Airline seat inventory control method and apparatus for computerized airline reservation systems

• US 5,255,184 A
• Filed: 12/19/1990
• Issued: 10/19/1993
• Est. Priority Date: 12/19/1990
• Status: Expired due to Term

• Current Assignee: Accenture Global Services Limited (Accenture PLC)
• Original Assignee: Andersen Consulting LLP (Accenture PLC)
• Inventors: Gupta, Subhash, Mehrotra, Ravi, Hong, Da D., Hornick, Scot W.
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): Shingala, Gita D.
• Application Number:
  US07/630,261
• Time in Patent Office: 1,035 Days
• Field of Search: 364/407, 364/401, 364/408
• US Class Current: 705/6
• CPC Class Codes: G06Q 10/02 Reservations, e.g. for tickets, services or events
• CPC Class Codes: G06Q 10/025 Coordination of plural reservations, e.g. plural trip segments, transportation combined with accommodation

Abstract

An airline reservation system implemented using a computer wherein reservations are controlled by a seat inventory control system. The inventory control system produces optimal reservation control using network-wide booking limits while taking into account the probabilistic nature of demand. The inventory control system, based on a concept termed Network-Based Expected Marginal Seat Revenue (EMSR), does not require the large number of variables required by the other network-based approaches, and it incorporates a probabilistic demand model without resorting to computationally intractable integer programming.

Claims

1. An airline seat reservation system, comprising:

(a) a programmed computer;

(b) data storage means, coupled to the programmed computer, for storing a database describing a flight network and seat reservation requests, the flight network comprised of a known seat capacity for each of a plurality of flight legs, a known demand distribution for each of a plurality of itinerary/fare classes, a known revenue yield for a seat reserved within each itinerary/fare class, and a booking limit for each itinerary/fare class;

(c) the programmed computer including itinerary/fare class assignment means for processing the data base describing the flight network to assign seats in a flight leg to a plurality of itinerary/fare class combinations in at least one itinerary of the flight network, the assignment means comprising;

(1) means for formulating a plurality of nonlinear equations representing a constrained optimization of an expected total revenue for the flight network and a plurality of Lagrange multipliers each representing an expected marginal seat revenue for an additional seat available on a flight leg;

(2) means for refining an initial approximation of a solution to the nonlinear equations using a Newton-Raphson iteration to provide a globally optimal set of expected marginal seat revenues and booking limits;

(d) reservation terminal means, operatively connected to the programmed computer, for entering a seat reservation request for a particular itinerary/fare class; and

(e) the programmed computer including reservation means for receiving the seat reservation request from the reservation terminal means, for accepting or rejecting the seat reservation request in accordance with the globally optimal set of expected marginal seat revenues and booking limits, for recording the seat reservation request in the data storage means, and for transmitting an electronic status signal of the seat reservation request to the reservation terminal means.

2. The system of claim 1, wherein the means for formulating comprises means for maximizing the expected total revenue for the flight network, the expected total revenue being defined by:

##EQU62## subject to a capacity constraint for each flight leg a so that;

##EQU63## by introducing a Lagrange multiplier λ

_b for each flight leg a so that;

##EQU64## wherein;

(a) z represents an itinerary/fare class;

(b) Z represents a set of all itinerary/fare classes;

(c) a represents a flight leg;

(d) A represents a set of all flight legs;

(e) V(a) represents the set of all itinerary/fare classes that use flight leg a;

(f) b represents a flight leg of the itinerary/fare class z;

(g) U(z) represents the set of all flight legs used by the itinerary/fare class z;

(h) d₂ (x) represents a demand distribution function for the itinerary/fare class z;

(i) S_z represents a booking limit for the itinerary/fare class z;

(j) C_a represents a capacity for the flight leg a;

(k) f_z represents a revenue yield for each seat sold in the itinerary/fare class z;

(1) R_z (S_z) represents an expected total revenue for the itinerary/fare class z so that;

##EQU65## (m) R_z '"'"' represents a derivative of R_z, i.e., an expected marginal seat revenue for satisfying additional demand in excess of S_z in the itinerary/fare class z, so that;

##EQU66## (n) Q_z represents an inverse of R_z '"'"' wherein Q_z (R_z '"'"'(S_z)/f_z)=S_z ; and

(o) λ

_b represents an expected marginal seat revenue generated by an additional seat in the flight leg b of the itinerary.

3. The system of claim 1, wherein the means for refining further comprises means for solving a plurality of linear equations for each Newton-Raphson iteration using a Conjugate Gradient iteration.

4. The system of claim 3, where in the means for solving comprises:

(a) means for representing a Jacobian matrix as a summation of a plurality of itinerary matrices, each itinerary matrix comprising a set of identical entries characteristic of an itinerary, the entries comprising partial derivatives of the residual capacity of the itinerary with respect to the expected marginal seat revenues of the flight legs of the itinerary; and

(b) means for accumulating the product of the itinerary matrices and an arbitrary vector into a product of the Jacobian matrix and the arbitrary vector on an itinerary-by-itinerary basis.

5. The system of claim 3, wherein the means for solving further comprises means for terminating the Conjugate Gradient iteration when an approximate solution to the linear system of equations satisfies the convergence criterion defined by:

##EQU67## so that ##EQU68## wherein;

(1) J is a Jacobian matrix comprised of a summation of a plurality of itinerary matrices, each itinerary matrix comprising a set of identical entries characteristic of an itinerary, the entries comprising partial derivatives of a residual capacity of the flight legs of the itinerary with respect to the expected marginal seat revenues of the flight legs of the itinerary;

(2) δ

₁ is a vector comprising adjustments to the expected marginal seat revenues;

(3) γ

₁ is a vector comprising the residual capacities of the flight legs;

(4) ι

denotes an ι

th iterated vector;

(5) η

₁ is a relaxation coefficient; and

(6) κ

is an arbitrary constant.

6. The system of claim 1, wherein the reservation means for accepting or rejecting the seat reservation request comprises means of accepting the seat reservation request when a total number of seats reserved in the itinerary/fare class does not exceed the booking limit for the itinerary/fare class.

7. The system of claim 1, wherein the reservation means for accepting or rejecting the seat reservation request comprises means for rejecting the seat reservation request when the seat reservation request would yield revenue less than a sum of the expected marginal seat revenues for all flight legs int he itinerary.

8. A yield management system, comprising:

(a) a programmed computer;

(b) data storage means, coupled to the programmed computer, for storing a data base describing known resource capacity for each of a plurality of resource categories, a known demand distribution for each of a plurality of demand categories, a known revenue yield for a resource reserved within each demand category, and a booking limit for each demand category;

(c) the programmed computer including assignment means for processing the database to assign resources in a resource category to a plurality of demand categories, the assignment means comprising;

(1) means for formulating a plurality of nonlinear equations representing a constrained optimization of an expected total revenue for the database and a plurality of Lagrange multipliers each representing an expected marginal resource revenue for an additional resource available on a resource category;

(2) means for refining an initial approximation of a solution to the nonlinear equations using a Newton-Raphson iteration to provide a globally set of expected marginal resource revenues and booking limits;

(d) reservation terminal means, operatively connected to the programmed computer, for entering a resource reservation request for a particular demand category; and

(e) the programmed computer including reservation means for receiving the resource reservation request from the reservation terminal means, for accepting or rejecting the resource reservation request in accordance with the globally optimal set of expected marginal resource revenues and booking limits, for recording the resource reservation request in the data storage means, and for transmitting to the reservation terminal means an electronic status signal of the resource reservation request.

9. The system of claim 8, wherein the means for formulating comprises means for maximizing the expected total revenue for the database, the expected total revenue being defined by:

##EQU69## subject to a capacity constraint for each resource category a so that;

##EQU70## by introducing a Lagrange multiplier λ

_b for each resource category a so that;

##EQU71## wherein;

(a) z represents a demand category;

(b) Z represents a set of all demand categories;

(c) a represents a resource category;

(d) A represents a set of all resource categories;

(e) V(a) represents the set of all demand categories that use resource category a;

(f) b represents a resource category of the demand category z;

(g) U(z) represents the set of all resource categories used by the demand category z;

(h) d_z (x) represents a demand distribution function for the demand category z;

(i) S_z represents a booking limit for the demand category z;

(j) C_a represents a capacity for the resource category a;

(k) f_z represents a revenue yield for each resource in the demand category z;

(l) R_z (S_z) represents an expected total revenue for the demand category z so that;

##EQU72## (m) R_z '"'"' represents a derivative of R_z, i.e., an expected marginal resource revenue for satisfying additional demand in excess of S_z in the demand category z, so that;

##EQU73## (n) Q_z represents an inverse of R_z '"'"' wherein Q_z (R_z '"'"'(S_z)/f_z)=S_z ; and

(o) λ

_b represents an expected marginal resource revenue generated by an additional resource in the resource category b.

10. The system of claim 8, wherein the means for refining further comprises means for solving a plurality of linear equations for each Newton-Raphson iteration using a Conjugate Gradient iteration.

11. The system of claim 10, wherein the means for solving further comprises means for terminating the Conjugate Gradient iteration when an approximate solution to the linear system of equations satisfies the convergence criterion defined by:

##EQU74## so that ##EQU75## wherein;

(1) J is a Jacobian matrix comprised of a summation of a plurality of demand category matrices, each demand category matrix comprising a set of identical entries characteristic of the demand category, the entries comprising partial derivatives of a residual capacity of the resource categories used by the demand category with respect to the expected marginal resource revenues of those resource categories;

(2 ) δ

₁ is a vector comprising adjustments to the expected marginal resource revenues;

(3) γ

₁ is a vector comprising the residual capacities of the resource categories;

(4) ι

denotes an ι

th iterated vector;

(5) η

₁ is a relaxation coefficient;

(6) κ

is an arbitrary constant.

12. The system of claim 8, wherein the reservation means for accepting or rejecting the resource reservation request comprises means for accepting the resource reservation request when a total number of resources reserved in the demand category does not exceed the booking limit for the demand category.

13. The system of claim 8, wherein the reservation means for accepting or rejecting the resource reservation request comprises means for rejecting the resource reservation request when the resource reservation request would yield revenue less than a sum of the expected marginal resource revenues for all resource categories.

14. An airline seat reservation system, comprising:

(a) a programmed computer;

(b) data storage means, coupled to the programmed computer, for storing a database describing a flight network, the flight network comprised of a known seat capacity for each of a plurality of flight legs, a known demand distribution for each of a plurality of itinerary/fare classes, a known revenue yield for a seat reserved within each itinerary/fare class, and a booking limit for each itinerary/fare class;

(c) the programmed computer including itinerary/fare class assignment means for processing the database describing the flight network to assign seats in a flight leg to a plurality of itinerary/fare class combinations in at least one itinerary of the flight network, the assignment means comprising;

(1) means for formulating a plurality of nonlinear equations representing a constrained optimization of an expected total revenue for the flight network and a plurality of Lagrange multipliers each representing an expected marginal seat revenue for an additional seat available on a flight leg; and

(2) means for refining an initial approximation of a solution to the nonlinear equations using a Newton-Raphson iteration to provide a globally optimal set of expected marginal seat revenues and booking limits.

15. The system of claim 14, wherein the means for formulating comprises means for maximizing the expected total revenue for the flight network, the expected total revenue being defined by:

##EQU76## subject to a capacity constraint for each flight leg a so that;

##EQU77## by introducing a Lagrange multiplier λ

_b for each flight leg a so that;

##EQU78## wherein;

(a) z represents an itinerary/fare class;

(b) Z represents a set of all itinerary/fare classes;

(c) a represents a flight leg;

(d) A represents a set of all flight legs;

(e) V(a) represents the set of all itinerary/fare classes that use flight leg a;

(f) b represents a flight leg of the itinerary/fare class z;

(g) U(z) represents the set of all flight legs used by the itinerary/fare class z;

(h) d_z (x) represents a demand distribution function for the itinerary/fare class z;

(i) S_z represents a booking limit for the itinerary/fare class z;

(j) C_a represents a capacity for the flight leg a;

(k) f_z represents a revenue yield for each seat in the itinerary/fare class z;

(l) R_z (S_z) represents an expected total revenue for the itinerary/fare class z so that;

##EQU79## (m) R_z '"'"' represents a derivative of R_z, i.e., an expected marginal seat revenue for satisfying additional demand in excess of S_z in the itinerary/fare class z, so that;

##EQU80## (n) Q_z represents an inverse of R_z '"'"' wherein Q_z (R_z '"'"'(S_z)/f_z)=S_z ; and

(o) λ

_b represents an expected marginal seat revenue generated by an additional seat in the flight leg b of the itinerary.

16. The system of claim 14, wherein the means for refining further comprises means for solving a plurality of linear equations for each Newton-Raphson iteration using a Conjugate Gradient iteration.

17. The system of claim 16, where in the means for solving comprises:

(a) means for representing a Jacobian matrix as a summation of a plurality of itinerary matrices, each itinerary matrix comprising a set of identical entries characteristic of an itinerary, the entries comprising partial derivatives of the residual capacity of the itinerary with respect to the expected marginal seat revenues of the flight legs of the itinerary; and

(b) means for accumulating the product of the itinerary matrices and an arbitrary vector into a product of the Jacobian matrix and the arbitrary vector on an itinerary-by-itinerary basis.

18. The system of claim 16, wherein the means for solving further comprises means for terminating the Conjugate Gradient iteration when an approximate solution to the linear system of equations satisfies the convergence criterion defined by:

##EQU81## so that ##EQU82## wherein;

(1) J is a Jacobian matrix comprised of a summation of a plurality of itinerary matrices, each itinerary matrix comprising a set of identical entries characteristic of an itinerary, the entries comprising partial derivatives of a residual capacity of the flight legs of the itinerary with respect to the expected marginal seat revenues of the flight legs of the itinerary;

(2) δ

₁ is a vector comprising adjustments to the expected marginal seat revenues;

(3) γ

₁ is a vector comprising the residual capacities of the flight legs;

(4) ι

denotes an ι

th iterated vector;

(5) η

₁ is a relaxation coefficient; and

(6) κ

is an arbitrary constant.

19. A yield management system, comprising:

(a) a programmed computer;

(b) data storage means, coupled to the programmed computer, for storing a database describing a known resource capacity for each of a plurality of resource categories, a known demand distribution for each of a plurality of demand categories, a known revenue yield for a resource reserved within each deamand category, and a booking limit for each demand category;

(c) the programmed computer including assignment means for processing the database to assign resources in a resource category to a plurality of demand categories, the assignment means comprising;

(1) means for formulating a plurality of nonlinear equations representing a constrained optimization of an expected total revenue for the database and a plurality of Lagrange multipliers each representing an expected marginal resource revenue for an additional resource available on a resource category; and

(2) means for refining an initial approximation of a solution to the nonlinear equations using a Newton-Raphson iteration to provide a globally optimal set of expected marginal resource revenues and booking limits.

20. The system of claim 19, wherein the means for formulating comprises means for maximizing the expected total revenue for the network, the expected total revenue being defined by:

##EQU83## subject to a capacity constraint for each resource category a so that;

##EQU84## by introducing a Lagrange multiplier λ

for each resource category a so that;

##EQU85## wherein;

(a) z represents a demand category;

(b) Z represents a set of all demand categories;

(c) a represents a resource category;

(d) A represents a set of all resource categories;

(e) V(a) represents the set of all demand categories that use resource category a;

(f) b represents a resource category of the demand category z;

(g) U(z) represents the set of all resource categories used by the demand category z;

(h) d₂ (x) represents a demand distribution function for the demand category z;

(i) S₂ represents a booking limit for the demand category z;

(j) C_a represents a capacity for the resource category a;

(k) f_z represents a revenue yield for each resource in the demand category z;

(l) R_z (S_z) represents an expected total revenue for the demand category z so that;

##EQU86## (m) R_z '"'"' represents a derivative of R_z, i.e., an expected marginal resource revenue for satisfying additional demand in excess of S_z in the demand category z, so that;

##EQU87## (n) Q_z represents an inverse of R_z '"'"' wherein Q_z (S_z '"'"')(S_z)/f_z)=S_z ; and

(o) λ

_b represents an expected marginal resource revenue generated by an additional resource in the resource category b.

21. The system of claim 19, wherein the means for refining further comprises means for solving a plurality of linear equations for each Newton-Raphson iteration using a Conjugate Gradient iteration.

22. The system of claim 21, wherein the means for solving further comprises means for terminating the Conjugate Gradient iteration when an approximate solution to the linear system of equations satisfies the convergence criterion defined by:

##EQU88## so that ##EQU89## wherein;

(1) J is a Jacobian matrix comprised of a summation of a plurality of demand category matrices, each demand category matrix comprising a set of identical entries characteristic of the demand category, the entries comprising partial derivatives of a residual capacity of the resource categories used by the demand category with respect to the expected marginal resource revenues of those resource categories;

(2) δ

₇₆ is a vector comprising adjustments to the expected marginal resource revenues;

(3) γ

₇₆ is a vector comprising the residual capacities of the resource categories;

(4) ι

denotes an ι

th iterated vector;

(5) η

₇₆ is a relaxation coefficient; and

(6) κ

is an arbitrary constant.