食品工程自动化控制（英文版）

Preface

Food quality quantization and process control are two important fields in
the automation of food engineering. Food quality quantization is a key technique
in automating food quality evaluation. Food quality process control is
the focus in food production lines. In the past 10 years, electronics and
computer technologies have significantly pushed forward the progress of
automation in the food industry. Research, development, and applications of
computerized food quality evaluation and process control have been accomplished
time after time. This is changing the traditional food industry. The
growth of applications of electronics and computer technologies to automation
for food engineering in the food industry will produce more nutritious,
better quality, and safer items for consumers.
The book describes the concepts, methods, and theories of data acquisition,
data analysis, modeling, classification and prediction, and control as
they pertain to food quality quantization and process control. The book
emphasizes the applications of advanced methods, such as wavelet analysis
and artificial neural networks, to automated food quality evaluation and
process control and introduces novel system prototypes such as machine
vision, elastography, and the electronic nose for food quality measurement,
analysis, and prediction. This book also provides examples to explain realworld
applications.
Although we expect readers to have a certain level of mathematical background,
we have simplified this requirement as much as possible to limit the
difficulties for all readers from undergraduate students, researchers, and engineers
to management personnel. We hope that the readers will benefit from
this work.
Outline of the Book
Six chapters follow the Introduction.
Chapter 2 concerns data acquisition (DAQ) from the measurement of
food samples. In Chapter 2, the issues of sampling are discussed with examples
of sampling for beef grading, food odor measurement, and meat quality
evaluation. Then, the general concepts and systems structure are introduced.
The examples of ultrasonic
A
-mode signal acquisition for beef grading, electronic
nose data acquisition for food odor measurement, and snack foodfrying data acquisition for process quality control are presented. Imaging
systems, as they are applied more and more in the area of food quality
characterization, are discussed in a separate section. Generic machine vision
systems and medical imaging systems are described. Image acquisition for
snack food quality evaluation, ultrasonic
B
-mode imaging for beef grading,
and elastographic imaging for meat quality evaluation are presented as
examples.
Chapter 3 is about processing and analysis of acquired data. In this
chapter, the methods of data preprocessing, such as data scaling, Fourier
transform, and wavelet transform are presented first. Then, the methods of
static and dynamic data analysis are described. Examples of ultrasonic
A
-
mode signal analysis for beef grading, electronic nose data analysis for food
odor measurement, and dynamic data analysis of snack food frying process
are presented. Image processing, including image preprocessing, image segmentation,
and image feature extraction, is discussed separately. The methods
of image morphological and textural feature extraction (such as Haralick抯
statistical and wavelet decomposition) are described. Examples of segmentation
of elastograms for the detection of hard objects in packaged beef rations,
morphological and Haralick抯 statistical textural feature extraction from
images of snack food samples, Haralick抯 statistical textural and gray-level
image intensity feature extraction from ultrasonic
B
-mode images for beef
grading, and Haralick抯 statistical and wavelet textural feature extraction
from meat elastograms are presented.
Chapter 4 concerns modeling for food quality quantization and process
control. Model strategies, both theoretical and empirical, are discussed first
in this chapter. The idea of an input杘utput model based on system identification
is introduced. The methods of linear statistical modeling and ANN
(artificial neural network) -based nonlinear modeling are described. In
dynamic process modeling, the models of ARX (autoregressive with exogenous
input) and NARX (nonlinear autoregressive with exogenous input) are
emphasized. In statistical modeling, examples of modeling based on ultrasonic
A
-mode signals for beef grading, meat attribute prediction modeling
based on Haralick抯 statistical textural features extracted from ultrasonic elastograms,
and snack food frying process ARX modeling are presented. In ANN
modeling, the examples of modeling for beef grading, modeling for food odor
pattern recognition with electronic nose, meat attribute prediction modeling,
and snack food frying process NARX modeling are presented.
Chapter 5 discusses classification and prediction of food quality. In this
chapter, the methods of classification and prediction for food quality quantization
are introduced first. Examples of beef sample classification for grading
based on statistical and ANN modeling, electronic nose data classification for
food odor pattern recognition, and meat attribute prediction based on statistical
and ANN modeling are presented. For food quality process control, the
methods of one-step-ahead and multiple-step-ahead predictions of linear and
nonlinear dynamic models, ARX and NARX, are described. The examples of one-step-ahead and multiple-step-ahead predictions for the snack food frying
process are presented.
Chapter 6 concentrates on food quality process control. In this chapter,
the strategies of IMC (internal model control) and PDC (predictive control)
are introduced. Based on the linear IMC and PDC, the ANN-based nonlinear
IMC and PDC, that is, NNIMC (neural network-based internal model control)
and NNPDC (neural network-based predictive control), are extended
and described. The algorithms for controller design also are described. The
methods of controller tuning are discussed. The examples of NNIMC and
neuro-fuzzy PDC for the snack food frying process are presented.
Chapter 7 concludes the work. This chapter is concerned with systems
integration for food quality quantization and process control. In this chapter,
based on the discussion and description from the previous chapters concerning
system components for food quality quantization and process control,
the principles, methods, and tools of systems integration for food quality
quantization and process control are presented and discussed. Then, the
techniques of systems development, especially software development, are
discussed for food quality quantization and process control.
Yanbo Huang
A. Dale Whittaker
Ronald E. Lacey
College Station, Texas
May 2001

============================

Contents


Chapter 1 Introduction
1.1 Food quality: a primary concern of the food industry
1.2 Automated evaluation of food quality
1.3 Food quality quantization and process control
1.4 Typical problems in food quality evaluation
and process control
1.4.1 Beef quality evaluation
1.4.2 Food odor measurement
1.4.3 Continuous snack food frying quality
process control
1.5 How to learn the technologies
References

Chapter 2 Data acquisition 11
2.1 Sampling
2.1.1 Example: Sampling for beef grading
2.1.2 Example: Sampling for detection of peanut
off-flavors
2.1.3 Example: Sampling for meat quality evaluation
2.1.4 Example: Sampling for snack food eating quality
evaluation
2.1.5 Example: Sampling for snack food frying quality
process control
2.2 Concepts and systems for data acquisition
2.2.1 Example: Ultrasonic
A
-mode signal acquisition
for beef grading
2.2.2 Example: Electronic nose data acquisition for food odor
measurement
2.2.3 Example: Snack food frying data acquisition
for quality process control
2.3 Image acquisition
2.3.1 Example: Image acquisition for snack food quality
evaluation
2.3.2 Example: Ultrasonic
B
-mode imaging for
beef grading
2.3.3 Example: Elastographic imaging for meat quality
evaluation
References

Chapter 3 Data analysis
3.1 Data preprocessing
3.2 Data analysis
3.2.1 Static data analysis
3.2.1.1 Example: Ultrasonic
A
-mode signal
analysis for beef grading
3.2.1.2 Example: Electronic nose data analysis for
detection of peanut off-flavors
3.2.2 Dynamic data analysis
3.2.2.1 Example: Dynamic data analysis of the
snack food frying process
3.3 Image processing
3.3.1 Image segmentation
3.3.1.1 Example: Segmentation of elastograms for
detection of hard objects in packaged
beef rations
3.3.2 Image feature extraction
3.3.2.1 Example: Morphological and Haralick抯
statistical textural feature extraction from
images of snack food samples
3.3.2.2 Example: Feature extraction from ultrasonic
B
-mode images for beef grading
3.3.2.3 Example: Haralick抯 statistical textural
feature extraction from meat elastograms
3.3.2.4 Example: Wavelet textural feature
extraction from meat elastograms
References

Chapter 4 Modeling
4.1 Modeling strategy
4.1.1 Theoretical and empirical modeling
4.1.2 Static and dynamic modeling
4.2 Linear statistical modeling
4.2.1 Example: Linear statistical modeling based on
ultrasonic
A
-mode signals for beef grading
4.2.2 Example: Linear statistical modeling for food
odor pattern recognition by an electronic nose
4.2.3 Example: Linear statistical modeling for meat attribute
prediction based on textural features extracted from
ultrasonic elastograms
4.2.4 Example: Linear statistical dynamic modeling
for snack food frying process control
?2001 by CRC Press LLC
4.3 ANN modeling
4.3.1 Example: ANN modeling for beef grading
4.3.2 Example: ANN modeling for food odor pattern
recognition by an electronic nose
4.3.3 Example: ANN modeling for snack food eating
quality evaluation
4.3.4 Example: ANN modeling for meat attribute
prediction
4.3.5 Example: ANN modeling for snack food frying
process control
References

Chapter 5 Prediction
5.1 Prediction and classification
5.1.1 Example: Sample classification for beef grading
based on linear statistical and ANN models
5.1.2 Example: Electronic nose data classification for
food odor pattern recognition
5.1.3 Example: Snack food classification for eating
quality evaluation based on linear statistical
and ANN models
5.1.4 Example: Meat attribute prediction based on
linear statistical and ANN models
5.2 One-step-ahead prediction
5.2.1 Example: One-step-ahead prediction for snack
food frying process control
5.3 Multiple-step-ahead prediction
5.3.1 Example: Multiple-step-ahead prediction for
snack food frying process control
References

Chapter 6 Control
6.1 Process control
6.2 Internal model control
6.2.1 Example: ANNIMC for the snack food
frying process
6.3 Predictive control
6.3.1 Example: Neuro-fuzzy PDC for snack food
frying process
References

Chapter 7 Systems integration
7.1 Food quality quantization systems integration
7.2 Food quality process control systems integration
7.3 Food quality quantization and process control systems
development
7.4 Concluding remarks
References