Vehicle scheduling optimization method and system based on real-time position and energy consumption

Abstract

The invention relates to the technical field of data processing, in particular to a vehicle scheduling optimization method and system based on real-time position and energy consumption, and the method comprises the steps: obtaining multi-dimensional data; obtaining a high-dimensional chromosome sequence based on the multi-dimensional data; arranging the high-dimensional chromosome sequences in rows, and constructing an analysis matrix; according to the cosine similarity between the column vector in the analysis matrix and other column vectors, determining a scheduling correlation quantity corresponding to the column vector; obtaining the element isolation value of the high-dimensional chromosome sequence by using the isolation value of each element in the column vector; based on the scheduling correlation quantity and the element isolation value, determining the scheduling characteristic correlation quantity of each gene locus in the high-dimensional chromosome sequence; according to the scheduling feature correlation quantity, obtaining a PCA dimension reduction weight of each gene locus in the high-dimensional chromosome sequence; a PCA dimensionality reduction weight is used to acquire an analysis matrix after dimensionality reduction; and obtaining an optimal scheduling scheme based on the analysis matrix after dimension reduction. According to the method, the calculation complexity of real-time scheduling is greatly reduced through dimension reduction.

Description

A Vehicle Dispatch Optimization Method and System Based on Real-Time Location and Energy Consumption Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a vehicle scheduling optimization method and system based on real-time location and energy consumption. Background Technology

[0002] Currently, vehicle dispatching in logistics, public transportation, and other sectors is growing rapidly, leading to continuously rising operating costs. Energy consumption is a significant component of these costs, and reducing energy consumption while ensuring transportation efficiency has become a key focus for the industry. With the widespread adoption of IoT and vehicle-to-everything (V2X) technologies, vehicles can transmit real-time data on location, speed, load, and road gradient during operation. This dynamic data enables more precise vehicle dispatching—theoretically, the dispatching system can dynamically plan the optimal route and speed based on the vehicle's current real-time location and predicted energy consumption, thereby achieving the dual goals of "energy saving + high efficiency."

[0003] In practical applications, such scheduling optimization problems typically require the use of intelligent algorithms. Genetic algorithms are one of the most representative, simulating a natural evolutionary process of "survival of the fittest" to gradually select the optimal solution from numerous possible path options. In scheduling scenarios based on real-time location and energy consumption, researchers usually encode factors such as path selection and velocity planning into the "chromosomes" of the genetic algorithm and calculate the energy cost of each path based on a real-time energy consumption model, using this as a basis for evaluating the merits of the solutions.

[0004] However, existing genetic algorithms still face a practical challenge when dealing with dynamic scheduling problems: because each scheduling scheme needs to comprehensively consider multiple dimensions of information such as path, energy consumption, duration, load, and gradient, the encoding of the "chromosomes" is extremely complex. When the system needs to evaluate the fitness of hundreds or thousands of candidate schemes, each scheme requires complex calculations with real-time data, resulting in a huge overall computational load and long computation time. In real-time scheduling environments that require fast response, this computational delay often affects the timeliness of scheduling, making it difficult to implement theoretically optimal schemes in practical applications. Therefore, how to optimize genetic algorithms to reduce their computational complexity under high-dimensional dynamic data while ensuring the quality of scheduling schemes has become a technical problem that needs to be solved. Summary of the Invention

[0005] This invention provides a vehicle scheduling optimization method and system based on real-time location and energy consumption to solve existing problems.

[0006] The vehicle scheduling optimization method and system based on real-time location and energy consumption of the present invention adopts the following technical solution: One embodiment of the present invention provides a vehicle scheduling optimization method based on real-time location and energy consumption, which includes the following steps: acquiring multi-dimensional data in a vehicle scheduling scenario; wherein, the multi-dimensional data includes path information, vehicle information, and delivery point information; acquiring high-dimensional chromosome sequences based on the multi-dimensional data; arranging the acquired multiple high-dimensional chromosome sequences by rows to construct an n×m analysis matrix; wherein, n is the number of high-dimensional chromosome sequences, and m is the length of the high-dimensional chromosome sequences; determining the cosine similarity of each column vector in the analysis matrix with each other column vector. The process involves: determining the scheduling correlation quantity corresponding to the quantity; obtaining the element isolation value of each high-dimensional chromosome sequence by using the isolated value of each element in each column vector of the analysis matrix; determining the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence based on the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence; obtaining the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence based on the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence; obtaining the dimensionality-reduced analysis matrix using the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence; and obtaining the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

[0007] Furthermore, the path information is road network data represented in the form of a distance matrix or a real-time road network map, the vehicle information includes the number of vehicles, the real-time location of each vehicle and the real-time remaining energy, and the delivery point information includes the location of the task point, the time window of the task, and the service duration of the task.

[0008] Furthermore, the specific steps for obtaining the high-dimensional chromosome sequence based on multidimensional data are as follows: arranging all delivery points according to the vehicle access order, and using the number "0" as a demarcation point or path separator to distinguish the paths of different vehicles, forming a basic path chromosome sequence; wherein, the sequence between two adjacent "0"s corresponds to the access order of a vehicle; based on the basic path chromosome sequence, adding the encoding information corresponding to the multidimensional data affecting vehicle scheduling, expanding the basic path chromosome sequence into a high-dimensional chromosome sequence.

[0009] Furthermore, the specific steps for determining the scheduling-related quantity corresponding to each column vector based on the cosine similarity between each column vector in the analysis matrix and each other column vector are as follows: obtaining the column vectors formed by each column in the analysis matrix, and calculating the cosine similarity between each column vector and each other column vector; summing the absolute values ​​of the cosine similarities between each column vector and each other column vector and taking the average value to obtain the scheduling-related quantity corresponding to each column vector.

[0010] Furthermore, the specific steps for obtaining the element isolation value of each high-dimensional chromosome sequence by utilizing the isolated values ​​of each element in each column vector of the analysis matrix are as follows: using the isolated forest method to obtain the isolated value of each element in each column vector, and using the isolated value of each element as the isolated value at each gene locus in each high-dimensional chromosome sequence; summing the isolated values ​​at each gene locus in each high-dimensional chromosome sequence and taking the average value to obtain the element isolation value of the high-dimensional chromosome sequence.

[0011] Furthermore, the specific steps for determining the scheduling characteristic correlation quantity of each gene locus in each high-dimensional chromosome sequence based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence are as follows: taking the scheduling correlation quantity corresponding to each column vector as the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence; dividing the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence by the element isolation value of the high-dimensional chromosome sequence to obtain the scheduling characteristic correlation quantity of each gene locus in the high-dimensional chromosome sequence.

[0012] Furthermore, the specific steps for obtaining the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence based on the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence are as follows: the normalized value of the reciprocal of the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence is determined as the PCA dimensionality reduction weight of each gene locus in the high-dimensional chromosome sequence.

[0013] Furthermore, the specific steps for obtaining the dimensionality-reduced analysis matrix by utilizing the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence are as follows: multiply the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence by each data in the analysis matrix to obtain the weighted analysis matrix; perform PCA dimensionality reduction on the weighted analysis matrix to obtain the dimensionality-reduced analysis matrix.

[0014] Further, the specific steps for obtaining the optimal scheduling scheme based on the dimensionality-reduced analysis matrix are as follows: Obtain the original high-dimensional chromosome sequence and its original scheduling scheme corresponding to each chromosome sequence in the dimensionality-reduced analysis matrix; determine whether each original scheduling scheme meets the scheduling feasibility condition: if it does, retain the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; if it does not, discard the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; wherein, the scheduling feasibility condition is that the remaining energy-capable distance of the vehicle is greater than or equal to the total path length of the scheduling scheme, and the remaining allowable driving time of the vehicle is greater than or equal to the scheduling scheme's... The total time required for the task is determined as follows: An ideal point is defined in the space of the reduced-dimensional analysis matrix. The Euclidean distance between the projection of each retained original high-dimensional chromosome sequence in the space of the reduced-dimensional analysis matrix and the ideal point is calculated, and the reciprocal of this Euclidean distance is used as the fitness value of the chromosome sequence. Based on the fitness value, selection, crossover, and mutation operations are performed on the retained original high-dimensional chromosome sequences to generate a new generation of original high-dimensional chromosome sequences. The generation of a new generation of original high-dimensional chromosome sequences is iteratively generated until the preset maximum number of iterations or the fitness value converges. The original high-dimensional chromosome sequence with the highest fitness value in the final population is decoded to obtain the optimal scheduling scheme.

[0015] One embodiment of the present invention provides a vehicle dispatch optimization system based on real-time location and energy consumption. The system includes the following modules: a data acquisition module for acquiring multidimensional data in a vehicle dispatch scenario; wherein the multidimensional data includes route information, vehicle information, and delivery point information; a data analysis module for acquiring high-dimensional chromosome sequences based on the multidimensional data; arranging the acquired multiple high-dimensional chromosome sequences in rows to construct an n×m analysis matrix; where n is the number of high-dimensional chromosome sequences and m is the length of the high-dimensional chromosome sequences; determining the dispatch-related quantity corresponding to each column vector based on the cosine similarity between each column vector and every other column vector in the analysis matrix; and utilizing the analysis... The isolation values ​​of each element in each column vector of the matrix are used to obtain the element isolation value of each high-dimensional chromosome sequence. Based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence, the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence is determined. According to the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence, the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence is obtained. Using the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence, the dimensionality-reduced analysis matrix is ​​obtained. The scheduling scheme acquisition module is used to obtain the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

[0016] The beneficial effects of the technical solution of this invention are as follows: This invention proposes a vehicle scheduling optimization method and system based on real-time location and energy consumption. For vehicle scheduling scenarios involving real-time location and energy consumption, a high-dimensional chromosome integrating multi-dimensional data is constructed. Dimensionality reduction weights are determined based on the correlation between chromosomes and the abnormal characteristics of each gene locus, achieving weighted dimensionality reduction of the high-dimensional chromosome. This allows fitness calculation to be completed quickly in a low-dimensional space, while the evolutionary operation of the genetic algorithm is still performed on the original chromosome, supplemented by feasibility screening. Compared with existing technologies, this invention significantly reduces the computational complexity of real-time scheduling through dimensionality reduction. Furthermore, it consciously preserves key abnormal characteristics such as sudden road conditions and energy consumption mutations during the dimensionality reduction process, avoiding their smoothing out as noise. Simultaneously, the combination of feasibility screening and evolution in the original space ensures the practical executability and global approximate optimality of the scheduling scheme, thereby guaranteeing the quality of the scheduling scheme. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 is a flowchart of the vehicle scheduling optimization method based on real-time location and energy consumption according to the present invention; Figure 2 is a block diagram of the vehicle scheduling optimization system based on real-time location and energy consumption according to the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the vehicle scheduling optimization method and system based on real-time location and energy consumption proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the vehicle scheduling optimization method and system based on real-time location and energy consumption provided by this invention.

[0022] It should be noted that this invention relates to a "vehicle scheduling optimization method and system based on real-time location and energy consumption," aiming to achieve real-time optimization of vehicle scheduling schemes in dynamic environments. Its technical approach includes: constructing a precise energy consumption model at the bottom layer that integrates multi-source dynamic data, providing an energy consumption calculation basis for scheduling decisions; and developing an intelligent algorithm at the upper layer that supports "trajectory-speed" collaborative optimization and dynamic "spatiotemporal matching," enabling dynamic generation and adjustment of scheduling schemes. Vehicle scheduling is a complex combinatorial optimization problem, and the core algorithm can employ intelligent optimization methods such as genetic algorithms, particle swarm optimization, and ant colony optimization. Among these, genetic algorithms, by simulating the natural evolutionary process for global optimization, are particularly suitable for such problems. In the scenario of this invention, the design of the genetic algorithm needs to integrate key elements such as real-time vehicle location and dynamic energy consumption models into every step, including encoding, fitness evaluation, and evolutionary operations, to achieve efficient scheduling optimization based on real-time location and energy consumption.

[0023] When applying genetic algorithms to vehicle scheduling scenarios based on real-time location and energy consumption, the key lies in: the encoding method must simultaneously express path selection and speed level to support "trajectory-speed" collaborative optimization; the fitness function must accurately invoke the dynamic energy consumption model and effectively handle real-time constraints such as time windows and remaining energy. Furthermore, to address environmental dynamism, a rolling optimization framework is typically employed, with each optimization based on the latest real-time location, road conditions, and energy data. In addition, genetic algorithms are easily combined with local search algorithms such as simulated annealing and tabu search to form hybrid algorithms, thereby accelerating convergence and improving solution quality.

[0024] When optimizing vehicle scheduling scenarios using genetic algorithms, the first step is to encode chromosomes based on actual optimization parameters (such as path selection and path energy consumption) to construct an initial population and establish a fitness function. The fitness function is used to standardize and evaluate chromosomes corresponding to different paths, serving as the basis for selection or elimination. Subsequently, through evolutionary operations such as crossover and mutation, the optimal scheduling path based on the genetic algorithm is iteratively obtained. In this process, many factors influence chromosome dimensionality, mainly including path code, basic energy consumption, path duration, load, and gradient, resulting in high chromosome encoding dimensionality. When each chromosome needs to perform spatial distance calculations with the fitness function, significant computational overhead occurs, thus affecting the overall efficiency of scheduling optimization.

[0025] To address the computational efficiency issues arising from high-dimensional chromosomes, this invention performs feature analysis on the scheduling environment, controls the dimensionality reduction of the generated chromosomes, and executes genetic computation based on the dimensionality reduction results to efficiently obtain optimized vehicle scheduling results. The specific implementation steps are as follows: First, acquire scene-dimensional data, including path information, number of vehicles, delivery point information, remaining vehicle energy, and real-time location, and perform path-based chromosome encoding to obtain high-dimensional chromosomes; Second, acquire several high-dimensional chromosomes, determine the weights for PCA dimensionality reduction by analyzing the horizontal scheduling feature correlation between chromosomes; calculate local fitness based on the dimensionality reduction results, input the fitness selection results into the fitness function, perform genetic algorithm evolution operations, and output an approximately optimal scheduling scheme after 500 iterations.

[0026] Please refer to Figure 1, which shows a flowchart of a vehicle scheduling optimization method based on real-time location and energy consumption provided by an embodiment of the present invention. The method includes the following steps: Step S001: Obtain multi-dimensional data in the vehicle scheduling scenario; wherein, the multi-dimensional data includes route information, vehicle information and delivery point information.

[0027] Route information is road network data represented in the form of a distance matrix or a real-time road network map. Vehicle information includes the number of vehicles, the real-time location of each vehicle, and the real-time remaining energy. Delivery point information includes the location of the task point, the time window of the task, and the service duration of the task.

[0028] It should be noted that: the distance matrix is ​​a table that records the fixed driving distance between any two locations and is used for basic path calculation in static scenarios; the real-time road network map is dynamically updated map data that includes dynamic information such as real-time traffic conditions, congestion levels, and traffic light waiting times, and is used to more accurately calculate the actual driving time and energy consumption between two points.

[0029] The real-time location of each vehicle refers to its real-time latitude and longitude or coordinates; the real-time remaining energy refers to the amount of electricity or fuel in each vehicle.

[0030] The task location in the delivery point information refers to the specific coordinates or address of each delivery location, the time window refers to the time period during which delivery is possible as requested by the customer, and the service duration refers to the dwell time required for the vehicle to complete unloading or handover after arrival.

[0031] Step S002: Obtain high-dimensional chromosome sequences based on multidimensional data.

[0032] It should be noted that after obtaining the above multidimensional data, each feasible vehicle scheduling scheme needs to be represented as a "chromosome" that can be operated on by the genetic algorithm.

[0033] Step S002 further includes steps S0021 and S0022: Step S0021: Arrange all delivery points according to the order of vehicle access, and use the number "0" as a parking lot or path separator to distinguish the paths of different vehicles, forming a basic path chromosome sequence; wherein, the sequence between two adjacent "0"s corresponds to the access order of a vehicle.

[0034] In this embodiment, a path-based coding method is adopted, and the specific steps are as follows: First, all delivery points (task points) to be visited are arranged in the order in which they are visited by the vehicles. In order to distinguish the driving paths of different vehicles, the number "0" is introduced as a special separator. This number usually represents a parking lot and also serves as the start and end point markers of each vehicle path.

[0035] Thus, a complete scheduling scheme is encoded as an integer sequence (i.e., a chromosome), which consists of several subsequences, with each subsequence between two adjacent "0"s corresponding to a vehicle's access path. The dimension of this chromosome is equal to "total number of delivery points + number of vehicles + 1".

[0036] For example, in a scenario with 2 delivery vehicles and 5 delivery points (numbered 1 to 5), the following is a chromosome encoding result: [0, 1, 3, 2, 0, 4, 5, 0].

[0037] The decoding meaning of this chromosome is as follows: Vehicle 1: Starts from the parking lot (first 0), visits delivery point 1, delivery point 3, delivery point 2 in sequence, and finally returns to the parking lot (second 0); Vehicle 2: Starts from the parking lot (second 0), visits delivery point 4, delivery point 5 in sequence, and finally returns to the parking lot (third 0).

[0038] In the example above, the total number of delivery points is 5, and the number of vehicles is 2. Therefore, the chromosome dimension is 5+2+1=8, which is consistent with the sequence length. This encoding method fully preserves the task allocation and access order information of vehicles in the scheduling scheme.

[0039] It should be noted that the above steps only construct a basic chromosome containing path information. In subsequent steps, the encoding information corresponding to the speed level of each road segment and other multi-dimensional feature parameters (such as load, gradient, etc.) will be further added to this basic chromosome to form a high-dimensional chromosome for subsequent optimization calculations.

[0040] Step S0022: Based on the basic path chromosome sequence, add the encoding information corresponding to the multi-dimensional data affecting vehicle scheduling, and expand the basic path chromosome sequence into a high-dimensional chromosome sequence.

[0041] In this embodiment, speed levels are used as an example for explanation: For the basic path chromosome sequence [0, 1, 3, 2, 0, 4, 5, 0] in the aforementioned example, its corresponding travel segments include: 0→1, 1→3, 3→2, 2→0, 0→4, 4→5, 5→0, a total of 7 segments. A speed level is added to each segment, for example, 1 represents low speed, 2 represents medium speed, and 3 represents high speed. The speed code can then be appended to the basic path chromosome sequence to form a high-dimensional chromosome sequence in the following form: [0, 1, 3, 2, 0, 4, 5, 0 | v1, v2, v3, v4, v5, v6, v7].

[0042] In this system, v1 corresponds to the speed level from 0 to 1, v2 corresponds to the speed level from 1 to 3, and so on, with v7 corresponding to the speed level from 5 to 0. By introducing speed information, the chromosome dimension increases from the original 8 dimensions to 15 dimensions (8 path gene loci plus 7 speed gene loci), thus carrying richer optimization information and enabling the genetic algorithm to perform collaborative optimization of "trajectory-speed".

[0043] It should be noted that speed level is only one example of multidimensional feature parameters. In practical applications, other parameters can be added to the chromosome as needed, such as the energy consumption estimate, load variation, and gradient coefficient for each road segment. Simply append the encoded values ​​of these parameters to the basic chromosome in the same way to form a chromosome with more dimensions to adapt to different optimization objectives.

[0044] Step S003: Arrange the obtained high-dimensional chromosome sequences in rows to construct an n×m analysis matrix; where n is the number of high-dimensional chromosome sequences and m is the length of the high-dimensional chromosome sequences.

[0045] It should be noted that numerous factors influence the optimization effect during vehicle scheduling, primarily including route information, vehicle information, and delivery point information. These factors, after encoding, are transformed into computationally achievable data sequences, resulting in a large amount of information involved in the optimization. To achieve precise scheduling in a dynamic environment, it is necessary to comprehensively analyze the real-time status information of different vehicles and generate the optimal scheduling scheme accordingly. Genetic algorithms, as an intelligent method suitable for complex combinatorial optimization problems, can solve for near-optimal scheduling based on the current state. However, because vehicle scheduling involves multiple factors such as route, energy consumption, duration, load, and gradient, the chromosome encoding has a high dimensionality, which can easily lead to the "curse of dimensionality" during fitness calculation, resulting in high computational overhead and low optimization efficiency.

[0046] After generating several high-dimensional chromosomes (denoted as...) Each chromosome corresponds to a complete scheduling scheme, and its sequence contains multi-dimensional information such as path, speed, time, and energy consumption. In subsequent fitness calculation and selection, if distance measurement is performed directly in the high-dimensional space, the computational load will increase sharply with the increase of dimensionality, which can easily reduce scheduling efficiency and solution accuracy. Therefore, this embodiment introduces PCA dimensionality reduction technology to compress the high-dimensional chromosomes.

[0047] Traditional PCA dimensionality reduction typically prioritizes data based on its global variance. However, in vehicle scheduling scenarios, location and energy consumption have higher priority. Therefore, this embodiment designs a weighted PCA dimensionality reduction method by performing feature analysis on chromosome sequence data, enabling the dimensionality reduction process to better preserve information closely related to the scheduling objective.

[0048] In vehicle dispatching, multiple indicators are often driven by the same underlying factors and exhibit strong correlations. For example: path length is positively correlated with travel time (under stable average speed); travel time is positively correlated with energy consumption (increased electric motor or engine running time); load is positively correlated with energy consumption (extra mass increases driving resistance); road gradient affects both travel time and energy consumption.

[0049] The aforementioned linear or nonlinear relationships indicate that the actual intrinsic dimension of the original index space is low. By reducing dimensionality, the original index can be approximated by a few comprehensive variables (principal components), thus replacing high-dimensional distance with low-dimensional distance in fitness calculation, significantly reducing computational complexity, especially suitable for fitness functions involving complex weighting or multiple distance metrics.

[0050] Based on this, it is necessary to... A cross-sectional analysis of chromosomes is performed to uncover the correlations between various indicators, and a scheduling correlation quantity is calculated for each element (i.e., each gene locus) in the chromosome sequence as the basis for subsequent weighted dimensionality reduction.

[0051] Specifically, assuming the current population contains A series of high-dimensional chromosome sequences, each with a gene length of [length missing]. (That is, each scheduling scheme is composed of) (Encoding a gene locus). This... A series of high-dimensional chromosome sequences are arranged in rows to form a Analysis matrix .

[0052] Taking seven high-dimensional chromosome sequences, each containing several genes, as an example, the analysis matrix can be represented as follows: In this matrix: each row represents a chromosome (i.e., a scheduling scheme), recording the specific values ​​of each gene locus under that scheme; each column represents the values ​​of all chromosomes at the same gene locus, reflecting the distribution of that gene locus in the current population. By constructing this analysis matrix, correlation analysis can be further performed on the column vectors to calculate the cosine similarity between gene loci, providing a data foundation for subsequent weighted PCA dimensionality reduction.

[0053] Step S004: Determine the scheduling-related quantity corresponding to each column vector based on the cosine similarity between each column vector in the analysis matrix and every other column vector.

[0054] It should be noted that: in the same analysis matrix, the first chromosome of different chromosomes... Each gene locus has the same physical meaning, such as representing a route code, speed level, energy consumption estimate, or load information for a specific road segment.

[0055] Step S004 further includes steps S0041 and S0042: Step S0041: Obtain the column vector formed by each column in the analysis matrix, and calculate the cosine similarity between each column vector and each other column vector.

[0056] Step S0042: Sum the absolute values ​​of the cosine similarity between each column vector and every other column vector and take the mean to obtain the scheduling-related quantity corresponding to each column vector.

[0057] Analysis of the first The chromosome sequence element (the th chromosome sequence element in the analysis matrix) Scheduling-related quantities corresponding to all elements in the list: ;in, Represents the first element in the analysis matrix. The scheduling-related quantities corresponding to each column vector. Represents the first element in the analysis matrix. column vectors, Represents the first element in the analysis matrix. column vectors, To represent the cosine similarity between two column vectors, summation is performed by iterating through the vectors. to That is, calculate the first The absolute value of the cosine similarity between a column and all subsequent columns is calculated, and then the average is taken.

[0058] First, for the summation part, the higher the cosine similarity (the closer to 1), the stronger the positive or negative correlation, indicating that the two columns of elements are numerically and physically correlated. For example, the two columns of elements might be path length and time taken. Generally, when the speed is constant, the path length and time taken by a vehicle are positively correlated. Therefore, when analyzing changes, either one can be chosen to represent the physical standard of correlation.

[0059] At the same time, the cumulative similarity of multiple columns of elements is summed and averaged. The closer the average result is to 1, the more it proves that the changing trends of multiple chromosome data are related to it. Therefore, the dimensionality reduction weight is lower when performing dimensionality reduction. The purpose is to retain the unrelated parts with data characteristics, while only one column of such highly correlated data needs to be retained.

[0060] Therefore, the closer this value is to 1, the higher the scheduling correlation, the higher the corresponding scheduling feature correlation, and the lower the PCA dimensionality reduction weight.

[0061] Step S005: Using the isolated values ​​of each element in each column vector of the analysis matrix, obtain the element isolation value of each high-dimensional chromosome sequence.

[0062] It should be noted that in real-world vehicle dispatching scenarios, the complex and ever-changing operating environment often presents nonlinear factors that deviate from the predictions of ideal models, such as real-time traffic fluctuations, sudden congestion, and severe weather. These variables are difficult to fully represent using conventional linear modeling, yet they significantly impact the feasibility and energy efficiency accuracy of the dispatching scheme. Therefore, this embodiment does not simply compress all variables uniformly during dimensionality reduction; instead, it consciously retains nonlinear characteristics that reflect these specific operating conditions, avoiding the loss of meaningful dispatching constraints due to excessive dimensionality reduction. By retaining these key variables, the optimization results better reflect the real-world operating environment, thereby improving the practicality and robustness of the dispatching scheme.

[0063] In real-world scheduling scenarios, there are various nonlinear disturbances. Taking traffic light waiting time as an example: the manifestation is that vehicles stop at intersections due to red lights, resulting in additional waiting time and energy consumption during the startup phase.

[0064] Characteristics: This variable is periodic but uncertain, and is affected by both the real-time signal timing scheme and the current traffic queue length. Its impact on time is manifested as discrete waiting time (usually on the order of tens of seconds), while its impact on energy consumption includes continuous consumption during idling and additional losses caused by frequent start-stop.

[0065] Severe weather (such as rain, snow, and fog) is also a typical nonlinear disturbance factor: it manifests as slippery roads and reduced visibility, leading to decreased vehicle speed and frequent braking, and may also cause local road congestion.

[0066] Characteristics: It is highly sudden and has a wide coverage, affecting all relevant road sections in the road network; it will not only prolong travel time, but also significantly increase energy consumption, such as increased rolling resistance in rainy or snowy weather, and additional energy consumption from turning on the air conditioner or defogging function.

[0067] Therefore, the correlation of scheduling features depends not only on the correlation between different gene loci, but also on the degree of abnormality of the values ​​of each gene within the chromosome. For example, due to the aforementioned nonlinear perturbation factors, data such as path length and travel time in some scheduling schemes may exhibit sudden anomalies. These outliers often reflect special circumstances in the real environment and need to be preserved during the optimization process. To this end, this embodiment further calculates the isolation value of each element in the chromosome sequence to quantify the degree of abnormality of the values ​​of each gene locus, serving as another important basis for determining the dimensionality reduction weights.

[0068] Step S005 further includes steps S0051 and S0052: Step S0051: The isolated value of each element in each column vector is obtained by using the isolated forest method, and the isolated value of each element is used as the isolated value of each gene locus in each high-dimensional chromosome sequence.

[0069] It should be noted that the Isolation Forest algorithm is a method for anomaly detection that randomly partitions the data space and measures the ease with which data points are isolated. Its core idea is that outliers are often isolated with only a few random partitions because they are sparsely distributed and significantly different from normal data.

[0070] When using the Isolation Forest method to obtain the isolated values ​​of each element in each column vector, each column in the analysis matrix is ​​treated as an independent data sample set, and anomaly detection is performed on all elements in that column. Specifically, for each column, a segmentation feature and segmentation value are randomly selected, and the data points in that column are recursively assigned to left and right child nodes, constructing multiple isolation trees. Because outliers have the characteristics of "sparse distribution and significant differences from normal data," they tend to have shorter path lengths in the isolation trees (i.e., they are isolated more quickly). By calculating the average path length of each data point in all isolation trees and normalizing it to between 0 and 1, the isolated value of that element can be obtained. The closer the value is to 1, the more anomalous the element; the closer it is to 0, the more normal the element.

[0071] The Isolation Forest algorithm is a well-known technique, and will not be elaborated on here.

[0072] Step S0052: Sum the isolated values ​​at each gene locus in each high-dimensional chromosome sequence and take the average to obtain the element isolation value of the high-dimensional chromosome sequence.

[0073] Specifically, Analysis matrix ,in For chromosome number, This represents the chromosome length. For each column... (i.e., each gene locus), the isolated forest algorithm is used to calculate the isolated value of each element in this column, resulting in... isolated value matrix ,in Indicates the first The high-dimensional chromosome sequence in the first Isolated values ​​at individual gene loci.

[0074] Based on this, the elemental isolation values ​​of the high-dimensional chromosome sequence are calculated: ;in, Indicates the first The elemental isolation value of a high-dimensional chromosome sequence.

[0075] By summing the isolated values ​​of each chromosome at each gene locus, the higher the sum, the greater the likelihood that the chromosome has anomalous values ​​at multiple gene loci, meaning the scheduling scheme contains more special working condition information. In this case, the scheduling feature correlation of the chromosome is relatively low, so its corresponding PCA dimensionality reduction weight should be increased accordingly to retain these anomalous features that reflect real environmental changes during dimensionality reduction and avoid the loss of effective information.

[0076] Step S006: Based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence, determine the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence.

[0077] Step S006 further includes steps S0061 and S0062: Step S0061: The scheduling correlation quantity corresponding to each column vector is used as the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence.

[0078] Step S0062: Divide the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence by the element isolation value of the high-dimensional chromosome sequence to obtain the scheduling feature correlation quantity at each gene locus in the high-dimensional chromosome sequence.

[0079] Specifically, the correlation coefficient of scheduling features for each gene locus in the high-dimensional chromosome sequence is calculated as follows: ;in, Indicates the first In the high-dimensional chromosome sequence, the first The correlation quantity of scheduling characteristics at each gene locus; it is directly proportional to the scheduling correlation quantity and inversely proportional to the element isolation.

[0080] Step S007: Based on the scheduling feature association of each gene locus in each high-dimensional chromosome sequence, obtain the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence.

[0081] Specifically, this includes: determining the normalized value of the inverse of the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence as the PCA dimensionality reduction weight of each gene locus in that high-dimensional chromosome sequence.

[0082] Specifically, the PCA dimensionality reduction weights for each gene locus in the high-dimensional chromosome sequence are calculated as follows: ;in, Indicates the first In the high-dimensional chromosome sequence, the first PCA dimensionality reduction weights at each gene locus This indicates normalization.

[0083] Step S008: Use the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence to obtain the dimensionality-reduced analysis matrix.

[0084] Step S008 further includes steps S0081 and S0082: Step S0081: Multiply the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence by each data in the analysis matrix to obtain the weighted analysis matrix.

[0085] Step S0082: Perform PCA dimensionality reduction on the weighted analysis matrix to obtain the dimensionality-reduced analysis matrix.

[0086] Specifically, after obtaining the PCA dimensionality reduction weights for each gene locus in each high-dimensional chromosome sequence, the weights are first multiplied element-wise with the original analysis matrix to obtain a weighted analysis matrix. Then, standard principal component analysis (PCA) is performed on the weighted analysis matrix to reduce its dimensionality. This is done by calculating its covariance matrix, solving for eigenvalues ​​and eigenvectors, and selecting the top... The principal component directions corresponding to the largest eigenvalues ​​are used to project the original data into this low-dimensional space, ultimately obtaining a dimensionality-reduced analysis matrix. This matrix significantly reduces the data dimensionality while preserving key scheduling features. PCA dimensionality reduction is a well-known technique and will not be elaborated upon here.

[0087] Step S009: Obtain the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

[0088] It should be noted that after obtaining the PCA dimensionality reduction weights for each gene locus in each high-dimensional chromosome sequence, these weights are used to perform weighted PCA dimensionality reduction on the original analysis matrix to obtain the dimensionality-reduced chromosome results. This dimensionality reduction result effectively filters out some highly correlated redundant elements while retaining the main information of the original data, and consciously preserves key gene loci carrying unique information or anomalous features. The dimensionality-reduced chromosomes will serve as data samples for fitness selection and calculation in subsequent genetic algorithms, providing support for efficient iterative evolution.

[0089] Step S009 further includes steps S0091-S0095: Step S0091: Obtain the original high-dimensional chromosome sequence and its original scheduling scheme corresponding to each chromosome sequence in the dimension-reduced analysis matrix.

[0090] It should be noted that after obtaining the dimensionality-reduced analysis matrix, each row in the dimensionality-reduced matrix (i.e., each dimensionality-reduced chromosome sequence) can be associated with the corresponding high-dimensional chromosome sequence in the original population through the maintained index mapping relationship. Specifically, before dimensionality reduction, each original high-dimensional chromosome retains a unique identifier or index; during dimensionality reduction, this index is recorded along with the data or maintains the row order. Therefore, for any row in the dimensionality-reduced matrix, the corresponding original high-dimensional chromosome sequence can be directly retrieved by backtracking based on its row number. Then, through decoding operations, the complete scheduling scheme represented by that chromosome can be obtained, including the path order of each vehicle, task allocation, speed level, and other multi-dimensional feature parameters.

[0091] Step S0092: Determine whether each original scheduling scheme meets the scheduling feasibility conditions: if it does, retain the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; if it does not, discard the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; wherein, the scheduling feasibility conditions are that the remaining energy available for the vehicle to travel is greater than or equal to the total path length of the scheduling scheme, and the remaining allowable travel time of the vehicle is greater than or equal to the total time required by the scheduling scheme.

[0092] It should be noted that after obtaining the dimensionality-reduced chromosome results, it is necessary to assess the feasibility of the corresponding original scheduling scheme, i.e., to introduce local fitness screening.

[0093] Local fitness refers to the logical anomalies that occur after dimensionality reduction of some chromosomes. For example, if the remaining energy in a chromosome after dimensionality reduction is insufficient to support the completion of a path, then the scheme corresponding to that chromosome is not schedulable and needs to be eliminated.

[0094] Therefore, the following criteria are used to screen the options: ;in, The range that a vehicle can travel with its remaining energy. This represents the total path length of the scheduling scheme; This represents the total time required for the plan. This is the maximum allowed duration. Only when both of the above conditions are met is the scheduling scheme corresponding to the chromosome considered feasible and proceed to the subsequent fitness calculation stage; otherwise, it is discarded.

[0095] The viable chromosomes retained after screening are used as the local fitness screening results and input into the fitness function of the genetic algorithm to drive subsequent evolutionary operations such as selection, crossover, and mutation.

[0096] Step S0093: Define an ideal point in the space of the dimension-reduced analysis matrix, calculate the Euclidean distance between the projection of each original high-dimensional chromosome sequence in the space of the dimension-reduced analysis matrix and the ideal point, and use the reciprocal of the Euclidean distance as the fitness value of the chromosome sequence.

[0097] It should be noted that after completing the local fitness screening, the fitness of the retained feasible chromosomes is calculated. Specifically, a fitness function is defined in the reduced-dimensional space, which uses the reciprocal of the Euclidean distance between the chromosome and the preset ideal point as the fitness value. Since the chromosome is now represented in the lower-dimensional space, the complexity of distance calculation is significantly reduced compared to the original high-dimensional space, thereby effectively improving the computational efficiency of the genetic algorithm in each generation of evolution.

[0098] Step S0094: Based on the fitness value, perform selection, crossover, and mutation operations on the preserved original high-dimensional chromosome sequence to generate a new generation of original high-dimensional chromosome sequences.

[0099] It should be noted that, based on the calculated fitness values, a selection strategy (such as roulette wheel selection or tournament selection) is used to retain superior individuals from the current population. Subsequently, crossover and mutation operations are performed on the original high-dimensional chromosomes corresponding to the selected superior individuals to generate a new generation of feasible scheduling schemes. It is important to emphasize that the dimensionality-reduced chromosomes are only used for fitness evaluation and do not participate in genetic operations to ensure the structural integrity and practical feasibility of the newly generated individuals.

[0100] Step S0095: Iteratively generate a new generation of original high-dimensional chromosome sequences until the preset maximum number of iterations or fitness value convergence is reached. Decode the original high-dimensional chromosome sequence with the highest fitness value in the final population to obtain the optimal scheduling scheme.

[0101] It should be noted that the fitness calculation and evolutionary operation steps described above are repeated until the preset maximum number of iterations is reached or the fitness converges. In each generation iteration, the original multidimensional indicators of the newly generated individuals must be recalculated and dimensionality reduction processed again to ensure the real-time performance and accuracy of subsequent evaluations.

[0102] After the iteration terminates, the individual with the highest fitness is selected from the final population, and its corresponding original high-dimensional chromosome is decoded to obtain specific vehicle paths, task access order, and speed suggestions for each road segment. This result is the approximate optimal scheduling scheme based on real-time location and energy consumption.

[0103] Thus, the efficient vehicle scheduling optimization based on genetic algorithms was completed, satisfying the dual constraints of real-time location and energy consumption in a dynamic environment.

[0104] Please refer to Figure 2, which shows a block diagram of a vehicle dispatch optimization system based on real-time location and energy consumption according to an embodiment of the present invention. The system includes the following modules: a data acquisition module 100, used to acquire multidimensional data in a vehicle dispatch scenario; wherein the multidimensional data includes route information, vehicle information, and delivery point information; a data analysis module 200, used to acquire high-dimensional chromosome sequences based on the multidimensional data; arrange the acquired multiple high-dimensional chromosome sequences by rows to construct an n×m analysis matrix; where n is the number of high-dimensional chromosome sequences and m is the length of the high-dimensional chromosome sequences; determine the dispatch corresponding to each column vector based on the cosine similarity between each column vector and every other column vector in the analysis matrix. The system employs several methods: 1) Analyzing the scheduling correlation of each gene locus in a high-dimensional chromosome sequence by using the isolated values ​​of each element in each column vector of the analysis matrix; 2) Determining the scheduling feature correlation of each gene locus in each high-dimensional chromosome sequence based on the scheduling feature correlation of each gene locus; 3) Obtaining the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence based on the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence; 4) Obtaining the dimensionality-reduced analysis matrix using the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence; and 5) A scheduling scheme acquisition module 300 is used to obtain the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

[0105] In summary, during the vehicle scheduling process based on location and energy consumption, the optimal scheduling result is obtained through a genetic algorithm. However, due to the high dimensionality of the scheduling reference, this embodiment improves the computational efficiency and accuracy of the algorithm for vehicle scheduling by solving the high-dimensional problem, thereby achieving the goal of real-time and high-efficiency scheduling.

[0106] Because the chromosome dimension is high during the algorithm's decomposition process, high-dimensional distance calculations arise during fitness selection, leading to increased computational load and reduced scheduling efficiency. This embodiment analyzes the lateral scheduling association features of chromosomes to determine PCA dimensionality reduction weights, aiming to improve scheduling efficiency while preserving important chromosome features.

[0107] This embodiment determines the local fitness by iteratively screening chromosomes longitudinally, obtains the corresponding fitness screening results, and then determines the optimal approximate result.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A vehicle scheduling optimization method based on real-time location and energy consumption, characterized in that, The method includes the following steps: acquiring multidimensional data in a vehicle scheduling scenario; wherein the multidimensional data includes route information, vehicle information, and delivery point information; acquiring high-dimensional chromosome sequences based on the multidimensional data; arranging the acquired multiple high-dimensional chromosome sequences by rows to construct an n×m analysis matrix; wherein n is the number of high-dimensional chromosome sequences and m is the length of the high-dimensional chromosome sequences; determining the scheduling correlation quantity corresponding to each column vector based on the cosine similarity between each column vector and each other column vector in the analysis matrix; obtaining the element isolation value of each high-dimensional chromosome sequence using the isolation value of each element in each column vector of the analysis matrix; determining the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence; obtaining the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence based on the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence; obtaining the dimensionality-reduced analysis matrix using the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence; and obtaining the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

2. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The path information is road network data represented in the form of a distance matrix or a real-time road network map. The vehicle information includes the number of vehicles, the real-time location of each vehicle, and the real-time remaining energy. The delivery point information includes the location of the task point, the time window of the task, and the service duration of the task.

3. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for obtaining a high-dimensional chromosome sequence based on multidimensional data are as follows: Arrange all delivery points according to the vehicle access order, and use the number "0" as a demarcation point or path separator to distinguish the paths of different vehicles, forming a basic path chromosome sequence; wherein, the sequence between two adjacent "0"s corresponds to the access order of a vehicle; based on the basic path chromosome sequence, add the encoding information corresponding to the multidimensional data affecting vehicle scheduling, and expand the basic path chromosome sequence into a high-dimensional chromosome sequence.

4. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for determining the scheduling-related quantity corresponding to each column vector based on the cosine similarity of each column vector in the analysis matrix with each other column vector are as follows: obtain the column vectors formed by each column in the analysis matrix, and calculate the cosine similarity of each column vector with each other column vector; sum the absolute values ​​of the cosine similarities of each column vector with each other column vector and take the average value to obtain the scheduling-related quantity corresponding to each column vector.

5. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for obtaining the element isolation value of each high-dimensional chromosome sequence by using the isolated values ​​of each element in each column vector of the analysis matrix are as follows: using the isolated forest method to obtain the isolated value of each element in each column vector, and using the isolated value of each element as the isolated value of each gene locus in each high-dimensional chromosome sequence; summing the isolated values ​​of each gene locus in each high-dimensional chromosome sequence and taking the average value to obtain the element isolation value of the high-dimensional chromosome sequence.

6. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The method for determining the scheduling characteristic correlation of each gene locus in each high-dimensional chromosome sequence based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence includes the following specific steps: taking the scheduling correlation quantity corresponding to each column vector as the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence; dividing the scheduling correlation quantity at each gene locus in each high-dimensional chromosome sequence by the element isolation value of the high-dimensional chromosome sequence to obtain the scheduling characteristic correlation quantity at each gene locus in the high-dimensional chromosome sequence.

7. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for obtaining the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence based on the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence are as follows: the normalized value of the inverse of the scheduling feature association quantity of each gene locus in each high-dimensional chromosome sequence is determined as the PCA dimensionality reduction weight of each gene locus in the high-dimensional chromosome sequence.

8. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for obtaining the dimensionality-reduced analysis matrix by utilizing the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence are as follows: multiply the PCA dimensionality reduction weights of each gene locus in each high-dimensional chromosome sequence by each data point in the analysis matrix to obtain the weighted analysis matrix; perform PCA dimensionality reduction on the weighted analysis matrix to obtain the dimensionality-reduced analysis matrix.

9. The vehicle scheduling optimization method based on real-time location and energy consumption according to claim 1, characterized in that, The specific steps for obtaining the optimal scheduling scheme based on the dimensionality-reduced analysis matrix are as follows: Obtain the original high-dimensional chromosome sequence and its original scheduling scheme corresponding to each chromosome sequence in the dimensionality-reduced analysis matrix; determine whether each original scheduling scheme meets the scheduling feasibility condition: if it does, retain the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; if it does not, discard the original scheduling scheme and its corresponding original high-dimensional chromosome sequence; wherein, the scheduling feasibility condition is that the remaining energy available for the vehicle to travel is greater than or equal to the total path length of the scheduling scheme, and the remaining allowable travel time of the vehicle is greater than or equal to the time required by the scheduling scheme. Total duration; Define an ideal point in the space of the dimensionality-reduced analysis matrix, calculate the Euclidean distance between the projection of each original high-dimensional chromosome sequence in the space of the dimensionality-reduced analysis matrix and the ideal point, and use the reciprocal of the Euclidean distance as the fitness value of the chromosome sequence; Based on the fitness value, perform selection, crossover and mutation operations on the retained original high-dimensional chromosome sequences to generate a new generation of original high-dimensional chromosome sequences; Iterate to generate a new generation of original high-dimensional chromosome sequences until the preset maximum number of iterations or the fitness value converges, decode the original high-dimensional chromosome sequence with the highest fitness value in the final population, and obtain the optimal scheduling scheme.

10. A vehicle dispatching optimization system based on real-time location and energy consumption, characterized in that, The system includes the following modules: a data acquisition module for acquiring multidimensional data in vehicle dispatching scenarios, including route information, vehicle information, and delivery point information; a data analysis module for acquiring high-dimensional chromosome sequences based on the multidimensional data; arranging the acquired high-dimensional chromosome sequences row-wise to construct an n×m analysis matrix, where n is the number of high-dimensional chromosome sequences and m is the length of each high-dimensional chromosome sequence; determining the dispatch-related quantity corresponding to each column vector based on the cosine similarity of each column vector in the analysis matrix with every other column vector; and utilizing the isolated values ​​of each element in each column vector of the analysis matrix. The system obtains the element isolation value of each high-dimensional chromosome sequence; based on the scheduling correlation quantity corresponding to each column vector in the analysis matrix and the element isolation value of each high-dimensional chromosome sequence, it determines the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence; based on the scheduling feature correlation quantity of each gene locus in each high-dimensional chromosome sequence, it obtains the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence; using the PCA dimensionality reduction weight of each gene locus in each high-dimensional chromosome sequence, it obtains the dimensionality-reduced analysis matrix; the scheduling scheme acquisition module is used to obtain the optimal scheduling scheme based on the dimensionality-reduced analysis matrix.

Images