A two-stage project-level maintenance decision-making method based on evidence and need

Abstract

The application discloses a two-stage project-level maintenance decision-making method based on evidence and demand, which comprises the following steps: S1, dividing standard road units; S2, constructing a successful maintenance case database; S3, establishing an evidence K-nearest neighbor model; S4, obtaining a set-valued decision-making result; S5, auditing and screening and making a first-stage decision; S6, constructing an optimization model considering subjective maintenance demand; and S7, making a second-stage decision to determine a final maintenance scheme. Through the two stages of set-valued decision-making and optimization decision-making, the application effectively reduces the rechecking quantity of road units, improves the artificial efficiency of maintenance experts, simultaneously, the set-valued decision-making result scientifically utilizes the valuable experience in the successful maintenance scheme library, realizes 100% rejection decision for road units with extremely abnormal detection information, and makes the maintenance scheme of each road unit not only reliable in evidence but also optimal in maintenance demand, thereby greatly improving the decision-making efficiency.

Description

Technical Field

[0001] This invention belongs to the field of road maintenance scheme decision-making technology, specifically involving a two-stage project-level maintenance decision-making method based on evidence and needs. Background Technology

[0002] With the gradual improvement of transportation systems, road transport plays a vital role in all sectors. However, due to the effects of loads and the environment, roads suffer irreversible damage, leading to decreased transport efficiency. Therefore, intelligent decision-making regarding maintenance plans for various sections of the vast road system, improving decision-making benefits and traffic efficiency, and narrowing the focus of maintenance experts on challenging road sections are of significant importance for addressing road maintenance decision-making issues.

[0003] In recent years, with the increasing number and service life of highways, the demand for highway maintenance has grown significantly, leading to increased attention on computer-aided intelligent maintenance decision-making methods. Currently, existing intelligent project-level maintenance decision-making technologies still suffer from the inability to reject road elements with abnormal detection information based on evidence from successful maintenance cases. This necessitates a comprehensive review of the intelligent maintenance decision-making results by experts, preventing a significant improvement in the efficiency of intelligent decision-making technology in reducing manual labor. Furthermore, current decision-making technologies for road elements with normal detection are either based solely on evidence from successful maintenance cases or solely on the needs of road maintenance management agencies, potentially leading to the failure of maintenance solutions derived from these technologies.

[0004] Therefore, intelligent decision-making methods that simultaneously consider evidence of successful maintenance cases and maintenance needs are crucial. Summary of the Invention

[0005] Based on the aforementioned background technology and addressing the shortcomings of existing technologies, the purpose of this invention is to propose a two-stage project-level maintenance decision-making method based on evidence and needs. By establishing an evidence-based K-nearest neighbor model that can output the mass function of each maintenance scheme, "outlier scheme," and scheme set based on evidence from successful maintenance cases, the method evaluates the strength of evidence for the adoption of each maintenance scheme for all road elements. Using the Hotz criterion combined with the mass function and benefit matrix, the set-valued decision result for each road element is determined. Road elements with a set of P are rejected and submitted to experts for review. Road elements with a set of non-P are used for optimization model building. An integer programming algorithm is used to solve this model to obtain maintenance schemes that consider subjective maintenance needs and are supported by evidence from successful maintenance cases. Using the intelligent maintenance decision-making algorithm proposed in this invention, the workload of expert review can be reduced, and the decision results can be reliable and optimal.

[0006] Technical solution: To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0007] A two-stage project-level maintenance decision-making approach based on evidence and needs includes the following steps:

[0008] S1, Standard Road Element Division:

[0009] Set the standard length of the detection unit, and divide the managed road segment into R road elements according to the standard length;

[0010] S2. Construct a database of successful maintenance cases:

[0011] Successful maintenance plans for various road elements within the managed road section over the years are collected to construct a database of successful maintenance cases composed of maintenance road elements. Each maintenance road element sample records the maintenance plan and corresponding detection information. The detection information includes F detection indicators. The detection indicators for any maintenance road element sample in this database are used... express;

[0012] S3. Establish the K-nearest neighbor model for evidence:

[0013] Obtain maintenance solutions from the database of successful maintenance cases, Each solution is assigned a unique label, P q Representing the corresponding solutions, a set of successful maintenance solutions is formed. An "outlier solution" P0 is added to this set to obtain a new solution set P, based on the detection information of the r-th path element. And a database of successful maintenance cases, generate the mass function when the road element is in any maintenance scheme or set of schemes according to the following steps;

[0014] S3.1 Determine the value of the nearest neighbor sample number K for the target road element, and obtain the K nearest neighbor road element sample set Φ with similar detection indicators to x in the successful maintenance history based on Euclidean distance. r If Φ r If the label of the k-th nearest neighbor sample is q, then this sample supports the r-th path element in P. q The mass function on P is as follows:

[0015]

[0016] In the formula, γ q =1 / d q d q To successfully maintain the historical scheme, the plan is P q The average Euclidean distance between every two sample detection metrics, d r,k Let be the Euclidean distance between the detection metrics of sample k and path element r, where e is the natural base, and α0, β, and are all hyperparameters;

[0017] S3.2, Order Represents the nearest neighbor sample set Φ r The set of samples whose labels are all q in the middle scheme;

[0018] According to the DS fusion law The r-th path element generated from different samples in set P q The fusion result with the mass function on P is as follows:

[0019]

[0020] In the formula, S k' express The sample with label q for the k'th scheme;

[0021] S3.3, According to the DS fusion law, Φ r Each sample set The r-th path element generated is in set P q The fusion result with the mass function on P is as follows:

[0022]

[0023] in, The normalization factor is expressed as follows:

[0024]

[0025] In the formula, Q represents the number of successful maintenance solutions in the successful maintenance case database, and q' represents a label different from q.

[0026] S4. Obtain the set-valued decision result:

[0027] Constructing a utility matrix The K-nearest neighbor model is used to output the mass function of the r-th road element on each maintenance scheme, "outlier scheme", and scheme set. The Hotz criterion is then used to combine the mass function with the utility matrix. To obtain expected utility E r (A t ), making the expected utility E r (A t The largest set The final set-valued decision result for the r-th path element;

[0028] S5. Review and screen applications and make the first-stage decision:

[0029] Analyze the set-valued decision results of all path elements in S4, and then analyze set A. t Perform a rejection decision for the path elements of set P, and for set At The decision results for the path elements in set P are not retained until the first stage of decision-making is completed;

[0030] S6. Construct an optimization model that considers subjective maintenance needs:

[0031] Based on the set-valued decision results of S5, an optimization model is constructed that includes the objective function, necessary constraints, and maintenance requirement constraints.

[0032] S7. Conduct the second phase of decision-making to determine the final maintenance plan:

[0033] The optimization model is solved using an integer programming algorithm. The final value of the decision variable in this model is the final maintenance plan, thus completing the second stage of decision-making.

[0034] Preferably, in S3, the expression for set P is as follows:

[0035] P = {P0, P1, ..., P} q ,…,P Q}

[0036] In the formula, P1 to P Q P0 represents the successful maintenance solutions in the successful maintenance case database, and P0 represents the "outlier solution".

[0037] Preferably, in S3, the mass function of the evidence K-nearest neighbor model on the "outlier scheme" satisfies:

[0038] m(P0)≡0

[0039] In the formula, m(P0) is the mass function of the "outlier scheme";

[0040] like Then let

[0041] Preferably, in S4, the utility matrix It is composed of (2) Q+1 A matrix consisting of -1)×(Q+1) real numbers satisfies:

[0042]

[0043] Utility matrix The construction process is as follows:

[0044] S4.1 Determine the inaccuracy tolerance γ, as shown in the following expression:

[0045]

[0046] In the formula, |A t| is a nonempty subset A t The base; It is to satisfy and The weight vector;

[0047] S4.2, Maximize information entropy to obtain the weight vector The expression is as follows:

[0048]

[0049] The above equation is constrained by

[0050] S4.3 Calculate the utility matrix The specific values ​​of each element are expressed as follows:

[0051]

[0052] In the formula, This indicates that when the state is the q-th category P q The t-th set The utility;

[0053] When the actual solution is the q-th maintenance solution, u (j)q For set {u i,q |P i ∈A t The j-th largest element in}, u i,q From which the identity matrix is ​​taken

[0054] Preferably, in S4, the utility calculation method for the set containing the "outlier solution" is the same as the utility calculation method for other sets.

[0055] Preferably, in S4, the desired utility E is obtained. r (A t When the mass function and utility matrix are combined, the Hotz criterion is expressed as follows:

[0056]

[0057] In the formula, v is the parameter to be trained, and v∈[0,1).

[0058] Preferably, the training process for parameter v is as follows:

[0059] S4.1. Let v = 0, and let the increment of v be 0.01. Let the maximum total utility on the successful maintenance case database be 0.

[0060] S4.2. Calculate the judgment value under the current value condition of v using the following expression, and proceed to S4.3;

[0061]

[0062] In the formula, N is the number of samples in the successful maintenance case database;

[0063] S4.3 Compare the obtained judgment value with the maximum total utility;

[0064] If the judgment value is greater than the maximum total utility, then set the maximum total utility to equal the judgment value, update v, and proceed to S4.4;

[0065] S4.4. Let v = v + 0.01. If v < 1, return to S4.2; otherwise, output the current v to complete the training of parameter v.

[0066] Preferably, the objective function expression of the optimization model is as follows:

[0067]

[0068] In the formula, Let r be the set-valued decision result of the r-th path element, and W is the set of weights, x r,q Let w be the decision variable for the q-th maintenance scheme corresponding to the r-th road element. j For the j-th weight, O j For the j-th maintenance requirement;

[0069] When the r-th path element adopts the q-th solution, x r,q It equals 1, otherwise it equals 0.

[0070] Preferably, the necessary constraints for the objective function of the optimization model are:

[0071] x r,q ∈{0,1}

[0072]

[0073] Preferably, the maintenance requirement constraints of the optimization model include at least budget constraints;

[0074] The constraint relationship between budget constraints and maintenance costs satisfies the following:

[0075]

[0076] In the formula, cost q Let q be the cost of the q-th maintenance plan, and budget be the total budget for maintenance costs.

[0077] Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0078] Through two stages: set-valued decision and optimization decision, the present invention directly rejects road elements with abnormal detection information based on evidence from the successful maintenance case database. This effectively reduces the number of road elements requiring review and improves the manual efficiency of maintenance experts. Simultaneously, the set-valued decision results scientifically utilize valuable experience from the successful maintenance scheme database, enabling a single road element to obtain multiple feasible successful maintenance schemes. In the optimization decision stage, the set-valued decision results of road elements that were not rejected are used as the range of values ​​for decision variables. This ensures that the maintenance scheme obtained through optimization not only meets the subjective needs of the road management agency but is also supported by evidence from successful maintenance cases. In summary, this invention can achieve 100% rejection of road elements with extremely abnormal detection information, and ensures that the maintenance schemes for each road element are not only supported by reliable evidence but also have optimal maintenance requirements, greatly improving decision-making efficiency. Attached image description:

[0079] Figure 1 It is a multi-stage project-level maintenance decision-making flowchart;

[0080] Figure 2 This is a structural diagram of a successful maintenance history program library;

[0081] Figure 3 It is a collection diagram of maintenance schemes that include "outlier solutions";

[0082] Figure 4 It is a set-valued decision result diagram of the path elements with anomaly detection information;

[0083] Figure 5 This is a diagram illustrating the effect of optimal decision-making. Detailed implementation method:

[0084] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and a specific example of the Yongjin Expressway in Zhejiang.

[0085] like Figure 1-5 As shown, a two-stage project-level maintenance decision-making method based on evidence and needs includes the following steps:

[0086] S1. Divide the standard road elements: Set the standard length of the detection unit. The detection unit is selected from the detection report of the managed road segment. In this embodiment, the standard length of the detection unit is set to 1km. Divide the managed road segment (i.e. the road segment within the management scope) into R road elements according to the standard length. In this embodiment, R = 179.

[0087] S2. Constructing a Successful Maintenance Case Database: Successful maintenance plans for the managed road sections over the years are collected. Based on the principle that each pavement inspection indicator specified in the "China Highway Technical Condition Assessment Standard" (JTG 5210-2018) shows a positive improvement, corresponding maintenance road element samples are selected to construct the Yongjin Expressway successful maintenance case database. This includes maintenance road element samples for each road element. Each maintenance road element sample records the maintenance plan and its corresponding inspection information. Any maintenance road element sample in the successful maintenance history database is in the form of "Inspection Information - Maintenance Plan." The inspection information consists of various inspection indicators on the maintenance road element in the maintenance year, and the maintenance plan is the actual maintenance plan adopted in that year. The inspection information contains F inspection indicators. The inspection indicators used by any maintenance road element sample in this database... express.

[0088] S3. Establish the K-nearest neighbor model for evidence: such as Figure 3 As shown, an "outlier" P0 is added to the maintenance scheme set obtained from the database of successful maintenance cases on the Yongjin Expressway, forming a new maintenance scheme set P = {P0, P1, ..., P...}. q ,…,P7}, where P q This represents the q-th maintenance plan. Based on the detection information of the r-th path element. Using the database of successful maintenance cases, generate the mass function for the road element when its state is any maintenance scheme or set of schemes, following these steps;

[0089] S3.1 Determine the value of the nearest neighbor sample number K for the target road element, and obtain the K nearest neighbor road element sample set Φ with similar detection indicators to x in the successful maintenance history based on Euclidean distance. r If Φ r If the label of the k-th nearest neighbor sample is q, then this sample supports the r-th path element in P. q The mass function on P is as follows:

[0090]

[0091] In the formula, γ q =1 / d q d q To successfully maintain the historical scheme, the plan is P q The average Euclidean distance between every two sample detection metrics, d r,k Let e ​​be the Euclidean distance between the detection metrics of sample k and path element r, where e is the natural base, and α0, β, and γ are... q All are hyperparameters; in this embodiment, α0 and β are set to 0.95 and 1, respectively.

[0092] S3.2, Order Represents the nearest neighbor sample set Φ r The set of samples whose labels are all q in the middle scheme;

[0093] According to the DS fusion law The r-th path element generated from different samples in set P q The fusion result with the mass function on P is as follows:

[0094]

[0095] In the formula, S k' express The sample with label q for the k'th scheme;

[0096] S3.3, According to the DS fusion law, Φ r Each sample set The r-th path element generated is in set P q The fusion result with the mass function on P is as follows:

[0097]

[0098] in, The normalization factor is expressed as follows:

[0099]

[0100] In the formula, Q represents the number of successful maintenance solutions in the successful maintenance case database, and q' represents a label different from q.

[0101] The mass function of the K-nearest neighbor model on the "outlier scheme" satisfies:

[0102] m(P0)≡0

[0103] In the formula, m(P0) is the mass function of the "outlier scheme";

[0104] like Then let Since no successful maintenance program in the history of successful maintenance falls under the P0 category, there is always a...

[0105] S4. Obtain the set-valued decision results: Construct the utility matrix The K-nearest neighbor model is used to output the mass function of all road elements on the road for each maintenance scheme, "outlier scheme", and scheme set. The Hotz criterion is then used to combine the mass function with the utility matrix. To obtain expected utility E r (A t), making the expected utility E r (A t The largest set The final set-valued decision result is supported by evidence from successful maintenance cases.

[0106] Utility matrix It is composed of (2) Q+1 A matrix consisting of -1)×(Q+1) real numbers satisfies:

[0107]

[0108] Utility matrix The construction process is as follows:

[0109] S4.1 Determine the inaccuracy tolerance γ, as shown in the following expression:

[0110]

[0111] In the formula, |A t | is a nonempty subset A t The base; It is to satisfy and The weight vector;

[0112] S4.2, Maximize information entropy to obtain the weight vector The expression is as follows:

[0113]

[0114] The above equation is constrained by

[0115] S4.3 Calculate the utility matrix The specific values ​​of each element are expressed as follows:

[0116]

[0117] In the formula, This indicates that when the state is the q-th category P q The t-th set The utility;

[0118] When the actual solution is the q-th maintenance solution, u (j)q For set {u i,q |P i ∈A t The j-th largest element in}, u i,q From which the identity matrix is ​​taken

[0119] The utility of a set containing "outliers" is calculated in the same way as the utility of other sets.

[0120] To obtain expected utility E r (A t When the mass function and utility matrix are combined, the Hotz criterion is expressed as follows:

[0121]

[0122] In the formula, v is the parameter to be trained, and v∈[0,1).

[0123] The training process for parameter v is as follows:

[0124] S4.1. Let v = 0, and let the increment of v be 0.01. Let the maximum total utility on the successful maintenance case database be 0.

[0125] S4.2. Calculate the judgment value under the current value condition of v using the following expression, and proceed to S4.3;

[0126]

[0127] In the formula, N is the number of samples in the successful maintenance case database;

[0128] S4.3 Compare the obtained judgment value with the maximum total utility;

[0129] If the judgment value is greater than the maximum total utility, then set the maximum total utility to equal the judgment value, update v, and proceed to S4.4;

[0130] S4.4. Let v = v + 0.01. If v < 1, return to S4.2; otherwise, output the current v to complete the training of parameter v.

[0131] S5. Review and screen, and make the first-stage decision: Analyze the set-valued decision results in S4, and apply them to set A. t For path elements in set P, a rejection decision is executed. During the analysis process, when executing the rejection decision, the decision-maker is prompted to manually review and confirm. For set A... t Path elements that are not in set P are retained. After evaluating all path elements, the first stage of decision-making is completed. Next, ... Figure 4 For example, Figure 4 As shown in Figure Ⅰ, 6 path elements are randomly selected. (The rest of the text appears to be a continuation of the previous sentence and can be left as is.) Figure 4 As indicated by box II, the detection information of two road elements was maliciously modified, causing the detection information of these road elements to be abnormal compared to the detection information of road elements in the successful maintenance case database, such as... Figure 4As shown in III and IV, the method proposed in this invention makes the set-valued decision result of the path element with abnormal information the entire set P, which means that this invention can effectively reject such abnormal path elements.

[0132] S6. Construct an optimization model that considers subjective maintenance needs: Based on the set-valued decision results of S5, specifically the set-valued decision results corresponding to the path elements that were not rejected, construct an optimization model that includes the objective function, necessary constraints, and maintenance requirement constraints.

[0133] The objective function expression for the optimization model is as follows:

[0134]

[0135] In the formula, Let r be the set-valued decision result of the r-th path element, and W is the set of weights, x r,q Let w be the decision variable for the q-th maintenance scheme corresponding to the r-th road element. j For the j-th weight, O j For the j-th maintenance requirement;

[0136] When the r-th path element adopts the q-th solution, x r,q It equals 1, otherwise it equals 0.

[0137] The necessary constraints for the objective function of the optimization model are:

[0138] x r,q ∈{0,1}

[0139]

[0140] Maintenance demand constraints refer to a series of maintenance needs that the road management department may subjectively require. The maintenance demand constraints of the optimization model should at least include budget constraints.

[0141] The constraint relationship between budget constraints and maintenance costs satisfies the following:

[0142]

[0143] In the formula, cost q Let q be the cost of the q-th maintenance plan, and budget be the total budget for maintenance costs.

[0144] S7. The second stage of decision-making involves determining the final maintenance plan: An integer programming algorithm is used to solve the optimization model. The final decision variables represent the final maintenance plan, completing the second stage of decision-making. The final maintenance plan possesses two important properties: support from successful maintenance case evidence and optimization of maintenance needs. The improvement in Pavement Condition Index (PCI) under the same budget is shown in the following figures compared to plans obtained using only successful maintenance case evidence. Figure 5 As shown, this result indicates that the proposed solution can improve PCI by 13% while ensuring the support of evidence.

Claims

1. A two-stage project-level maintenance decision-making method based on evidence and needs, characterized in that, Includes the following steps: S1, Standard Road Element Division: Set the standard length of the detection unit, and divide the managed road segment into R road elements according to the standard length; S2. Construct a database of successful maintenance cases: Successful maintenance plans for the managed road sections over the years are collected, and a successful maintenance case database is constructed using these plans for multiple road elements. This database includes maintenance road element samples for each road element, each recording a maintenance plan and its corresponding detection information. The detection information includes F detection indicators. The detection indicators for any maintenance road element sample in this database are used... express; S3. Establish the K-nearest neighbor model for evidence: Obtain maintenance solutions from the database of successful maintenance cases, Each solution is assigned a unique label, P q Representing the corresponding solutions, a set of successful maintenance solutions is formed. An "outlier solution" P0 is added to this set to obtain a new solution set P, based on the detection information of the r-th path element. And a database of successful maintenance cases, generate the mass function when the road element state is any maintenance scheme or set of schemes according to the following steps; S3.1 Determine the value of the nearest neighbor sample number K for the target road element, and obtain the K nearest neighbor road element sample set Φ with similar detection indicators to x in the successful maintenance history based on Euclidean distance. r If Φ r If the label of the k-th nearest neighbor sample is q, then the corresponding sample supports the r-th path element in P. q The mass function on P is as follows: In the formula, γ q =1 / d q d q To successfully maintain the historical scheme, the plan is P q The average Euclidean distance between every two sample detection metrics, d r,k Let e ​​be the Euclidean distance between the detection metrics of sample k and path element r, where e is the natural base, and a0, β, and are all hyperparameters. S3.2, Order Represents the nearest neighbor sample set Φ r The set of samples whose labels are all q in the middle scheme; According to the DS fusion law The r-th path element generated from different samples in set P q The fusion result with the mass function on P is as follows: In the formula, S k' express The sample with label q for the k'th scheme; S3.3, According to the DS fusion law, Φ r Each sample set The r-th path element generated is in set P q The fusion result with the mass function on P is as follows: in, The normalization factor is expressed as follows: In the formula, Q represents the number of successful maintenance solutions in the successful maintenance case database, and q' represents a label different from q; S4. Obtain the set-valued decision results: Constructing a utility matrix The K-nearest neighbor model is used to output the mass function of the r-th road element on each maintenance scheme, "outlier scheme", and scheme set. The Hotz criterion is then used to combine the mass function with the utility matrix. To obtain expected utility E r (A t ), making the expected utility E r (A t The largest set The final set-valued decision result for the r-th path element; S5. Review and screen applications and make the first-stage decision: Analyze the set-valued decision results of all path elements in S4, and then analyze set A. t Perform a rejection decision for the path elements of set P, and for set A t The decision results for the path elements in set P are not retained until the first stage of decision-making is completed; S6. Construct an optimization model that considers subjective maintenance needs: Based on the set-valued decision results of S5, an optimization model is constructed that includes the objective function, necessary constraints, and maintenance requirement constraints. S7. Conduct the second phase of decision-making to determine the final maintenance plan: The optimization model is solved using an integer programming algorithm. The final value of the corresponding input model decision variable is the final maintenance plan, thus completing the second stage of decision-making.

2. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S3, the expression for set P is as follows: P={P0,P1,…,P q ,…,P Q } In the formula, P1 to P Q P0 represents the successful maintenance program in the successful maintenance case database, and P0 represents the "outlier program".

3. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S3, the mass function of the evidence K-nearest neighbor model on the "outlier scheme" satisfies: m(P0)≡0 In the formula, m(P0) is the mass function of the "outlier scheme"; like Then let 4. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S4, the utility matrix It is composed of (2) Q+1 A matrix consisting of -1)×(Q+1) real numbers satisfies: Utility matrix The construction process is as follows: S4.1 Determine the inaccuracy tolerance γ, as shown in the following expression: In the formula, |A t | is a nonempty subset A t The base; It is to satisfy and The weight vector; S4.2, Maximize information entropy to obtain the weight vector The expression is as follows: The above equation is constrained by S4.3 Calculate the utility matrix The specific values ​​of each element are expressed as follows: In the formula, This indicates that when the state is the q-th category P q The t-th set The utility; When the actual solution is the q-th maintenance solution, u (j)q For set {u i,q |P i ∈A t The j-th largest element in}, u i,q Taken from the identity matrix 5. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S4, the utility of a set containing "outlier solutions" is calculated in the same way as the utility of other sets.

6. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S4, the expected utility E is obtained. r (A t When the mass function and utility matrix are combined, the Hotz criterion is expressed as follows: In the formula, v is the parameter to be trained, and v∈[0,1).

7. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 6, characterized in that, The training process for parameter v is as follows: S4.

1. Let v = 0, and let the increment of v be 0.

01. Let the maximum total utility on the successful maintenance case database be 0. S4.

2. Calculate the judgment value under the current value condition of v using the following expression, and proceed to S4.3; In the formula, N is the number of samples in the successful maintenance case database; S4.3 Compare the obtained judgment value with the maximum total utility; If the judgment value is greater than the maximum total utility, then set the maximum total utility to equal the judgment value, update v, and proceed to S4.4; S4.

4. Let v = v + 0.

01. If v < 1, return to S4.2; otherwise, output the current v to complete the training of parameter v.

8. The two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 1, characterized in that, In S6, the objective function expression for the optimization model is as follows: In the formula, Let r be the set-valued decision result of the r-th path element, and W is the set of weights, x r,q Let w be the decision variable for the q-th maintenance scheme corresponding to the r-th road element. j For the j-th weight, O j For the j-th maintenance requirement; When the r-th path element adopts the q-th solution, x r,q It equals 1, otherwise it equals 0.

9. A two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 8, characterized in that, The necessary constraints for the objective function of the optimization model are: x r,q ∈{0,1 10. A two-stage project-level maintenance decision-making method based on evidence and needs as described in claim 8, characterized in that, The maintenance requirement constraints for the optimization model should include at least budget constraints; The constraint relationship between budget constraints and maintenance costs satisfies the following: In the formula, cost q Let q be the cost of the q-th maintenance plan, and budget be the total budget for maintenance costs.

Images