of the product significantly (p<0.05). The values of hardness, cohesiveness, springiness, chewiness, and resilience varied from 2.048 to 42.030 N, 1.002 to 5.003, 1.138 to 1.69, 2.773 to 228.146N, and 0.301 to 0.779, respectively. The highest values of hardness and chewiness were attained for the product with 70:30 sugar and shelled sunflower seed proportion respectively. Similarly the highest values of cohesiveness, springiness and resilience were observed in 50 : 30, 50 : 40, and 50 : 50 proportions respectively. The lowest values of hardness and chewiness were observed in 50 : 50 (sugar: shelled sunflower seed) proportion respectively. Similarly the lowest values of cohesiveness were observed in 70: 50 whereas the lowest values of springiness and resilience were observed in 70 : 30 proportions respectively. Hardness, cohesiveness, and chewiness trended to increase whereas springiness and resilience decreased with increase in sugar proportion.

Product lifecycle-oriented quality and productivity improvement based on stream of variation methodology

Manufacturers in the 21st century will face increased customization, product proliferation, shorter product lifecycle and development time, responsiveness, frequent and unpredictable market changes. To gain the competitive advantage, manufacturing companies must be able to analyze and predict product quality during the product design phase and identify root causes of all faults for productivity improvement during ramp-up time and production time. Therefore, a new manufacturing strategy, namely, stream of variation (SoV) methodology, has been proposed, developed and applied. The SoV methodology is a generic math model for modeling, analysis, prediction and control of product quality and productivity improvement in complex multistage manufacturing systems such as automotive, aerospace, appliance, and electronics industries. The methodology integrates multivariate statistics, control theory and design/manufacturing knowledge into a unified framework and can help in eliminating costly trial-and-error fine-tuning of new-product manufacturing processes throughout the product design and manufacturing. The related issues of the state-of-the-art practice, goals, benefits and future directions related to SoV methodology are discussed, which include rationale of SoV, state space model, root causes identification, etc. An application example is also provided.

Real-time monitoring and chemical profiling of a cultivation process

A method for at-line quality assessment of a cultivation process is developed in order to (1) enable improved process control, (2) enable faster detection of batch end point, and (3) enable immediate quality assessment of final product. Fluorescence excitation–emission measurements are used because they are known to reflect important properties of the fermentation process. Focus is also on important sampling issues—mainly structurally sub-optimal primary sampling methods affecting the representativity obtainable relative to the lot characteristics. Several different calibration approaches are investigated for predicting product quality (enzymatic activity). The fluorescence data are also modeled by a PARAFAC model, providing a chemically interpretable visualization of the process variation thereby enhancing the possibilities for gaining in-depth process understanding. The results of our investigations are that enzyme activity can be predicted directly from fluorescence measurements with an uncertainty comparable to the uncertainty of the chemical reference analysis for the enzyme activity. Chemical profiling of the cultivation process using PARAFAC verifies basic understanding of the cultivation process. An enzyme marker profile as well as a tryptophan (protein marker) profile is identified.

Quantitative model for evaluating the quality of an automotive business process

One of the key issues to business process control is the identification of measurable process attributes. For manufacturing processes these are typically physical parameters of the process (e.g. temperature, set points) or physical attributes of the manufactured product (e.g. dimension, functional performance). However, for business processes the metrics are more abstract. The challenge has been to develop metrics that capture the contributing subtle and hard to measure factors for business process control. This paper presents an analytical model that uses the weights-of-evidence concept to convert answers to audit or self-assessment questions into a single numerical process quality index. This index is used to forecast process success or failure and monitor its performance from start to end. The application of the approach is illustrated with an automotive industry product development sub-process where the process performance metric is the field warranty data, i.e. incidents per thousand vehicles (IPTV). The analytical model converts process self-assessment (failure mode and effect analysis) questions into a single numeric process quality index. The validity of the model is reflected in the strength of the correlation between the index and the IPTV results. Also, in this paper a measure is developed for identifying critical process quality assessment questions. This measure quantifies the deviation in the automotive business process that should have more focus. The significance of the analytical model proposed in this research is that the project managers or quality assurance auditors may be able to use the metric to predict product quality at any point in the product development process.

Apple firmness: Creating a tool for product evaluation based on a sensory–instrumental relationship

In studies with trained and untrained participants, we have demonstrated that people can discriminate differences in texture between apples that differ by more than ∼5 N in instrumental puncture tests [Harker, F.R., Gunson, F.A., Brookfield, P.L., White, A., 2002a. An apple a day: the influence of memory of consumer judgement of quality. Food Qual. Prefer. 13, 173–179; Harker, F.R., Maindonald, J., Murray, S.H., Gunson, F.A., Hallett, I.C., Walker, S.B., 2002b. Sensory interpretation of instrumental measurements 1: texture of apple fruit. Postharvest Biol. Technol. 24, 225–239]. Using this information, it is possible to estimate the probability that a consumer will perceive a difference between apples selected at random from two lines of fruit (e.g. apples harvested from different orchards) based on knowledge of the distribution of instrumental values in each product line. The data we present demonstrates the overwhelming impact of biological variability on attempts to use instrumental–sensory relationships to predict product quality in apples. Calculations predicted that two lines of fruit needed to differ by about 12 N (mean puncture force) in order for consumers to perceive a difference in sensory texture ( p = 0.95). At a practical level, curves are provided to enable researchers to infer the probability with which consumers may perceive differences between apples from treatments that differ in mean puncture force.

Towards a Crystalline Product Quality Prediction Method by Combining Process Modeling and Molecular Simulations

AbstractIt is shown for the first time that a combination of molecular simulations and process modeling can be used to predict crystalline product quality aspects such as the polymorphic fraction and crystal size distribution. Using molecular simulations the interfacial energy of 3D nuclei and the step free energy of 2D nuclei on crystal surfaces of different polymorphs are determined. These result in the nucleation and growth behavior of the polymorphs as a function of the supersaturation. The supersaturation development during polymorph crystallizations can be accounted for using dynamic process modeling. Furthermore, this study clearly demonstrates that a deeper understanding of polymorph crystallization can be obtained by combining research on a molecular scale with that on a process scale. Molecular simulations show, for instance, that there is no simple relationship between experimental solubility and apparent interfacial energy of the polymorphs.

Product quality prediction by a neural soft-sensor based on MSA and PCA

A novel soft-sensor model which incorporates PCA (principal component analysis), RBF (Radial Basis Function) networks, and MSA (Multi-scale analysis), is proposed to infer the properties of manufactured products from real process variables. PCA is carried out to select the most relevant process features and to eliminate the correlations of input variables; multi-scale analysis is introduced to acquire much more information and to reduce uncertainty in the system; and RBF networks are used to characterize the nonlinearity of the process. A prediction of the melt index (MI), or quality of polypropylene produced in an actual industrial process, is taken as a case study. Research results show that the proposed method provides promising prediction reliability and accuracy.

Quality prediction in manufacturing system design

Expected product quality is affected by manufacturing system design decisions. Product quality prediction would allow a manufacturer to make better choices of system parameters at the early design stage and, hence, enhance competitiveness through achieving higher quality levels. A Configuration Capability Indicator (CCI) that maps the manufacturing system configuration parameters into an expected product quality level has been developed. A hierarchical fuzzy inference system was developed for modeling the relationship between manufacturing system design parameters and the resulting product quality level. A configuration capability zone is proposed to graphically represent the configuration capability, for a system configuration that produces more than one product, and compare it to the benchmark six-sigma capability. The developed model has been applied to two case studies (Test Parts ANC-90 and ANC-101, and Rack Bar Machining) with different system configurations for illustration and verification. The results demonstrate the ability of the CCI to compare different system configurations from a quality point of view and to support the decision-making during the early stages of manufacturing system planning and design.

FEED-FORWARD CONTROL OF WATER AND NUTRIENT SUPPLY IN GREENHOUSE HORTICULTURE: DEVELOPMENT OF A SYSTEM

Minimisation of nutrient and water loss to the environment, prediction and planning of production and product quality, such that they meet the demands of customers, and increased financial return are some of the goals of current greenhouse horticulture. This requires an integrated system for monitoring and control of the nutrient solution, plant growth and product quality. This paper describes the components of such a system. Tomato was selected as example crop. Plant sensors for photosynthesis, radiation interception and fresh growth rate were developed and tested, and the combined plant-substrate model was validated against greenhouse experiments and commercial growth data. Techniques for auto-calibration of the crop model with sensor information were developed. Scenarios describing various fertigation regimes were defined, and a technique for the generation of set points for water and nutrient application on the basis of pre-defined optimisation goals was developed. These goals included drainage volume and nutrient concentration, nutrient application rate, fruit dry matter concentration, and fruit growth rate. The overall system was evaluated in a greenhouse experiment. In comparison with simulated ‘normal’ cultivation, optimisation of water and nutrient application indeed resulted in reduced application and drainage rates, increased fruit growth rate, and a dry matter concentration that met the optimisation goal. In conclusion, the monitoring and control system offers good prospects for efficient control of water and nutrient use, and control of crop growth in future control of greenhouse cropping systems.

Fine-tuning of the rolling mill using an Internet-based on-line application-processing interface

In order to meet the challenge of steady growing pressure to improve product quality, rolling mills employ extensive automation and sophisticated on-line data sampling techniques. However, the steel sheet quality is influenced by the entire ‘life history’ of the rolled strip. Application of the self-organizing maps helps to discover hidden dependencies influencing the quality parameters, such as flatness, profile, wedge, thickness and width deviations and surface quality, as well as mechanical and magnetic properties. A dataset collected from the measured parameters in a strip location corresponds to a proper process state, described by a point in the state space. Time series of points corresponding to each sampled data set show how the system ‘moves’ in the state space and how the predicted product quality is changing within the strip.

Batch process monitoring and its application to polymerization systems

AbstractSlight changes in raw material properties or operating conditions during critical periods of operation of batch and semi‐batch polymerization reactors may have a strong influence on reaction mechanism and impact final product quality. Online process monitoring, fault detection, fault diagnosis, and product quality prediction in real‐time ensure safe reactor operation and warn operators about excursions from normal operation that may lead to deterioration in product properties. Multivariate statistical process monitoring and quality prediction using multiway principal components analysis and multiway partial least squares have been successful in detecting abnormalities in process operation and product quality. When abnormal process operation is detected, fault diagnosis tools are used to determine the source cause of the deviation. Illustrative case studies are presented via simulated polyvinyl acetate polymerization.

Digital Imaging for Process Monitoring and Control with Industrial Applications

Abstract The development of on-line digital imaging systems for process monitoring and control is illustrated through two industrial applications: i) the control of coating concentration and distribution on snack food products. and ii) the monitoring of boiler systems through imaging of the combustion processes. Feature information extracted from images using Multivariate Image Analysis (MIA) based on Principal Component Analysis (PCA). is used to develop models to predict product quality and process property variables. The imaging systems are used to monitor these product quality and process property variables, to detect and diagnose operational problems in the plants, and to directly implement closed-loop feedback control.

Recursive data-based prediction and control of product quality for a PMMA batch process

In many batch processes, frequent process/feedstock disturbances and unavailability of direct on-line quality measurements make it very difficult to achieve tight control of product quality. Motivated by this, we present a simple data-based method in which measurements of other process variables are related to end product quality using a historical data base. The developed correlation model is used to make on-line predictions of end quality, which can serve as a basis for adjusting the batch condition/time so that desired product quality may be achieved. This strategy is applied to a methyl methacrylate (MMA) polymerization process. Important end quality variables, the weight average molecular weight and the polydispersity, are predicted recursively based on the measurements of reactor cooling rate. Subsequently, a shrinking-horizon model predictive control approach is used to manipulate the reaction temperature. The results in this study show promise for the proposed inferential control method.

APPLIED GENOMICS - AN INNOVATIVE TOOL TO IMPROVE QUALITY IN CHAINS

Genomics technologies can be applied for unravelling the biological processes that are important for pre- and post-harvest quality. Such knowledge will allow precise monitoring of physiological condition and can be used as a tool for optimising production chains. A combination with quality change models will allow prediction of future product quality with much more precision and certainty than is currently possible. To demonstrate the validity of this approach we set up a project to identify, via cDNA microarray analysis, gene expression patterns that are correlated to the development of mealiness during storage of Cox apples (Malus x domestica `Cox?s Orange Pippin?). The experimental set-up encompassed, next to mRNA profiling, both instrumental and sensory analyses of apples at various time points before and during storage. We were able to select a subset of genes the expression of which is indicative of reduced sensory quality (mealiness and off-flavour). Moreover we detected batch differences in initial profiles that could be related to storage quality. Such indicators can be used as decision support tools for growers and traders.

Neural virtual sensor for the inferential prediction of product quality from process variables

A predictive Fuzzy ARTMAP neural system and two hybrid networks, each combining a dynamic unsupervised classifier with a different kind of supervised mechanism, were applied to develop virtual sensor systems capable of inferring the properties of manufactured products from real process variables. A new method to construct dynamically the unsupervised layer was developed. A sensitivity analysis was carried out by means of self-organizing maps to select the most relevant process features and to reduce the number of input variables into the model. The prediction of the melt index (MI) or quality of six different LDPE grades produced in a tubular reactor was taken as a case study. The MI inferred from the most relevant process variables measured at the beginning of the process cycle deviated 5% from on-line MI values for single grade neural sensors and 7% for composite neural models valid for all grades simultaneously.

Novel computational tools in bakery process data analysis: a comparative study

We studied the potential of various machine learning and statistical methods in the prediction of product quality in industrial bakery processes. The methods included classification and regression tree, decision list, neural network, support vector machine and Bayesian learning algorithms as well as statistical multivariate methods. Our data originated from two industrial bakery processes: a sourdough rye bread and a Danish pastry process. In our studies, the Naive Bayesian algorithm turned out to be the best classifier building algorithm while the partial least squares (PLS) method was the best regression method. The prediction accuracy of these models improved significantly by pruning the original set of variables. In this study, two response variables could be predicted on a level that justifies further study: rye bread pH could be predicted with high accuracy with Naive Bayesian Classifier, and Danish pastry height could be predicted with a moderately high correlation with PLS.

Principal Surface Model Based Quality Control Approach for Batch/Semi-Batch Processes

Batch/semi-batch quality control has attracted much attention from chemical engineering researchers recently. In this work, a novel empirical model based quality control approach for batch processes is developed. The model implemented is based on independent nonlinear principal components obtained by means of principal surface analysis- a very hot topic in the area of statistical modeling. The partial initial conditions and intermediate measurements obtained from each batch are analyzed using this approach to extract the independent nonlinear principal components. An empirical model with the nonlinear principal components as inputs is built to predict product quality. Correcting actions based on the initial conditions and intermediate measurements are taken based on the predictions obtained using the empirical model. In this work, the problem formulation and solution method are presented. The illustrative examples include a simulation example and an experimental study on the batch reactive distillation process. The simulation and experimental results show that the proposed approach is highly promising.

Inferential control of product quality attributes—application to food cooking extrusion process

Reducing product quality variability is one of the most important issues for the process industry. As product quality attributes (PQAs) are often measured in a laboratory using off-line sensors, direct feedback control is not feasible unless expensive on-line sensors are used. Inferential control provides a means to the solution of this type of problem without the cost of expensive sensors. This paper reports a successful application of control of product quality attributes in a food cooking extruder. It covers the essential steps in achieving the control objective: (1) identifying the influential variables that can be measured on-line and have significant impact on the product quality attributes; (2) estimating a dynamic model between the influential variables and the PQAs for on-line prediction of product quality; (3) on-line feedback control using the influential variables. Experimental results demonstrate that the designed control system not only maintains the quality of the products but also automatically brings the PQAs to user's specification.

Adaptive receptive fields for radial basis functions

We propose a network architecture based on adaptive receptive fields and a learning algorithm that combines both supervised learning of centers and unsupervised learning of output layer weights. This algorithm causes each group of radial basis functions to adapt to regions of the clustered input space. Networks produced by this algorithm appear to have better generalization performance on prediction of non-linear input-output mappings than corresponding back-propagation algorithms and requires a fraction of the number of connection weights required by fixed center radial basis. For a test problem of predicting product quality of a reverse osmosis desalination plant, the network learns much faster than a three-layer perceptron trained with back-propagation, but requires additional computational burden.

Recursive data‐based prediction and control of batch product quality

AbstractIn typical batch and semibatch processes, process/feedstock disturbances occur frequently and on‐line measurements of product quality variables are not available. As a result, most batch processes have not been able to achieve tight quality control. Empirical, data‐driven approaches are very attractive for dealing with this problem because of the difficulties associated with developing accurate process models from first principles. An approach for recursive on‐line quality prediction was developed around data‐based model structures. Techniques designed to incorporate the predictive models into on‐line monitoring and control of batch product quality were also examined. The proposed control approach can be viewed as shrinking‐horizon model‐predictive control based on empirical models. The effectiveness of the proposed prediction and control methods are illustrated by using an industrially relevant simulated polymerization example.