Mine muck pile shoveling efficiency statistical method and system based on multi-dimensional data

Abstract

The invention relates to the field of mining working efficiency analysis, in particular to a mine muck pile shoveling efficiency statistical method and system based on multi-dimensional data, and the method comprises the following steps: obtaining multi-dimensional real-time operation data of a carry-scraper in a mine muck pile shoveling process; obtaining multi-modal fusion feature information data of the carry-scraper based on the multi-dimensional real-time operation data; obtaining operation cycle information data of the carry-scraper according to the multi-modal fusion feature information data; constructing a shovel loading efficiency statistical optimization model of the carry-scraper in combination with the operation cycle information data; and obtaining a shoveling efficiency statistical result of the carry-scraper according to the shoveling efficiency statistical optimization model, and completing statistics of the mine muck pile shoveling efficiency. The multi-modal feature information is obtained based on the multi-dimensional operation data of the carry-scraper, the working state of the carry-scraper can be analyzed to obtain the operation cycle, then the shovel loading efficiency statistical optimization model is constructed, and the accuracy of shovel loading efficiency statistics is improved.

Description

Technical Field

[0001] The invention relates to the field of mining efficiency analysis, and in particular to a method and system for statistically analyzing mining blast pile shoveling efficiency based on multi-dimensional data. Background Art

[0002] At present, in the ore mining and transportation process of underground mines, loaders are the main mechanical equipment. With their excellent loading capacity and highly flexible maneuverability, loaders can efficiently dig ore from the blast pile and transport it to the designated location, which greatly promotes the improvement of the overall efficiency of mining. However, it is difficult to accurately calculate the loading efficiency of loaders during the blast pile loading operation in mines.

[0003] The traditional method of shoveling efficiency statistics mainly relies on manual calibration, which not only consumes a lot of human resources and increases unnecessary workload, but also due to the interference of human factors, the accuracy and reliability of the statistical results are often difficult to guarantee. The manual calibration process is cumbersome and time-consuming, and the statistical efficiency is low, making it difficult to provide accurate and effective data support for mine management decisions in a timely manner.

[0004] In view of this, improving the statistical method of shoveling efficiency of shovel loaders and realizing efficient and accurate automated statistics have become technical problems that need to be solved urgently in the field of mining. The present invention proposes a statistical method and system for mining blast pile loading efficiency based on multidimensional data; making full use of the precise measurement capability of sensors, it can monitor the various operating parameters of shovel loaders in real time and accurately during the loading process, thereby realizing high-precision statistics of loading efficiency; avoiding the tediousness of traditional manual calibration methods, and showing the ability of efficient real-time monitoring and optimization of operating processes in complex and changeable mining environments. Engineering personnel can more accurately grasp the working status of shovel loaders, and promptly discover and solve potential problems, thereby further improving the overall efficiency and safety of mining. Summary of the invention

[0005] In view of the defects in the prior art, the present invention provides a method and system for calculating the efficiency of mine blasting pile shoveling based on multi-dimensional data.

[0006] In order to achieve the above-mentioned objectives, in a first aspect, the present invention provides a method for statistically analyzing mine blast pile shoveling efficiency based on multidimensional data, the method comprising the following steps: obtaining multidimensional real-time operation data of a shovel loader during the mine blast pile shoveling process; obtaining multimodal fusion feature information data of the shovel loader based on the multidimensional real-time operation data; obtaining operation cycle information data of the shovel loader according to the multimodal fusion feature information data; constructing a statistical optimization model of the shoveling efficiency of the shovel loader in combination with the operation cycle information data; obtaining the statistical results of the shoveling efficiency of the shovel loader according to the statistical optimization model of the shoveling efficiency, and completing the statistics of the mine blast pile shoveling efficiency. The present invention comprehensively and in real time captures the multi-dimensional operating data of the scraper during the mine blast pile shoveling process, and then obtains multi-modal feature information through data fusion, which provides a data basis for accurately analyzing the working status of the scraper; using the multi-modal feature information to obtain the operating cycle of the scraper helps to deeply understand its operating rhythm and efficiency bottleneck; the loading efficiency statistical optimization model constructed in combination with the operating cycle information improves the accuracy of the loading efficiency evaluation, and provides a scientific basis for optimizing the operating efficiency of the scraper; it effectively promotes the intelligent management of mine blast pile shoveling operations.

[0007] Optionally, the acquisition of multi-dimensional real-time operation data of the shovel loader during the mine blast pile shoveling process includes: configuring and installing a three-dimensional sensor on the shovel loader, and setting the data sampling frequency of the three-dimensional sensor; based on the data sampling frequency, using the three-dimensional sensor to collect the multi-dimensional real-time operation data of the shovel loader during the mine blast pile shoveling process, and the multi-dimensional real-time operation data includes angle, angular velocity and acceleration. The present invention accurately captures the dynamic changes of the shovel loader during the mine blast pile shoveling process by configuring a three-dimensional sensor and reasonably setting the data sampling frequency; using high-precision three-dimensional sensors to collect the multi-dimensional operation data of the shovel loader in real time, including angle, angular velocity and acceleration, providing a rich and accurate data basis for subsequent data analysis and efficiency evaluation; ensuring the real-time and comprehensiveness of the data, and improving the precision and accuracy of the monitoring of the shoveling operation process; better grasping the working status of the shovel loader through real-time data, timely discovering potential problems, and laying a solid foundation for statistical shoveling efficiency, improving operation safety and overall operational efficiency.

[0008] Optionally, the method of obtaining the multimodal fusion feature information data of the shovel loader based on the multidimensional real-time operation data includes: preprocessing the multidimensional real-time operation data to obtain the multidimensional real-time operation feature data of the shovel loader; and fusing the multidimensional real-time operation feature data to obtain the multimodal fusion feature information data of the shovel loader. The present invention preprocesses the multidimensional real-time operation data of the shovel loader to effectively remove noise and outliers, thereby obtaining multidimensional real-time operation feature data that reflects the actual working state of the shovel loader; and then completes data fusion to generate multimodal fusion feature information data, thereby enhancing the representativeness and availability of the data, and providing comprehensive and accurate data support for subsequent analysis and modeling. Multimodal fusion feature information data helps to more accurately understand the operating characteristics of the shovel loader, provides a scientific basis for the statistics of shoveling efficiency, and provides strong support for intelligent management decisions of mine shoveling operations.

[0009] Optionally, the performing data fusion on the multi-dimensional real-time operation feature data to obtain multi-modal fusion feature information data of the scraper includes:

[0010]

[0011] in, is the comprehensive feature of multimodal fusion. is the traversal count flag, It is the dynamic weighting coefficient of multi-dimensional real-time running characteristic data. The present invention realizes the comprehensive processing of multi-dimensional real-time operation characteristic data of the scraper to obtain multi-modal fusion characteristic information data; introduces dynamic weighting coefficients, and flexibly adjusts according to the importance and real-time performance of different features to ensure the accuracy and reliability of the fusion results; improves the efficiency and accuracy of data fusion, and provides accurate data support for the optimization of the scraper loading efficiency; multi-modal fusion characteristic information data helps to deeply explore the potential laws in the scraper operation process, and provides a scientific basis for the statistics of the loading efficiency of mining operations.

[0012] Optionally, the operation cycle information data of the shovel loader is obtained according to the multimodal fusion feature information data, including: obtaining a comprehensive feature vector based on the multimodal fusion feature information data, thereby identifying the complete action cycle of the shovel loader; performing anomaly detection on the complete action cycle to obtain the operation cycle information data of the shovel loader. The present invention uses multimodal fusion feature information data to obtain a comprehensive feature vector, accurately identifies the complete action cycle of the shovel loader, and provides support for in-depth analysis of the working status of the shovel loader; performs anomaly detection on the complete action cycle to promptly discover possible faults or anomalies that may occur during the operation of the shovel loader, thereby obtaining operation cycle information data; enhances the controllability and safety of the shovel loader operation process, and provides a basis for subsequent maintenance management and efficiency statistics.

[0013] Optionally, obtaining a comprehensive feature vector based on the multimodal fusion feature information data to identify a complete action cycle of the scraper includes:

[0014]

[0015] in, is the comprehensive feature vector, Indicates the angle change amplitude, Indicates the speed of angle change, is the mean angular velocity, is the standard deviation of angular velocity, is the mean acceleration, The present invention uses multimodal fusion feature information data to obtain a comprehensive feature vector, and uses the angle change amplitude, angle change speed, angular velocity mean and standard deviation, acceleration mean and peak value to jointly reflect the dynamic characteristics of the scraper during operation; it provides strong data support for accurately identifying the complete action cycle of the scraper; it not only improves the accuracy and reliability of action cycle identification, but also provides an important basis for in-depth analysis of the work efficiency of the scraper; it more accurately grasps the operating status of the scraper, and provides data support for subsequent statistical shovel efficiency.

[0016] Optionally, the construction of the loading efficiency statistical optimization model of the shovel loader in combination with the operation cycle information data includes: performing statistical analysis on the operation cycle information data to obtain the loading times and operating efficiency of the shovel loader, and using the loading times and operating efficiency as output targets; using the comprehensive feature vector as input feature, and constructing a mapping function between the input feature and the output target as the loading efficiency statistical model of the shovel loader; and training and optimizing the loading efficiency statistical model to obtain the loading efficiency statistical optimization model of the shovel loader. The present invention constructs a loading efficiency statistical optimization model in combination with the operation cycle information data of the shovel loader; accurately calculates the loading times and operating efficiency of the shovel loader as the output target of the model through statistical analysis of the operation cycle information; uses the comprehensive feature vector as input feature, and establishes a mapping function between input and output as the loading efficiency statistical model; obtains the loading efficiency statistical optimization model after training and optimization; improves the evaluation accuracy of the loading efficiency of the shovel loader, and provides a scientific basis for the operation optimization and efficiency improvement of the shovel loader.

[0017] Optionally, the taking the comprehensive feature vector as an input feature and constructing a mapping function between the input feature and the output target as a statistical model of the loading efficiency of the scraper includes:

[0018]

[0019] in, is the model output, represents the mapping function, is the comprehensive feature vector, The present invention uses the comprehensive feature vector as the input feature, constructs a statistical model of the shoveling efficiency of the scraper, accurately maps the relationship between the input feature and the shoveling efficiency, realizes the accurate statistics and evaluation of the shoveling efficiency of the scraper, improves the scientificity and accuracy of the statistics of the shoveling efficiency, and provides strong data support and theoretical basis for the optimization of the operating efficiency of the scraper.

[0020] Optionally, the shoveling efficiency statistical results of the shoveling loader are obtained according to the shoveling efficiency statistical optimization model, and the statistics of the mine blast pile shoveling efficiency are completed, including: fitting and analyzing the mine blast pile shoveling process of the shovel loader according to the shoveling efficiency statistical optimization model to obtain the shoveling efficiency statistical results. The present invention uses the comprehensive feature vector as input and the shoveling efficiency statistical model as output to construct a mapping function; through the adjustment and optimization of model parameters, it can accurately reflect the relationship between the dynamic characteristics and shoveling efficiency during the operation of the shovel loader; it realizes the precise mapping from the shovel loader operation data to the shoveling efficiency, more accurately evaluates the operation efficiency of the shovel loader, provides strong support for subsequent optimization control and operation management, and helps to improve the intelligence level of mine operations and the overall operational efficiency.

[0021] In the second aspect, the present invention provides a mine blast pile shoveling efficiency statistics system based on multidimensional data, the system executes the mine blast pile shoveling efficiency statistics method based on multidimensional data provided by the present invention, the system includes an input device, an output device, a processor and a memory, and its gain lies in: the hardware facilities integrated by the present invention have excellent performance, the input device, the output device, the processor and the memory are interconnected, the information transmission between the various components is smooth, and an efficient information processing system is constructed through the interaction of multiple hardware facilities. The present invention constructs an efficient information processing platform by integrating high-performance hardware facilities, ensuring that the system can quickly process a large amount of multidimensional data. Through the collaborative work of multiple hardware facilities, the system can accurately and quickly count and analyze the mine blast pile shoveling efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of a method for statistically analyzing the mining pile blasting efficiency based on multi-dimensional data according to an embodiment of the present invention;

[0023] Figure 2 A schematic diagram of the structure of a three-dimensional sensor according to an embodiment of the present invention;

[0024] Figure 3 A schematic diagram of the installation position of a three-dimensional sensor according to an embodiment of the present invention;

[0025] Figure 4 It is a schematic diagram of the lifting stage of the scraper according to an embodiment of the present invention;

[0026] Figure 5 This is a schematic diagram of a return stage of a scraper according to an embodiment of the present invention;

[0027] Figure 6 is a histogram of the number of shoveling times of the scraper according to an embodiment of the present invention;

[0028] Figure 7 is an operating efficiency distribution curve diagram of a scraper according to an embodiment of the present invention;

[0029] Figure 8 A time stacked bar chart of the action phases of a scraper according to an embodiment of the present invention;

[0030] Fig. 9 A pie chart showing the time proportion of the action phases of the scraper according to an embodiment of the present invention;

[0031] Fig.10 A time series diagram of a scraper according to an embodiment of the present invention;

[0032] Fig.11 A feature correlation heat map of a scraper according to an embodiment of the present invention;

[0033] Fig.12 A paired scatter matrix diagram of a scraper according to an embodiment of the present invention;

[0034] Fig.13 is a feature importance ranking diagram of a scraper according to an embodiment of the present invention;

[0035] Fig.14 This is a framework diagram of a mine blast pile shoveling efficiency statistics system based on multidimensional data in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0037] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0038] See also Figure 1 An embodiment of the present invention provides a method for calculating mining pile shoveling efficiency statistics based on multidimensional data, the method comprising the following steps:

[0039] S1. Obtain multi-dimensional real-time operation data of the loader during the mine blast pile loading process.

[0040] Among them, S1 specifically includes the following steps:

[0041] S11. Install a three-dimensional sensor on the scraper and set a data sampling frequency of the three-dimensional sensor.

[0042] See also Figure 2 , the figure is a schematic diagram of the three-dimensional sensor structure. The three-dimensional sensor in the figure is marked with three directions of X, Y and Z. The three-dimensional sensor can collect angle, angular velocity and acceleration change data in the three directions of X, Y and Z, providing solid data support.

[0043] See also Figure 3 The figure shows the installation position of the 3D sensor. First, fully charge the 3D sensor to ensure normal operation, and then install the 3D sensor at the middle and rear end of the arm of the scraper. During the installation of the 3D sensor, the X direction needs to be pointed directly in front of the scraper.

[0044] Specifically, the data sampling frequency of the three-dimensional sensor satisfies the following relationship:

[0045]

[0046] in, For the Sampling time points, is the sampling time interval, is the sampling frequency.

[0047] S12. Based on the data sampling frequency, the three-dimensional sensor is used to collect multi-dimensional real-time operation data of the shovel loader during the mine blast pile shoveling process, wherein the multi-dimensional real-time operation data includes angle, angular velocity and acceleration.

[0048] In this embodiment, a three-dimensional sensor with a set data sampling frequency is used to collect the angle, angular velocity and acceleration of the upper arm of the scraper during the loading process to form multi-dimensional real-time operation data of the scraper, providing a data basis for subsequent analysis.

[0049] Specifically, the time series of the angles output by the three-dimensional sensor satisfies the following relationship:

[0050]

[0051] in, is the angle of the upper arm, is the angle of the upper arm in the X direction, is the angle of the upper arm in the Y direction, is the angle of the upper arm in the Z direction.

[0052] Furthermore, the angular velocity indicates how fast the angle of the arm changes per unit time, reflecting the dynamic characteristics of the scraper during the loading process. The time series of the angular velocity satisfies the following relationship:

[0053]

[0054] in, is the angular velocity of the upper arm, is the angular velocity of the upper arm in the X direction, is the angular velocity of the upper arm in the Y direction, is the angular velocity of the upper arm in the Z direction.

[0055] More specifically, acceleration represents the acceleration of the arm of the scraper during operation, which is used to describe the start, stop and vibration characteristics of the scraper's shoveling action. The time series of the acceleration satisfies the following relationship:

[0056]

[0057] in, is the acceleration of the upper arm, is the acceleration of the upper arm in the X direction, is the acceleration of the arm in the Y direction, is the acceleration of the arm in the Z direction.

[0058] S2. Obtain multimodal fusion feature information data of the scraper based on the multidimensional real-time operation data.

[0059] Among them, S2 specifically includes the following steps:

[0060] S21. Preprocess the multi-dimensional real-time operation data to obtain multi-dimensional real-time operation characteristic data of the scraper.

[0061] Among them, S21 specifically includes the following steps:

[0062] S211. Data preprocessing.

[0063] Specifically, the state variables satisfy the following relationship:

[0064]

[0065] in, is the state vector, is the angle, is the angular velocity, is the acceleration, Indicates transpose.

[0066] Furthermore, the prediction of the state vector satisfies the following relationship:

[0067]

[0068] in, is the current state vector, is the state transfer matrix, is the previous state vector, is the control input matrix, is the control input, is the process noise.

[0069] Furthermore, the observation correction of the state vector satisfies the following relationship:

[0070]

[0071] in, is the corrected state vector, Based on time About the moment The predicted estimated value of is the Kalman gain, is the current observation value, is the observation matrix.

[0072] S212: Normalization processing.

[0073] In this embodiment, the data is normalized to satisfy the following relationship:

[0074]

[0075] in, is the normalized data, is the original data, is the minimum value of the original data, is the maximum value of the original data.

[0076] S213. Time alignment.

[0077] Specifically, in this embodiment, the multi-dimensional real-time operation data of the scraper is time-aligned by linear interpolation to satisfy the following relationship:

[0078]

[0079] in, For in time The data value after top alignment, For at time point The original data value on is the current time point, is the time point of known data, For immediately following The next known time after For at time point The original data value on .

[0080] S214. Feature extraction.

[0081] In this embodiment, feature extraction is performed on the processed multi-dimensional real-time operation data to obtain angle features, angular velocity features and acceleration features; the angle features include angle difference and angle change rate, the angular velocity features include average value and standard deviation, and the acceleration features include peak value.

[0082] Specifically, the angle difference satisfies the following relationship:

[0083]

[0084] The angle change speed satisfies the following relationship:

[0085]

[0086] in, is the angle difference, is the maximum angle in the original angle data, is the minimum angle value in the original angle data, is the angle change speed, is the angle change amplitude, is the sampling time interval.

[0087] Furthermore, the average value satisfies the following relationship:

[0088]

[0089] The standard deviation satisfies the following relationship:

[0090]

[0091] in, is the mean angular velocity, is the total amount of angular velocity data collected, is the traversal count flag, is the original angular velocity data value, is the standard deviation of the angular velocity.

[0092] Furthermore, the peak value satisfies the following relationship:

[0093]

[0094] in, is the peak value of acceleration, Indicates taking the maximum value, is the original acceleration data value.

[0095] S22: performing data fusion on the multi-dimensional real-time operation feature data to obtain multi-modal fusion feature information data of the scraper.

[0096] In this embodiment, a dynamic Kalman filter fusion algorithm is used to fuse multi-dimensional real-time operation feature data to satisfy the following relationship:

[0097]

[0098] in, is the comprehensive feature of multimodal fusion. is the traversal count flag, It is the dynamic weighting coefficient of multi-dimensional real-time running characteristic data. Run feature data in real-time for multiple dimensions.

[0099] Specifically, the dynamic weighting coefficient satisfies the following relationship:

[0100]

[0101]

[0102] in, is the dynamic weighting coefficient, is the natural exponential function, is the error between the observed value and the predicted value, is the traversal count flag, is the error value of the overall sample, is the observed value, is the predicted value.

[0103] S3. Obtaining the operation cycle information data of the scraper according to the multimodal fusion feature information data.

[0104] Among them, S3 specifically includes the following steps:

[0105] S31. Obtain a comprehensive feature vector based on the multimodal fusion feature information data, so as to identify a complete action cycle of the scraper.

[0106] In this embodiment, an action cycle recognition module is constructed. The action cycle recognition module combines signal decomposition, dynamic time warping and state transition model to deal with irregular periodic characteristics under complex working conditions. The action cycle recognition module is used to accurately identify the complete action cycle of the shoveling of the scraper based on the comprehensive feature vector.

[0107] Wherein, S31 specifically includes the following steps:

[0108] S311. Obtain a comprehensive feature vector.

[0109] Specifically, a comprehensive feature vector is obtained based on the multimodal fusion feature information data, and the comprehensive feature vector satisfies the following relationship:

[0110]

[0111] in, is the comprehensive feature vector, Indicates the angle change amplitude, Indicates the speed of angle change, is the mean angular velocity, is the standard deviation of angular velocity, is the mean acceleration, is the peak acceleration.

[0112] S312: Define action characteristics for different stages in a complete action cycle.

[0113] In this embodiment, the complete action cycle includes five stages: shoveling, transporting, lifting, unloading and returning.

[0114] Specifically, the action characteristics of the shoveling stage: In the shoveling stage, when the bucket enters the ore, the arm maintains a low angle and the linear acceleration increases. At this time, the angle change of the arm is relatively small, and the acceleration gradually increases, indicating that the bucket contacts the ore and starts to shovel. A threshold is set to determine whether it is in the shoveling stage, satisfying the following relationship:

[0115]

[0116] in, is the change range of the upper arm angle, is the acceleration of the arm along the X direction, is the lower limit amplitude threshold of the upper arm angle, is the threshold of the arm acceleration.

[0117] Specifically, the action characteristics of the transport phase are: the boom gradually rises, keeps the angle stable, the angular velocity is close to zero, and the acceleration decreases. At this time, the boom is in a stable position, mainly driven by the movement of the vehicle, and the changes in angular velocity and acceleration are small, satisfying the following relationship:

[0118]

[0119] in, is the change range of the upper arm angle, is the lower limit amplitude threshold of the upper arm angle, is the upper limit amplitude threshold of the arm angle, is the angular velocity of the arm along the X direction.

[0120] Specifically, the action characteristics of the lifting stage are: the boom is quickly lifted to the highest point, and the angular velocity increases. This stage is the process of the boom rising rapidly during the shovel loading operation. The angle changes greatly, and the angular velocity and acceleration increase sharply, satisfying the following relationship:

[0121]

[0122] in,, is the change range of the upper arm angle, is the upper limit amplitude threshold of the arm angle, is the angular velocity of the upper arm along the X direction, is the lower threshold of angular velocity.

[0123] For further information, see Figure 4 , the figure is a schematic diagram of the lifting stage of the scraper loader, showing the state of the scraper loader's boom lifted to the highest position.

[0124] Specifically, the action characteristics of the unloading stage are: the bucket unloads the ore, the arm remains high and gradually descends. At this time, the angle change of the arm tends to be stable, and the acceleration gradually decreases, satisfying the following relationship:

[0125]

[0126] in, is the change range of the upper arm angle, is the acceleration of the arm along the X direction, is the acceleration threshold during unloading.

[0127] Specifically, the action characteristics of the return phase are: the arm returns to the initial position, the angle returns to the lowest point, the angular velocity and acceleration are close to zero, at this time, the movement of the arm tends to be stable, the dynamic changes in the return process are small, and the following relationship is satisfied:

[0128]

[0129] in, is the change range of the upper arm angle, is the lower limit amplitude threshold of the upper arm angle, is the angular velocity of the upper arm along the X direction, is the acceleration of the arm along the X direction.

[0130] For further information, see Figure 5 ,The figure is a schematic diagram of the return phase, showing the status of the scraper during the return phase.

[0131] S313, identification and statistics of action cycles.

[0132] In this embodiment, the action cycle identification module combines signal decomposition, dynamic time warping and state transition model, and combines the comprehensive feature vector and the action feature definition to perform stage identification on the shoveling process of the scraper.

[0133] Specifically, the state transition model is based on a hidden Markov model and defines the state transition probability to satisfy the following relationship:

[0134]

[0135] in, Indicates that the prerequisite is the current action stage The next action stage is The probability of is the number of phase transitions, For the next action phase, is the action stage at the current moment, is the number of occurrences in the current stage.

[0136] Furthermore, according to the action characteristics of different stages, a complete action cycle is defined, which includes five state sequences of shoveling, transporting, lifting, unloading and returning, satisfying the following relationship:

[0137]

[0138] in, Represents a complete state sequence. If a complete state sequence is detected, the cycle counting function ,otherwise .

[0139] Furthermore, the number of complete cycles during the loading process of the scraper satisfies the following relationship:

[0140]

[0141] in, is the number of complete cycles, is the total number of time steps, Represents a complete sequence of states.

[0142] S32: Perform an abnormality detection on the complete action cycle to obtain operation cycle information data of the scraper.

[0143] In this embodiment, abnormal cycle information is eliminated through abnormal detection to obtain the operation cycle information data of the scraper.

[0144] Wherein, S32 specifically includes the following steps:

[0145] S321. Eliminate abnormal cycles.

[0146] An abnormal cycle is defined as an incomplete action cycle that lacks certain action phases, such as not entering the unloading phase or the return phase; the abnormal cycle detection condition is that the complete state sequence is not met.

[0147] In this embodiment, statistical features are used to perform abnormality detection on the action cycle, the feature mean and standard deviation are calculated, and abnormal values ​​beyond the range are eliminated.

[0148] Specifically, the characteristic mean and the standard deviation satisfy the following relationship:

[0149]

[0150]

[0151] in, is the feature mean, is the total number of action cycles, For the The characteristic value of the period, is the standard deviation.

[0152] Furthermore, an abnormal elimination rule is established based on the feature mean and the standard deviation, and the abnormal elimination rule satisfies the following relationship:

[0153]

[0154] in, For the The characteristic value of the period, is the feature mean, is the standard deviation.

[0155] In an optional embodiment, machine learning is used to classify the cycles using a supervised learning model (such as a random forest) to satisfy the following relationship:

[0156]

[0157] in, Indicated in The probability of abnormal cycle occurring under the condition, is the characteristic vector of the current period, Express the probability of The largest , Indicates that the condition The following categories are The probability of is a normal category or an abnormal category.

[0158] Furthermore, the number of valid cycles after eliminating abnormal cycles satisfies the following relationship:

[0159]

[0160] in, is the effective cycle number, is the total number of cycles, is the number of abnormal cycles.

[0161] In this embodiment, by combining feature threshold detection, hidden Markov model and statistical analysis, the complete cycle of the scraper is accurately identified and abnormal cycles are eliminated, which significantly improves the reliability of cycle statistics and provides an accurate data basis for subsequent efficiency analysis. It should be noted that the features can be automatically identified or manually set.

[0162] The method uses data-driven automatic identification of thresholds. It has strong adaptability and can automatically identify thresholds through machine learning and data analysis. It can adapt to different working environments and changes in working conditions. The thresholds can be dynamically adjusted according to patterns in the data. At the same time, the automatic identification of thresholds reduces the need for human intervention and can be adjusted according to real-time data to provide higher accuracy.

[0163] S322: Identify non-working time periods.

[0164] Specifically, when the arm of the scraper remains stationary or swings in a non-working position, the equipment runs without load. The characteristic performance satisfies the following relationship:

[0165]

[0166] in, is the change range of the upper arm angle, is the minimum threshold of the arm angle change, is the angular velocity of the upper arm along the X direction, is the acceleration of the arm along the X direction.

[0167] Furthermore, data noise or mutation is manifested as abnormal peaks or mutations in 3D sensor data, and the characteristic performance satisfies the following relationship:

[0168]

[0169] in, is the acceleration of the arm along the X direction, is the maximum threshold of acceleration noise, is the angular velocity of the upper arm along the X direction, is the maximum threshold of angular velocity noise.

[0170] Through the above feature definition and threshold setting, the five key stages of the shoveling operation can be accurately identified: shoveling, transportation, lifting, unloading and returning. In the cycle identification module, the combination of multimodal data (such as the angle, angular velocity and acceleration of the boom) and dynamic time warping and state transition models can cope with irregular cycle changes in complex environments; in addition, the use of anomaly detection algorithms further eliminates incomplete cycles, non-working time and noise data to ensure the accuracy and reliability of cycle statistics; this method can provide high-precision cycle identification and operation efficiency analysis, providing strong data support for the optimization of mining operations.

[0171] S4. Constructing a statistical optimization model of the loading efficiency of the scraper loader in combination with the operation cycle information data.

[0172] Wherein, S4 specifically includes the following steps:

[0173] S41. Statistically analyze the operation cycle information data to obtain the number of shoveling times and the operating efficiency of the scraper, and use the number of shoveling times and the operating efficiency as output targets.

[0174] Specifically, the number of shoveling times is the statistical result of the effective cycle excluding abnormal action cycles; the operation efficiency is the shoveling efficiency per unit time within the effective working time, and the operation efficiency satisfies the following relationship:

[0175]

[0176] in, For operational efficiency, is the effective cycle number, Effective working time.

[0177] S42: Taking the comprehensive feature vector as an input feature, constructing a mapping function between the input feature and the output target as a statistical model of the loading efficiency of the scraper.

[0178] In this embodiment, the mapping function is used as a statistical model of shoveling efficiency, satisfying the following relationship:

[0179]

[0180] in, is the model output, represents the mapping function, is the comprehensive feature vector, is the model parameter.

[0181] In an optional embodiment, a regression model is constructed to perform comparative analysis with the shoveling efficiency statistical model; the regression model includes a linear regression model, a random forest regression model, and a gradient boosting regression tree model.

[0182] Specifically, the linear regression model is used to evaluate the linear relationship between the input features and the output target, satisfying the following relationship:

[0183]

[0184] in, is the model output, is the intercept, is the regression coefficient, is the input feature.

[0185] Specifically, the random forest regression model is used to capture the nonlinear relationship between features, is robust to high-dimensional features and noisy data, and satisfies the following relationship:

[0186]

[0187] in, is the model output, is the total number of decision trees, It is The output of the tree.

[0188] Specifically, the gradient boosting regression tree model works well in small sample data, and the model accuracy is optimized through weighted iteration to meet the following relationship:

[0189]

[0190] in, Indicated in After rounds of iterations, the model is input The predicted value of Indicated in After rounds of iterations, the model is input The predicted value of is the learning rate, For the The residuals of the predictions of the trees.

[0191] S43, training and optimizing the shoveling efficiency statistical model to obtain a shoveling efficiency statistical optimization model of the scraper.

[0192] In this embodiment, the original data set is divided into a training set, a validation set and a test set. The training set is used for model training, the validation set is used for model hyperparameter optimization, and the test set is used for model performance evaluation. The training set, validation set and test set satisfy the following relationship:

[0193]

[0194] in, is the original data set, is the training set, is the validation set, For the test set.

[0195] Wherein, S43 specifically includes the following steps:

[0196] S431. Model training optimization.

[0197] In this embodiment, a loss function is constructed using random forests, and the model is recursively optimized based on the training set combined with the residual. The feature weights are dynamically adjusted by calculating the split contributions of different features. The recursive optimization satisfies the following relationship:

[0198]

[0199] in, For the The residual at the iteration, For the The true value of the samples, For the front The predicted value for the iteration.

[0200] S432. Model hyperparameter optimization.

[0201] In this embodiment, grid search and Bayesian optimization are used to adjust model hyperparameters, which include random forest parameters (such as the number of trees and the maximum depth) and GBDT parameters (learning rate and number of sub-models). The model hyperparameter optimization satisfies the following relationship:

[0202]

[0203] in, is the optimal hyperparameter set, It means to find the set of hyperparameters that minimize the loss function. are model parameters, is the loss function value on the validation set.

[0204] S433. Model performance evaluation.

[0205] In this embodiment, the model performance is evaluated using the test set, including cross-validation, mean square error, mean absolute error, and coefficient of determination.

[0206] Specifically, the cross validation adopts -Fold cross validation evaluates model performance and satisfies the following relationship:

[0207]

[0208] in, As the evaluation index, For the fold, For the The folded loss function value.

[0209] Specifically, the mean square error satisfies the following relationship:

[0210]

[0211] in, is the mean square error, is the total sample size, For the The true value of samples; For the The predicted value of a sample.

[0212] Specifically, the mean absolute error satisfies the following relationship:

[0213]

[0214] in, is the mean square error, is the total sample size, For the The true value of samples; For the The predicted value of a sample.

[0215] Specifically, the determination coefficient satisfies the following relationship:

[0216]

[0217] in, is the coefficient of determination, For the The true value of the samples, For the The predicted value of samples, is the mean of the true values ​​of the samples.

[0218] S5. Obtain the shoveling efficiency statistical result of the scraper according to the shoveling efficiency statistical optimization model, and complete the statistics of the shoveling efficiency of the mine blast pile.

[0219] Specifically, according to the shoveling efficiency statistical optimization model, a fitting analysis is performed on the mine blast pile shoveling process of the shovel loader to obtain the shoveling efficiency statistical results.

[0220] In this embodiment, the shoveling efficiency statistics are obtained by conducting an in-depth analysis of the action cycle and operating efficiency of the shovel loader, and the shoveling efficiency statistics are visualized; the shoveling efficiency statistics include a histogram of the number of shoveling times, a distribution curve of operating efficiency, a stacked bar chart of the action stage time, a pie chart of the action stage time proportion, a time series chart, a feature correlation heat map, a paired scatter matrix chart, and a feature importance ranking chart.

[0221] S51, histogram of shoveling times.

[0222] See also Figure 6 , the figure is a histogram of the shoveling times of the scraper; in order to count the distribution of the shoveling times per hour and intuitively understand the operating frequency of the scraper, the shoveling times data is divided into several intervals, and the shoveling times distribution is plotted. The figure shows the shoveling times per hour within 12 hours, and the horizontal axis time can be adjusted according to specific needs.

[0223] Specifically, the shoveling times distribution satisfies the following relationship:

[0224]

[0225] in, is the number of shovel loadings, is the total number of times, is the number of intervals.

[0226] S52, operation efficiency distribution curve.

[0227] See also Figure 7 , the figure shows the distribution curve of the operation efficiency of the scraper; in order to draw the distribution trend of the operation efficiency and show the proportion of efficient and inefficient working conditions, the operation efficiency expression calculated based on the effective working time is used, and the kernel density estimation is used to generate a probability density curve as the operation efficiency distribution curve.

[0228] Specifically, the kernel density estimation satisfies the following relationship:

[0229]

[0230] in, Indicates that at the data point The estimated value of the probability density function at , is the sample size, is the bandwidth, is the kernel function, represents a data point, represents the sample data points, Indicates that each sample data point Through bandwidth Scaled and relative to the data point To standardize.

[0231] S53, action phase time stacked bar chart.

[0232] See also Figure 8 , the figure is a stacked bar chart of the action phase time of the scraper; to analyze the distribution of each action phase of the scraper (loading, transporting, lifting, unloading, returning) in the total cycle time. The cumulative time of each action phase is counted to meet the following relationship:

[0233]

[0234] in, For the The total time of the phase, is the effective cycle number, For the In the cycle Stage time.

[0235] S54, pie chart of action phase time percentage.

[0236] See also Fig. 9 , the figure is a pie chart of the time proportion of the action stage of the scraper; the time proportion of each action stage is calculated by the cumulative time and total time of each action stage; the time proportion satisfies the following relationship:

[0237]

[0238] in, For the The time proportion of the phase, For the The total time of the phase, The total time of a complete action cycle.

[0239] S55, time series graph.

[0240] See also Fig.10 The figure is a time series diagram of a scraper, with time on the horizontal axis and operating efficiency on the vertical axis. It describes the relationship between operating efficiency and time, and explores the existence of trends or periodic changes. The operating efficiency time series satisfies the following relationship:

[0241]

[0242] in, Indicates at time The efficiency of the work within Indicates at time The number of operating cycles within is the total time period.

[0243] S56. Feature correlation heat map.

[0244] See also Fig.11 ,The figure shows the feature correlation heat map of the scraper and the color depth indicates the ,degree of correlation between features. It shows the correlation between ,loading, transportation, lifting, unloading, return, cycle time and ,operation efficiency.

[0245] Specifically, the correlation between features is calculated, and a correlation matrix is ​​constructed as a feature correlation heat map, satisfying the following relationship:

[0246]

[0247] in, Features and The correlation coefficient of is the covariance, and is the standard deviation.

[0248] S57, paired scatter matrix plot.

[0249] See also Fig.12 ,The figure shows a pairwise scatter matrix diagram of a scraper, showing the relationship between the scraper cycle time, features, and operating efficiency; it shows the relationship between multiple features and operating efficiency. Especially when there are many features, a pairwise scatter matrix diagram is used to show the pairwise relationship between each feature, as well as the correlation between features and efficiency.

[0250] S58, feature importance ranking diagram.

[0251] See also Fig.13 ,The figure shows the feature importance ranking diagram of the scraper, which shows the importance of features to the prediction of work efficiency, which is crucial for feature selection and model optimization. The features include loading, transporting, lifting, unloading, returning and cycle time.

[0252] See also Fig.14 In an optional embodiment, in order to efficiently execute the mine blast pile shoveling efficiency statistics method based on multidimensional data provided by the present invention, the present invention provides a mine blast pile shoveling efficiency statistics system based on multidimensional data, the system includes an input device, an output device, a processor and a memory, the hardware facilities are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the steps of the specific embodiment of the method of the present invention. The mine blast pile shoveling efficiency statistics system based on multidimensional data provided by the present invention has a complete structure, objective and stable, and improves the overall applicability and practical application ability of the method of the present invention.

[0253] In summary, the method of the present invention provides a method and system for statistically analyzing the efficiency of mine blast pile shoveling based on multidimensional data. By collecting multidimensional operation data of the shovel loader, integrating multimodal feature information, analyzing the operation cycle of the shovel loader, and constructing a shoveling efficiency optimization model, the accuracy of efficiency statistics is improved, and a reference basis is provided for optimizing operation efficiency, promoting intelligent management of mine shoveling operations. The method of the present invention is easy to understand, simple to calculate, with a small workload, and convenient for engineering application, providing a theoretical basis and technical support for the further development of the field of mine mining efficiency analysis.

[0254] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A statistical method for mining pile shoveling efficiency based on multidimensional data, characterized in that: The steps include: Obtain multi-dimensional real-time operation data of the loader during the mine blast pile loading process; Obtaining multimodal fusion feature information data of the scraper based on the multidimensional real-time operation data; Obtaining operation cycle information data of the scraper according to the multimodal fusion feature information data; Building a statistical optimization model of the shoveling efficiency of the scraper in combination with the operation cycle information data; The shoveling efficiency statistical results of the scraper are obtained based on the shoveling efficiency statistical optimization model, and the statistics of the shoveling efficiency of the mine blast pile are completed.

2. The method for calculating the efficiency of mine blasting pile shoveling based on multidimensional data according to claim 1 is characterized in that: The multi-dimensional real-time operation data of the scraper during the mine blast pile shoveling process is obtained, including: A three-dimensional sensor is installed on the scraper, and a data sampling frequency of the three-dimensional sensor is set; Based on the data sampling frequency, the three-dimensional sensor is used to collect multi-dimensional real-time operation data of the shovel loader during the mine blast pile shoveling process, and the multi-dimensional real-time operation data includes angle, angular velocity and acceleration.

3. The method for calculating the efficiency of mine blasting pile shoveling based on multidimensional data according to claim 1 is characterized in that: The method of obtaining the multimodal fusion feature information data of the scraper based on the multidimensional real-time operation data includes: Preprocessing the multi-dimensional real-time operation data to obtain multi-dimensional real-time operation characteristic data of the scraper; The multi-dimensional real-time operation characteristic data is subjected to data fusion to obtain multi-modal fusion characteristic information data of the scraper.

4. The method for calculating the efficiency of mine blast pile shoveling based on multidimensional data according to claim 3 is characterized in that: The step of fusing the multi-dimensional real-time operation feature data to obtain multi-modal fusion feature information data of the scraper includes: ； in, is the comprehensive feature of multimodal fusion. is the traversal count flag, It is the dynamic weighting coefficient of multi-dimensional real-time running characteristic data. Run feature data in real-time for multiple dimensions.

5. The method for calculating the efficiency of mine blast pile shoveling based on multidimensional data according to claim 1 is characterized in that: The step of obtaining the operation cycle information data of the scraper according to the multimodal fusion feature information data includes: Obtaining a comprehensive feature vector based on the multimodal fusion feature information data, thereby identifying a complete action cycle of the scraper; Anomaly detection is performed on the complete action cycle to obtain operation cycle information data of the scraper.

6. The method for calculating the efficiency of mine blast pile shoveling based on multidimensional data according to claim 5 is characterized in that: The step of obtaining a comprehensive feature vector based on the multimodal fusion feature information data, thereby identifying a complete action cycle of the scraper, includes: ； in, is the comprehensive feature vector, Indicates the angle change amplitude, Indicates the speed of angle change, is the mean angular velocity, is the standard deviation of angular velocity, is the mean acceleration, is the peak acceleration.

7. The method for calculating the efficiency of mine blast pile shoveling based on multidimensional data according to claim 5 is characterized in that: The step of constructing a loading efficiency statistical optimization model of the scraper in combination with the operation cycle information data includes: Performing statistical analysis on the operation cycle information data to obtain the number of shoveling times and the operating efficiency of the scraper, and the number of shoveling times and the operating efficiency are used as output targets; Taking the comprehensive feature vector as an input feature, constructing a mapping function between the input feature and the output target as a loading efficiency statistical model of the scraper; The loading efficiency statistical model is trained and optimized to obtain the loading efficiency statistical optimization model of the scraper.

8. The method for calculating the efficiency of mine blast pile shoveling based on multidimensional data according to claim 7 is characterized in that: The method of taking the comprehensive feature vector as an input feature and constructing a mapping function between the input feature and the output target as a statistical model of the loading efficiency of the scraper includes: ； in, is the model output, represents the mapping function, is the comprehensive feature vector, is the model parameter.

9. The method for calculating mining pile shoveling efficiency based on multidimensional data according to claim 1 is characterized in that: The method of obtaining the shoveling efficiency statistical result of the scraper according to the shoveling efficiency statistical optimization model and completing the statistics of the shoveling efficiency of the mine blast pile includes: The loading efficiency statistical optimization model is used to fit the mine blast pile loading process of the scraper to obtain the loading efficiency statistical results.

10. A mine blast pile shoveling efficiency statistics system based on multi-dimensional data, characterized in that: The system includes an input device, an output device, a processor and a memory, wherein the input device, the output device, the processor and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the mine blasting pile shoveling efficiency statistics method based on multidimensional data as described in any one of claims 1 to 9.

Images