WO2021045638A1 - Method and system for optimizing cash servicing of cash transaction points - Google Patents

Abstract

The invention relates to the automated optimization of efficient cash servicing of cash transaction points by means of the automated prediction of the financial demand on transaction points for a given time period, taking into consideration data about the state of each transaction point. The invention describes a computerized method for optimizing the cash servicing of cash transaction points, which is performed with the aid of a processor and includes: obtaining data from transaction points over a given period of time, said transaction points being self-service devices and/or banking centres and said data including information about the location of the transaction points, the status and operation history of the transaction points and the amount of funds at each transaction point; processing the status and operation history data of the transaction points; predicting, with the aid of a machine learning algorithm, the flow of cash at each transaction point on the basis of the processed status and operation history data of the transaction points; optimizing the cash servicing of the transaction points on the basis of the prediction made for a given time period, which includes generating a matrix of the state of the transaction points that contains cash servicing implementation parameters.

Description

METHOD AND SYSTEM FOR OPTIMIZATION OF COLLECTOR SERVICE OBJECTS OF CASH CIRCULATION

FIELD OF TECHNOLOGY

[0001] The present technical solution relates to the field of computing, in particular to a method and system for automated optimization of collection services for objects of cash circulation.

LEVEL OF TECHNOLOGY

[0002] At present, the process of organizing cash collection services in financial organizations is pushed away from manual processing of data received from various objects of cash circulation (CDO), which include various types of self-service devices (CS) (ATMs, terminals, etc.) , and internal structural divisions (VSP), providing services to clients of financial institutions.

[0003] Proceeding from the numerous fleet of CS and VSP, there is a problem of predicting the load on the objects of the NDO, which leads to an increase in the processing time of information on the need for collection and maintenance of these objects to ensure their efficient operation. Also, the process of encashment of NDO objects is a laborious operation and is associated with a share of risk based on the human factor, while minimization of collection procedures should be clearly defined for priority NDO objects, in order to exclude the likelihood of their prolonged downtime, which leads to financial costs and the inability to service customers in terms of issuing funds.

[0004] At present, the process of managing cash liquidity of NDO objects is organized using the AS "OptiCash / OptiNet", however, this system does not provide an adequate level of automation, taking into account changes in the process of managing cash liquidity and necessitates the use of additional tools (including .n. "small automation") to increase employee productivity and improve the efficiency of the cash liquidity management process.

[0005] In the changing conditions of economic, financial and political factors surrounding the participants in the process of managing cash liquidity, existing schemes of interaction of the OptiCash system with other automated systems and their modules do not allow for a prompt response to emerging force majeure circumstances, incidents or other critical changes in process parameters.

[0006] Automated cash flow control systems are also known in the art. From US patent N ° 10,169,947 (Innovative Tech Ltd, 01.01.2019), a system is known for calculating and predicting the exhaustion of a given type of currency in ATMs for their timely provision with cash.

[0007] US Patent Application JM ° 20190034842 (NCR Corp., 01/31/2019) discloses an approach to optimize cash collection service for RS, which consists in modeling and predicting the financial burden of RS in a given time interval.

[0008] The disadvantages of existing solutions is the lack of efficiency in predicting the financial burden on NDO objects.

DISCLOSURE OF THE INVENTION

[0009] The technical problem to be solved consists in the elimination of the disadvantages inherent in their prior art analogues.

[0010] The technical result is to increase the efficiency of optimizing the collection of CDN objects, due to automated forecasting of the financial load of CDN objects in a given time interval, taking into account the data on the state of each CDN object.

[OOP] An additional technical result is a reduction in the time spent on servicing NDO facilities, by ensuring the construction of data on the financial load of NDO facilities and prioritizing the maintenance of these facilities with the required amount of cash.

[0012] To achieve the claimed technical result in a preferred embodiment of the claimed solution, a computer-implemented method for optimizing the collection service of objects of cash circulation (NCO) is presented, performed using a processor and containing the steps at which:

- receive data from NDO objects for a given time period, and NDO objects are self-service devices (CS) and / or bank branches and the data will include at least information about the place location of NDO objects, history of statuses and operations of NDO objects, as well as the amount of funds at each NDO object;

- carry out processing of data on the history of statuses and operations of NDO objects;

- on the basis of the processed data of the history of statuses and operations of NLO objects, forecasting the movement of cash in each NLO object using at least one machine learning algorithm;

- carry out the optimization of cash collection services for CDN objects based on the performed forecasting for a given time period, during which a matrix of states of CDN objects with the parameters of cash collection execution is formed.

[0013] In one of the particular embodiments of the method, the IDs are selected from the group: an ATM, a payment terminal, an information-payment terminal, an automatic currency exchange terminal, or a combination thereof.

[0014] In another particular embodiment of the method, according to the data of bank branches, the data is filtered by statuses that do not change or change in a predetermined range in different time intervals the amount of funds for a given branch.

[0015] In another particular embodiment of the method, the idle dates of the devices are determined for the CD history data as the time of the inactive status of the device during the day.

[0016] In another particular embodiment of the method, a first historical slice is constructed for a CA with a small amount of downtime, containing points reflecting the status of the movement of cash in the CA in the context of a given period of time.

[0017] In another particular embodiment of the method, for the CA with the number of downtime above a predetermined threshold value, a second time slice is built, reflecting the cash flow status in the CA in the context of a given period of time, and for said second historical slice, normalization is performed based on comparison with the first slice.

[0018] In another particular embodiment of the method, during normalization, at least part of the second historical slice is replaced with indicators from the first historical slice, and the replacement is performed based on the correlation of such indicators as: US. [0019] In another particular embodiment of the method, if there are points in the first historical slice that characterize the days of DC idle, such days are replaced based on a moving average based on the indicators of neighboring points.

[0020] In another particular embodiment of the method, at the stage of forecasting, the analysis of future volumes of cash transactions for each NDO object is carried out.

[0021] In another particular embodiment of the method, the volume of transactions is determined based on the amount of funds and / or the number of sheets of notes. [0022] In another particular embodiment of the method, in a given prediction time interval, anomalies are determined that characterize periodic mass customer receipts / withdrawals of funds - salary payments.

[0023] In another particular embodiment of the method, the optimization step is performed using a stochastic optimization algorithm.

[0024] In another particular embodiment of the method, the number of downloads / unloads of funds for each NDO object is calculated.

[0025] In another particular embodiment of the method, based on the optimization results, a digital map is generated indicating the NDO objects with the date of collection and additional information.

[0026] In another particular embodiment of the method, the additional information includes at least one of: the amount of funds, the time of availability of the NDO object, the balance of funds.

[0027] Achievement of the claimed technical result is also achieved by a system for optimizing cash collection services for NDO objects, which contains at least one processor and at least one memory device containing machine-readable instructions that, when executed by the processor, perform the above method.

BRIEF DESCRIPTION OF DRAWINGS

[0028] Features and advantages of the present invention will become apparent from the following detailed description of the invention and the accompanying drawings, in which:

[0029] FIG. 1 illustrates a general diagram of device interaction.

[0030] FIG. 2 illustrates a block diagram of the steps of the claimed method.

[0031] FIG. 3 illustrates an example of a processing schedule for normalizing a historical slice of the financial history of an asset. [0032] FIG. 4 illustrates the schedule of financial cash flow for NPO objects with CFI.

[0033] FIG. 5 illustrates an event generation table.

[0034] FIG. 6 illustrates a plot of anomaly fixation.

[0035] FIG. 7 illustrates a display table of standard features.

[0036] FIG. 8 illustrates an example of a forecast graph.

[0037] FIG. 9 illustrates an example of a computing device for implementing a technical solution.

CARRYING OUT THE INVENTION

[0038] The following will describe the concepts and terms necessary to understand this technical solution.

[0039] In this technical solution, a system means, including a computer system, a computer (electronic computer), a CNC (numerical control), a PLC (programmable logic controller), computerized control systems and any other devices capable of performing a given , a well-defined sequence of operations (actions, instructions).

[0040] By a command processor is meant an electronic unit or an integrated circuit (microprocessor) that executes machine instructions (programs). [0041] A command processor reads and executes machine instructions (programs) from one or more storage devices. The role of data storage devices can be, but are not limited to, hard disks (HDD), flash memory, ROM (read only memory), solid state drives (SSD), optical drives.

[0042] A program is a sequence of instructions for execution by a computer control device or command processing device.

[0043] FIG. 1 shows a general diagram (100) of the interaction of the elements of the claimed solution. Optimization of cash collection service is carried out for NDO objects, which include various types of RS (110) and

VSP (120). US (software) can be: an ATM, a payment terminal, an information and payment terminal, an automatic currency exchange terminal and another type of device that allows you to carry out transaction operations using cash. VSP (120) means, for example, branches financial organizations, in particular banks that provide services to individuals / legal entities and accept / withdraw cash.

[0044] Each of the NDO objects (110, 120) has a unique set of parameters, which includes such information as: the location of NDO objects, the history of statuses and operations of NDO objects, as well as the amount of funds at each NDO object. This information is transmitted from NDO objects (PO, 120) through a data transmission network (130) to a central computing device (150), for example, a server, server cluster or other suitable type of computing device to perform the required functions.

[0045] As a data transmission network (130), the Internet is used, which can be organized at each of the NDO objects (PO, 120) using appropriate technical means, for example, by means of data transmission via a cellular network ( 2G / 3G / 4G / 5G), wireless WLAN, WAN, etc. The specific principle of organizing the reception / transmission of data streams between the NDO objects (110, 120) and the server (150) is selected based on the specific type of NDO object, for example, the type of the US (PO), or the most preferred type of communication for the VSP (120), for example, based on the territorial location, using well-known technical means that provide the required functionality in terms of ensuring information interaction.

[0046] Next, consider the main process of the claimed method. FIG. 2 shows a method (200) for optimizing cash collection services for NDO objects. At the initial stage (201), the information is collected by the server (150) received from the NDO objects (110, 120). As mentioned above, such information is the geographical location of NPO objects, the history of statuses and operations of NPO objects, as well as the amount of funds at each NPO object.

[0047] As information about the location, the address of the CS installation / location (110), the name of the installation object (shopping center, store, VSP, etc.), the CS (PO) location floor, geo-coordinates can be used. Also, for each NDO object (110, 120), a corresponding unique identifier (UID) is assigned. Based on the UID, a corresponding set of data is stored in the server database (150), which is necessary for performing the method (200) in terms of performing the required computational operations.

[0048] Based on the information received, at step (202), data processing on NDO objects (110, 120) is performed. During processing, for each object, the NDO (110, 120), based on the information received at step (201), is determined in time b cutaway information displaying cash flow. For US (110), such information can be historical information of device downtime, in particular, indications of activity and content of funds in a given time period. For VSP (120), in this case, the data is filtered by statuses that do not change or change in a given range in different time intervals the amount of funds. This filtering makes it possible to exclude types of cash transactions that do not significantly change the cash balance in the VSP (120) at a given threshold value. For example, such operations can be: utility payments, withdrawal of funds from a deposit and repayment of a loan with the same funds (or re-entry of these funds), etc.

[0049] During data processing at step (202), the data received from the DC (110) is analyzed and replaced with historical data to generate weighted information on their historical statuses in a time slice. Based on the information on the turnover of cash on each CS (110), the analysis of downtime and / or movements of devices is carried out, in particular, the date of the inactive status in the daily context and the movement of cash for the CS (PO) are taken into account. The analysis is carried out for each cash processing module of each CS (110), in particular, the cash dispensing module (dispenser) and cash acceptance (deposit) module.

[0050] A set of CDs (110) with a small amount of downtime is identified from the total data set obtained in step (201), in particular, the percentage of downtime of which does not exceed a predetermined threshold value. SOs (POs) with a low amount of downtime are marked as devices with a good fictitious history (HFI). The first historical slice is constructed according to the mentioned set of CAs (PO) with HFI, containing points reflecting the status of the movement of cash in the CA in the context of a given period of time.

[0051] Finding a dataset for the CA (PO) with HFI can be performed based on such historical CA data that contains a small amount of downtime, for example, <5% of the entire history, and also there are no long periods of downtime (gaps or percentage downtime above the threshold <14 consecutive days). Additionally, downtime data processing can be performed for US (software) with HFI, since there may be small gaps or a percentage of downtime above the threshold on some days. To do this, the days with downtime are replaced with moving averages based on the readings of neighboring points. Normalization (Min-Max Scaling) of the changed history based on the unmodified one can also be performed. [0052] As one example, the calculation of the DC downtime readings (110) is determined as a percentage based on the DC downtime on a particular day, at which the summation of the DC downtime minutes (110) is performed within the range of availability of a certain AC (ON) and dividing the obtained value by the total availability time (due, for example, to the operating time of the object on which the DC is located, etc.).

[0053] For the CS (ON) with the number of downtime above a predetermined threshold value, with a bad financial history, a second time slice is built, reflecting the status of cash flow in the CS (110) in the context of a given period of time. For the second historical slice, its performance is normalized based on comparison with the first historical slice. During normalization, at least a part of the second historical slice is replaced with indicators from the first historical slice, and the replacement is performed based on the correlation of such indicators as: frequency of access to the RS, the mode of accessibility of the RS, the territorial proximity between the RS. For US (110) with a bad financial history, the entire part of the historical cut before the last long downtime can be replaced with a similar part with a similar one. [0054] FIG. 3 shows graphs with a normalized financial history according to US (110). Graph (A) shows an example of a historical slice for the selected set of EOs (110), graph (B) shows a smoothed historical series for a set of EOs (110) with HFI. [0055] At step (203), based on the processed information of the history of statuses and operations of the NLO objects, at step (202), cash flow forecasting in each NLO object is performed using one or more machine learning algorithms. The input to the algorithm is information about the normalized HFI of NDO objects (PO, 120), event data and the time of availability of NDO objects (110, 120).

[0056] FIG. 4 is a graph of the input cash flow forecasting data. For each type of NDO object, the mechanism of cash flow inherent in it is analyzed. The analysis of the information at step (203) is necessary for the purposes of modeling the forecast for the withdrawal / deposit of cash to predict the objects of transactions (in the amount or number of sheets) for a given future period (for example, 14 days).

[0057] In the case of CS (PO), executed in the form of ATMs and / or payment terminals, the movement of cash may consist in issuing cash, depositing cash. For VSP (120) and RATM (ATM with a recirculation function), the key characteristics for the purpose of predicting the financial burden are the following indicators: balance and bottom (minimum cumulative balance with an aggregation period of 30 minutes) of customer cash.

[0058] FIG. 5 shows an example of a table with the formation of events, according to which anomalies (bursts / surges of activity) are recorded in the time section. Anomalies characterize periodic massive client receipts / withdrawals of funds - salary payments, etc. For NDO objects (110, 120), periodic anomalies are analyzed. Periodic anomalies are not processed. Non-periodic ones are smoothed using a moving average or a simplified forecast. The processed data is then used in the prediction stage [0059] In the prediction stage (203), the processing of the incoming data is performed using one or more (ensemble, committee) machine learning models. The following machine learning algorithms can be used: linear regression, LASSO linear regression, RIDGE linear regression, decision tree, random forest, gradient boosting on trees, recurrent neural network, etc.

[0060] At step (203), when constructing features for each subsequent predicted day in a given time interval, all values for the previous day, including those already predicted, are used. This approach is used in order to predict the day T + X, where T is the current day, X is an arbitrary day in the forecasting horizon. To do this, you need to get the characteristics of the day T + X - 1 in order to use them as relevant key figures in the calculation of data for a given time range.

[0061] FIG. 6 shows an example of a table of reflections of events in the time domain. Based on the identified events at specified time intervals, in particular, the identified anomalies, further enrichment (expansion) of the feature space is performed with calendar signs - days of the week and detected periodic anomalies.

[0062] Next, the generation of standard features is performed:

- Lags of a given depth (demand values shifted in time by a given depth);

- Moving statistics in the window (mean, std, median, min, max);

- Moving statistics by day of the week (mean, std, median, min, max);

- Statistics for the month (mean, std, median, min, max).

[0063] An example of feature generation is shown in FIG. 7. Based on the generated standard features, a forecast is built for a fixed number of days at each point in time, using the already predicted data (the forecast is made both gross for all denominations and for each denomination separately).

[0064] At the output of the work of stage (203), the forecasting of the movement of cash at the objects of the NDO (PO, 120) for a given time period is carried out. At the same time, a retro forecast is also carried out to assess the quality of the forecast model. On the basis of this estimate, the optimal parameters of the forecast models are selected, at which the forecast error is minimized. FIG. 8 is a graph showing the generated forecast. With the help of the generated indicators in the time interval, it is possible to further optimize the collection of NDO objects (110, 120).

[0065] At step (204), the optimization of the cash collection service of the NDO objects (110, 120) is performed based on the performed prediction for a given time interval, during which a matrix of states of the NDO objects with the parameters of the collection execution is formed. The input parameters for the implementation of stage (204) are:

- current funding rate (for example, 0.7);

- the selected optimization period, for example, a week, a month, etc .;

- limits (maximum cash capacity in the US (110), taking into account the maximum amount and number of bills, the number of US cassettes (110);

- cash pre-order limits for GSP (120);

- time series of cash center capacity, in particular, the maximum number of encashments per day;

- list of UID objects of NDO;

- incoming balance at the beginning of the period of optimization of NPO objects;

- costs of collection for each NDO object;

- availability of collection;

- forecasted daily gross balance and bottom (for GSP and RATM) or gross daily withdrawals / replenishments (for CA with a deposit function), taking into account the limits.

[0066] The optimization algorithm is performed using a stochastic optimization algorithm. Taking into account the analysis of the input data, the state matrix is initialized, displaying the NDO objects (110, 120), for which it is required to carry out collection. Using a stochastic optimization algorithm, valid indicators of the load on NDO facilities (110, 120) are constructed in a given time interval, taking into account the limits and the bottom (minimum daily load of the US / VSP) and optimal from the point of view of total costs.

где Yΐ прогноз снятий на день или дна (для кэшцентра в таблице состояний количество единиц в столбцах не превышает мощности за день), N - величина максимальной загрузки. С учётом входящих параметров рассчитываются загрузки/выгрузки и остатки. По прогнозным значениям сальдо и дна, а также с помощью вектора состояния рассчитываются минимальные загрузки/выгрузки для объектов НДО (НО, 120), по которым строится таблица инкассаций и последующая матрица состояний. На основании минимизации затрат на осуществление инкассаций для конкретных объектов НДО (110, 120) принимается решение о проведении инкассации. [0068] Информация о проведении инкассации на объектах НДО (110, 120) может формироваться в виде различного рода данных, отображаемых, например, в виде отчетов, графиков и т.п. Данные по оптимизации инкассации объектов НДО (110, 120) могут отображаться на цифровой ГИС - карте с указанием адресов, УИД объектов НДО и иной дополнительной информацией, например, количество денежных средств, время доступности объекта НДО, остаток денежных средств. [0067] For each NDO object (110, 120), a state vector is calculated (for the cache center, the state table), in which for the distance d between units it satisfies the inequality

where Y is the forecast of withdrawals for the day or the bottom (for the cash center in the table of states, the number of units in the columns does not exceed the capacity per day), N is the maximum load. Loading / unloading and balances are calculated taking into account the input parameters. Based on the predicted values of the balance and the bottom, as well as using the state vector, the minimum downloads / unloads for NDO objects (BUT, 120) are calculated, according to which the collection table and the subsequent matrix of states are built. On the basis of minimizing the costs of cash collection for specific NDO objects (110, 120), a decision is made to carry out cash collection. [0068] Information about cash collection at NDO facilities (110, 120) can be generated in the form of various kinds of data displayed, for example, in the form of reports, graphs, and the like. Data on optimizing the collection of NDO objects (110, 120) can be displayed on a digital GIS - map indicating the addresses, UID of NDO objects and other additional information, for example, the amount of funds, the time of availability of the NDO object, the balance of funds.

[0069] FIG. 9 shows an example of a general view of a computing system (300) based on a computing device that provides the implementation of the claimed method (200). The computing device can be a server (150), indicated earlier in the materials of the application.

[0070] In the General case, the system (300) contains one or more processors (301) united by a common bus (310), memory means such as RAM (302) and ROM (303), input / output interfaces (304) , input / output devices (305), and a device for networking (306).

[0071] The processor (301) (or multiple processors, multi-core processor) can be selected from a range of devices currently widely used, for example, Intel ™, AMD ™, Apple ™, Samsung Exynos ™, MediaTEK ™, Qualcomm Snapdragon ™ and etc. Under the processor, it is also necessary to take into account a graphics processor, for example, an NVIDIA or ATI GPU, which is also suitable for complete or partial execution of the method (200). In this case, the memory means can be the available amount of memory of the graphics card or graphics processor. [0072] RAM (302) is a random access memory and is intended for storing machine-readable instructions executed by the processor (301) for performing the necessary operations for logical processing of data. RAM (302), as a rule, contains executable instructions of the operating system and corresponding software components (applications, software modules, etc.).

[0073] ROM (303) is one or more persistent storage devices, for example, hard disk drive (HDD), solid state data storage device (SSD), flash memory (EEPROM, NAND, etc.), optical storage media ( CD-R / RW, DVD-R / RW, BlueRay Disc, MD), etc.

[0074] Various types of I / O interfaces (304) are used to organize the operation of the components of the computing system (300) and to organize the operation of external connected devices. The choice of the appropriate interfaces depends on the specific version of the computing device, which can be, but are not limited to: PCI, AGP, PS / 2, IrDa, FireWire, LPT, COM, SATA, IDE, Lightning, USB (2.0, 3.0, 3.1, micro, mini, type C), TRS / Audio jack (2.5, 3.5, 6.35), HDMI, DVI, VGA, Display Port, RJ45, RS232, etc.

[0075] To ensure user interaction with the computing system (300), various means (305) I / O information are used, for example, a keyboard, display (monitor), touch display, touch-pad, joystick, mouse manipulator, light pen, stylus, touch panel, trackball, speakers, microphone, augmented reality, optical sensors, tablet, light indicators, projector, camera, biometric identification (retina scanner, fingerprint scanner, voice recognition module), etc.

[0076] The means of networking (306) allows the system (300) to transmit data via an internal or external computer network, for example, Intranet, Internet, LAN, and the like. One or more means (306) may be used, but not limited to: Ethernet card, GSM modem, GPRS modem, LTE modem, 5G modem, satellite communication module, NFC module, Bluetooth and / or BLE module, Wi-Fi module and dr.

[0077] In addition, satellite navigation aids can also be used as part of the device (200), for example, GPS, GLONASS, BeiDou, Galileo.

[0078] The submitted application materials disclose preferred examples of the implementation of the technical solution and should not be interpreted as limiting other, particular examples of its implementation that do not go beyond the requested legal protection that are obvious to specialists in the relevant field of technology.

Claims

FORMULA 1. A computer-implemented method for optimizing cash collection services for objects of cash circulation (NCO), performed with the help of a processor and containing the stages at which: - receive data from NDO objects for a given time period, and NDO objects are self-service devices (CS) and / or bank branches and the data will include at least information about the location of NDO objects, history of statuses and operations of NDO objects, as well as the number funds at each NDO facility; - carry out processing of data on the history of statuses and operations of NDO objects; - on the basis of the processed data of the history of statuses and operations of NLO objects, forecasting the movement of cash in each NLO object using at least one machine learning algorithm; - carry out the optimization of cash collection services for CDN objects based on the performed forecasting for a given time period, during which a matrix of states of CDN objects with the parameters of cash collection execution is formed. 2. The method according to claim 1, characterized in that the RS are selected from the group: an ATM, a payment terminal, an information-payment terminal, an automatic currency exchange terminal, or a combination thereof. 3. The method according to claim 1, characterized in that according to the data of bank branches, the data is filtered by statuses that do not change or change in a predetermined range in different time intervals the amount of funds for a given branch. 4. The method according to claim 1, characterized in that the dates of device downtime are determined for the data of the RS history as the time of the inactive status of the device during the day. 5. The method according to claim 4, characterized in that for the CA with a small amount of downtime, the first historical slice is built, containing points reflecting the status of the movement of cash in the CA in the context of a given period of time. 6. The method according to claim 5, characterized in that for the CA with the number of downtime above a predetermined threshold value, a second time slice is built, reflecting the status of the movement of cash in the CA in the context of a given period of time, and for the said second historical slice is normalized based on comparison with the first historical slice. 7. The method according to claim 6, characterized in that during normalization, at least part of the second historical slice is replaced with indicators from the first historical slice, and the replacement is performed based on the correlation of such indicators as: territorial proximity between US. 8. The method according to claim 4, characterized in that if there are points in the first historical slice that characterize the days of idle RS, such days are replaced based on a moving average based on the indicators of neighboring points. 9. The method according to claim 1, characterized in that at the forecasting stage, the analysis of future volumes of cash transactions for each NDO object is carried out. 10. The method according to claim 9, characterized in that the volume of transactions is determined on the basis of the amount of funds and / or the number of sheets of notes. 11. The method according to claim 9, characterized in that in a given forecasting time interval anomalies are determined that characterize periodic mass customer receipts / withdrawals of funds - salary payments. 12. The method according to claim 1, characterized in that the optimization step is performed using a stochastic optimization algorithm. 13. The method according to claim 12, characterized in that the number of downloads / unloads of funds is calculated for each NDO object. 14. The method according to claim 1, characterized in that, based on the results of the optimization, a digital map is generated indicating NDO objects with the date of collection and additional information. 15. The method according to claim 14, characterized in that the additional information includes at least one of: the amount of funds, the time of availability of the NDO object, the balance of funds. 16. The system for optimizing the collection service of NDO objects, containing at least one processor and at least one memory device containing machine-readable instructions, which, when executed by the processor, perform the method according to any one of claims. 1-15.