【正文】 relations, mapped to a standardized theory field. Usually, the corresponding relationship between off for quantitative factors. For ease of handling, BiaoZhunLun domain such as bulk chemical separation, and then to fuzzy partition of discourse, define the fuzzy subset, such as NB, PZ, PS, etc. The same fuzzy control rule base, fuzzy partition to the fundamental theory of domain is different, the control effect is also different. Specifically, correspondence, BiaoZhunLun domain, the number of fuzzy subset, and the membership function of fuzzy subset has a great influence on the control effect. These three kinds of parameters and the fuzzy control rules have the same importance, therefore to merge them into fuzzy controller parameter database, together with the fuzzy control rule base of the knowledge base. 2. The blur Convert accurate input into fuzzy quantity F there are two ways: (1) converts gauged BiaoZhunLun fuzzy single point sets on the domain. Gauged by the corresponding relation between x x G into BiaoZhunLun domain on the basic elements, then the elements of the fuzzy single point set F UF (u) = 1 if u = G (x) (2) converts gauged BiaoZhunLun domain of fuzzy subsets. Gauged by the corresponding relationship into BiaoZhunLun domain on the basic elements, on the element has the maximum membership degree of fuzzy subsets, namely for the precise amount corresponding fuzzy subset. 3. The fuzzy inference 12 The most basic form of fuzzy reasoning is: 1 IF A THEN B Premise 2 IF A 39。 Among them, A, A 39。 for the theory of fuzzy subset V on the domain. Premise 1 is called the fuzzy implication relations, to A and B. In practice, the general rules of first in view of the individual reasoning, then the reasoning result sum and eventually reasoning results are obtained. 4. Accurate Reasoning of fuzzy subset to convert accurate value, to get the final control output y. Two accurate methods monly used at present: (1) the maximum membership degree method. In the reasoning of fuzzy subset, the selection membership degree of the largest BiaoZhunLun domain element as the average of the accurate results. (2) the gravity method. Will get the membership function of fuzzy subset reasoning abscissa and the area around the center of gravity of the corresponding BiaoZhunLun domain elements as accurate results. After the reasoning results precision value, still should according to corresponding relation, get the final control output y. the classification of the fuzzy controller The type of fuzzy control are: (1) the basic fuzzy controller: after fuzzy control table was determined, control rules are fixed。 (3) the intelligent fuzzy controller, it brings people, artificial intelligence and neural work, and realize the integrated information processing, make the system both with flexible inference mechanism, the heuristic knowledge and production rule says, but also has many layers, choice of multiple types of control law. the design of fuzzy controller 13 Fuzzy controller plays an important role in fuzzy automatic control system, so the fuzzy control system, the design and adjustment of fuzzy controller is very important. The design of fuzzy controller includes the following content: 1, to determine the fuzzy controller input variables and output variables。 3, establish fuzzy and the fuzzy method。 j = 1, 2,..., m), type of Ai Bj Cij respectively defined in the theory of error, error change and control the amount of X, Y, Z in fuzzy sets。但由于模糊規(guī)則是人們對過程或?qū)ο竽：畔⒌臍w納，對高階、非線性、大時滯、時變參數(shù)以及隨機(jī)干擾嚴(yán)重的復(fù)雜控制過程，人們的認(rèn)識往往比較貧乏或難以總結(jié)完整的經(jīng)驗，這就使得單純的模糊控制在某些情況下很粗糙，難以適應(yīng)不同的運行狀態(tài)，影響了控制效果。在實際應(yīng)用中，往往是將模糊控制或模糊推理的思想，與其它相對成熟的控制理論或 方法結(jié)合起來，發(fā)揮各自的長處，從而獲得理想的控制效果。對模糊控制的改進(jìn)方法可大致的分為模糊復(fù)合控制，自適應(yīng)和自學(xué)習(xí)模糊控制，以及模糊控制與智能化方法的結(jié)合等三個方面。一種簡便有效的做法是模糊控制器和 I 調(diào)節(jié)器共同合成控制作用。 史密斯 模糊控制器 :針對系統(tǒng)的純滯后特性設(shè)計，用模糊控制器替代 PID 可以解決常規(guī)史密斯 PID 16 控制器對參數(shù)變化適應(yīng)能力較弱的缺陷 。 三維模糊控制器 :一種是利用誤差 E，誤差變化 Ec和誤差變化速率 Ecc作為三維變量，可以解決傳統(tǒng)二維模糊控制器的 快速響應(yīng)與穩(wěn)定性要求之間的矛盾 。 多變量模糊控制 :一般采用結(jié)構(gòu)分解和分層分級結(jié)構(gòu)，利用多個簡單的模糊控制器進(jìn)行組合，并兼顧多規(guī)則集之間的相互關(guān)系。基于模糊模型的自校正模糊控制器，包括 利用模糊集理論辨識系統(tǒng)模型的語言化方法，基于參考模糊集的系統(tǒng)模糊關(guān)系模型辨識方法，以及由 I/O 數(shù)據(jù)建立模糊規(guī)則模型，并以此作為自校正控制器設(shè)計的基礎(chǔ)等。基于模糊推理的 PID 自整定控制，如參數(shù)自整定模糊 PD控制，以及類似的 PI 及 PID 控制等。 具有自學(xué)習(xí)功能的模糊控制 :包括多種對外擾影響或重復(fù)任務(wù)的性能具有自學(xué)習(xí)功能的模糊控制方法，以及自尋優(yōu)模糊控制器等，其關(guān)鍵在于學(xué)習(xí)和尋優(yōu)算法的設(shè)計，尤其是提高其速度和效率。 模糊控制與其它智能控制方法的結(jié)合： 盡管模糊控制在概念和理論上仍然存在著不少爭議，但進(jìn)入 90 年代以來，由于國際上許多著名學(xué)者的參與，以及大量工程應(yīng)用上取得的成功，尤其是對無法用經(jīng) 典與現(xiàn)代控制理論建立精確數(shù)學(xué)模型的復(fù)雜系統(tǒng)特別顯得成績非凡，因而導(dǎo)致了更為廣泛深入的研究，事實上模糊控制已作為智能控制的一個重要分支確定了下來。二者的結(jié)合還能夠擁有過程控制復(fù)雜的知識，并能夠在更為復(fù)雜的情況下對這些知識加以有效利用。自適應(yīng)神經(jīng)網(wǎng)絡(luò)模糊控制，利用神經(jīng)網(wǎng)絡(luò)的學(xué)習(xí)功能作為模型辨識或直接用作控制器 。模糊系統(tǒng)與遺傳算法相結(jié)合的控制器設(shè)計方法則提供了更為新穎的思路。 模糊控制研究方向展望 模糊控制仍然是一個充滿爭議的領(lǐng)域。 模糊系統(tǒng)理論還有一些重要的理論課題沒有解決。以及如何保證模糊系統(tǒng)的穩(wěn)定性?？刂破鞯聂敯粜苑治觯到y(tǒng)的可控性和可觀測性判定方法等。 模糊控制器參數(shù)最優(yōu)調(diào)整理論的確定，以及修正推理規(guī)則的學(xué)習(xí)方式和算法等。 模糊預(yù)測系統(tǒng)的設(shè)計方法和提高計算速度的方法。 模糊控制算法改進(jìn)的研究 :由于模糊邏輯的范疇很廣，包含大量的概念和原則 。這方面的嘗試有待深入。 The third chapter fuzzy control research situation and prospects of application the fuzzy control application research status: Fuzzy control has better control effect of the key is to have a perfect control rules. But due to the fuzzy rules are summarized of or object to process fuzzy information, the high order, nonlinear, large delay, timevarying parameters of the random disturbance and severe plex control 18 process, the understanding of the people tend to be poor or difficult to summarize the experiences of plete, this makes the simple fuzzy control, in some cases, very rough, difficult to adapt to different running status, has affected the control effect. Conventional fuzzy control of the two main problem is: to improve the steady state control precision and improve the level of intelligence and ability to adapt. In practical application, often is the thought of fuzzy control and fuzzy reasoning, and other relatively mature control theory or method, pla