A method and system for integrating mobile intelligent logistics operations

Abstract

The present invention relates to the field of smart logistics technology, and discloses a method and system for mobile terminal smart logistics operation integration. The method for mobile terminal smart logistics operation integration comprises the following steps: step S101, obtaining M different destinations through statistical analysis; step S102, obtaining cargo information of the cargo to be delivered at the M different destinations; step S103, obtaining vehicle information of N vehicles to be dispatched; step S104, obtaining road condition information of the M different destinations, and performing preprocessing to generate a feature sequence; step S105, obtaining road condition scores of the M different destinations through a road condition prediction model; step S106, generating a scheduling plan through an improved tuna swarm optimization algorithm; the present invention predicts and scores the road conditions in the future time period through a road condition prediction model, and generates a scheduling plan in combination with the improved tuna swarm optimization algorithm, thereby completing the unified scheduling of multiple logistics vehicles, thereby improving the distribution efficiency.

Description

Technical Field

[0001] The present invention relates to the field of smart logistics technology, and more specifically, to a method and system for integrating smart logistics operations on a mobile terminal. Background Art

[0002] Traditional logistics distribution usually adopts the method of regional centralized distribution, that is, the distribution area is divided according to the destination of the goods to be delivered. Each distribution area is responsible for a specific warehouse or distribution center. The goods are transported from the central warehouse to the sorting center in each area. After classification and packaging, the delivery personnel drive the vehicle to deliver the goods to the customer. However, traditional logistics distribution mainly relies on manual decision-making, and there may be a situation where the vehicle loading rate is not high, resulting in low efficiency of logistics distribution.

[0003] With the development of artificial intelligence and autonomous driving technology, driverless vehicles are gradually being used in logistics distribution. Driverless vehicles can operate around the clock and are almost not restricted by working hours, which can improve the efficiency of logistics distribution to a certain extent. However, the above solution still lacks the unified scheduling of multiple delivery vehicles at the same time. For example, multiple delivery vehicles may be concentrated in the same area for delivery at the same time, resulting in a high degree of overlap in the delivery routes in the area, resulting in low delivery efficiency.

[0004] Therefore, there is an urgent need for an intelligent logistics operation method to solve the above problems. Summary of the invention

[0005] The present invention provides a method and system for integrating smart logistics operations on mobile terminals to solve the technical problems in the above-mentioned background technology.

[0006] The present invention provides a method for integrating mobile terminal intelligent logistics operations, comprising the following steps:

[0007] Step S101, obtaining M different destinations through statistical analysis according to the destinations of the goods to be delivered;

[0008] Step S102, obtaining cargo information of cargo to be delivered to M different destinations;

[0009] Cargo information includes: unique number, weight, volume and delivery time;

[0010] Step S103, obtaining vehicle information of N vehicles to be dispatched in the distribution center;

[0011] Vehicle information includes: unique number, maximum load weight, maximum load volume, total mileage, average fuel consumption, maximum speed and historical maintenance times;

[0012] Step S104, within a first preset time period T1, traffic information of M different destinations is obtained at preset time intervals t, and preprocessed to generate a feature sequence;

[0013] The feature sequence of the mth destination includes R sequence units, and the rth sequence unit represents the pre-processed road condition information of the mth destination at the rth time point, where 1≤m≤M, 1≤r≤R, and R=T1 / t;

[0014] Traffic information includes: distance to the distribution center, road conditions, traffic events, number of traffic lights, temperature, rainfall, wind speed and visibility;

[0015] Step S105, inputting the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2;

[0016] The road condition score ranges from 0 to 10, and the larger the road condition score, the better the road condition;

[0017] Step S106, generating a dispatching plan by using an improved tuna swarm optimization algorithm according to the vehicle information of the N vehicles to be dispatched, the cargo information of the cargo to be delivered at M different destinations, and the road condition scores within a preset time period T2;

[0018] According to the scheduling plan, the goods to be delivered at M different destinations are allocated to N vehicles to be scheduled.

[0019] Furthermore, the number M of different destinations, the number N of vehicles to be dispatched, the first preset time period T1, the preset time interval t and the second preset time period T2 are all custom parameters, and the unique number of the goods to be delivered and the unique number of the vehicle to be dispatched are both represented by self-increasing positive integer codes.

[0020] Furthermore, the cargo information of the cargo to be delivered is stored in a database, and statistical analysis is performed through a query tool provided by the database to obtain different destinations.

[0021] Further, the road condition information is preprocessed, including the following steps:

[0022] Step S201, converting road conditions and traffic events into numerical representations;

[0023] Road conditions include good, average, and poor, represented by positive integers 3, 2, and 1 respectively;

[0024] Traffic incidents are represented by the number of traffic incidents that occurred;

[0025] Step S202: For missing values ​​in the traffic condition information, interpolation filling is performed by taking the average value of the traffic condition information of two adjacent time points of the missing value;

[0026] Step S203: normalize the traffic information using a Z-Score method.

[0027] Furthermore, the road condition prediction model includes: a first hidden layer, a concatenated layer, R second hidden layers and a first classifier;

[0028] The first hidden layer inputs the feature sequence and outputs the global vector;

[0029] The concatenation layer is used to concatenate the global vector output by the first hidden layer with each sequence unit of the feature sequence to obtain a combined vector;

[0030] The r-th second hidden layer inputs the combination vector corresponding to the r-th sequence unit of the feature sequence and outputs the update vector;

[0031] The second hidden layer is built based on gated recurrent network units;

[0032] The update vector output by the Nth second hidden layer is input to the first classifier, and the classification space of the first classifier represents the road condition score within the second preset time period T2;

[0033] The training of the traffic condition prediction model includes: pre-training and formal training;

[0034] During the pre-training process, the update vector output by the Nth second hidden layer is input into the second classifier and the third classifier, and the classification spaces of the second classifier and the third classifier respectively represent the road condition and the temperature in the second preset time period T2;

[0035] During the formal training process, the sample labels of the training samples used to train the road condition prediction model are obtained through manual annotation.

[0036] Furthermore, the calculation formula of the first hidden layer includes:

[0037] ;

[0038] ;

[0039] ;

[0040] in represents the global vector output by the first hidden layer, seq represents the feature sequence of the first hidden layer input, W represents the weight parameter, b represents the bias parameter, d represents the number of dimensions of a sequence unit of the feature sequence of the first hidden layer input, and the value is 8. and denote the first intermediate matrix and the second intermediate matrix respectively, and represent the first weight parameter and the second weight parameter corresponding to the first intermediate matrix respectively, and They respectively represent the first weight parameter and the second weight parameter corresponding to the second intermediate matrix, T represents a transpose operation, concat represents a concatenation operation, cos represents a cosine function, sin represents a sine function, and Swish represents a Swish activation function.

[0041] Furthermore, a scheduling scheme is generated by using an improved tuna swarm optimization algorithm, including the following steps:

[0042] Step S301, randomly generate an initialization population consisting of K individuals that meet the constraints, and initialize the current number of iterations to 1;

[0043] Where K is a custom parameter;

[0044] The constraints include:

[0045] Constraint 1: The total weight of the goods to be delivered in a single transport cannot exceed the preset load threshold of the vehicle to be dispatched, where the preset load threshold is a custom parameter;

[0046] Constraint 2: The total volume of the goods to be delivered in a single transport cannot exceed the preset load volume threshold of the vehicle to be dispatched, where the preset load volume threshold is a custom parameter;

[0047] Constraint 3: The number of vehicles to be dispatched for a single transport of goods to the same destination cannot exceed the preset vehicle threshold, where the preset vehicle threshold is a custom parameter;

[0048] Constraint 4: The number of different destinations that the same dispatched vehicle can transport simultaneously cannot exceed the preset destination number threshold, where the destination number threshold is a custom parameter;

[0049] Constraint 5: Load the goods to be delivered in ascending order of their arrival time;

[0050] The individual is represented by a matrix code of M rows and N columns. The element value of the matrix code is represented by 0 or 1. The element value of the mth row and nth column is 0, which means that the goods to be delivered at the mth destination are not allocated to the nth vehicle to be dispatched. Otherwise, it means that the goods to be delivered at the mth destination are allocated to the nth vehicle to be dispatched.

[0051] Step S302, calculating the fitness values ​​of K individuals in the initialization population through the objective function;

[0052] Step S303, calculate the balance coefficient according to the current number of iterations and the maximum number of iterations, and judge that the balance coefficient is greater than or equal to the balance coefficient threshold, then update the matrix codes of K individuals in the initialized population through the parabolic foraging strategy, otherwise proceed to step S304;

[0053] The calculation formula of the balance coefficient Balance is as follows:

[0054] ;

[0055] Where t means the current iteration number is t, Indicates the maximum number of iterations;

[0056] The maximum number of iterations and the balance coefficient threshold are both custom parameters;

[0057] Step S304, updating the matrix codes of random K / 2 individuals in the initialized population by using the first spiral foraging strategy, and updating the matrix codes of the remaining K / 2 individuals in the initialized population by using the second spiral foraging strategy;

[0058] Step S305, respectively calculating the average value of the element value of the matrix code of each individual in the initialized population, and respectively determining that if the element value of the matrix code of each individual is greater than or equal to the corresponding average value, then modify the element value to 1, otherwise modify it to 0;

[0059] Step S306, if it is determined that the iteration termination condition is met, the matrix code of the individual with the largest fitness value in the initialized population is output as the scheduling solution, otherwise the current iteration number is accumulated by 1, and the process returns to step S302 to continue execution;

[0060] The iteration termination conditions include: the current iteration number is greater than or equal to the maximum iteration number; within three consecutive iterations, the change in the maximum fitness value in the initialized population is less than the change threshold, where the change threshold is a custom parameter.

[0061] Furthermore, the calculation formula for calculating the individual fitness value Fit through the objective function is as follows:

[0062] ;

[0063] in The element value of the mth row and nth column of the matrix encoding of the individual is represented by: , They represent the distance between the mth destination and the nth vehicle to be dispatched and the road condition score within the preset time period T2, respectively. and They represent the maximum speed and average fuel consumption of the nth vehicle to be dispatched when the goods to be delivered at the mth destination are assigned to the nth vehicle to be dispatched. and Respectively represent the custom weight coefficients, and and The sum of is 1.

[0064] Furthermore, the calculation formula of the improved tuna school optimization algorithm includes:

[0065] The calculation formula of the parabolic foraging strategy includes:

[0066] ;

[0067] ;

[0068] Where 1≤t≤ , 1≤i≤K, and They represent the matrix codes of the i-th individual whose current iteration number is t and whose current iteration number is t+1, respectively. Represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, represents the first random number in the range of 0 to 1, TF represents a random number between 1 and -1, and p represents a nonlinear decreasing coefficient;

[0069] The calculation formula for the first spiral foraging strategy includes:

[0070] ;

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] in , , They represent the first decision coefficient, the second decision coefficient and the third decision coefficient respectively, f represents the foraging factor, e represents the natural constant, and Respectively represent the second random number and the third random number whose values ​​range from 0 to 1;

[0076] The calculation formula of the second spiral foraging strategy is as follows:

[0077] ;

[0078] in Represents the matrix encoding of a random individual with the current iteration number t.

[0079] The present invention provides a system for integrating mobile terminal intelligent logistics operations, comprising:

[0080] A destination statistics module is used to obtain M different destinations through statistical analysis according to the destinations of the goods to be delivered;

[0081] A cargo information collection module, which is used to obtain cargo information of cargo to be delivered to M different destinations;

[0082] The vehicle information collection module is used to obtain the vehicle information of N vehicles to be dispatched in the distribution center;

[0083] A feature sequence generation module, which is used to obtain the road condition information of M different destinations at a preset time interval t within a first preset time period T1, and perform preprocessing to generate a feature sequence;

[0084] A traffic score prediction module, which is used to input the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2;

[0085] The scheduling plan generation module is used to generate a scheduling plan through an improved tuna school optimization algorithm according to the vehicle information of N vehicles to be scheduled, the cargo information of M cargoes to be delivered at different destinations, and the road condition score within a preset time period T2.

[0086] The beneficial effect of the present invention is that the present invention predicts and scores the road conditions in the future time period through a road condition prediction model, and generates a scheduling plan in combination with an improved tuna school optimization algorithm to complete the unified scheduling of multiple logistics vehicles, thereby improving distribution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a flow chart of a method for integrating mobile terminal intelligent logistics operations according to the present invention;

[0088] Figure 2 is a flow chart of preprocessing road condition information of the present invention;

[0089] Figure 3 is a flow chart of generating a scheduling plan through an improved tuna school optimization algorithm of the present invention;

[0090] Figure 4 It is a schematic diagram of a system for integrating mobile intelligent logistics operations according to the present invention.

[0091] In the figure: destination statistics module 401, cargo information collection module 402, vehicle information collection module 403, feature sequence generation module 404, road condition score prediction module 405, and scheduling plan generation module 406. DETAILED DESCRIPTION

[0092] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0093] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0094] like Figure 1 to Figure 4 As shown, a method for integrating mobile intelligent logistics operations includes the following steps:

[0095] Step S101, obtaining M different destinations through statistical analysis according to the destinations of the goods to be delivered;

[0096] Step S102, obtaining cargo information of cargo to be delivered to M different destinations;

[0097] Cargo information includes: unique number, weight, volume and delivery time;

[0098] Step S103, obtaining vehicle information of N vehicles to be dispatched in the distribution center;

[0099] Vehicle information includes: unique number, maximum load weight, maximum load volume, total mileage, average fuel consumption, maximum speed and historical maintenance times;

[0100] Step S104, within a first preset time period T1, traffic information of M different destinations is obtained at preset time intervals t, and preprocessed to generate a feature sequence;

[0101] The feature sequence of the mth destination includes R sequence units, and the rth sequence unit represents the pre-processed road condition information of the mth destination at the rth time point, where 1≤m≤M, 1≤r≤R, and R=T1 / t;

[0102] Traffic information includes: distance to the distribution center, road conditions, traffic events, number of traffic lights, temperature, rainfall, wind speed and visibility;

[0103] Step S105, inputting the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2;

[0104] The road condition score ranges from 0 to 10, and the larger the road condition score, the better the road condition;

[0105] Step S106, generating a dispatching plan by using an improved tuna swarm optimization algorithm according to the vehicle information of the N vehicles to be dispatched, the cargo information of the cargo to be delivered at M different destinations, and the road condition scores within the preset time period T2;

[0106] According to the scheduling plan, the goods to be delivered at M different destinations are allocated to N vehicles to be scheduled.

[0107] In one embodiment of the present invention, the number M of different destinations, the number N of vehicles to be dispatched, the first preset time period T1, the preset time interval t and the second preset time period T2 are all custom parameters, and the unique number of the goods to be delivered and the unique number of the vehicle to be dispatched are both represented by self-increasing positive integer codes, and can also be generated by a snowflake algorithm.

[0108] It should be noted that the number M of different destinations is obtained by statistical analysis of the destinations of the goods to be delivered, and the number N of vehicles to be dispatched is set according to the actual number of idle vehicles in the distribution center. Preferably, the first preset time period T1 is set to 1 hour, and the preset time interval t is set to 5 minutes, then R=T1 / t=12, and the second preset time period is set to 1 hour.

[0109] In one embodiment of the present invention, the cargo information of the cargo to be delivered is stored in a database, and statistical analysis is performed through a query tool provided by the database to obtain different destinations.

[0110] For example, the cargo information of the goods to be delivered is stored in the MySQL database, and the query statement for statistical analysis is as follows:

[0111] SELECT destination, COUNT(*) AS cargo_count, SUM(weight) AS total_weight, SUM(volume) AS total_volume FROM cargo GROUP BY destination;

[0112] Where destination represents the destination, cargo_count represents the number of different destinations, total_weight represents the total weight of cargo at different destinations, total_volume represents the total volume of cargo at different destinations, and cargo represents a data table storing cargo information of cargo to be delivered. In addition, cargo information of cargo to be delivered can also be stored in other types of databases, such as MongoDB or ES (Elasticsearch).

[0113] In one embodiment of the present invention, the road section between the distribution center and different destinations is divided into multiple sub-sections according to intersections, and the road conditions, traffic events and the number of traffic lights of the sub-sections are obtained through the Baidu Map API, and then the traffic events and the number of traffic lights of the sub-sections are respectively accumulated as the traffic events and the number of traffic lights in the road condition information; the road conditions of the sub-sections are averaged as the road conditions in the road condition information; the temperature, rainfall, wind speed and visibility of the sub-sections are obtained through existing weather websites, and similarly, the temperature, rainfall, wind speed and visibility of the sub-sections are respectively averaged as the temperature, rainfall, wind speed and visibility in the road condition information. The existing weather website can be Weather.net or Hefeng Weather, etc., which will not be elaborated here.

[0114] In one embodiment of the present invention, Figure 2 As shown, preprocessing the road condition information includes the following steps:

[0115] Step S201, converting road conditions and traffic events into numerical representations;

[0116] Road conditions include good, average, and poor, represented by positive integers 3, 2, and 1 respectively;

[0117] Traffic incidents are represented by the number of traffic incidents that occurred;

[0118] Step S202: For missing values ​​in the traffic condition information, interpolation filling is performed by taking the average value of the traffic condition information of two adjacent time points of the missing value;

[0119] Step S203: normalize the traffic information using a Z-Score method.

[0120] It should be noted that the road condition information is represented numerically to facilitate the processing of the road condition prediction model. The interpolation and filling processing is used to ensure the integrity of the data and reduce noise. The normalization processing is used to eliminate the dimensionality effect, so that the road condition prediction model performs more stably on data in different ranges and improves its generalization ability and robustness.

[0121] In one embodiment of the present invention, the road condition prediction model includes: a first hidden layer, a concatenated layer, R second hidden layers and a first classifier;

[0122] The first hidden layer inputs the feature sequence and outputs the global vector;

[0123] The concatenation layer is used to concatenate the global vector output by the first hidden layer with each sequence unit of the feature sequence to obtain a combined vector;

[0124] The r-th second hidden layer inputs the combination vector corresponding to the r-th sequence unit of the feature sequence and outputs the update vector;

[0125] The second hidden layer is built based on GRU (Gated Recurrent Unit);

[0126] The update vector output by the Nth second hidden layer is input to the first classifier, and the classification space of the first classifier represents the road condition score within the second preset time period T2.

[0127] In one embodiment of the present invention, the training of the road condition prediction model includes: pre-training and formal training;

[0128] During the pre-training process, the update vector output by the Nth second hidden layer is input into the second classifier and the third classifier, and the classification spaces of the second classifier and the third classifier respectively represent the road condition and the temperature in the second preset time period T2;

[0129] During the formal training process, the sample labels of the training samples used to train the road condition prediction model are obtained through manual annotation.

[0130] It should be noted that the training data used for pre-training is easier to collect, and pre-training can help the model learn the characteristics of road condition information. The training samples required for formal training will be greatly reduced, thereby saving a lot of manual labeling costs. In addition, fine-tuning the parameters of the pre-trained road condition prediction model can speed up the training of the road condition prediction model during formal training, thereby reducing the risk of overfitting and improving its generalization ability and robustness.

[0131] In one embodiment of the present invention, the calculation formula of the first hidden layer includes:

[0132] ;

[0133] ;

[0134] ;

[0135] in represents the global vector output by the first hidden layer, seq represents the feature sequence of the first hidden layer input, W represents the weight parameter, b represents the bias parameter, d represents the number of dimensions of a sequence unit of the feature sequence of the first hidden layer input, and the value is 8. and denote the first intermediate matrix and the second intermediate matrix respectively, and represent the first weight parameter and the second weight parameter corresponding to the first intermediate matrix respectively, and They respectively represent the first weight parameter and the second weight parameter corresponding to the second intermediate matrix, T represents a transpose operation, concat represents a concatenation operation, cos represents a cosine function, sin represents a sine function, and Swish represents a Swish activation function.

[0136] It should be noted that the weight parameters and bias parameters in the road condition prediction model are all learnable parameters. The dimension number of a sequence unit of the feature sequence input to the first hidden layer is 8, which correspond to the distance to the distribution center, road conditions, traffic events, number of traffic lights, temperature, rainfall, wind speed and visibility, respectively. Then seq is represented by a matrix of size R×8. The first weight parameters and the second weight parameters corresponding to the first intermediate matrix and the second intermediate matrix can both be designed as matrices of size 8×16. Then the first intermediate matrix and the second intermediate matrix are both matrices of size R×32. The first intermediate matrix multiplied by the transpose of the second intermediate matrix obtains a matrix of size R×R. W can be designed as a vector of size R×1, so the size of the global vector output by the first hidden layer is R×1.

[0137] It should be noted that during the training process of the road condition prediction model, the mean square error between the value output by the road condition prediction model at each iteration and the sample label of the training sample is specified as the loss function, and the weight parameters and bias parameters in the road condition prediction model are reversely updated through the chain rule and gradient descent algorithm. The training of the neural network model is a conventional technical means and will not be elaborated here.

[0138] In one embodiment of the present invention, the second hidden layer can also be constructed based on RNN (recurrent neural network model) or LSTM (long short-term memory recurrent neural network model), etc., which will not be described in detail here.

[0139] In one embodiment of the present invention, Figure 3As shown, the scheduling scheme is generated by the improved tuna swarm optimization algorithm, including the following steps:

[0140] Step S301, randomly generate an initialization population consisting of K individuals that meet the constraints, and initialize the current number of iterations to 1;

[0141] Where K is a custom parameter, preferably, K is set to 20;

[0142] The constraints include:

[0143] Constraint 1: The total weight of the goods to be delivered in a single transport cannot exceed the preset load threshold of the vehicle to be dispatched, where the preset load threshold is a custom parameter. Preferably, the preset load threshold is set to 80% of the maximum load of the vehicle to be dispatched;

[0144] Constraint 2: The total volume of the goods to be delivered in a single transport cannot exceed the preset load volume threshold of the vehicle to be dispatched, where the preset load volume threshold is a custom parameter. Preferably, the preset load volume threshold is set to 80% of the maximum load volume of the vehicle to be dispatched;

[0145] Constraint 3: The number of vehicles to be dispatched for a single transport of goods to be delivered to the same destination cannot exceed the preset vehicle threshold, where the preset vehicle threshold is a custom parameter. Preferably, the preset vehicle threshold is set to 3;

[0146] Constraint 4: The number of different destinations transported by the same dispatched vehicle at the same time cannot exceed the preset destination number threshold, where the destination number threshold is a custom parameter. Preferably, the destination number threshold is set to 6;

[0147] Constraint 5: Load the goods to be delivered in ascending order of their arrival time;

[0148] The individual is represented by a matrix code of M rows and N columns. The element value of the matrix code is represented by 0 or 1. The element value of the mth row and nth column is 0, which means that the goods to be delivered at the mth destination are not allocated to the nth vehicle to be dispatched. Otherwise, it means that the goods to be delivered at the mth destination are allocated to the nth vehicle to be dispatched.

[0149] Step S302, calculating the fitness values ​​of K individuals in the initialization population through the objective function;

[0150] Step S303, calculate the balance coefficient according to the current number of iterations and the maximum number of iterations, and judge that the balance coefficient is greater than or equal to the balance coefficient threshold, then update the matrix codes of K individuals in the initialized population through the parabolic foraging strategy, otherwise proceed to step S304;

[0151] The calculation formula of the balance coefficient Balance is as follows:

[0152] ;

[0153] Where t means the current iteration number is t, Indicates the maximum number of iterations;

[0154] The maximum number of iterations and the balance coefficient threshold are both custom parameters. Preferably, the maximum number of iterations is set to 50, and the balance coefficient threshold is set to 0.5;

[0155] Step S304, updating the matrix codes of random K / 2 individuals in the initialized population by using the first spiral foraging strategy, and updating the matrix codes of the remaining K / 2 individuals in the initialized population by using the second spiral foraging strategy;

[0156] Step S305, respectively calculating the average value of the element value of the matrix code of each individual in the initialized population, and respectively determining that if the element value of the matrix code of each individual is greater than or equal to the corresponding average value, then modify the element value to 1, otherwise modify it to 0;

[0157] Step S306, if it is determined that the iteration termination condition is met, the matrix code of the individual with the largest fitness value in the initialized population is output as the scheduling solution, otherwise the current iteration number is accumulated by 1, and the process returns to step S302 to continue execution;

[0158] The iteration termination conditions include: the current iteration number is greater than or equal to the maximum iteration number; within three consecutive iterations, the change in the maximum fitness value in the initialized population is less than the change threshold, where the change threshold is a custom parameter, preferably, the change threshold is set to 0.1.

[0159] It should be noted that the constraints are set to ensure the rationality and safety of the loading of the goods to be delivered, and the goods to be delivered to the same destination can be transported by multiple vehicles to be dispatched, mainly to reduce empty vehicles. In addition, the present invention is aimed at vehicle scheduling for a single transportation. After the single transportation is completed, the scheduling plan can be re-generated by the improved tuna school optimization algorithm.

[0160] In one embodiment of the present invention, the calculation formula for calculating the fitness value Fit of an individual through the objective function is as follows:

[0161] ;

[0162] in The element value of the mth row and nth column of the matrix encoding of the individual is represented by: , They represent the distance between the mth destination and the nth vehicle to be dispatched and the road condition score within the preset time period T2, respectively. and They represent the maximum speed and average fuel consumption of the nth vehicle to be dispatched when the goods to be delivered at the mth destination are assigned to the nth vehicle to be dispatched. and Respectively represent the custom weight coefficients, and and The sum of the values ​​is 1, preferably, Set to 0.6, Set to 0.4.

[0163] In one embodiment of the present invention, the calculation formula of the improved tuna school optimization algorithm includes:

[0164] The calculation formula of the parabolic foraging strategy includes:

[0165] ;

[0166] ;

[0167] Where 1≤t≤ , 1≤i≤K, and They represent the matrix codes of the i-th individual whose current iteration number is t and whose current iteration number is t+1, respectively. Represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, represents the first random number in the range of 0 to 1, TF represents a random number between 1 and -1, and p represents a nonlinear decreasing coefficient;

[0168] The calculation formula for the first spiral foraging strategy includes:

[0169] ;

[0170] ;

[0171] ;

[0172] ;

[0173] ;

[0174] in , and They represent the first decision coefficient, the second decision coefficient and the third decision coefficient respectively, f represents the foraging factor, e represents the natural constant, and Respectively represent the second random number and the third random number whose values ​​range from 0 to 1;

[0175] The calculation formula of the second spiral foraging strategy is as follows:

[0176] ;

[0177] in Represents the matrix encoding of a random individual with the current iteration number t.

[0178] In one embodiment of the present invention, Figure 4 As shown, a system for integrating mobile intelligent logistics operations includes:

[0179] A destination statistics module 401 is used to obtain M different destinations through statistical analysis according to the destinations of the goods to be delivered;

[0180] A cargo information collection module 402 is used to obtain cargo information of cargo to be delivered to M different destinations;

[0181] The vehicle information collection module 403 is used to obtain the vehicle information of N vehicles to be dispatched in the distribution center;

[0182] The feature sequence generating module 404 is used to obtain the road condition information of M different destinations at a preset time interval t within a first preset time period T1, and perform preprocessing to generate a feature sequence;

[0183] A traffic score prediction module 405 is used to input the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2;

[0184] The dispatching scheme generating module 406 is used to generate a dispatching scheme by using an improved tuna school optimization algorithm according to the vehicle information of N vehicles to be dispatched, the cargo information of M cargoes to be delivered at different destinations, and the road condition scores within a preset time period T2.

[0185] It should be noted that, in addition to being applied to the logistics scheduling of unmanned vehicles, the present invention can also be applied to the logistics scheduling of manually driven vehicles, and can be integrated on the mobile terminal to display the delivery progress and delivery route of each courier, etc., which will not be elaborated here.

[0186] The specific implementation methods of this embodiment are described above, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make many forms inspired by this embodiment, all of which are within the protection of this embodiment.

Claims

1. A method for integrating mobile terminal intelligent logistics operations, characterized in that: The following steps are involved: Step S101, obtaining M different destinations through statistical analysis according to the destinations of the goods to be delivered; Step S102, obtaining cargo information of cargo to be delivered to M different destinations; Cargo information includes: unique number, weight, volume and delivery time; Step S103, obtaining vehicle information of N vehicles to be dispatched in the distribution center; Vehicle information includes: unique number, maximum load weight, maximum load volume, total mileage, average fuel consumption, maximum speed and historical maintenance times; Step S104, within a first preset time period T1, traffic information of M different destinations is obtained at preset time intervals t, and preprocessed to generate a feature sequence; The feature sequence of the mth destination includes R sequence units, and the rth sequence unit represents the road condition information of the mth destination after preprocessing at the rth time point, where 1≤m≤M, 1≤r≤R, and R=T1 / t; Traffic information includes: distance to the distribution center, road conditions, traffic events, number of traffic lights, temperature, rainfall, wind speed and visibility; Step S105, inputting the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2; The road condition score ranges from 0 to 10, and the larger the road condition score, the better the road condition; Step S106, generating a dispatching plan by using an improved tuna swarm optimization algorithm according to the vehicle information of the N vehicles to be dispatched, the cargo information of the cargo to be delivered at M different destinations, and the road condition scores within a preset time period T2; According to the dispatch plan, the goods to be delivered at M different destinations are assigned to N vehicles to be dispatched; The scheduling scheme is generated by the improved tuna school optimization algorithm, which includes the following steps: Step S301, randomly generate an initialization population consisting of K individuals that meet the constraints, and initialize the current number of iterations to 1; Where K is a custom parameter; Step S302, calculating the fitness values ​​of K individuals in the initialization population through the objective function; Step S303, calculate the balance coefficient according to the current number of iterations and the maximum number of iterations, and judge that the balance coefficient is greater than or equal to the balance coefficient threshold, then update the matrix codes of K individuals in the initialized population through the parabolic foraging strategy, otherwise proceed to step S304; The calculation formula of the balance coefficient Balance is as follows: Where t represents the current iteration number t, t max Indicates the maximum number of iterations; The maximum number of iterations and the balance coefficient threshold are both custom parameters; Step S304, updating the matrix codes of random K / 2 individuals in the initialized population by using the first spiral foraging strategy, and updating the matrix codes of the remaining K / 2 individuals in the initialized population by using the second spiral foraging strategy; Step S305, respectively calculating the average value of the element value of the matrix code of each individual in the initialized population, and respectively determining that if the element value of the matrix code of each individual is greater than or equal to the corresponding average value, then modify the element value to 1, otherwise modify it to 0; Step S306, if it is determined that the iteration termination condition is met, the matrix code of the individual with the largest fitness value in the initialized population is output as the scheduling solution, otherwise the current iteration number is accumulated by 1, and the process returns to step S302 to continue execution; The iteration termination conditions include: the current iteration number is greater than or equal to the maximum iteration number; within three consecutive iterations, the change in the maximum fitness value in the initialized population is less than the change threshold, where the change threshold is a custom parameter.

2. The method for integrating mobile intelligent logistics operations according to claim 1 is characterized in that: The number M of different destinations, the number N of vehicles to be dispatched, the first preset time period T1, the preset time interval t and the second preset time period T2 are all custom parameters. The unique number of the goods to be delivered and the unique number of the vehicle to be dispatched are both represented by self-increasing positive integer codes.

3. The method for integrating mobile intelligent logistics operations according to claim 1 is characterized in that: The cargo information of the goods to be delivered is stored in the database, and statistical analysis is performed through the query tools provided by the database to obtain different destinations.

4. The method for integrating mobile intelligent logistics operations according to claim 1 is characterized in that: Preprocessing the traffic information includes the following steps: Step S201, converting road conditions and traffic events into numerical representations; Road conditions include good, average, and poor, represented by positive integers 3, 2, and 1 respectively; Traffic incidents are represented by the number of traffic incidents that occurred; Step S202: For missing values ​​in the traffic condition information, interpolation filling is performed by taking the average value of the traffic condition information of two adjacent time points of the missing value; Step S203: normalize the traffic information using a Z-Score method.

5. The method for integrating mobile intelligent logistics operations according to claim 1 is characterized in that: The road condition prediction model includes: a first hidden layer, a concatenation layer, R second hidden layers and a first classifier; The first hidden layer inputs the feature sequence and outputs the global vector; The concatenation layer is used to concatenate the global vector output by the first hidden layer with each sequence unit of the feature sequence to obtain a combined vector; The r-th second hidden layer inputs the combination vector corresponding to the r-th sequence unit of the feature sequence and outputs the update vector; The second hidden layer is built based on gated recurrent network units; The update vector output by the Nth second hidden layer is input to the first classifier, and the classification space of the first classifier represents the road condition score within the second preset time period T2; The training of the traffic condition prediction model includes: pre-training and formal training; During the pre-training process, the update vector output by the Nth second hidden layer is input into the second classifier and the third classifier, and the classification spaces of the second classifier and the third classifier respectively represent the road condition and the temperature in the second preset time period T2; During the formal training process, the sample labels of the training samples used to train the road condition prediction model are obtained through manual annotation.

6. A method for integrating mobile intelligent logistics operations according to claim 5, characterized in that: The calculation formula for the first hidden layer includes: F Global =Swish(F Q ×F K T ×W+b); where F Global represents the global vector output by the first hidden layer, seq represents the feature sequence of the first hidden layer input, W represents the weight parameter, b represents the bias parameter, d represents the number of dimensions of a sequence unit of the feature sequence of the first hidden layer input, and the value is 8, F Q and F K denote the first intermediate matrix and the second intermediate matrix respectively, and represent the first weight parameter and the second weight parameter corresponding to the first intermediate matrix respectively, and They respectively represent the first weight parameter and the second weight parameter corresponding to the second intermediate matrix, T represents a transpose operation, concat represents a concatenation operation, cos represents a cosine function, sin represents a sine function, and Swish represents a Swish activation function.

7. The method for integrating mobile intelligent logistics operations according to claim 1 is characterized in that: The constraints include: Constraint 1: The total weight of the goods to be delivered in a single transport cannot exceed the preset load threshold of the vehicle to be dispatched, where the preset load threshold is a custom parameter; Constraint 2: The total volume of the goods to be delivered in a single transport cannot exceed the preset load volume threshold of the vehicle to be dispatched, where the preset load volume threshold is a custom parameter; Constraint 3: The number of vehicles to be dispatched for a single transport of goods to the same destination cannot exceed the preset vehicle threshold, where the preset vehicle threshold is a custom parameter; Constraint 4: The number of different destinations that the same dispatched vehicle can transport simultaneously cannot exceed the preset destination number threshold, where the destination number threshold is a custom parameter; Constraint 5: Load the goods to be delivered in ascending order of their arrival time; The individual is represented by a matrix code of M rows and N columns. The element value of the matrix code is represented by 0 or 1. The element value of the mth row and nth column is 0, which means that the goods to be delivered at the mth destination are not allocated to the nth vehicle to be dispatched. Otherwise, it means that the goods to be delivered at the mth destination are allocated to the nth vehicle to be dispatched.

8. The method for integrating mobile intelligent logistics operations according to claim 7 is characterized in that: The calculation formula for calculating the individual's fitness value Fit through the objective function is as follows: in The element value of the mth row and nth column of the matrix encoding of the individual is represented by: They represent the distance between the mth destination and the nth vehicle to be dispatched and the road condition score within the preset time period T2, respectively. and They respectively represent the maximum speed and average fuel consumption of the goods to be delivered at the mth destination allocated to the nth vehicle to be dispatched. U1 and U2 respectively represent the custom weight coefficients, and the sum of U1 and U2 is 1.

9. The method for integrating mobile terminal intelligent logistics operations according to claim 7 is characterized in that: The calculation formula of the improved tuna school optimization algorithm includes: The calculation formula of the parabolic foraging strategy includes: Where 1≤t≤t max , 1≤i≤K, and They represent the matrix codes of the i-th individual whose current iteration number is t and whose current iteration number is t+1, respectively. It represents the matrix encoding of the individual with the largest fitness value at the current iteration number t, rand1 represents the first random number ranging from 0 to 1, TF represents a random number between 1 and -1, and p represents the nonlinear decreasing coefficient; The calculation formula for the first spiral foraging strategy includes: Where c1, c2 and c3 represent the first decision coefficient, the second decision coefficient and the third decision coefficient respectively, f represents the foraging factor, e represents the natural constant, rand2 and rand3 represent the second random number and the third random number in the range of 0 to 1 respectively; The calculation formula of the second spiral foraging strategy is as follows: in Represents the matrix encoding of a random individual with the current iteration number t.

10. A system for integrating mobile intelligent logistics operations, characterized in that: Executing a method for integrating mobile intelligent logistics operations as claimed in any one of claims 1 to 9, comprising: A destination statistics module is used to obtain M different destinations through statistical analysis according to the destinations of the goods to be delivered; A cargo information collection module, which is used to obtain cargo information of cargo to be delivered to M different destinations; The vehicle information collection module is used to obtain the vehicle information of N vehicles to be dispatched in the distribution center; A feature sequence generation module, which is used to obtain the road condition information of M different destinations at a preset time interval t within a first preset time period T1, and perform preprocessing to generate a feature sequence; A traffic score prediction module, which is used to input the feature sequences of M different destinations into the traffic prediction model respectively, and the output value represents the traffic score in the second preset time period T2; The scheduling plan generation module is used to generate a scheduling plan through an improved tuna school optimization algorithm according to the vehicle information of N vehicles to be scheduled, the cargo information of M cargoes to be delivered at different destinations, and the road condition score within a preset time period T2.

Images