RU2140660C1 - Method detecting low fluxes of ionizing radiations - Google Patents

Abstract

FIELD: detection of low fluxes of ionizing radiation from mobile and immobile, open and closed beta-, gamma- and neutron sources with use of ionization counters ( proportional or Geiger counters ) or scintillation detectors. SUBSTANCE: threshold q₀ is determined before start of test of objects in correspondence with probability of false operations specified by operator by tables for normal distribution, then average rate b of background counting for time t_b and average rate n of counting of additive mixture of signal and background obtained for time t_n of object presence in field of vision of detector. Later parameter η is found by processing of measurement results by formula

which are compared with threshold q₀ to make decision on result of detection. If η > q₀ then decision on detection is taken, otherwise decision on failure to detect is taken. EFFECT: minimal detected signal decreased to level close to theoretical limit, achievement of greatest probability of detection with preset probability of false alarm. 1 tbl

Description

 The inventive method relates to the field of operational radiation monitoring using ionization counters (proportional or Geiger counters) or scintillation detectors and is designed to detect weak fluxes of ionizing radiation (AI) from moving and stationary, open and closed beta, gamma and neutron sources. The inventive method can be used in stationary or portable (portable) devices for radiation monitoring on railways and highways, at border checkpoints, at customs inspection and storage points, in warehouses of fresh nuclear fuel from nuclear power plants, at entry and exit control posts of nuclear enterprises and other industries, for state, departmental and private control of territories, dumps, warehouses (scrap metal and others), equipment, cars, banking upyur, securities and all types of food.

 Known methods for detecting weak streams of ionizing radiation used in portable devices for operational radiation monitoring [1-3]. Such devices are, as a rule, portable radiometers-dosimeters into which an electronic threshold device is inserted, and the corresponding sound and (or) light signaling of exceeding the threshold.

 In the mentioned devices (systems) various counting detectors of ionizing radiation are used (either scintillation counters of gamma radiation, or neutron counters, etc.). This is one of the key differences between ionizing radiation registration devices and detection systems used in radio engineering (in communication systems) and in radar [4, 5]. The essence of this difference comes from the nature of ionizing radiation, which is always a quantized stream (gamma rays, neutrons, etc.) that obeys Poisson statistics, while in radio engineering they almost always deal with normal white noise.

So, in one of the latest developments - the search dosimeter type ДРС-РМ1401 (Polimaster joint venture, Minsk) [1], built on the basis of a scintillation counter with a photodiode, the simplest detection method is used, based on a strict control algorithm and reduced to the following sequence of operations:
before the start of the control of the checked objects, the number of pulses of background radiation N _b is measured for a time t _b = 36 s;
the device switches to operational control mode, i.e. measured the number of radiation pulses from the monitored object N _o during the monitoring interval t _n (in [1] t _n = 2 s); (physically, the value of N _{о is} proportional to the average count rate of the additive mixture of the signal and background n obtained during the time t _n when the control object is in the field of view of the detector - N _o = n • t _n ).

the threshold q = b • t _n + m σ is calculated,
where σ = (N _b ) ^1/2 is the standard deviation of the value of N _b ;
m is a number equal to the number of standard deviations;
b = N _b / t _b is the average speed of light of the background.

the obtained N _{о is} compared with the threshold q; if N _о > q, then a decision is made on detection (the light or sound alarm signal is turned on); otherwise, a decision is made to not detect the source of the AI.

 The main disadvantage of the described detection method [1] is the rigidity of the set thresholds for detecting AI sources, which does not allow to obtain low detection thresholds.

 The detection method used in the DRBP-03 type dosimeter-radiometer (VNIIFTRI, Moscow) [2] is based on the use of an even more stringent algorithm than that used in [1]; in it even the number of standard deviations (rms) m does not change and is always equal to m = 4. This leads to roughening of the detection thresholds and to an increase in the probability of skipping a weak AI source, all other things being equal. A method was constructed similarly for detecting weak sources of AI in the Inspector MKS-06N radiometer-dosimeter (manufactured by Green Star LLP (Moscow) [3].

 Obviously, in order to reduce the level of the minimum detectable signal when developing devices for detecting weak sources, AI seeks to increase the dimensions of the detectors and thereby increase the registration efficiency. This is the main difference between the device [6] and [1-3]. However, the detection method used by the authors of the development is not known for [6]. At the same time, indirectly, according to the figures given in the information leaflet [6], it can be stated with a high degree of certainty that far from the optimal detection method is used here, which leads to an unjustified overestimation of the minimum detectable signal by 1.5–2 times.

 A highly sensitive helicopter search system [7], built on large NaI-Tl scintillation crystals (120 mm in diameter), is also known. However, as in [6], in [7] there is no description of the detection method used.

 Known methods for detecting weak flows of AI [8], which use some basic provisions and conclusions from the theory of detection [9 - 12]. However, in [8] the mentioned provisions are not used consistently and accurately enough, and therefore also lead to overestimation of detection thresholds.

 It is appropriate to describe in more detail the most important provisions and conclusions used for practical (engineering) tasks and conclusions from the fundamental theory of detection, long and thoroughly worked out in relation to radio communications and radar [4, 5, 9 - 12].

 In accordance with the theory of detection, in conditions of complete a priori uncertainty (i.e., when there is no information about the detected object either about the presence or absence of an AI source on it, or about the magnitude of the signal from the AI source) in the detector, only the criterion can be used Neumann-Pearson, which is the most powerful criterion. The latter means that the application of the Neumann-Pearson criterion makes it possible to obtain the highest theoretically possible detection probability for a given probability of false alarm.

где все обозначения совпадают с введенными выше.The second most important point of the theory of detection, which should be used in the engineering implementation of the theory, is the need to calculate the likelihood ratio. Without going into details of the theory, we emphasize that for practical purposes it is more convenient to use the expression equivalent to the likelihood ratio for sufficient statistics [12], which in our particular case of detecting weak AI flows takes the form

where all the notations coincide with those introduced above.

 In physical terms, η is the normalized value of the average signal counting rate from the detected source a = n - b, i.e., signal a is presented in such a way that its dispersion is equal to unity.

The obtained value of η is compared with the threshold q _o , the value of which depends on the probability of false alarm given by the operator P _lt and is determined from the tables of the normal distribution law.

In the detection method described in [8], the Neumann-Pearson criterion is applied, however, the detection threshold is not calculated using sufficient statistics (1), but in the form
q = b • t _n + m σ, (2)
where σ = (N _b ) ^1/2 is the standard deviation of the value of N _b .

Obviously, the described method is almost identical to [1], except that the values of t _b and t _n can be set by the operator to any, depending on the required accuracy and restrictions on the measurement time.

 It is easy to show that the use of (2) leads to one and a half to two-fold overestimation of the minimum detectable signal compared to theory (i.e., according to (1)), since the number of s. k.o. m in (2) is established by intuition (usually m> 4), and not on the basis of strictly specified detection parameters and detection theory, leading to (1) for the problem under consideration.

Closest to the claimed is the detection method used in the systems (installations) "Amber" (production of the SPC "Aspect", Dubna) [13]. (Note that the Aspect research and development center, Dubna, produces about 6 different types of installations of the Yantar family, which differ only in the number and size of the countable gamma and neutron radiation detectors used in them. All types of systems [13] use the same method discovery, which boils down to the following sequence of operations:
before the start of monitoring objects (cars and pedestrians), the number of pulses of background radiation N _b is measured, and the background measurement time t _b can be any depending on the "operational situation" (ie, given restrictions on the monitoring time);
automatically (at the signal of the "presence sensor") the device switches to the operational control mode, i.e. the number of radiation pulses from the controlled object N _о is measured during the monitoring interval t _n equal to the time the object was in the control zone plus additional intervals (the so-called “look back” lasting about 1 s and “looking ahead” are also about 1 s);
the threshold q = b • t _n + m σ is calculated,
where σ = (N _b ) ^1/2 is the standard deviation of the value of N _b ;
m is a number equal to the number of standard deviations (SD) defined by the operator and stored in RAM;
b = N _b / t _b is the average background count rate;
the obtained N _{о is} compared with the threshold q; if N _о > q, then a decision is made on detection (the light or sound alarm signal is turned on); otherwise, a decision is made about non-detection.

The disadvantage of the described prototype method, as well as analogues, is that the number of s.c.o. m is assigned by the operator rigidly either according to the directions (recommendations) from the technical description, or intuitively, and therefore the detection threshold is too high. (Note that in [13] it is recommended to set m in the range from 4 to 8). In this case, of course, a low probability of false alarms P _lt <10 ^{-4 is} achieved, but due to a significant increase in the probability of skipping weak signals (i.e., reducing the probability of detection).

 The present invention solves the problem of reducing the minimum detectable signal to a level close to the theoretical limit.

To solve the above problem, the claimed method for detecting weak flows of AI consistently uses the optimal detection method, which is based on the theory of detection [9 - 12] (the main useful conclusions from which are briefly described above) and which reduces to the following sequence of operations:
before the start of the monitoring of the objects, the threshold q ₀ is determined in accordance with the probability of false positives set by the operator according to the tables for the normal distribution;
measuring the average count rate b caused by background radiation, the background measurement time t _b being any large and determined by the operational environment or entered by the operator;
the device is switched to the operational control mode, i.e. measure the number of radiation pulses from the controlled object N _about during the monitoring interval t _n specified by the operator; calculate the average count rate of the additive mixture of signal and background n = N _o / t _n ;
determine the value of the parameter η by (1);
the obtained value of the parameter η is compared with the threshold q ₀ .

If η> q _o , then decide on the detection (light or sound alarm signal is turned on); otherwise, a decision is made about non-detection, which is recorded in the countdown according to the control procedure.

 Notes.

1. The main essence of the proposed method and its difference from the prototype is that in the proposed method, when calculating the parameter η according to (1), the total dispersion of the sum of the signal and background n is taken into account, including the absolute monitoring time t _n , and therefore no the need to artificially inflate the threshold; the threshold is assigned in strict accordance with the specified probability of false alarms P _lt . As a result, the maximum probability of detection in the given control conditions is achieved.

2. The difference between the threshold q _o used in the proposed detection method and the threshold q used in the analogs and prototype is that q _o represents the quantile of the normal distribution (ie, equal to the number of scd for distribution with a single dispersion), and q is the number of dispersions σ = (N _b ) ^1/2 , i.e. the threshold q in analogues and prototype is expressed in absolute units (in units of the count rate 1 / s).

The table shows the results of comparative calculations of the minimum detectable signal a _min for the proposed method and for the prototype. The values are calculated for identical conditions, i.e. and the claimed method and the prototype are applied to the processing of measurement information for the same equipment and under the same detection conditions. As an example, the case of using a scintillation detector based on CsI-T1 with dimensions ⌀ 63 x 63 mm was considered. Specific figures for the conditions and detection parameters are given in the notes to the table. From the above calculations it follows that the proposed method achieves more than 1.5-2-fold reduction in the level of the minimum detectable signal a _min compared to the prototype. This ensures the achievement of the maximum power of the detection criterion, i.e. maximum probability of detection at a given level of probability of false alarms P _lt . The above calculations are confirmed by experiment. The experiment showed a very good agreement with the expected (calculated) values of a _min .

The proposed method has several additional advantages compared to the prototype and analogues, the main of which are:
the operator enters into the measurement information processing program a specific value of the probability of false alarms P _lt , i.e. the most important parameter that characterizes the quality of the solution to the detection problem (and not the number of scrams associated with the detection parameters by a rather complex and substantially non-linear dependence, as in the analogs and prototype);
as a result of control, the operator receives not only information about detection or non-detection, but also the resulting value of the probability of detection P _obn . The latter is easily found from the tables of normal distribution and corresponds to a quantile equal to q _β = η-q _o .
Briefly about additional advantages, we can summarize that when using the proposed method, the operator "works" not with some not always clear constants, but with specific detection parameters (P _lt and P _obn ) that it needs to achieve.

Table notes
1. The designations used in the table are deciphered in the text.

2. The values of the minimum detectable signal a _min calculated for identical conditions as in the present method, and the prototype:
CsI-Tl single crystal scintillation detector (diameter 63 mm, height 63 mm);
probability of false alarms P _lt :
- for the proposed method - P _lt = 0.05,
- for the prototype P _lt = 3 • 10 ^-5 (corresponds to q _o = 4.0);
probability of detection of P _obn :
- when calculating a _min1 - P _obn1 = 0.95;
- when calculating a _min2 - P _obn2 = 0.50.

 3. The background value corresponds to the experimentally measured value for the scintillation detector indicated in Section 2, equipped with a lead collimator, when setting the energy range of 50–3000 keV.

Literature
1. Search dosimeter microprocessor-based DRS-RM1401. Technical description and instruction manual. - Minsk: SP Polimaster, 1997.

 2. Dosimeter-radiometer DRBP-03. Passport GKPS 14.00.00.000 PS. - M.: VNIIFTRI, 1996.

 3. The radiometer-dosimeter MKS-06N INSPECTOR. Passport. - M .: Green Star LLP, 1996.

 4. Repin V. G., Artakovsky G. P. Statistical synthesis with a priori uncertainty and adaptation of information systems. - M .: Owls. Radio 1977, 432 pp.

 5. Sheremetyev AG, Statistical theory of laser communication. - M.: Communication, 1979.

 6. Stationary radiation threshold alarm "Dozor". Information sheet. - M.: VNIIFTRI, 1997.

 7. Jobst J. E., A history of aerial surveys radiological incidents and accidents: CONF-860932. - 1987, p. 79 - 84.

 8. Kirillov V.M., Suprunov V.I. Detection of moving sources of ionizing radiation. Measuring technique. - 1994. N 8, p. 63 and 64.

 9. Van Tris, G., Theory of Detection, Estimation, and Modulation. T. 1. - M .: Sov. Radio, 1972, 744 p.

 10. Tikhonov V.I. Statistical radio engineering. -M.: Radio and Communications, 1982, 624 p.

 11. Akimov P.S., Bakut P.A., Bogdanovich V.A. et al. Theory of signal detection. -M .: Radio and communications, 1984, 440 p.

 12. Sachs S. Theory of statistical conclusions. -M .: Mir, 1975, 776 p.

 13. Stationary customs system for the detection of fissile and radioactive materials "Yantar-1A". Technical description and instruction manual. DKTSI.425713.004 TO. - (Dubna, Moscow Region: SPC Aspect, 1997.

Claims (1)

A method for detecting weak ionizing radiation from moving and stationary, open and closed beta, gamma - and neutron sources using ionization counters or scintillation detectors, which consists in measuring the average count rate of the background b for time t _b and the average count rate of the additive mixture of signal and background n, obtained in the time t _n finding control object in the field of view of the detector, determining a detection threshold object, and deciding the detection result, wherein the detection threshold obe q _o determine that before the control in accordance with the defined probability of false alarm, as determined by the parameters at the end of control

and compare it with the detection threshold q _o , and the decision to detect an object is made when η> q _o .

Images