CN114444809A - Data-driven multi-target strip mine card path optimization method - Google Patents

Data-driven multi-target strip mine card path optimization method Download PDF

Info

Publication number
CN114444809A
CN114444809A CN202210124843.7A CN202210124843A CN114444809A CN 114444809 A CN114444809 A CN 114444809A CN 202210124843 A CN202210124843 A CN 202210124843A CN 114444809 A CN114444809 A CN 114444809A
Authority
CN
China
Prior art keywords
mine
model
mine card
ant
transportation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210124843.7A
Other languages
Chinese (zh)
Inventor
顾清华
王倩
江松
阮顺领
冯治东
李学现
陈露
常朝朝
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Architecture and Technology
Original Assignee
Xian University of Architecture and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian University of Architecture and Technology filed Critical Xian University of Architecture and Technology
Priority to CN202210124843.7A priority Critical patent/CN114444809A/en
Publication of CN114444809A publication Critical patent/CN114444809A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/02—Agriculture; Fishing; Forestry; Mining
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40—Business processes related to the transportation industry

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Health & Medical Sciences (AREA)
  • Tourism & Hospitality (AREA)
  • Primary Health Care (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Operations Research (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Game Theory and Decision Science (AREA)
  • Molecular Biology (AREA)
  • Quality & Reliability (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Agronomy & Crop Science (AREA)
  • Animal Husbandry (AREA)
  • Marine Sciences & Fisheries (AREA)
  • Mining & Mineral Resources (AREA)
  • Development Economics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A multi-target strip mine card path optimization method under data driving is characterized in that a multi-target strip mine card path optimization model which aims at the shortest total transportation distance and the smallest time loss penalty cost is constructed according to actual production requirements and indexes of a mine; based on historical track data of the mine cards, a data-driven random forest auxiliary optimization algorithm is adopted to solve the multi-target strip mine card path optimization model, the algorithm is improved under the optimization framework of the ant colony algorithm, and the transportation scheme more conforming to the current production situation of the mine is searched efficiently. The invention starts from the actual situation of mine production, and has important significance for improving the transportation efficiency, reducing the transportation cost and improving the economic benefits of mine enterprises.

Description

Data-driven multi-target strip mine card path optimization method
Technical Field
The invention belongs to the technical field of mining system engineering and mine optimization, and particularly relates to a data-driven multi-target strip mine card path optimization method.
Background
In recent years, with the increasingly advanced application of new technologies such as artificial intelligence, big data, internet of things, cloud computing and the like to the field of mining industry, the global mining industry is undergoing a new technical revolution. The open-pit mining is an important mode for mining mineral resources in China, and the transportation cost of the open-pit mining accounts for about 50% of the total production and operation cost. Meanwhile, mine truck transportation is the main mode of strip mine production transportation in China at present. Therefore, global planning of the transportation path, reasonable distribution of mine cards and transportation routes, and realization of cost reduction and efficiency improvement in production and operation are problems which are closely concerned and urgently need to be solved by each mine enterprise.
However, in the related art, real factors such as mine road conditions and weather which have great influence on the transportation efficiency are rarely considered, so that the obtained scheme is difficult to meet the actual production requirement. Meanwhile, as the demand of enterprises on the transportation scheme is more diversified, the ideal transportation scheme often needs to consider a plurality of indexes simultaneously, so as to achieve the optimal effect. Therefore, in order to solve this problem, it is necessary to study a multi-objective surface mine card path optimization method based on trajectory data information.
Disclosure of Invention
In order to overcome the defects of the prior art and solve the problem of the transportation path of the current metal strip mine card team, the invention aims to provide a multi-target strip mine card path optimization method under data driving, establish an strip mine card path optimization model with the shortest total transportation distance and the smallest time loss penalty cost as the target, and design a random forest auxiliary optimization algorithm under data driving to solve the multi-target strip mine card path optimization model.
In order to achieve the purpose, the invention adopts the technical scheme that:
a multi-target strip mine card path optimization method under data driving comprises the following steps:
step 1, according to mine production requirements and indexes, combining a balance relation between transportation cost and transportation efficiency, and constructing a multi-target strip mine card path optimization model by taking the shortest total transportation distance and the smallest time loss penalty cost as targets;
step 2, training by using an actual production data set of the strip mine at least comprising mine card numbers, position elevations, mine card speeds, driving time and distances, establishing a random forest agent auxiliary model for an optimization individual, namely a fleet mine card route scheme set, wherein a model predicted value is a target value of the optimization model;
step 3, improving a multi-target ant colony solving algorithm around the multi-target strip mine card path optimization model characteristics;
and 4, solving the multi-target strip mine card path optimization model by adopting a random forest auxiliary optimization algorithm under data driving.
Compared with the prior art, the invention has the beneficial effects that:
the method has the advantages that the random forest auxiliary optimization algorithm under data driving has a positive feedback mechanism, the solving efficiency is high, the feasibility and the superiority are realized when complex problems are processed, and the transportation scheme which is more in line with the current mine production situation can be obtained in the process of solving the multi-target strip mine card path optimization model. The method provided by the invention starts from the actual mine production situation, and has important significance in the aspects of improving the transportation efficiency, reducing the transportation cost and improving the economic benefits of mine enterprises.
Drawings
FIG. 1 is a flow chart of a model solving method using a random forest aided optimization algorithm under data driving in the present invention.
Detailed Description
The embodiments of the present invention will be described in detail below with reference to the drawings and examples.
The invention provides a data-driven multi-target strip mine card path optimization method, which introduces a data-driven random forest agent auxiliary model on the basis of the existing mine card path model, uses historical data containing uncertain factors such as road conditions, weather and the like to guide the search of an improved multi-target ant colony algorithm (an adaptive pheromone updating mechanism, an improved ant circumference model and a Pareto fitness evaluation mechanism), and mainly comprises the following steps as shown in figure 1:
step 1, according to mine production requirements and indexes, combining a balance relation between transportation cost and transportation efficiency, and constructing a multi-target strip mine card path optimization model by taking the shortest total transportation distance and the smallest time loss penalty cost as targets.
(1) The shortest total distance of transportation target:
Figure BDA0003499950730000031
wherein, F1Representing the transport cost of the ore cards, R being the total number of ore cards dispatched, M being the set of ore loading points, N being the set of ore unloading points, c1Representing the cost per unit of transportation in the line, dijRepresents the length of the route between nodes i and j in the road network of the mining area, xijrIs a variable from 0 to 1 when xijrWhen the value is 1, the value represents that the R-th mine card runs from the ith loading point to the jth unloading point, i belongs to {1,2, …, M }, j belongs to {1,2, …, N }, and R belongs to {1,2, …, R };
(2) time loss penalty cost minimum objective:
Figure BDA0003499950730000032
wherein, F2Representing the sum of the time loss cost of each loading point and each unloading point,
Figure BDA0003499950730000033
representing the time when the mine card reaches the ith loading point, i.e. representing the starting time at that loading point,
Figure BDA0003499950730000034
the time window requirement for the ith load point,
Figure BDA0003499950730000035
represents earlier than
Figure BDA0003499950730000036
The unit penalty cost of the arrival of the time,
Figure BDA0003499950730000037
is later than
Figure BDA0003499950730000038
Unit punishment cost of time arrival;
Figure BDA0003499950730000039
representing the time when the mine card reaches the jth unloading point, i.e. representing the starting time at that unloading point,
Figure BDA00034999507300000310
the time window requirement for the jth unload point,
Figure BDA00034999507300000311
represents earlier than
Figure BDA00034999507300000312
The unit penalty cost of the arrival of the time,
Figure BDA00034999507300000313
is later than
Figure BDA00034999507300000314
Unit punishment cost of time arrival;
(3) and (3) carrying load restraint of the mine trucks:
Figure BDA00034999507300000315
wherein, gjRepresenting the weight of ore delivered by the r-th card at the j-th unloading point in a single transit, WrRepresenting the maximum load capacity of the r-th mine card;
(4) transit time limit constraints:
Figure BDA0003499950730000041
wherein, wiIndicating the queue waiting time, st, at the ith load pointiDenotes the loading time at the ith loading point, tijRepresenting the journey time of the mine card from the ith loading point to the jth unloading point;
(5) flow conservation constraint
Figure BDA0003499950730000042
xi*r∈{0,1}
x*jr∈{0,1}
Wherein x isi*r1 represents the r-th mine car completing the loading task at the i-th loading point, x*jr1 represents that the r vehicle is unloaded at the jth unloading point; in order to ensure that the mine cards from each loading point must arrive at each unloading point for transportation service, thereby forming a complete transportation route, the number of mine cards from the loading point must be equal to the total number of mine cards served at each unloading point;
6) decision variable constraints
xijr∈{0,1}
Wherein the value of the decision variable is 0 or 1;
(7) non-negative variable constraints
Figure BDA0003499950730000043
Wherein all variables involved in the optimization model are non-negative values.
And 2, training by using an actual production data set of the strip mine at least comprising mine card numbers, position elevations, mine card speeds, driving time and distances, and establishing a random forest agent auxiliary model for an optimization individual, namely a fleet mine card route scheme set, wherein a model predicted value is a target value of the optimization model. The specific method of model management is as follows:
(1) establishing a random forest agent auxiliary model, comprising the following steps: firstly, based on a historical data training set of a mine card, adopting a bootstrap method to perform replaced random sampling to generate C sub-sample sets, and using the C sub-sample sets to train C classification regression trees. In the process of training the classification regression tree, feature node division is carried out according to mine card trajectory data information such as mine section node distance, time and the like in each sample subset, the obtained part with the minimum mean square sum error is regarded as the best split in all split points, binary tree splitting is carried out in sequence, and after a termination condition is reached, the classification regression tree stops growing. And C classification regression trees are sequentially constructed according to the steps. And finally, calculating the average value of all classification regression trees, and taking the average value as the final output of the random forest model. Since the optimization model in step 1 has multiple targets, the random forest agent auxiliary model needs to be constructed for each path optimization target.
(2) And estimating and correcting the approximate error of the random forest agent auxiliary model by adopting an individual-based model management strategy so as to improve the prediction precision. First, the difference between the true and predicted values of the y-th target for the non-dominant individual is evaluated by the following equation, denoted as ery. Non-dominant individuals here represent the currently optimal mine card transportation scenario.
Figure BDA0003499950730000051
Wherein, the first and the second end of the pipe are connected with each other,
Figure BDA0003499950730000052
the true value for the h-th non-dominant individual at the y-th optimization objective,
Figure BDA0003499950730000053
the predicted value of the H non-dominant individual in the y optimization target, and H is the number of the non-dominant individuals.
Then, the target predicted values of all the non-dominant individuals are corrected based on the error, and the target predicted values are obtainedIf the objective function is minimized, it is corrected by subtracting the corresponding error value from the value of the y-th objective function, i.e. Fy-er。
And finally, taking the corrected non-dominant individual as a new sample and putting the new sample into a training data set, thereby completing the updating of the random forest agent auxiliary model.
Step 3, in order to improve the optimization efficiency of the algorithm and the quality of the transportation scheme, the multi-target ant colony solving algorithm is improved around the characteristics of the multi-target strip mine card path optimization model, and in the optimization process, on the basis of an ant colony algorithm optimization framework, a self-adaptive pheromone updating mechanism and an improved ant circumference model are introduced, so that the solution efficiency and the scheme quality are improved; and a Pareto fitness evaluation mechanism was introduced to select more potential nondominant individuals. The method comprises the following specific steps:
(1) adaptive pheromone update mechanism
Combining different search biases presented by the evolutionary algorithm in different search stages, a pheromone volatilization factor capable of being adaptively adjusted is designed, and the pheromone volatilization quantity can be adaptively adjusted according to the requirement of the search progress, as shown in the following formula:
Figure BDA0003499950730000061
where ρ is the pheromone volatility factor and FE represents the current function evaluation times. Here, each ant represents a group of mine car transportation scenarios. Based on the updating mechanism, the ants have larger pheromone volatilization amount in the early stage of searching, can search in as wide an area as possible, and is favorable for developing global exploration on the optimal transportation path. With the iteration, the ant colony gradually converges to the vicinity of the Pareto optimal transportation scheme in the middle and later stages of the search, and the pheromone volatilization amount is adaptively adjusted to a small value at the moment so as to accelerate the convergence speed of the ant colony algorithm and save the computing resources.
(2) Improved ant surrounding model
According to the established path optimization model, the ant circumference model is improved, and the increment of each ant pheromone is evaluated from the aspects of transportation distance and waiting time penalty, so that the ant is guided to explore the optimal path more accurately.
Figure BDA0003499950730000062
Wherein Q is1And Q2Respectively the total quantity of pheromones on two targets of minimum time loss penalty cost and minimum total distance cost of transportation in the path optimization model, LeIs the total path length, P, traveled by the e-th anteThe e-th ant receives a time window violation penalty on its travel path.
(3) Pareto fitness evaluation mechanism
Aiming at the characteristics of the multi-objective optimization problem, the evaluation mechanism in the SPEA2 is introduced into the ant colony algorithm, the idea of rapid non-dominant ordering is used, and meanwhile, the relation between a non-dominant individual and a dominated individual is considered, so that the quality of each individual is comprehensively evaluated. In addition, the fitness mechanism also introduces a k-nearest neighbor method, and the distribution condition of individuals is taken into account, so that the diversity and convergence of ant populations can be effectively balanced. Overall, the fitness evaluation mechanism considering the individual domination relationship and the position distribution information can evaluate the quality of the transportation scheme more comprehensively and scientifically. The specific formula is as follows:
F(e)＝R(e)+D(e)
Figure BDA0003499950730000071
S(e)＝|{u|u∈P+Q∧e＞u}|
Figure BDA0003499950730000072
Figure BDA0003499950730000073
wherein, F (e), R (e) and D (e) respectively represent the adaptability value, the dominance grade value and the position distribution information of the e-th ant. S (e) represents the number of ants dominated by ant e in population P and external archive set Q, and u is the ant in the set of P and Q.
Figure BDA0003499950730000074
To calculate the euclidean distance from the e-th ant to the k-th adjacent ant, the distances of ant e from both population P and other ants in external archive set Q are calculated and sorted in ascending order.
And 4, solving the multi-target strip mine card path optimization model by adopting a data-driven random forest auxiliary ant colony optimization algorithm, wherein the process is as follows:
step 1, initializing algorithm parameters and a tabu table, wherein the parameters comprise a population size NpThe number C of classification regression trees, the splitting stop condition T of the classification regression trees, the pheromone importance degree factor alpha, the heuristic information importance degree factor beta, the total pheromone Q, the pheromone volatilization coefficient rho and the maximum evaluation times FEmaxEtc.;
step 2: the method comprises the following steps that initial ants randomly search a transportation route from any loading point, and select a next route node to be visited according to transition probability;
step 3: recording route nodes visited by ants in a taboo table;
step 4: and judging whether the ants reach any unloading point in the road network of the mining area. If the condition is met, executing Step 5, otherwise, returning to Step 2;
step 5: predicting and evaluating all route schemes by using a random forest agent auxiliary model;
step 6: evaluating the quality of the scheme according to a fitness evaluation mechanism, and dividing Pareto grades;
step 7: globally updating pheromones on the paths based on a self-adaptive pheromone updating mechanism and an improved ant periphery model;
step 8: storing the current optimal path scheme, and recording the total transportation distance and time cost of the current optimal path scheme;
step 9: using the current non-dominated individual to carry out error correction and update on the agent model;
step 10: judging the iteration state, and if the iteration state reaches a termination condition, outputting the current optimal Pareto path scheme; otherwise, the taboo list is emptied and the operation returns to Step 2.
Aiming at the problem of the transportation of the strip mine card, according to the actual production requirements and indexes of the mine, a multi-target strip mine card path optimization model which aims at the shortest total transportation distance and the smallest time loss penalty cost is constructed; and the model is solved by adopting a data-driven random forest auxiliary optimization algorithm, so that a transportation scheme which is more in line with the current situation of mine production can be obtained, the transportation efficiency is effectively improved from the actual situation of mine production, the transportation cost is reduced, and the method has important significance for obviously improving the economic benefit of mine enterprises.

Claims (8)

1. A data-driven multi-target strip mine card path optimization method is characterized by comprising the following steps:
step 1, according to mine production requirements and indexes, combining a balance relation between transportation cost and transportation efficiency, and constructing a multi-target strip mine card path optimization model by taking the shortest total transportation distance and the smallest time loss penalty cost as targets;
step 2, training by using an actual production data set of the strip mine at least comprising mine card numbers, position elevations, mine card speeds, driving time and distances, establishing a random forest agent auxiliary model for an optimization individual, namely a fleet mine card route scheme set, wherein a model predicted value is a target value of the optimization model;
step 3, improving a multi-target ant colony solving algorithm around the multi-target strip mine card path optimization model characteristics;
and 4, solving the multi-target strip mine card path optimization model by adopting a random forest auxiliary optimization algorithm under data driving.
2. The data-driven multi-objective strip mine card path optimization method according to claim 1, wherein the multi-objective strip mine card path optimization model in the step 1 is expressed by the following formula:
(1) the shortest total distance of transportation target:
Figure FDA0003499950720000011
wherein, F1Representing the transport cost of the ore cards, R being the total number of ore cards dispatched, M being the set of ore loading points, N being the set of ore unloading points, c1Representing the cost per unit of transportation in the line, dijRepresents the length of the route between nodes i and j in the road network of the mining area, xijrIs a variable from 0 to 1 when xijrWhen the value is 1, the value represents that the R-th mine card runs from the ith loading point to the jth unloading point, i belongs to {1,2, …, M }, j belongs to {1,2, …, N }, and R belongs to {1,2, …, R };
(2) time loss penalty cost minimum objective:
Figure FDA0003499950720000021
wherein, F2Representing the sum of the time loss cost of each loading point and each unloading point,
Figure FDA0003499950720000022
representing the time when the mine card reaches the ith loading point, i.e. representing the starting time at that loading point,
Figure FDA0003499950720000023
the time window requirement for the ith load point,
Figure FDA0003499950720000024
represents earlier than
Figure FDA0003499950720000025
Unit punishment cost of time arrival，
Figure FDA0003499950720000026
Is later than
Figure FDA0003499950720000027
Unit punishment cost of time arrival;
Figure FDA0003499950720000028
representing the time when the mine card reaches the jth unloading point, i.e. representing the starting time at that unloading point,
Figure FDA0003499950720000029
Figure FDA00034999507200000210
the time window requirement for the jth unload point,
Figure FDA00034999507200000211
represents earlier than
Figure FDA00034999507200000212
The unit penalty cost of the arrival of the time,
Figure FDA00034999507200000213
is later than
Figure FDA00034999507200000214
Unit punishment cost of time arrival;
(3) and (3) carrying load restraint of the mine trucks:
Figure FDA00034999507200000215
wherein, gjRepresenting the weight of ore delivered by the r-th card at the j-th unloading point in a single transit, WrRepresenting the maximum load capacity of the r-th mine card;
(4) transit time limit constraints:
Figure FDA00034999507200000216
wherein, wiIndicating the queue waiting time, st, at the ith load pointiDenotes the loading time at the ith loading point, tijRepresenting the journey time of the mine card from the ith loading point to the jth unloading point;
(5) flow conservation constraint
Figure FDA00034999507200000217
xi*r∈{0,1}
x*jr∈{0,1}
Wherein x isi*r1 represents the r-th mine car completing the loading task at the i-th loading point, x*jr1 represents that the r vehicle is unloaded at the jth unloading point;
(6) decision variable constraints
xijr∈{0,1}
Wherein the value of the decision variable is 0 or 1;
(7) non-negative variable constraints
Figure FDA0003499950720000031
Wherein all variables involved in the optimization model are non-negative values.
3. The data-driven multi-objective strip mine card path optimization method according to claim 1, wherein in the step 2, a random forest agent auxiliary model is respectively established for each objective of the optimization model; the random forest agent auxiliary model is established by the following steps: based on a mine card historical data training set, adopting a bootstrap method to perform replaced random sampling to generate C sub-sample sets, and using the C sub-sample sets to train C classification regression trees; and calculating the average value of all classification regression trees, and taking the average value as the final output of the random forest agent auxiliary model.
4. The data-driven multi-target strip mine card path optimization method according to claim 3, wherein in the process of training the classification regression tree, feature node division is performed according to mine card trajectory data in various subsets, the obtained part with the minimum mean square sum error is regarded as the best split among all split points, binary tree splitting is performed in sequence, and after a termination condition is reached, the classification regression tree stops growing.
5. The data-driven multi-target strip mine card path optimization method according to any one of claims 1 to 4, wherein in the step 2, the approximation error of the random forest agent auxiliary model is estimated and corrected by adopting an individual-based model management strategy so as to improve the prediction accuracy, and the method comprises the following steps:
the error between the true and predicted values of the y-th optimization objective for non-dominant individuals is first estimated by the following equation, denoted as eryThe non-dominant individual represents a current optimal mine card transportation scenario;
Figure FDA0003499950720000032
wherein the content of the first and second substances,
Figure FDA0003499950720000033
the true value for the h-th non-dominant individual at the y-th optimization objective,
Figure FDA0003499950720000034
the predicted value of the H non-dominant individual in the y optimization target, wherein H is the number of the non-dominant individuals;
then, the target predicted values of all the non-dominant individuals are corrected according to the error, and the target predicted values are obtainedIf the objective function is minimized, it is corrected by subtracting the corresponding error value from the value of the y-th objective function, i.e. Fy-ery；
And finally, taking the corrected non-dominant individual as a new sample and putting the new sample into a training data set, thereby completing the updating of the random forest agent auxiliary model.
6. The data-driven multi-target strip mine card path optimization method according to claim 1, wherein in the optimization process of the step 3, on the basis of an ant colony algorithm optimization framework, a self-adaptive pheromone updating mechanism is introduced and an ant circumference model is improved, so that the solving efficiency and the scheme quality are improved; and a Pareto fitness evaluation mechanism was introduced to select more potential nondominant individuals.
7. The data-driven multi-objective strip mine card path optimization method of claim 6, wherein the adaptive pheromone update mechanism is as follows:
combining different search weights presented by the ant colony algorithm in different search stages, the pheromone volatilization factor which is self-adaptively adjusted is adopted, and the pheromone volatilization quantity can be self-adaptively adjusted according to the requirement of the search progress, and is shown as the following formula:
Figure FDA0003499950720000041
wherein rho is an pheromone volatilization factor, FE represents the current function evaluation times, and each ant represents a group of mine car transportation schemes;
the method for improving the periant model comprises the following steps:
according to the optimization model, the ant circumference model is improved, and the increment of each ant pheromone is evaluated from two angles of transportation distance and waiting time punishment, so that the ant is guided to explore the optimal path more accurately;
Figure FDA0003499950720000042
wherein Q is1And Q2Respectively the total quantity of pheromones on two targets of minimum time loss penalty cost and minimum total distance cost of transportation in the path optimization model, LeIs the total path length, P, traveled by the e-th antePunishment of violation time window for the e-th ant on the walking path;
the Pareto fitness evaluation mechanism is as follows:
aiming at the characteristics of the multi-objective optimization problem, the evaluation mechanism in the SPEA2 is introduced into the ant colony algorithm, so that the potential of each individual is comprehensively evaluated, and meanwhile, a k-nearest neighbor method is introduced, so that the distribution condition of the individual is taken into consideration, and the formula is shown as follows:
F(e)＝R(e)+D(e)
Figure FDA0003499950720000051
Figure FDA0003499950720000052
Figure FDA0003499950720000053
Figure FDA0003499950720000054
wherein, F (e), R (e) and D (e) respectively represent the adaptability value, dominance level value and position distribution information of the e-th ant, S (e) represents the number of ants dominated by ant e in the population P and the external archive set Q, u is the ant in the set P and Q,
Figure FDA0003499950720000055
is the euclidean distance from the e-th ant to the k-th adjacent ant.
8. The data-driven multi-objective strip mine card path optimization method according to claim 1, wherein the solving process of the step 4 is as follows:
step 1, initializing algorithm parameters and a tabu table, wherein the parameters comprise a population size NpThe number C of classification regression trees, the splitting stop condition T of the classification regression trees, the pheromone importance degree factor alpha, the heuristic information importance degree factor beta, the total pheromone Q, the pheromone volatilization coefficient rho and the maximum evaluation number FEmax；
Step 2: the method comprises the following steps that initial ants search a transportation route from any loading point randomly, and select a route node to be visited next according to transition probability;
step 3: recording route nodes visited by ants in a taboo table;
step 4: judging whether the ants reach any unloading point in the road network of the mining area, if so, executing Step 5, otherwise, returning to Step 2;
step 5: predicting and evaluating all route schemes by using a random forest agent auxiliary model;
step 6: evaluating the quality of the scheme according to a fitness evaluation mechanism, and dividing Pareto grades;
step 7: globally updating pheromones on the paths based on a self-adaptive pheromone updating mechanism and an improved ant periphery model;
step 8: storing the current optimal path scheme, and recording the total transportation distance and time cost of the current optimal path scheme;
step 9: using the current non-dominated individual to carry out error correction and update on the agent model;
step 10: judging the iteration state, and if the iteration state reaches a termination condition, outputting the current optimal Pareto path scheme; otherwise, the taboo list is emptied and the operation returns to Step 2.
CN202210124843.7A 2022-02-10 2022-02-10 Data-driven multi-target strip mine card path optimization method Pending CN114444809A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210124843.7A CN114444809A (en) 2022-02-10 2022-02-10 Data-driven multi-target strip mine card path optimization method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210124843.7A CN114444809A (en) 2022-02-10 2022-02-10 Data-driven multi-target strip mine card path optimization method

Publications (1)

Publication Number Publication Date
CN114444809A true CN114444809A (en) 2022-05-06

Family

ID=81371156

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210124843.7A Pending CN114444809A (en) 2022-02-10 2022-02-10 Data-driven multi-target strip mine card path optimization method

Country Status (1)

Country Link
CN (1) CN114444809A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113393055A (en) * 2021-07-05 2021-09-14 苏州清研捷运信息科技有限公司 Preprocessing and using method of freight car navigation along-the-way traffic control data
CN115409293A (en) * 2022-10-31 2022-11-29 宁波长壁流体动力科技有限公司 Digital twin-based intelligent mine management and control method and management and control facility

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113393055A (en) * 2021-07-05 2021-09-14 苏州清研捷运信息科技有限公司 Preprocessing and using method of freight car navigation along-the-way traffic control data
CN113393055B (en) * 2021-07-05 2023-07-25 苏州清研捷运信息科技有限公司 Pretreatment and use method of truck navigation along-route data
CN115409293A (en) * 2022-10-31 2022-11-29 宁波长壁流体动力科技有限公司 Digital twin-based intelligent mine management and control method and management and control facility

Similar Documents

Publication Publication Date Title
CN111191847B (en) Distribution path planning method and system considering order polymerization degree
CN114444809A (en) Data-driven multi-target strip mine card path optimization method
US20220156693A1 (en) Computerized system and method for developing optimized cargo transportation solutions
CN113222463B (en) Data-driven neural network agent-assisted strip mine unmanned truck scheduling method
Tirado et al. Heuristics for dynamic and stochastic routing in industrial shipping
CN112183812B (en) Finished cigarette logistics vehicle scheduling method considering short-time and low-cost
Mutar et al. An efficient improvement of ant colony system algorithm for handling capacity vehicle routing problem
CN111860989B (en) LSTM neural network short-time traffic flow prediction method based on ant colony optimization
CN115759917A (en) Logistics path planning method based on improved mixed ant colony algorithm
CN113361073B (en) Inventory path joint optimization method based on improved Lagrange relaxation algorithm
CN115907254A (en) Vehicle route planning method based on evolutionary multi-objective reinforcement learning
CN114037335B (en) Vehicle and cargo matching intelligent decision method and system for maximizing carrying capacity
CN116151499A (en) Intelligent multi-mode intermodal route planning method based on improved simulated annealing algorithm
Kravets et al. Multimodal freight transportation based on multicriteria optimization by time indicators
CN116629424A (en) Novel automatic double-yard bridge intelligent scheduling method for container yard operation mode
CN115186930A (en) Method and system for planning path problem of simultaneously-taking and delivering goods vehicle with multiple time windows under uncertain goods taking demand
CN113687651A (en) Path planning method and device for delivering vehicles according to needs
CN116050752A (en) Data-driven neural network agent-assisted strip mine unmanned truck scheduling method
CN112016750A (en) Improved method for solving problem of vehicle path with constraint
CN116629479A (en) AGV multi-target traffic control method and system for single-way collision-free AGV
CN112734111B (en) Horizontal transport task AGV dynamic time prediction method
CN112836846B (en) Multi-depot and multi-direction combined transportation scheduling double-layer optimization algorithm for cigarette delivery
CN112836858B (en) Multi-type intermodal transportation emission reduction path selection method, system and device for containers
CN111445100B (en) Vehicle and goods matching method based on self-adaptive time window under limited transport capacity
CN115470651A (en) Ant colony algorithm-based vehicle path optimization method with road and time window

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination