Co pHi ghl © IF.-\C Identification and SI' SlCm Parameler b limalion. l kiji ng. PRC AN ALGORITHM FOR ESTIMATING CROSS AND MACHINE DIRECTION MOISTURE PROFILES FOR PAPER MACHINES K. Natarajan*, G. A. Dumont** and M. S. Davies*** *Plllp alld Paper Crn tre, L'lIh 'ersit)· of British Columbia, Canada ** Plllp (I//(I Pap/'/' Rl'.lmrc/i IlI stitute of CQllQda and Department of Electrical Ellgilleerillg, Pll lp ami Papn Ce llt re, 2385 East Mall, University of British Collllllbw, l 'a l/ co II I'er, BC, V6T 1 W5 , Canada *** Departllll'llt of Elrctriml ElIgilll'lTi ll g, L' lIi vers itJ of British Columbia , Canada Abstract. In this paper we develop an algorithm for estimating, in the presence of noise, cross and machine direction (CD and MD) moisture profiles based on a model described in the literature . The algorithm consists of a least squares parameter identifier for estimating CD profile deviations and a Kalman filter for estimating MD profiles. Simulation results of the algorithm are given. The preliminary results of the application of the algorithm to industrial data are also given. The algorithm developed can be extended to other measurement models. Keywords . Parameter estimation; state estimation; paper industry. 1. INTRODUCTION The moisture content of paper, as it is being made, is measured using a single infrared scan- ning sensor located after the paper machine dry- ers. The sensor is mounted and guided to move in the cross machine direction (CD) perpendicular to the movement of the paper in the machine di- rection (MD). The sensor traverses from front to back and back to front continuously, taking mea- surements at fixed intervals of time. Due to the MD movement of the paper and the CD move- ment of the sensor, the resultant measurement points, on the paper, form a zig-zag pattern as shown in Figure 1. The measured profile is a composite of CD and MD profiles. Lindeborg [1) postulated and experimentally verified the mea- sured profile to be a nonlinear function of CD and M D profiles. This relationship forms the model for the present work. __ Sensor movement direction (CD) • - measurement point • • • • • Paper • • • • • • • . • • • (k + 1 T • • kT- (k - 1)T • • • • • • • • • I p n p e r m o v e 111 e n t d I r e c t The ultimate object of the measurement of moisture on paper is its control. The primary mechanism to control the MD profile is the dryer steam pressure. The CD profile is controlled ei- ther by remoisturizing showers and/or by infrared heating boxes at intervals across the machine's width . Therefore, in order to control the mois- ture, we must first estimate the CD and MD pro- files from the composite measured profile . • 1 In this paper we develop an algorithm to de- duce the CD and MD profiles from the zig-zag measurements. 1091 • 0 n 111111111111 I I11 111I I 11 I I (MD) 1 n-lllln+l N n Figure 1: Zig-zag measurements due to CD and MD movements of sensor and paper