Underground mine truck transportation cycle operation working hour integrated prediction method

Abstract

The application provides an underground mine truck transportation cycle operation work hour integrated prediction method, including the following steps: S1, the trackless transportation truck cycle operation process is refined into six stages; S2, an operation work hour influencing factor index system under different transportation stages is constructed; S3, truck operation work hour data and influencing factor index data of different transportation stages are collected; S4, based on the Stacking integrated learning framework, a Stacking integrated learning work hour prediction model of multi-model fusion is constructed; S5, the collected operation work hour data and the corresponding work hour influencing factor data are divided into a training set and a test set, and the influencing factor index is taken as input and the operation work hour is taken as output, so that training and testing are carried out, and the prediction model corresponding to the six stages is obtained; the operation work hours of the six stages are respectively predicted through the to-be-predicted parameters, so that the trackless transportation truck cycle operation work hour prediction value is obtained, which is more in line with the actual transportation situation of the mine, and the operation work hour prediction precision is improved.

Description

Technical Field

[0001] This invention belongs to the field of mining equipment operation time prediction technology, specifically relating to an integrated prediction method for the operation time of truck transportation cycles in underground mines. Background Technology

[0002] Mineral resources are a crucial material foundation and guarantee for human societal development. Depending on their burial depth, mineral resources are primarily acquired through two methods: open-pit mining and underground mining. Globally, underground mining still dominates mineral deposit development and utilization. With the increasing demand for mineral resources, the scale of underground mining is also expanding, and the application of extensive mechanized equipment has become an inevitable trend in global underground mine production and construction. Constructing a real-time, precise scheduling system for mine transportation equipment is key to achieving continuous and efficient transportation operations with multiple pieces of equipment. However, current underground mine vehicle scheduling methods do not consider the impact of transportation equipment travel time on scheduling accuracy, easily leading to problems of scheduling plans being ahead of or behind schedule in practical applications. Therefore, predicting the travel time of underground mine trucks is an essential requirement for building a real-time, precise scheduling system.

[0003] Currently, research on predicting the operating time of mining transportation equipment is relatively limited, with a large body of literature focusing primarily on predicting vehicle travel time in urban road and highway transportation systems. Based on prediction models, travel time prediction methods in the transportation field can be categorized into time series prediction models, traffic parameter prediction models, and influencing factor prediction models. Underground mine roads suffer from low traffic density and weak spatiotemporal correlation, and the travel time of transport trucks is significantly affected by environmental and personnel factors, making it prone to fluctuations. Prediction methods based on time series or traffic parameters are not entirely applicable to predicting the travel time of vehicles in underground mines, while prediction models based on travel time influencing factors are closer to the realities of mining. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by providing an integrated prediction method for the cyclical operation time of truck transportation in underground mines. Based on the location and status of the trucks, the cyclical operation process of the trucks is broken down into six stages. Combining the characteristics of each stage, the influencing factors of travel time in each stage are determined, and the truck travel time of each of the six stages is predicted separately. This method is more in line with the actual transportation situation in mines and improves the accuracy of operation time prediction.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] An integrated prediction method for cyclical truck transportation operations in underground mines, comprising the following steps:

[0007] S1. For the trackless transportation process in underground mines, the cyclical operation process of trackless transportation trucks is refined into 6 stages based on the status of the transportation trucks and the operating environment.

[0008] S2. Analyze the factors affecting truck operation time and construct an index system of factors affecting operation time under different transportation stages;

[0009] S3. Collect truck operation time data for different transportation stages, and collect corresponding operation time influencing factor data.

[0010] S4. Based on the Stacking ensemble learning framework, construct a Stacking ensemble learning work time prediction model that integrates multiple models;

[0011] S5. Divide the work time data and corresponding work time influencing factor data collected in step S3 into training set and test set. Use the influencing factor index as input and work time as output to train and test the model corresponding to the 6 stages, and obtain the prediction model corresponding to the 6 stages respectively. Predict the work time of the 6 stages respectively through the parameters to be predicted, so as to obtain the predicted value of trackless transportation cycle operation work time.

[0012] Preferably, in step S1, the process of the truck traveling from the loading point to the unloading point and then back to the loading point is considered as a cyclical operation. This process includes three road sections: an underground horizontal tunnel section, a sloping road section, and a surface road section. Based on the working status of the transport truck on each road section, the cyclical operation is divided into six stages: loaded truck horizontal tunnel operation stage, loaded truck sloping road uphill stage, loaded truck surface operation stage, empty truck surface return stage, empty truck sloping road downhill stage, and empty truck horizontal tunnel return stage.

[0013] Preferably, in step S2, based on the actual trackless transportation situation in the mine and combined with the characteristics of each transportation stage, factors affecting the working hours of the transport trucks are selected, and Pearson correlation analysis is performed on the selected factors to finally obtain several independent influencing factor indicators for each stage.

[0014] Ideally, in determining the influencing factors for each transportation stage, based on the actual transportation situation in the mine, on the sloping road section, considering the need for vehicles going up and down to pass each other, the waiting time at the red light for empty vehicles going down the sloping road is considered an important factor affecting working hours during the downhill phase of the empty vehicle sloping road.

[0015] Preferably, in step S3, the waiting time for a passing empty truck at a red light is allocated to the operating hours of each truck by calculating the probability of a passing truck encountering another truck, including the following process:

[0016] Assuming there are N trucks operating on the route, the length of the flat alley section is L1, the length of the sloping section is L2, and the length of the surface road section is L3, then the probability P that a truck needs to wait to pass another truck on the sloping section is: In the formula, ρ1, ρ2, and ρ3 are the time coefficients for the truck on the three road segments, representing the difficulty of the truck traveling on the corresponding road segments; Let t represent the number of combinations in which two vehicles among N vehicles meet; then the red light waiting time allocated to each truck's travel time is: t w =T w ×P, where T w The red light waiting time is the time specified for vehicles to pass each other on the downhill section of a ramp.

[0017] Preferably, in step S3, the index data of factors affecting working hours includes road surface roughness index, and the road surface roughness of each section of the underground mine trackless transportation is calculated by using road surface image grayscale processing.

[0018] Preferably, the grayscale processing of the road surface image includes the following steps:

[0019] S31. Select five measuring points at equal intervals along the road section, and set up a camera two meters directly above the measuring point and perpendicular to the road surface to take images of the road surface.

[0020] S32. Perform grayscale processing on the obtained road surface image to obtain the grayscale value of each pixel in the image;

[0021] S33. Let the number of pixels in each image be N×M, p ij (i = 1, 2, ..., N; j = 1, 2, ..., M) represents the grayscale data for each pixel, p max Given the maximum grayscale value of each pixel, the formula for calculating the road surface roughness R in each image is:

[0022] S34. The average of the road surface roughness values ​​corresponding to the five measuring points is the road surface roughness of the measured section. The calculation formula is as follows: Where R L To measure the road surface roughness of the road section, R1, R2, R3, R4, and R5 are the road surface roughness at five measuring points.

[0023] Preferably, in step S4, the process of constructing a multi-model fusion Stacking ensemble learning time prediction model includes:

[0024] S41. Based on the Stacking ensemble learning framework, Random Forest (RF), Lightweight Gradient Boosting Machine (LightGBM), and Least Squares Support Vector Machine (LSSVM) are selected as the base models in the Stacking ensemble learning framework, and the prediction results of each base model are used as the input of the meta-model.

[0025] S42. Select the extreme gradient boosting tree XGBoost, which has good stability and accuracy, as the meta-model in the Stacking ensemble learning framework. The output of the meta-model is the final prediction result.

[0026] Preferably, in step S5, the factors affecting working hours are used as input parameters and the working hours are used as output parameters. A loop structure is constructed using Python to iterate the hyperparameters of each single model. The mean absolute error (MAE) is used as the evaluation index of the prediction results to seek the optimal hyperparameters of each model.

[0027] Preferably, in step S5, the process for predicting the working hours of the trackless transport cycle operation is as follows:

[0028] Let the predicted values ​​for the six transportation stages be t1, t2, t3, t4, t5, and t6, then the predicted working hours for the trackless transportation cycle operation are: Among them, t w Red light waiting time allocated to each truck's travel time.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] (1) Based on the location and status of the truck, the present invention refines the truck operation cycle into 6 stages. Combining the characteristics of each stage, the indicators of the travel time influencing factors of each stage are determined, and the truck running time of each of the 6 stages is predicted, which is more in line with the actual transportation situation in the mine.

[0031] (2) In the downhill phase of the ramp, the present invention takes into account the red light waiting time of the downhill transport trucks. By calculating the probability of passing, the red light waiting time is evenly distributed to the running time of each vehicle, which is close to the actual mining production and can more accurately reflect the actual transportation situation of mining trucks.

[0032] (3) Based on the Stacking learning framework, XGBoost, which has the best generalization ability, is used as the meta-model, and RF, LightGBM and LSSVM are used as the base models in the Stacking integrated learning framework. A multi-model fusion transportation operation time integrated prediction model is constructed, which is more adaptable to the characteristics of trackless transportation roads in underground mines and adaptable to the prediction of transportation truck time in different transportation stages. Attached Figure Description

[0033] Figure 1 This is a flowchart of an integrated prediction method for the working hours of underground mine truck transportation cycles, as described in an embodiment of the present invention.

[0034] Figure 2 This is a summary representation of the factors influencing working hours at each stage of transportation in an embodiment of the present invention.

[0035] Figure 3 This is a schematic diagram illustrating the influence of each base model hyperparameter on the prediction performance in an embodiment of the present invention.

[0036] Figure 4 This is a schematic representation of the optimal hyperparameters of each model in the embodiments of the present invention.

[0037] Figure 5 This is a schematic diagram showing the comparison between the predicted and actual working hours of the test sample in an embodiment of the present invention.

[0038] Figure 6 This is a schematic representation of the evaluation index parameters of the test samples in an embodiment of the present invention. Detailed Implementation

[0039] The present invention will be further described below with reference to the embodiments shown in the accompanying drawings.

[0040] As attached Figure 1 As shown in the figure, this embodiment discloses an integrated prediction method for the working hours of truck transportation cycles in underground mines. The prediction method includes the following steps:

[0041] Step S1: For the trackless transportation process in underground mines, the trackless transportation truck cycle operation process is refined into 6 stages based on the status of the transport trucks and the operating environment.

[0042] In step S1, the process of driving the truck from the loading point to the unloading point and then back to the loading point is a cyclical operation. This process includes three road sections: an underground horizontal tunnel section, a sloping road section, and a surface road section. Based on the working status of the transport truck on each road section, the cyclical operation is divided into six stages: loaded truck horizontal tunnel operation stage, loaded truck sloping road uphill stage, loaded truck surface operation stage, empty truck surface return stage, empty truck sloping road downhill stage, and empty truck horizontal tunnel return stage.

[0043] Step S2: Analyze the factors affecting truck operation time and construct an index system of factors affecting operation time under different transportation stages.

[0044] In step S2, based on the actual trackless transportation situation in the mine and combined with the characteristics of each transportation stage, the influencing factors affecting the working hours of the transport trucks are selected. Pearson correlation analysis is performed on the selected influencing factors to finally obtain several independent influencing factor indicators for each stage.

[0045] In another embodiment, during the process of determining the influencing factors for each transportation stage in step S2, based on the actual transportation situation in the mine, traffic signal devices are specifically installed on the sloping road sections where vehicles going up and down need to pass each other. In this case, the principle of empty cars yielding to loaded cars and downhill cars yielding to uphill cars should be followed, and passing should take place in a passing chamber. Therefore, the waiting time for passing empty cars at the red light on the downhill sloping road is considered an important factor affecting the working hours during the downhill phase of the empty car sloping road.

[0046] S3. Collect truck operation time data for different transportation stages, and collect corresponding operation time influencing factor data.

[0047] In step S3, the indicators of factors affecting work hours include quantitative indicators and qualitative indicators. Quantitative indicators can be obtained by measuring with relevant instruments or by consulting relevant data.

[0048] For example, the truck's load capacity can be measured in real time by a weighbridge system. The truck's speed can be recorded by the display panel in the truck's cab at the time of testing. The gradient of the ramp is measured by a slope meter. The oxygen concentration in the tunnel is measured and recorded by an underground oxygen detector. The truck driver's length of service, the truck's age, the number of turns, and the number of times pedestrians were avoided can all be obtained from relevant mine data.

[0049] Regarding driver skill levels, according to China's national vocational qualification certificate system, truck drivers are classified into four levels: entry-level, intermediate-level, advanced-level, and technician, represented by the numbers 1, 2, 3, and 4. The corresponding data can be obtained by consulting the vocational qualification certificate for mining truck drivers. Rain and snow weather indicators divide the weather into three levels: no rainfall, light rainfall, and heavy rainfall, represented by the numbers 0, 1, and 2, obtained by consulting the monitoring data from the China Meteorological Administration for that day.

[0050] In addition, regarding the key factor indicator of the downhill empty truck passing red light waiting time identified in step S2, the downhill empty truck passing red light waiting time can be allocated to the working hours of each truck by calculating the probability of the transport trucks passing each other. The specific process includes the following:

[0051] Assuming there are N trucks operating on the route, the length of the flat alley section is L1, the length of the sloping section is L2, and the length of the surface road section is L3, then the probability P that a truck needs to wait to pass another truck on the sloping section is: In the formula, ρ1, ρ2, and ρ3 are the time coefficients for the truck on the three road segments, representing the difficulty of the truck traveling on the corresponding road segments; Let t represent the number of combinations in which two vehicles among N vehicles meet; then the red light waiting time allocated to each truck's travel time is: t w =T w ×P, where Tw The red light waiting time is the time specified for vehicles to pass each other on the downhill section of a ramp.

[0052] In another embodiment, in step S3, the data on factors affecting working hours may also include road surface roughness index, and the road surface roughness of each section of the underground mine trackless transportation is calculated using road surface image grayscale processing.

[0053] Specifically, the grayscale processing of road surface images includes the following steps:

[0054] S31. Select five measuring points at equal intervals along the measurement section, and set up a camera two meters directly above the measuring point and perpendicular to the road surface to take images of the road surface.

[0055] S32. Perform grayscale processing on the obtained road surface image to obtain the grayscale value of each pixel in the image.

[0056] S33. Let the number of pixels in each image be N×M, p ij (i = 1, 2, ..., N; j = 1, 2, ..., M) represents the grayscale data for each pixel, p max Given the maximum grayscale value of each pixel, the formula for calculating the road surface roughness R in each image is:

[0057] S34. The average of the road surface roughness values ​​corresponding to the five measuring points is the road surface roughness of the measured section. The calculation formula is as follows: Where R L To measure the road surface roughness of the road section, R1, R2, R3, R4, and R5 are the road surface roughness at five measuring points.

[0058] S4. Based on the Stacking ensemble learning framework, construct a Stacking ensemble learning time prediction model that integrates multiple models.

[0059] In step S4, the process of constructing a multi-model fusion Stacking ensemble learning time prediction model includes:

[0060] S41. Based on the Stacking ensemble learning framework, Random Forest (RF), Lightweight Gradient Boosting Machine (LightGBM), and Least Squares Support Vector Machine (LSSVM) are selected as the base models in the Stacking ensemble learning framework, and the prediction results of each base model are used as the input of the meta-model.

[0061] S42. Select the extreme gradient boosting tree XGBoost, which has good stability and accuracy, as the meta-model in the Stacking ensemble learning framework. The output of the meta-model is the final prediction result.

[0062] S5. Divide the work time data and corresponding work time influencing factor data collected in step S3 into training set and test set. Use the influencing factor index as input and work time as output to train and test the model corresponding to the 6 stages, and obtain the prediction model corresponding to the 6 stages respectively. Predict the work time of the 6 stages respectively through the parameters to be predicted, so as to obtain the predicted value of trackless transportation cycle work time.

[0063] In step S5, using the factors affecting working hours as input parameters and the working hours as output parameters, a loop structure is constructed using Python to iterate the hyperparameters of each single model. The mean absolute error (MAE) is used as the evaluation index of the prediction results to seek the optimal hyperparameters of each model.

[0064] In addition, in step S5, the predicted values ​​of the six transportation stages are added together to obtain the predicted value of the cyclic operation time. The process of predicting the cyclic operation time of trackless transportation is as follows:

[0065] Let the predicted values ​​for the six transportation stages be t1, t2, t3, t4, t5, and t6, then the predicted working hours for the trackless transportation cycle operation are: Among them, t w Red light waiting time allocated to each truck's travel time.

[0066] The prediction method in this embodiment refines the truck operation cycle into six stages based on the truck's location and status. It identifies various influencing factors for travel time in each stage and predicts truck travel time for each of the six stages separately, better reflecting actual mine transportation conditions. Furthermore, in the downhill phase of the ramp, the red light waiting time for passing trucks is considered. By calculating the probability of passing situations, the red light waiting time is evenly distributed across each truck's travel time, closely aligning with actual mine production and more accurately reflecting the actual transportation situation of mine trucks. Additionally, based on the Stacking learning framework, using XGBoost (which has the best generalization ability) as the meta-model and RF, LightGBM, and LSSVM as base models in the Stacking ensemble learning framework, a multi-model fusion transportation operation time prediction model is constructed. This model is more adaptable to the characteristics of trackless transportation roads in underground mines and can better predict truck travel time at different transportation stages.

[0067] The following section of a trackless transportation operation in a domestic underground mine is selected as the experimental section to verify the prediction method of this embodiment.

[0068] Step S1: In the trackless transportation process of underground mines in this experimental section, the truck travels from the loading point to the unloading point and then back to the loading point, constituting a cyclical operation. The process mainly involves three sections: an underground horizontal tunnel section, a sloping road section, and a surface section. Based on the truck's operating status on each section (loaded or empty), the cyclical operation is divided into six stages: loaded truck horizontal tunnel operation stage, loaded truck sloping road uphill stage, loaded truck surface operation stage, empty truck surface return stage, empty truck sloping road downhill stage, and empty truck horizontal tunnel return stage.

[0069] Step S2: Through investigation of trackless transportation in the mine and interviews with relevant management personnel, the factors affecting the operating hours of transport trucks were analyzed from multiple perspectives. Preliminary selection of influencing factors for each transportation stage was made, and Pearson correlation analysis was conducted on these preliminary factors to obtain several independent influencing factor indicators for each stage. The specific influencing factor indicators for the six transportation stages are detailed below. Figure 2 .

[0070] Step S3: Track the trackless transportation cycle operation process of the underground mine experimental section, continuously collect the working hours data of the transportation trucks according to the 6 divided transportation stages, and collect the corresponding working hours influencing factor data. 200 sets of sample data are collected for each transportation stage.

[0071] For the waiting time at a red light when passing other trucks, the probability of a truck passing another truck is calculated, and this waiting time is then allocated to each truck's operating hours. Assuming there are N trucks operating on the route, the length of the flat road section is L1, the length of the sloping road section is L2, and the length of the surface road section is L3, then the probability P of a truck needing to wait to pass another truck on the sloping road is: In the formula: ρ1, ρ2, and ρ3 are the time coefficients for the truck on the three road segments, representing the difficulty of the truck traveling on the corresponding road segments; Let t represent the number of combinations in which two trucks among N trucks meet. Then the red light waiting time allocated to each truck's travel time is: t w =T w ×P, where T w This refers to the red light waiting time stipulated for vehicles to pass each other during the downhill phase of the ramp. According to mine data, the number of trucks operating on the experimental section is 5. The length of the horizontal tunnel in section 2822 is approximately 3800 meters, the length of the ramp is approximately 3600 meters, and the length of the surface road section is approximately 3000 meters. ρ1, ρ2, and ρ1 are taken as 1, 1.2, and 0.8 respectively. The statistical analysis is performed on T. w The value is approximately 120 seconds. Therefore, t can be calculated using the formula above. w It lasted 40.46 seconds.

[0072] Step S4: Based on the Stacking ensemble learning framework, select RF (Random Forest), LightGBM (Lightweight Gradient Boosting Machine), and LSSVM (Least Squares Support Vector Machine) as base models in the Stacking ensemble learning framework. The initial predictions of each base model are used as the input of the meta-model. Select XGBoost (Extreme Gradient Boosting Tree), which has better stability and accuracy, to replace the conventional linear regression algorithm as the meta-model in the Stacking ensemble learning framework. The output of the meta-model is the final prediction result.

[0073] Step S5: The underground mine transport truck operation process is divided into three sections: underground horizontal tunnels, underground inclined roads, and surface roads. The underground inclined road section has a more complex transportation environment and involves more influencing factors than the other two sections. Therefore, more representative and comprehensive inclined road sample data is selected as the training sample. Using the factors affecting work hours as input parameters and the work hours as output parameters, a loop structure is constructed using Python to iterate the hyperparameters of each single model. The mean absolute error (MAE) is used as the evaluation index for the prediction results to seek the optimal hyperparameters for each model. Finally, the influence of the hyperparameters of each base model on the prediction effect is obtained (see [link to relevant documentation]). Figure 3 As shown, the optimal hyperparameters for each model are listed below. Figure 4 As shown.

[0074] The optimal hyperparameters of each model are substituted into the fusion model, and the operation time data and corresponding influencing factor data collected in step S3 are divided into training and testing sets, with 160 groups in the training set and 40 groups in the testing set. Using influencing factor indicators as input and operation time as output, the model is trained and tested to predict the operation time for each of the six transportation stages. Finally, the predicted operation time for the six stages is added together to obtain the predicted value of the trackless transportation cycle operation time. Mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R²) are selected as evaluation indicators for the test samples to evaluate the model's accuracy.

[0075] Let the predicted values ​​for the six transportation stages be t1, t2, t3, t4, t5, and t6, then the cyclical working hours of the transportation trucks are: Among them, t w The red light waiting time allocated to each truck's travel time is calculated as 40.46 seconds by the formula in step S3.

[0076] The predicted values ​​at each stage and t w The sum is the predicted cycle time T for transport truck operations. P , Figure 5The image shows a comparison between predicted and actual values, including the mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R²) for the predicted samples. 2 (See indicators) Figure 6 As shown. By Figure 5 and Figure 6 It can be seen that the predicted values ​​and the actual values ​​have a high degree of fit, and R0 is high. 2 It reached 0.9762, with a mean absolute error of 146.74 seconds.

[0077] The above experiments demonstrate that the integrated prediction model and method for underground mine transportation operation time proposed in this invention has strong practicality and can meet the data requirements of the intelligent mine scheduling system within the allowable error range.

[0078] The scope of protection of this invention is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its scope and spirit. If these modifications and variations fall within the scope of the claims of this invention and their equivalents, then the intent of this invention also includes these modifications and variations.

Claims

1. A method for integrated prediction of working hours in cyclical truck transportation operations in underground mines, characterized in that, The prediction method includes the following steps: S1. For the trackless transportation process in underground mines, the cyclical operation process of trackless transportation trucks is refined into 6 stages based on the status of the transportation trucks and the operating environment. S2. Analyze the factors affecting truck operation time and construct an index system of factors affecting operation time under different transportation stages; S3. Collect truck operation time data for different transportation stages, and collect corresponding operation time influencing factor data. S4. Based on the Stacking ensemble learning framework, construct a Stacking ensemble learning work time prediction model that integrates multiple models; S5. Divide the work time data and corresponding work time influencing factor data collected in step S3 into training set and test set. Use the influencing factor index as input and work time as output to train and test the model corresponding to the 6 stages, and obtain the prediction model corresponding to the 6 stages respectively. Predict the work time of the 6 stages respectively through the parameters to be predicted, so as to obtain the predicted value of trackless transportation cycle work time. In step S1, the process of driving the truck from the loading point to the unloading point and back to the loading point is a cyclical operation. This process includes three road sections: an underground horizontal tunnel section, a sloping road section, and a surface road section. Based on the working status of the transport truck on each road section, the cyclical operation is divided into six stages: loaded truck horizontal tunnel operation stage, loaded truck sloping road uphill stage, loaded truck surface operation stage, empty truck surface return stage, empty truck sloping road downhill stage, and empty truck horizontal tunnel return stage. In step S2, based on the actual trackless transportation situation in the mine and combined with the characteristics of each transportation stage, factors affecting the working hours of the transport trucks are selected, and Pearson correlation analysis is performed on the selected factors to finally obtain several independent influencing factor indicators for each stage. In determining the influencing factors for each transportation stage, based on the actual transportation situation in the mine, on the sloping road section, considering the need for vehicles going up and down to pass each other, the waiting time at the red light for empty vehicles going down the sloping road is taken as an important factor affecting the working hours during the downhill stage of empty vehicles. In step S3, the waiting time for a passing empty truck at a red light is allocated to the operating hours of each truck by calculating the probability of a passing truck meeting another truck. This includes the following process: Assume the number of trucks running on the route is The length of the flat alley section is The length of the sloping road section is The length of the surface road section is The probability that a truck needs to wait to pass another vehicle on a slope is... for: In the formula, , , The time coefficient for the truck on the three road segments indicates the difficulty of the truck traveling on the corresponding road segments; express The number of combinations in which two vehicles meet; then the red light waiting time allocated to each truck's travel time is: In the formula, The red light waiting time specified for passing other vehicles on the downhill section of a ramp; In step S5, the process for predicting the working hours of the trackless transportation cycle operation is as follows: Let the predicted values ​​for the 6 transportation stages be as follows: , , , , , The predicted working hours for trackless transport cyclic operations are: ,in, Red light waiting time allocated to each truck's travel time.

2. The integrated prediction method for cyclic operation time of truck transportation in underground mines according to claim 1, characterized in that: In step S3, the index data of factors affecting working hours include road surface roughness index, and the road surface roughness of each section of the underground mine trackless transportation is calculated by using road surface image grayscale processing.

3. The integrated prediction method for cyclic operation time of truck transportation in underground mines according to claim 2, characterized in that: The grayscale processing of the road surface image includes the following steps: S31. Select five measuring points at equal intervals along the road section, and set up a camera two meters directly above the measuring point and perpendicular to the road surface to take images of the road surface. S32. Perform grayscale processing on the obtained road surface image to obtain the grayscale value of each pixel in the image; S33. Let the number of pixels in each image be... , Grayscale data for each pixel, Given the maximum grayscale value of each pixel, the formula for calculating the road surface roughness R in each image is: ; S34. The average of the road surface roughness values ​​corresponding to the five measuring points is the road surface roughness of the measured section. The calculation formula is as follows: ,in To measure the road surface roughness of the road section, , , , , The road surface roughness is measured at five points.

4. The integrated prediction method for cyclical operation time of truck transportation in underground mines according to claim 1, characterized in that: In step S4, the process of constructing a multi-model fusion Stacking ensemble learning time prediction model includes: S41. Based on the Stacking ensemble learning framework, Random Forest (RF), Lightweight Gradient Boosting Machine (LightGBM), and Least Squares Support Vector Machine (LSSVM) are selected as the base models in the Stacking ensemble learning framework, and the prediction results of each base model are used as the input of the meta-model. S42. Select the extreme gradient boosting tree XGBoost, which has good stability and accuracy, as the meta-model in the Stacking ensemble learning framework. The output of the meta-model is the final prediction result.

5. The integrated prediction method for cyclical operation time of truck transportation in underground mines according to claim 1, characterized in that: In step S5, using the factors affecting working hours as input parameters and the working hours as output parameters, a loop structure is constructed using Python to iterate the hyperparameters of each single model. The mean absolute error (MAE) is used as the evaluation index of the prediction results to seek the optimal hyperparameters of each model.

Images