Gold mine filling automatic control method based on Internet of Things

Abstract

The invention relates to the technical field of gold mine filling automatic control, in particular to a gold mine filling automatic control method based on the Internet of Things. The method comprises the steps that original multi-modal data are collected and preprocessed, the preprocessed multi-modal data are subjected to tensor nonlinear transformation, and disturbance characteristic quantities are obtained; constructing a modal input vector based on the disturbance characteristic quantity, and processing the modal input vector to obtain a hidden layer embedding vector; based on the implicit layer embedding vector, carrying out implicit variable path evolution processing to obtain inter-modal path tension; constructing a control instruction based on the inter-modal path tension; after the control instruction is sent to the control equipment, original multi-modal data are collected in real time and predicted; and calculating a composite quantitative feedback error based on the multi-modal data predicted value, and updating and optimizing the control instruction. The problems that the filling quality control precision is low, the filling efficiency is not high, the manual dependence degree is large, filling parameter adjustment lags behind, and the sudden disturbance response capacity is poor in the current gold mine filling operation are solved.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for gold mine filling, and in particular to an automatic control method for gold mine filling based on the Internet of Things. Background Technology

[0002] In current gold mining operations, due to the complex structure of underground goaf areas, variable geological conditions, and limited working environments, backfilling operations have become a crucial link in ensuring underground safety and achieving efficient resource recovery. However, existing backfilling systems generally rely on manual experience and fixed logical rules for operation and control, making it difficult to adapt to nonlinear changes and real-time disturbances during the backfilling process. This results in difficulties in accurately coordinating key parameters such as backfill slurry concentration, flow rate, and pressure, leading to sluggish system adjustment response, unstable compaction and strength of the backfill, and potential safety hazards such as local collapse or insufficient backfilling. Furthermore, existing backfilling systems have limited monitoring methods for the backfilling process, lacking dynamic sensing and data fusion capabilities for key areas, and failing to promptly reflect the impact of external factors such as underground disturbances and water pressure fluctuations on the backfilling effect. In addition, the low frequency of control strategy updates and rigid parameter adjustment mechanisms make it difficult to meet the technical requirements of high precision, low energy consumption, adaptability, and unmanned operation for backfilling operations in the construction of green and intelligent mines.

[0003] In summary, current gold mine backfilling operations still face technical challenges such as low precision in backfilling quality control, low backfilling efficiency, high reliance on manual labor, delayed adjustment of backfilling parameters, and poor response to sudden disturbances. Summary of the Invention

[0004] This invention provides an automated control method for gold mine backfilling based on the Internet of Things, in order to solve the technical problems existing in current gold mine backfilling operations, such as low backfilling quality control accuracy, low backfilling efficiency, high dependence on manual labor, lag in backfilling parameter adjustment, and poor response to sudden disturbances.

[0005] The present invention provides an automated control method for gold mine filling based on the Internet of Things, which specifically includes the following technical solutions:

[0006] An automated control method for gold mine filling based on the Internet of Things includes the following steps:

[0007] S1. Collect raw multimodal data and preprocess it to obtain preprocessed multimodal data; perform tensor nonlinear transformation on the preprocessed multimodal data to obtain perturbation features; construct modal input vectors based on perturbation features and process them to obtain hidden layer embedding vectors; perform hidden path evolution processing based on hidden layer embedding vectors to obtain intermodal path tension.

[0008] S2. Based on the intermodal path tension, generate control parameters and construct control commands; after sending the control commands to the control device, collect raw multimodal data in real time and make predictions to obtain multimodal data prediction values; based on the multimodal data prediction values, calculate the composite quantization feedback error and update and optimize the control parameters.

[0009] Preferably, S1 specifically includes:

[0010] Based on the preprocessed multimodal data, a perturbation amplification term, a frequency modulation tangent term, and an exponential decay term are constructed, and a perturbation feature quantity is constructed by combining them with a logarithmic function. All perturbation feature quantities are combined to generate a modal input vector, and the modal input vector is processed by a nonlinear compression network to obtain the hidden layer embedding vector.

[0011] Preferably, S1 specifically includes:

[0012] Based on the hidden layer embedding vector, the energy distance and modal coupling strength of modal differences are quantified, and construction parameters are introduced. Combined with integral operations, the path tension between modes is calculated.

[0013] Preferably, S2 specifically includes:

[0014] Based on the intermodal path tension, a coupling driving contribution term and an exponential suppression function term of the mode to the control parameters are constructed, and combined with the time oscillation term to generate control parameters; based on the control parameters, control commands are constructed.

[0015] Preferably, S2 specifically includes:

[0016] The control parameters in the control command are transmitted to the control device, and the raw multimodal data is collected in real time. After the raw multimodal data is preprocessed as described in step S1, it is combined with historical monitoring data to make predictions and obtain the predicted values ​​of the multimodal data.

[0017] Preferably, S2 specifically includes:

[0018] Based on the control parameters, a penalty term for the rate of change of the control parameters is constructed, and the composite quantization feedback error is calculated by combining the predicted values ​​of multimodal data.

[0019] Preferably, S2 specifically includes:

[0020] Based on the composite quantization feedback error, the construction parameters of the intermodal path tension are updated through gradient and optimization control, and the evolution path is optimized.

[0021] Preferably, S2 specifically includes:

[0022] Calculate the difference between the composite quantization feedback error at the current moment and the previous moment. When the absolute value of the difference between the composite quantization feedback errors is less than or equal to a preset error difference threshold, and the composite quantization feedback error at the current moment is less than or equal to a preset absolute error threshold, the control parameters at the current moment are used as the final control parameters. The final control parameters are restored to the actual control range of the control equipment through a linear mapping method and sent to the control equipment for automated control of gold mine filling.

[0023] The beneficial effects of the technical solution of the present invention are:

[0024] 1. By performing tensor-based nonlinear transformation on the original multimodal data (such as slurry flow rate, slurry concentration, pipeline pressure, and conveying velocity), key features under complex coupled conditions such as geological disturbance and mechanical oscillation can be effectively extracted, so that steady-state identification capability can still be maintained under extreme environments. By constructing intermodal path tension and combining it with dynamic feedback optimization, the filling state changes can be adaptively identified and control parameters can be finely adjusted, thereby achieving uniform control of the filling strength and significantly improving the stability and reliability of the filling quality.

[0025] 2. This invention is based on hidden layer embedded vectors to perform implicit path evolution processing, construct intermodal path tension, and deeply integrate control logic with perception data. It no longer relies on fixed rules or experience models, but automatically optimizes control parameters based on composite quantized feedback error through self-learning and self-evolution mechanisms. This avoids rigid logic set by humans, improves the level of intelligence, and reduces reliance on manual adjustment. Especially in gold mine filling scenarios with strong nonlinearity and high time-varying characteristics, it can dynamically adjust the control strategy according to feedback data to achieve an automated control closed loop with self-optimization capabilities. Attached Figure Description

[0026] Figure 1 This is a flowchart of an automated control method for gold mine filling based on the Internet of Things, as described in this invention. Detailed Implementation

[0027] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0029] The following description, in conjunction with the accompanying drawings, details a specific scheme for an automated control method for gold mine filling based on the Internet of Things (IoT) provided by this invention.

[0030] See attached document Figure 1 The diagram illustrates a flowchart of an IoT-based automated control method for gold mine filling, according to an embodiment of the present invention. The method includes the following steps:

[0031] S1. Collect raw multimodal data and preprocess it to obtain preprocessed multimodal data; perform tensor nonlinear transformation on the preprocessed multimodal data to obtain perturbation features; construct modal input vectors based on perturbation features and process them to obtain hidden layer embedding vectors; perform hidden path evolution processing based on hidden layer embedding vectors to obtain intermodal path tension.

[0032] Raw multimodal data is collected by deploying sensing devices at key nodes in the filling system, such as pump stations, mixing tanks, delivery pipelines, and wellheads. This raw multimodal data includes slurry flow rate, slurry concentration, pipeline pressure, and delivery velocity. The sensing devices include concentration sensors, temperature sensors, pressure sensors, flow meters, vibration monitoring devices, liquid level monitoring devices, and stress monitoring devices. The raw multimodal data undergoes preprocessing such as denoising, standardization, and normalization to obtain preprocessed multimodal data. All preprocessing processes employ techniques well-known to those skilled in the art and will not be elaborated upon here.

[0033] Furthermore, to uniformly model the preprocessed multimodal data and enhance its frequency domain perturbation characteristics and physical representation capabilities, a tensor nonlinear transformation is introduced. This transforms the preprocessed multimodal data into high-dimensional input variables, i.e., perturbation features, suitable for subsequent path evolution calculations. The specific implementation formula is as follows:

[0034]

[0035] in, It is the first The modality at time... The enhanced features of the perturbation, namely the perturbation feature quantity, are used for subsequent path evolution calculations; It is the first The modality at time... The preprocessed modal data; It is a very small constant, and can take values ​​of ; It is the first The frequency perturbation factors for each mode are used to excite the periodic behavior of the tangent function. These factors are determined through spectral analysis, with a reference range of values. The spectrum analysis method is a well-known technique in the art and will not be described in detail here. This is the time decay weight, used to control the exponential decay of the mode over time. It is determined through adaptive feature normalization, and the reference value range is [value range missing]. The adaptive feature normalization method is a well-known technique in the art and will not be described in detail here. It is a perturbation amplification term, used to enhance the amplitude sensitivity of preprocessed multimodal data to ensure that small perturbations are not ignored; It is the frequency modulation tangent, which represents the periodic burst behavior of the preprocessed multimodal data under high-frequency disturbances, and is used to simulate periodic disturbances such as mine pressure impact, slurry oscillation, and pump vibration. It is an exponential decay term, which means that as the value of the preprocessed multimodal data increases, its interference or control value decreases rapidly, and is used to simulate the phenomenon of reduced control sensitivity. It can reflect the dual disturbance effect of "intense oscillation + steady-state mitigation"; The perturbation amplification term and the period decay factor are described. The nonlinear comparison is used to compress the range of extreme variations by taking the detailed differences in the logarithmic stretching ratio.

[0036] The above-mentioned tensor-quantized nonlinear transformation can highlight the nonlinear propagation characteristics of filling process parameters under the influence of sudden factors (such as mine tremors, water pressure fluctuations, etc.) in complex underground environments while preserving the modal amplitude characteristics and disturbance frequency characteristics.

[0037] After the above tensor-quantized nonlinear transformation, all perturbation features are combined to form the modal input vector: ,in, Indicates at time modal input vector, This represents the total number of modes; subsequently, the modal input vectors are processed through an existing nonlinear compression network to obtain the hidden state vectors in higher-order representation. , is used to represent the implicit representation of modal input vectors in high-dimensional semantic space, namely the hidden layer embedding vector. The physical meaning of the hidden layer embedding vector is to compress redundant information and highlight the coupling behavior between dominant modes while preserving the correlation between modal input vectors.

[0038] Furthermore, based on the hidden layer embedding vectors, implicit path evolution processing is performed. That is, instead of explicitly modeling the mapping relationship between modal input vectors, the spatiotemporal dynamics of modal cooperation are implicitly described by calculating the "evolutionary path tension" between modal input vectors. The path tension between any two modes is defined as:

[0039]

[0040] in, It is the intermodal path tension, representing the first... The first mode and the first The modalities at time... Path tension; , These are the th hidden layer embedding vectors. The modality, the first The latent representation of each modality, derived from the hidden layer embedding vector. Obtain from; It is the squared term of the state difference between modes, representing the energy distance of the modal difference. The larger the value, the stronger the difference in modal behavior. It is a nonlinear disturbance factor used to control the weight of sinusoidal disturbances in the path, with a reference value range of [value missing]. ; This is the damping coefficient, used to control the weight of the time window. A larger damping coefficient indicates a smaller influence from recent states. The reference range is... ; This is the coupling balance factor, used to control the amplification effect of modal coupling strength in the denominator. The reference value range is... ; , , The construction parameters that constitute the intermodal path tension are all updated based on feedback, which will be explained later; It is a time disturbance term, representing periodic energy fluctuations caused by pipeline vibration and pump speed pulsation; Describes the time The actual effective energy tension amplitude between the two modes; It represents the modal coupling strength, used to measure the degree of synchronous change between two modes; It is an energy compression term used to logarithmically suppress high-amplitude coupling and prevent resonance peak distortion; It is a time-domain damping term used to simulate the decay characteristics of energy over time, similar to the exponential response damping effect of a control system. It is the effective coupling tension factor between modes, describing the relationship between two modes at time [time value missing]. The coupled energy response efficiency is measured; finally, integral calculation is introduced to demonstrate the energy interaction and accumulation effect.

[0041] S2. Based on the intermodal path tension, generate control parameters and construct control commands; after sending the control commands to the control device, collect raw multimodal data in real time and make predictions to obtain multimodal data prediction values; based on the multimodal data prediction values, calculate the composite quantization feedback error and update and optimize the control parameters.

[0042] intermodal path tension As a driving factor, control instructions are constructed through tensor mapping. ,in, At any moment The One control parameter, It refers to the number of control parameters, such as... The corresponding control parameters are pump station frequency, mixing ratio, and pipeline control valve opening. This indicates transpose; the formula for generating any one of the control parameters is:

[0043]

[0044] in, At any moment The Each control parameter represents a control parameter of a certain control device; At any moment With the The intermodal path tensions related to each control parameter are selected by professional staff. Is with the first The mapping weights of the intermodal path tensions related to each control parameter to the control result are determined based on expert experience, with a reference range of values. ; Is with the first The oscillation frequency factor between modes related to each control parameter is obtained through frequency analysis, with a reference range of values. The frequency analysis method is a well-known technique in the art and will not be described in detail here. It is the first The baseline offset of each control parameter is set according to the characteristics of the control equipment to which the control parameter belongs, and will not be elaborated here. It is an exponential suppression function term, used to suppress the path tension between modes to avoid output oscillations caused by excessive mode coupling; It is a time oscillation term used to simulate periodic oscillations between modal paths caused by mechanical inertia or hydraulic dynamic response; It is the first The first mode and the first The coupling drive contribution of each mode to the control parameters; finally, the hyperbolic tangent function As an activation function, it is used to summarize all path contributions, perform nonlinear mapping, and output control parameters, which can prevent control commands from jumping or exceeding the device's capacity.

[0045] Furthermore, by directly calling the communication interface of the hardware controller (such as Modbus or CANopen), the control parameters in the control command are transmitted to the control device to control the action of the field equipment and to collect monitoring data in real time, i.e., raw multimodal data, such as slurry flow rate, slurry concentration, pipeline pressure, and conveying velocity. After preprocessing the raw multimodal data as described in step S1, it is combined with historical monitoring data extracted from the existing monitoring database and predicted using an existing prediction model such as Long Short-Term Memory (LSTM) network to obtain the predicted value of the multimodal data. Based on the predicted value of the multimodal data and combined with the rate of change penalty term of the control parameters, the composite quantized feedback error is calculated. The specific implementation formula is as follows:

[0046]

[0047] in, It is the current moment. The composite quantization feedback error; At any moment No. Preprocessed multimodal data for each modality; At any moment No. Multimodal data prediction values ​​for each modality; It is a very small constant used to prevent division by zero and to maintain stable numerical calculations; it can take values ​​of... ; At any moment The One control parameter; This is a penalty factor for controlling changes, determined based on expert experience, with a reference range of values. ; It is the relative sum of squared prediction errors, used to measure the deviation between the current multimodal data predictions and the actual monitoring data; It is a penalty term for the rate of change of control parameters, used to constrain drastic changes in the control signal at consecutive moments;

[0048] Based on the aforementioned composite quantization feedback error, the construction parameters for intermodal path tension are determined through gradient and optimization control (chain rule). (including nonlinear disturbance factors) Damping coefficient and coupling balance factor The parameters for constructing intermodal path tension are updated to continuously approach the optimal control strategy; the update formula for these parameters is:

[0049]

[0050] in, It is the current moment. The construction parameters for intermodal path tension; At any moment The construction parameters for intermodal path tension; The step size factor controls the speed or magnitude of parameter updates for intermodal path tension. It is determined using the adaptive gradient method, with a reference value range of [value missing]. The adaptive gradient method is a well-known technique in the art and will not be described in detail here. It is the partial derivative of the composite quantization feedback error with respect to the construction parameters of the intermodal path tension. It is obtained by automatic differentiation mechanism (such as TensorFlow, PyTorch) or by analytical formula differentiation. It is a well-known technique in the art and will not be described in detail here.

[0051] Finally, calculate the composite quantization feedback error at the current moment. Composite quantization feedback error with the previous moment The difference is calculated, and the absolute value of the difference is compared with the preset error difference threshold and absolute error threshold. If both conditions are met: and If the control parameters at this point are used as the final control parameters, then the control parameters at this point will be taken as the final control parameters; where, This is the error difference threshold, and the reference value range is: , This is the absolute error threshold, and the reference range is: The error difference threshold and absolute error threshold are determined by existing statistical analysis methods. By collecting the composite quantitative feedback error under each control cycle in the historical operation cycle and establishing the composite quantitative feedback error sequence, the error range under the steady state is obtained by performing mathematical statistical analysis on the composite quantitative feedback error sequence. The final control parameters are restored to the actual control range of the control equipment by existing linear mapping methods. For example, the interval [−1,1] is mapped to the pump frequency [30Hz,60Hz] or the valve angle [10°,90°] and sent to each control equipment to realize the automated control of gold mine filling.

[0052] In summary, an automated control method for gold mine filling based on the Internet of Things has been developed.

[0053] The order of the embodiments is for illustrative purposes only and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0054] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0055] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An automated control method for gold mine filling based on the Internet of Things, characterized in that, Includes the following steps: S1. Collect raw multimodal data and preprocess it to obtain preprocessed multimodal data; Tensor-quantized nonlinear transformation is performed on the preprocessed multimodal data to obtain perturbation features; Based on the perturbation features, a modal input vector is constructed and processed to obtain the hidden layer embedding vector; Based on the hidden layer embedding vector, implicit path evolution processing is performed to obtain the intermodal path tension; S2. Based on the intermodal path tension, generate control parameters and construct control commands; after sending the control commands to the control device, collect raw multimodal data in real time and make predictions to obtain multimodal data prediction values; based on the multimodal data prediction values, calculate the composite quantization feedback error and update and optimize the control parameters.

2. The automated control method for gold mine filling based on the Internet of Things according to claim 1, characterized in that, S1 specifically includes: Based on the preprocessed multimodal data, a perturbation amplification term, a frequency modulation tangent term, and an exponential decay term are constructed, and a perturbation feature quantity is constructed by combining them with a logarithmic function. All perturbation feature quantities are combined to generate a modal input vector, and the modal input vector is processed by a nonlinear compression network to obtain the hidden layer embedding vector.

3. The automated control method for gold mine filling based on the Internet of Things according to claim 2, characterized in that, S1 specifically includes: Based on the hidden layer embedding vector, the energy distance and modal coupling strength of modal differences are quantified, and construction parameters are introduced. Combined with integral operations, the path tension between modes is calculated.

4. The automated control method for gold mine filling based on the Internet of Things according to claim 1, characterized in that, S2 specifically includes: Based on the intermodal path tension, a coupling driving contribution term and an exponential suppression function term of the mode to the control parameters are constructed, and combined with the time oscillation term to generate control parameters; based on the control parameters, control commands are constructed.

5. The automated control method for gold mine filling based on the Internet of Things according to claim 4, characterized in that, S2 specifically includes: The control parameters in the control command are transmitted to the control device, and the raw multimodal data is collected in real time. After the raw multimodal data is preprocessed as described in step S1, it is combined with historical monitoring data to make predictions and obtain the predicted values ​​of the multimodal data.

6. The automated control method for gold mine filling based on the Internet of Things according to claim 5, characterized in that, S2 specifically includes: Based on the control parameters, a penalty term for the rate of change of the control parameters is constructed, and the composite quantization feedback error is calculated by combining the predicted values ​​of multimodal data.

7. The automated control method for gold mine filling based on the Internet of Things according to claim 6, characterized in that, S2 specifically includes: Based on the composite quantization feedback error, the construction parameters of the intermodal path tension are updated through gradient and optimization control, and the evolution path is optimized.

8. The automated control method for gold mine filling based on the Internet of Things according to claim 7, characterized in that, S2 specifically includes: Calculate the difference between the composite quantization feedback error at the current moment and the previous moment. When the absolute value of the difference between the composite quantization feedback errors is less than or equal to a preset error difference threshold, and the composite quantization feedback error at the current moment is less than or equal to a preset absolute error threshold, the control parameters at the current moment are used as the final control parameters. The final control parameters are restored to the actual control range of the control equipment through a linear mapping method and sent to the control equipment for automated control of gold mine filling.

Images