Customer Reviews

System Identification: Theory for the User (2nd Edition)

5 star:		(4)
4 star:		(2)
3 star:		(2)
2 star:		(1)
1 star:		(1)

 

 

Let me be the first to slam this book as having little practical use. The author uses a very unconventional, inconsistent notation that renders this book virtually unreadable. This is coming from somebody who has taken various courses in probability/stochastic processes, system dynamics, controls, and Kalman filtering.

The "theory" part in the title is dead on, but what about the "for the user" part? What user is this referring to, an ivory tower professor? Most of the problems at the end of each chapter begin with "prove", which in my opinion makes this a math book and not an engineering text. A practical example never surfaces in the text.

This book is very through in presenting methods of system identification...mathematically. However as with most good mathematical oriented books there are not a lot of practical design type problems. One other mark aginst the book is that similar notation is used for different topics, which can sometimes be a bit confusing. One thing I might recommend, to a potential buyer is that you take a course in random process, and possess a through understanding of signals and transforms. All in all though if a little more meat in terms of practicality were added I'd rate this book a five.

I have two system identification books on my table right now: One is by Prof Ljung and the other is 'Time Series: Theory and Methods' by Profs. Brockwell and Davis, and I must say that I refer to the latter much more often than the former. The primary reason for this is that Prof. Ljung's book uses notation very inconsistently. This makes for very difficult reading. On the other hand 'Time Series: T&M' is consistent and an excellent resource. Both books are of an advanced nature, suitable for graduate students. Again, 'Time Series T&M' is more rigorous.

Nonetheless, if you plan on using Matlab's Sys Id toolbox, I would recommend Prof. Ljung's book. He wrote the toolbox, and he refers heavily in the Matlab documentation to the book.

I have taught system identification at graduate level for over 15 years. Ljung's System Identification is my first choice as a thorough reference book. However, I have still to find a person who finds the book helpful to the beginner. If you are looking for a text-book to teach system identification to beginners or if you are such a beginner, even at a graduate level, you better look for something more readable. On the other hand, keep Ljung's book handy, it will be hard to find a topic in LINEAR system identification which is not covered by this book. As for readability, I would give it one or two (max) stars (and the second edition improved on the first!). As for thoroughness, it is a five-star book.

The book is not readable because the notaion
is very confusing and it is very easy to get lost due to this careles
notation. The objective is not clear at most places and the book does not provide
any intitution. It is not suitable for self-study.
Poeople in the field of system identification should try other books.

This book is in all sense the angular stone for anyone who desire to start in a serious manner to develop skills in the System Identification subject.

This book has all the options, the "theory for the user" are the "kitchen recipies" proper of the topic, this is the techniques that you can use without understand the background, and in the other hand it has all the neccesary theoretical support to develop a clear and complete serious idea of the subject, in which case the posibility of gain a much more profound comprehension of the System Identification field is at hand in the text of the book.This is a very specific speciallity book.

The notation can be confusing at first, but it should not be a decisive point for someone who is ready to learn the System Identification subject, in other words, if you have the necessary prerrequisites for study this topic (and you must have it if you has reached this far) the notation wont be a major problem.

Excellent binding, a high quality product in all aspects.

(Sorry for mi English)

Contrary to other reviewers, I have used this book for very PRACTICAL USER applications. Yes, the book is "heavy" in mathematical proofs and the notation will keep you focused. However, you do not need the proofs to apply the material. They are there for those (I am NOT one of them) who would like to read/understand them. PRACTICAL System Identification is by its nature an application of stable recursive computations and thus is indeed "heavy" in mathematics. I applied very successfully the recursive methods, which I studied in the first edition of this book, in my early System Identification Control projects. Was it easy? No, not by any means. It took some very careful studying and very careful programming of the recursive least square algorithms. Then, with the advent of the Matlab System Identification Toolbox, the topic of system identification (using Ljung's 2nd edition of "System Identification: Theory for the User") became indeed very accessible for the USER. Yes, you will need to have a relative good understanding of Stochastic Processes. However, since the topic of system ID is graduate material, I would anticipate that most readers will have studied stochastic processes as part of their undergrad studies.
Many years ago I was so fortunate to be able to attend a Matlab organized conference in Boston near MIT. As part of this conference a System ID Toolbox workshop was given by Prof. Ljung. It was this workshop, this Toolbox and this text, that took systems ID from my "basement" levels to the very applied levels. Even though I am now retired, I still spend most my days trying to keep up with the advances in Control Systems. I gave my old copy of this book to an ex-colleague. Now I miss the book; so I just ordered another copy.

This book is a fundamental volume for whoever is interested in identification techniques. Theory is explained clearly.

Covred basic and advanced topics in system Id. sield. It is good to read it while reading the System Id user's guide in MATLAB.

An excellent textbook for engineers, and a necessary reference for Matlab toolbox (System Identification).