US3795008A - Method for the discrete sampling of co-related values of two or more variables - Google Patents
Method for the discrete sampling of co-related values of two or more variables Download PDFInfo
- Publication number
- US3795008A US3795008A US00243131A US3795008DA US3795008A US 3795008 A US3795008 A US 3795008A US 00243131 A US00243131 A US 00243131A US 3795008D A US3795008D A US 3795008DA US 3795008 A US3795008 A US 3795008A
- Authority
- US
- United States
- Prior art keywords
- sampling
- variables
- amplitude
- variable
- values
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01P—MEASURING LINEAR OR ANGULAR SPEED, ACCELERATION, DECELERATION, OR SHOCK; INDICATING PRESENCE, ABSENCE, OR DIRECTION, OF MOVEMENT
- G01P3/00—Measuring linear or angular speed; Measuring differences of linear or angular speeds
- G01P3/64—Devices characterised by the determination of the time taken to traverse a fixed distance
- G01P3/80—Devices characterised by the determination of the time taken to traverse a fixed distance using auto-correlation or cross-correlation detection means
- G01P3/803—Devices characterised by the determination of the time taken to traverse a fixed distance using auto-correlation or cross-correlation detection means in devices of the type to be classified in G01P3/66
Definitions
- ABSTRACT Automatically controlled, discrete sampling of corelated values of two or more variables is triggered in response to at least two of the variables each time any of them reaches or passes predetermined values.
- the predetermined values are preferably equidistant and may be defined by one or more preceding samples.
- An apparatus used to practice the method comprises means for triggering sampling of the variables when any of two or more of them is subjected to a predetermined variation in its magnitude, and the apparatus further includes means for supplying the sampled values to indicating, recording, or appropriate controlling means.
- the present invention relates generally to automati- Another object of the present invention is to provide an improved method of sampling in which additional cally controlled sampling of variable processes and,
- the advantage of effecting a discrete sampling of data, when compared with data monitoring: methods which are continuous in operation, is that the sampling technique achieves a significant reduction in the vol ume of data concerned, and this in turn' permits a corresponding reduction in the data handling capacity of the recording or control means employed.
- This advantage is, however, subject to a related disadvantage, i.e., going from a continuous monitoring of information to discrete sampling involves a simultaneous change from an exact to an approximate method of data monitoring since sampling techniques do not yield any information about the observed function during the interval between two consecutive samplings.
- sampling frequency While the approximation can be reduced by increasing the sampling frequency, a comparatively small increase in sampling frequency causes a significant uncertainty to persist; and if the sampling frequency is increased to a very high value, the resultant method becomes equivalent to a continuous monitoring operation which, as explained above, increases the cost of the equipment employed due to the higherdata handling capacity which it must exhibit, i.e., higher sampling frequencies necessarily require use of more sophisticated and expensive electronic equipment and, in addition, require a considerably larger volume of incoming data to be properly handled.
- the main object of the present invention is to provide a method for automatically achieving discrete sampling of (:o-related values of two or more variables which yield an increase in accuracy without considerably increasing the sampling frequency.
- samplings of a variable parameter are triggered only when such a sampling is necessary.
- the method of the present invention is characterized by a sampling technique in which two or more variables are continually monitored, with the variables being sampled each time any of two or more of said variables reaches or surpasses predetermined values which are preferably equidistant from one another.
- the two variables are monitored simultaneously, and each variable is capable of triggering sampling of the other variable when the actual variable reaches or surpasses a predetermined value.
- each variable is capable of triggering a sampling of the instantaneous value of the said other variable.
- FIG. 1 is a graphical representation showing a typical prior art sampling technique
- FIG. 2 is a graphical representation of an improved sampling technique in accordance with the present invention.
- FIG. 3 is a block diagram of a sampling arrangement for use in practicing the method of the present invention.
- FIG. 4 is a graphical representation of a modified sampling technique in accordance with the present invention.
- FIG. 5 is a block diagram similar to FIG. 3, showing another form of apparatus capable of practicing the present invention.
- FIG. 6 is a schematic diagram of an apparatus which embodies the method of the present invention.
- FIG. 1 which is exemplary of the prior art, illustrates graphically an arbitrary function y f(x).
- the dependent variable y is sampled whenever the independent variable x reaches certain predetermined discrete values, here designated x x x x x
- the values x, etc. are spaced equidistantly from one another.
- the independent variable x is time, and such a variable will accordingly be assumed for purposes of the subsequent description.
- the broken line in FIG. 1 illustrates, in a very rough and approximate fashion, the interrelationship between the variables x and y depicted by the full line curve.
- the dependent variable may be subjected to very large fluctuations in value which are never detected.
- the picture of the real function represented by the sampling curve is, accordingly, distorted and involves a considerable loss in relevant information. It must be admitted that, for purposes of illustrating this inherent limitation in the prior art sampling method, the deviations between sampling points have been exaggerated. In practical applications, when it has become apparent that the methodis insufficiently exact, the accuracy can be increased by a decrease in the spacing of the xintervals. On the other hand, it is equally obvious that this way of attacking the problem is essentially one involving a difference in degree rather than a change in principle, and efforts to improve accuracy by increasingthe sampling frequency are accompanied by the significant disadvantages which have already been described.
- FIG. 2 illustrates the improved sampling method of the present invention. It depicts the same curve, y f(x) as is shown in FIG. 1; and a comparison of the broken lines in FIGS. land 2 illustrates the substantially increased accuracy which is achieved in FIG. 2.
- This increase in accuracy is effected, in accordance with the present invention, by triggering the sampling not only in response to predetermined variations of the xvariable, but by also triggering sampling whenever the y-variable reaches or passes certain predetermined threshold values y y etc.
- the values y y etc. are also equidistant.
- the increase in sampling frequency is relatively modest and, above all, illustrates that the increase in the number of sampling points selected occurs only when a corresponding need for additional sampling exists.
- the latter characteristic may be alternatively defined in such a way that, if the observed function during the abscissa interval x,x which is within the ordinate interval y -y should not reach or pass the value y (or y,), then no additional samplings are triggered. If, on the other hand, the deviation of the function in the ordinate direction appearing within the primary sampling interval x -x are even greater in amplitude than has been depicted in FIG.
- FIG. 2 illustrates in diagrammatic form an apparatus embodying the new method.
- the curve in FIG. 2 represents the function y f(x); and, for purposes of the subsequent description, it will be assumed that x denotes time and that y denotes a voltage.
- the circuit illus ied one unit, say one volt.
- Signal transmitter B is a time signal transmitter which emits a pulse when a predetermined time unit, for instance one second, has elapsed.
- the information from transmitters A and B are coupled to two counters C and D, which are of the digital or binary types.
- the counter C has the capacity to count forwards or backwards depending on the information about the sign of the derivative receivedfrom.
- Counter D counts every pulse which is received from signaltransmitter B.
- the outputs of the two counters C and D are coupled to an output device E, which may comprise a paper punch, a tape recorder, a plotter, or the like.
- the out put device E receives information about which variable triggered the sampling, and also receives the actual value of the sampled variable, i.e., how many pulses, at
- the counter C will record ten pulses from transmitter A.
- the counter D will record ten pulses from transmitter B when ten seconds have elapsed between x and x
- the counters are arranged to be reset to their zero positions after having recorded ten pulses.
- the x and yaxes have been subdivided into units so that there are ten units (i.e., seconds) between x and x etc., and there are ten units (i.e.,volts) between y, and y etc.
- counter D when the registering operation starts, counter D will first reach the value zero at point m. This causes sampling of the actual value of counter C to be triggered. The actual value is 2. Both counters will then go on to count pulses. Counter C reaches, at point n, the value 10 0, and sampling of the actual value of counter D is accordingly triggered.
- the actual value is 3.
- the cgLter Q has a value of 5. No sampling gered. However, there is recorded a variation of sign in the derivative of the function to correspond with the fact that the voltage will reduce beyond the point y and counter C accordingly receives an order to count in a backwards direction. At the point 0, the value of counter C has been reduced from 6 to 0. Sampling of the actual valued the counter D is accordingly triggered; and the actual value is 8.
- the counter D reaches the value 10 0 at the point p. Sampling of the actual value of counter C isaccordingly triggered, and the value-5 is recorded. At the point 8, the counter C reaches the value IO 0 and sampling of the actual value of counter D is triggered.
- counter D has the value 5 and counter C has a value 4. No sampling is triggered. However, there is again indicated a variation of sign of the derivative of the function, which means that counter C starts to count in a forwards direction once more. At the point r the actual value of counter C has increased to 0, and the value of counter D (i.e., 8) is recorded. At the point s counter D has a value 10 O and sampling of the corresponding value (2) of counter C is triggered. At the point t counter D has the value 10 0, and the value 7 of counter C is recorded.
- FIG. 4 depicts, graphically, a further embodiment of the invention wherein each of the values of the controlling variables triggering a sampling are determined by the next preceding sampling.
- the sampling triggering values are equidistant and, as was the case in FIG. 2, these equidistant xand yvalues have been repro' symbolized on the same scale. This implies that, after a certain sampling has taken place, the function may, without causing a further sampling, at maximum vary within a square the sides of which are parallel to the xand yaxes, with one of the four corners of the square being located at a point corresponding to the latest sampling.
- This principle may be defined alternatively as a successive displacement of origin in several dimensions, and in such a way that any square representing the permitted deviation will be located in the one of the four quadrants the position of which corresponds to the assumption that the continued development of the function on the basis of an extrapolation would probably, as far as its general direction is concerned, be oriented toward the diagonally opposite corner of the square.
- both counters C and D will have a value of zero after each sampling operation.
- the apparatus depicted in FIG. 3 can be modified to add two resetting-to-zero channels between the output device E and the counters C and D respectively, as designated in broken line in FIG. 3.
- the increased accuracy of the method depicted in FIG. 2, over that shown in FIG. 1, is achieved by an increase in the number of samplings from four (in FIG. 1) to eight (in FIG. 2).
- the FIG. 4 method when compared with FIG. 2, shows that essentially the same fidelity can be achieved with a smaller number of samplings, i.e., six samplings instead of eight.
- the reduction in sampling frequency due to successive displacement of origin, makes it feasible to attain savings or simplifications in the stage of the process or in the corresponding portion of the apparatus, respectively, where recordation and/or control is to take place.
- This advantage is achieved by an arrangement, however, wherein the sampled values are somewhat more difficult to evaluate. While the number of samplings in the FIG. 4 method is still higher than in FIG.
- the invention is generally applicable in connection with the sampling of composite functions for realizing one or more of the results, i.e., indication, recordation, and control. It should also be obvious that the practical design of the apparatus employed will vary within very wide limits in response to the actual conditions. For these reasons, the embodiments of the invention to be described hereinafter in reference to FIG. 5 and 6 are intended to be merely illustrative.
- FIG. 5 illustrates, in block diagrammatic form, a system adapted to transfer data from the speed and distance recorder of a motor vehicle to an apparatus for automatic data processing.
- Reference numerals A and B designate two signal transmitters each associated with one of two sampling controlling variables, in the present case distance and time.
- the transmitters may be programmed in such a way that the sampling pulses are triggered at equidistant points, e.g., when the vehicle has traveled two yards or a certain time, say 30 milliseconds, has elapsed.
- the pulses a and b are each supplied to a binary counter C and D in which the signals are converted from analog to digital form.
- Block G represents a unit the object of which is to start the tape when a corresponding order, 0, has been received from block E.
- a sampling pulse is triggered each time any of the number of pulses a or b, respectively, supplied to either of the binary counters C, D, has reached a predetermined value.
- FIG. 6 shows, in greater detail, an arrangement of the general type referred to above in reference to FIG. 5, adapted to sample data from a motor vehicle.
- the circuit of FIG. 6 depicts an apparatus the function of which is to record, on an eight-channel punched tape, the relationship between the distance covered by a vehicle and the corresponding time interval. Internationally adopted symbols have been used.
- a binary l is represented by a low level potential, and a binary by a high level potential.
- Element K which constitutes one of the inputs being monitored, comprises a contact mounted at the universal driving shaft of the vehicle, e.g., an automobile.
- the normally open contact K is closed once for each completed turn of the shaft.
- Contact K is coupled to a circut MVl which serves as a pulse former operative to filter away spurious signals generated in response to contact bouncing. Accordingly, when the universal drive shaft rotates, MVl will deliver pulses which pass through a gate G, (providing that the punch mechanism, used to record data, is unoccupied); and the pulses passing through gate 0, are coupled to counter BRl where they are counted.
- the aforementioned time function is derived from a clock pulse generator CP the output of which passes through a gate G (again assuming that the aforementioned punch is unoccupied) to a further counter BR2, where they are also counted. 7
- the punch when operating, delivers a signal TU to gates G3 and G7 to indicate that the punch is occupied; and the punch also delivers a signal T? to gates G14 and G15 indicating that the punch pins are ready.
- Units MV3 and MV4 are coupled, along with unit MV2, to the output of gate G4 and are shifted simultaneously with MV2.
- the output of MV3 is coupled via gate G16 to each of gates G1 and G5 and operates, as mentioned above, to block the system via Gland G5. The time period of this block is somewhat longer than that which is effected by MV4.
- MV4 operates, via gate G17, to control the output conditions of gates G8-G13 (high level). Since the outputs of BRl are at high level, the outputs of BR2 will now determine the output levels from gates G8-G13' i.e., the data supplied to the punch.
- counter BR2 receives pulses from a clock pulse generator CP, e.g., an astable'flipflop circuit, and its operation'is analogous to that of BRl.
- the information punched on the tape due to the sequence of operations described above, will accordingly correspond to the position of counter BR2, i.e., to the number of clock pulses counted in BR2 when counterBRl reaches the value 32.
- each variable being monitored is adapted tocontrol sampling of the other variable.
- the particular counter which has reached 32, and which has accordingly triggered the sampling is identified in channel 7 of the tape by means of a signal which appears on line K7.
- the signal on line K7 is also coupled, along with the aforementioned TP signal, to the inputs of gates G14 and G15 to reset BV1 and BV2 preparatory to a further sampling operation.
- the operation will be as follows: the output of gate G2 assumes its low value, and BV1 assumes its 1 position. Gates G1 and G5 are closed, and counters BRl and BR2 stop counting. No signal passes through gate G3 since the punch is, at this time, delivering an occupied" signal TU. After the TU signal disappears, however, the signal from BV1 passes through gate G3 to gate G4 and thence to MV2, MV3 and MV4 which starts a punching operation, as described above. It should be noted that, when interference does occur, the system canbe blocked for about 15 microseconds corresponding to the time necessary for carrying out a punching opera tion.
- counters BRl and BR2 it is, of course, also possible for counters BRl and BR2 to reach their values 32 in exact simultaneity with positions, counters BRl and BR2 are stopped, and a pulse passes simultaneously through gates G3 and G7. Units MV2, MV3 and MV4 are triggered. information is punched in channels Kl-KS (BRl and BR2 are both now at their 0 level). On the other hand, punching takes place in K7 since BV2 controls gate G13. The punch supplies an occupied signal TU; and as soon as punching in K7 has been indicated, BV2 takes up its 0 position. BVl is not affected, and the system accordingly is maintained in its blocked condition.
- the MV units of FIG. 6 can be of the type commercially available under the designation Teledyne 342C]; the BR units can comprise Teledyne 312 C] devices; the various gates G1-G3 and GS-GIS can comprise Teledyne 303C] devices; the gates G4, G16 and G17 can comprise conventional diode gates; and the various differentiators can have the circuit configuration shown in the legend forming a portion of FIG. 6.
- contact K can be replaced by any suitable analog-digital converter if other phenomena are to be recorded.
- counter BR can be directed to count backwards when the derivative is negative, and to count in the forward direction when said derivative is positive, etc.
- a method for automatically controlling the discrete sampling of co-related amplitude values of an unknown function of at least two variables comprising the steps of establishing a number of predetermined values associated with each of said variables respectively and defining amplitude limits within which each of said at least two variables may vary individually without triggering sampling of the other of said variables, causing either of said variables upon reaching or surpassing the amplitude of each of its associated predetermined values to trigger sampling of the instantaneous amplitude value of the other of said variables, and supplying said sampled amplitude values to data receiving means.
- one of said two variables comprises a regularly spaced time function, and the other of said two variables comprises a parameter other than time.
- An apparatus for effecting automatically controlled discrete sampling of co-related amplitude values of an unknown function of at least two variables comprising first means responsive to said unknown function for monitoring the amplitude of the first variable in said function, means for defining first amplitude limits within which the amplitude of said first variable may vary without triggering a sampling operation, second means responsive to said unknown function for moni' toring the amplitude of the second variable in said function, means for defining second amplitude limits within which the amplitude of said second variable may vary without triggering a sampling operation, means responsive to each occurrence of said first variable reaching or exceeding one of said first limits in amplitude for sampling the then-existing instantaneous amplitude of said second variable, and means responsive to each occurrence of said second variable reaching or exceeding one of said second limits in amplitude for sampling the then-existing instantaneous amplitude of said first variable.
- first and second means comprise first and second electrical signal generators.
- the apparatus of claim 8 including first and second counters coupled to said first and second signal generators respectively.
- the apparatus of claim 7 including recording means coupled to said first means and to said second means for recording said sampled instantaneous amplitudes of said first and second variables.
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Feedback Control In General (AREA)
Abstract
Automatically controlled, discrete sampling of co-related values of two or more variables is triggered in response to at least two of the variables each time any of them reaches or passes predetermined values. The predetermined values are preferably equidistant and may be defined by one or more preceding samples. At the time of each sampling, the particular variable which has triggered the sampling is determined. An apparatus used to practice the method comprises means for triggering sampling of the variables when any of two or more of them is subjected to a predetermined variation in its magnitude, and the apparatus further includes means for supplying the sampled values to indicating, recording, or appropriate controlling means.
Description
United States Patent 1 1 Kolsrud et al.'-
[ 1 Feb. 26, 1974 1 1 METHOD FOR THE DISCRETE SAMPLING OF CO-RELATED VALUES OF TWO OR MORE VARIABLES [22] Filed: Apr. 12, 1972 [21] Appl. No.: 243,131
Related US. Application Data [63] Continuation-in-part of Ser. No. 869,768, Oct. 27,
1969, abandoned.
[52] US. Cl. 346/33 R, 328/132, 328/150, 328/151, 235/l50.3, 235/150.53, 235/197,
[51] Int. Cl G01d 9/00, H03k 17/00, H03k 5/00 [58] Field of Search... 328/132, 114, 150, 151, 154, 328/157; 340/408, 411, 413, 415; 346/65;
[56] i I References Cited UNITED STATES PATENTS 2,921,740. 1/1960 Dobbins et a1 235/150.53
12/1970 Walsh 235115053 12/1968 Takano 346/33 R Primary Examiner-Richard B. Wilkinson Assistant ExaminerVit N. Miska Attorney, Agent, or Firm-Fred C. Philpitt [57] ABSTRACT Automatically controlled, discrete sampling of corelated values of two or more variables is triggered in response to at least two of the variables each time any of them reaches or passes predetermined values. The predetermined values are preferably equidistant and may be defined by one or more preceding samples. At the time of each sampling, the particular variable which has triggered thesampling is determined. An apparatus used to practice the method comprises means for triggering sampling of the variables when any of two or more of them is subjected to a predetermined variation in its magnitude, and the apparatus further includes means for supplying the sampled values to indicating, recording, or appropriate controlling means.
10 Claims, 6 Drawing Figures PAiENTED 3,795,008
SHE! 2 0F 3 FIG. 3.
. Signal. Slgnu' Transmitter Transmitter A B i Counter j Counter F C o -i I (Reset) cgfm L i (Output Device Signal Si nal Tran mitter B Transmitter Binary Binary Counter Counter Decoding Unit TaDe Start Unit K4 K5 K6 K7 K8 F Tape Recorder FIG. 5.
CROSS REFERENCE TO RELATED APPLICATION This application is a continuation-in-part of our prior copending application Ser. No. 869,768, filed Oct. 27,
1969, now abandoned, for Improvements in a Method for Automatically Controlled Discrete Sampling of Co- R'elated Values of Two or More Variables and an Apparatus for Carrying Out Said Method.
BACKGROUND OF THE INVENTION The present invention relates generally to automati- Another object of the present invention is to provide an improved method of sampling in which additional cally controlled sampling of variable processes and,
more particularly, is concerned with the discrete sampling of co-related values of two or more variables.
In many technical and scientific applications it is well known, when sampling variable processes, to use an apparatus which does'not operate continuously but which samples the function under observation at discrete points only. The sampled values may then be applied to indicating means or, which is more often the case, can be recorded. It is also common to feed the corresponding information back to the input side of the system under observation to be used there for control purposes. However, before recording or control takes place, thesampled variables are generally transformed into some other condition, e.g., from analog to digital form, so that the recording or control can be carried out by means of punched tapes, magnetic tapes, or ac cording to any other well known or suitable technique for automatic data processing.
The advantage of effecting a discrete sampling of data, when compared with data monitoring: methods which are continuous in operation, is that the sampling technique achieves a significant reduction in the vol ume of data concerned, and this in turn' permits a corresponding reduction in the data handling capacity of the recording or control means employed. This advantage is, however, subject to a related disadvantage, i.e., going from a continuous monitoring of information to discrete sampling involves a simultaneous change from an exact to an approximate method of data monitoring since sampling techniques do not yield any information about the observed function during the interval between two consecutive samplings. While the approximation can be reduced by increasing the sampling frequency, a comparatively small increase in sampling frequency causes a significant uncertainty to persist; and if the sampling frequency is increased to a very high value, the resultant method becomes equivalent to a continuous monitoring operation which, as explained above, increases the cost of the equipment employed due to the higherdata handling capacity which it must exhibit, i.e., higher sampling frequencies necessarily require use of more sophisticated and expensive electronic equipment and, in addition, require a considerably larger volume of incoming data to be properly handled.
The main object of the present invention is to provide a method for automatically achieving discrete sampling of (:o-related values of two or more variables which yield an increase in accuracy without considerably increasing the sampling frequency.
samplings of a variable parameter are triggered only when such a sampling is necessary.
SUMMARY OF THE INVENTION The method of the present invention is characterized by a sampling technique in which two or more variables are continually monitored, with the variables being sampled each time any of two or more of said variables reaches or surpasses predetermined values which are preferably equidistant from one another. In practicing the method, in conjunction with two variables, the two variables are monitored simultaneously, and each variable is capable of triggering sampling of the other variable when the actual variable reaches or surpasses a predetermined value. Moreover, each variable is capable of triggering a sampling of the instantaneous value of the said other variable.
By employing this technique, a major increase in the accuracy of the sampled data is achieved with only a minimal increase in the number of sampling points involved.
BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a graphical representation showing a typical prior art sampling technique;
FIG. 2 is a graphical representation of an improved sampling technique in accordance with the present invention;
FIG. 3 is a block diagram of a sampling arrangement for use in practicing the method of the present invention;
FIG. 4 is a graphical representation of a modified sampling technique in accordance with the present invention;
FIG. 5 is a block diagram similar to FIG. 3, showing another form of apparatus capable of practicing the present invention; and
FIG. 6 is a schematic diagram of an apparatus which embodies the method of the present invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1, which is exemplary of the prior art, illustrates graphically an arbitrary function y f(x). The dependent variable y is sampled whenever the independent variable x reaches certain predetermined discrete values, here designated x x x x In accordance with the example shown in FIG. 1, the values x, etc. are spaced equidistantly from one another. In most cases, the independent variable x is time, and such a variable will accordingly be assumed for purposes of the subsequent description.
The broken line in FIG. 1 illustrates, in a very rough and approximate fashion, the interrelationship between the variables x and y depicted by the full line curve. In other words, it is apparent that, during the time interval between any two consecutive samplings which are triggered by variations of the independent variable 2: only, the dependent variable may be subjected to very large fluctuations in value which are never detected. The picture of the real function represented by the sampling curve is, accordingly, distorted and involves a considerable loss in relevant information. It must be admitted that, for purposes of illustrating this inherent limitation in the prior art sampling method, the deviations between sampling points have been exaggerated. In practical applications, when it has become apparent that the methodis insufficiently exact, the accuracy can be increased by a decrease in the spacing of the xintervals. On the other hand, it is equally obvious that this way of attacking the problem is essentially one involving a difference in degree rather than a change in principle, and efforts to improve accuracy by increasingthe sampling frequency are accompanied by the significant disadvantages which have already been described.
FIG. 2 illustrates the improved sampling method of the present invention. It depicts the same curve, y f(x) as is shown in FIG. 1; and a comparison of the broken lines in FIGS. land 2 illustrates the substantially increased accuracy which is achieved in FIG. 2. This increase in accuracy is effected, in accordance with the present invention, by triggering the sampling not only in response to predetermined variations of the xvariable, but by also triggering sampling whenever the y-variable reaches or passes certain predetermined threshold values y y etc. For illustrative purposes, the values y y etc. are also equidistant. FIG. 2 illustrates that the increase in sampling frequency is relatively modest and, above all, illustrates that the increase in the number of sampling points selected occurs only when a corresponding need for additional sampling exists. The latter characteristic may be alternatively defined in such a way that, if the observed function during the abscissa interval x,x which is within the ordinate interval y -y should not reach or pass the value y (or y,), then no additional samplings are triggered. If, on the other hand, the deviation of the function in the ordinate direction appearing within the primary sampling interval x -x are even greater in amplitude than has been depicted in FIG. 2, e.g., so that the function actually reaches or surpasses the value y then two or more further samplings would have been triggeredso that this variation as well would have been detected and represented in the data information collected for indication purposes, recordation and/or control. If one chooses instead to characterize the reproduced curve starting from the applicable error margins, it can be laid down that the maximum area of errors equals the area of each of the squares shown in the Figure.
An important advantage of the method depicted in FIG. 2 is that the relation between the accuracy of the reproduction and the capacity of the electronic equipment employed is optimized, and this in turn means that a greater reliability is attained without any substantial augmentation insofar as the volume of registered data is concerned. Moreover, the increased complexity of the equipment required, due to the fact that two or more variables trigger samplings instead of only one variable, as was the case in FIG. 1, is of entirely negligible importance.
The method depicted in FIG. 2 may be more readily appreciated by considering that Figure along with FIG. 3, which illustrates in diagrammatic form an apparatus embodying the new method.
As already described, the curve in FIG. 2 represents the function y f(x); and, for purposes of the subsequent description, it will be assumed that x denotes time and that y denotes a voltage. The circuit illus ied one unit, say one volt. The transmitter A is, in addition, capable of transmitting information as to whether the voltage is increasing or decreasing, i.e., varying negatively or positively. In other words, it can record the sign of the derivative of the function y =f(x). Signal transmitter B is a time signal transmitter which emits a pulse when a predetermined time unit, for instance one second, has elapsed.
The information from transmitters A and B are coupled to two counters C and D, which are of the digital or binary types. The counter C has the capacity to count forwards or backwards depending on the information about the sign of the derivative receivedfrom.
signal transmitter A. Accordingly, if the voltage increases, the counter index will move one step for every received pulse from A forwardly or backwardly depending on whether the variation of the voltage is positive or negative. Counter D counts every pulse which is received from signaltransmitter B. The outputs of the two counters C and D are coupled to an output device E, which may comprise a paper punch, a tape recorder, a plotter, or the like.
Every time the counter C or D reaches a predetermined value, such as O, 10, 20, 30 etc., sampling of the actual value of the other variable is triggered. The out put device E receives information about which variable triggered the sampling, and also receives the actual value of the sampled variable, i.e., how many pulses, at
that time, were recorded in the corresponding counter.
More particularly, and referring to FIG. 2, let us assume that the distance between x, and x 2 corresponds to ten seconds, and that the distance between y, and y corresponds to ten volts. When the voltage varies from y, to 31,, the counter C will record ten pulses from transmitter A. Correspondingly, the counter D will record ten pulses from transmitter B when ten seconds have elapsed between x and x In order to simplify the timing operation, the counters are arranged to be reset to their zero positions after having recorded ten pulses. In FIG. 2, the x and yaxes have been subdivided into units so that there are ten units (i.e., seconds) between x and x etc., and there are ten units (i.e.,volts) between y, and y etc. For the particular function shown in FIG. 2, when the registering operation starts, counter D will first reach the value zero at point m. This causes sampling of the actual value of counter C to be triggered. The actual value is 2. Both counters will then go on to count pulses. Counter C reaches, at point n, the value 10 0, and sampling of the actual value of counter D is accordingly triggered. The actual valueis 3.
At the point ymizz the counter Chas a value of 6, and
the cgLter Q has a value of 5. No sampling gered. However, there is recorded a variation of sign in the derivative of the function to correspond with the fact that the voltage will reduce beyond the point y and counter C accordingly receives an order to count in a backwards direction. At the point 0, the value of counter C has been reduced from 6 to 0. Sampling of the actual valued the counter D is accordingly triggered; and the actual value is 8.
The counter D reaches the value 10 0 at the point p. Sampling of the actual value of counter C isaccordingly triggered, and the value-5 is recorded. At the point 8, the counter C reaches the value IO 0 and sampling of the actual value of counter D is triggered.
The actual value, I, is recorded. At the point y,,,,,, the
counter D has the value 5 and counter C has a value 4. No sampling is triggered. However, there is again indicated a variation of sign of the derivative of the function, which means that counter C starts to count in a forwards direction once more. At the point r the actual value of counter C has increased to 0, and the value of counter D (i.e., 8) is recorded. At the point s counter D has a value 10 O and sampling of the corresponding value (2) of counter C is triggered. At the point t counter D has the value 10 0, and the value 7 of counter C is recorded.
The various different points (m, n, 0, p, q, r, s, t) at which sampling has been triggered are interconnected by-broken line in FIG. 2 to illustrate, graphically, the approximation in the graphical representation of the actual function y =f(x) which is recorded by this technique. A comparison of the broken line in FIG. 2 with the broken line in FIG. 1 illustrates that the approximation received is much closer to the real shape of the curve than has been effected by the prior art.
FIG. 4 depicts, graphically, a further embodiment of the invention wherein each of the values of the controlling variables triggering a sampling are determined by the next preceding sampling. To simplify the figure, it has again been assumed that the sampling triggering values are equidistant and, as was the case in FIG. 2, these equidistant xand yvalues have been repro' duced on the same scale. This implies that, after a certain sampling has taken place, the function may, without causing a further sampling, at maximum vary within a square the sides of which are parallel to the xand yaxes, with one of the four corners of the square being located at a point corresponding to the latest sampling. This principle may be defined alternatively as a successive displacement of origin in several dimensions, and in such a way that any square representing the permitted deviation will be located in the one of the four quadrants the position of which corresponds to the assumption that the continued development of the function on the basis of an extrapolation would probably, as far as its general direction is concerned, be oriented toward the diagonally opposite corner of the square.
The foregoing principle will be more readily understood by reference to FIG. 4. More particularly, when a sampling has taken place at point a, the area for permitted deviations up to the next sampling becomes located within the square a b c d a. In practice, the extrapolation is carried out by determination of the derivative of the function. When the next higher ythreshold value has been attained, a further sampling is triggered at point e whereupon the next area of permitted deviations is automatically determined in the way above described, and is defined by the square e f g h c. Obviously, the next sampling will take place at point i since the line e h is there passed. Due to the fact that the derivative of the function has now changed its sign, the succeeding square will be located below the sampling point i, and its corners will be i j k l i. The subsequent operations will become apparent from the zero of the counter which did not trigger the sampling operation immediately after its actual value has been recorded. Accordingly, if an apparatus of the general type shown in FIG. 3 is employed to practice the method of FIG. 4, both counters C and D will have a value of zero after each sampling operation. For such purpose, the apparatus depicted in FIG. 3 can be modified to add two resetting-to-zero channels between the output device E and the counters C and D respectively, as designated in broken line in FIG. 3.
The increased accuracy of the method depicted in FIG. 2, over that shown in FIG. 1, is achieved by an increase in the number of samplings from four (in FIG. 1) to eight (in FIG. 2). The FIG. 4 method, when compared with FIG. 2, shows that essentially the same fidelity can be achieved with a smaller number of samplings, i.e., six samplings instead of eight. The reduction in sampling frequency, due to successive displacement of origin, makes it feasible to attain savings or simplifications in the stage of the process or in the corresponding portion of the apparatus, respectively, where recordation and/or control is to take place. This advantage is achieved by an arrangement, however, wherein the sampled values are somewhat more difficult to evaluate. While the number of samplings in the FIG. 4 method is still higher than in FIG. 1 i.e., six samplings instead of four, it nevertheless achieves a reproductional fidelity which could not be reached by use of the conventional principle illustrated in FIG. 1, based on the use of only one sampling triggering variable, unless the sampling frequency were to be increased to a value considerably higher than that used in FIG. 2 and to a much higher value than that employed in FIG. 4.
As has been discussed above, the invention is generally applicable in connection with the sampling of composite functions for realizing one or more of the results, i.e., indication, recordation, and control. It should also be obvious that the practical design of the apparatus employed will vary within very wide limits in response to the actual conditions. For these reasons, the embodiments of the invention to be described hereinafter in reference to FIG. 5 and 6 are intended to be merely illustrative.
FIG. 5 illustrates, in block diagrammatic form, a system adapted to transfer data from the speed and distance recorder of a motor vehicle to an apparatus for automatic data processing. Reference numerals A and B designate two signal transmitters each associated with one of two sampling controlling variables, in the present case distance and time. The transmitters may be programmed in such a way that the sampling pulses are triggered at equidistant points, e.g., when the vehicle has traveled two yards or a certain time, say 30 milliseconds, has elapsed. The pulses a and b are each supplied to a binary counter C and D in which the signals are converted from analog to digital form. The information appearing on channels c-h and i-n is passed to a decoding unit E which in turn, via data channels Kl-K8, controls a device F which, by way of example, may constitute a puncher for punching the information onto a paper tape. Block G represents a unit the object of which is to start the tape when a corresponding order, 0, has been received from block E. In accordance with the embodiment here selected it is assumed that a sampling pulse is triggered each time any of the number of pulses a or b, respectively, supplied to either of the binary counters C, D, has reached a predetermined value.
When they invention is carried out in practice, it is possible in many respects to modify the mode of operation discussed above. It must be emphasized in particular that it is not necessary to derive the sampling pulses directly from each sampling controlling variable since, in many cases, it could be more advantageous instead to base the pulses on a derivative, or on one or more other functions of the variables.
FIG. 6 shows, in greater detail, an arrangement of the general type referred to above in reference to FIG. 5, adapted to sample data from a motor vehicle. The circuit of FIG. 6 depicts an apparatus the function of which is to record, on an eight-channel punched tape, the relationship between the distance covered by a vehicle and the corresponding time interval. Internationally adopted symbols have been used. A binary l is represented by a low level potential, and a binary by a high level potential.
Element K, which constitutes one of the inputs being monitored, comprises a contact mounted at the universal driving shaft of the vehicle, e.g., an automobile. The normally open contact K is closed once for each completed turn of the shaft. Contact K is coupled to a circut MVl which serves as a pulse former operative to filter away spurious signals generated in response to contact bouncing. Accordingly, when the universal drive shaft rotates, MVl will deliver pulses which pass through a gate G, (providing that the punch mechanism, used to record data, is unoccupied); and the pulses passing through gate 0, are coupled to counter BRl where they are counted.
The aforementioned time function is derived from a clock pulse generator CP the output of which passes through a gate G (again assuming that the aforementioned punch is unoccupied) to a further counter BR2, where they are also counted. 7
When 32 pulses have been counted in counter BRl, all of the zero outputs of the counter are at their high potential level. Thes outputs are coupled to gate G2 which, at this time, assumes a low level value; and the output of G2 is coupled via a differentiator to a bistable circuit BV1, which is brought into its binary 1 condition due to the differentiation. If the punch is still unoccupied, the signal from BV1 passes through gatesGS and G4 to circuit MV2, which orders the punch to start its operations; and the punching cycle is accordingly initiated. Unit BV1 also functions at this time, as will be described, to block gates G1 and G5 so that, from this point on, the counters BRl and BR2 are rendered insensitive to incoming pulses.
The punch, when operating, delivers a signal TU to gates G3 and G7 to indicate that the punch is occupied; and the punch also delivers a signal T? to gates G14 and G15 indicating that the punch pins are ready.
Units MV3 and MV4 are coupled, along with unit MV2, to the output of gate G4 and are shifted simultaneously with MV2. The output of MV3 is coupled via gate G16 to each of gates G1 and G5 and operates, as mentioned above, to block the system via Gland G5. The time period of this block is somewhat longer than that which is effected by MV4. MV4 operates, via gate G17, to control the output conditions of gates G8-G13 (high level). Since the outputs of BRl are at high level, the outputs of BR2 will now determine the output levels from gates G8-G13' i.e., the data supplied to the punch.
As'mentioned earlier, counter BR2 receives pulses from a clock pulse generator CP, e.g., an astable'flipflop circuit, and its operation'is analogous to that of BRl. The information punched on the tape, due to the sequence of operations described above, will accordingly correspond to the position of counter BR2, i.e., to the number of clock pulses counted in BR2 when counterBRl reaches the value 32.
As shown in FIG. 6, the two portions of the system coupled respectively to counters BRl and BR2 are symmetrical; and therefore, by an operation analagous to that already described above, counter BR2, if it should reach 32 first, will operate to record the information which then exists in BRl. In this way each variable being monitored is adapted tocontrol sampling of the other variable. The particular counter which has reached 32, and which has accordingly triggered the sampling, is identified in channel 7 of the tape by means of a signal which appears on line K7. The signal on line K7 is also coupled, along with the aforementioned TP signal, to the inputs of gates G14 and G15 to reset BV1 and BV2 preparatory to a further sampling operation. If no punching occurs in channel 7 (K7), the signal from MV4 applied via gate G15 to BV1 operates, after a predetermined period of time, to switch BV1 to its zero position whereupon gates G1 and G5 are again opened. Counters BR] and BR2 thereupon continue counting pulses until one of them has reached the value 32, at which time another punching cycle is'initiated.
that BR2 has reached the 32 value, and that punching of the position, eg 31, of BR2 has commenced. The system is now locked against incoming pulses during a period of approximately 350 microseconds, during which period the information is gated out to the buffer register of the punch. The punch is totally occupied for about 15 microseconds during each punching cycle, corresponding to the time lapse necessary to have the tape punched and fed on. When the 350 microseconds have elapsed, BRl and BR2 resume their counting operation, and BRl can now immediately assume a value 32 which, normally, would initiate a punching operation. However, this cannot be permitted now since the punch is still occupied with carrying out the punching orders received from BR2.
In the foregoing interference situation, the operation will be as follows: the output of gate G2 assumes its low value, and BV1 assumes its 1 position. Gates G1 and G5 are closed, and counters BRl and BR2 stop counting. No signal passes through gate G3 since the punch is, at this time, delivering an occupied" signal TU. After the TU signal disappears, however, the signal from BV1 passes through gate G3 to gate G4 and thence to MV2, MV3 and MV4 which starts a punching operation, as described above. It should be noted that, when interference does occur, the system canbe blocked for about 15 microseconds corresponding to the time necessary for carrying out a punching opera tion.
It is, of course, also possible for counters BRl and BR2 to reach their values 32 in exact simultaneity with positions, counters BRl and BR2 are stopped, and a pulse passes simultaneously through gates G3 and G7. Units MV2, MV3 and MV4 are triggered. information is punched in channels Kl-KS (BRl and BR2 are both now at their 0 level). On the other hand, punching takes place in K7 since BV2 controls gate G13. The punch supplies an occupied signal TU; and as soon as punching in K7 has been indicated, BV2 takes up its 0 position. BVl is not affected, and the system accordingly is maintained in its blocked condition. When TU disappears, a signal passes through G3, and another series of 0 punchings appear in channels Kl-KS (BRl and BR2 remaining in their locked condition). However, at this moment no punching takes place in K7 because BV2 is at O. The remainder of the operation follows the sequence already described above.
The MV units of FIG. 6 can be of the type commercially available under the designation Teledyne 342C]; the BR units can comprise Teledyne 312 C] devices; the various gates G1-G3 and GS-GIS can comprise Teledyne 303C] devices; the gates G4, G16 and G17 can comprise conventional diode gates; and the various differentiators can have the circuit configuration shown in the legend forming a portion of FIG. 6.
It should be observed that contact K can be replaced by any suitable analog-digital converter if other phenomena are to be recorded. Similarly, counter BR] can be directed to count backwards when the derivative is negative, and to count in the forward direction when said derivative is positive, etc. Many other variations will be apparent to those skilled in the art, and it must therefore be emphasized that the foregoing description is intended to be illustrative only and not limitative of the present invention. All such variations and modifications as are in accord with the principles described are meant to fall within the scope of the appended claims.
Having thus described our invention, we claim:
1. A method for automatically controlling the discrete sampling of co-related amplitude values of an unknown function of at least two variables comprising the steps of establishing a number of predetermined values associated with each of said variables respectively and defining amplitude limits within which each of said at least two variables may vary individually without triggering sampling of the other of said variables, causing either of said variables upon reaching or surpassing the amplitude of each of its associated predetermined values to trigger sampling of the instantaneous amplitude value of the other of said variables, and supplying said sampled amplitude values to data receiving means.
2. The method of claim 1 wherein said predetermined values are equidistantly spaced.
3. The method of claim 1 wherein the magnitudes of said predetermined values are determined by at least one preceding sampling.
4. The method of claim 1 wherein one of said two variables comprises a regularly spaced time function, and the other of said two variables comprises a parameter other than time.
5. The method of claim 1 wherein said variables are derived from two different signal transmitters coupled respectively to two binary counters.
6. The method of claim 1 wherein said data receiving means comprises a data recorder.
7. An apparatus for effecting automatically controlled discrete sampling of co-related amplitude values of an unknown function of at least two variables, comprising first means responsive to said unknown function for monitoring the amplitude of the first variable in said function, means for defining first amplitude limits within which the amplitude of said first variable may vary without triggering a sampling operation, second means responsive to said unknown function for moni' toring the amplitude of the second variable in said function, means for defining second amplitude limits within which the amplitude of said second variable may vary without triggering a sampling operation, means responsive to each occurrence of said first variable reaching or exceeding one of said first limits in amplitude for sampling the then-existing instantaneous amplitude of said second variable, and means responsive to each occurrence of said second variable reaching or exceeding one of said second limits in amplitude for sampling the then-existing instantaneous amplitude of said first variable.
8. The apparatus of claim 7 wherein said first and second means comprise first and second electrical signal generators.
9. The apparatus of claim 8 including first and second counters coupled to said first and second signal generators respectively.
10. The apparatus of claim 7 including recording means coupled to said first means and to said second means for recording said sampled instantaneous amplitudes of said first and second variables.
Claims (10)
1. A method for automatically controlling the discrete sampling of co-related amplitude values of an unknown function of at least two variables comprising the steps of establishing a number of predetermined values Associated with each of said variables respectively and defining amplitude limits within which each of said at least two variables may vary individually without triggering sampling of the other of said variables, causing either of said variables upon reaching or surpassing the amplitude of each of its associated predetermined values to trigger sampling of the instantaneous amplitude value of the other of said variables, and supplying said sampled amplitude values to data receiving means.
2. The method of claim 1 wherein said predetermined values are equidistantly spaced.
3. The method of claim 1 wherein the magnitudes of said predetermined values are determined by at least one preceding sampling.
4. The method of claim 1 wherein one of said two variables comprises a regularly spaced time function, and the other of said two variables comprises a parameter other than time.
5. The method of claim 1 wherein said variables are derived from two different signal transmitters coupled respectively to two binary counters.
6. The method of claim 1 wherein said data receiving means comprises a data recorder.
7. An apparatus for effecting automatically controlled discrete sampling of co-related amplitude values of an unknown function of at least two variables, comprising first means responsive to said unknown function for monitoring the amplitude of the first variable in said function, means for defining first amplitude limits within which the amplitude of said first variable may vary without triggering a sampling operation, second means responsive to said unknown function for monitoring the amplitude of the second variable in said function, means for defining second amplitude limits within which the amplitude of said second variable may vary without triggering a sampling operation, means responsive to each occurrence of said first variable reaching or exceeding one of said first limits in amplitude for sampling the then-existing instantaneous amplitude of said second variable, and means responsive to each occurrence of said second variable reaching or exceeding one of said second limits in amplitude for sampling the then-existing instantaneous amplitude of said first variable.
8. The apparatus of claim 7 wherein said first and second means comprise first and second electrical signal generators.
9. The apparatus of claim 8 including first and second counters coupled to said first and second signal generators respectively.
10. The apparatus of claim 7 including recording means coupled to said first means and to said second means for recording said sampled instantaneous amplitudes of said first and second variables.
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US24313172A | 1972-04-12 | 1972-04-12 |
Publications (1)
Publication Number | Publication Date |
---|---|
US3795008A true US3795008A (en) | 1974-02-26 |
Family
ID=22917463
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US00243131A Expired - Lifetime US3795008A (en) | 1972-04-12 | 1972-04-12 | Method for the discrete sampling of co-related values of two or more variables |
Country Status (1)
Country | Link |
---|---|
US (1) | US3795008A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3936740A (en) * | 1974-02-13 | 1976-02-03 | Coulter Electronics, Inc. | Method and apparatus for automatically sampling pulses a predetermined average number of times for storage and subsequent reproduction |
US4001768A (en) * | 1973-05-07 | 1977-01-04 | Geophysical Systems Corporation | Data acquisition, transport and storage system |
US4006348A (en) * | 1973-10-01 | 1977-02-01 | Tokyo Denryoku Kabushiki Kaisha | Apparatus for monitoring two electric quantities by combining three consecutive samples of each quantity |
US4075697A (en) * | 1973-10-01 | 1978-02-21 | Tokyo Denryoku Kabushiki Kaisha | Apparatus for monitoring two electric quantities by combining three consecutive samples of each quantity |
US4213134A (en) * | 1979-02-26 | 1980-07-15 | The University Of Akron | Circuit and method for the recorder display of high frequency periodic signals |
US4772847A (en) * | 1985-04-17 | 1988-09-20 | Hitachi, Ltd. | Stroboscopic type potential measurement device |
US20120209568A1 (en) * | 2011-02-14 | 2012-08-16 | International Business Machines Corporation | Multiple modeling paradigm for predictive analytics |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US2921740A (en) * | 1949-12-19 | 1960-01-19 | Northrop Corp | Binary incremental slope computer |
US3419880A (en) * | 1965-12-17 | 1968-12-31 | Jeol Ltd | X-ray diffraction recording system |
US3551655A (en) * | 1968-12-23 | 1970-12-29 | Honeywell Inc | Control apparatus for approximating a signal waveform |
-
1972
- 1972-04-12 US US00243131A patent/US3795008A/en not_active Expired - Lifetime
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US2921740A (en) * | 1949-12-19 | 1960-01-19 | Northrop Corp | Binary incremental slope computer |
US3419880A (en) * | 1965-12-17 | 1968-12-31 | Jeol Ltd | X-ray diffraction recording system |
US3551655A (en) * | 1968-12-23 | 1970-12-29 | Honeywell Inc | Control apparatus for approximating a signal waveform |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4001768A (en) * | 1973-05-07 | 1977-01-04 | Geophysical Systems Corporation | Data acquisition, transport and storage system |
US4006348A (en) * | 1973-10-01 | 1977-02-01 | Tokyo Denryoku Kabushiki Kaisha | Apparatus for monitoring two electric quantities by combining three consecutive samples of each quantity |
US4075697A (en) * | 1973-10-01 | 1978-02-21 | Tokyo Denryoku Kabushiki Kaisha | Apparatus for monitoring two electric quantities by combining three consecutive samples of each quantity |
US3936740A (en) * | 1974-02-13 | 1976-02-03 | Coulter Electronics, Inc. | Method and apparatus for automatically sampling pulses a predetermined average number of times for storage and subsequent reproduction |
US4213134A (en) * | 1979-02-26 | 1980-07-15 | The University Of Akron | Circuit and method for the recorder display of high frequency periodic signals |
US4772847A (en) * | 1985-04-17 | 1988-09-20 | Hitachi, Ltd. | Stroboscopic type potential measurement device |
US20120209568A1 (en) * | 2011-02-14 | 2012-08-16 | International Business Machines Corporation | Multiple modeling paradigm for predictive analytics |
US20130086431A1 (en) * | 2011-02-14 | 2013-04-04 | International Business Machines Corporation | Multiple modeling paradigm for predictive analytics |
GB2499535A (en) * | 2011-02-14 | 2013-08-21 | Ibm | Multiple modeling paradigm for predictive analytics |
CN103354924A (en) * | 2011-02-14 | 2013-10-16 | 国际商业机器公司 | Multiple modeling paradigm for predictive analytics |
GB2499535B (en) * | 2011-02-14 | 2014-12-10 | Ibm | Multiple modeling paradigm for predictive analytics |
CN103354924B (en) * | 2011-02-14 | 2016-05-11 | 国际商业机器公司 | For monitoring the method and system of performance indications |
US10558545B2 (en) * | 2011-02-14 | 2020-02-11 | International Business Machines Corporation | Multiple modeling paradigm for predictive analytics |
US10558544B2 (en) * | 2011-02-14 | 2020-02-11 | International Business Machines Corporation | Multiple modeling paradigm for predictive analytics |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US3333247A (en) | Digital recording of seismic data | |
US2700149A (en) | Polarity selector | |
US4023085A (en) | Numerical control system having a digitized phase loop | |
US3795008A (en) | Method for the discrete sampling of co-related values of two or more variables | |
IL46206A (en) | Method and apparatus for automatically determining a point on a data carrier | |
US3267435A (en) | Multiple shift registers | |
US3209268A (en) | Phase modulation read out circuit | |
US3512146A (en) | Magnetic tape recording methods | |
US3471652A (en) | System for recording operational failure history | |
US3096506A (en) | Graphic character recognition | |
US3805175A (en) | Retrospective pulse modulation decoding method and apparatus | |
US2860323A (en) | Means for synchronizing a pair of data handling devices | |
US3159840A (en) | Pattern sensitivity compensation in high pulse density recording | |
US3412215A (en) | Digital-to-audio readout system | |
US2760063A (en) | Magnetic pulse recording | |
US3188648A (en) | Method of tape recording pulses occurring in very near coincidence | |
US3575658A (en) | Method for measuring incremental speed of tape transport | |
US3404260A (en) | Analyzing varying signals | |
GB1364598A (en) | Data record merging apparatus | |
US2976517A (en) | Data readout system | |
US2968026A (en) | Stores, especially for digital computers | |
US3087101A (en) | Multiplex recording and playback system | |
US3475740A (en) | Magnetic recording and playback apparatus for analytical signals | |
GB906789A (en) | Improved magnetic tape system | |
US3230542A (en) | Data recording system |