FR2562267A1 - METHOD AND APPARATUS FOR PROCESSING SURVEY DATA WITH HIGH RELIABILITY - Google Patents

Abstract

THE INVENTION RELATES TO THE PROCESSING OF DATA FROM REDUNDANT SENSORS. IT RELATES TO A METHOD AND APPARATUS IN WHICH, IN A SURVEY 16, THE SIGNALS OF TWO SENSORS ARE COMPARED AND, WHEN ONE OF THEM IS DETERMINED TO BE BAD, IT IS REPLACED BY THAT OF THE OTHER SENSOR . THE TWO SIGNALS ARE THEN COMBINED SO THAT THEY FORM ONLY ONE SIGNAL TRANSMITTED TO THE SURFACE. IN THIS WAY, THE CAPACITY OF TRANSMISSION OF INFORMATION FROM THE BOTTOM OF THE SURFACE TO THE SURFACE IS NOT INCREASED. APPLICATION TO THE EXECUTION OF OIL DRILLS.

Description

The present invention relates to the use of redundant sensors

  for the measurement of data in boreholes during drilling and more precisely it concerns the processing of data obtained from redundant sensors so that the reliability of the parameters

measured is increased.

  During the drilling and production of

  oil dump in the ground formations, the logging of a survey and the accuracy of the surveys carried out in the surveys are essential for determining the presence and location of

  oil products in formations, along the survey.

  Logging has generally been carried out so far by lowering a measuring instrument attached to a conducting cable in the open borehole and measuring various parameters associated with the characteristics of the formations

  when moving the instrument along the sounding.

  Such logging tools often include several sensors for simultaneously measuring various parameters associated with the earth formations along the borehole, a single sensor being generally used for the measurement.

of each parameter.

  Recently, technology has made it possible

  survey parameters during the drilling operation.

  The possibilities of measurement during drilling have much

  increased the efficiency of a drilling operation by

  the operator to have direct, real-time information on the direction and inclination of the borehole, on the pressures and temperatures at the bottom of the borehole and on the forces applied to the bit, as well as on various parameters directly associated with characteristics

  terrestrial formations in which the survey is conducted.

  The technology of measurements being drilled has a lot

  increased the efficiency of drilling by eliminating

  formerly necessary to stop the drilling and

  removal of the drill string for measurements. In addition,

  measurement technology being drilled has increased

  the safety of the drilling operation because it gives a real-time indication of the conditions in the borehole, and thus enables corrective measures to be taken immediately, as soon as dangerous conditions occur during drilling operations. In measuring operations during drilling, it is also conventional to use a sensor only for the measurement of each of the parameters of the sounding to be controlled. For example, a single sensor is used for measuring each of the parameters such as the natural gamma radiation, the inclination of the survey, the azimuth of the survey, the speed of the bit and torque applied to the drill bit, the forces exerted on the bit, the drilling fluid pressure, the temperature and the resistivity of the drilling fluid as well as many

  other parameters giving information relevant to the operation

  tor. The use of a single sensor for each of these parameters indicates that, when a sensor exhibits a malfunction or a total failure, the operator either receives erroneous data or does not receive any at all on the particular parameter

  concerned. When the parameter is essential for the operation

  drilling, it may need to be interrupted for the replacement of the deteriorated or failed sensor or more often drilling operations continue and the sensor with a failure is not discovered before analysis of the data received, when it is too late to obtain adequate data on the same part of the survey. Failure of a sensor at the bottom of the borehole, particularly in the case of measurements being drilled, may not only be expensive because of the loss of important data, but may also pose increased hazards to life and property. that a very important indication of the presence of emergency conditions is not given while the operator foresees the receipt of this information and relies on it for the conduct of the operation

drilling.

  For this reason, it is extremely useful

  to have redundant sensors of the different parameters

  meters of drilling, in a survey. Thus, in the event of failure of a single sensor or a malfunction of this sensor with production of erroneous data, alternative measurements allow to obtain a greater reliability on the total amount of data.

obtained during logging.

  The use of redundant sensors for measuring parameters in a survey, while providing increased security through data reliability, also requires greater storage and / or recording capabilities in the survey or may require remote transmission of more data to the wellhead, before surface treatment. The use of greater storage capacity, recording or telemetry can be very expensive and difficult to achieve given the conditions under which the equipment must be. Thus, it is desirable for a signal processing device to operate in the survey

  so that information from redundant sensors

  only one parameter is processed and, when both sensors are working properly, the two output signals are combined to form a single data element. In addition, the signal processing capability is advantageous for the detection

  the moment one of the redundant sensors has fallen

  or faulty data, so that the sensor data working properly

  can be used in their entirety at this time, an indica-

  the other sensor indicating failure or fault

Operating.

  Different test patterns can be used

  and evaluation, known in the methodology, for the evaluation

  the fact that a sensor gives reliable data or not, these data can be recorded or transmitted to the surface and can be considered reliable

  by the operator. For example, one can realize algo-

  rithms that use: a) the known physical / statistical properties of the measured variable, b) the known limits, determined empirically,

  the measurement and its speed of variation, and -

  (c) the degree of correlation of the data channels

  redundant corresponding to the same parameter.

  Known techniques for producing redundant instruments have been used but generally in areas such as computer control of rocket systems and manned spaceships. Redundancy is also commonly used in maintenance systems of environmental conditions when human life

  depends on it in extremely hostile conditions. How-

  In these systems, it is not necessary at the same time

  to obtain redundant data and to reduce

  the data processing capacity.

  The processing of several data channels for borehole logging is generally of the type in which two surveys obtained from different soundings in a particular area are compared and correlated so that certain conclusions can be

  based on the relationship between the two surveys.

  This can be done for example for the construction of a statistical model of a complete field by correlation of the separate well records in the same field. Such a system is described in the United States patent

  No. 4,310,887. Similarly, a single set of diag-

  probes from sensors can be retained in the sounding and transmitted to the surface for immediate use. Subsequently, the stored and transmitted data are compared so that the reliability of the transmission system is evaluated. This gives some security against the loss of valuable data and

  unrecoverable in case of failure of the transmission process.

  U.S. Patent No. 4,216,536 describes such a system. However, in all these applications, there is no redundancy of the individual sensors of the same parameter giving increased reliability of the measurements or data processing of the two sensors of the same parameter for the evaluation of the possibility of inaccuracy or of the failure of one of the two sensors, so that the reliability of the unique set of data from the two sensors, which is actually recorded and transmitted

on the surface, be increased.

  The method and apparatus according to the invention provide an increased degree of reliability to the measured data during logging while drilling, and simultaneously reduce

  at least the additional storage facilities

  recording and / or transmission of data

  necessary for redundant data processing.

  The method and apparatus combine reliability data

  equality from redundant single-channel sensors or select the most reliable and give an indication of the failure of one of the sensors or the transmission of unreliable data by one or the other

sensors.

  The invention relates to the use of redundant sensors for increasing the reliability of data obtained in a survey and data processing techniques for analyzing redundant data so that the signal storage and data processing capabilities are minimal. . More specifically, the invention relates to a method and an apparatus for using multi-channel sensor data processing for increasing the reliability of the data derived from a sounding, and in particular data obtained in measurements made in progress. drilling. The method and the apparatus implement the redundancy of the measurement sensors with the analysis of the sensor data, in the survey, so that the erroneous data points are detected and only the appropriate data are retained, so that the reliability information obtained

  in the survey is increased without the ability to memorize

  data storage or telemetry is increased.

  The invention also relates to a method and an apparatus for giving greater reliability to the measurement of the soundings parameters, comprising two detectors intended to each detect a parameter to be measured and to create data point currents representative of these parameters. . Each stream of points

  data is analyzed against statistical standards.

  provided for data points and parameters

  which are associated in a known manner with the characteristics

  the fact that the data points are in the parameters and ranges provided is determined. A flag is raised for data points that are outside the expected parameters and ranges. The data point currents that are in the expected parameters and ranges are then combined to be recorded with the flags of the data points that are outside the ranges and parameters. Also provided is a method and apparatus in which the analysis of each data point stream includes analyzing each data stream from the sensor pair so that good data points are determined and the bad data points are determined. Data points, which are outside the range of statistical standards predicted from previous measurements of the same data stream, are isolated. Data points that are bad, in any individual current, are then replaced by the good data points of the other current or by an earlier data point of the same current. In addition, the data points are compared so that a correlation between them or a scale factor is determined. The data streams are then analyzed according to the scale factor and a flag is raised when the variation of the scale factor giving the correlation of the two signals is outside the expected standard. A new scale factor, relating to the degree of correlation between the data point currents, is also created. The invention also relates to a method and an apparatus for processing several channels of measurement points of signal data obtained from several redundant sensors for the same parameter.

  the survey, including the memorization of a time series

  of corresponding parts of each of the channels, and the analysis of each data point, by its temporal response. The analysis is a determination that the point value is within a prescribed standard range of historical values and that the deviation of the point value from the standard value

  corresponds to the standards of historical difference. The analysis

  also takes the determination that the point value is within a predetermined range, based on physical and stastistic standards for the measurement parameter. Each data point is then compared to the corresponding associated point in time of each of the other channels. A scale factor is formed as a result of the comparison and gives a relationship between each of the data points and the corresponding data point of each of the other channels. Each scale factor is then analyzed in time response so that the variation of the value of the scale factor over time, between each pair of channels, is

  in a previously chosen range, be determined.

  Each data point considered bad because it does not meet the criteria for any analysis

  is then replaced by a valid data point corresponding to

  from another channel after multiplying the value of the valid point by the scale factor that connects the two channels. A flag is then raised as an indication of each bad data point and the channel in which it appeared. Each good or replaced data point of each channel is then combined with the others as a single stream of good data which is recorded along with the flag indications of the bad data points and the associated channel.

  Other features and advantages of the invention

  The following description will become more apparent from the following description.

  with reference to the accompanying drawings, in which: FIG. 1 is a diagrammatic side elevation;

  illustrating a drilling operation of a sounding

  both the description of the measurement of the survey parameters

  according to the invention; Figure 2 is a block diagram of an apparatus according to the invention; FIG. 3 is a block diagram, in

  part in the form of a flow chart illustrating the function

  ment of the apparatus according to the invention; and FIG. 4 is a block diagram illustrating

  the operations implemented according to the invention.

  FIG. 1 shows a drilling installation 11 placed above a borehole 12. A first embodiment of a redundant sensor data processing apparatus 10, with several channels, is carried by a connection 14 comprising a part A drill bit 15 is placed in the borehole 12. A drill bit 22 is mounted at the lower end of the drill string 18 and digs the borehole 12 in the earth formations 24 while drilling mud 26 is pumped to the Wellhead 28. A metal casing 29 is shown in the borehole 12 at its upper portion so that the borehole keeps its integrity close to the surface. The annular space 16 separating the drill string 18 from the bore wall 20 creates a return sludge path, which path is theoretically closed. The sludge is pumped from the head 28 by a pumping system 30 into a mud supply pipe 31 which is coupled to the casing 18. The sludge is thus driven into the central axial passage of the drill string 18 and out level

  bit 22 to carry the debris, including

  the pieces of earth, rock and

  which have been torn off so that they come up from the trephine to the surface. A conduit 32 channels the sludge from the borehole 18 to a tank 34 placed near the wellhead. The drilling mud is for example treated and transmitted by various devices (not shown) such as degassing units and circulation tanks for maintaining the viscosity and

  the consistency chosen for the mud.

  The connector 14, in this example, contains several

  sieurs redundant sensors and other instruments for measuring different parameters in the sounding, a first pair of sensors that can be arranged for the measurement of each parameter. For example, strength

  applied to the trephine, the natural gamma radiation, the inclination

  survey, survey azimuth, resistivity,

  the temperature and the pressure of the mud are

  interesting meters. The connector 14 also contains the

  multi-channel data processing line which is part of the apparatus according to the invention, as well as the apparatus for recording data and

  remote data transmission.

  As shown in Figure 2, the apparatus according to the invention comprises a pair of sensors such as sensors A and B 31 and 32, for measuring the same parameter of the sounding. For example, the sensors 31 and 32 may each include a pair of gamma radiation detectors placed in the connector. The sensor pair 31, 32 transmits output signals which are connected by a respective data channel A and B 33 and 34 to a processing apparatus 35 which evaluates the signal of the signal.

  output of the two sensors. After treatment according to the invention

  The data is transmitted by a single output 36 to a single recorder. The data recorded in the recorder 37 requires practically no more space than when using a single sensor, but they have a much higher degree of reliability and accuracy because the signal comes from a pair of

redundant sensors 31 and 32.

  In the method and apparatus according to the invention, the redundancy of the sensors is used for increasing the overall reliability of an individual sensor module

  only when a function fault is detected.

  a sensor module. For example, when each of the sensors has a satisfactory operating probability equal to P, where P is less than 1.0, the probability that the two sensors will work satisfactorily is then P. In addition, the probability that one at least sensors work satisfactorily is 2P-P2. Thus, we have the relationship 2P-P2 >> P2. When the process and the processing algorithms always allow

  the detection of the presence of a sensor having a function

  incorrectly and allow the identification of the wrongly functioning sensor, the reliability of the system is 2P-P2. However, when the system can only detect a fault but can not identify the particular sensor that is failing, the reliability is P. In addition, in the absence of the disconnection of faulty sensors, the reliability is simply p2. The algorithm

  implemented according to the invention gives a value of reliability

  the sensor module which is between 2P-P2 and P, that is to say it is much greater than the value of reliability obtained when two redundant channels are simply added in the sounding, without deleting the defective channel, by using the techniques according to the invention, that is to say that this value is p2. According to the technique of the invention, the serial data over time are measured by two separate channels which, when the two sensors work satisfactorily, must represent the same reproducible physical measurement. The apparatus of the present invention provides for the processing of these redundant data channels so that (1) the bad data points are detected, (2) one or more defective channels are detected, and (3) the correct data is recorded in only one channel and the channels are combined when both are considered reliable. For this purpose, the apparatus requires the following information: (1) the known statistical and / or physical properties of the measured variable, (2) the known limits of the measurement and its velocities

  variation, these parameters must be determined

  and (3) the degree of correlation of the two channels

  redundant since they both measure the same parameter.

  The first criterion for evaluating and detecting bad data points is determined using a Kalman filter. According to well-known techniques, historically determined parameters are used for flag raising and for rejecting bad data points. A Kalman filter is used for the empirical determination of mean amplitude and variance of data discontinuities and

  variance of the data around the mean at each step.

  Thus, abnormal data points are identified using the following criteria: (a) when the difference between the considered value and the average value, regardless of the size of this value,. significantly exceeds the average step size, the point is considered bad, or (b) when the difference is of the order of the average step size or is larger but when the point is isolated, that is, say when the average values of both sides are the same, the data point is wrong. As information on the following and previous values around an individual data point to be evaluated is needed, the current filtering requires a "prior consultation" by delaying the evaluation of the data for the formation of a drawing around the point. of data to be evaluated. According to the invention, a memory maintains a time series of corresponding portions of data from all channels. However, in one embodiment, advance consultation of a point is appropriate for isolation determination when using Kalman filtering to obtain estimated values of the expected variance for the difference

  between the measured value and a predicted value.

  The second possibility of evaluating the data of a pair of redundant sensors uses the known, statistical and physical range of the parameters whose

  variations may appear for a given set of para-

  meters measured. For example, when detecting natural gamma radiation, it is known that, when the number given by the sensor is of the order of 100, depending on the physical principles of radiation detection, the expected variation is 100 or 10. so that the number expected in an interval of one second must be between 90 and 110. However, if the number varies abruptly between 80 and 120, that is to say presents a variation of 20%, it is manifest that the sensor poses a problem - since the physical behavior of gamma radiation is such that uncertainties in the number counted must not reach 20%. It can thus be deduced that the sensor has failed or that the electronic circuit

has a fault.

  In addition, the measurement of each physical parameter varies in certain individual statistical and physical ranges. This information is used to perform a statistical-physical test for each of the two data channels so that the fact that the value

  is within acceptable ranges or not determined.

  In addition, the data of the different channels of the redundant sensors are further processed according to the method of the invention by correlating the data obtained from the two redundant sensors. Thus, in the case of individual sensors, there is still some difference in the relative efficiency of the sensors so that the output signal representative of a given state of the parameter can

  vary slightly. In the case of the detection of

  In the meantime, the sensors can change over time so that the output of the output signal can be output without failure or without giving really erroneous data. For example, in place of the transmission at any time of a value equal to 30 by one sensor and a value equal to 35 by the other sensor, the sensor giving the value 35 can go to 40 under

  the effect of thermal effects. Thus, during the correlation

  the two sensors are connected to each other by a scale factor which is generally of the order of 1, but allowing to take into account the particular behavior of each sensor, so that the reading

  one of the sensors can be connected to that of the other.

  Referring now to FIG.

  in the form of a synoptic diagram and a

  gramme, the operation of the apparatus according to the invention.

  The output signals of two redundant sensors called channel A and channel B are respectively transmitted by

  Routes 33a and 34a to Kalman filters 41 and 42.

  The output signals of these filters are transmitted to

  data insertion circuits 43 and 44 which enable

  attempt to replace the erroneous data points that are detected. The raw data of the two channels A and B are also directly transmitted to the circuits 43 and 44 for inserting data via paths 33b and 34b

  which pass bypass compared to Kalman filters.

  In the Kalman filters 41 and 42, the data of each channel is evaluated against a reference and using selected algorithms so that the data points are in the prescribed ranges or not and the deviation from any point in relation to the reference range corresponds to the deviation standards,

  given the historical parameters or not,

  mined. So, when a particular data point varies a lot with respect to the data point that's

  according to the preceding and following points, the conclusion

  drawn is that the point is bad. In this case, a "flag 1" is raised so that it identifies the fact of determining a bad data point and

  the point in the channel in which the event occurred.

  The suitable data input circuit 43, 44 is then controlled so that it replaces the wrong point

  the wrong channel by selecting the corresponding point

  of data in the other channel, modified by the

  appropriate correlation scale factor, and by replacing

  the wrong point by the new point.

  When the data of the channels are good or when the data of a bad point, detected by the Kalman filter, have been replaced, the data of the two channels are transmitted by the circuits 45 and

  46 statistical-physical test. These circuits are

  algorithms that contain information about the statistical standards of the data of the type controlled by the sensors in question and furthermore include information relating to the physical range in which the data of the type controlled by the sensors may vary. . Thus, in the case where a data stream in which points have been replaced as previously described appears to vary outside the standards or ranges in which the data must vary, from the point of view of physical and statistical restrictions, the individual data channel which is identified as being bad is marked with a "flag 3", and the data are replaced by those of the channel that

  is considered good. When the statistical tests

  are applied to each of the two data streams, the lack of satisfaction of the criteria imposed by the test poses to the circuit 47, 48 the question of knowing

  whether the other channel has passed the physical test or not.

  So when both channels are in

  outside the limits set by the statistical parameters-

  This event is important for data analysis and a decision for the transmission of a signal representative of a yes or no through lines 51 and 52 is recorded by the data logging circuit 60. If channel A does not satisfy the physical-statistical test and channel B satisfies this test, a state "flag 3a" is noted. In the opposite case and if the B-channel is defective while the data of the channel A are satisfactory, a state "flag 3b" is noted. When both channels A and B do not satisfy the

  statistical-physical test, a "flag 4" state is pre-

  and prepares the alternate recording of the two channels (with the state of the flag) so that the survey analyst

  more information for his diagnosis.

  When both channels satisfy the statistical-physical test in the circuits 45 and 46, they are each transmitted to a correlation circuit 55 in which each data point of the two channels A and B is mutually correlated so that the possible variation of the factor of scale connecting the two individual circuits is determined and, in case of variation, in which range this variation is located so that the probability that the channels still function properly is determined. When both signals satisfy the correlation, they are combined into one and recorded. When the two data signals do not satisfy the correlation, within an acceptable margin, one or other of the channels is erroneous. If the faulty channel can not be identified, a "flag 5" is raised and the data is recorded alternately for both channels. These data do not show a good correlation and if it is known that the channel A is defective, according to other tests, a state "flag 2a" is raised and the channel B is recorded. Conversely, in the absence of correlation and when it is known that the B-channel is defective, a state "flag 2b" is raised and the channel A is registered. The scaling factor between the two acceptable data channels is recurrently generated and is returned to both data point replacement circuits 43 and 44 as well as to the physical-statistical test circuits 45 and 46 in which scale relating to the signals of the two sensors are necessary. Referring now to FIG. 4, which shows the entire apparatus in rectangle 61, the treatment performed in the survey is considered

  as a state I in the case where both channels are considered

  as reliable, and the results are simply added and recorded. In state II, one of the channels is bad and the correct channel is multiplied by the scale factor and recorded, and a flag identifying that state is raised. Similarly, in state III, the other channel is bad, the right channel is simply changed by the scale factor and a flag identifying the wrong channel is raised. In states IV and V, both channels are considered defective, a flag is raised, and the signals of channels A and B are recorded alternately or, when considered that the two channels are defective but that one of 'they are less defective than the other, only the signals of the best channel are

  saved and a flag identifying this fact is raised.

  The redundant processed data is transmitted to the surface by remote transmission or recorded and processed on the surface as indicated by the rectangle

  62, using several conventional techniques.

  These techniques include, for example, smoothing which reduces readings such as those made with

  very high resolution, so with significant variations

  parameters, so that variations are more regular and continuous, based on reference For example, in the case of surveys obtained during drilling, smoothing may be used so that the survey can be analyzed by conventional techniques for analyzing less sensitive readings obtained with drill ropes. A deconvolution is used to reestablish net survey boundaries that have been rounded by various physical parameters that affect the data. In addition, a time conversion is used so that the readings obtained by the redundant sensors are connected to the depth rather than the passage of time. Finally, additional processing or information that an operator may wish may be used for the data that is obtained with the redundant sensors, with greater reliability than can be obtained

with a single sensor.

  It should also be noted that the method and the apparatus according to the invention allow the use of a pair of sensors for each parameter to be measured in the sounding, with preservation of a high statistical reliability of the redundant sensors, without, however, increasing the

  data processing capability of the device that transmits

  puts the data to the surface or increases the storage or recording capacity for

  implementation of the reliability of the redundant sensors.

Claims (19)

  1. Method for measuring parameters of a sounding, inside thereof, with great reliability, characterized in that it comprises: the arrangement of a pair of detectors (31, 32) for the detection of each parameters to be measured, and the creation of data point currents representative thereof, analyzing each current of data points against expected statistical standards of data points and physical parameters known to be associated with the characteristics detectors so that the data points are in ranges corresponding to the parameters is determined, the creation of a flag for the data points that are out of range for the parameters, and the recording of the data currents. data points that are in the acceptable ranges of the parameters   so that they are registered with the corresponding flags.   data points determined to be in outside the parameter ranges.   The method of claim 1, characterized in that analyzing each of the data point streams includes analyzing each data stream from said pair of sensors so that the correct data points are determined and the bad data points, which lie outside the beach   statistical standards based on prior   the same data stream, be isolated, these bad data points being replaced, in each individual stream, either by the correct data point of the other current or by a previous data point the same current.   The method of claim 1, further comprising comparing the data points of the data streams to determine their correlation, and determining a scaling factor. corresponding to their relationship.   4. Method according to claim 3, characterized in that it further comprises -analysis of the data currents according to the scale factor and the creation of a flag when the variation of the scale factor which represents the correlation of the two currents is outside the expected standard, and the creation of a new scale factor corresponding to the degree of correlation data point currents.   Method according to claim 1, characterized in that alternating currents of data points are recorded when both currents are determined as unacceptable.   A device for measuring parameters of a sounding, inside thereof, with a high reliability, characterized in that it comprises: a pair of detectors (31, 32) intended to detect each parameter to be measured and creating data point currents representative thereof, a device (41, 42) for analyzing each current of data points against statistical standards provided for the data points and physical parameters known to be associated the characteristics of the detectors so that the fact that the data points are in parameter ranges is determined, a flag generating device (35) for the data points determined to be outside said parameter ranges, and a device (35) for combining the data point currents that are in said acceptable ranges of parameters so that they are recorded with   flags corresponding to specific data points   as being outside the parameter ranges.   7. Apparatus according to claim 6, characterized in that the device for analyzing each current of data points comprises a device (41, 42) for analyzing each data stream coming from the pair of data points.   sensors so that good data points are   and that bad data points, which lie outside the range of statistical standards predicted from previous measurements of the same data stream, are isolated, this device being intended to replace the bad data points of a any individual current either by the good data point of the other current or by an earlier data point of the same current. 8. Apparatus according to claim 6, characterized in that it comprises a device (35) for comparing the data points of the data streams so that a correlation is determined between them, and determining   a scaling factor that connects the two currents.   9. Apparatus according to claim 6, characterized in that it further comprises a device (35) for analyzing the data streams according to said scale factor, and creating a flag when the variation of the factor d the connecting scale, the two currents are outside the expected norm, and a device generating a new scale factor relating to the degree of correlation data point currents.   Apparatus according to claim 6, characterized   in that data points of the alternating currents are recorded when the two currents are determined as unacceptable.   A method for measuring survey parameters with high reliability, characterized by comprising: providing a pair of redundant sensors (31,32) for forming a pair of data point streams; each point being representative of the value of the same survey parameters as a function of time, the analysis of each stream of data points during a selected period so that the fact that each   point is within a prescribed range of historical values.   and the fact that the deviation from any point is within a prescribed range of historical deviations shall be determined, the creation, following the analysis of the points, of an indication when a data point of one or the other current is determined to be bad because it is outside of said range of historical values and said range of historical deviations, with indication of the current in which said point is located, the replacement of all the wrong data points following the first analysis step, with a corresponding data point in time from the other current, after multiplying the value of the data point by an associated scale factor or with the last previous data point of the same current   bad data points when the corresponding point is   laying of data is bad in both currents, analyzing each current of data points so that the fact that each data point is located   within a previously programmed range of values, deter-   based on physical and statistical standards of the parameter measured by the sensor pair, be determined, the creation of an indication, following the second analysis step, when a data point of one or the other current is determined to be bad because it is outside a range of physical and statistical standards, the replacement of all the bad data, as a result of the second analysis step, by the corresponding data in time from the other current, after multiplying this value of the data point by an associated scale factor, comparing corresponding data points in the two currents so that the degree of correlation between the respective points be determined, creating an associated scale factor for   each data point representative of the degree of corre-   between the values produced by the sensors, the comparison of the scale factor created to a prescribed range of values so that the fact that the scale factor has varied outside this range is determined, the creation, as a result of the second comparison step, an indication that a data point of one or other of the currents is considered bad because its correlation scale factor falls outside the prescribed range; replacing all the bad data points, following the second comparison step, with the corresponding point in time from the other current after multiplying the value of the data point by an associated scale factor, the combination of all the corresponding good or replacement data points in the two data streams into a single current, the recording of the combined combined current and all the indications of the bad data points when u'at least one current is good, and alternating recording of data streams   when all points of both currents are bad.   12. Method according to claim 11, characterized in that the analysis of each data stream with respect to its historical values is carried out by filtering in a Kalman filter.   Apparatus for measuring survey parameters with high reliability, characterized in that it comprises: a pair of redundant sensors (31, 32) for forming a pair of data point streams, each point being representative the value of the same probe parameters as a function of time, a first device (41, 42) for analyzing each stream of data points for a selected period so that each point is within a prescribed range of values and the fact that the deviation from any point is within a prescribed range of historical deviations is determined, a device controlled by the first analyzer and intended to give an indication when a data point of the one or the other current is determined to be bad because it is outside the range of historical values and range of historical deviations, with indication of the current in which the bad a device (43, 44) controlled by the first analysis device and intended to replace all the   bad data points by the corresponding data point.   of the other current after multiplication of the data point value by an associated scale factor or by the last previous data point in the same current of bad data points when the corresponding data point is bad in both current, a second device (45, 46) for analyzing each current of data points so that each data point is in a previously programmed range of values, determined according to physical and statistical standards, for the parameter measured by the pair of sensors is determined, a device (47, 48) controlled by the second analyzing device and intended to give an indication when a data point of one or the other current is determined as being bad because it is outside the range of physical and statistical standards, a device controlled by the second analysis device and intended to replace all the bad points data ts by the corresponding data point in time from the other current, after multiplying the value of the data point by an associated scale factor, a first device (35) for comparing corresponding data points of the data points. two currents and determining the degree of correlation between the respective points, a device (55) for forming a factor   associated scale for each data point, represents   of a degree of correspondence between the values produced by the sensors, a second device (55) for comparing the scale factor formed with a prescribed range of values so that the fact that the scale factor has varied and has come out of this a device controlled by the second comparison device and intended to form an indication when a data point of one or the other current is determined to be bad because its correlation scale factor has been determined. varied outside the prescribed range, a device controlled by the second comparator and intended to replace all the bad data points with the corresponding data point   in time from the other stream, after multi-   plotting the value of the data point by an associated scale factor, a combination device of all corresponding data points which are good or replaced, in both currents, in a single current, a device (60) of recording of the combined combined current and all indications of bad data points when at least one of the currents is good, and a device (60) for alternately recording the two data streams when all the points of the two currents are bad.   14. A method of processing multiple measured point channels of signal data obtained from   several redundant sensors (31, 32) of the same para-   meter of a sounding, characterized in that it comprises: memorizing a time series of corresponding parts of each of the channels,   analysis of each data point, successively   time, so that the point value falls within a prescribed standard range of historical values and the deviation from the point value from the standard value is within the historical deviation standards, and the fact that the value of the point is in a predetermined range, according to physical and statistical norms of the measured parameter, are determined, the comparison of each data point with the corresponding point associated in time in each of the other channels creating a scaling factor as a result of this comparison, this scale factor connecting each data point to the corresponding data point of each of the other channels,   the analysis of each scale factor,   sivemient in time, so that the fact that the variation of the value of the scale factor over time, between each pair of channels, is in a previously chosen range, is determined,   the replacement of each data point considered   deregistered as bad by not satisfying the criteria of any stage of analysis, by a corresponding good point of data from another channel after multiplying the value of the good data point by the scale factor that binds the two channels , the creation of an indication in the form of a   flag for each bad data point, with indica-   the channel in which it is located, the combination of good data points or data points replaced by all channels in a single stream of good data, and the single current record of good data and all indications under Flag shape indicating bad data points and associated channels. The method according to claim 14, characterized in that determining that each data point, successively in time, is in a predetermined standard range of historical values and that the deviation of the values of the points from the standard value is found in standard deviation standards, is achieved by filtering each data channel in a Kalman filter.   Method according to claim 14, characterized   in that it implements two redundant data channels   pendent. 17. Apparatus for processing a plurality of measured data point signal channels corresponding to signals from a plurality of redundant sensors of the same sounding parameters, characterized in that it comprises: a device for conserving a succession in time of parts corresponding means of each of the channels, a device (41, 42) for analyzing each data point, successively in time, so that the fact that the value of the point is in a previously selected standard range of historical values and that the deviation of the point value from the standard value is Y-in the historical deviation norms, and the fact that the point value is within a predetermined range according to the physical and statistical norms of the measured parameter, determined, a device (43, 44) for comparing each data point with the corresponding time associated point of each of the other channels, a device (55) for creating a scale factor according to the comparison, this device connecting each data point to the corresponding point of each of the other channels, a device (45, 46) for analyzing each scale factor, successively in the time, for determining whether the change in value of the scale factor over time between each pair of channels is in a predetermined range, a device (43, 44) for replacing each data point considered bad because that it does not satisfy the criteria of any stage of analysis, by a good data point corresponding to another channel after multiplication of the value of the good point by the scale factor linking the two channels, a flag generating device indicating each bad data point and the channel in which it is located, a device (55) for combining each good data point or replacing all the channels in a single current of good data, and a device (60) for recording the single current of good data and all the flags indicating bad data points and the channels which are associated.   Apparatus according to claim 17, characterized   characterized in that the device for determining that each data point is in a predetermined standard range of historical values and that the deviation of the dot values from the standard value is in the standard deviation standards, is carried out using a Kalman filter (41, 42) for each data channel.   Apparatus according to claim 17, characterized   in that it implements two redundant data channels.