Revolution water potential evaluation method based on machine learning and multi-factor comprehensive analysis

Abstract

The invention discloses a revolution water potential assessment method based on machine learning and multi-factor comprehensive analysis, and belongs to the field of revolution water potential assessment, and the method comprises the steps: obtaining cargo flow and flow direction data, and carrying out the classification of the cargo flow direction; factors influencing revolution water are collected; extracting main influence factors through a principal component analysis method; based on the obtained main influence factors, calculating a main influence factor comprehensive score of each cargo flow direction, and performing dynamic correction on the influence factor comprehensive score; adopting machine learning to generate simulation data, and introducing random disturbance; and setting a water transfer threshold value by adopting a K-means clustering method, analyzing a high-potential flow direction in combination with a thermodynamic diagram, and outputting decision classification of priority water transfer, moderate water transfer and unsuitable water transfer. According to the method, simulation data generation is combined with machine learning optimization, the problem of lack of historical data in water transfer potential evaluation is solved, meanwhile, scientificity and practicability of the data generation process are guaranteed, and distribution and characteristics of high potential flow directions are displayed based on model potential scores.

Description

Technical Field

[0001] The present invention belongs to the technical field of planetary water potential assessment, and more specifically, relates to a planetary water potential assessment method based on machine learning and multi-factor comprehensive analysis. Background Art

[0002] Currently, optimizing transport structures and promoting green transformation are key areas of focus for the transportation and logistics industry. With China's vigorous efforts to optimize transport structures, the shift from road to rail and road to water transport has become a key area of ​​green logistics transformation and a hot topic for industry research. Existing research and technologies for assessing water transport diversion potential primarily employ the following methods: By assigning fixed weights to each mode of transport, the attractiveness and priority of each mode is calculated; by using historical data and expert experience to assess water transport diversion potential using qualitative or simple quantitative methods; and by establishing static mathematical models based on factors such as transport costs, timeliness, and cargo flow to assess potential.

[0003] The above methods have the following main flaws: insufficient historical data, especially in the early stages of the "public-to-water" policy, lack of relevant data on the potential for transport flow diversion, insufficient coverage of existing data, and limitations in the assessment results; existing methods consider water diversion potential more from a theoretical perspective, and the applicability and flexibility of the assessment model are poor. The assessment model lacks scientific optimization and makes it difficult to balance the randomness and complexity in the data. Summary of the Invention

[0004] In view of the problems mentioned above or existing methods for evaluating the potential of water resources conversion based on machine learning and multi-factor comprehensive analysis, the present invention is proposed.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] The embodiment of the present invention provides a method for evaluating the potential of public water transportation based on machine learning and multi-factor comprehensive analysis, including: obtaining cargo flow direction data from logistics management systems, port systems and transportation department channels, analyzing the data using GIS software to extract transportation flow direction, traffic density and port facility information, classifying the cargo flow direction using machine learning cluster analysis, and dividing high-frequency transportation routes and key flow directions; collecting factors affecting public water transportation and dividing them into five dimensions, including economy, timeliness, accessibility, safety and environmental protection; extracting influencing factors through principal component analysis, first standardizing the data, then calculating the covariance matrix, then performing eigenvalue decomposition, and finally selecting the principal components; calculating the comprehensive score of the influencing factors of each cargo flow direction based on the obtained main influencing factors, adjusting the influencing factors, combining the adjustment factors with actual economic, timeliness and infrastructure conditions, and dynamically revising the comprehensive score of the influencing factors to obtain a public water transportation potential value close to the actual value, and defining the final public water transportation potential value P for each cargo flow direction. j, which is calculated as follows:

[0007] P j =S j ×A j

[0008] Among them, S j is the comprehensive score of the impact factor of this flow, A j The method uses machine learning to generate simulated data, introduces random disturbances, and reduces dependence on the initial formula. The K-means clustering method is used to set the water diversion threshold, and the water suitability of the water diversion is visualized based on the GIS map. The high-potential flow direction is analyzed in combination with the heat map, and the decision classification of priority water diversion, moderate water diversion, and unsuitable water diversion is output in combination with the potential value.

[0009] As a preferred solution of the method for evaluating the public water potential based on machine learning and multi-factor comprehensive analysis described in the present invention, the cargo flow direction is classified by cluster analysis based on machine learning, including: extracting characteristics that affect the cargo flow direction, including transportation mode characteristics, flow direction characteristics, economic characteristics and environmental characteristics, and using K-Means clustering to classify the cargo flow direction, dividing it into high-frequency transportation paths and key flow directions.

[0010] As a preferred solution of the method for evaluating the potential of water transport based on machine learning and multi-factor comprehensive analysis described in the present invention, K-Means clustering is used to classify cargo flows and divide high-frequency transportation paths and key flows, including: selecting the number of clusters k, calculating the characteristic vector of each cargo flow direction, randomly initializing k cluster centers, performing iterative optimization until convergence, calculating the center of each cluster, and determining the characteristics of each cluster; based on the clustering results, screening high-frequency transportation paths, combining transportation cost and time factors, screening key water transport directions, and combining hierarchical analysis method to give the priority of each category of flow.

[0011] As a preferred solution of the method for evaluating the potential of water resources transfer based on machine learning and multi-factor comprehensive analysis described in the present invention, the priority of each category of flow direction is given in combination with the hierarchical analysis method, including:

[0012] According to the evaluation target of cargo flow priority, a three-layer decision hierarchy is constructed, including the target layer, the criterion layer, and the solution layer. The solution layer includes L1 high priority flow, L2 medium priority flow, and L3 low priority flow. According to the performance of different flow categories on each criterion, the evaluation matrix R is constructed. The matrix R is expressed as follows:

[0013]

[0014] Among them, L1 category: economy 0.9, timeliness 0.8, infrastructure 0.7, policy 0.5; L2 category: economy 0.7, timeliness 0.6, infrastructure 0.5, policy 0.3; L3 category: economy 0.4, timeliness 0.3, infrastructure 0.2, policy 0.1.

[0015] As a preferred solution of the public water transport potential assessment method based on machine learning and multi-factor comprehensive analysis described in the present invention, the influencing factors of economy include unit freight, transportation cost volatility, freight volume and logistics surcharges; the influencing factors of timeliness include transportation time, delay rate and punctuality rate; the influencing factors of accessibility include terminal distance, kilometer freight station distance and route density; the influencing factors of safety include cargo damage rate, accident rate and risk response capability; and the influencing factors of environmental protection include carbon emissions and energy consumption.

[0016] As a preferred solution of the method for evaluating the potential of water resources for planetary transformation based on machine learning and multi-factor comprehensive analysis described in the present invention, the data is first standardized, the covariance matrix is ​​calculated, the eigenvalue decomposition is performed, and finally the principal components are selected, including:

[0017] The data is standardized so that the scores of each factor have the same scale. The standardized expression is as follows:

[0018]

[0019] Among them, x is the original data, μ is the data mean, and σ is the data standard deviation;

[0020] The standardized data set is used to calculate the covariance matrix. The standardized data matrix is ​​X, and the covariance matrix calculation formula is as follows:

[0021]

[0022] Where n is the number of samples, X T is the transpose of X;

[0023] Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors. The eigenvalues ​​represent the variance of each principal component, and the eigenvectors represent the direction of each principal component. The principal components are identified through eigenvalue decomposition.

[0024] The variance contribution rate of the principal component indicates the importance of each principal component in the original data. By calculating the variance contribution rate of each principal component and arranging them in descending order, the principal component with the highest contribution rate is obtained. The calculation formula of the variance contribution rate is as follows:

[0025]

[0026] Among them, λi is the eigenvalue corresponding to the i-th principal component, ∑λ i is the sum of all eigenvalues.

[0027] As a preferred solution of the method for evaluating the potential of public transport water resources based on machine learning and multi-factor comprehensive analysis of the present invention, the comprehensive score of the influencing factors of each cargo flow direction is calculated based on the obtained principal components, including:

[0028] The calculation formula for the comprehensive impact factor score is as follows:

[0029]

[0030] Among them, the weight ω i The Score is determined by the analytic hierarchy process and expert opinion method. j is the comprehensive score of factors affecting the j-th cargo flow direction, ω i is the weight of the i-th impact factor, X ij is the value of the j-th flow direction on the i-th influencing factor, which is normalized and mapped to [0,100].

[0031] As a preferred embodiment of the method for evaluating the water transfer potential based on machine learning and multi-factor comprehensive analysis described in the present invention, the adjustment factors are combined with actual economic, timeliness, and infrastructure conditions to dynamically modify the comprehensive score of the influencing factors to obtain a water transfer potential value close to the actual value, including:

[0032] Based on the impact factor scores and combined with the dynamic adjustment factors, the orbital water potential is calculated. The expression for calculating the orbital water potential is as follows:

[0033] P j =Score j g(ΔC j ,ΔT j ,Infra j ,Policy j )

[0034] Where g(ΔC j ,ΔT j ,Infra j ,Policy j ) is a comprehensive adjustment factor used to reflect the impact of economics, timeliness, infrastructure and policy subsidies on potential;

[0035]

[0036] ΔC j It is a cost advantage. Among them, C roadjis the cost of road transport for goods flowing to j, C waterj is the cost of water transport for goods flowing to j, if ΔC j >0, then the cost advantage of water transportation is large;

[0037] ΔT j It's a time disadvantage. Among them, T roadj is the transportation time required for goods to flow to j by road, T waterj is the transportation time required for goods to flow to j by water transport, if ΔT j >0, it means that the time advantage of water transport is large;

[0038] Infra j is the adaptability of infrastructure, including port distance and port throughput capacity, Among them, D J is the port distance of the jth flow direction, Q J is the port throughput capacity of the jth flow direction, with weights of 0.7 and 0.3 respectively, max(D) is the maximum port distance among all cargo flows, and max(Q) is the maximum throughput capacity among all ports;

[0039] Policy j The impact of policy subsidies on water diversion potential, including direct subsidies, tax exemptions and policy preferences, Among them, the weights are 0.6, 0.3 and 0.1 respectively, P 1j is the direct subsidy amount for direction j, max(P1) is the highest direct subsidy amount among all directions, P 2j is the tax exemption ratio enjoyed by flow j, max(P2) is the highest tax exemption ratio among all flows, P 3j is the policy bias strength applicable to flow j, and max(P3) is the highest policy bias score among all flows.

[0040] As a preferred solution of the method for evaluating the potential of water resources in the earth's surface based on machine learning and multi-factor comprehensive analysis described in the present invention, machine learning is used to generate simulated data, random disturbances are introduced, and dependence on the initial formula is reduced, including: improving the random terms and adding the potential evaluation formula of the random terms. The potential evaluation formula is as follows:

[0041] P j =Score j g(ΔC j ,ΔT j ,Infra j ,Policy j )+ε j

[0042] Among them, the random term ε j The following rules apply: the strength of the random item is fixed and proportional to the score; high-scoring flows have superior conditions, with smaller random fluctuations in scores, and cannot be excessively disturbed by randomness; low-scoring flows have low scores themselves, allowing for greater fluctuations, simulating more uncertainty in reality;

[0043] The specific expression of random items is:

[0044]

[0045] in, is the original score calculated by the formula, η is the random item influence coefficient, which is 0.2, that is, the random item affects the score by 20% at most. It is to reduce silver and limit the fluctuation of high score, σ 2 is the preset random item strength.

[0046] As a preferred solution of the method for evaluating the potential of public water transfer based on machine learning and multi-factor comprehensive analysis described in the present invention, the water transfer threshold is set by using the K-means clustering method, the public water transfer suitability is visualized based on the GIS map, and the high potential flow direction is analyzed in combination with the heat map. Combined with the potential value, the decision classification of priority water transfer, appropriate water transfer and unsuitable water transfer is output, including: calculating the potential value P of each cargo flow direction j In K-means clustering, the sum of squares of clustering errors and the silhouette coefficient of each sample under different K values ​​are calculated; the geographic coordinate data of cargo flow directions are combined with water transport accessibility data to generate a water-suitable flow map; a heat map is used to visualize high-potential water diversion flows, showing high-frequency transportation routes, cargo flow density, and key nodes with high water transport potential; based on the K-means clustering results, the decision is made in combination with the hierarchical analysis method: P ≥ 80 is priority for water diversion, 50 ≤ P < 80 is moderate water diversion, and P < 50 is not suitable for water diversion.

[0047] The beneficial effects of the present invention are as follows: the present invention proposes a method of utilizing simulation data generation combined with machine learning optimization, which solves the problem of lack of historical data in water diversion potential assessment, while ensuring the scientificity and practicality of the data generation process. Dynamic adjustment factors are introduced for model construction, and the evaluation results are closer to actual needs. Through the improved random item design, the fairness and rationality of the scoring results are balanced, and the deviation of high-potential or low-potential samples due to random items is avoided. Based on the potential score predicted by the model, GIS software is used for visualization analysis to intuitively display the distribution and characteristics of high-potential flow directions, clearly distinguish between flow directions suitable for water diversion and those not suitable for water diversion, and facilitate quick decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0049] Figure 1 A flow chart of a method for evaluating the potential of water resources for planetary transformation based on machine learning and multi-factor comprehensive analysis provided in an embodiment of the present invention.

[0050] Figure 2 A K-Means clustering cargo flow classification flow chart of the public water potential assessment method based on machine learning and multi-factor comprehensive analysis provided in an embodiment of the present invention.

[0051] Figure 3 A flow chart of extracting influencing factors through principal component analysis in a method for evaluating water potential based on machine learning and multi-factor comprehensive analysis provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0055] Example

[0056] Reference below Figure 1 , Figure 1 This is a flow chart of a method for evaluating the potential of water resources for planetary transformation based on machine learning and multi-factor comprehensive analysis, provided in one embodiment of the present invention. It should be noted that the embodiments of the present invention can be applied to any applicable scenario.

[0057] Figure 1 The process of the method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis provided by one embodiment of the present invention includes:

[0058] S1: Obtain cargo flow data from logistics management systems, port systems, and transportation department channels, use GIS software to analyze the data, extract transportation flow direction, traffic density, and port facility information, and use machine learning cluster analysis to classify cargo flow directions and divide high-frequency transportation routes and key flow directions.

[0059] Preferably, the features that affect the flow of goods are extracted, including transportation mode features, flow features, economic features and environmental features, and K-Means clustering is used to classify the flow of goods and divide them into high-frequency transportation routes and key flows.

[0060] Preferably, the number of clusters k is selected, the characteristic vector of each cargo flow direction is calculated, k cluster centers are randomly initialized, and iterative optimization is performed until convergence, the center of each cluster is calculated, and the characteristics of each cluster are determined;

[0061] Based on the clustering results, high-frequency transportation routes are screened, and the key public water flow directions are screened in combination with transportation cost and time factors. Combined with the hierarchical analysis method, the priority of each category of flow direction is given.

[0062] Preferably, according to the cargo flow priority evaluation target, a three-layer decision hierarchy is constructed, including a target layer, a criterion layer, and a solution layer. The solution layer includes L1 high priority flow, L2 medium priority flow, and L3 low priority flow. According to the performance of different flow categories on each criterion, an evaluation matrix R is constructed. The matrix R is expressed as follows:

[0063]

[0064] Among them, L1 category: economy 0.9, timeliness 0.8, infrastructure 0.7, policy 0.5; L2 category: economy 0.7, timeliness 0.6, infrastructure 0.5, policy 0.3; L3 category: economy 0.4, timeliness 0.3, infrastructure 0.2, policy 0.1.

[0065] Furthermore, each cargo flow direction is represented by a feature vector. The features include transport economy, transport timeliness, transport frequency, cargo volume, port accessibility, port throughput capacity, and policy support. The features of all flows are standardized to obtain the standardized data matrix:

[0066] X={x1,x2,...,xn},xi=(xi1,xi2,...,xim)

[0067] Among them, n is the number of cargo flow directions, and m is the number of feature dimensions;

[0068] Initialize K cluster centers, randomly select K initial cluster centers μ1, μ2, ..., μk, calculate the distance from the sample to the cluster center, assign each cargo flow direction to the nearest cluster, update the cluster center, recalculate the mean of each cluster as the new center, repeat the steps until the cluster centers converge, obtain K final clusters, and calculate the characteristic mean of each cluster; through the K-Means clustering results, analyze the freight volume, transportation frequency, and economy of each cluster, screen high-frequency transportation routes, and determine the key transit water flow directions based on transportation cost and time factors.

[0069] Furthermore, the hierarchical analysis method is used to construct the judgment matrix, determine the weights of the criteria layer indicators, and construct the judgment matrix A:

[0070]

[0071] Calculate the weight vector W:

[0072] W=[w1,w2,w3,w4]=[0.45,0.26,0.18,0.11]

[0073] Among them, economic efficiency (w1) = 0.45, timeliness (w2) = 0.26, infrastructure (w3) = 0.18, and policy support (w4) = 0.11. A consistency test was performed and CR < 0.1 was met. The matrix is ​​valid;

[0074] The evaluation matrix R reflects the performance of the three types of flows, L1, L2, and L3, in various criteria:

[0075]

[0076] Among them, r ij represents the score of the i-th flow category on the j-th criterion;

[0077] According to the weight vector W and the evaluation matrix R, the comprehensive score S of L1, L2, and L3 is calculated:

[0078] S=W·R T

[0079] Calculate the priority scores for each category: S1 = 0.782, S2 = 0.582, S3 = 0.292;

[0080] Finally, the score of L1 (high priority flow) = 0.782, the score of L2 (medium priority flow) = 0.582, and the score of L3 (low priority flow) = 0.292;

[0081] Based on the scores, the priority classification standards for public water transfer are set: priority water transfer (L1): S≥0.75; moderate water transfer (L2): 0.5≤S<0.75; unsuitable water transfer (L3): S<0.5.

[0082] S2: Collect factors that affect public water resources and divide them into five dimensions, including economy, timeliness, accessibility, safety and environmental protection.

[0083] Preferably, the factors affecting economy include unit freight, transportation cost volatility, freight volume and logistics surcharges; the factors affecting timeliness include transportation time, delay rate and punctuality rate; the factors affecting accessibility include terminal distance, kilometer freight station distance and route density; the factors affecting safety include cargo damage rate, accident rate and risk response capability; and the factors affecting environmental protection include carbon emissions and energy consumption.

[0084] S3: Extract influencing factors through principal component analysis. First, standardize the data, then calculate the covariance matrix, perform eigenvalue decomposition, and finally select the principal components.

[0085] Preferably, the data is standardized so that the scores of each factor have the same scale. The standardized expression is as follows:

[0086]

[0087] Among them, x is the original data, μ is the data mean, and σ is the data standard deviation;

[0088] The standardized data set is used to calculate the covariance matrix. The standardized data matrix is ​​X, and the covariance matrix calculation formula is as follows:

[0089]

[0090] Where n is the number of samples, X T is the transpose of X;

[0091] Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors. The eigenvalues ​​represent the variance of each principal component, and the eigenvectors represent the direction of each principal component. The principal components are identified through eigenvalue decomposition.

[0092] The variance contribution rate of the principal component indicates the importance of each principal component in the original data. By calculating the variance contribution rate of each principal component and arranging them in descending order, the principal component with the highest contribution rate is obtained. The calculation formula of the variance contribution rate is as follows:

[0093]

[0094] Among them, λ i is the eigenvalue corresponding to the i-th principal component, ∑λ i is the sum of all eigenvalues.

[0095] Furthermore, assuming there are five cargo flows, the data includes four indicators: economy (C1), timeliness (C2), infrastructure (C3), and policy support (C4). The original data is as follows:

[0096] Cargo flow Economy C1 Timeliness C2 Infrastructure C3 Policy support C4 A 90 75 80 50 B 85 65 70 40 C 70 60 60 30 D 60 50 50 20 E 55 45 15 15

[0097] Calculate the mean μ and standard deviation σ for each indicator:

[0098] index Economy C1 Timeliness C2 Infrastructure C3 Policy support C4 Mean μ 72.0 59.0 61.0 31.0 Standard deviation σ 14.28 12.12 13.71 13.71

[0099] Use the normalization formula to calculate the normalized data:

[0100] Cargo flow Economy C1 Timeliness C2 Infrastructure C3 Policy support C4 A 1.26 1.32 1.39 1.39 B 0.91 0.49 0.65 0.65 C -0.14 0.08 -0.07 -0.07 D -0.84 -0.74 -0.80 -0.80 E -1.19 -1.15 -1.17 -1.17

[0101] Perform eigenvalue decomposition on the covariance matrix and obtain the eigenvalues: λ1 = 3.50, λ2 = 0.40, λ3 = 0.07, λ4 = 0.03;

[0102] Calculate the contribution of each principal component:

[0103] Total eigenvalue sum = 3.50 + 0.40 + 0.07 + 0.03 = 4.00

[0104] Cumulative variance contribution PC1 + PC2 = 97.5% → The first two principal components can explain 97.5% of the information, so PC1 and PC2 can be retained as the main influencing factors;

[0105] The contribution rate of principal component 1 (PC1) is 87.5%, which is comprehensively determined by economy, timeliness, infrastructure, and policy support; the contribution rate of principal component 2 (PC2) is 10.0%, which is mainly determined by policy support and infrastructure; the contribution rates of PC3 and PC4 are relatively low (<2%) and can be ignored; principal component 1 (PC1) is the main decision factor, and the decision on the direction of the water flow can be adjusted based on the PC1 value.

[0106] S4: Based on the obtained principal components, calculate the comprehensive score of the influencing factors of each cargo flow direction, adjust the influencing factors, combine the adjustment factors with the actual economic, timeliness and infrastructure conditions, and dynamically correct the comprehensive score of the influencing factors to obtain a public water potential value close to the actual value.

[0107] Preferably, the formula for calculating the comprehensive impact factor score is as follows:

[0108]

[0109] Among them, the weight ω i The Score is determined by the analytic hierarchy process and expert opinion method. j is the comprehensive score of factors affecting the j-th cargo flow direction, ω iis the weight of the i-th impact factor, X ij is the value of the j-th flow direction on the i-th influencing factor, which is normalized and mapped to [0,100].

[0110] Preferably, the revolution water potential is calculated based on the impact factor score and combined with the dynamic adjustment factor. The expression for calculating the revolution water potential is as follows:

[0111] P j =Score j g(ΔC j ,ΔT j ,Infra j ,Policy j )

[0112] Where g(ΔC j ,ΔT j ,Infra j ,Policy j ) is a comprehensive adjustment factor used to reflect the impact of economics, timeliness, infrastructure and policy subsidies on potential;

[0113]

[0114] ΔC j It is a cost advantage. Among them, C roadj The cargo flow to j is transported by road, C waterj is the cost of water transport for goods flowing to j, if ΔC j >0, then the cost advantage of water transportation is large;

[0115] ΔT j It's a time disadvantage. Among them, T roadj is the transportation time required for goods to flow to j by road, T waterj is the transportation time required for goods to flow to j by water transport, if ΔT j <0, it means that the time advantage of water transportation is large;

[0116] Infra j is the adaptability of infrastructure, including port distance and port throughput capacity, Among them, D J is the port distance of the jth flow direction, Q J is the port throughput capacity of the jth flow direction, with weights of 0.7 and 0.3 respectively, max(D) is the maximum port distance among all cargo flows, and max(Q) is the maximum throughput capacity among all ports;

[0117] Policy jThe impact of policy subsidies on water diversion potential, including direct subsidies, tax exemptions and policy preferences, Among them, the weights are 0.6, 0.3 and 0.1 respectively, P 1j is the direct subsidy amount for direction j, max(P1) is the highest direct subsidy amount among all directions, P 2j is the tax exemption ratio enjoyed by flow j, max(P2) is the highest tax exemption ratio among all flows, P 3j is the policy bias strength applicable to flow j, and max(P3) is the highest policy bias score among all flows.

[0118] Furthermore, the calculation of the earth-to-water potential is based on the scores of the impact factors and the dynamic adjustment factors. The potential calculation formula is as follows:

[0119] P j =C·(0.7×(C cost )+0.2×(C time )+0.1×(C infra ))·(1+P policy )

[0120] Among them, P j is the public water transfer potential of the jth cargo flow, C is a comprehensive adjustment factor, which is used to reflect the impact of economy, timeliness, infrastructure and policy subsidies on potential, C cost is a cost advantage. If the cost advantage is large, then C cost The value is large, C time is a time disadvantage. If the time disadvantage is large, then C time The value is large, C infra is the infrastructure adaptability, including port distance and throughput capacity, P polic is the impact of policy subsidies. If the subsidies are large, then P policy The value is larger.

[0121] Assume that the dynamic factors of flow direction A are as follows:

[0122] C cost =0.8, C time =0.5, C infra =0.7, P policy =0.4;

[0123] Substitute into the potential formula to calculate:

[0124] PA=1.2·(0.7×0.8+0.2×0.5+0.1×0.7)·(1+0.4)

[0125] PA=1.2·(0.56+0.10+0.07)·1.4=1.2·0.73·1.4=1.226

[0126] At this time, the water potential of flow direction A is 1.226. The same calculation is performed on other flow directions to obtain their potential values ​​and prioritize them.

[0127] S5: Use machine learning to generate simulated data, introduce random perturbations, and reduce dependence on the initial formula.

[0128] Preferably, the random item is improved and a potential evaluation formula of the random item is added. The potential evaluation formula is as follows:

[0129] P j =Score j g(ΔC j ,ΔT j ,Infra j ,Policy j )+ε j

[0130] Among them, the random term ε j The following rules apply: the strength of the random item is fixed and proportional to the score; high-scoring flows have superior conditions, with smaller random fluctuations in scores, and cannot be excessively disturbed by randomness; low-scoring flows have lower scores themselves, allowing for greater fluctuations, simulating more uncertainty in reality;

[0131] The specific expression of random items is:

[0132]

[0133] in, is the original score calculated by the formula, η is the random item influence coefficient, which is 0.2, that is, the random item affects the score by 20% at most. It is to reduce silver and limit the fluctuation of high scores, and σ is the preset random item strength.

[0134] Further, the code is as follows:

[0135]

[0136]

[0137]

[0138] (2) Data training

[0139] The data is divided into training set, validation set and test set with the proportions of 70%, 20% and 10% respectively, and random forest is used for data training. The code is as follows:

[0140]

[0141]

[0142]

[0143] S6: The K-means clustering method is used to set the water diversion threshold, and the water suitability of the public water diversion is visualized based on the GIS map. The high-potential flow direction is analyzed in combination with the heat map. Combined with the potential value, the decision classification of priority water diversion, appropriate water diversion and unsuitable water diversion is output.

[0144] Preferably, calculate the potential score P of each cargo flow direction j In K-means clustering, the sum of squares of clustering errors and the silhouette coefficient of each sample under different K values ​​are calculated; the geographic coordinate data of cargo flow directions are combined with water transport accessibility data to generate a water-suitable flow map; a heat map is used to visualize high-potential water diversion flows, showing high-frequency transportation routes, cargo flow density, and key nodes with high water transport potential; based on the K-means clustering results, the decision is made in combination with the hierarchical analysis method: P ≥ 80 is priority for water diversion, 50 ≤ P < 80 is moderate water diversion, and P < 50 is not suitable for water diversion.

[0145] Furthermore, assuming the potential score P of the following cargo flow direction j :

[0146]

[0147]

[0148] According to the decision criteria, the water flow is classified into the following categories: flow direction A (P≥80): priority water diversion, flow direction B (50≤P<80): moderate water diversion, flow direction C (50≤P<80): moderate water diversion, flow direction D (P<50): not suitable for water diversion, and flow direction E (P<50): not suitable for water diversion.

[0149] Furthermore, using geographic coordinate data and water transport accessibility data, the starting and ending points of each cargo flow are plotted on a geographic map, and the water transport potential score is superimposed. Each flow is color-coded according to its potential score to form a heat map, with high-potential flows displayed in hot red and low-potential flows in cool blue. Using the matplotlib library, a heat map with potential scores is generated:

[0150]

[0151]

[0152] In the description of the present invention, it should be noted that the terms "first", "second" and "third" are only used for descriptive purposes and should not be understood as indicating or implying relative importance.

[0153] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0154] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. There may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interface, indirect coupling or communication connection of devices or units, which may be electrical, mechanical or other forms.

[0155] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0156] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0157] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0158] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

[0159] Furthermore, although the operations of the method of the present invention are described in a particular order in the accompanying drawings, this does not require or imply that these operations must be performed in this particular order, or that all illustrated operations must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

Claims

1. A method for evaluating the potential of water resources for planetary transformation based on machine learning and multi-factor comprehensive analysis, characterized in that: include: Cargo flow data is obtained from logistics management systems, port systems, and transportation departments. GIS software is used to analyze the data, extracting information on transportation flow direction, traffic density, and port facilities. Machine learning cluster analysis is used to classify cargo flows and identify high-frequency transportation routes and key flow directions. Collect factors that affect public water resources and divide them into five dimensions, including economy, timeliness, accessibility, safety and environmental protection; The influencing factors were extracted by principal component analysis, which first normalized the data, then calculated the covariance matrix, performed eigenvalue decomposition, and finally selected the principal components. Based on the obtained principal components, the comprehensive score of the influencing factors of each cargo flow direction is calculated, the influencing factors are adjusted, and the adjustment factors are combined with the actual economic, timeliness and infrastructure conditions to dynamically correct the comprehensive score of the influencing factors to obtain a value close to the actual public water potential value, and define the final public water potential value P for each cargo flow direction. j , which is calculated as follows: P j =S j ×A j Among them, S j is the comprehensive score of the impact factor of this flow, A j is the actual adjustment factor for the flow direction; Use machine learning to generate simulated data, introduce random perturbations, and reduce reliance on the initial formula; The K-means clustering method was used to set the water diversion threshold, and the water suitability of the public water diversion was visualized based on the GIS map. The high-potential flow directions were analyzed in combination with the heat map. Combined with the potential value, the decision classification of priority water diversion, appropriate water diversion and unsuitable water diversion was output.

2. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The cluster analysis using machine learning to classify cargo flows includes: The characteristics that affect the flow of goods are extracted, including transportation mode characteristics, flow characteristics, economic characteristics and environmental characteristics, and K-Means clustering is used to classify the flow of goods and divide them into high-frequency transportation routes and key flow directions.

3. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 2, wherein: The K-Means clustering is used to classify the cargo flow, dividing the high-frequency transportation routes and key flow directions, including: Select the number of clusters k, calculate the characteristic vector of each cargo flow direction, randomly initialize k cluster centers, perform iterative optimization until convergence, calculate the center of each cluster, and determine the characteristics of each cluster; Based on the clustering results, high-frequency transportation routes are screened, and the key public water flow directions are screened in combination with transportation cost and time factors. Combined with the hierarchical analysis method, the priority of each category of flow direction is given.

4. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 3, wherein: The above method combines the hierarchical analysis method to give the priority of each category flow, including: According to the evaluation target of cargo flow priority, a three-layer decision hierarchy is constructed, including the target layer, the criterion layer, and the solution layer. The solution layer includes L1 high priority flow, L2 medium priority flow, and L3 low priority flow. According to the performance of different flow categories on each criterion, the evaluation matrix R is constructed. The matrix R is expressed as follows: Among them, L1 category: economy 0.9, timeliness 0.8, infrastructure 0.7, policy 0.5; L2 category: economy 0.7, timeliness 0.6, infrastructure 0.5, policy 0.3; L3 category: economy 0.4, timeliness 0.3, infrastructure 0.2, policy 0.

1.

5. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The factors affecting economy include unit freight, transportation cost volatility, freight volume and logistics surcharges; the factors affecting timeliness include transportation time, delay rate and punctuality rate; the factors affecting accessibility include terminal distance, kilometer freight station distance and route density; the factors affecting safety include cargo damage rate, accident rate and risk response capability; the factors affecting environmental protection include carbon emissions and energy consumption.

6. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The data is first standardized, the covariance matrix is ​​calculated, the eigenvalue decomposition is performed, and finally the principal components are selected, including: The data is standardized so that the scores of each factor have the same scale. The standardized expression is as follows: Among them, x is the original data, μ is the data mean, and σ is the data standard deviation; The standardized data set is used to calculate the covariance matrix. The standardized data matrix is ​​X, and the covariance matrix calculation formula is as follows: Where n is the number of samples, X T is the transpose of X; Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors. The eigenvalues ​​represent the variance of each principal component, and the eigenvectors represent the direction of each principal component. The principal components are identified through eigenvalue decomposition. The variance contribution rate of the principal component indicates the importance of each principal component in the original data. By calculating the variance contribution rate of each principal component and arranging them in descending order, the principal component with the highest contribution rate is obtained. The calculation formula of the variance contribution rate is as follows: Among them, λ i is the eigenvalue corresponding to the i-th principal component, ∑λ i is the sum of all eigenvalues.

7. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: Based on the obtained principal components, the comprehensive score of the influencing factors of each cargo flow direction is calculated, including: The calculation formula for the comprehensive impact factor score is as follows: Among them, the weight ω i The Score is determined by the analytic hierarchy process and expert opinion method. j is the comprehensive score of the factors affecting the j-th cargo flow direction, ω i is the weight of the i-th impact factor, X ij is the value of the j-th flow direction on the i-th influencing factor, which is normalized and mapped to [0,100].

8. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The adjustment factors are combined with the actual economic, timeliness and infrastructure conditions to dynamically modify the comprehensive score of the influencing factors to obtain a value close to the actual public water potential, including: Based on the impact factor score and combined with the dynamic adjustment factor, the orbital water potential is calculated. The expression for calculating the orbital water potential value is as follows: P j =Score j ·g(ΔC j ,ΔT j ,Infra j ,Policy j ) Where g(ΔC j ,ΔT j ,Infra j ,Policy j ) is a comprehensive adjustment factor used to reflect the impact of economics, timeliness, infrastructure and policy subsidies on potential; ΔC j It is a cost advantage. Among them, C roadj is the cost of road transport for goods flowing to j, C waterj is the cost of water transport for goods flowing to j, if ΔC j >0, then the cost advantage of water transportation is large; ΔT j It's a time disadvantage. Among them, T roadj is the transportation time required for goods to flow to j by road, T waterj is the transportation time required for goods to flow to j by water transport, if ΔT j >0, it means that the time advantage of water transportation is large; Infra j is the adaptability of infrastructure, including port distance and port throughput capacity, Among them, D J is the port distance of the jth flow direction, Q J is the port throughput capacity of the jth flow direction, with weights of 0.7 and 0.3 respectively, max(D) is the maximum port distance among all cargo flows, and max(Q) is the maximum throughput capacity among all ports; Policy j The impact of policy subsidies on water diversion potential, including direct subsidies, tax exemptions and policy preferences, Among them, the weights are 0.6, 0.3 and 0.1 respectively, P 1j is the direct subsidy amount for direction j, max(P1) is the highest direct subsidy amount among all directions, P 2j is the tax exemption ratio enjoyed by flow j, max(P2) is the highest tax exemption ratio among all flows, P 3j is the policy bias strength applicable to flow j, and max(P3) is the highest policy bias score among all flows.

9. The method for evaluating the potential of water resources for revolution based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The method of using machine learning to generate simulated data, introducing random disturbances, and reducing dependence on the initial formula includes: The random item is improved and a potential evaluation formula for the random item is added. The potential evaluation formula is as follows: P j =Score j ·g(ΔC j ,ΔT j ,Infra j ,Policy j )+ε j Among them, the random term ε j The following rules apply: the strength of the random item is fixed and proportional to the score; high-scoring flows have superior conditions, with smaller random fluctuations in scores, and cannot be excessively disturbed by randomness; low-scoring flows have low scores themselves, allowing for greater fluctuations, simulating more uncertainty in reality; The specific expression of random items is: in, is the original score calculated by the formula, η is the random item influence coefficient, which is 0.2, that is, the random item affects the score by 20% at most. It is to reduce silver and limit the fluctuation of high score, σ 2 is the preset random item strength.

10. The method for evaluating the potential of water resources for planetary transformation based on machine learning and multi-factor comprehensive analysis according to claim 1, wherein: The K-means clustering method is used to set the water diversion threshold, and the water suitability of the public water diversion is visualized based on the GIS map. The high-potential flow direction is analyzed in combination with the heat map. Combined with the potential value, the decision classification of priority water diversion, moderate water diversion and unsuitable water diversion is output, including: Calculate the potential value P of each cargo flow direction j , calculate the sum of squares of clustering errors and the silhouette coefficient of each sample under different K values ​​in K-means clustering; combine the geographic coordinate data of cargo flow directions and overlay water transport accessibility data to generate a water-suitable flow map; use heat maps to visualize high-potential water transfer flows, showing high-frequency transportation routes, cargo flow density, and key nodes with high water transport potential; Based on the K-means clustering results, the decision was made in combination with the analytic hierarchy process: P ≥ 80 was considered as priority water transfer, 50 ≤ P < 80 was considered as moderate water transfer, and P < 50 was considered as unsuitable water transfer.

Images