RU2632124C1 - Method of predictive assessment of multi-stage process effectiveness - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: method records the necessary initial data in the memory, analyzes the correspondence of the solutions and situations, determines the likelihood of timely and correct decision-making, classifies the variants of the initial state, the results of the implementation of the intermediate stages and the results of the final stage of the multi-stage process, sets the apriori probability of situations corresponding to the classes of the initial state of the multi-stage process, determines the probability of occurrence of situations, which correspond to the classes of intermediate states of the multi-stage process and to the classes of the final state of the multi-stage process, as well as for the multi-stage process, calculates the apriori probability of achieving the targets, forms a predictive characteristics of the classes of the initial state, receives information about the values of the predictive characteristics of the classes of state, and calculates a posteriori probability of the initial state and the predictive value of the efficiency indicator.

EFFECT: providing an automated calculation of the predicted value of the efficiency indicator of multi-stage processes.

1 dwg

Description

The invention relates to methods for processing digital data for special applications in the field of forecasting and control of multi-stage processes characterized by a priori uncertainty of situations arising from the implementation of the stages.

At the same time, multi-stage processes are understood to mean processes in which management decisions are made at each stage, depending on the results of the previous stage.

A known method of providing assistance to the design and production processes (RU 2321886, 2008), comprising using a data input module to receive a plurality of product characteristics values associated with a set of products having a deviation range with respect to a plurality of product characteristics obtained in the process; using a correlation module to select a predictor characteristic from a plurality of product characteristics; and using a regression module to determine the regression model (s) between the predictor characteristic and at least one of the remaining product characteristics in said plurality of product characteristics. The present invention provides knowledge of how many of the characteristics of a given end result of a process are related to each other, with specification limits and input for preprocessing. This knowledge reduces the cost of measurement, analysis and reporting, both before and during manufacture. It also identifies the changes that are needed for the input to produce pre-processing in order to achieve manufacturing for design purposes.

The disadvantage of this method is the relatively narrow scope, due to the assumption that the process can be described by a regression equation that connects the output to input data, which does not allow taking into account the results of managerial decisions at intermediate stages of multi-stage processes.

Also known is a method of statistical regulation of a technological process (RU 2470352, 2012), in which product quality indicators are determined, technological process parameters are selected, measured, a matrix of quantitative and qualitative parameters control matrix is formed according to the first embodiment of the method, or, according to the second embodiment of the method, form a matrix of a control card of quantitative and qualitative parameters and a matrix of a control card of qualitative parameters, carry out statistical accounting and analysis h deviations of the parameters of the technological process and indicators of product quality in comparison with the selected values, while determining by statistical methods the calculation of the correlation coefficient estimates and evaluating the state of the technological process, and performing actions regulating the technological process, including through acceptance quality control, identify types of defects that correspond to technological operations and equipment and carry out corrective control actions.

The disadvantage of this method is the relatively narrow scope, due to the amount of data required for statistical accounting and analysis of deviations of process parameters and product quality indicators.

Also known is a method of evaluating the effectiveness of the process of developing military equipment facilities (RU 2282243, 2006), based on the formation of data arrays on the technical and economic indicators of the developed military equipment with the display and combination of this information in the windows on the display screen. The technical result is the provision of a method of computer simulation of the process of warfare, providing a simplified simulation of the process of warfare.

The disadvantage of this method is the relatively narrow scope, due to the features of the process of warfare and making it difficult to use to predict the effectiveness of multi-stage processes in other areas.

Also known is a unified method for evaluating the effectiveness of large systems and an automated device for its implementation (RU 2453912, 2012), including the presentation of a specific large system (BS) in the form of a hierarchy of its structural elements, recording performance indicators in memory of each element of the structure BS, standard values and importance factors corresponding to each particular performance indicator, after which they receive the value of a generalized performance indicator, pre were lent by the convolution of partial performance.

The disadvantage of this method is the relatively narrow scope, due to the requirement for independence of private performance indicators, which does not allow the use of this method for the predictive evaluation of the effectiveness of multi-stage processes, in which the result of the next step depends on the state of the process at the previous stage.

The closest analogue of the claimed method is a method for evaluating the effectiveness of control (RU 2517409, 2014), in which the necessary initial data is written to the storage devices, the sum of the values of the weight coefficients of importance Ki of the basic argument is calculated, the weight coefficient K is calculated, the correspondence of the solutions and situations is analyzed , take into account the importance coefficient of the first argument, calculate the number of coincidence of solutions and situations, display on the screen of the display unit, analyze the results of estimating the number of of falling decision options and situations and by the largest number of coincidences choose the best solution, estimate the time spent on making a decision by the i-th decision maker in the j-th situation, determine the number of decisions Rj made in a timely manner for the j-th situation, determine the correctly made decisions for j- th situation, determine the number of correctly made decisions for the j-th situation Sj, determine the total number of correctly made decisions, determine the total relative number of correctly made decisions, determine the relative amount in a timely manner and correctly made decisions, determine the total relative number of timely and correct decisions, determine the likelihood of making decisions correctly, determine the likelihood of timely and correct decisions, determine the value of the management efficiency indicator, display on the screen the display unit and analyze the resulting assessment of management effectiveness, simultaneously with recording the necessary source data in the storage device also record data on the number of communication directions Q, data permissible time td.p for the implementation of decisions, the values of time t ij spent on the implementation of the i-th solution in the j-th situation, at the same time as estimating the time spent on the decision of the i-th decision maker in the j-th situation, evaluate the time t ij using the block for estimating the time of implementation of decisions, simultaneously with determining the number of decisions Rj made in a timely manner for the jth situation, determine the number of timely decisions Cj using the block for determining the number of timely decisions, simultaneously with By dividing the total relative number of timely and correctly made decisions, determine the total relative number of timely implemented decisions using the unit for determining the total relative number of timely implemented decisions, simultaneously with determining the probability of correct decision making and the probability of timely and correct decision making, determine the probability of timely implementation of decisions made using the block determining the likelihood of timely implementation of accepted solution, characterized in that, in addition to simultaneously recording the necessary initial data in the storage devices, they also record data on the admissible time for collecting information tdop.i, and the value of time tuj spent on collecting all the information necessary for making a decision in the jth situation, simultaneously with the assessment the time spent on making a decision by the i-th decision maker in the j-th situation, estimate the time tuj spent on collecting all the information necessary for making a decision in the j-th situation, using the block for estimating the time of collecting information in the j-th situation the number of situations Msv.inf, j, for making decisions in which all the necessary information was collected in a timely manner, using the unit for determining the number of situations with timely collected information, simultaneously with determining the probability of a correct decision making, determine the probability of timely collection of all the information necessary for decision-making. Lok determine the likelihood of the timely collection of information.

The disadvantage of this method is the relatively narrow scope, due to the fact that:

- the sequence of decision-making at various stages of multi-stage processes is not taken into account;

- many situations are considered formed in advance;

- only the indicator of management efficiency is calculated (the probability of timely and correct decision-making), while the effectiveness of the managed process is not evaluated (the probability of achieving the goals of the process).

The required technical result is the provision of automated calculation of the predicted value of the efficiency indicator of multi-stage processes, taking into account:

- the uncertainty of situations arising from the implementation of the stages of multi-stage processes;

- the consequences of decision-making at various stages of multi-stage processes;

- current information about the state of the process.

The desired result is achieved through the application of the proposed method, which differs from the already known in that:

- carry out the classification of options for the initial state of the multi-stage process, the classification of the results of the implementation of the intermediate stages of the multi-stage process, the classification of the results of the implementation of the final stage of the multi-stage process;

- set the a priori probability of the onset of situations corresponding to the classes of the initial state of a multi-stage process;

- determine the probability of occurrence of situations corresponding to classes of intermediate states of a multi-stage process;

- determine the probability of occurrence of situations corresponding to the classes of the final state of a multi-stage process;

- calculate the a priori probability of achieving the goals of the multi-stage process;

- form predictive signs of the classes of the initial state of the multi-stage process;

- receive information about the values of the predicted signs of the classes of the state of the multi-stage process;

- calculate the posterior probability of the initial state of the multi-stage process and the predicted value of the efficiency indicator of the multi-stage process.

In this case, the operations of classification and calculation of probabilities are performed using a specialized computing device or personal computer software.

The essence of the proposed method is as follows.

A multi-stage process is represented as a set of indicators reflecting its initial state and state based on the results of each of the stages. Since the proposed method is intended for the predictive assessment of the effectiveness of multi-stage processes, these indicators are probabilistic in nature, that is, they are random variables whose values depend on previously unknown uncontrolled random factors and decisions made at each stage. The construction of a complete probabilistic model of a multi-stage process is difficult, since it requires a joint assessment of the probability of the values of all indicators of a multi-stage process. Therefore, to evaluate the predicted effectiveness of a multi-stage process, it is proposed to present the state of a multi-stage process at each stage in the form of a corresponding set of classes and use the values of the conditional probability of occurrence of situations corresponding to the state classes of a multi-stage process, depending on the class of its state at the previous stage.

To achieve the desired technical result, the necessary initial data is recorded in the storage devices. Moreover, the multi-stage process is described by the corresponding indicators at each of j = 0..k stages (j = 0 is the index of the initial state of the multi-stage process, j = k is the index of the state of the multi-stage process according to the results of the final stage).

, таким образом, чтобы каждый из классов мог быть идентифицирован при контроле многоэтапного процесса. Результатом выполнения данного этапа является множество классов исходного состояния многоэтапного процесса S⁰:Next, carry out the classification of options for the initial state of a multi-stage process. For this, the set of values of the indicators of the initial state of a multi-stage process is divided into disjoint subsets corresponding to classes

, so that each of the classes can be identified by monitoring the multi-step process. The result of this step is the set of classes of the initial state of the multi-stage process S ⁰ :

where i ⁰ = 1..I ⁰ is the set of class indices of the initial state of a multi-stage process.

, таким образом, что каждому классу соответствует своя задача принятия управленческого решения. Результатом выполнения данного этапа является множество классов промежуточных состояний многоэтапного процесса {S^j} для каждого из j=1..(k-1) этапов:Then, the results of the implementation of the intermediate stages of the multi-stage process are classified. For this, the set of values of the indicators of the intermediate states of a multi-stage process is divided into disjoint subsets corresponding to the classes

Thus, each class has its own task of making managerial decisions. The result of this stage is the set of classes of intermediate states of the multi-stage process {S ^j } for each of j = 1 .. (k-1) stages:

where i ^j = 1..I ^j is the set of class indices of the intermediate state of the multi-stage process at the jth stage.

, таким образом, что каждому классу соответствует различная вероятность достижения целей многоэтапного процесса. Результатом выполнения данного этапа является множество классов конечного состояния многоэтапного процесса S^k:Then, the results of the implementation of the final stage of the multi-stage process are classified. For this, the set of values of the indicators of the final state of a multi-stage process is divided into disjoint subsets corresponding to classes

Thus, each class has a different probability of achieving the goals of the multi-stage process. The result of this step is the set of classes of the final state of the multi-stage process S ^k :

where i ^k = 1..I ^k is the set of class indices of the final state of a multi-stage process.

. Данная вероятность характеризует данные о процессе, полученные до начала его контроля и прогнозной оценки эффективности. В случае отсутствия таких данных ситуации, соответствующие классам исходного состояния многоэтапного процесса, считаются равновероятными:Next, the a priori probability of the occurrence of situations corresponding to the classes of the initial state of the multi-stage process is set

. This probability characterizes the data on the process obtained before the start of its control and predictive evaluation of efficiency. In the absence of such data, situations corresponding to the classes of the initial state of a multi-stage process are considered equally probable:

, где

- множество вариантов решений на j-м этапе процесса в i^j-м классе состояния процесса.Then analyze the conformity of options for decisions and situations. For this, for each of the situations corresponding to the classes of the initial and intermediate states of the multi-stage process, the value of the utility function of management decisions is calculated

where

- a lot of solution options at the jth stage of the process in the i ^jth class of the state of the process.

с применением методов теории вероятности и теории игр определяют вероятность своевременного и правильного принятия решений

при условии того, что процесс на j-м этапе находится в i^j-м классе состояния.Further, based on the values of the utility function of management decisions

using the methods of probability theory and game theory determine the probability of timely and correct decision-making

provided that the process at the jth stage is in the i ^jth class of state.

. Данная вероятность является условной и рассчитывается исходя из того, что принято своевременное и правильное решение

и процесс находится в определенном состоянии

по результатам предыдущего этапа.Then, on the basis of the methods of probability theory and game theory, the probability of occurrence of situations corresponding to the classes of intermediate states of a multi-stage process is determined

. This probability is conditional and is calculated on the basis that a timely and correct decision has been made.

and the process is in a certain state

according to the results of the previous stage.

.After that, on the basis of the methods of probability theory and game theory, the probability of occurrence of situations corresponding to the classes of the final state of a multi-stage process is determined

.

Next, calculate the a priori probability of achieving the goals of the multi-stage process P (Y), where Y is the event that consists in achieving the goals of the multi-stage process. This operation can be performed as follows:

where P (Y⎪S ^k ) is the probability of achieving the goals of the multi-stage process obtained by classifying the results of the implementation of its final stage.

, где

- множество индексов прогнозных признаков. Данные признаки используются для идентификации класса исходного состояния многоэтапного процесса при его контроле и прогнозной оценке эффективности и формируются таким образом, чтобы была известна условная вероятность их проявления в зависимости от класса исходного состояния многоэтапного процесса Р(X⎪S⁰)After this, the prognostic signs of the classes of the initial state of the multi-stage process are formed.

where

- many indices of forecast signs. These signs are used to identify the class of the initial state of a multi-stage process during its control and predictive evaluation of efficiency and are formed in such a way that the conditional probability of their manifestation depending on the class of the initial state of the multi-stage process P (X⎪S ⁰ ) is known

.Next, information is obtained on the values of the forecast features of the classes of the initial state of the multi-stage process

.

:After that, using the methods of probability theory (Bayesian formulas), the posterior probability of realizing the initial state of a multi-stage process is calculated

:

.Next, using the methods of probability theory, the predicted value of the efficiency indicator of a multi-stage process is calculated, which is defined as the probability of achieving the goals of the process taking into account current information about the values of the forecast features of the classes of the initial state of the multi-stage process

.

The proposed method can be illustrated by the following example.

The predicted effectiveness of space activities aimed at solving emergency monitoring tasks is evaluated. Events are presented in the form of the following two-stage process: “production and launch of spacecraft; operation of a constellation of spacecraft. "

The initial state of the process is represented by three classes of states:

- степень готовности предприятий промышленности к производству и запуску космических аппаратов «высокая»;-

- the degree of readiness of industrial enterprises for the production and launch of spacecraft is “high”;

- степень готовности предприятий промышленности к производству и запуску космических аппаратов «средняя»;-

- the degree of readiness of industrial enterprises for the production and launch of spacecraft is “average”;

- степень готовности предприятий промышленности к производству и запуску космических аппаратов «низкая»;-

- the degree of readiness of industrial enterprises for the production and launch of spacecraft is “low”;

According to the results of the production and launch of spacecraft, the process can go into the following classes of states:

- выведено на орбиту 4-5 космических аппаратов;-

- launched into orbit 4-5 spacecraft;

- выведено на орбиту 2-3 космических аппарата;-

- 2-3 spacecraft were put into orbit;

The results of the operation of a grouping of spacecraft (taking into account their possible breakdown during operation) can transfer the process to the following classes of final states:

- периодичность обзора заданного участка местности более 10 раз в сутки;-

- the frequency of review of a given area more than 10 times a day;

- периодичность обзора заданного участка местности 5-10 раз в сутки;-

- the frequency of review of a given area 5-10 times a day;

- периодичность обзора заданного участка менее 5 раз в сутки.-

- the frequency of review of a given area less than 5 times a day.

,

,

.The probability of achieving the goals of the process depending on the classes of final states is defined as follows:

,

,

.

.The initial conditions of the process are considered equally probable and the a priori probability is calculated as:

.

The probabilities of correct and timely decision-making are defined for the presented situations (process state classes) as follows:

,

,

,

,

,

.

,

.

The probabilities of occurrence of situations corresponding to the classes of intermediate states of the process in question are calculated on the basis of modeling:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

.

,

,

,

.

The data presented allow us to calculate the a priori probability of achieving the goals of the process in question - P (Y) = 0.395.

In FIG. Figure 1 presents a graph model with weights corresponding to the conditional probabilities of the process in question.

As predictive signs of the classes of the initial state of the multi-stage process X are used:

- the share of R&D in the total volume of work is more than 40%;

- the share of research in the volume of research and development more than 20%;

- the level of use of production capacity of more than 80%;

- the rate of renewal of fixed assets over 15%.

Suppose that at the time of the predictive assessment only the value of the first sign can be obtained, and using expert assessment it is established that the conditional probability of its manifestation, depending on the class of the initial state of the process under consideration, is:

.

.

If the predictive sign is observed according to the control results, the posterior probability of the initial state classes is calculated:

,

,

,

,

.

.

And accordingly, the value of the efficiency indicator of a multi-stage process is calculated, defined as the probability of achieving the goals of the process, taking into account current information about the values of the predicted attribute: P (Y⎪x _l ) = 0.48.

Upon receipt of information on the value of other forecast signs, the adjustment of the value of the performance indicator is carried out in a similar way.

Thus, a predictive assessment of the effectiveness of the multi-stage process of space activity was carried out taking into account the uncertainty of situations that arise during the implementation of the process steps, the consequences of decision-making at various stages of multi-stage processes and current information about the state of the process, which confirms the required technical result.

The proposed technical solution is industrially applicable, as it is based on operations that are widely used in automated computing and control systems, and can be implemented either as a device with specialized units, or on the basis of a personal computer with appropriate software for performing the functions provided.

Claims (1)

The method of calculating the predicted value of the efficiency indicator of multi-stage processes, which consists in writing the necessary source data to the storage devices, analyzing the correspondence of the solutions and situations, determining the probability of timely and correct decision-making, characterized in that using a specialized computing device or personal software computers carry out the classification of options for the initial state of a multi-stage process, classification June the results of the implementation of the intermediate stages of the multi-stage process, the classification of the results of the final stage of the multi-stage process, write the classification results to the storage device, set the a priori probability of the occurrence of situations corresponding to the classes of the initial state of the multi-stage process, determine the probability of occurrence of situations corresponding to the classes of intermediate states of the multi-stage process, determine the probability occurrence of situations corresponding to classes of finite conditions of the multi-stage process, calculate the a priori probability of achieving the goals of the multi-stage process, write the probability of occurrence of situations into the storage device, generate the predicted signs of the classes of the initial state of the multi-stage process, write the values of the signs of the classes of the initial state of the multi-stage process into the memory, obtain information about the values of the predicted signs of the stages of the multi-stage process process using a specialized computing device or personal computer software calculates the posterior probability of implementing the initial state of a multi-stage process, calculates the predicted value of the efficiency indicator of a multi-stage process.

Images