Abstract
The problems of simplified synthesis of scalar energy efficient systems of frequency control induction motor are discussed. The method of special frequency control laws calculated for a given load range is considered as a method to provide energy efficiency.
The calculation procedure of the frequency control law of induction motor is performed by iterative optimization, where the number of iterations can reach several thousand. So some more appropriate simplified procedure for the method of control law forming is needed. This procedure should be suitable for engineering calculations. The hypothesis of "similarity" of control laws is introduced for motors with similar parameters of the equivalent circuit. It is assumed that the control law (quasi-optimal law) is obtained for one motor and applied to other ones with similar parameters of the equivalent circuit with a certain error. Identification of motors with similar parameters is performed via self-organizing artificial neural networks (Kohonen map). The number of clusters is chosen according to the desired dispersion of the motor parameters, which is determined by the deviation from the values of the optimization criterion extremum. The result is presented as hexagonal Kohonen map of 64 clusters where each cluster has its own set of quasi-optimal laws. Thus, it is enough to match the motor with a certain cluster (or classify). As a result the motor gets the suitable quasi-optimal control law automatically.
The paper describes the procedure of the network forming and training. The examples of optimized control laws are given.
Keywords
Optimization, scalar control, induction motor, neural network clustering.
1. Braslavskii I.Ya. Energosberegayuschiy asinhronnyiy elektroprivod [Energy saving induction electric drive]. Braslavskii I.Ya., Ishmatov Z.Sh., Polyakov V. N.. Moscow: ACADEMIA, 2004. 202 p.
2. Braslavskii I.Ya., Kostylev A.V, Tsibanov D.V. Study of optimal startup processes in the real network-FC-IM system. Russian Electrical Engineering, 83(9), 2012, pp. 499-503.
3. Braslavskii I.Ya., Kostylev A.V, Tsibanov D.V. The use of cluster analysis for the synthesis of optimal frequency control law for induction drive. Power Electronics, Electrical Drives, Automation and Motion (SPEEDAM), International Symposium on, 2014, pp. 478-482.
4. Zyuzev A.M.. Metelkov V.P. Termodinamicheskiye modeli dlya proverki asinkhronnogo dvigatelya po nagrevaniyu [Thermodynamic models for temperature control of induction electric drive]. Elektrotekhnika [Electrical engineering]. 2012. no.9, pp. 48–52.
5. Braslavskii I.Ya., Ishmatov Zh.Sh., Kostylev A.V., Plotnikov Yu.V., Polyakov V.N., Erman G.Z., Antonov D.L. Energy efficiency of laws of scalar frequency control of induction electric drives. Russian Electrical Engineering. 2012, vol.83, no.9, pp 508-511.
6. Shreiner R.T., Kostylev A.V., Shilin S.I., Khabarov A.I. Optimization of a variable-frequency induction motor drive with a scalar control system. Russian Electrical Engineering. 2012, vol. 83, no.9, pp 490-493.
7. Braslavskii I.Ya., Kostylev A.V, Stepanyuk D.P. Starting processes in the frequency-regulated asynchronous drive during optimal control. Russian Electrical Engineering. 2007, vol.78, no.11, pp. 607-610.
8. Braslavskii I.Ya., Kostylev A.V, Stepanyuk D.P. Optimization of starting Process of the Frequency Controlled Induction Motor. Proceedings of the 13th International Power Electronics and Motion Control conference, Poznan, Poland. 2008, pp. 97-101.
9. Janos Abonyi, Balázs Feil, Cluster Analysis for Data Mining and System Identification, Springer Science & Business Media, 2007, 324 p.
10. Pokorný P., Dostál P. Cluster analysis and neural network. In Technical Computing, Prague, 2008. Sborník příspěvků 16. ročníku konference. Praha: Humusoft. 2008. pp. 25-34.
11. Braslavskii I.Ya., Kostylev A.V., Stepanyuk D.P., Mezeusheva D.V. Synthesis of asynchronous-drive control systems using neural networks. Russian Electrical Engineering. No.76(9), 2005, pp. 65-68.