mathematics body of frames of knowledge base of expert system (ES) "Energy-saving control by means of dynamic objects" is considered. The frames allow to develop operatively algorithms of energy-saving regulators for apparatuses, installations in which processes of body heating and cooling play an important role. To intellectual energy-saving regulators we relate microprocessor controlling devices that solve a synthesis problem of optimum control realized by one positional strategy in real time. Expenses of energy, specific fuel consumption or combined functionals taking into account speed of response are taken as minimizing functionals.
The task of optimal control (TOC) is usually formulated as a task with a control restriction, fastened ends of change trajectory of phase coordinates vector and fixed time interval. The ES frame name is set by a cortege of 4 symbols < M, F, S, N > where M - object model, F - functional, S - strategy, N - problem peculiarities.
Practically all thermal objects are non-linear due to changing of properties and characteristics of heated bodies at temperature increase. object models are presented in the form of segment-linear ordinary differential equations for operative solving TOC. An object model and a functional ascertain the from of a synthesizing function which determine the dependence of optimal control at the given moment of time on the current object phase coordinates value and remaining time. Results of full analysis of optimal control on the set of functioning states are used for determining the form of a synthesizing function and calculation of its parameters.
The analysis is carried out with the help of maximum principle and a method of synthesizing variables. optimal control realization strategies differ in ways of taking into account changes of functioning states. These changes can occur at the random moments of time and they are caused by change technological regimes, a breakdown in the serviceability of technical means and etc. Editional restrictions, e.g., maximum speed restriction, temperature increase, interval task of model parameters and etc. relate to the peculiarities of TOC to be solved. separate working modulus of ES are connected to Internet network.
R.T.Gumbatov PERSPECTIVES OF APPLICATION OF ARTIFICIAL INTELLIGENCE SYSTEM FOR THE INDUSTRIAL WASTERWATER CLARIFICATION PROCESS CONTROL
As control objects, the industrial wasterwater clarification processes are described by varity, lack of possibility of direct or even indirect measurement of many technological parameters, difficulties & sometimes impossibility of establishing the adequate model, search of controlling influeces on the objects with using of models & methods, assuming integration of mixed differential equations with non-linear boundary conditions etc. In conditions of significant changes of contaminators quantitative & qualitative indices & variety of clarifying facilities working regime more efficient is to realize the control tasks on the basis of description of control objects work using qualitative information. As the latter using of information, generalizing clarifying facilities staff working experience concerning latter’s working regime & outcomes of scientific researches, carried out on these objects is proposed.
Last years creation of intelligence control system by such objects with use of fuzzy sets has the special actuality. Such systems are realized by means of creation the logic-linguistic description of object with use of linguistic variables, characterizing the objects position parameters, indignation & controlling influences. Linguistic variables are determined from appropriate universal sets, which contain not concrete values, & qualitative evaluation, for example, “very big”, “average”, “very small” etc. By means of generalizing of working & knowledge experience, gathered by the specialists - experts, the object control fuzzy algorithm is created. Appropriate programming software, worked out on system programming language Pascal, allows to realize the efficient control of technological complexes on biochemical clarification & neutralization of buffer wasterwater of chemical productions.
It’s perspective to realize such systems as it allows refusing from expensive measurement complex, which exploitation is is connected with large amount of expenses as well.
In procedure of electrical measurements and measuring transformations of the information data, submitted by set (or by vector) of electrical signals, together with the main purpose of measurements (of measuring transformation) it is often necessary to fulfil a set of accompanying tasks: a rejection (suppression) and a selecting (allocating) in correspondence with given tag of one of several signals, also a sorting and a arranging of signals according to information attribute, grouping of set of signals, a arranging of a signals on transmitting channels pursuant to a given attribute, a address identification of one input channel, onto which a signal with a given attribute is sent, address recommutation of signals, weighted summation of signals etc.
It is accepted to name intelligent all this indicated measurements (measuring transformations) with accompanying operations and (or) algorithms of processing.
The reproduction of indicated operations and algorithms of processing by conventional means of computer facilities if necessary of functioning in a real time scale and (or) at parallel registration of result of measurement (transformation) often is not possible because of sequential algorithm of processing of signals by microprocessors.
The indicated defect may be eliminated at use of a commutative simulation principle, which is used in analog computers, where the given function or algorithm of processing are reproduced by the appropriate electrical circuit. The same principle is used and in natural neural networks.
The stated approach is illustrated on an example of accompanying operations of address-rank processing of electrical signals, which are submitted by set of voltages or by vector of variable (signals), at use of logic-algebraic apparatus of predicative algebra of choice .
Collection of address-rank operations was considered in article . Hardware is realised in basis of relator components as modules, which are built in measuring devices (measuring-computer complexes, and measuring-informational systems).
On circuits C1 and C2 (fig) the examples of "intelligent" digital voltmeters - DV (of analog-digital converters - ADCs) are respectively represented. On the circuit C1 is presented a relator rank processor (PR), which fulfils a processing of analog signals pursuant to the formula (7) in . The measurand is being any of signals, which is sent onto informative inputs both DV and PR (this signal for C1 is x,). Other signals in any order are moved onto inputs of comparator PR. An identifying voltage Y is sent onto a input D of demultiplexor (switching) channel.
On the circuit C2 is presented a analog relator address processor, which reproduces function of selection (allocation) of variable of given rank by a switching of channels of multiplexor M of a processor PA. On demultiplexor channel D a function of identification of address of wire (number of a source channel), onto which variable of given rank is sent, is reproduced. The connection of inputs of PA to the signal bus is executed in any order (property of invariancy of result of processing to rearrangement of variables).
Relator realisation of processors PR and PA was described in [3,4]. Other opportunities of expansion of "intelligence" of electric devices are concluded in use of a mathematical apparatus of continuous logic [5,6] and hybrid logic [7,8]. In particular, at use of algorithms of continuous logic the ADC with extremely achievable speed (ADC with peak convolution) was constructed etc.
The work is executed due to a grant "Development of the theoretical bases of both logic synthesis and circuitry of relator converters and processors for expansion of intelligent possibilities of analog measuring devices", which is financed by Department of Education.
Volgin L.I. Complementary algebra and predicative algebra of choice .-Ulyanovsk: USTU, 1996.-68 p. (in Russian)
Volgin L.I. Structural properties of the Pascal column // Relator and continuous-logic systems and models: Work ISTC "Neural, relator and continuous-logic systems and models".- 1998.- Vol.2.- p. 13-17. (in Russian).
Volgin L.I. Relator neural processors and commutation-logic converters of analog signals with coding by the number of channel .- Ulyanovsk: USTU, 1996.-76 p.
Volgin L.I. Relator-based processors for ranging and ordinal processing of analog signals // Journal of New Generation Computer Systems .- 1991.- Vol.4.
Levin V. 1. Multiplevalued logic in tasks of cybernetics. - M.:Radio and communication.- 176 p. (in Russian).
Volgin L.I., Levin V.I. Continuous logic. The theory and applications .-Tallinn:Estonia AS, 1990.- 210 p. (in Russian).
Shimbirev P.N. Hybrid continuous-logic devices. -M.: Energoatomizdat, 1990.-174 p. (in Russian).
Shimbirev P.N. The theory and principles of construction hybrid continuous-logic (soft) computing facilities and from application in systems of processing of the information and management. -M.: MIET. 1997.- 37 p. (in Russian).