Искусственная вентиляция легких (ИВЛ) считается одним из важнейших методов интенсивной терапии, входящим в комплекс мер по поддержанию жизненно важных функций организма в критических состояниях. В связи с созданием интеллектуальных режимов вентиляции легких, повышающих эффективность управления аппаратами ИВЛ, необходимо разработать и применить различные вычислительные схемы обработки данных значений текущих показателей пациента при ИВЛ. В статье рассматривается проблема выявления аномальных выбросов и нивелирования их отрицательного влияния на выделяемые значимые характеристики рассчитываемых показателей, необходимых для принятия оптимальных значений параметров вентиляционного потока, обеспечивающих наиболее эффективное лечение пациента. Для решения поставленной задачи в статье рассматриваются и применяются несколько так называемых робастных методов и основанных на них вычислительных схем выделения аномальных выбросов в значениях показателей состояния пациента и определения их будущих значений.

An optimal decision support system under the decommissioning of nuclear facilities (a brief description of the structure is presented in the Bulletin of the National Research Nuclear University Moscow Engineering Physics Institute (NRNU MEPhI)) is presented. The main problem (choosing the best option for decommissioning nuclear facilities based on a multicriteria approach) is solved in the system. To solve the main problem under conditions of uncertainty, partial indicators are calculated using the Monte Carlo method for each option under the decommissioning of the nuclear facility. The cost and duration of decommissioning, as well as radiation exposure to personnel and the environment, are proposed to be taken as partial indicators. A histogram of the distribution of the complex indicator, which combines partial indicators according to their priority, is plotted using the calculated partial indicators for each decommissioning option. The best option for decommissioning the nuclear facility is chosen on the basis of the priority of the distribution histogram of the complex indicator for the best option in relation to other options under consideration.

ОБОСНОВАНИЕ ОПТИМАЛЬНЫХ ТЕХНИЧЕСКИХ И ОРГАНИЗАЦИОННЫХ РЕШЕНИЙ ПРИ ВЫВОДЕ ИЗ ЭКСПЛУАТАЦИИ ОИАЭ С УЧЕТОМ ОБЕСПЕЧЕНИЯ ЯРБВ. В. Бочкарев 1, 2 , Б

The paper presents the structure of the optimal Decision Support System (DSS) for planning, preparing and implementation of nuclear facilities decommissioning. The proposed structure is intended to implement the following DSS functions: modelling of nuclear facilities and decommissioning activities carried out according to various possible decommissioning options; quantitative assessment of a specific set of partial indicators that characterise the possible decommissioning strategies; identifying the preferences of the decision-maker and redacting them to a formalised form (the quantitative restrictions on the values of partial indicators, the weighting functions of significance for partial indicators); identification of the values of the complex indicator which is a measure for optimality of the considered possible decommissioning strategies. Possible decommissioning strategies are described using a set of the following partial indicators: potential radiation hazards, decommissioning costs, duration of the decommissioning stage. The DSS is designed to be used for decommissioning of nuclear facilities in the Russian Federation. However, it can be used for decommissioning of nuclear facilities in other countries after appropriate modifications.

Scintillation detectors with organic scintillators are widely used for fast neutrons detection in high gamma ray background. The peculiarity of this type of detector is that the pulse shape depends on the type of the detected particle. Traditionally, the Pulse Shape Discrimination (PSD) histogram is used to determine the number of detected neutrons. The PSD parameter is calculated from the shape of the detector pulse and assigned to each pulse. A typical PSD histogram contains two peaks corresponding to neutrons and gamma rays that overlap in the region between the peaks. With this approach, it is impossible to identify each individual signal in the area between the peaks. Therefore, it is not possible to calculate the overall signal identification coefficient. We have proposed a new method for the identification of neutrons and gamma quanta, which includes a combination of three signal separation algorithms: the traditional histogram PSD, the dependence of the area of signals on their amplitude, Tau histogram (tau means the fall constant of the detector pulses). This combination of three algorithms makes it possible to calculate the value of the signal identification coefficient. To test a new method for identifying neutrons and gamma quanta, we used a Pu-Be neutron source, a scintillation detector with a p-terphenyl crystal and a CAEN DT5730 Digitizer (14 bit, 500 MHz). When a scintillation detector registered neutron from a Pu-Be source, the signal identification coefficient was 91.6%. A new method for identifying signals from a scintillation detector is used to register neutrons at the light ion accelerator.

The solution of the problem how to register fast neutrons in the presence of intense gamma radiation is required when solving such fundamental and applied problems as registration of the neutron and gamma background in underground low-background experiments (the low background detectors of the neutrino and dark matter); beam diagnostic at particle accelerators; radiation monitoring at nuclear facilities, nuclear medicine; environmental monitoring. To separate signals from neutrons and gamma quanta, scintillation detectors with organic scintillators are used. The best scintillators are organic crystals of stilbene and p-terphenyl. The efficiency of separating signals from neutrons and gamma quanta can be increased using various methods of digital signal processing of the pulse shapes of the registered signals. A parameter traditionally called the Figure of Merit (FOM) is used to compare these methods. The experimental setup consisted of a Pu-Be neutron source, a scintillation detector with organic crystal p-terphenyl, a Hamamatsu R6094 photomultiplier, a CAEN DT5730 Digitizer (500 MHz, 14bit), which store the shape of each pulse for the following digital processing. A new “method of normalized signals” was developed. Three variants of the new method of normalized signals are described, which give the following FOM values: 1.6, 1.7, and 2.1. The traditional method of signals separation on the same array of experimental data showed the efficiency FOM = 1.6. The new method of signal separation is used to register fast neutrons in the installation dedicated for the development of a compact neutron generator, which is necessary for the calibration of low-background detectors of neutrinos and dark matter particles.

Problems for shares of resources to be distributed in conditions of constraints, including the process of dynamic change of shares are considered. The tasks of forming and dynamically changing the composition of portfolios are considered, in particular, in the presence of group restrictions. A mathematical model for predicting the dynamics of shares for a transition process from one equilibrium state of the system to another equilibrium state is proposed. The case of the presence of a noise chaotic component in such a model is considered. An example of the application of the proposed model of the dynamic process of resource allocation is given.

In this paper, we investigated the efficiency of several known and new methods of digital pulse shape discrimination for neutrons and gamma quanta. Experimental data were obtained on a setup consists of a Pu-Be neutron source, organic p-terphenyl scintillation detector and 14 bits, 500 MHz sampling rate flash-ADC with capability to store and upload to the host computer long waveforms for further analysis. A comparison is made in between the results of using traditional and new methods for calculating the signal separation efficiency of Figure of Merit (FOM). The best known from the literature value of the efficiency of neutron and gamma quanta discrimination for the Pu-Be source is FOM = 1.5. We obtained the separation efficiency FOM = 1.77 in the scintillation detector with the p-terphenyl crystal, by a new method. Note also that for the known liquid scintillator BC-501A FOM≈1. A new method of scintillation detector pulse shape discrimination from neutrons and gamma quanta is used to detect the neutron yield from compact neutron generator that is created on the basis of carbon nanotubes.

In the presented work, we investigated several digital methods of a discrimination signals from fast neutrons and gamma quanta The experimental setup consists of a Pu-Be neutron source, a scintillation detector with an organic para-terphenyl monocrystal, and a digitizer (CAEN DT5730, 500 MS/s). Mixed waveform sequences were stored and then separated by pulse shape. Four methods were used for signals separation. Comparison of the traditional and the new methods of Figure of Merit (FOM) calculation is given. FOM = 1.5 was obtained in our setup for the minimum threshold value. A scintillation detector with a para-terphenyl crystal was used to measure neutron yield in the neutron generator with carbon nanotubes.

The influence of a sampling rate of ADC on the efficiency of the pulse shape discrimination procedure (PSDP) developed for gamma-neutron discrimination was studied. Pu-Be neutron source and two types of digitizers (CAEN DT5730 and CAEN DT5743) were used. Both digitizers together with application software allow to store sequences of waveforms from a scintillation detector. The functional features of the CAEN DT5730 and CAEN DT5743 are described, and experimental characteristics of their operation are compared. Experimental values of an efficiency of neutron/gamma signal discrimination using two ADCs with different sampling frequencies are presented.

Results of the experimental study of electron emission from liquid xenon via electroluminescence of the gas phase are presented. We report on observation of a peculiar kind of delayed electroluminescent signal following initial electroluminescence caused by ionizing particles. We also present the results of a study of spontaneous single electron emission following cosmic muon signals. It was found that the rate of spontaneous single electron signals strongly depends on the time passed since the initial electroluminescence happened. The analysis of experimental data showed that both spontaneous single electron signals and delayed electroluminescent signals are associated with ionization electrons which are trapped by the potential barrier at the interface.

A two-phase emission detector containing 5 kg of liquid Xe is installed at the horizontal experimental channel of the research nuclear reactor IRT MEPhI to measure the liquid Xe response to nuclei recoils with kinetic energies below 1 keV. Preliminary tests have demonstrated that ≥ 15 μs electron lifetime in liquid Xe and ~ 10 photoelectrons single ionization electron signal are achieved. These parameters are sufficient to detect and identify events at the single electron level.

Рассматриваются задачи выделения компонент и прогнозирования динамических процессов. Исследованы схемы прогнозирования хаотических временнх рядов с помощью предварительного выделения регулярной, аномальной и хаотической компонент и дальнейшего применения к регулярной компоненте одного из представленных методов прогнозирования. Для выделения регулярных компонент используются робастные линейные сплайны и сингулярно-спектральный анализ. Приведенные в работе примеры показывают, что применение представленных схем прогнозирования позволяет получать прогнозируемые значения исследуемых динамических процессов с приемлемой точностью. Библ. 9. Фиг. 4. х рядов с помощью предварительного выделения регулярной, аномальной и хаотической компонент и дальнейшего применения к регулярной компоненте одного из представленных методов прогнозирования. Для выделения регулярных компонент используются робастные линейные сплайны и сингулярно-спектральный анализ. Приведенные в работе примеры показывают, что применение представленных схем прогнозирования позволяет получать прогнозируемые значения исследуемых динамических процессов с приемлемой точностью. Библ. 9. Фиг. 4.

The problems of detecting components and predicting dynamical processes are considered. Schemes for predicting chaotic time series that are based on detecting their regular, anomalous, and chaotic components followed by applying one of the described prediction methods to the regular component are presented. Regular components are detected using robust linear splines and singular spectrum analysis. Provided examples show that the presented schemes allow predicting dynamical processes with acceptable accuracy.