A rural highway grade promotion decision method based on a greedy genetic algorithm

Abstract

The application discloses a rural road grade promotion dynamic decision method based on a greedy genetic algorithm, comprising the following steps: acquiring data of each road section in a road network to be decided, constructing a candidate scheme set containing a status maintaining scheme and at least one grade promotion scheme for each road section, and determining construction costs of the schemes; and the rural road grade promotion decision method based on the greedy genetic algorithm can automatically filter out a project combination maximizing the overall comprehensive benefit of the road network from hundreds of candidate schemes by integrating multi-dimensional benefit quantification, budget constraints and regional balance requirements into a combination optimization model and solving the model by using a greedy genetic hybrid algorithm, thereby avoiding the limitation that a traditional manual sorting or single project evaluation method is difficult to consider the global optimization, and providing a scientific and efficient quantitative basis for the rural road grade promotion decision.

Description

Technical Field

[0001] This application relates to the technical field of highway upgrading decision-making, and in particular to a rural highway upgrading decision-making method based on a greedy genetic algorithm. Background Technology

[0002] In current rural road upgrade decision-making, related technologies mostly focus on single-dimensional evaluation or engineering structure design. For example, safety facility evaluation methods only identify weak points in road network safety, connectivity evaluation methods only measure accessibility, and road structure upgrades only provide physical modification schemes. These discrete tools cannot answer the core planning question: "Given a limited budget, which road sections should be prioritized for upgrades, and to what level?" Traditional analytic hierarchy process (AHP) and similar methods rely on manual weighting, making it difficult to handle the global optimization challenges arising from the explosion of hundreds of candidate road sections. Furthermore, once a planning scheme is formulated, it lacks the ability to dynamically respond to external changes. Therefore, there is an urgent need for a decision-making method that can integrate multi-source data, automatically output the globally optimal project combination, and possess dynamic adjustment capabilities. Summary of the Invention

[0003] This application aims to at least partially address one of the technical problems in the related art.

[0004] Therefore, one objective of this application is to provide a decision-making method for upgrading rural roads based on a greedy genetic algorithm. By incorporating multi-dimensional benefit quantification, budget constraints, and regional equilibrium requirements into a combinatorial optimization model and using a greedy genetic hybrid algorithm to solve the problem, this method can automatically select the project combination that maximizes the overall comprehensive benefits of the road network from hundreds or thousands of candidate solutions. This avoids the limitations of traditional manual sorting or single-project evaluation methods that cannot take into account the global optimum, and provides a scientific and efficient quantitative basis for decision-making on upgrading rural roads.

[0005] To achieve the above objectives, the first aspect of this application proposes a dynamic decision-making method for upgrading the grade of rural roads based on a greedy genetic algorithm, comprising: Acquire data for each road segment in the road network to be decided, construct a candidate scheme set for each road segment that includes a scheme to maintain the status quo and at least one scheme to upgrade the level, and determine the construction cost of each scheme; Based on indicators from multiple evaluation dimensions, the comprehensive benefit value of each road segment under each candidate scheme is calculated; With the goal of maximizing the sum of the comprehensive benefits of the selected options and with the constraint that the total construction cost does not exceed the predetermined budget, a combinatorial optimization model is established. The combined optimization model is solved using a greedy genetic hybrid algorithm to obtain the optimal combination of road segment upgrade schemes.

[0006] This application presents a rural road grade improvement decision-making method based on a greedy genetic algorithm. By incorporating multi-dimensional benefit quantification, budget constraints, and regional equilibrium requirements into a combinatorial optimization model and solving it using a greedy genetic hybrid algorithm, this method can automatically select the project combination that maximizes the overall comprehensive benefits of the road network from hundreds or thousands of candidate schemes. This avoids the limitations of traditional manual sorting or single-project evaluation methods that cannot take into account the global optimum, and provides a scientific and efficient quantitative basis for rural road grade improvement decisions.

[0007] In addition, the rural road grade upgrading decision method based on a greedy genetic algorithm proposed above in this application may also have the following additional technical features: In one embodiment of this application, the indicators of the multiple evaluation dimensions include at least safety benefits, connectivity benefits, economic incentive benefits, and road condition service benefits; the calculation of the comprehensive benefit value includes: The original benefit values ​​of each evaluation dimension were normalized to obtain the normalized benefit values ​​of each dimension. The normalized benefit values ​​of each dimension are weighted and summed using a dynamically adjustable weight vector to obtain the comprehensive benefit value. The normalization process uses the following formula for positive indices. For contrarian indicators, use the formula In the formula The original benefit value, , These are the minimum and maximum values ​​of the indicator among all candidate solutions, respectively. , They are the 95th percentile and the 5th percentile, respectively. When the value exceeds the range [0, 100], it is truncated to 100 or 0 respectively.

[0008] In one embodiment of this application, the dynamically adjusted weight vector is determined in the following manner: A human-computer interaction interface is provided, on which weight adjustment controls corresponding one-to-one with the multiple evaluation dimensions are displayed, and the sum of the current values ​​of each control is constrained to a constant value. The interface also simultaneously displays the comparison results of the road segment level improvement scheme combinations obtained by the combined optimization model under at least two different weight vector configurations. In response to the decision-maker's operation of adjusting the weight value of at least one dimension through the control, the comparison results are updated in real time until the decision-maker confirms a set of target weight vectors; The confirmed target weight vector is used as the weight vector for the current planning period, and substituted into the weighted summation calculation of the comprehensive benefit value.

[0009] In one embodiment of this application, the constraint further includes a regional equilibrium constraint, which is applied in the following manner: The road network to be decided is divided into multiple regional units according to preset administrative or planning regions; a regional equilibrium penalty term is introduced into the objective function of the combined optimization model, and the penalty term is calculated according to the following formula:

[0010]

[0011] in, This is the regional equilibrium penalty value. The penalty coefficient is... This represents the total number of regional units. For belonging to the first A collection of road sections in each area For a binary variable, when the road segment The value is 1 when the grade improvement plan is selected, and 0 otherwise; the objective function is modified to maximize the sum of the comprehensive benefits minus the penalty value; The penalty coefficient The initial value is automatically set according to the ratio of the predetermined budget to the average construction cost of a single road segment, and is gradually increased during the iteration of the greedy genetic hybrid algorithm to ensure that the final solution satisfies that at least one road segment in each region obtains a grade upgrade scheme.

[0012] In one embodiment of this application, the greedy genetic hybrid algorithm performs a greedy repair operation on individuals that do not meet the budget constraint during the solution process. The greedy repair operation iteratively removes the selected solution with the lowest benefit-cost ratio in the current solution until the budget constraint is met.

[0013] In one embodiment of this application, the greedy genetic hybrid algorithm also performs a greedy local search enhancement operation on the best individual in the population during the solution process. The enhancement operation includes attempting to add unselected high-efficiency cost ratio solutions to the current solution and replacing the selected low-efficiency cost ratio solutions with them, so as to improve the overall efficiency while meeting the budget constraints.

[0014] In one embodiment of this application, when initializing the population, the greedy genetic hybrid algorithm generates at least some individuals through a greedy strategy. The greedy strategy selects schemes in descending order of their benefit-cost ratios until the budget is exhausted.

[0015] In one embodiment of this application, the method further includes: When an external trigger event is received, the implemented road segment plan is locked into an unchangeable state, the candidate plan set and associated parameters are updated, the combined optimization model is solved again, and the adjusted combination of road segment level improvement plans is output.

[0016] The second aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the dynamic decision-making method for upgrading rural road grades based on a greedy genetic algorithm proposed in the first aspect.

[0017] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of a dynamic decision-making method for rural road grade improvement based on a greedy genetic algorithm proposed in the first aspect.

[0018] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0019] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a rural road grade improvement decision-making method based on a greedy genetic algorithm according to an embodiment of this application; Figure 2 The flowchart below shows the solution of a greedy genetic hybrid algorithm for a rural road grade improvement decision-making method based on a greedy genetic algorithm according to another embodiment of this application. Figure 3 This is a flowchart illustrating the dynamic adaptive update of a rural road grade improvement decision-making method based on a greedy genetic algorithm, according to another embodiment of this application. Detailed Implementation

[0020] Embodiments of this application are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. Rather, embodiments of this application include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0021] The following is in conjunction with the appendix Figure 1 - Appendix Figure 3 This application describes a dynamic decision-making method for upgrading rural road grades based on a greedy genetic algorithm, comprising: Acquire data for each road segment in the road network to be decided, construct a candidate scheme set for each road segment that includes a scheme to maintain the status quo and at least one scheme to upgrade the level, and determine the construction cost of each scheme; Based on indicators from multiple evaluation dimensions, the comprehensive benefit value of each road segment under each candidate scheme is calculated; With the goal of maximizing the sum of the comprehensive benefits of the selected options and with the constraint that the total construction cost does not exceed the predetermined budget, a combinatorial optimization model is established. A greedy genetic hybrid algorithm is used to solve the combinatorial optimization model and obtain the optimal combination of road segment upgrade schemes.

[0022] This method can be executed by a computer or server with the corresponding software installed, to assist transportation planning departments in scientifically selecting a combination of road sections to be prioritized for upgrading from a number of rural road sections to be renovated, given limited funds.

[0023] Step 1: Obtain data and build a set of candidate solutions.

[0024] First, obtain basic data for each road segment in the road network to be decided. This data can come from existing highway condition monitoring reports, traffic volume observation station records, local socio-economic statistical yearbooks, and safety risk survey ledgers, etc.

[0025] For each road segment, decision-makers pre-determine several feasible solutions based on its current technical level, traffic demand, and terrain conditions, forming a candidate solution set. For example, for a road segment currently classified as a Class IV highway, the candidate solutions might include: Option 0: Maintain the status quo and perform only routine maintenance; Option 1: Carry out major road repairs and improve local safety facilities without changing the highway grade; Option 2: Upgrade to a Class III highway; Option 3: Upgrade to a Class II highway.

[0026] For each of the above options, calculate the required construction cost. This cost can be estimated based on local construction engineering quotas, road width, mileage, and historical settlement data of similar projects.

[0027] Step 2: Calculate the overall benefit value of each option.

[0028] The purpose of this step is to quantify the overall benefits that each option can bring. The selected evaluation dimensions include at least four: safety benefits, connectivity benefits, economic benefits, and road condition service benefits.

[0029] The method for quantifying safety benefits is as follows: based on the safety risk assessment model, calculate the reduction in the predicted accident risk value of the road section before and after the implementation of a certain plan.

[0030] The quantification of connectivity benefits is as follows: based on the road network connectivity evaluation model, calculate the reduction in travel time from major villages along the route to the nearest county town, school, or hospital after the road section is upgraded.

[0031] The economic benefits are quantified by: calculating the number of permanent residents in the area directly affected by the road section, the level and scale of the industrial parks or tourist attractions covered, and taking into account the predicted value of the induced traffic volume for comprehensive scoring.

[0032] The way to quantify the benefits of road condition services is to directly measure them by the improvement in the road surface performance index after implementing a certain plan.

[0033] The original benefit values ​​for the four dimensions mentioned above are normalized to eliminate dimensional differences. Normalization can be achieved using quantile methods. For example, for positive indicators, the formula (original value - minimum value) / (95th percentile value - minimum value) * 100 is used, and for negative indicators, the formula (maximum value - original value) / (maximum value - 5th percentile value) * 100 is used, with the results limited to between 0 and 100.

[0034] Finally, the four normalized benefit values ​​are weighted and summed according to preset weights (e.g., each initially accounts for 25%), and the result is the comprehensive benefit value of the road segment under this scheme.

[0035] Step 3: Establish a combinatorial optimization model.

[0036] This step transforms a real-world planning and decision-making problem into a mathematical model that can be solved by a computer.

[0037] Decision variable: The final choice of all road segments in the road network. For example, if there are N roads, each road selects one option from its own set of candidate options, and all choices are combined to form a decision option.

[0038] Objective function: To maximize the sum of the overall benefits of all selected options.

[0039] Constraint: The total construction cost of all selected options shall not exceed the total fiscal budget for this year.

[0040] Step 4: Solve using a greedy genetic hybrid algorithm.

[0041] When the number of candidate road segments is large, the number of possible combinations is extremely large, making exhaustive comparison impossible. Therefore, a specially designed greedy genetic hybrid algorithm is used for efficient optimization. The solution process is executed iteratively within the computer and includes the following key steps: Encoding and Initialization: Each decision combination is encoded as a "chromosome," with each gene on the chromosome corresponding to the selected solution number. When generating the initial population, some individuals are generated using a "greedy" strategy, prioritizing solutions with high "return on investment per 10,000 yuan" until the funds are exhausted. This ensures that there are high-quality solutions in the initial population; the remaining individuals are generated randomly to maintain solution diversity.

[0042] Constraint handling and fitness evaluation: Calculate the total benefit for each individual in the population as its fitness. For "infeasible" individuals whose total cost exceeds the budget, initiate a "greedy repair" procedure: repeatedly find the project with the lowest "input-output ratio per 10,000 yuan" among the current options, downgrade its option to "maintain the status quo," which is equivalent to withdrawing this investment, and check the total cost until the total cost meets the budget constraint.

[0043] Population Evolution: Through operations such as "selection," "crossover," and "mutation," the natural evolutionary process is simulated, allowing the population to produce new individuals with better efficiency generation after generation. If the new individuals produced in each generation exceed the budget, the aforementioned "greedy repair" procedure will be run again to correct the situation. At the same time, the best individuals in each generation are directly retained to the next generation (elite retention strategy).

[0044] Local Enhancement: For the best remaining individual, perform a "greedy local search" fine-tuning. That is, try to add some currently unadopted but highly cost-effective solutions, and correspondingly remove some currently adopted but low-cost-effective solutions. As long as this replacement can improve the overall benefit without increasing the total budget, it is retained, and this operation is repeated until no further optimization is possible.

[0045] The algorithm stops after reaching a preset number of iterations (e.g., 500 generations). At this point, the individual with the highest fitness in the decoding population yields a list of road segment upgrade plans that maximize the overall benefits of the road network. This list clearly indicates "which roads should be upgraded and to which level," and includes corresponding calculations of total cost and expected total benefits.

[0046] By following the four steps outlined above, and given the data, benefit model, and budget constraints, a combination of highway upgrading schemes with decision-making reference value can be obtained. Subsequently, based on this result and in accordance with actual circumstances, a specific annual construction plan can be developed.

[0047] In one embodiment of this application, the indicators for multiple evaluation dimensions include at least safety benefits, connectivity benefits, economic incentive benefits, and road condition service benefits; the calculation of the comprehensive benefit value includes: The original benefit values ​​of each evaluation dimension were normalized to obtain the normalized benefit values ​​of each dimension. The normalized benefit values ​​of each dimension are weighted and summed using a dynamically adjustable weight vector to obtain the comprehensive benefit value. Among them, the normalization process uses the formula for positive indicators. For contrarian indicators, use the formula In the formula The original benefit value, , These are the minimum and maximum values ​​of the indicator among all candidate solutions, respectively. , They are the 95th percentile and the 5th percentile, respectively. When the value exceeds the range [0, 100], it is truncated to 100 or 0 respectively.

[0048] The original benefit values ​​of the four dimensions are uniformly transformed into standardized scores that can be weighted and summed.

[0049] 1. Obtaining the initial benefit value.

[0050] For each candidate solution for each road segment, the raw benefit values ​​in four dimensions are obtained: Safety benefits: The reduction in the road network safety risk index after implementing this plan; Connectivity benefits: The reduction in average travel time from major nodes along the route to the nearest service center, in minutes; Economic impact: An economic impact index comprehensively evaluated based on factors such as the population along the route and the coverage of industrial parks; Road service benefits: the expected increase in the Road Surface Quality Index (PQI).

[0051] All four indicators are positive; the higher the value, the better the benefit.

[0052] 2. Normalization process.

[0053] Taking safety benefits as an example, assuming that all candidate solutions for all road segments generate N original safety benefit values, the processing steps are as follows: Sort N values ​​in ascending order; Determine the minimum value 95th percentile ; For each original value According to the formula Calculate the normalized value; If the calculation result is greater than 100, take 100; if it is less than 0, take 0.

[0054] use Using the non-maximum value as the denominator is to avoid compressing the distribution range of most data with a few extreme high values, so that the normalized score maintains a reasonable degree of discrimination. The other three indicators are processed in the same way.

[0055] 3. Weighted summation.

[0056] The four normalized scores of each option are weighted and summed according to a preset weight vector to obtain the overall benefit value of that option. Initially, the weight of each dimension is 0.25 by default. Decision-makers can adjust the weights of each dimension as needed, and the system will recalculate the overall benefit value of all options according to the new weights.

[0057] In one embodiment of this application, the dynamically adjusted weight vector is determined in the following manner: It provides a human-computer interaction interface, which displays weight adjustment controls that correspond one-to-one with multiple evaluation dimensions. The sum of the current values ​​of each control is constrained to a constant value. The interface also displays the comparison results of the road segment level improvement schemes obtained by the combined optimization model under at least two different weight vector configurations. In response to the decision-maker's operation of adjusting the weight value of at least one dimension through the control, the comparison results are updated in real time until the decision-maker confirms a set of target weight vectors; The confirmed target weight vector is used as the weight vector for the current planning period, and substituted into the weighted summation calculation of the comprehensive benefit value.

[0058] Specifically, this is an interactive method that assists decision-makers in determining the weight values ​​for each evaluation dimension.

[0059] 1. The presentation of the weight adjustment interface.

[0060] The computer display interface provides a weight setting area. This area contains four adjustment controls (such as sliders or numerical input boxes) corresponding to the four evaluation dimensions: safety, connectivity, economy, and road conditions. The current weight value is displayed next to each control, and the sum of the four values ​​is always constrained to 100%. When the decision-maker increases the weight of a certain dimension, the weights of the other dimensions are automatically reduced proportionally to maintain the total weight.

[0061] 2. Presentation of the results of the comparison of multiple options.

[0062] In another area of ​​the interface, at least two sets of model calculation results under different weight configurations are displayed simultaneously. Specifically, the system runs the combined optimization model once in the background for the currently set weights and once for a preset set of control weights, then displays the two sets of results side-by-side. Each column lists the selected road segment number, upgrade level, required cost, and expected benefits under the corresponding weight. Differences between the two sets of results are highlighted, allowing decision-makers to intuitively compare the changes in project selection brought about by different weight strategies.

[0063] 3. Real-time updates and confirmations.

[0064] When a decision-maker drags the slider to adjust the weight of a certain dimension, the system detects the change in value and automatically invokes a simplified solution process to quickly update the item list and benefit data in the "Current Weight" column of the comparison results area. Thanks to the use of fast operators such as greedy repair and local enhancement, a single update can be completed in a short time. Decision-makers can repeatedly adjust the weights, observe the changes in the scheme, and click the confirmation button when satisfied with a certain set of weights and its corresponding item combination.

[0065] 4. Substituting the weight vector.

[0066] After clicking "Confirm," the system will record the set of weight values ​​on the current interface as the target weight vector and substitute it into the weighted summation calculation of the comprehensive benefit value. The comprehensive benefit value of all subsequent schemes will be calculated using this set of weights.

[0067] In one embodiment of this application, the constraints further include a regional equilibrium constraint, which is applied in the following manner: The road network to be decided is divided into multiple regional units according to preset administrative or planning areas; a regional equilibrium penalty term is introduced into the objective function of the combined optimization model, and the penalty term is calculated according to the following formula:

[0068] in, This is the regional equilibrium penalty value. The penalty coefficient is... This represents the total number of regional units. For belonging to the first A collection of road sections in each area For a binary variable, when the road segment The value is 1 when the grade improvement plan is selected, and 0 otherwise; the objective function is modified to maximize the sum of comprehensive benefits minus the penalty value; Penalty coefficient The initial value is automatically set based on the ratio of the predetermined budget to the average construction cost of a single road segment, and is gradually increased during the iterative process of the greedy genetic hybrid algorithm to ensure that the final solution satisfies the requirement that at least one road segment in each region obtains a grade upgrade scheme.

[0069] In practical decision-making, a common problem arises where the optimal solution derived from the model might concentrate most of the funds in two or three highly profitable townships, while neglecting to upgrade any road sections in other townships. While this may offer the best overall engineering benefits, it is difficult to implement in practical management. To address this issue, this method introduces a soft-processing mechanism for regional equilibrium constraints.

[0070] 1. Division of regional units.

[0071] First, based on actual management needs, the road network to be evaluated is divided into R regional units according to administrative divisions (such as townships) or planning areas. For example, if a county has 12 townships, the entire road network is divided into 12 regions, each containing all road sections to be evaluated within its territory.

[0072] 2. Calculation of penalty items.

[0073] When evaluating the merits of a decision, the system checks each of the 12 regions. For any region, if at least one road segment is selected for a grade upgrade, the region's penalty contribution is zero; if no road segment is selected in a region, the region incurs a penalty contribution.

[0074] In the specific calculation, for each region r, the number of selected road segments within it is counted. If the number is 0, then (1-0)=1, and the penalty contribution for that region is 1; if the number is greater than or equal to 1, then max(0, 1-number) is 0, and there is no penalty. The penalty contributions of all R regions are summed, and then multiplied by the penalty coefficient. The total regional equilibrium penalty value P is obtained.

[0075] 3. Modification of the objective function.

[0076] The algorithm originally evaluated the merits of a solution solely based on its overall benefit. However, by introducing a penalty term, the evaluation criterion is modified to: Fitness = Overall Benefit - P. This means that if a solution results in certain areas not improving certain items, its fitness will be lowered by the penalty, putting it at a disadvantage during the evolutionary process.

[0077] 4. Penalty coefficient Automatic adjustment.

[0078] Penalty coefficient The size of the area determines the stringency of the regional equilibrium requirements. The penalty is too small, and the final solution may still have areas missing. If it's too large, it overemphasizes balance and sacrifices overall efficiency. In this method, Adopt an automatic adjustment strategy: Initial value setting: estimated based on the ratio of the total budget to the average construction cost per road. This ratio roughly reflects the number of projects the budget can cover; the more projects, the easier it is to achieve regional coverage. The initial value can be set to be relatively small.

[0079] The iterations gradually increase: In the early stages of algorithm iteration, When the value is relatively small, the algorithm prioritizes finding solutions with high overall efficiency; as iterations proceed, As the number of solutions gradually increases, infeasible solutions (solutions with missing regions) are forced to be eliminated. In the later stages of iteration, only those solutions that satisfy both regional equilibrium and have high overall benefits will survive.

[0080] Through this dynamic penalty mechanism, the algorithm can automatically find a reasonable balance between "maximizing global benefits" and "ensuring balanced regional coverage" without the need for repeated manual adjustment of constraint parameters.

[0081] In one embodiment of this application, the greedy genetic hybrid algorithm performs a greedy repair operation on individuals that do not meet the budget constraint during the solution process. The greedy repair operation iteratively removes the selected solution with the lowest benefit-cost ratio in the current solution until the budget constraint is met.

[0082] Specifically, after the crossover and mutation operations of the algorithm, new individuals (i.e., new combinations of solutions) are generated. These new individuals often result in total construction costs exceeding the budget constraint, and are therefore infeasible solutions. The purpose of this step is to repair these infeasible solutions into feasible solutions.

[0083] 1. Calculation of benefit-cost ratio.

[0084] The repair operation relies on a key indicator: the benefit-cost ratio, which is the overall benefit value of each selected upgrade plan divided by its construction cost, representing "how much benefit can be generated from an investment of 10,000 yuan".

[0085] 2. Iterative removal process.

[0086] The specific repair steps are as follows: Check if the total construction cost for the current unit exceeds the budget. If not, no repair is needed.

[0087] If the ratio exceeds the limit, calculate the benefit-cost ratio of all selected options within that individual and find the option with the smallest ratio. This option represents the one with the lowest capital utilization efficiency in the current combination.

[0088] Changing the gene locus value of the proposed scheme to 0 would downgrade it from "level upgrade" to "maintain status quo," essentially abandoning the investment.

[0089] Recalculate the total construction cost. If it still exceeds the budget, recalculate the benefit-cost ratio of the remaining selected options under the new conditions, and remove the one with the lowest ratio again.

[0090] Repeat the above steps, removing the least cost-effective option each time, until the total construction cost is reduced to within the budget.

[0091] This method of gradually eliminating inefficient investments can retain high-efficiency solutions as much as possible while fixing constraints, and reduce the loss of overall solution quality caused by the fixing process.

[0092] In one embodiment of this application, the greedy genetic hybrid algorithm also performs a greedy local search enhancement operation on the best individual in the population during the solution process. The enhancement operation includes trying to add the unselected high-efficiency cost ratio scheme to the current scheme and replacing the selected low-efficiency cost ratio scheme with it, so as to improve the overall efficiency while meeting the budget constraint.

[0093] Specifically, after each generation of evolution in the genetic algorithm, although the best individual in the population is already the optimal solution for the current generation, there may still be room for local optimization. This step performs a fine-grained local adjustment on the best individual retained from each generation, with the aim of further exploring potential within the budget constraint and improving the overall efficiency of the solution.

[0094] 1. Try incorporating cost-effective solutions.

[0095] First, iterate through all unselected candidate solutions (i.e., road segments with gene position 0 and their corresponding level upgrade options) in the current optimal individual, calculate the benefit-cost ratio of each unselected solution, and sort them from highest to lowest ratio. In this order, attempt to add each unselected solution to the current individual one by one. After each attempt, check if the total cost exceeds the budget. If it does not exceed the budget, retain the addition operation; if it does, abandon the addition. After the iteration is complete, the overall benefit of the current individual is initially improved.

[0096] 2. Try replacing the less cost-effective solution.

[0097] After the addition operation is completed, the following situation may occur: the total budget for the current plan is full, but some of the selected plans are not cost-effective, while there are still unselected plans with higher cost-effectiveness that have not been included. Therefore, the replacement operation should be performed.

[0098] Iterate through all unselected options. For each high-efficiency unselected option, try replacing a low-efficiency selected option in the current individual with it. Specifically, cancel the investment in a low-efficiency option (change its gene locus to 0) and add the high-efficiency option (change its gene locus to the corresponding level). If the total benefit after replacement is improved and the total cost does not exceed the budget, retain the replacement; otherwise, cancel the replacement and restore the original state.

[0099] Repeat the above replacement operation until no replacement combination can be found that can further improve the overall efficiency.

[0100] After the above two-step adjustment of addition and replacement, the overall benefit of the optimal individual is further optimized without exceeding the budget. Then, the enhanced individual is put back into the population to continue to participate in the evolution of the next generation.

[0101] In one embodiment of this application, when initializing the population, the greedy genetic hybrid algorithm generates at least some individuals through a greedy strategy. The greedy strategy selects schemes in descending order of their benefit-cost ratios until the budget is exhausted.

[0102] Specifically, at the start of the algorithm, a set of initial individuals needs to be generated to form the first generation population. Randomly generated individuals are usually of poor quality, which can lead to slow convergence of the algorithm. Therefore, this method uses a greedy strategy to generate some initial individuals to improve the overall quality of the initial population.

[0103] 1. Construct a global benefit-cost ratio ranking table.

[0104] All upgrade plans for all road sections are compiled together, and the benefit-cost ratio of each plan (i.e., the overall benefit value divided by the construction cost) is calculated. Then, the plans are sorted from high to low according to the ratio to form a global ranking table.

[0105] 2. Greedy filling method to generate individuals.

[0106] Create a new empty individual, and initially set all gene positions to 0 (i.e., all road segments are temporarily set to maintain the status quo).

[0107] Starting with the solution with the highest ratio in the sorting table, try adding it to the current individual one by one. After each addition, accumulate the construction cost already used.

[0108] If the total cost after incorporating the current plan does not exceed the budget, then the plan is accepted, and the corresponding gene loci are set to the corresponding level values.

[0109] If the total cost exceeds the budget after adding the option, skip that option and continue to check the next option in the sorting table.

[0110] Repeat this process until the entire sorting table has been traversed, or the remaining budget is insufficient to support any of the remaining schemes in the sorting table.

[0111] Individuals generated using this method incorporate as many cost-effective solutions as possible within a budget constraint, and their overall efficiency is typically significantly better than that of randomly generated individuals. In generating the initial population, this method uses the aforementioned greedy strategy to generate approximately half of the individuals, while the other half is still generated randomly, thus balancing the quality of the initial solution with the diversity of the population.

[0112] In one embodiment of this application, the method further includes: When an external trigger event is received, the implemented road segment plan is locked into an unchangeable state, the candidate plan set and associated parameters are updated, the combined optimization model is solved again, and the adjusted combination of road segment level improvement plans is output.

[0113] In practice, five-year plans or annual plans for upgrading highway grades are rarely finalized and set in stone. Various changes may occur during the planning period, such as the establishment of a new industrial park in a township, severe damage to a road due to a natural disaster, or additional special subsidies from higher-level government. When these external events occur, adjustments to the original plan are necessary based on the new circumstances.

[0114] 1. Receiving externally triggered events.

[0115] The system reserves a data interface to receive externally input event information. Typical triggering events include: New industrial planning: If a new industrial park or tourist destination is added in a certain area, the economic benefits of the relevant road sections need to be reassessed; Sudden change in road conditions: Due to factors such as water damage, the road service benefits and safety benefits indicators of a certain road section change significantly; Budget adjustment: The total amount of available funds for the year increases or decreases.

[0116] 2. Locking in implemented projects.

[0117] When a trigger event is received, some road sections may have already completed upgrade construction during the planning period. These implemented projects are established facts and should not be canceled or downgraded in the new round of optimization.

[0118] The system maintains a list of implemented projects. Before resolving the problem, the decision variables corresponding to these implemented road segments are locked, that is, the corresponding gene loci are fixed to the values ​​of the implemented level scheme. In subsequent population initialization, crossover, mutation and repair operations, these gene loci will not participate in any change operations.

[0119] 3. Parameter update and re-solution.

[0120] Update the relevant data based on the type of the triggered event: If it is a new industrial plan, update the original value of the economic driving effect of the relevant road sections and recalculate its comprehensive benefits; If there is a sudden change in road conditions, update the original values ​​of road conditions and safety benefits for the affected road sections; If it is a budget adjustment, update the budget constraint parameters.

[0121] Simultaneously, the construction costs of implemented projects are deducted from the total budget, and the remaining budget serves as the budget constraint for the new round of solution. After completing the above update, the combined optimization model of steps S3 and S4 and the greedy genetic hybrid algorithm are invoked. Under the premise of locking the implemented projects and using only the remaining budget, all unimplemented road segments are re-solved, and the adjusted combination of road segment grade improvement schemes is output.

[0122] 4. Practical application examples.

[0123] For example, a county formulated an annual upgrade plan at the beginning of the year, including 10 upgrade projects. Mid-year, a large agricultural product processing park was newly established in a township, expected to significantly increase traffic volume and economic benefits on related roads. Upon receiving this event, the system locked the three projects already under construction, used the remaining budget as a constraint, updated the economic benefit values ​​of the roads surrounding the park, and recalculated the solution. The results showed that the new plan replaced a road section far from the park in the original plan with a road adjacent to the park, and the overall benefits were improved. This adjusted plan is the new annual implementation plan.

[0124] Through this mechanism, the decision-making method has the ability to dynamically respond to changes in reality, avoiding the problem that one-time planning loses its guiding significance when encountering unexpected situations.

[0125] This application discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a dynamic decision-making method for upgrading rural road grades based on a greedy genetic algorithm.

[0126] This application discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of a dynamic decision-making method for upgrading the grade of rural roads based on a greedy genetic algorithm.

[0127] Specifically, the implementation process of this application is as follows: Step 1: Construct a set of candidate solutions.

[0128] Basic data for each road segment in the road network to be decided is obtained, sourced from highway condition monitoring reports, traffic volume observation station records, socio-economic statistical yearbooks, and safety risk ledgers. For each road segment, several candidate schemes are pre-set based on its current technical level, traffic demand, and terrain conditions. For example, for a Class IV highway segment, Scheme 0 (maintain the status quo), Scheme 1 (major road surface repair), Scheme 2 (upgrade to Class III highway), and Scheme 3 (upgrade to Class II highway) can be set. For each scheme, construction costs are calculated based on engineering quotas, mileage, and historical settlement data.

[0129] Step 2: Calculate the overall benefit value.

[0130] Four evaluation dimensions were selected: safety benefits, connectivity benefits, economic benefits, and road condition service benefits. The raw benefit values ​​for each dimension were normalized: positive indicators were evaluated using the following formula. The contrarian indicator uses a formula. , If the value exceeds [0, 100], it is truncated to 100 or 0. and As a benchmark, extreme values ​​are avoided to compress data discrimination. The four normalized scores are weighted and summed according to preset weights to obtain the comprehensive benefit value.

[0131] The weight vector is dynamically adjusted via a human-computer interaction interface. The interface provides adjustment controls corresponding to the four dimensions, with the sum of all control values ​​always equal to 100%. The interface simultaneously displays comparison results of at least two schemes with different weight configurations. When decision-makers adjust the weights, the system updates the comparison results in real time and incorporates them into the calculation of the comprehensive benefit value after confirmation.

[0132] Step 3: Establish a combinatorial optimization model.

[0133] The objective is to maximize the sum of the comprehensive benefits of all selected options, with the constraint that the total construction cost does not exceed the budget. If regional equilibrium needs to be considered, a penalty term P=λ·Σmax(0,1-Σyi) is introduced into the objective function, where λ is the penalty coefficient. The initial value of λ is set according to the ratio of the budget to the average cost per route, and it is gradually increased during iteration to ensure that the final solution satisfies that each region has at least one improvement project.

[0134] Step 4: Solve using a greedy genetic hybrid algorithm.

[0135] During population initialization, approximately half of the individuals are generated using a greedy strategy: selecting solutions in descending order of cost-benefit ratio until the budget is exhausted; the remaining individuals are generated randomly. For individuals that do not meet the budget constraint, a greedy repair process is performed: iteratively removing the solution with the lowest cost-benefit ratio until the constraint is satisfied. After each generation, the optimal individual undergoes local enhancement: attempting to add high-cost-effective unselected solutions and replacing low-cost-effective selected solutions to improve overall efficiency within the budget. After a preset number of iterations, the optimal individual is decoded, and the combination of road segment level improvement solutions is output.

[0136] Dynamically adaptive updates.

[0137] When receiving external events such as new industrial plans, sudden changes in road conditions, or budget adjustments, the system locks down implemented projects, deducts the budget already used, updates relevant parameters, re-solves the model, and outputs the adjusted solution combination.

[0138] The above methods can be stored on a computer-readable medium or run within an electronic device.

[0139] The dynamic decision-making method for upgrading rural roads based on a greedy genetic algorithm in this application incorporates multi-dimensional benefit quantification, budget constraints, and regional equilibrium requirements into a combinatorial optimization model and uses a greedy genetic hybrid algorithm to solve it. This method can automatically select the project combination that maximizes the overall comprehensive benefits of the road network from hundreds or thousands of candidate schemes, avoiding the limitations of traditional manual sorting or single-project evaluation methods that cannot take into account the global optimum. This provides a scientific and efficient quantitative basis for decision-making on upgrading rural roads.

[0140] In the description of this specification, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0141] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0142] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A dynamic decision-making method for upgrading the grade of rural roads based on a greedy genetic algorithm, characterized in that, include: Data on each road segment in the road network to be decided is obtained, a set of candidate schemes is constructed for each road segment, including the option to maintain the status quo and at least one option to upgrade the level, and the construction cost of each scheme is determined. Based on indicators from multiple evaluation dimensions, the comprehensive benefit value of each road segment under each candidate scheme is calculated; With the goal of maximizing the sum of the comprehensive benefits of the selected options and with the constraint that the total construction cost does not exceed the predetermined budget, a combinatorial optimization model is established. The combined optimization model is solved using a greedy genetic hybrid algorithm to obtain the optimal combination of road segment upgrade schemes.

2. The method according to claim 1, characterized in that, The indicators for the multiple evaluation dimensions include at least safety benefits, connectivity benefits, economic incentive benefits, and road condition service benefits; the calculation of the comprehensive benefit value includes: The original benefit values ​​of each evaluation dimension were normalized to obtain the normalized benefit values ​​of each dimension. The normalized benefit values ​​of each dimension are weighted and summed using a dynamically adjustable weight vector to obtain the comprehensive benefit value. The normalization process uses the following formula for positive indices. For contrarian indicators, use the formula In the formula The original benefit value, , These are the minimum and maximum values ​​of the indicator among all candidate solutions, respectively. , They are the 95th percentile and the 5th percentile, respectively. When the value exceeds the range [0, 100], it is truncated to 100 or 0 respectively.

3. The method according to claim 2, characterized in that, The dynamically adjusted weight vector is determined in the following way: A human-computer interaction interface is provided, on which weight adjustment controls corresponding one-to-one with the multiple evaluation dimensions are displayed, and the sum of the current values ​​of each control is constrained to a constant value. The interface also simultaneously displays the comparison results of the road segment level improvement scheme combinations obtained by the combined optimization model under at least two different weight vector configurations. In response to the decision-maker's operation of adjusting the weight value of at least one dimension through the control, the comparison results are updated in real time until the decision-maker confirms a set of target weight vectors; The confirmed target weight vector is used as the weight vector for the current planning period, and substituted into the weighted summation calculation of the comprehensive benefit value.

4. The method according to claim 1, characterized in that, The constraints also include regional equilibrium constraints, which are applied in the following manner: The road network to be decided is divided into multiple regional units according to preset administrative or planning regions; a regional equilibrium penalty term is introduced into the objective function of the combined optimization model, and the penalty term is calculated according to the following formula: in, This is the regional equilibrium penalty value. The penalty coefficient is... This represents the total number of regional units. For belonging to the first A collection of road sections in each area For a binary variable, when the road segment The value is 1 when the level upgrade plan is selected, and 0 otherwise; the objective function is modified to maximize the sum of the comprehensive benefits minus the penalty value; The penalty coefficient The initial value is automatically set according to the ratio of the predetermined budget to the average construction cost of a single road segment, and is gradually increased during the iteration of the greedy genetic hybrid algorithm to ensure that the final solution satisfies that at least one road segment in each region obtains a grade upgrade scheme.

5. The method according to claim 1, characterized in that, During the solution process, the greedy genetic hybrid algorithm performs a greedy repair operation on individuals that do not meet the budget constraint. The greedy repair operation iteratively removes the selected solution with the lowest benefit-cost ratio in the current solution until the budget constraint is met.

6. The method according to claim 1 or 5, characterized in that, During the solution process, the greedy genetic hybrid algorithm also performs a greedy local search enhancement operation on the best individual in the population. The enhancement operation includes trying to add the unselected high-efficiency cost ratio scheme to the current scheme and replacing the selected low-efficiency cost ratio scheme with it, so as to improve the overall efficiency while meeting the budget constraint.

7. The method according to claim 1, characterized in that, In the greedy genetic hybrid algorithm, at least some individuals are generated through a greedy strategy when initializing the population. The greedy strategy selects schemes in descending order of their benefit-cost ratios until the budget is exhausted.

8. The method according to claim 1, characterized in that, The method further includes: When an external trigger event is received, the implemented road segment plan is locked into an unchangeable state, the candidate plan set and associated parameters are updated, the combined optimization model is solved again, and the adjusted combination of road segment level improvement plans is output.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1 to 8.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

Images