Computer aided risk management in multiple-parameter physical systems

• US 5,930,762 A
• Filed: 09/24/1996
• Issued: 07/27/1999
• Est. Priority Date: 09/24/1996
• Status: Expired due to Fees

• Current Assignee: RCO Software Limited
• Original Assignee: RCO Software Limited
• Inventors: Masch, Vladimir A.
• Primary Examiner(s): Voeltz, Emanuel Todd
• Assistant Examiner(s): Hughet, William N.
• Application Number:
  US08/717,821
• Time in Patent Office: 1,036 Days
• Field of Search: 705/7, 705/8
• US Class Current: 705/7.28
• CPC Class Codes: G06F 17/10 Complex mathematical operations function generation by table look-up G06F1/03; evaluation of elementary functions by calculation G06F7/544
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 10/0635 Risk analysis of enterprise or organisation activities
• CPC Class Codes: G06Q 10/0637 Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals

Abstract

A computer method manages risk in multiple-parameter physical systems performing interrelated activities, where at least one of such activities is risk-related in that it may have an outcome level which may fall outside of boundary limits. The method establishes a course of action (strategy) for the physical systems that facilitates preventing any outcome levels for risk-related activities from falling outside of boundary limits. The method assumes the existence of a computational multiscenario decision-making model that describes the physical systems and determines, under some set of criteria, both feasible and desirable levels of their activities. The method may assume either certainty, under which the value of each model parameter is known, or uncertainty, under which the values of one or more of model parameters are unknown. The method finds a set of satisfiable boundary limits in computer memory, develops in computer memory a multitude of candidate strategies that satisfy these limits, describes the strategies in computer memory in formats of multidimensional outcome and regret matrices, and jointly applies to such matrices multiple optimization criteria: first, to exclude inferior strategies and, second, to aid decision-makers in selecting a good and risk-protected implementable strategy.

Claims

1. A computer method for managing risk in multiple parameter physical systems performing interrelated activities, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said method developing a strategy for said physical systems preventing any outcome levels for said risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in computer memory, said model including values for parameters of each of said physical systems;

(2) producing a solution of said initial model in said computer memory;

(3) deriving from said solution of said initial model a set of outcome levels for said risk-related activities and displaying said set of outcome levels in an at least three-dimensional matrix in the form of a graph;

(4) comparing said set of outcome levels for said risk-related activities with said boundary limits and, if any of said boundary limits are violated, then deriving at least one risk-limiting derivative constraint from such comparison;

(5) expanding said initial model in said computer memory by adding said risk-limiting derivative constraint to said initial model in said computer memory;

(6) producing a solution of said expanded model with said added risk-limiting derivative constraint in said computer memory; and

(7) deriving from said solution of said expanded model a new set of outcome levels for said risk-related activities and displaying said new set of outcome levels in an at least three-dimensional matrix in the form of a graph.

2. The computer risk management method of claim 1 where each risk-limiting derivative constraint is in the form of mathematical equations and inequalities.

3. The computer risk management method of claim 1 where each risk-limiting derivative constraint is in the form of fines and penalties.

4. A computer method for managing risk in multiple parameter physical systems performing interrelated activities, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said method developing a strategy for said physical systems preventing any of said outcome levels for said risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in said computer memory, said model including both a set of decision variables and values for parameters of each of said physical systems, where members of a subset of said decision variables are strategic variables and where a set of values specified for all said strategic variables constitutes a strategy;

(2) producing a solution of said initial model in said computer memory;

(3) deriving from said solution of said initial model a set of outcome levels for said risk-related activities and displaying said set of outcome levels in an at least three-dimensional matrix in the form of a graph;

(4) comparing said set of outcome levels for said risk-related activities with said boundary limits and, if the outcome level of at least one said risk-related activity falls outside said boundary limits for that activity, then deriving at least one risk-limiting derivative constraint from such comparison;

(5) expanding said initial model in said computer memory by adding said risk-limiting derivative constraint to said initial model in said computer memory;

(6) producing a solution of said expanded model with said added risk-limiting derivative constraint in said computer memory; and

(7) deriving from said solution of said expanded model a new set of outcome levels for said risk-related activities and displaying said new set of outcome levels in an at least three-dimensional matrix in the form of a graph.

5. The computer risk management method of claim 4 in which said computational model of said physical systems is used to locate regions and groups of risky decisions and strategies and to cut off said regions and groups by adding to said model at least one derivative constraint.

6. A computer method for managing risk under certainty in multiple parameter physical systems performing interrelated activities, where certainty means that none of the parameters of said systems have a plurality of possible values, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said method developing a strategy for said physical systems preventing any of, said outcome levels for risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in computer memory, said model including values for parameters of each of said physical systems;

(2) producing a solution of said initial model in said computer memory;

(3) deriving from said solution of said initial model a set of outcome levels for said risk-related activities and displaying said set of outcome levels in an at least three-dimensional matrix in the form of a graph;

(4) comparing said set of outcome levels for said risk-related activities with said boundary limits;

(5) if the outcome of at least one said risk-related activity falls outside said boundary limits for that activity, then forming a set of risk-limiting derivative constraints for all of said boundary limits;

(6) expanding said initial model in said computer memory by adding said set of risk-limiting derivative constraints to said initial model in said computer memory;

(7) producing a solution of said expanded model with said set of added risk-limiting derivative constraints in said computer memory;

(8) deriving from said solution of said expanded model a second set of outcome levels for said risk-related activities, and displaying said second set of outcome levels from said expanded model in said matrix;

(9) comparing said second set of outcome levels for said risk-related activities with said boundary limits;

(10) changing said expanded model by changing some of said boundary limits and changing said risk-limiting derivative constraints for said changed boundary limits;

(11) producing a solution of said changed model with said changed risk-limiting derivative constraints in, said computer memory;

(12) deriving from said solution of said changed model a third set of outcome levels for said risk-related activities, and displaying said third set of outcome levels from said changed model in said matrix; and

(13) repeating steps (9) through (12) until no outcome level of any of said risk-related activities falls outside said boundary limits for that activity.

7. A computer method for managing risk in multiple parameter physical systems performing interrelated activities, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said method developing a strategy for said physical systems preventing any of said outcome levels for said risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in said computer memory, said model including both a set of decision variables and values for parameters of each of said physical systems, where members of a subset of said decision variables are strategic variables; and

where a set of values specified for all said strategic variables constitutes a strategy;

(2) producing a solution of said initial model in said computer memory;

(3) deriving from said solution of said initial model a set of outcome levels for said risk-related activities and displaying said set of outcome levels in an at least three-dimensional matrix in the form of a graph;

(4) comparing said set of outcome levels for said risk-related activities with said boundary limits and, if the outcome level of at least one said risk-related activity falls outside said boundary limits for that activity, then deriving at least one risk-limiting derivative constraint from such comparison;

(5) expanding said initial model in said computer memory by adding said risk-limiting derivative constraint to said initial model in said computer memory;

(6) producing a solution of said expanded models with said added risk-limiting derivative constraint in said computer memory;

(7) deriving from said solution of said expanded model a new set of outcome levels for said risk-related activities, and displaying said new Bet of outcome levels in said matrix;

(8) deriving from said solution of said expanded model a discretionary boundary limit for at least one sa d risk-related activity;

(9) comparing said new set of outcome levels with said set of discretionary boundary limits and deriving at least one additional discretionary risk-limiting constraint from such comparison;

(10) further expanding said expanded model in said computer memory by adding said discretionary risk-limiting constraint to said expanded model in said computer memory;

(11) producing a solution of said further expanded model with said discretionary risk-limiting constraint in said computer memory; and

(12) repeating steps (7) through (11) until outcome levels of all risk-related activities fall within boundary limits.

8. A computer method for managing risk under uncertainty in multiple parameter physical systems performing interrelated activities, where uncertainty means that at least one of said parameters has a plurality of different possible values, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said systems having parameters with certain values and at least one parameter with uncertain values, said method developing a strategy for said physical systems preventing any of said outcome levels for said risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in computer memory, said model including both a set of decision variables and values for parameters of each of said physical systems, where members of a subset of said decision variables are strategic variables and where a set of values specified for said strategic variables constitutes a strategy;

(2) producing solutions of said initial model in said computer memory for a plurality of respectively different combinations of said parameter values, where each of said different combinations includes a respectively different one of said possible parameter values;

(3) deriving from said solutions of said initial model a set of outcome levels for said risk-related activities for each of said different combinations and displaying said set of outcome levels in an at least three-dimensional matrix in the form of a graph;

(4) comparing said set of outcome levels for said risk-related activities with said boundary limits and, if any of said boundary limits are violated, then deriving at least one risk-limiting derivative constraint from such comparison;

(5) expanding said model in said computer memory by adding said risk-limiting derivative constraint to said model in said computer memory;

(6) producing solutions of said expanded model with said added risk-limiting derivative constraint in said computer memory;

(7) repeating iteratively steps (2) through (6) until the outcome levels of all risk related activities fall within boundary limits;

(8) forming strategies for said physical systems from the values of said strategic variables in said solution of said expanded model and forming scenarios for different combinations of alternative values for said parameter with uncertain values; and

(9) recording outcome levels for all scenario versus strategy versus risk-related activity combinations into said outcome matrix.

9. The computer risk management method of claim 8 which includes the additional step of:

(10) deriving from entries in said outcome matrix all scenario versus strategy versus risk-related activity combinations of regret values and forming a regret matrix in the form of a graph, where regret values, for each risk-related activity, are computed by the equation

space="preserve" listing-type="equation">REGRET H! J!=BEST H!-O H! J!,

where BEST H! is the best scenario outcome for scenario H and O H! J! is the outcome for strategy J under scenario H.

10. The computer risk management method of claim 9 in which both of said matrices have n dimensions, where n is an integer greater than two.

11. A computer method for managing risk under uncertainty in multiple parameter physical systems performing interrelated activities, where uncertainty means that at least one of said parameters has a plurality of different possible values, at least two of such activities being designated as risk-related activities and having outcome levels which may fall outside of boundary limits, said method developing a strategy for said physical systems preventing any of said outcome levels for said risk-related activities from falling outside of said boundary limits and comprising the steps of:

(1) storing an initial model of said systems in, computer memory, said initial model including values for parameters of each of said physical systems, and said model comprising interconnected submodels for a plurality of respectively different combinations of said parameter values, where each of said different combinations is a respectively different scenario;

(2) entering values for each of said certain parameters into said scenario submodels in said computer memory;

(3) entering a respectively different one of said possible values for said uncertain parameter into each of said scenario submodels in said computer memory;

(4) producing a solution of said model in said computer memory, said solution including a subsolution for each of said scenario submodels;

(5) deriving from said subsolutions for each of said scenario submodels respective outcome levels for said risk-related activities for each said scenario and displaying said outcome levels in an at least three-dimensional matrix in the form of a graph;

(6) comparing said set of outcome levels for said risk-related activities from each of said scenario submodels with said boundary limits and, if any of said boundary limits are violated, then deriving at least one risk-limiting derivative constraint from such comparison;

(7) expanding at least some of said scenario submodels in said computer memory by adding said risk-limiting derivative constraint to such scenario submodels;

(8) producing solutions of said expanded submodels with said added risk-limiting derivative constraint in said computer memory; and

(9) repeating iteratively steps (4) through (8) until the outcomes of all risk related activities fall within boundary limits.