| 摘要 |
1,111,731. Seismic prospecting. TEXAS INSTRUMENTS Inc. 12 Aug., 1965 [12 Aug., 1964], No. 34566/65. Heading H4D. [Also in Divisions G1 and G4] A method of processing a set of related signals having desired signal components and noise components with different time and amplitude distributions comprises the steps of (a) auto-correlating each constituent of the set and cross-correlating each constituent with every other constituent of the set to produce a first suite of correlation functions characterizing noise components in the set, (b) deriving a second like suite of correlation functions characterizing desired signal components in the set, (c) deriving a set of correlation functions representing the best estimate of noise-free signals in the set of related signals, (d) modifying each of the related signals in dependence on the two suites and the set of correlation functions, and (e) combining and recording the modified signals to provide signals in which the desired signal is enhanced relative to noise. In a seismic signal processing embodiment, seismic waves generated successively at a plurality of shot points are detected by a group of geophones whose outputs are corrected for static and normal moveout errors and recorded on drums 10 and 11 (Fig. 1), on which traces recorded on channels 13 and 17 correspond to the first shot point, those on channels 14 and 18 correspond to the second shot point, and so on. To provide correlation functions characterizing noise the traces are reproduced from the drums and portions in which noise predominates (e.g. portions at late record time) are selected by time gates 38, 39 and applied to correlators 40 to 49. Incremental time delays between the output signals from the two drums are introduced by relative angular displacements of the drums and define the delay intervals associated with the correlation function coefficients produced by the correlators 40 to 79. The correlator outputs are recorded on drum 36, the noise auto-correlation functions for traces associated with the shot points being recorded on channels 70 to 73 respectively and the noise cross-correlation functions on channels 74 to 79. To provide correlation functions characterizing desired signals, traces reproduced from drums 10 and 11 are applied to summing units 101 to 104 for compositing all signals from a given shot. The composite signals are then recorded on drums 30 and 31. Early record portions of the composite signals, in which desired signals predominate, are selected by gate 106 and applied to correlators 50 to 59. Drums 30 and 31 provide the correlation delay intervals, as above, and the auto- and cross-correlation function coefficients from correlators 50 to 59 are recorded on channels 80 to 89. Correlation functions representing substantially noise-free signals are provided by summing the composite signals from units 101 to 104, and correlating this sum signal with itself and with each composite signal separately in correlators 60 to 63. These correlation functions are recorded on channels 90 to 93. Parameters of a timedomain filter for inverse filtering of the detected seismic signals to enhance the signal/noise ratio are computed in a filter synthesizer 130 from a matrix equation relating the correlation functions characterizing the noise, the desired signals and the substantially noise-free signals. These parameters determine the gains of amplifiers (163) of a magnetic delay line filter (Fig. 2) and hence the proportions in which the detected seismic signals, subjected to multiple delays by means of a delay drum (160), are combined by summing units (164 to 167) to provide filtered signals which are subsequently composited and recorded. Solution of the matrix equation may be performed by digital or analogue computer. Details are given of an analogue system (Fig. 3) in which the correlation functions from drum 36 are combined by multipliers 210, 211, 226, 228, 234, 235, adders 214, 227, 228 and dividers 216, 236 to provide voltages for controlling the gains of inverse-filter amplifiers 232 and 239 and which are analogues of solutions of a simplified exemplary matrix equation. |
