CN102262702A - Decision-making method for maintaining middle and small span concrete bridges - Google Patents

Abstract

The invention provides a decision-making method for maintaining middle and small span concrete bridges. The decision-making method provided by the invention comprises the following steps of: (2) determining a decision-making object and a predicated age limit, wherein the decision-making object is a bridge network OD (Outside Dimension) model; (2) indentifying the bridge network OD model: converting an image-formed OD model into a numerical-value-formed OD model and identifying a minimum path set matrix P which can be used for a decision-making calculation; (3) establishing a cost model of a maintenance policy and an effect model which can be used for processing an event tree analysis on the OD model to obtain the maintenance policy; (4) carrying out an improved non-dominated sorted genetic algorithm: defining a coding method of the maintenance policy and taking a connecting probability Rnw and a maintenance cost Cnw of the bridge connection as two targets to be participated in a genetic evolution; carrying out a Pareto optimality and searching for the globally optimal solution. The method provided by the invention has the advantages of simple principle, convenience in using, wide application range, good reliability and the like.

Description

A kind of medium and small decision-making technique of striding footpath concrete-bridge maintenance

Technical field

The present invention is mainly concerned with the maintenance field of bridge, refers in particular to medium and small decision-making technique of striding the footpath concrete curing in a kind of zone.

Background technology

Along with the continuous development of national economy, improve day by day the effect and the status of highway communication.Bridge is a throat of guaranteeing that highway is unimpeded, and its load-bearing capacity and the traffic capacity are to link up key completely, directly influence economic benefit and social benefit.China has built a large amount of bridges at nearly 30 years, in use, and owing to vehicular load, the volume of traffic constantly increase, the load of bridge increases the weight of day by day, and extraneous various factors role and influence, makes bridge structure produce disease, defective occurs, have a strong impact on its normal use.In the increasing rapidly of bridge scale, a large amount of disease bridges appear, and unsafe bridge quantity shows a rising trend.

Bridge structure is carried out rational maintenance, can significantly improve the service level of bridge, improve the serviceable life of structure.But in fact the maintenance of bridge is not effectively carried out.Be that bridge administrative authority still rests on the backward management mode and ladder of management on the one hand, lack the guiding theory of science, the detection evaluation method of system.Be the maintenance requirements that the maintenance fund of bridge often is not enough to finish all bridges on the other hand, the resources allocation aspect lacks the decision-making means of objective judgment criteria and science.

Present Bridge Maintenance Management System, generally the technical scheme of Cai Yonging is: by bridge member is checked detection, utilization analytical hierarchy process (AHP) is assessed bridge structure, and artificially sets a certain limit value and judge whether bridge structure not will be safeguarded.Check and the existence of method that detects and artificial limit value, mean that there are human factor in the raw data of bridge state estimation and decision principle.The AHP method is by artificial definite bridge member importance, and relatively the importance of bridge member obtains weight matrix, has departed from truth.The object of the strategy of maintenance simultaneously only is a bridge block often, can't consider the optimum state of the transportation network overall situation, rationally arranges limited fund.

Therefore, there is following problem in above-mentioned decision-making technique:

(1) the reasonable assessment means of bridge structural state.The appraisal procedure of generally using is visual examination, instrument detecting bridge structure member at present, and utilization AHP method calculates the score of bridge structure, and supposes its deterioration law.Because therefore the existence that human factor cannot be avoided in checking the process that detects, drafts AHP weight matrix and supposition deterioration law need seek objective assessment method more;

(2) object of Maintenance Decision making.For competent authorities, often be faced with and in the zone, rationally arrange budget fund, solve the problem that when what bridge is adopted what maintenance means.Present maintenance policy optimization object is single bridge project, can't carry out rational fund allocation and optimization from the overall situation of transportation network efficient;

(3) balance maintenance cost and bridge network service performance.People always expect that the cost of the enough minimums of energy acquires best service, yet this is unpractical.Does the raising of bridge network service performance often require to increase the maintenance cost, and these two targets are mutual conflicts, balance how between them? be exactly to solve the problem of the good and bad discrimination standard of maintenance strategy based on these two targets to a certain extent;

(4) calculating means.For the Maintenance Decision making of single bridge, can exhaustive All Policies in certain time limit and compare and obtain optimal strategy; After object rises to the bridge network, every increase by one bridge block, possible maintenance strategy becomes geometric growth, causes so-called shot array problem, and relying on the method for exhaustion to seek optimum global policies is very a poor efficiency and a painful thing.

Summary of the invention

The technical problem to be solved in the present invention just is: at the technical matters that prior art exists, the invention provides that a kind of principle is simple, easy to use, applied widely, the medium and small decision-making technique of striding footpath concrete-bridge maintenance of good reliability.

For solving the problems of the technologies described above, the present invention by the following technical solutions:

A kind of medium and small decision-making technique of striding footpath concrete-bridge maintenance is characterized in that step is:

(1), determines decision object and predicts the time limit that decision object is a bridge network OD model;

(2), the identification of bridge network OD model: with the OD model conversation of graphic form is the OD model of numerical value form, and is identified as and can be used to the minimal path collection matrix P that makes a strategic decision and calculate;

(3), set up the cost model of maintenance strategy and the OD model is carried out the effect model that ETA obtains the maintenance strategy: be used to calculate two target function values of arbitrary maintenance strategy, bridge network-in-dialing reliability R _NwWith maintenance cost C _NwAnd defined the discrimination standard of non-dominance relation as maintenance strategy quality;

(4), improved non-domination ordering genetic algorithm: define the coding method of maintenance strategy, with bridge network-in-dialing reliability R _NwWith maintenance cost C _NwParticipate in genetic evolution as two targets, carry out Pareto optimization, seek globally optimal solution; That is, obtain a Pareto optimality disaggregation, separating the concentrated suitable Pareto optimality strategy of choosing by Pareto optimization.

As a further improvement on the present invention:

The step of described bridge network OD modelling is:

(1), bridge is that node, road are arc in the bridge network OD model;

(2), the connected relation between the actual bridge, in bridge network OD model, couple together with the node of arc with correspondence;

(3), road never lost efficacy in the bridge network OD model;

(4), only have connected relation in the bridge network OD model, the information such as length, width of ignoring bridge and road.

The step of described bridge network OD Model Identification is:

(1), generates the adjacency matrix A of OD model according to the connected relation between the node;

(2), search is represented with T from being communicated with all paths that may exist between the OD in A;

(3), in path tree further search minimal path collection, P represents with matrix.

Describedly the OD model carried out the effect model that ETA obtains the maintenance strategy be:

(1), constitute situation, generate the complete event tree, E represents with matrix according to OD model node;

(2), the minimal path collection P that obtains according to step 106, whether the branch of E is analyzed inefficacy one by one, branch support platform Ef obtains losing efficacy;

(3), according to minimal path collection P and the set E of inefficacy branch _f, generate maintenance strategy effect model.

Described improved non-domination ordering genetic algorithm is: according to the node scale and the optimization time limit of OD model, at first determine population scale, and according to the chromosome coding rule, finish the initialization of population; According to the cost model of maintenance strategy and the effect model of maintenance strategy, calculate two target function values of all individualities, bridge network failure probability and maintenance cost that promptly should the individuality correspondence; Based on this to calculating individual non-domination layering, and population is carried out layer sorting, finish the generative process in mating pond in conjunction with the concentration class comparison operator; In the mating pond father population with and merge by the sub-population of intersecting, the variation computing obtains, the selection by the partial ordering relation operator produces population of future generation.

Compared with prior art, the invention has the advantages that:

1, the present invention adopts RELIABILITY INDEX to describe the configuration state of bridge, has taken all factors into consideration reinforced concrete performance degradation and operating environment for Structural Influence, and science is objective more with respect to means such as visual examinations;

2, decision object of the present invention is all medium and small OD models of striding the footpath concrete-bridge in the inclusion region, considers the maintenance problem from " net ", puts into practice need of work with respect to satisfying more on " point ";

3, among the present invention by setting up maintenance strategy effect model, cost model, define non-dominance relation, obtain different maintenance strategies cost, effect this to the superior and inferior evaluating problem on the conflict objective;

4, adopt improved non-domination ordering genetic algorithm among the present invention, make counting yield be guaranteed, overcome " shot array " problem; Further, non-domination ordering is particularly suitable for handling the optimization of a plurality of conflict objectives, finds Pareto optimality; Genetic algorithm has very strong ability of searching optimum, cooperates with non-domination ordering and can find overall Pareto optimality maintenance strategy.

Description of drawings

Fig. 1 is a schematic flow sheet of the present invention;

Fig. 2 is the schematic flow sheet that the non-domination ordering of employed improvement genetic algorithm is carried out the maintenance policy optimization among the present invention;

Fig. 3 is the synoptic diagram that adopts chromosome coding in the concrete application example of the present invention;

Fig. 4 is the synoptic diagram that adopts population to distribute in the concrete application example of the present invention;

Fig. 5 is the synoptic diagram that adopts the Pareto forward position to evolve in the concrete application example of the present invention.

Embodiment

Below with reference to Figure of description and specific embodiment the present invention is described in further details.

The decision-making technique institute that the present invention is medium and small to stride footpath concrete-bridge maintenance at be bridge network in the zone, it is carried out overall Maintenance Decision making judges, the Pareto optimality of the searching maintenance fund and the bridge network traffic capacity.

As shown in Figure 1, detailed step of the present invention is:

1. step 101: determine the decision object and the prediction time limit;

As required, determine the regional traffic network, from Bridge Management Information System, obtain bridge, road distributed model in the zone, and obtain the member composition and the maintenance means of corresponding bridge.Consider price variation, operability, calculating scale etc., suggestion prediction year was limited in 10 years.

2. step 102: with decision object abstract be bridge network OD model;

The complicated connected network that transportation network is made up of bridge and road, bridge all has different attributes and state with road.Step 102 is carried out corresponding abstract and simplification with the bridge network of reality, and principle is as follows: 1, bridge is that node, road are arc in the OD model; 2, the connected relation between the actual bridge couples together with the node of arc with correspondence in the OD model; 3, road never lost efficacy in the OD model; 4, only there is connected relation in the OD model, ignores the information such as length, width of bridge and road.

3. the identification of step 103:OD model;

The OD model that obtains by above-mentioned steps 102 is to exist with the form of scheming, though can express the bridge network intuitively, is difficult for computing.What step 103 was finished is that the OD model conversation that will exist with the form of scheming is the process of numerical value OD model, comprises 3 sub-steps: 104,105,106.Substep 104 generates the adjacency matrix A of OD model according to the connected relation between the node; Substep 105 is searched in A from being communicated with all paths that may exist between the OD, represents with T; Substep 106 is further searched for the minimal path collection in the path tree, P represents with matrix.

A) step 104: adjacency matrix A

All nodes of OD model (comprising node O, all bridge nodes, node D) constitute set V, and corresponding element is v _iAll arcs constitute set L, and corresponding element is l _i, adjacency matrix A:

A＝[a _ij] (1)

Element a _IjRepresentation node v _iTo node v _jConnectedness:

$a_{\mathrm{ij}}=\left\{\begin{array}{cc}1,& <v_i,v_j>\&Element;\mathrm{<figure-callout\; id="0"\; label="L"\; filenames="HDA0000080654020000041.png"\; state="\{\{state\}\}">L</figure-callout>}\\ 0,& <v_i,v_j>\&NotElement;L\end{array}\right.$ (v _i，v _j∈V) (2)

Promptly as node v _iTo node v _jThere is road l _kDuring connection, a _IjGet 1, otherwise get 0.

The generation method of A is: at first element of initialization is 0 n (element number of set V) rank matrix entirely; Each element is l among the pair set L _kCirculation is according to l _kStarting point v _iWith terminal point v _j, with a of correspondence _IjBe set to 1.The capable nonzero element of i among the A, the node set that representative can arrive from the i node; The nonzero element of j row, representative can arrive the node set of j node.

B) step 105: path tree matrix T

Each row t of path tree matrix T _iRepresent a path between OD, element t _IjRepresent t _iA configuration node, corresponding value is a node serial number.t _iFirst element be node O, last element is node D.The path that has the loop, promptly same node once above path occurs and are not counted among the T.

The generation method of T is: from first all nonzero element (a of row of A _1j=1 (j=1,2 ..., n)) set out, search for all nonzero element (a in the row that is numbered these element column numberings _Jk=1 (k=1,2 ..., n)), constantly become column number and be line number, carry out this search procedure and preserve searching record, up to the place of nonzero element classify A as last classify as and end.

C) step 106: minimal path collection matrix P

Minimal path collection P is the subclass of T, each bar path p among the P _iAll be minimum, wherein after any one node, this path lost efficacy in deletion.

The generation method of P is: at first give P with T; To each bar path p among the P _iThe circulation, respectively with T in all interstitial contents less than p _iThe path t of interstitial content _kRelatively, if there is a path t _kMake p _iComprise t _kAll nodes, p then _iDelete from P.

4. step 107: maintenance strategy effect model;

Step 107 pair OD model carries out the effect model that ETA obtains the maintenance strategy, and it comprises three sub-steps: substep 108 constitutes situation according to OD model node, generates the complete event tree, and E represents with matrix; Substep 109, whether the minimal path collection P that obtains according to step 106 analyzes inefficacy one by one to the branch of E, and the set E of branch obtains losing efficacy _fSubstep 110 is according to minimal path collection P and the set E of inefficacy branch _f, generate maintenance strategy effect model.

A) effect of bridge RELIABILITY INDEX

After the bridge structure maintenance, the corresponding variation takes place in the RELIABILITY INDEX β of bridge and deterioration law thereof, and the present invention adopts three parameters to describe the influence of each maintenance method to RELIABILITY INDEX:

γ: the β value of rising to, the moment of β promotes after the description maintenance;

α: β year degradation ratio minimizing value is described the minimizing of fiduciary level curvature of curve;

t _α: effective time, the action time of describing α.

And think:

1) γ must not make the texture ratio structure original state after the maintenance that higher RELIABILITY INDEX is arranged.The γ value can not change the fiduciary level slope of a curve, and after RELIABILITY INDEX β value promoted, structure was degenerated according to original degradation ratio.

2) maximal value of α must not make fiduciary level degradation ratio after the maintenance greater than 0, and has t effective acting time _α, at t _αOutside, bridge is degenerated according to original degradation ratio.

Bridge structure is made up of different members, and different members can adopt different maintenance means, and every breeding handguard section can be used γ, α, t _αThree parameters are described its effect.

B) step 108: event tree matrix E

The RELIABILITY INDEX of all bridges in the known network obtains the connection fiduciary level of bridge network, needs to rely on event tree to analyze.The event tree of bridge network A is in respectively normally and the combination of failure state for all nodes, can use one 2 ^N-2* (n-2) the matrix E on rank (thinking that node O, node D never lost efficacy) is described event tree, each row e _iRepresented the branch of event tree, element e _IjRepresented bridge j to prop up e in incident _iIn state.e _Ij=0 expression bridge lost efficacy e _Ij=1 expression bridge operate as normal.e _IjValue can express with following formula:

$e_{\mathrm{ij}}=\left\{\begin{array}{c}\frac{\left[\frac{i-1}{2^{n-2-j}}\right]}{2}\\ \end{array}\right\}---\left(3\right)$

[] is the rounding operation symbol in the following formula, and { } is the complementation symbol.

C) step 109: the failure event tree E of branch _f

Any e of delegation among the event tree E _iIn all values be the bridge node of 1 element correspondence, if be enough to constitute any minimal path of minimal path collection P, think the e of this incident branch _iBe effectively, from node O via e _iThe bridge node of middle operate as normal can arrive node D; On the contrary, if e _iIn all values be the bridge node of 1 element correspondence, deficiency is enough to constitute at least one minimal path of minimal path collection P, thinks the e of this incident branch _iLost efficacy, from node O via e _iThe bridge node of middle operate as normal can not arrive node D.Set E according to the failure event that mentioned above principle generates _fBe the subclass of incident E, its numbers of branches is by the structures shape of bridge network.

D) step 110: the effect model of maintenance strategy

The probability of bridge network failure equals the probability sum of all inefficacy branches in the event tree:

$P_{\mathrm{dc}}=\underset{i=1}{\overset{N_{\mathrm{dc}}}{\mathrm{\&Sigma;}}}\left(\underset{j=1}{\overset{N}{\mathrm{\&Pi;}}}{P}_{\mathrm{ij}}\right)---\left(4\right)$

In the following formula:

P _Dc, the failure probability of bridge network;

N _Dc, event tree inefficacy number of branches, the i.e. line number of Ef;

N, the number of all bridge nodes in the bridge network, N=n-2;

P _Ij, the e of event tree branch _iThe state probability of the corresponding bridge node of middle j.

The computing method of bridge network-in-dialing fiduciary level are:

R _nw＝-Φ ^-1(P _dc) (5)

In the formula: Φ ^-1() is the inverse function of Standard Normal Distribution.

When step 112 uses genetic algorithm to carry out multiple-objection optimization, P _DcPerhaps-R _NwWill be as first objective function of Pareto optimization.

5. step 111: maintenance strategy cost model;

Application of the present invention is the bridge network that is in the military service phase, takes after the maintenance measure, will produce corresponding bridge network maintenance cost present worth:

$C_{\mathrm{nw}}=\underset{i=1}{\overset{N_b}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_i^m}{\mathrm{\&Sigma;}}}\frac{C_{\mathrm{ij}}^m\left(t_{\mathrm{ij}}\right)}{{(1+v)}^{t_{\mathrm{ij}}}}---\left(6\right)$

Symbolic significance is as follows in the formula:

N, the bridge interstitial content N=n-2 in the bridge network;

Optimize the maintenance number of times of bridge i in the time limit;

t _Ij, the j time maintenance of bridge i is apart from the time of optimizing time limit starting point;

The cost of the j time maintenance of bridge i;

V, the currency rate of discount.

When step 112 uses genetic algorithm to carry out multiple-objection optimization, C _NwWill be as second objective function of Pareto optimization.

6. step 112: improved non-domination ordering genetic algorithm

Step 112 is finished the Pareto searching process of maintenance strategy with improved non-domination ordering genetic algorithm (NSGA-II), and detailed process as shown in Figure 2.Be to select the different of operator with the maximum difference of common genetic algorithm.The present invention calculates two target function values of maintenance strategy, and in population strategy is carried out non-domination layering according to effect model (step 107) and cost model (step 111), calculates the gathering distance and the constraint destruction value of each strategy.The selection operator of genetic algorithm has been considered following partial ordering relation: 1, non-dominance relation; 2, assemble distance relatively; 3, individual constraint destroys.For guaranteeing the outstanding gene of elitism strategy, parent population and progeny population merge the follow-on evolution of participation.

According to the node scale and the optimization time limit of OD model, at first determine population scale, and, finish the initialization of population according to the chromosome coding rule; The effect model of determining according to formula (4), the cost model that formula (6) is determined calculate two target function values of all individualities, bridge network failure probability and maintenance cost that promptly should the individuality correspondence; Based on this to calculating individual non-domination layering, and population is carried out layer sorting, finish the generative process in mating pond in conjunction with the concentration class comparison operator.In the mating pond father population with and merge by the sub-population of intersecting, the variation computing obtains, the selection by the partial ordering relation operator produces population of future generation.

A) step 201: be the beginning of genetic algorithm, want the work between the completing steps 101～111 before this stage, obtain effect model and cost model;

B) step 202: chromosome coding;

This step with the feasible solution of practical problems from its solution space be transformed into genetic algorithm the process of treatable search volume.The formation unit of bridge network is a bridge structure, and the formation unit of bridge structure is a structural elements, the different maintenance means of definition on the member aspect.The present invention adopts the integer coding method, and the chromosome structure of generation is shown in Fig. 3 (optimizing time limit Y=5), and each time of each member takies a gene position, and genic value is the maintenance means that this member adopted in the corresponding time.Suppose that the OD model comprises the Building N bridge, every bridge i number of components is n _i(N), the maintenance means number of member j is m for i=1,2... _{I, j, k}(i=1,2..., N; J=1,2 ..., n _i), the gene interval be [0, m _{I, j, k}], 0 representative need not to adopt maintenance strategy, 1～m in current this year of this structure _{I, j, k}Representative is when the maintenance strategy numbering of front part this year employing.Chromosomal gene figure place is:

$L=Y\&times;\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{n}_i---\left(7\right)$

Fig. 3 is the chromosome coding synoptic diagram that the present invention obtains.

C) step 203: initialization population P ₁

The population P that scale of initialization is Pop _t(t=1) (t is an evolutionary generation) begins to evolve.Pop directly has influence on convergence and counting yield, the too small locally optimal solution that converges to easily; Excessive, can cause computing velocity to reduce.The present invention adopts random algorithm to finish initialization of population, 2～5 times of chromogene figure places of the value of Pop.

N＝a×L(2≤a≤5) (8)

D) step 204: intersect, make a variation;

With P _tBe the father population, produce sub-population Q by operations such as intersection, variations _t

Crossover operator is selected two individualities by bigger probability from colony, exchange base because of, inherited the essential characteristic of parent, determined the ability of searching optimum of algorithm.Intersection is to produce new individuality, keeps the multifarious main method of population.The present invention adopts the algorithm of even intersection, and two each individual genes of pairing all exchange with identical crossover probability, and crossover probability is p _c=0.5, concrete calculation process is as follows:

Step 1: generate the sub-W=w of shielding ₁w ₂... w _i... w _LL is a chromosome length, w _iRandom assignment is 0 or 1.

Step 2: from two parent individualities of A, B, produce two new offspring individual A ', B ' according to following rule:

If w _i=0, then the i gene of A ' is inherited A correspondence position genic value, and the i gene of B ' is inherited B correspondence position genic value;

If w _i=1, then the i gene of A ' is inherited B correspondence position genic value, and the i gene of B ' is inherited A correspondence position genic value;

The variation computing is to produce new individual householder method, the local search ability of decision genetic algorithm.The variation algorithm that the present invention adopts is evenly sudden change, with meeting the uniform random number that the constraint of gene position value is wanted, with the variation Probability p _m=0.1 replaces the gene in the individual coded strings.

E) step 205: merge population;

The present invention considers elite's retention mechanism, with father population P _tWith sub-population Q _tMerge and participate in evolving.

F) step 206: non-domination ordering;

According to the dominance relation between the individuality to P _tSort.

The condition that individual p arranges individual q is: P _{Dc, p}≤ P _{Dc, q}, C _{Nw, p}≤ C _{Nw, q}, above-mentioned two inequality have at least one should satisfy strict less than, be expressed as p＞q.P＞q shows: individual p compares with individual q, and effect target, cost objective can be not poorer, and at least one is better.

The definition vector n _iAnd s _i(i ∈ P _t) write down the individual number of the individual i of domination and the individual collections that individual i is arranged respectively, promptly have:

n _i＝|{k|k＞i?i，k∈P _t}| (9)

s _i＝{j|i＞j?i，j∈P _t} (10)

Go out each individual s by double cycle calculations _i, n _i, then:

F ₁＝{i|n _i＝0，i∈{1，2，...，N}} (11)

Try to achieve F then successively _k(k=2,3 ..., m):

F _k＝{i|n _i-k+1＝0，i∈P _t} (12)

G) step 207～215: according to non-domination layering and concentration class ordering, from P _tIn produce P of future generation according to qualifications _T+1

The non-domination layering of step 207 initialization i=1 is initialized as empty father population P _T+1Step 208 is judged layering F _iWith P _T+1Individual sum whether reach population scale N: if not, calculate F _iThe gathering distance (step 209) of middle element is with F _iMerge to P _T+1In (step 210), step 211 switches to F _I+1, repeating step 208; If show P _T+1With F _iThe individual number of intersection will above or equal N, write down the position of current layering.Step 213 is judged P _T+1With F _LThe individual number of intersection whether just be N: if then carry out step 215, with F _LMerge to P _T+1In; Otherwise operating procedure 214 is according to assembling distance to F _LIndividuality carries out descending sort, and chooses preceding N-|P _T+1| individuality is incorporated into P _T+1In (step 215).

The gathering distance of individual i is:

D _i＝(P _dc，i+1-P _dc，i-1)+(C _nw，i+1-C _nw，i-1) (13)

P _{Dc, i+1}, P _{Dc, i-1}Be respectively in the population with the most close P of individual i _DcBig value and little value; C _{Nw, i+1}, C _{Nw, i-1}Be respectively in the population with the most close C of individual i _NwBig value and little value; P when individuality _DcOr (C _Nw) when value was population minimum (maximum), its D value was infinitely great, to guarantee the population border.

H) step 216: judge the evolution end condition;

The condition that the present invention stops evolving has three: 1, maximum evolutionary generation; 2, maximum working time; 3, the non-domination layering number of population.

I) step 217: evolve for t=t+1, evolve from step 204 beginning is of future generation.

J) step 218: chromosome decoding;

Evolution obtains the Pareto optimality individuality and decodes according to coding rule, obtain the maintenance strategy set, suitable strategy can be from this set be selected in conjunction with capital budgeting, the conditions such as bridge network failure probability that can bear by maintenance unit, arranges to optimize the maintenance work in the time limit.

Fig. 4 (for conveniently drawing, horizontal ordinate uses the fiduciary level with the failure probability equivalence) is in the evolutionary process, the population distribution schematic diagram, and the scope that population distributes is comparatively wide, and near the population density the close Pareto optimal solution is obviously greater than other zones.Fig. 5 is the synoptic diagram of evolving in the Pareto forward position, evolutionary generation t=10,50 o'clock, Pareto forward position gap is bigger, and the Pareto forward position between the t=50,75 is approaching relatively, t=75, the 100th, the Pareto forward position is very approaching, and the present invention has desirable speed of convergence.

7. step 113: the application of Pareto optimality strategy

Step 112 has obtained a Pareto optimality disaggregation, and administrative authority is from parameters such as capital budgeting arrangement, expection bridge network service levels, according to self-demand, chooses suitable Pareto optimality strategy separating to concentrate, and is used to instruct carrying out of maintenance work.

Below only be preferred implementation of the present invention, protection scope of the present invention also not only is confined to the foregoing description, and all technical schemes that belongs under the thinking of the present invention all belong to protection scope of the present invention.Should be pointed out that for those skilled in the art the some improvements and modifications not breaking away under the principle of the invention prerequisite should be considered as protection scope of the present invention.

Claims (5)

1. stride the directly decision-making technique of concrete-bridge maintenance for one kind medium and small, it is characterized in that step is:

(1), determines decision object and predicts the time limit that decision object is a bridge network OD model;

(2), the identification of bridge network OD model: with the OD model conversation of graphic form is the OD model of numerical value form, and is identified as and can be used to the minimal path collection matrix P that makes a strategic decision and calculate;

(3), set up the cost model of maintenance strategy and the OD model is carried out the effect model that ETA obtains the maintenance strategy: be used to calculate two target function values of arbitrary maintenance strategy, bridge network-in-dialing reliability R _NwWith maintenance cost C _NwAnd defined the discrimination standard of non-dominance relation as maintenance strategy quality;

(4), improved non-domination ordering genetic algorithm: define the coding method of maintenance strategy, with bridge network-in-dialing reliability R _NwWith maintenance cost C _NwParticipate in genetic evolution as two targets, carry out Pareto optimization, seek globally optimal solution; That is, obtain a Pareto optimality disaggregation, separating the concentrated suitable Pareto optimality strategy of choosing by Pareto optimization.

2. medium and small decision-making technique of striding footpath concrete-bridge maintenance according to claim 1 is characterized in that the step of described bridge network OD modelling is:

(1), bridge is that node, road are arc in the bridge network OD model;

(2), the connected relation between the actual bridge, in bridge network OD model, couple together with the node of arc with correspondence;

(3), road never lost efficacy in the bridge network OD model;

(4), only have connected relation in the bridge network OD model, ignore length, the width information of bridge and road.

3. medium and small decision-making technique of striding footpath concrete-bridge maintenance according to claim 2 is characterized in that the step of described bridge network OD Model Identification is:

(1), generates the adjacency matrix A of OD model according to the connected relation between the node;

(2), search is represented with T from being communicated with all paths that may exist between the OD in A;

(3), in path tree further search minimal path collection, P represents with matrix.

4. medium and small decision-making technique of striding footpath concrete-bridge maintenance according to claim 1 is characterized in that, describedly the OD model is carried out the effect model that ETA obtains the maintenance strategy is:

(1), constitute situation, generate the complete event tree, E represents with matrix according to OD model node;

(2), the minimal path collection P that obtains according to step 106, whether the branch of E is analyzed inefficacy one by one, branch support platform E obtains losing efficacy _f

(3), according to minimal path collection P and the set E of inefficacy branch _f, generate maintenance strategy effect model.

5. medium and small decision-making technique of striding footpath concrete-bridge maintenance according to claim 1, it is characterized in that, described improved non-domination ordering genetic algorithm is: according to the node scale and the optimization time limit of OD model, at first determine population scale, and, finish the initialization of population according to the chromosome coding rule; According to the cost model of maintenance strategy and the effect model of maintenance strategy, calculate two target function values of all individualities, bridge network failure probability and maintenance cost that promptly should the individuality correspondence; Based on this to calculating individual non-domination layering, and population is carried out layer sorting, finish the generative process in mating pond in conjunction with the concentration class comparison operator; In the mating pond father population with and merge by the sub-population of intersecting, the variation computing obtains, the selection by the partial ordering relation operator produces population of future generation.

Images