AU2020100769A4

AU2020100769A4 - Infrastructure asset management system and/or method

Info

Publication number: AU2020100769A4
Application number: AU2020100769A
Authority: AU
Inventors: Craig Hunter
Original assignee: Smart Infrastructure Asset Management Australia Res And Development Pty Ltd
Current assignee: Smart Infrastructure Asset Management Australia Research And Development Pty Ltd
Priority date: 2017-12-19
Filing date: 2020-05-18
Publication date: 2020-06-25
Anticipated expiration: 2026-12-19

Abstract

The present invention relates to a system for monitoring the condition of one or more civil structures and forecasting, using one or more processors, the 5 deterioration of each civil structure. The system comprises at least one computer readable medium for storing historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more civil structures, and for storing environmental parameters. At least one processor is arranged to process the condition state data 10 for each component and the environmental parameters to predict information indicative of future performance-related condition states associated with the component. 94-09-14 - -16 96-07-3D 98-09-14 99-10-26 00-09-24 01-11-12 C2-12-01 D_ 67 0.6667 C.6667 0.6667 0 _6667 0.66,67 056667 11-17 04-11-02 05-10-10 0 3 033 0 3333 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 11-10-25 08333 12-07-08 13-05-13 7991 0.7991 14-07-08 15-05-11 16-03-10 16-11-21 667 0.6667 0.6667 0.6667 2011 2012 2013 2014 2015 2016 2017

Description

INFRASTRUCTURE ASSET MANAGEMENT SYSTEM AND/OR METHOD

FIELD OF THE INVENTION

The present invention relates to a civil infrastructure asset management system and/or method for managing and/or forecasting infrastructure asset conditions. The present invention has particular, although not exclusive application to a bridge related infrastructure asset management system and/or method and can be used for other types of infrastructure.

BACKGROUND TO THE INVENTION

Infrastructure asset management involves strategies for sustaining and prolonging the lifetime of infrastructure assets such as roads, utility grids, bridges, railways and the like. The process utilises software tools to determine appropriately timed 15 maintenance, repair and replacement of assets to achieve preservation and extension of lifetime in an economical and sustainable manner.

One application of an asset management system is a bridge related asset management system. The efficient use of maintenance funds and of budgeting for 20 the well-being of bridges requires effective bridge asset management technology and its application. A Bridge Management System (BMS) is nowadays essential and helps determine the complexity of decision-making for bridge maintenance, repair and rehabilitation (MR&R) strategies for bridge authorities. The most well-known commercial version of BMS software was developed in the early 1990s and has 25 become a common tool for many bridge agencies worldwide. However, in current asset management practice, there still remain some fundamental shortcomings associated with the health status of a bridge for long-term planning of asset management strategies. Reliable long-term forecasting of bridge performance is crucial and can be used as input information for various key functions in a BMS, i.e. 30 cost-related and MR&R as a priority etc. However, the reliability of currently available long-term performance modelling is still doubtful thus, it requires further development. Some underlying problems exist with respect to the development of a long term performance/deterioration model. They are elaborated below:

(1) Insufficient historical condition ratings: The deterioration rate is calculated based on historical condition ratings obtained from routine bridge inspections, i.e. the structural component-level bridge inspections. Commercial BMS software has

2020100769 18 May 2020 only been used for less than 20 years and even those bridge agencies that have implemented BMSs from an early stage, have only approximately 7 to 9 inspection records per structure. Although most bridge authorities have previously conducted inspections, these past inspection records are incompatible as input with what is 5 required by a typical BMS. Such incompatibility is one of the causes for the deficiency of the current BMS outcomes. As a result of limited bridge condition rating records, it is very difficult to use typical stochastic-based deterioration models to accurately predict future condition ratings.

(2) Overall Condition Rating (OCR) methods: OCR methods are used in most existing bridge management technologies. The condition rating information is collected via a quantitative bridge inspection procedure. It is then converted into OCR in a subjective manner. The conditions of bridge components collected using the component-level bridge inspection process are expressed quantitatively via the 15 conventional grading system, i.e. the health index or the four condition states (CSs 1 to 4). The overall condition of one or more component types of a bridge is calculated with the aid of a weighted average condition state (CS) numbering system. Thus, the OCR is incapable of capturing the condition status of individual structural components (i.e. individual beams, piers etc), be it at CSI (i.e. condition as new or excellent), CS2 (fair), CS3 (poor) or CS4 (very poor). This is a key drawback because a bridge may collapse as a result of the failure of a single component. In view of this, each of the four CSs for individual components needs to be evaluated in order to reduce the risks of total bridge failure. A further drawback with this stepwise grading system is that there are only four CSs with graduation 25 of 1/4 or 25%. Such a step is too large to be used effectively in deterioration modelling. This indeterminacy seriously increases the degree of uncertainty of time in predicting long-term bridge component performance. Note also that it is too expensive to change the current inspection method, which has been used for many years and already produced massive amounts of historical condition rating records.

Any change to the current inspection method will be cost ineffective and will also create data-incompatibility issues.

Among many existing research outcomes, stochastic bridge deterioration modelling is one of the most prominent techniques. It can be classified into two types, state35 based and time-based modelling. Some limitations in state-based modelling are:

(1) initial condition ratings are independent from the historical condition ratings (a stationary process that is memory-less) and transition probabilities are constant;

2020100769 18 May 2020 (2) a lack of knowledge of the hidden nature of deterioration; (3) failing to account for maintenance issues; and (4) it only handles an ideal condition rating data distribution. On the other hand, time-based modelling overcomes many of the disadvantages of state-based modelling: (1) it considers the time spent in an initial 5 condition state - meaning that it overcomes the limitations of the stationary process; and (2) it provides more reliable long-term prediction than the state-based model if the condition rating data is available over a long period of time. However, the decision of which type of modelling is more appropriate for deterioration prediction is also highly dependent on the nature of the available condition ratings 10 coupled with bridge life spans. In other words, the stochastic approaches cannot guarantee workable modelling and/or reliable long-term prediction for various situations of condition rating input. Consequently, such fundamental problems as modelling input requirements still remain and need to be overcome.

In this specification where reference has been made to patent specifications, other external documents, or other sources of information, this is generally for the purpose of providing a context for discussing the features of the invention. Unless specifically stated otherwise, reference to such external documents is not to be construed as an admission that such documents, or such sources of information, in 20 any jurisdiction, are prior art, or form part of the common general knowledge in the art.

Embodiments of the present invention to provide an improved deterioration modelling system and/or method for a civil infrastructure asset management 25 system, or to at least provide the public with a useful choice.

SUMMARY OF THE INVENTION

In one aspect, the invention may broadly be said to consist of a system for monitoring the condition of one or more civil structures and forecasting, using one 30 or more processors, the deterioration of each civil structure, the system comprising:

at least one computer readable medium for storing historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more civil structures, and for 35 storing environmental parameters,

2020100769 18 May 2020 at least one processor arranged to process the condition state data for each component and the environmental parameters to predict information indicative of future performance-related condition states associated with the component.

Preferably the at least one processor is configured to process the condition state data for each component to predict a deterioration profile indicative of future performance-related condition states over time.

Preferably the at least one processor is configured to execute a machine learning 10 model to process the condition state data and predict the deterioration profile for each component.

Preferably when there is insufficient condition state data to determine the deterioration profile information for a particular component, the at least one 15 processor is configured to:

determine one or more similar components to the particular component;

acquire from the at least one readable medium data indicative of one or more performance-related condition states obtained over time for the one or more similar components; and process the condition state data associated with the similar components to determine the deterioration profile information for the particular component.

Preferably the system comprises a categorisation function comprising categorisation software executable by the at least one processor to categorise in the at least one 25 computer-readable medium multiple components into groups based on same or similar values of one or more categorisation parameters associated with the components.

Preferably the categorisation parameters include at least a component type and/or 30 a material-type parameter. Preferably the categorisation parameters further include any one or more of: a network identification parameter; a structure identification parameter; an associated traffic volume parameter; and/or an associated environmental condition parameter.

Preferably the step of determining one or more similar components to the particular component comprises identifying one or more similar components in the same group as the particular component as categorised by the categorisation function.

2020100769 18 May 2020

Preferably the condition state data includes one or more values of a condition state parameter obtained over time.

In a preferred embodiment, the system further comprises a condition rating module comprising software executable by the at least one processor to determine, using the condition state data of one or more components of a structure, output information relating to any one or more of:

• an average condition state for each of one or more groups of same or similar 10 components within the structure;

• an average condition state of the structure;

• a unit cost for each of one or more components within the structure as affected by condition of the one or more components;

• an asset value of the structure as affected by condition of the one or more 15 components;

• a risk value for one or more components within the structure associated with failure of the one or more components; and/or • a risk value associated with failure of the structure.

Preferably the condition rating module is configured to determine, using the condition state data of the one or more components of the structure, a value or values for one or more of the following output condition-rating parameters:

• An overall condition rating (OCR.) parameter indicative the average condition state for one or more components within the structure;

· A structure health index parameter indicative of the average condition of the structure;

• A component value parameter indicative of a unit cost of one or more components in the structure as affected by condition;

• A structure value parameter indicative of the asset value of the structure as 30 affected by the condition of components within the structure;

• A component risk score parameter indicative of the risk associated with one or more components within the structure; and/or • A structure risk score parameter indicative of the risk associated with failure of the structure.

2020100769 18 May 2020

Preferably the at least one processor is configured to utilise the condition rating module to predict one or more future values of one or more of the output conditionrating parameters using the predicted future condition state information.

In a preferred embodiment, the system further comprises a maintenance module comprising software executable by the at least one processor to determine, using the condition state data and/or the output information of the condition rating module, a list of components of a structure to be repaired or replaced depending on financial budget constraints.

to

Preferably the maintenance module is configured to determine the list of components to be repaired or replaced by:

• prioritising the maintenance of components of a structure based on the component risk score associated with the components from t5 highest priority to lowest priority;

• accumulating a cost of repairing or replacing each component from highest priority to lowest priority until the accumulated cost is equal or similar to a predetermined budget value; and • extracting and outputting a list of prioritised components within the 20 predetermined budget value.

Preferably the maintenance module is further configured to predict future maintenance costs for one or more future time instances using the predicted future condition state data.

Preferably the maintenance module is further configured to output information indicative of future maintenance costs based on two or more different predetermined maintenance actions for the list of prioritised components to be repaired or replaced.

Preferably the method further comprises using the maintenance module to generate future maintenance scenarios for one or more future time instances using the predicted future condition state data.

In a preferred embodiment the system further comprises a budget planning module comprising software executable by the at least one processor to determine a cost of maintaining a desired health index level of a structure or a network of two or more

2020100769 18 May 2020 structures based on the condition state information, the predicted future condition state information and/or the output condition rating information of the condition rating module.

Preferably the budget planning module is configured to determine a target network level performance based on the future health index from the deterioration module.

Preferably the budget planning module is configured to determine a cost of maintenance required at a current and/or future time instance to achieve the 10 desired health index level by comparing a future network level performance obtained using the deterioration module, with the target network level performance.

Preferably the budget planning module is configured to output information 15 indicative of budget requirements for achieving one or more desired health index levels, e.g. network level performance, over a period of time in the future.

In a second aspect the invention may broadly be said to consist of a method for monitoring the condition of one or more civil structures and forecasting, using one 20 or more processors, the deterioration of each civil structure, the method comprising the steps of:

obtaining, from at least one computer readable medium, historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more civil structures and 25 environmental parameters, and predicting, using at least one processor arranged to process the condition state data for each component, information indicative of future performance-related condition states associated with the component and the environmental parameters.

Preferably the step of predicting includes processing the condition state data for each component to predict a deterioration profile indicative of future performancerelated condition states overtime.

Preferably the step of predicting includes executing a machine learning model to 35 process the condition state data and predict the deterioration profile for each component.

2020100769 18 May 2020

Preferably the method further comprises, prior to predicting the future condition state information for a component:

determining one or more similar components to the component;

acquiring from the at least one readable medium data indicative of one or 5 more performance-related condition states obtained over time for the one or more similar components; and processing the condition state data associated with the similar components to determine the deterioration profile information for the component.

Preferably the prior to predicting the future condition state information, the method comprises the step of categorising in the at least one computer-readable medium multiple components into groups based on same or similar values of one or more categorisation parameters associated with the components.

Preferably the categorisation parameters include at least a component type and/or a material-type parameter. Preferably the categorisation parameters further include any one or more of: a network identification parameter; a structure identification parameter; an associated traffic volume parameter; and/or an associated environmental condition parameter.

Preferably the step of determining one or more similar components to the particular component comprises identifying one or more similar components in the same group as the particular component as categorised by the categorisation process.

In a preferred embodiment, the method further comprises utilising a condition rating module comprising software executable by the at least one processor to 30 determine, using the condition state data of one or more components of a structure, output condition rating information relating to any one or more of:

• an average condition state for each of one or more groups of same or similar components within the structure;

• an average condition state of the structure;

«a unit cost for each of one or more components within the structure as affected by condition of the one or more components;

2020100769 18 May 2020 • an asset value of the structure as affected by condition of the one or more components;

• a risk value for one or more components within the structure associated with failure of the one or more components; and/or · A risk value associated with failure of the structure.

Preferably the output condition rating information relates to a value or values for one or more of the following output condition-rating parameters:

• An Overall Condition Rating (OCR) parameter indicative the average 10 condition state for a group of same or similar components within the structure;

• A structure health index parameter indicative of the average condition of the structure;

• A component value parameter indicative of a unit cost of one or more 15 components in the structure as affected by condition;

• A structure value parameter indicative of the asset value of the structure as affected by the condition of components within the structure;

• A component risk score parameter indicative of the risk associated with one or more components within the structure; and/or · A structure risk score parameter indicative of the risk associated with failure of the structure.

Preferably the method further comprises utilising the condition rating module to predict one or more future values of one or more of the output condition-rating 25 parameters using the predicted future condition state information.

In a preferred embodiment, the method further comprises utilising a maintenance module comprising software executable by the at least one processor to determine, using the condition state data and/or the output information of the condition rating 30 module, a list of components of a structure to be repaired or replaced depending on budget constraints.

· prioritising the maintenance of components of a structure based on the component risk information associated with the components from highest priority to lowest priority;

2020100769 18 May 2020 • accumulating a cost of repairing or replacing each component from highest priority to lowest priority until the accumulated cost is equal or similar to a predetermined budget value; and • extracting and outputting a list of prioritised components within the 5 predetermined budget value.

In a preferred embodiment the method further comprises utilising a budget planning module, comprising software executable by the at least one processor, to determine a cost of maintaining a desired health index level of a network of two or more structures based on the predicted future condition state information determined by the deterioration module.

Preferably the method comprises utilising the budget planning module to determine a target network level performance based on the desired network performance.

Preferably the method comprises utilising the budget planning module to determine 30 a cost of maintenance required at a current and/or future time instance to achieve the desired health index level by comparing a future network level performance obtained using the deterioration module, with the target network performance level.

Preferably the method comprises utilising the budget planning module to determine information indicative of budget requirements for achieving one or more desired health index levels over a period of time in the future.

2020100769 18 May 2020

In a third aspect the invention may broadly be said to consist of a civil infrastructure asset management system comprising at least one computer readable medium for storing one or more software applications and at least one processor for executing the software applications, the system comprising a performance deterioration module configured to predict deterioration of one or more components of one or more structures, the deterioration module being configured to:

obtain, from the at least one computer readable medium, historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more structures and environmental parameters, process, using the at least one processor, the condition state data for each component and the environmental parameters to predict information indicative of future performance-related condition states associated with the component.

Preferably the deterioration module is configured to process the condition state data for each component to predict a deterioration profile indicative of future performance-related condition states over time.

Preferably the deterioration module is configured to execute a machine learning model to process the condition state data and predict the deterioration profile for each component.

Preferably when there is insufficient condition state data to determine the deterioration profile information for a particular component, the deterioration module configured to:

determine one or more similar components to the particular component;

Preferably the deterioration module comprises categorisation software executable by the at least one processor to categorise in the at least one computer-readable

2020100769 18 May 2020 medium multiple components into groups based on same or similar values of one or more categorisation parameters associated with the components.

Preferably the categorisation parameters include at least a component type and/or 5 a material-type parameter. Preferably the categorisation parameters further include any one or more of: a network identification parameter; a structure identification parameter; an associated traffic volume parameter; and/or an associated environmental condition parameter.

Preferably the condition state data includes one or more values of a condition state 15 parameter obtained over time.

In a preferred embodiment, the system further comprises a condition rating module comprising software executable by the at least one processor to determine, using the condition state data of one or more components of a structure, output 20 information relating to any one or more of:

• an average condition state of the structure;

• an asset value of the structure as affected by condition of the one or more components;

• a risk value for one or more components within the structure associated with failure of the one or more components; and/or · A risk associated with failure of the structure.

· An overall condition rating (OCR.) parameter indicative the average condition state for a group of same or similar components within the structure;

2020100769 18 May 2020 • A structure health index parameter indicative of the average condition of the structure;

· A structure value parameter indicative of the asset value of the structure as affected by the condition of components within the structure;

• A component risk score parameter indicative of the risk associated with one or more components within the structure; and/or • A structure risk score parameter indicative of the risk associated with failure 10 of the structure.

In a preferred embodiment, the system further comprises a maintenance module comprising software executable by the at least one processor to determine, using the condition state data and/or the output information of the condition rating module, a list of components of a structure to be repaired or replaced depending on 20 budget constraints.

• prioritising the maintenance of components of a structure based on 25 the component risk information associated with the components from highest priority to lowest priority;

• accumulating a cost of repairing or replacing each component from highest priority to lowest priority until the accumulated cost is equal or similar to a predetermined budget value; and · extracting and outputting a list of prioritised components within the predetermined budget value.

Preferably the method further comprises using the maintenance module to predict future maintenance costs for one or more future time instances using the predicted 35 future condition state data.

2020100769 18 May 2020

In a preferred embodiment the system further comprises a budget planning module comprising software executable by the at least one processor to determine a cost of maintaining a desired health index level of a structure or a network of two or more structures based on the condition state information, the predicted future condition 10 state information and/or the output condition rating information of the condition rating module.

Preferably the budget planning module is configured to determine a target structure performance based on the desired network level performance.

Preferably the budget planning module is configured to determine a cost of maintenance required at a current and/or future time instance to achieve the desired health index level by comparing a future network level performance obtained using the deterioration module, with the target network level 20 performance.

Preferably the budget planning module is configured to output information indicative of budget requirements for achieving one or more desired health index levels over a period of time in the future.

In one embodiment the system further comprises one or more servers including the at least one computer readable medium and the at least one processor for executing the software applications and a communication interface for communicating via a communications network with one or more software 30 applications executable on one or more field operator devices.

Preferably the one or more servers are configured to communicate inspection and/or management information to the software applications executable on the field operator devices.

Preferably the one or more servers are configured to receive data indicative of one or more performance-related condition states for one or more components of the

2020100769 18 May 2020 one or more structures from the software applications executable on the one or more field operator devices.

Any one or more of the above embodiments or preferred features can be combined 5 with any one or more of the above aspects.

The term comprising as used in this specification means consisting at least in part of. When interpreting each statement in this specification that includes the term comprising, features other than that or those prefaced by the term may also 10 be present. Related terms such as comprise and comprises are to be interpreted in the same manner.

Number Ranges

It is intended that reference to a range of numbers disclosed herein (for example, 1 15 to 10) also incorporates reference to all rational numbers within that range (for example, 1, 1.1, 2, 3, 3.9, 4, 5, 6, 6.5, 7, 8, 9 and 10) and also any range of rational numbers within that range (for example, 2 to 8, 1.5 to 5.5 and 3.1 to 4.7) and, therefore, all sub-ranges of all ranges expressly disclosed herein are hereby expressly disclosed. These are only examples of what is specifically intended and 20 all possible combinations of numerical values between the lowest value and the highest value enumerated are to be considered to be expressly stated in this application in a similar manner.

As used herein the term and/or means and or or, or both.

As used herein (s) following a noun means the plural and/or singular forms of the noun.

This invention may also be said broadly to consist in the parts, elements and 30 features referred to or indicated in the specification of the application, individually or collectively, and any or all combinations of any two or more said parts, elements or features, and where specific integers are mentioned herein which have known equivalents in the art to which this invention relates, such known equivalents are deemed to be incorporated herein as if individually set forth.

The invention consists in the foregoing and also envisages constructions of which the following gives examples only.

2020100769 18 May 2020

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred embodiments of the invention will be described by way of example only and with reference to the drawings, in which:

Figures la and lb are schematics showing a preferred form infrastructure management system of an embodiment of the invention and the associated modules of the system;

Figure 2 is a diagram showing the functions or stages of a preferred form deterioration modelling module of an embodiment of the invention;

Figure 3 is a flowchart showing the stages or functions of categorisation and support vector regression in the deterioration module of figure 2;

Figure 4 is a diagram showing the functions or stages of a preferred form maintenance planning module of an embodiment of the invention;

Figure 5 is a graph of accumulated maintenance cost vs. maintenance funding, 20 showing the process of determining maintenance cut-off in the maintenance planning module of figure 4;

Figure 6 is a process diagram showing the stages involved in preferred form costbenefit analysis function of the maintenance planning module of figure 4;

Figure 7 is a graph of Network Health Index vs. Budget, showing an output of a preferred form budget planning module of an embodiment of the invention;

Figure 8a is a performance based network level budget planning graph as 30 determined by a performance-based funding estimation function of the budget planning module of figure 7;

Figure 8b is an exemplary graph of time vs. maintenance budget requirement;

Figure 9a is an exemplary graph showing the process of long-term prediction without the categorisation function of the deterioration module of figure 2;

2020100769 18 May 2020

Figure 9b is an exemplary graph showing the process of long-term prediction with the categorisation function of the deterioration module of figure 2; and

Figure 10 is a schematic showing communication between the preferred form 5 infrastructure management system of an embodiment of the invention and users of this embodiment;

Figure 11 is an example graph of OCR. data when Quantity is 1. (Bridge ID: 89850, Element ID: 311);

Figure 12 is an example graph of OCR Data where Quantity is 23.77. (Bridge ID:

82815, Element ID: 102); and

Figure 3 shows example deterioration curves. (Yellow: a=0.1, c=0.001. Blue: 15 a = 0.001, c=0.1).

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

In the following description, specific details are given to provide a thorough understanding of the embodiments. However, it will be understood by one of 20 ordinary skill in the art that the embodiments may be practiced without these specific details. For example, software modules, functions, circuits, algorithms etc., may be shown in block diagrams in order not to obscure the embodiments in unnecessary detail. In other instances, well-known modules, structures and techniques may not be shown in detail in order not to obscure the embodiments.

Also, it is noted that the embodiments may be described as a process that is depicted as a flowchart, a flow diagram, a structure diagram, or a block diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be rearranged. A process is terminated when its operations 30 are completed. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc., in a computer program. When a process corresponds to a function, its termination corresponds to a return of the function to the calling function or a main function.

2020100769 18 May 2020

Aspects of the systems and methods described below may be operable on any type of general purpose computer system or computing device, including, but not limited to, a desktop, laptop, notebook, tablet or mobile device. The term mobile device includes, but is not limited to, a wireless device, a mobile phone, a mobile 5 communication device, a user communication device, personal digital assistant, mobile hand-held computer, a laptop computer, an electronic book reader and reading devices capable of reading electronic contents and/or other types of mobile devices typically carried by individuals and/or having some form of communication capabilities (e.g., wireless, infrared, short-range radio, etc.).

Furthermore, embodiments may be implemented by hardware, software, firmware, middleware, microcode, or any combination thereof. When implemented in software, firmware, middleware or microcode, the program code or code segments to perform the necessary tasks may be stored in a machine-readable medium such as a storage medium or other storage(s). A processor may perform the necessary tasks. A code segment may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a class, or any combination of instructions, data structures, or program statements. A code segment may be coupled to another code segment by passing and/or receiving information, data, arguments, parameters, or memory contents. Information, 20 arguments, parameters, data, etc. may be passed, forwarded, or transmitted via any suitable means including memory sharing, message passing, token passing, network transmission, etc.

Referring to figures la and lb, an overview of a preferred form smart civil infrastructure asset management system and method 1000 of an embodiment of 25 the invention is shown. The system 1000 is implemented using various sub-systems and/or modules including, but not limited to: a condition rating module 1100; a deterioration modelling module 1200; a maintenance planning module 1300 and a budget planning module 1400. The operation and purpose of each module and how its inputs and outputs relate to other modules of the system will be described in 30 further detail below.

The management system 1000 comprises at least one computer readable medium for storing information relating to the inputs and outputs of the various modules, as well as for storing software associated with each module. In the preferred embodiment, the at least one computer readable medium includes at least a

2020100769 18 May 2020 database 1500 storing data relating to one or more civil infrastructure assets. Referring to figure lb, this data may include inventory information (for example the location, the specification of a structure/asset, the related road and traffic conditions etc), cost related data including for example unit cost of each component 5 of each structure/asset and/or unit repair cost, condition data including for example the condition of one or more components of at least one civil infrastructure asset and/or data relating to the overall condition of at least one civil infrastructure asset, and/or analysis data including for example computed deterioration rates for each component in a structure/asset. As will be described in further detail below, this 10 data is utilized by one or more modules to determine information that is useful for planning and management of the civil infrastructure assets including for example: deterioration profile information, maintenance planning information and budget planning information. The management system 1000 further comprises at least one processor arranged to execute one or more software programs stored in the 15 computer readable medium and configured to process data stored in the database and/or other input data to obtain the relevant outputs of the various modules accordingly.

In the preferred embodiment, the infrastructure asset management system 1000 20 and its various sub-systems/modules are implemented as a Bridge Management System (BMS) for determining the current and future conditions of bridges and for planning the maintenance and allocation of budgets associated with those bridges. It will be appreciated however, that the principles of operation as will be apparent to those skilled in the art can be applied to other infrastructure assets and the 25 invention is not intended to be limited to this preferred implementation.

The condition rating 1100 module is configured to process an overall condition rating (OCR) function 1110 and a risk scoring function 1120 to help determine the condition of a bridge and its components in a network and the risk factor associated 30 with the bridge and its components in their current conditions. The parameters output from these functions are input into the maintenance planning 1300 and budget planning 1400 modules.

The deterioration modelling module 1200 is configured to process similar OCR and 35 risk scoring functions to the condition rating module but with the aim of predicting future conditions states of a bridge and its components and associated condition

2020100769 18 May 2020 risks. The parameters output from these functions are also input into the maintenance planning 1300 and budget planning 1400 modules.

The maintenance planning module 1300 is configured to assess the current and 5 future states and risk scores of multiple bridges in a network and their associated components to help prioritize and plan maintenance of the network. It utilizes functions including the maintenance prioritization function 1310, the maintenance options function 1320, the cost-benefit analysis function 1330 and maintenance effects function 1340 to achieve this.

The budget planning module 1400 is configured to assess the outputs of the condition rating 1100, deterioration modelling 1200 and maintenance planning 1300 modules to manage financial budgets and forecast required long term budget for achieving a target health index.

All modules 1100-1400 utilise a system database 1500 to store parameter/variable values 1510 that may be pre-determined, pre-set or dynamically changing to enable the necessary calculation and computation as will be defined in further detail below. The database 1500 preferably also stores information relating to any 20 combination of one or more of:

• bridge/asset inventory - structure details and present/past information and analysis results;

• condition ratings - historical structure components' condition ratings;

• maintenance records - historical structure components' maintenance 25 records;

• Geographic Information System (GIS) Information - electronic graphical map linked to actual structure information;

• Traffic - past and present traffic conditions associated with structures;

• Cost Accounting - historical expenditures and all component material unit 30 costs for maintenance, repair and rehabilitation activities; and • Analysis outcomes - all outcomes from analytical modules including deterioration module, maintenance planning module and/or budget planning module.

In a preferred embodiment the system 1000 also comprises a GIS based program that displays on a user's device a map having various structures displayed throughout based on their location. The structures are colour coded on the map

2020100769 18 May 2020 depending on their current health index (health index described in detail further below). Each 'locator pin' on the map provides a condition rating/health index, as well as the ability to access detailed structure information on the user's device via an appropriate link.

Referring to figure 10, in the preferred embodiment the civil infrastructure asset management system 1000 is implemented using a client-server distributed application structure 2000. This structure provides user applications, accessible via user devices, such as desktop computers 2110 and/or mobile devices 2120 that 10 enable information to be shared between users and the management system 1000.

A central server 2200 hosts the various modules 1100-1400 and the database 1500 of the management system 1000 and communicates with the user devices 2110, 2120 via an appropriate communication system 2300.

1 Condition Rating Module

The condition rating module 1100 is configured to calculate the condition rating of bridge components to help determine and/or quantify the deterioration or deterioration profile of a structure/asset. The condition rating of each type of component in a structure is the primary factor used in other core BMS modules to 20 classify structural deficiency. The condition rating module is operable to receive as an input, for each of one or more structures, condition status/rating data indicative of the condition/performance status of one or more components of the structure. In this specification reference a component of a structure is intended to include distinct components or elements and groups of related components or elements in 25 the structure. The condition rating module 1100 is operable to process the condition rating data and determine the value(s) of one or more component-level output parameters and/or one or more structure-level output parameters indicative of the condition of the structure.

In the preferred embodiment, the component-level output parameters include any one or more of:

• The overall condition rating (OCR) indicative of the overall condition of a plurality of same/similar components in the structure;

• The component risk score indicative of the risk associated with one or more 35 same/similar components in the structure; and • The current component value (CCV) indicative of the value of one or more same/similar components based on their condition.

2020100769 18 May 2020

In the preferred embodiment the component-level output parameters includes the OCR., the average component risk score and the CCV.

In the preferred embodiment, the structure-level output parameters include any 5 one or more of:

• The structure health index indicative of the overall condition of the structure;

• The structure risk score indicative of the risk associated with the structure; and · The current asset value (CAV) indicative of the value of the structure based on its condition.

In the preferred embodiment, the structure-level output parameters include the structure health index, the structure risk score and the CAV.

As mentioned above, the input to the condition rating module 1100 is condition rating data. Condition rating data indicative of the condition of one or more components of a structure (e.g. bridge structure) is acquired in the preferred embodiment during a data acquisition stage from routine bridge inspections (condition assessment). During the data acquisition stage, inspectors or other authorised and qualified personnel inspect the various components in a structure to determine the conditions of those components. The components that are analysed are preferably the structural components of a structure. For example, in the embodiment of a bridge management system, the components may include any one or more of abutments, decks, piers, headstocks, bearings, joins, guardrails etc.

The condition is quantified and given a value indicative of a component's current condition. The value is preferably indicative of a status relative to a new or unused condition. The value may be numerical or symbolic. The value may be selected from a predetermined number of discrete values, each indicating a different condition, or it may be calculated using a predetermined equation or similar. In the preferred embodiment, the condition rating value of a structural component is acquired using a number of predetermined discrete condition statuses. Table 1 outlines the preferred predetermined values of the condition status/rating variable; however, it will be appreciated that other discrete ratings may be implemented. For example, there may be a higher or lower number of discrete values to heighten or 35 lower the resolution.

2020100769 18 May 2020

Condition Status Value	Indicative Condition
CSI	New, near new or excellent condition
CS2	Fair working condition
CS3	Poor working condition
CS4	Very poor working condition

Table 1: Structural Component Condition Statuses

Condition status information relating to the structural components of a particular structure (e.g. the structural components of a bridge) are input and stored in a 5 database 1500 associated with the condition rating module 1100. The date of acquisition of the data is also stored in memory to provide a historical collection of condition data for each structural component over time.

1.1 Overall Condition Rating and Structure Health Index

Bridge agencies handle enormous amounts of bridge-related data. Even with the 10 best modelling and computer programmes, the outcomes must be clearly communicated to funding agencies and top-level decision makers in a manner that is simple and easy to understand. The condition rating module 1100 of the preferred embodiment is configured to execute an overall condition rating (OCR.) function 1110 to calculate the average condition rating of a plurality of the 15 same/similar components which can be used in the classification of structural deficiencies.

In the preferred embodiment, the OCR of various types of component groups is utilised by the condition rating module 1100 to determine the value of a structure 20 health index (and in this particular example the Bridge Health Index (BHI)) variable as a performance indicator of the structural health of the asset. The BHI variable enables decision makers to easily comprehend and compare the conditions of various bridges in the network. The OCR and BHI variables may have values ranging from zero to 100, with an OCR or BHI of 100 representing a new condition 25 and an OCR or BHI of zero representing the worst condition. It will be appreciated that in other implementations, the OCR or BHI variables can be expressed using other scales and/or numerical ranges to represent relative conditions between the associated components/structures. For example, the OCR or BHI may be represented by a scale of zero to ten, and/or it may be represented with zero 30 indicating a new component/bridge for instance. The OCR and BHI variables are preferably represented numerically.

2020100769 18 May 2020

The BHI is calculated using information relating to the condition of one or more bridge components. As previously described, during a data acquisition stage condition status values are assigned to the various bridge components. This is used to determine the OCR. and in turn, the BHI. The BHI is based on the financial value 5 of the structure and the total replacement value of the structure. The financial value, i.e. asset value, of the structure is dependent on the OCR of the various component groups and the total replacement value of each component group.

Original asset values (new/unused values) are assigned to the various bridge 10 components (by bridge agencies for example) and input into the system database

1500 in an initial set up stage of the system. This will help determine the current (financial) values of components to thereby determine BHI.

In the preferred embodiment, the value of the BHI variable is calculated by the 15 condition rating module 1100 as follows:

BHI = (Σ CCV / Σ TRV) x 100 (1) where TRV means total component replacement value and CCV means current component value. They are calculated as follows:

Total Component Replacement Value (TRV) = Total Component Quantity x Unit 20 Replacement Cost.

The CCV of a component is dependent on its condition state. As the condition state deteriorates, the asset's value declines. Table 2 shows an exemplary, but preferred, representation of the value of a component as a fraction of its 'new asset' value for 25 various condition states, CSI to CS4. It will be appreciated other declining asset value factors may be assigned to various condition states depending on the application and the invention is not intended to be limited to those shown in Table 2. The CCV is expressed as follows by the condition rating module 1100:

Current Component Value (CCV) = Σ (Quantity Condition State i x WFi) x TRV

WFi is the condition state weighting factor for state T as presented in Table 2.

Condition State (i)	CSI	CS2	CS3	CS4
WFi(Declining Asset Value	1	0.8	0.6	0.2

2020100769 18 May 2020

Factor)

Table 2: Declining Asset Value Factor (WFi)

The current asset value (CAV) is the value of the structure which is determined by aggregating the CCV of each component group in the structure. The ratio between 5 the CAV of a bridge and the asset value of a new bridge expressed as a percentage will be the health index (BHI) of that particular bridge. This identical process can be extended to all of the bridges in a network and an aggregated (or averaged) health index can be derived for that network. This will be expressed as the Network Health Index (NHI).

The following is an example of a BHI calculation for a bridge having multiple components. Condition status data (CSs) relating to multiple bridge components having various identifications (IDs) is analysed to determine OCR.. Predetermined information that is stored in memory, including unit replacement cost is then 15 utilised to determine the total component replacement value (based on the number of units and the cost of each unit) and the CCV for each component ID. The total component replacement values are aggregated to determine the total replacement value for the bridge. The CCV are also aggregated to determine the CAV of the entire bridge. The ratio between the CAV of the bridge and the total replacement 20 value for the bridge is then determined to obtain the BHI.

Example #1: BHI worksheet

ID	Total Qty.	Condition States (CSs)	OCR (%)	Unit Replaceme nt Cost ($)	Total Component Replaceme nt Value (TRV)($)	Current Componen t Value (CCV) ($)	Total Replaceme nt Value (ZTRV) ($)	Total Current Compone nt Value (ZCCV) ($)	BHI (%)
1	2	3	4
lc	10	5	1	2	2	74.00	2,500.00	25,000.00	18,500.00	417,500.00	324,000.0 0	77.6 0
2s	15	5	5	5	0	80.00	2,500.00	37,500.00	30,000.00
3p	20	10	5	5	0	85.00	2,500.00	50,000.00	42,500.00
4s	5	1	1	1	2	56.00	2,500.00	12,500.00	7,000.00
12o	30	5	10	10	5	66.67	2,500.00	75,000.00	50,000.00

2020100769 18 May 2020

31C	20	10	0	5	5	70.00	2,500.00	50,000.00	35,000.00
51P	22	12	5	5	0	86.36	2,500.00	55,000.00	47,500.00
59C	32	15	15	2	0	88.13	2,500.00	80,000.00	70,500.00
71c	10	3	3	3	1	74.00	2,500.00	25,000.00	18,500.00
84p	3	0	0	3	0	60.00	2,500.00	7,500.00	4,500.00

„„„ Λνΐ (csixWFii+icszxWF^+Ccsaxw^+Ccs+xHT^_nn

ULK \ zO/ — Λ 1UU =

Total Quantity or Component (5xi)+(lxa.8} + (2x0L6)+(2xa.i „„

-------------------= 74

TRV($) = Unit Replacement Cost ($) X Total Quantity of Component = 2, 500 X 10 = 25,000

CCV(i) = TRV (ί) x OCR = 25, 000 x 74/100 = 18, 500

Total Current Component Value (YCCVJ ($J 324.000

BHl(%) =---------------------------,--- , ---x 100 =-------x 100 = 77.60

Total Replacement Value(£TRV) ($) 417,500

1.2 Risk Scoring

The risk scoring function 1120 considers the significance of components when calculating the health of a structure. The results are values of risk score variables at component and structure levels which can be used to prioritise stocks for listing required maintenance work. The risk scoring function 1120 is configured to 15 determine an average risk score value of a component/group of components and an average risk score value of a structure based on the condition status values of the various components in the structure and other structure related variables.

The factors that influence the risk score may include, for example:

· The health of bridge components and component significance rating;

• the traffic volume of commercial (heavy) vehicles and road significance; and/or • the environmental condition and asset value of a structure.

The risk scoring function can be varied over time to improve the output of results. In a preferred embodiment of the invention, the existing Department of Transport

2020100769 18 May 2020 and Main Roads (TMR) risk-based asset management processes are implemented into the risk scoring function 1120 of the condition rating module 1100 to. It will be appreciated that in alternative embodiments other existing or predefined processes may be implemented to determine risk scores and the invention is not intended to 5 be limited to the following exemplary implementation.

In the preferred embodiment, the condition rating module 1100 calculates the values of risk score variables as follows:

RISK = Probability of Failure x Consequence of Failure (2)

Component Risk =Probability of Component Failure(PCF) x Consequence of (3) Failure(COF)

Structure Risk = Probability of Structure Failure (PSF) x Consequence of Failure (COF) (4) to

Probability of Component Failure (PCF) = LF x SF x CF x IF x XF

where,	Loading Factor (LF)	(e.g.	LF = 1 to
		24);
	Resistance Factor (SF)	(e.g. 16);	SF = 1 to
	Condition Factor (CF)	(e.g. 8);	CF = 1 to
	Inspection Factor (IF)	(e.g. 2);	IF = 1 to
	Exposure Factor (XF)	(e.g. to 2).	XF = 1 (5)

As previously mentioned, the ranges given for the above factors are only exemplary and other ranges and scales may be used for each factor depending on the particular application.

y PCF

Probability of Group Failure (PGF) = (1 + MDFp) X------------------------------Number of component in the Group

MDF = 1 +

MDFp (6)

F_c = 1.84

2020100769 18 May 2020

PGF . . ^(1+MDFp) MDF z_7>

Probability of Structure Failure (PSF) = ------------------------------x---- < ')

Number of group m a structure F_c

Consequence of Failure (COF) = HC + EC + TC + NC + RS + AC where, Human Consequence (HC) (HC = 1 to 8); Environmental Consequence (EC) (EC = 0 to 2); Traffic Access Consequence (TC) (TC = 1 to 5); Economic Consequence (NC) (NC = 1 to 5); Road Significance (RS) (RS = 1 to 6); Local Industry Access Consequence (AC) (AC = 0 or 1.5).
Example #2: Risk Score worksheet • structure Id • Bridge design class • Standard component ID • Classification • Annual Average Daily Traffic (Commercial only) • Structure Replacement Costs • Road Class • Inspection Factor • Local industry access consequence • Pedestrian bridge	vehicle	xxxx BM 150, 2C, 20P, 31C, 21C, 54C 2 600 242800 Motorway/Freeway 1 0 No

ID	Total Quantity	CS1	CS2	CS3	CS4	COF	PCF	MDFp	MDF	PGF	PSF	Average Component Risk	Average Structure Risk
150	25.20	0.00	8.40	8.40	8.40	13.00	631.68	5.21	6.21	11891.37	6462.70	8211.84	84015.10
2C	72.00	0.00	72.00	0.00	0.00	90.24	1173.12
20P	34.00	0.00	20.00	14.00	0.00	573.29	7452.76
31C	12.00	0.00	2.00	7.00	3.00	1895.04	24635.52
21C	8.00	0.00	1.00	1.00	6.00	4872.96	63348.48
54C	4.00	1.00	1.00	0.00	2.00	3429.12	44578.56

The condition rating module 1100 is therefore configured to receive condition status data indicative of the status of one or more components of a structure (such a

2020100769 18 May 2020 bridge) to determine the value of a health index variable (e.g. BHI) of that asset and the a value of an average risk score variable associated with the asset. In the preferred embodiment, the condition rating module 1100 utilises predefined and/or existing asset management processes for determining the average risk score value.

The health index and risk score values for each asset in the network are then stored in memory for further analysis by the maintenance and budget planning modules of the system.

As will be described further below, the deterioration module 1200 also utilises the condition rating module 1100 to determine future structure health index and future 10 average risk score values based on predicted future condition status values.

2. Deterioration Module

The deterioration module 1200 of the infrastructure management system for bridges 1000 has advanced functions to produce reliable and meaningful results of long-term performance of bridge components. Deterioration modelling in any 15 Bridge Management System (BMS) is important for analysing future bridge needs, including maintenance planning and budget forecasting as will be described in further detail below. Without this information, the software is unable to analyse essential future bridge needs for bridge owners.

The deterioration module 1200 is configured to receive condition status data for one or more components of a structure, and predict based on this data, and optionally additional information, the future condition statuses of the one or more components. In this manner, predicted future condition statuses can be obtained for various time instances in the future and the condition rating module 1100 can 25 be utilised to predict other important information for these future time instances, including for example future OCRs, future BHIs, future component average risk scores, future structure average risk scores and future component and structure values.

In the preferred embodiment, the deterioration module 1200 utilises a machine learning method called Support Vector Regression (SVR) to achieve long-term performance output data. This technique has the ability to model the deterioration process of each component of every bridge in a network. It is robust and much faster than Artificial Neural Networks (ANNs). It will be appreciated however that 35 other machine learning methods may be used in alternative embodiments if need

2020100769 18 May 2020 be to meet particular design requirements. The SVR technique is particularly suited for dealing with large condition rating data simultaneously.

The major stages/functions of the deterioration modelling in the BMS 1000 are 5 presented in Figure 2. The figure includes the stages/functions of preparing input condition rating data 1210, categorisation of bridge component inspection 1220 records, calculation of component OCRs 1230 and SVR prediction 1240.

2.1 Data Gathering/Preparation

The first stage of deterioration modelling is data preparation for available condition rating data 1210. During this stage, the deterioration module 1210 obtains and/or receives data indicative of current and/or past condition states of one or more components of a bridge and/or of multiple bridges in a network. This data may be received via input by an operator, or may be obtained from a database 1500 in memory associated with the deterioration and/or condition rating module 1100 for example. In the preferred embodiment, a minimum of two cycles of inspection data per bridge is recommended for appropriate long-term prediction modelling. It will be appreciated however that any number of two or more cycles may be utilised, with the higher number of cycles improving the accuracy of the output of the prediction stage 1240. Therefore, the data preparation stage 1210 comprises, for one or more structures in a network, the acquisition (either from memory or via collected by an operator) of condition state data indicative of the value of the condition status variable of one or more components of the structure.

In the case of insufficient condition rating data availability, a Backward Prediction Model (BPM) as described in A Methodology for Developing Bridge Condition Rating Models Based on Limited Inspection Records by Lee, JH (2007), the entire contents of which are incorporated herein by reference, can be optionally utilised 1211 to increase the amount of condition status data. It helps to significantly improve the reliability of the long-term prediction results. The use of this model provides a prediction of historical condition status values for one or more components of a structure based on one or more current or recently acquired values.

Alternatively, or in addition, if insufficient condition status data is available for long 35 term performance prediction, a categorisation process 1220 may be utilised to estimate previous/past condition status values for a component to improve/enhance long term prediction.

2020100769 18 May 2020

2.2 Categorisation

In the preferred embodiment, the deterioration module is configured to execute a categorisation process 1220 when it decides there is insufficient condition status 5 data for a particular component. The categorisation process 1220 is configured to identify components that are the similar to the particular component based on one or more factors. This identification is then used to extract current or past condition status information of those similar components and supplement the condition status information of the particular component. This improves the SVR prediction function 10 and enhances long term performance prediction for the particular component. The categorisation process provides supplementary condition status data using components that have similar situations to reflect their prior experienced deterioration pattern into the machine learning process via the SVR. This is to compute reliable long-term performance prediction outcomes. The categorisation 15 process takes into consideration factors such as environmental conditions, bridge and material types, structure location, construction era and/or traffic volumes associated with the component or structure. It will be appreciated that other factors directly or indirectly affecting the condition of components of a structure may also be considered by the categorisation process and the invention is not intended to be 20 limited to the examples given herein. The network-level categorisation process 1220 provides additional input data (condition status values/ratings) for the SVR training process.

In the preferred embodiment, the deterioration module identifies similar 25 components based on one or more parameters that affect (either directly or indirectly) the performance and/or condition of a component of the associated structure. For example, the categorisation process identifies similar components that have similar values for location, construction era, component type and/or material type parameters. This is achieved by comparing the value of each 30 parameter with a threshold value or threshold criteria representative of that component.

Referring to figure 3, the deterioration module 1200 utilises the categorisation process to obtain condition status data of similar bridge components during one or 35 more previous time instances when there is insufficient data for a particular bridge component (step 1221). For example, if the deterioration module identifies there is insufficient data to create a long term performance prediction of a particular

2020100769 18 May 2020 component, the categorisation process (step 1222) will be initiated. Categorisation is used for long term performance prediction, and the accuracy of the prediction is dependent on the quality and quantity of the input data. During this process, the module 1200 will first identify the bridge network associated with that component 5 (step 1222a), then identify the location of the bridge associated with that component (step 1222b), then the material type of the component (step 1222c) and finally the categorisation process may consider/identify the associated traffic conditions to which the component is normally exposed (step 1222d). It will be appreciated that only some of these factors may be considered or other factors may 10 be considered either alternatively or in combination. In the preferred embodiment factors 1222a-1222c are of particular importance and will be considered with factor 1222d being optional. From this identification, the categorisation function 1220 can search for other components having values that meet certain threshold criteria. For example, the categorisation function 1220 may look for only components in the 15 same network, within certain proximity of the location of the associated bridge having the same or similar material type, and/or having the same or similar traffic conditions. It will be appreciated that other parameter values and associated threshold criteria may be utilised by the categorisation function to identify similar components and the invention is not intended to be limited to this particular 20 example. Once these similar components have been identified, condition state data associated with those components (past and/or current condition state data) will be extracted from database 1500 and input into the prediction module for the currently analysed component. This process helps provide additional common deterioration pattern information as an SVR training input so that the SVR can 25 achieve better and reliable long-term performance prediction of each bridge component (steps 1231-1232).

As an example, when condition data for deterioration modelling has only a limited amount of inspection records available and no deterioration detected as shown in 30 Figure 9a:

• Bridge age: 34 years old; and • Available identical condition rating data: 2 cycles of inspection records (year 2010 & 2014) without condition rating changes.

As shown in Figure 9b, the categorisation process provides supplementary condition status data using bridge components that have similar situations to reflect their

2020100769 18 May 2020 prior experienced deterioration pattern into the machine learning process via the SVR. This is to enhance the computation of reliable long-term prediction outcomes.

2.3 Computation of Health Indices and Risk Scores

The SVR prediction function enables the prediction of future condition statuses of one or more components of a structure. This information can be input into the condition rating module 1100 to predict other future parameters including future OCR, future BHI and/or future component and structure risk scores for example.

Once the available inspection records are categorised at step 1220, OCR is computed (using the condition rating module 1100) for the categorised component for one or more time instances to be used as input in the SVR prediction model at stage 1230. It will be appreciated in alternative embodiments the condition status data of each similar component may alternatively be input in the SVR prediction model at stage 1230. OCR provides a means for reducing the number of input data points by averaging them first.

In the preferred embodiment, the main parameters for the SVR are - Kernel: polynomial; Degree: 2; coefficient: 0; Gamma: 1; capacity: 100000; and epsilon: 20 le-3. It will be appreciated that other parameter values may be incorporated depending on the particular application, and the invention is not intended to be limited to this implementation.

In the case of the condition status data being insufficient or having no 25 deterioration, the SVR gets assistance from additional data points by utilising the categorisation process as described above. Additional data points obtained from the categorisation process as shown in Figure 3, can be included in order to reinforce certain situations where the number of available condition status data points for the SVR training process is insufficient or no deterioration is detected. For example, if a 30 component does not have sufficient condition status values for a given period of time, additional data can be obtained by utilising the categorisation process to determine similar components and acquire indicative condition status values stored against those components for/within that period.

The deterioration module can then predict the condition status values for one or more components of an asset during one or more future time instances. These future condition status values can be used to predict one or more future parameter

2020100769 18 May 2020 values for each component and/or asset, including for example associated OCR. values, asset values and BHI value as described for the condition rating module 1100. Future risk score values may also be determined by the deterioration module as described for the condition rating module 1100. These future parameter values 5 can be determined using the condition rating module 1100, using the predicted future condition status values as inputs. Preferably the condition status value is predicted for periodic future time instances, for example every year, for a predefined term, for example for a term of 25 years. It will be appreciated that other periodic time instances and predefined terms may be utilised.

In the preferred embodiment, the predicted future condition rating values associated with one or more components of a structure are used to determine any one or more of the following parameters:

• Future overall condition rating of a group of like components;

· Future structure health index (or future BHI) of a structure associated with the one or more components;

• Future component value of a group of like components;

• Future asset value of a structure associated with the one or more components;

· Future component risk value of a group of like components; and/or • Future structure risk value of a structure associated with the one or more components.

This computed future information is useful for prioritising and establishing long25 term maintenance and budget planning.

Maintenance Planning Module

Referring now to figure 4, in the preferred embodiment the management system 1000 further comprises a maintenance planning module 1300 configured to assist 30 agencies in planning maintenance schedules for various structures, such as bridges, and their related components in a network.

For example, bridge structures that pose a safety risk must be repaired without delay. Bridge components that are in a critical condition must also be seen as a 35 priority and repaired immediately. The complexities involved in the assessment of bridges in a large network pose significant challenges to decision makers with regard to planning and budgeting for repair and maintenance works. The

2020100769 18 May 2020 prioritisation of repair and replacement works is governed by risk management standards. Figure 4 shows the general procedures for analysing and calculating maintenance required bridge components in the maintenance planning module 1300 of an embodiment of the invention.

The maintenance planning module 1300 is configured to receive inputs from the condition rating module 1100 and the deterioration module 1200. The condition rating module 1100 provides the maintenance planning module 1300 with information/data relating to the current health status of a bridge and/or bridge component (step 1301), including for example any combination of one or more of:

current OCRs, BHIs, CAVs, CCVs and average component and structure risk scores. This information preferably includes the risk score associated with each bridge structure of interest and in particular the risk score associated with one or more, preferably all, structural components of interests as determined by the risk scoring function of the condition rating module 1100. The deterioration module 1200 provides the maintenance planning module 1300 with information/data relating to predicted future status of a bridge and/or bridge component (step 1302), for example periodic future values of one or more of: future OCRs, future BHIs, future CAVs, future CCVs and future average component and structure risk scores. Predetermined funding and/or financial constraints provide a third input to the maintenance planning module 1300.

3.1 Maintenance Prioritisation

The structure maintenance prioritisation is determined by magnitude of the risk score of all components in a network and the level of maintenance funding constraints.

The critical condition state will be determined by the bridge agency. The critical condition state has been set as CS3 and below as a default in the preferred embodiment, however other states may be set as the critical condition state in alternative embodiments.

The maintenance planning module 1300 is operable to execute a maintenance prioritisation function 1310 configured to determine an order of priority for maintenance of bridge components to thereby assist with the overall planning of maintenance. For maintenance prioritisation for the current year, the maintenance prioritisation function is configured to receive as an input the current or recent

2020100769 18 May 2020 component risk scores of bridge components determined using the condition rating module 1100. These risk scores may be determined and received in real time or accessed from a pre-stored database 1500 associated with the system 1000. Upon receiving these inputs the maintenance prioritisation function is then configured to:

a) Prioritise maintenance required components using the risk scores calculated in the condition rating module (for example components are prioritised from highest component risk score to lowest component risk score);

b) List the maintenance required components in a network and then group by structure ID in order of high risk structures to low risk structures (use 10 average structure risk values for prioritisation);

c) determine accumulated maintenance cost by adding up the cost of maintenance of each individual component and utilise available funding criteria to determine a threshold for the accumulated maintenance cost, then determine a cutoff point for the ordered maintenance required components based on the threshold 15 (see Figure 5);

d) output prioritised components within the threshold to trigger repair and/or replacement work.

3.2 Maintenance Planning Options

The maintenance planning module 1300 further comprises a maintenance planning 20 options function 1320 that generates a reasonable condition improvement programme for a predetermined number of years of the planning period (for example, but not limited to 25 years). This function 1320 operates to provide predetermined maintenance planning options 1350 based on the output of the maintenance prioritisation function 1310. In particular, the function 1320 outputs 25 predicted future outcomes for multiple planning scenarios. Each planning scenario is dependent on a particular maintenance action for a specified planning period associated with the prioritised components output by the maintenance prioritisation function 1310. For instance, one maintenance action may be not to replace or repair any of the prioritised components for the specified planning period, and 30 another maintenance action may be to repair or replace the prioritised components for the specified planning period. It will be appreciated in alternative embodiments there may be any number and type of maintenance actions, including for example

2020100769 18 May 2020 repair or replace a predetermined percentage of the prioritised components. The planning period may be of any length set by the user. The maintenance planning options function 1320 will generate predicted future maintenance costs for each planning scenario based on the associated maintenance action for the planning period, and based on the predicted future states of the various components as determined using the deterioration module 1200. In the preferred embodiment, three planning scenarios are utilised by the maintenance planning options function: 'do-nothing', 'auto-repair' or 'auto-replacement'. It will be appreciated that the planning option parameter may comprise other scenarios/values and those provided, although preferred, are only exemplary. The values are explained below:

• Do-nothing: no maintenance action performed during the planning period.

• Auto-repair: implement a repair action when the condition rating of a component reaches the critical condition state (set by CS3 and below as default in the preferred embodiment).

• Auto-replacement: implement a repair and replacement action when the condition rating of a component reaches the critical condition state (CS3: repair; CS4 and below: replacement as default in the preferred embodiment).

3.3 Maintenance Costs

Once the maintenance required components are determined in a bridge network, and their associated maintenance action values determined, the maintenance planning module may be controlled to execute a cost function 1330 to determine the maintenance costs for the repair and replacement works for each component. This calculation is based on multiplication of the predetermined unit maintenance costs (that is, unit repair or/and replacement cost) and component quantity within the repair and/or replacement threshold (e.g. those with condition statuses of CS3 and/or CS4). Examples 3 and 4 below demonstrate the calculation of the total maintenance costs in the auto-repair and auto-repair/replacement programme.

Example #3: Auto-repair programme worksheet

Total Compone nt Quantity	Condition States (CSs)	Repair Unit Cost of component (FC) ($)	Total maintena nee Cost ($)
1	2	3	4

2020100769 18 May 2020

10 ”15	5 5”	1 5”	2 ”5''	2 'θ”'	2,000.00 ’ϊ’δΟΟ’.ΟΟ””	8,000.00 ”7,500.00”
20	10	5	5	0	2,200.00	11,000.0 0
5	1	1	1	2	1,300.00	3,900.00
30	5	10	10	5	1,000.00	15,000.0 0
20	10	0	5	5	1,500.00	15,000.0 0
22	12	5	5	0	800.00	4,000.00
32	15	15	2	0	1,100.00	2,200.00

Total Maintenance Cost® = 2,000 X (2 + 2) = 8,000

Example #4: Auto-replacement programme worksheet

Total Compon ent Quantity 10 15	Condition States (CSs)	Repair Unit Cost of componen t (FC) ($) 2,000.00 1,500.00	Replacem ent Unit Cost of compone nt (FC) ($) 3,000.00 2,500.00	Total mainten ance Cost ($) ”13,000’.”’ 00 ”12,5’00’.’” 00
1 4 3	2 1 5	3 2 5	4 3 2
20	7	5	5	3	2,200.00	3,800.00	22,400. 00
5	1	1	1	2	1,300.00	3,000.00	7,300.0 0
30	5	10	10	5	1,000.00	2,200.00	21,000. 00
20	10	0	5	5	1,500.00	3,100.00	23,000. 00
22	10	5	5	2	800.00	1,500.00	7,000.0 0

2020100769 18 May 2020

32

12

15

2

3

1,100.00

2,500.00

9,700.0 0

Total Maintenance Cost (S) = (2,000 X 2) + (3,000 X 3) = 13,000

Any function of the maintenance planning module herein-described can be utilised 5 to predict future maintenance actions and costs using the output information of the deterioration module. For example, expected maintenance costs and/or actions for a particular year in the future may be determined by using the predicted condition ratings for that year.

3.4 Cost Benefit Analysis (CBA)

CBA is applied into the established maintenance planning options—such as 'do nothing' and 'auto-repair'—to compare the economic worth of the proposed maintenance works for the community's costs and benefits of a maintenance decision. CBA is a consistent approach to ensure a robust measure of the economic merit and benefits of a project. The results of a CBA can also be used to compare 15 competing maintenance projects to assess the social worth of project options to deliver specific outcomes. Hence, CBA outcomes will help asset engineers/managers make reliable decisions under funding constraints.

CBA can be expressed by benefit-cost ratio (BCR.) and net present value (NPV) 20 parameters. These are indicators that summarise the overall benefits of the maintenance project. Figure 6 presents the CBA structural process. For the BCR calculation, travel time costs 1341, vehicle operating speed and costs 1342 and accident rates and costs 1343 are considered to calculate the net benefits 1345 — that is, the user costs. The inspection costs per structure and unit 25 repair/replacement cost per component type 1346 are considered to compute the net costs—that is, the agency costs 1347.

A BCR is a comparison between a base case (without Maintenance, Repair & Rehabilitation (MR&R) work) and a project case (with MR&R work). The base case is 30 the same as the 'do nothing' scenario in this development work. The base case usually involves routine/periodic maintenance of existing infrastructure. The base case can include assumptions about further maintenance and replacement of the current structure in the absence of the proposed maintenance options when bridge

2020100769 18 May 2020 agencies have fully scheduled MR&R of each type of bridge components. However, the current BCR process is only considered a routine Level 2 bridge inspection and is not considered scheduled repair and replacement work. A project case is appraised against the base case. This is one of the established maintenance 5 options—that is, repair and replacement work to compare with the base case to compute the BCR. After computation of the base and project case for the benefits and costs, the BCR is the total of the discounted benefits divided by the total of the discounted costs. If the BCR of the project is greater than one, the project has greater benefits than costs.

In order to calculate the present value, P is the discounted benefit and cost of the future value, the future value (Ft) is the value of the future amount in time (t), i is the discount rate, and t is the year, as shown in Equation 8.

The end user can change the required discount rates per annum to reflect a variety 15 of views of the risks associated with realising the potential project benefits and costs. As the TMR CBA manual recommended, the discount rate in this development work has 6% as the default value. Further extending Equation 8, the Net Present Value (NPV) is derived from the following equation:

n

ΣΒ, - ^ct t=o^{k J} where the discount rate is r, the benefit in year t is Bt, the cost in year t is Ct, and n 20 is the time for planning years. The NPV of a stream is equivalent to the cost that would have to be invested today in order to obtain a compounded return of r per cent over n years.

Cost Benefit Analysis (CBA): Input Parameters (Table number and title from 25 Transport and Main Roads, Cost-benefit Analysis Manual - Road Project)

1. Vehicle Operating Cost • VCR (Volume capacity ratio)

- CBAT1: Passenger car equivalent factors

- CBA T2: Hourly PCE capacity

- CBA T3: Road type and peak hour capacity factor

2020100769 18 May 2020 • VOS(Vehicle operating speed)

- CBA T4: Free speed array

- CBA T5: Terrain grade percentages

- CBA T6: FSRG1 - Pavement speed condition factor at 110NRM

- CBA T7: FSRG2 - Pavement speed condition factor at 250NRM

- CBAT8: Final operating speed parameters • VOC(Vehicle Operating Cost)

- CBAT10: Fuel costs and consumption factors

- CBA Til: Fuel consumption gradient adjustment array

- CBAT12: Fuel consumption curvature adjustment

- CBA T13: FCGRVF fuel consumption roughness adjustment array

- CBA T14: Oil costs and consumption factors

- CBA T15: Tyre wear and cost parameters

- CBA T16: Curvature and gradient tyre cost adjustments

- CBA T17: Pre-set gradient and curvature proportions

- CBA T18: Tyre roughness adjustment array

- CBA T19: Repairs and servicing cost (RMUC)

- CBA T20: Pavement condition index

- CBA T21: Time and depreciation factors

- CBA T22: Surface type factor

2. Travel Time Cost · CBA T23: Estimated values of travel time - occupant and freight payload values

3. Accident Costs • CBA 24: Unit costs per crash type for Queensland · CBA 25: Severity of road crashes in Queensland (2005) • CBA 26: Average crash costs • CBA 27: Crash rate per MRS

Example #5: CBA calculation worksheet

	Project Period (0 to n)
Year 0 ($)	Year 1 ($)	Year 2 ($)	Year n

2020100769 18 May 2020

				($)
Base Case
User Costs	- Vehicle operating	248,819.03	258,850.90	269,290.50
- Travel time	137.32	137.32	137.32
- Accident	8,177.97	8,505.08	8,845.29
Total Benefit	257,134.32	267,493.31	278,273.11
Discounted Total Benefit	257,134.32	252,352.18	247,662.08
Agency Costs	Routine Inspection	5,000.00	0	5,000
Total Costs	5,000.00	0	5,000
Discounted Total Costs	5,000.00	0	4,449.98
Project Case
User Costs	- Vehicle operating	248,819.03	258,850.90	269,290.50
- Travel time	137.32	137.32	137.32
- Accident	8,177.97	8,505.08	8,845.29
Total Benefit	257,134.32	267,493.31	278,273.11
Discounted Total Benefit	257,134.32	252,352.18	247,662.08
Agency Costs	Routine Inspection	5,000.00	0	5,000.00
- Maintenance	100,000.00	0	0
Total Costs	105,000.00	0	5,000.00
Discounted Total Costs	105,000.00	0	4,449.98
Discounted Net Benefit	0	0	0
Discounted Net Cost	100,000.00	0	0
Benefit Cost Ratio (BCR)	204.43
Net Present Value (NPV)	$22,905,262.00
Discount Rate (i)	6.00%

Workings for the NPV and BCR. are provided below:

BCR = Total discounted net benefit / Total discounted net costs = (0.00 + 0.00 + 0.00 + ...)/(100,000.00 + 0.00 + 0.00 + .....) = Z04.43

2020100769 18 May 2020

NPV = Total discounted net benefit - Total discounted net costs = (0.00 + 0.00 + 0.00 + .,.} - (100,000.00 + 0.00 + 0.00 + ...} = $ΖΖ,90Ξ, Z6Z, 00

Budget Planning Module

In the preferred embodiment, the system 1000 further comprises a budget planning module 1400. The budget planning module takes as an input a desired 5 Network Health Index (NHI) level. The NHI is an average health index across multiple structures (bridges) in the network, calculated by averaging the BHIs of all bridges in a network for example. Bridge agencies and funding agencies, for example, should come to an agreement as to what is an acceptable NHI level for their network. This will depend on the condition of the bridge stock, repair costs 10 and the availability of funds in some cases. The BMS budget planning module 1400 is a useful tool for enabling informed decision making in this regard. The budget planning module 1400 is configured to consider the budget requirements for achieving the desired NHI by generating a plot, as shown in Figure 7 for example. This gives a snapshot of the task and funding requirements to achieve the desired 15 outcomes. For example, the agency could set a target to achieve a certain minimum NHI by a certain year and the asset managers could then plan and implement a programme to achieve these policy objectives. The budget planning module may also alternatively be configured to allow users to determine budget required for achieving a desired BHI for a particular bridge/structure within the 20 network.

4.1 Performance-based Funding Estimation

Users are required to identify the budgets that are currently available for each bridge network. The module 1400 can plan the network level budget allocation for a predetermined number of years, for example up to 10 years. This is because the 25 deterioration module 1200 has been designed to provide long-term performance of each component for the predetermined number of years. The total annual maintenance budget can be determined by the network level structural performance—that is, the NHI and target NHI needed by the bridge agency. Figure 8a presents the concept of costs for a network level condition improvement. In 30 Figure 8a, Point (A) indicates the predicted NHI and the corresponding network asset value. This can be determined using the output data of the deterioration module. For example the predicted NHI for a particular future year can be determined by averaging the BHI values predicted by the deterioration module in that year. The network asset value is the sum of all asset values and therefore can

2020100769 18 May 2020 also be predicted for a particular year by summing the asset values of bridges predicted for that year. Point (B) indicates the target NHI and the corresponding network asset value as set by the agency. The target asset value (B) for a particular year minus the predicted asset value (A) for that year is the cost required 5 for network level condition improvement.

In the preferred embodiment, the budget planning module is further configured to output information indicative of budget requirements, at predetermined time intervals for a predefined period (for example every year for the next 10 years), 10 that are necessary for achieving a target health index (preferably NHI) level. For example, for a target NHI of 90%, the budget planning module will output information indicative of the budget requirements now for achieving a NHI of 90%, information indicative of the budget requirements for achieving a NHI of 90% 1 year from now, information indicative of the budget requirements for achieving a 15 NHI of 90% 2 years from now, and so on until a predefined term (e.g. 10 years but could be more or less). For the current year, the budget planning module would utilise the maintenance cost outputs of maintenance planning module 1300 based on the current/recent condition rating values as inputs to determine required budget. For future years, the budget planning module would utilise the 20 maintenance cost outputs of the maintenance planning module 1300 based on predicted future condition rating values as inputs to determine the required budget. In the preferred embodiment, the budget planning module is configured to output information indicative of the budget requirements for achieving multiple different target health index values over a predetermined number of years. For example, 25 and referring to figure 8b the budget planning module may be configured to output information indicative of the budget required to achieve a target network performance of 90%, 80%, 70% and 60% thereby automatically generating various budget scenarios to improve the NHI.

Example #6: Budget planning worksheet

Network ID	Future Years	Value For Predicted Network Performance (M$)	Value For Target Network Performance (M$)	Maintenance Budget Requirement (M$)
90%	80%	70%	60%	90%	80%	70%	60%

2020100769 18 May 2020

1	1	80	95	85	-	-	15	5	-	-
2	77	101	90	79	-	24	13	2	-
3	74	107	95	83	-	33	21	9	-
4	71	114	101	88	76	43	30	17	5
5	68	120	107	94	80	52	39	26	12
6	65	128	113	99	85	63	48	34	20
7	62	135	120	105	90	73	58	43	28
8	59	143	128	112	96	84	69	53	37
9	56	152	135	118	101	96	79	62	45
10	49	161	143	125	107	112	94	76	58

Figure 8b describes workings for the maintenance budget planning:

Maintenance budget requirement (M$) at year t = Value for target network performance (M$) - Value for predicted network performance (M$).

The required maintenance budget at the second future year for the 80% of target network performance = 90 - 77 = M$ 13 10

5. Client-Server Application Structure

Referring to figure 10, in a preferred embodiment the civil infrastructure asset management system 1000 is implemented using a client-server distributed application structure/system 2000. The structure 2000 comprises software 15 applications 2210/2220 that are accessible and/or executable via one or more user devices 2110/2120. Such user devices may include for example computing systems such as desktop computers 2110 or mobile computing systems 2120 such as smart phones etc. It will be appreciated that the invention is not meant to be limited to any particular type of computing system as would be understood by a person 20 skilled in the art. The software applications 2210/2220 allow users, such as field operators (i.e. inspectors and maintenance contractors) to access the application on their devices and receive designated tasks to be completed and optionally reported on. For example, field operators may be prompted to acquire and input inspection data for one or more structures in a network for use by the management system 25 1000 as described above. Field operators may also be prompted to perform maintenance jobs on one or more structures as output by the management system

2020100769 18 May 2020

1000. The results of a maintenance job may be input back into the system for further decision making.

One or more management-end software applications including the modules 11005 1400 of the management system 1000 described above are implemented at a central server 2200 to enable other users of the system (such as managers, engineers etc) to manage work flow, budgets and maintenance using the described modules 1100-1400. The memory component 1500 is also associated with the central server. The modules 1100-1400 and optionally other software applications 10 are accessible via one or more computing devices such as desktop 2110 and/or mobile computing systems 2120. A desktop application sub-system 2210 and a mobile clients' sub-system 2220 may be implemented within the server 2200 to enable access of the software applications by the desktop and mobile users 2110 and 2120 respectively. Users can use these software applications and modules to 15 analyse inspection data, plan maintenance work and send information to the field operator applications. Such information may include for example, instructions for acquiring further inspection data and/or conducting maintenance work for example. The management end software applications are also configured to receive data from the field operator applications, including for example inspection data and 20 maintenance reports.

A suitable communication system and protocol 2300 is utilised for linking the central server 2200 to the user devices 2110, 2120 to enable the transfer of information/data between the management-end and field-operator-end software 25 applications. For example, the internet or an intranet maybe used as a protocol for this communication but it will be appreciated that any other protocol known in the art may be utilised and the invention is not intended to be limited to a particular communication system/protocol.

The following is an exemplary workflow using the structure 2000 herein described: 1. The management system desktop user 2110 designates scheduled new inspections and/or maintenance jobs for a structure to appropriate field operators (it will be appreciated in alternative embodiments, this can alternatively or additionally be done by a mobile user 2120);

2. The field operator receives this notification on their mobile device 2120 with details of the assigned activities including job types, worklist, deadline etc

2020100769 18 May 2020 (it will be appreciated in alternative embodiments, this can alternatively or additionally be received by a desktop user 2110);

3. The field operator conducts the assigned activities while entering the required feedback information (i.e. new routine inspection or maintenance 5 job) on the accessible software application; and

4. The completed assigned job is then sent back to the management mobile sub-system 2220 in the server through a communications network 2300 where the completed assigned job can be reviewed by asset engineer/manager for approval before being placed into the management 10 system database 1500.

Advantages of this structure 2000 include instantaneous work updates from field operators while improving efficiency in operating asset management practice.

6. Variations and Advantages

The advantages of implementing the management system 1000 software include:

1) The deterioration modelling used in the management system for bridges assists bridge owners to identify the life cycle of their assets and future 20 bridge needs;

2) The methods used in the deterioration module shorten the total BMS implementation time to reach the BMS target reliability. Therefore, bridge authorities can benefit through early stabilisation of the BMS operation leading to continuous dependable budget planning and MR&R decisions;

3) The deterioration module can recognise the detailed health status of individual structural components, together with risk-based methodology, thereby helping eliminate the probability of overall bridge failures;

4) Although recurrent BMS operational costs such as biennial bridge inspections will be similar, the total costs, including the time for reaching the target 30 reliability will be much less; and

5) Enhanced efficiency in terms of computational effort will provide a fast and user-friendly application;

In alternative embodiments of the invention, the management methodologies used in the management system for bridges 1000 described above can be applied to many other infrastructure assets such as:

2020100769 18 May 2020 o pavement, o high rise buildings, o tall towers, o electricity/communication transmission towers, 5 o gantry structures, o dams & reservoirs, o tunnels, o underground shelters, o underground mines o etc.

In the foregoing, a storage medium may represent one or more devices for storing data, including read-only memory (ROM), random access memory (RAM), magnetic disk storage mediums, optical storage mediums, flash memory devices and/or other machine readable mediums for storing information. The terms machine 15 readable medium and computer readable medium include, but are not limited to portable or fixed storage devices, optical storage devices, and/or various other mediums capable of storing, containing or carrying instruction(s) and/or data.

The various illustrative logical blocks, modules, circuits, elements, and/or components described in connection with the examples disclosed herein may be 20 implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic component, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor 25 may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, circuit, and/or state machine. A processor may also be implemented as a combination of computing components, e.g., a combination of a DSP and a microprocessor, a number of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such 30 configuration.

The methods or algorithms described in connection with the examples disclosed herein may be embodied directly in hardware, in a software module executable by a processor, or in a combination of both, in the form of processing unit, programming instructions, or other directions, and may be contained in a single device or

2020100769 18 May 2020 distributed across multiple devices. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD- ROM, or any other form of storage medium known in the art. A storage medium may be coupled to the processor such that the processor 5 can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor.

One or more of the components and functions illustrated the figures may be rearranged and/or combined into a single component or embodied in several components without departing from the invention. Additional elements or 10 components may also be added without departing from the invention. Additionally, the features described herein may be implemented in software, hardware, as a business method, and/or combination thereof.

In its various aspects, the invention can be embodied in a computer-implemented process, a machine (such as an electronic device, or a general purpose computer or 15 other device that provides a platform on which computer programs can be executed), processes performed by these machines, or an article of manufacture. Such articles can include a computer program product or digital information product in which a computer readable storage medium containing computer program instructions or computer readable data stored thereon, and processes and 20 machines that create and use these articles of manufacture. The invention is typically hosted in a cloud-based environment, with the option for an on-premises installation if required by customers with specific enterprise requirements.

The foregoing description of the invention includes preferred forms thereof. 25 Modifications may be made thereto without departing from the scope of the invention.

7. Modelling of Component Health Deterioration and Model-Based Prediction

The Support Vector Regression model and associated computation of health indicies and risk scores, as previously described in section 2 above, can be substituted with alterative modelling approaches which plots an estimation of a component's deterioration whilst considering various environmental parameters from historical 35 inspection records. Two such modelling approaches are as follows:

2020100769 18 May 2020 (1) Element OCR. Time Series Prediction - involving inputting Big Data inspection records and environmental parameters into machine learning and outputting future predictions.

(2) Health Deterioration modelling - when data is proven inaccurate and unreliable, using this approach will help generalise a natural deterioration of the asset element by applying a model equation with adjustable variables to produce a line of best fit against data of the historical inspection records.

A detailed description of these approaches is provided below:

7.1. Definitions of Health and Its Relationship to Condition States and OCR.

In discussing the condition of building components, each component or each subset of component has a health, which is defined as the ideal health status of the component or subset of component. Health, H, is a function of time, t. That means Health will change over time, the process of which is the deterioration.

The health is defined as a function of time (t):

H=h(t) | te[0,+oo),He[0,l]

When H=l, the component is in perfect ideal health condition; when H=0, the component is totally corrupted.

Inspection record is not the faithful record of health, but the compromised observation of health. Building component inspections usually summarize the status of a component in limited number of Condition States. Currently, the 4states system is widely adopted in many countries. When describing an individual unit of component, condition states are not accurate description of the component health. Each condition status actually corresponds to a range of health, and the mapping from condition state to health might not be even linear (see Table 3 as an example). In real operation, health is not allowed to drop below some threshold, which is the upper limit of CS4, and thus CS4 does not mean component health is 0.

Condition State	Corresponding Health
CSI	80%-100%
CS2	60%-80%

2020100769 18 May 2020

CS3	40%-60%
CS4	0%-40%

Table 3. An Example of Condition State Mapping to Health.

A Rating function R is defined that converts the health value to overall condition rating (OCR) for a unit of component (which can be considered as the OCR when 5 quantity is 1 unit):

However, since ^unit rounds the continuous ^Hunit value into only 4 states (Typically, 100%, 66.7%, 33.3%, 0% in OCR calculation), the ^unit function is irreversible conversion.

That's why the conditions states are actually compromised observations of health.

For example, when a component (Reinforced Concrete Deck) has n Units (Square Feet), each unit has a health - ^wi, the OCR is also not equivalent to the average health.

EJ ___ ^{1 L}

But OCR is approximately equal to Rating function result of average health, where the conversion of average health to OCR is defined as ^{— — n}L?rsi’cii [η] I _n J^— [nJ ^CL’ercgsJ

When there is ⁿ units of components, the OCR can have ^4,1 states in the 4-state system. Here ^ο-^ηζίί[nj is defined as the function to convert ^average to OCR.

Therefore, when a component has large quantity of units, OCR j_S finely divided into states and thus can be used to estimate ^cvsrcgs for deterioration study (where all the quantity of that component are considered as one object for study); while when quantity is small, OCR j_{S on}|y rough observation of (see Figure 11).

However, in real inspections, even quantity is not very small, the majority of OCR historical curves are still polygonal lines (see Figure 12) because of the rounding nature of ^unit and approximate percentages assigned to each condition state by inspectors for practical reasons.

One known method in scientific studies to handle inaccurate observations is to use the average of multiple observations and estimate the error with standard deviation. However, since a specific bridge does not get inspected by multiple

2020100769 18 May 2020 inspectors at the same time there are no multiple observation data to suppress the observation errors. Furthermore, as each bridge is unique considering traffic amount, building materials and standards, and other environmental conditions, given our goal as categorization of bridges/components by environmental conditions, overlapping OCR. historical curves of similar components from different bridges is against our goal of clustering and categorization.

In all, as the result of approximation and rounding natures of inspection data and uniqueness of each bridge and component, studying the internal data manifold inside the inspection dataset (observations) is not very likely to reveal the real component health (facts). The modeling and prediction of facts, health is more important than that of the approximated observations - OCR.

7.2. Mathematic Modelling of Accelerated Deterioration Process.

To build the mathematical model of deterioration, the manner that bridge component deteriorates needs to be defined.

As long as the component is in use, there should be a basic constant rate of deterioration, c:

df ⁼

If the rate is only constant, it is a linear deterioration. However, linear deterioration should only fit mechanical components that get worn in a one-dimensional way, such as that the wearing of a type reduces the tread depth. However, by checking the nature of each bridge components, the Applicant discovered that all components tend to deteriorate faster when they already have some sort of deterioration or defects. For example, cracks will reduce the strength of surrounding structure and lead to easier formation of new cracks; rusts on metal surface, decay in timber, spalling in concrete or stone will grow in size and depth as well. Those accelerated deterioration lead to the concept that existing deterioration will accelerate future deterioration. The acceleration rate, a, is used to describe the deteriorated part's contribution to future deterioration.

Therefore:

—— = f(c+a(l - H)} at

2020100769 18 May 2020

However, since the deterioration only effect the part that is healthy, those rates should only apply on the healthy part. As a result, the following differential equation is derived:

= ₊ = + a(l - h(t))) tit

In this equation, the behavior that deterioration rates + ^αθ· ^_ applied on H is defined as linear. Thus, theoretically the constant deterioration rate ^c and acceleration rate ^a may not be constant. They could also be functions of time (i.e. changing overtime), as they depend on traffic amount, building materials and standards, climate, humidity, geological conditions, characteristics of different types 10 of deteriorations (such as crack, rust, chemical/biological decay, spalling), etc.

Except those detailed behaviors that can be included when further high-level details and highly accurate observations are available, this differential equation should be able to describe the general behaviors of most component deteriorations.

By solving the differential equation with the boundary conditions (h(0) - Ι,Μτ®) - 0), following function is obtained:

The curve is a sigmoid in nature, which has a turning point:

Before the turning point, the component deteriorates faster and faster; while after the turning point, the component deteriorates slower and slower.

Figure 13 shows how those factors affect the deterioration of health. It can be found that when ^{c < D}, the turning point is below 0, the deterioration slows down from the very beginning; while when ^c > °,the turning point is above 0, the deterioration speed increases first and then slows down.

The model above is named as the Acceleration Model.

In general, this model can fit many types of deteriorations including the most widely studied reinforced concrete models. Although many specific models for concrete deterioration have been suggested, researches have found that their fitness to real data is still limited. The present generic model has the advantage to

2020100769 18 May 2020 simulate accelerated deterioration process with parameters that can be fully understood, where the accelerated deterioration process is a basis for optimizing timing. If maintenance shall be done to avoid the need of replacement, the maintenance shall be done at early stage before the deterioration reaches the turning point, where the deterioration speed is the highest. Maintenance at early stage means paying for running cost of lower average deterioration rates, which is thus the goal of sustainable maintenance planning.

7.3. Build Health Curve from Inspection OCR Data.

To rebuild the health curve from OCR. points, there are two cases:

1) Deterioration starts from ^H = i. In this case, there are two parameters to resolve: ^Q'^c· fc + _c}_e-^^+CJf _Qe-Ca₊d_f+c

2) Deterioration start from H < 1_ in this case, the deterioration time ^> °, so there are three parameters to resolve:

_ (b + ug-tc+dCt+tn) _c

Since the constant deterioration rates and acceleration rates must be above zero and ⁱo should be no less than 0 when ^{f >} °, the boundary conditions for those parameters are:

a > 0,c > 0, t_D > 0

In addition, as all component have estimated range of lifespans, boundary conditions for life spans and turning points were also set up and used penalty in the neural network algorithms that builds the health deterioration curves from OCR curves.

_f . W_C)_e-^LC+CJr“A ~ hVlifs) ~ -U+dt_[i/c+ - °'⁴

By solving the equation:

(1 ^Ioe----ΪΓΤΤ---^iifs ~

In addition, the turning point is:

I°g+ —_____L.

an f^irn „ > „

JU a + c

So the boundary conditions are:

W⁰min tiifs Ltf^erzaj:

2020100769 18 May 2020

Twn _{mE R} < if urn —

Where we assume turning point should be relative to life span as all component should deteriorate slower when they are intact:

min « QALifs^ « O.SLife^

To solve the curve, we are fitting the curve to a series of OCR. points:

OCR,) 11 < i < ni e N

Fitting the curve is to find the °·· ^c> ⁱD that minimizes the cost (deviation):

cost = p (a + cjg-Uz+cJltj+tnJ ^{i = L}' _ag-C4Z4-^iti+toJ _{+ c}

- OCR/}² n

The boundary conditions were set up as linear penalty when °·· ^c·· are out of the boundary.

A generic algorithm is set up to adjust to minimize the cost. In order to minimize the cost instead of the penalty, the algorithm must start with a proper set of ⁿ··^c that sits within the boundary:

^^ι/®ιπίη — ^ίί/s —

OALife^n *Turn_mirl < i_tLirn «O.SZife^

So assume the following as a typical solution of the starting point:

(1 - h_Eifo)a + c

E-----------‘-----------_ lif + Lif ^[i/e a+c 2 a

_ °^c _ Llfe^in + W^enazjt ^turn a + c 4

From which is derived:

(a 4- a)(c - h_Ei^a) = 0

Given:

a > 0, c > 0

We can get:

Then obtaining the starting parameters for generic algorithm :

-4log h_EE/e -4fc_Ei/e log Λ_ΕΕ/ί

Q. ⁼ ---------------------------------“j C ⁼---------------------------------¢1 + + Rife^ax} (1 + ^ii/s p_ +

7.4. Definition of Lifespan Boundaries for Component Type.

define whether some curves can fit into the lifespan limits or hit the boundary. A further machine learning process is to set up to fit proper

2020100769 18 May 2020 f₀|- each specific component type. In this machine learning, we define Curves^ife^^Life,^'} _as the assembly of all curves within the boundary of —'U/^rrcojt and ® < tfurn — Si 0.8Zif β, gpnj then we can define the machine learning of function:

def:F(.CompoentType') -* (Fife„,_ia,Life_mcx) by: flitntmizs: raI;dCompoTiff?ii:CitrT?0sO/Typff eCwves(Lif „, Life^ )

This machine learning will set up lifespan boundaries for each component type. 10

7.5. Subcategorization of Component Type into Subcategories and Environment Parameters Classification of Component.

The concept of which will be further applied to subcategorization of components 15 with clustering.

d ef: (fc, metm_;, Δ _;) 11 < i < fc, i E JV by:

il m;

Af iMtinizi?: y (ii/Sj — τΐΐ^ατί;}² — msnnj < Δ_; i=ij=i then:

Δ; Δ( ^Lifei ™ = ^mean i - J ^Lif^e i = + 7

By this clustering approach, the components in a component type are subcategorized and define their minimal life span range.

Furthermore, with the subcategories, classification machine learning will be applied to find relationship between bridge and component parameters and environmental parameters to the subcategorization.

def:F(bridge, component, enviromeni} -* Subcategory by:

{/^lf^ecompDnsnt ^— l^lf^en.axsabcatsgery । Component ^> ^^lf^emax' subcatsgery 0 | Hf^ecamponmt itf⁹rrdn_Safc_atSg_ary ^f⁰ntinsu,0category ~ ltf⁰conq>imsnt I usu&ccts-gory ^> Component

Minimizeiyilife^^^f -mean_SlLi,_cetegory')² + PenaltyRate / .(Penalty(_life_CBmpBns!lt)}²

2020100769 18 May 2020

This classifier will enable us to assign a bridge without any historical data to a proper subcategory of component for generic model based health prediction.

By further analyzing the importance/contribution of each parameter in categorization by machine learning, we will be able to achieve the life insurance type subcategorization for each component type.

7.6. Conclusion.

This approach reflects the nature of the components while netrualizes the affects of 10 changing parameters such as traffic changes, climate changes, geological changes, etc. in each deterioration cycle.

8. Machine Learning Time Series Prediction of OCR Values based on the Inspection Big Data

8.1. Importance of Irregular Changing Parameters in Asset Management.

Except for the regular maintenance of components that naturally deteriorate to some level, a lot of maintenance effort aims toward fighting against irregular disasters or accidents. Natural disasters and accidental damage in one year can be reflected in the following year's inspections, which were also caught by the inspection records as irregularities. However, in a broader scope, irregularities are relative to many additional varying factors, such as climate, traffic or even economy. Therefore, a comprehensive time series machine learning prediction should also be able to prediction future inspections based on current knowledge of changing data in climate, traffic and even economy indices.

8.2. Machine Learning Scope of Time Series OCR Prediction.

In order to take irregular parameters such as climate, geological changes/disasters, 30 traffic and even economy indices into the prediction for inspection and maintenance decisions into consideration, Time Series OCR. prediction should consume many historic data in a broad scope.

To build up the relationship from changing parameters to future OCR, the input of 35 the function should be last few years' data and the output should be future data.

2020100769 18 May 2020 def:F„ last n years OCR last n years local climate info last η years traffice data last n years stock and enconomy indices bridge info component info constant environment info

-» next Ξ years OCR by:

mtni-mizg: ^(τι^χί 5 years OCR — predicted OCR)¹

Theoretically,⁷¹ should be relative large to allow machine learning to discover insights over long term climate fluctuations. Given the limitation of data, can typically be 10.

8.3. Optimization of Machine Learning Scope.

Although a generic time series model can be achieved by machine learning of nation-wide data, accuracies may increase when generic prediction is divided into zones (geological/climate zones or bridge/component subgroups) where factors with less effects can be removed to reduce noise. Furthermore, the impacts of additional detailed data such as max daily precipitation, annual precipitation, max annual temperature, min annual temperature, etc can be also evaluated.

The optimization of the prediction success rate will be achieved by further subcategorization of geological/climate zones or bridge/component subgroups or by incorporation of additional valuable data such as max daily precipitation, annual 20 precipitation, max annual temperature, min annual temperature, etc.

of factors') -s next Ξ years OCR

3.4. Fluctuation Detection in the Changing Environmental Factors.

In order to judge when fluctuation should be considered to adjust the health modelling prediction, the threshold ⁸ of irregularity should be determined. We define as the threshold. is the standard deviation, ⁿ is the factor applied to the standard deviation, ⁸i is the threshold between 0 and 1.

, -, f 0|Δ< n&: std Exception^) = L . „ 0 < < 1

The cost function is designed as following to ensure ⁸i is reasonably large to avoid take all noises as irregularity.

2020100769 18 May 2020 maximize:^ R ε JVJ e JV

The nature of this irregularity detection is unsupervised learning, so results are cross-checked by performance. The irregularity conditions will determine where the significant amount ofchanging factors should be considered. When multiple parameters reach irregular threshold, decision makers will be also notified with irregularity warnings such as potential flood damage.

8.5. Conclusion.

The foregoing time series prediction approach is based on observations where irregular factors were also incorporated, and thus the prediction output will also reflect irregular factors in the previous years. Those predictions are built on the basis of neural network algorithms or decision tree algorithms, which sometimes can not be as clearly understood as the deterioration model, but can adjust the health deterioration model when fluctuation of factors can be detected.

Maintenance costs

The total cost of maintenance discussed in section 3.3 above is no longer determined by the cumulative sum of repair or replace costs. Instead, the cost of maintenance is divided into Labour, Equipment and Materials cost based on the user's professional input. However, subject to the provision of sufficient historical maintenance records provided by the customer, machine learning can be used to provide a calculated estimate of the maintenance activity cost automatically.

10. Cost benefit analysis

The Cost-Benefit Analysis as described in section 3.4 can been superseded by the foregoing system. Instead, a machine learning function can determine the most optimal time for performing MR&R (Maintenance, Repair & Rehabilitation) on a specified infrastructure asset for reduced long-term costs whilst maximising infrastructure lifespan.

For example, there are many methods to repair a concrete crack, but each are most effective depending on the extent of the crack and have different applied 35 costs.

2020100769 18 May 2020

By investigating the behaviours of all defect deterioration, it is possible to discover the most optimal action for maintaining a network of assets.

11. Additional enhancements

The foregoing system can incorporate a Maintenance Planner Budget Planning module. The module applies budget planning by allowing the user to input the costs for each activity scheduled for a particular time period then calculating the total cost for all activities. In addition, the module allows users to create multiple What

If scenarios to compare and contrast maintenance plans to determine the best solution in maintaining their asset network to a targeted overall condition rating.

The foregoing system can also generate detailed and informative analytical reports, for example:

- A list of the most critical assets requiring attention

- Records of previous years expenditure

- Printouts of planned maintenance activities for budget planning

- Current condition of individual assets

Claims

1. A system for monitoring the condition of one or more civil structures and forecasting, using one or more processors, the deterioration of each civil structure, 5 the system comprising:

at least one computer readable medium for storing historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more civil structures, and for storing environmental parameters; and

10 at least one processor arranged to process the condition state data for each component and the environmental parameters to predict information indicative of future performance-related condition states associated with the component.

2. A system as claimed in claim 1, wherein the information is predicted using 15 deterioration modelling to produce a line of best fit against the historical condition state data.

3. A system as claimed in claim 2, wherein the deterioration modelling is applied responsive to determining that the historical condition state data is 20 inaccurate or unreliable.

4. A system as claimed in claim 2, wherein the deterioration modelling involves using the component health (/7) function:

(_O + c)_S-^lfl+tJr

H = =-----_r—v:----11 > 0, c > 0, a > 0 _Q0-la+cJt _c

25 where c is the deterioration rate;

a is the acceleration rate; and t is time.

30 5. A system as claimed in claim 2, wherein producing the line of best fit involves minimizing the cost as follows:

where c is the deterioration rate;

35 a is the acceleration rate;

2020100769 18 May 2020 t is time; and

OCR. is an overall condition rating.

6. A system as claimed in claim 2, wherein the deterioration modelling involves 5 using machine learning to establish lifespan boundaries for each type of component.

7. A system as claimed in claim 6, wherein the deterioration modelling further involves subcategorizing each type of component.

8. A system as claimed in claim 1, wherein the information is predicted using machine learning by minimize: ^(next 5 years OCR — predicted OCT)² where:

15 OCR is an overall condition rating.

9. A system as claimed in claim 8, wherein next 5 years OCR includes the environmental parameters, in turn, including one or more of local climate information, traffic data, stock and economy indices, bridge information and other 20 environment information.

10. A system as claimed in claim 1, wherein the environmental parameters are subcategorized according to one or more of: geological zones, climate zones, precipitation and temperature.

11. A system as claimed in claim 1, further involving warning means for generating a warning when the environmental parameters are subject to irregular fluctuations.

12. A method for monitoring the condition of one or more civil structures and 30 forecasting, using one or more processors, the deterioration of each civil structure, the method comprising the steps of:

obtaining, from at least one computer readable medium, historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more civil structures and 35 environmental parameters, and

2020100769 18 May 2020 predicting, using at least one processor arranged to process the condition state data for each component, information indicative of future performance-related condition states associated with the component and the environmental parameters.

5 13. A civil infrastructure asset management system comprising at least one computer readable medium for storing one or more software applications and at least one processor for executing the software applications, the system comprising a performance deterioration module configured to predict deterioration of one or more components of one or more structures, the deterioration module being 10 configured to:

obtain, from the at least one computer readable medium, historical condition state data indicative of one or more performance-related condition states obtained over time for a plurality of components of the one or more structures and environmental parameters,

15 process, using the at least one processor, the condition state data for each component and the environmental parameters to predict information indicative of future performance-related condition states associated with the component.