Intelligent sensor-based automated sampling system for sticky gold concentrates

Abstract

This invention relates to the field of viscous ore sampling technology and discloses an automated sampling system for viscous gold concentrate based on intelligent sensing. The system includes a ore flow state sensing module, a ore sample characteristic fusion module, a ore topology mapping module, a ore feature screening module, and a sampling anomaly detection module. The ore flow state sensing module collects dynamic parameters of rheological characteristics through a multi-source sensor array and outputs a slurry rheological parameter matrix through heterogeneous data fusion. The ore sample characteristic fusion module decouples its multi-dimensional features, generates a slurry feature tensor, and constructs a local correlation topology. The ore topology mapping module extracts key ore feature nodes, calculates cross-dimensional correlation strength, and integrates them to form a global rheological feature topology map. The ore feature screening module analyzes the feature vector entropy values ​​and, combined with the cross-dimensional correlation strength, screens out the core entropy sequence of ore features. The sampling anomaly detection module triggers sampling anomaly judgment based on this sequence and generates sampling action control instructions.

Description

Technical Field

[0001] This invention relates to the field of viscous ore sampling technology, specifically to an automated sampling system for viscous gold concentrate based on intelligent sensing. Background Technology

[0002] In the processing of viscous gold concentrate, sampling is a crucial means of obtaining mineral information, and its results directly affect the adjustment of subsequent processes and the control of mineral quality. Traditional sampling methods mostly rely on manual operation or simple mechanical sampling devices, which are difficult to adapt to the complex physical characteristics of viscous gold concentrate.

[0003] Viscous gold concentrate slurry has high viscosity and yield stress, and is prone to uneven velocity distribution and localized accumulation during flow. When sampling manually, operators struggle to accurately assess the homogeneity of the slurry, and the selection of sampling points is often subjective, resulting in samples that fail to accurately reflect the overall composition and characteristics of the slurry. Furthermore, the frequency and timeliness of manual sampling are significantly affected by human factors, making it difficult to meet the demands for real-time mineral information in continuous production processes.

[0004] Existing mechanical sampling devices typically employ fixed sampling cycles and volumes, lacking the ability to dynamically sense the rheological state of the slurry. When parameters such as slurry viscosity and flow rate change, the fixed sampling pattern can easily lead to excessive or insufficient sample volumes, or even clogging of the sampling device by the slurry. Furthermore, these devices cannot perform in-depth processing of the collected slurry information, making it difficult to identify abnormal states in the slurry, thus significantly reducing the reliability of the sampling results.

[0005] As mining production moves towards intelligence and automation, higher demands are being placed on the accuracy and adaptability of sampling systems. Traditional sampling methods are showing increasingly obvious limitations when faced with the complex and variable characteristics of viscous gold concentrate, failing to provide stable and accurate basic data for mineral analysis. Therefore, an automated sampling system capable of dynamically sensing the state of the slurry and intelligently adjusting the sampling strategy is needed. Summary of the Invention

[0006] The purpose of this invention is to provide an automated sampling system for viscous gold concentrate based on intelligent sensing, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides an automated sampling system for viscous gold concentrate based on intelligent sensing, the system comprising:

[0008] The ore flow state sensing module is configured to collect dynamic parameters of rheological properties of viscous gold concentrate slurry in real time through a multi-source sensor array, perform heterogeneous data fusion processing on the dynamic parameters of rheological properties, and output a slurry rheological parameter matrix.

[0009] The mineral sample characteristic fusion module is connected to the mineral flow state sensing module and is configured to perform multi-dimensional feature decoupling on the mineral slurry rheological parameter matrix, generate a mineral slurry feature tensor, and construct a local correlation topology for each rheological dimension based on the mineral slurry feature tensor.

[0010] The mineral topology map module is configured to extract key mineral feature nodes in each rheological dimension, calculate the cross-dimensional correlation strength between the key mineral feature nodes, and integrate all local correlation topologies to form a global rheological feature topology map of viscous gold concentrate.

[0011] The mineral feature screening module is configured to parse the feature vector entropy value of each rheological dimension from the slurry feature tensor, and combine it with the cross-dimensional correlation strength in the global rheological feature topology map to screen out the core entropy sequence of mineral features.

[0012] The sampling anomaly detection module is connected to the mineral feature screening module and is configured to trigger the sampling anomaly determination of viscous gold concentrate based on the core entropy sequence of the mineral features, and generate sampling action control instructions.

[0013] Preferably, the rheological dynamic parameters included in the mineral flow state sensing module specifically cover slurry viscosity gradient data, mineral particle size distribution data, solid-liquid two-phase concentration data, pipeline flow velocity pulsation data, and temperature field distribution data.

[0014] Preferably, the mineral sample characteristic fusion module performs multi-dimensional feature decoupling operations including:

[0015] The slurry rheological parameter matrix is ​​subjected to noise suppression and data normalization to generate a standardized slurry rheological parameter matrix;

[0016] The feature decoupling engine separates the independent feature components of each dimension in the standardized slurry rheological parameter matrix;

[0017] The slurry feature tensor is generated by aggregating the independent feature components of all dimensions.

[0018] Preferably, the process of calculating the cross-dimensional correlation strength by the mineral topology map module includes:

[0019] Select any two key mineral feature nodes and extract the feature response sequences of the key mineral feature nodes in the slurry feature tensor;

[0020] The morphological similarity measure of the feature response sequence is calculated using a dynamic time warping algorithm;

[0021] The morphological similarity metric is mapped to a cross-dimensional association strength weighting coefficient.

[0022] Preferably, the operation of parsing the feature vector entropy value by the mineral feature screening module includes:

[0023] Obtain the dispersion coefficient of mineral distribution for each rheological dimension;

[0024] Calculate the statistical weight distribution of the internal feature components for each dimension of the feature vector;

[0025] The quantization result of the feature vector entropy value is generated based on the discrete coefficient of the mineral distribution and the statistical weight distribution.

[0026] Preferably, the step of the mineral feature screening module in screening the core entropy sequence of mineral features includes:

[0027] Extract the cross-dimensional correlation strength weight coefficients of all key mineral feature nodes in the global rheological feature topology map;

[0028] The quantization result of the feature vector entropy value is weighted and fused with the corresponding cross-dimensional correlation strength weight coefficient;

[0029] Highly significant mineral feature nodes are selected by setting a preset entropy threshold.

[0030] The core entropy sequence of the mineral features is generated by sorting all highly significant mineral feature nodes in descending order of entropy value.

[0031] Preferably, the sampling anomaly detection module further includes:

[0032] The sampling execution feedback unit is configured to collect the motion status parameters of the sampling robotic arm in real time and generate motion status feedback signals.

[0033] The instruction optimization unit, connected to the sampling execution feedback unit, is configured to fuse the core entropy sequence of mineral features with the action state feedback signal to generate a dynamic compensation instruction.

[0034] The instruction fusion unit is configured to superimpose and integrate the sampling action control instruction and the dynamic compensation instruction to output the final sampling control signal.

[0035] Preferably, the action status feedback signal specifically includes displacement trajectory deviation data of the sampling robot arm, hydraulic actuation pressure fluctuation data, sampling container filling rate data, and mechanical vibration spectrum data.

[0036] Preferably, the system further includes:

[0037] An adaptive sampling actuator, connected to the sampling anomaly detection module, is configured to receive the sampling action control command and perform physical sampling operations;

[0038] The adaptive sampling actuator includes a rotary cutting sampling head, a vacuum negative pressure stabilization unit, a multi-stage filtration separation chamber, and a sample packaging robot.

[0039] Preferably, the operating parameters of the rotary cutting sampling head include cutting speed gradient, cutting angle offset, reciprocating motion frequency, and cutting depth increment;

[0040] The parameters that the vacuum negative pressure stabilization unit adjusts in real time include the negative pressure intensity threshold, the airflow pulsation suppression coefficient, and the air pressure balance value of the sample transmission pipeline.

[0041] The separation parameters configured in the multi-stage filtration separation chamber include the screen aperture combination sequence, vibration separation amplitude curve, and mineral residue cleaning cycle.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] The ore flow state sensing module utilizes a multi-source sensor array to collect dynamic parameters of the rheological properties of viscous gold concentrate slurry in real time, and performs heterogeneous data fusion processing to output a slurry rheological parameter matrix. This process can comprehensively capture various physical property changes of the slurry during the flow process, making the system's perception of the slurry state more comprehensive and detailed, thus overcoming the problem of one-sided perception of the slurry state in traditional sampling methods.

[0044] The mineral sample characteristic fusion module decouples the slurry rheological parameter matrix into multi-dimensional features, generates a slurry feature tensor, and constructs the local correlation topology of each rheological dimension. This decomposes complex slurry parameters into feature information of different dimensions, clearly presents the internal correlations of each rheological dimension, and enables the system to understand the characteristics of the slurry from multiple perspectives, providing richer feature evidence for subsequent mineral analysis.

[0045] The mineral topology mapping module extracts key mineral feature nodes, calculates cross-dimensional correlation strength, and integrates them to form a global rheological feature topology map. By constructing a global map, the system can grasp the intrinsic connections between different rheological dimensions, breaking the isolation of information in each dimension and making the understanding of the overall characteristics of the slurry more systematic and comprehensive.

[0046] The mineral characteristic screening module parses the eigenvector entropy values ​​from the slurry characteristic tensor and filters out the core entropy sequence of mineral characteristics by combining cross-dimensional correlation strength. This screening process can eliminate redundant information and focus on the core features that play a key role in the characteristics of the slurry, making the system more targeted in processing information and reducing the interference of invalid data on subsequent analysis.

[0047] The sampling anomaly detection module triggers sampling anomaly judgment and generates control instructions based on the core entropy sequence of mineral characteristics. When an abnormal state occurs in the slurry, the system can promptly identify and adjust the sampling action to match the actual state of the slurry, avoiding adverse effects of abnormal slurry on the sampling results and making the sampling process more closely aligned with the dynamic changes of the slurry. Attached Figure Description

[0048] Figure 1 This is a schematic diagram illustrating the working principle of the automated sampling system for viscous gold concentrate based on intelligent sensing as described in this invention.

[0049] Figure 2 A flowchart for decoupling features from the mineral sample characteristic fusion module;

[0050] Figure 3 A flowchart for calculating the cross-dimensional correlation strength of the mineral topology map module;

[0051] Figure 4 A flowchart for screening core entropy sequences of mineral characteristics;

[0052] Figure 5 A flowchart for optimizing the instructions of the sampling anomaly detection module. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0054] Please see Figure 1 This invention provides an automated sampling system for viscous gold concentrate based on intelligent sensing, the system comprising:

[0055] The mineral flow state sensing module collects dynamic parameters of rheological properties of viscous gold concentrate slurry in real time through a multi-source sensor array, performs heterogeneous data fusion processing on the dynamic parameters of rheological properties, and outputs a mineral slurry rheological parameter matrix.

[0056] The mineral sample characteristic fusion module is connected to the mineral flow state perception module. It performs multi-dimensional feature decoupling on the slurry rheological parameter matrix, generates a slurry feature tensor, and constructs a local correlation topology for each rheological dimension based on the slurry feature tensor.

[0057] The mineral topology map module extracts key mineral feature nodes in each rheological dimension, calculates the cross-dimensional correlation strength between key mineral feature nodes, and integrates all local correlation topologies to form a global rheological feature topology map of viscous gold concentrate.

[0058] The mineral feature screening module parses the feature vector entropy value of each rheological dimension from the slurry feature tensor, and combines it with the cross-dimensional correlation strength in the global rheological feature topology map to screen out the core entropy sequence of mineral features.

[0059] The sampling anomaly detection module is connected to the mineral quality feature screening module. Based on the core entropy sequence of mineral quality features, it triggers the sampling anomaly judgment of viscous gold concentrate and generates sampling action control instructions.

[0060] Example 1: See Figure 2 The rheological dynamic parameters processed by the ore flow state sensing module specifically cover slurry viscosity gradient data, mineral particle size distribution data, solid-liquid two-phase concentration data, pipeline flow velocity pulsation data, and temperature field distribution data. Slurry viscosity gradient data is collected through multiple sets of viscosity sensors arranged radially along the pipeline. Each set of sensors contains three probes at different depths, enabling simultaneous acquisition of real-time viscosity values ​​at different radial locations within the pipeline cross-section. The viscosity gradient change is then calculated using the viscosity difference between adjacent locations. These sensors employ a high-frequency response design, and the sampling interval can be automatically adjusted according to the slurry flow state. The sampling interval is extended when the slurry flow is stable and shortened when flow fluctuations are severe, ensuring the timeliness and completeness of the data.

[0061] Mineral particle size distribution data is collected by a laser particle size analyzer installed beside the pipeline. This device emits a laser beam of a specific wavelength that passes through the slurry flow. Different particle sizes in the slurry scatter the laser beam at different angles. An array of photodetectors captures the intensity and angle distribution of the scattered light. After signal conversion and data processing, the particle size distribution range of the mineral particles and the proportion of particles in each size range are obtained. To avoid the influence of air bubbles in the slurry on the measurement, both the laser transmitter and receiver are equipped with bubble detection and rejection algorithms. When a bubble is detected passing through the measurement area, the data for that time period is automatically marked and corrected in subsequent processing.

[0062] Solid-liquid two-phase concentration data were acquired using microwave transmission measurement. A microwave transmitter and receiver were installed on opposite sides of the pipe. The microwave signal emitted by the transmitter passed through the slurry and was captured by the receiver. Due to the different absorption and reflection characteristics of microwaves by solid particles and liquids in the slurry, the solid-liquid two-phase concentrations in the slurry were calculated by measuring the attenuation and phase change of the microwave signal and combining this with a pre-established concentration calibration model. The microwave frequency of the measuring device can be adjusted according to the viscosity characteristics of the slurry. A lower frequency is used for high-viscosity slurries to enhance signal penetration, while a higher frequency is used for low-viscosity slurries to improve measurement accuracy.

[0063] Pipeline flow velocity pulsation data were acquired using an ultrasonic Doppler velocimeter. This instrument has two pairs of ultrasonic transducers installed on the outer wall of the pipe: one pair emits ultrasonic waves, and the other receives echo signals reflected by particles in the slurry. The Doppler effect is used to calculate the slurry's flow velocity. Simultaneously, the instrument's built-in high-frequency sampling module records velocity changes on a millisecond-level timescale, thus capturing the pulsation characteristics of the flow velocity. To cover the flow velocity in different areas of the pipe, the transducers are installed at multiple angles, monitoring the flow velocity in the central region and near the pipe wall separately. Data fusion is then used to obtain the flow velocity pulsation distribution across the entire pipe.

[0064] Temperature field distribution data is acquired by an infrared thermal imaging sensor installed in the non-contact measurement area of ​​the pipeline. This sensor captures real-time images of the temperature distribution on the pipeline surface, and then uses a temperature field reconstruction algorithm to retrieve the temperature field distribution of the slurry inside the pipeline. The sensor's field of view can be adjusted via an electric adjustment mechanism to ensure complete coverage of the pipeline's cross-section. Equipped with dustproof and cooling devices, it ensures continuous and stable operation in the complex environment of the mine, preventing dust accumulation and high temperatures from affecting measurement accuracy.

[0065] When the mineral sample characteristic fusion module performs multi-dimensional feature decoupling, it first performs noise suppression processing on the slurry rheological parameter matrix. Noise suppression employs a wavelet transform-based denoising method, decomposing the data in the matrix into wavelet coefficients of different frequencies. High-frequency coefficients are thresholded to remove noise, and then the denoised slurry rheological parameter matrix is ​​reconstructed through inverse wavelet transform. Different thresholding strategies are used for different types of parameter data. For flow velocity fluctuation data and temperature field distribution data that are significantly affected by random noise, a stricter hard thresholding process is used. For viscosity gradient data and particle size distribution data that are primarily affected by system noise, a soft thresholding process is used to retain more detailed information.

[0066] Data normalization transforms the data in the noise-suppressed slurry rheological parameter matrix into a unified data range. Specifically, the min-max normalization method is used, which calculates the minimum and maximum values ​​of each parameter to map the original data to the range [0,1], generating a standardized slurry rheological parameter matrix. During normalization, for some parameters that may exhibit extreme outliers, a truncation method is employed, forcibly setting values ​​exceeding the reasonable range to their maximum or minimum values ​​to avoid the impact of outliers on the overall normalization result.

[0067] The feature decoupling engine separates the independent feature components of each dimension in the standardized slurry rheological parameter matrix. This engine is based on an independent component analysis algorithm, which removes second-order correlations between data in each dimension by whitening the standardized matrix. It then iteratively optimizes the matrix to find a separation matrix that maximizes the statistical independence of each component, thus separating the mixed feature components into independent individual feature components. For each separated independent feature component, its effectiveness is verified by calculating its correlation coefficient with the original parameter matrix, retaining components with high correlation, and removing interfering components with low correlation.

[0068] When aggregating independent feature components across all dimensions to generate a slurry feature tensor, the tensor's dimensional structure is first determined, using sampling time, feature type, and spatial location as three basic dimensions. Each independent feature component is then filled into its corresponding position within the tensor according to its time, type, and location information. For parameter data with different sampling frequencies, interpolation or downsampling is used to unify the resolution of the time dimension, ensuring the tensor's consistency across the time dimension. The final generated slurry feature tensor is stored as a three-dimensional array, which intuitively reflects the temporal and spatial relationships of the features across each dimension.

[0069] Example 2: See Figure 3 When calculating the cross-dimensional correlation strength, the mineral topology mapping module first extracts key mineral feature nodes for each rheological dimension from the slurry feature tensor. The extraction of key mineral feature nodes is based on feature importance assessment. By analyzing the response sensitivity of each feature during the slurry rheological process, feature points that significantly reflect changes in slurry properties are selected as nodes. These nodes are distributed across different rheological dimensions, covering multiple aspects such as slurry viscosity, particle size distribution, and solid-liquid concentration. Each node corresponds to the characteristic behavior of the slurry under specific conditions.

[0070] After selecting any two key mineral feature nodes, the feature response sequences of these nodes are extracted from the slurry feature tensor. The extraction of the feature response sequences is based on the time axis, arranging the feature data related to each node in the slurry feature tensor in chronological order to form a continuous sequence. The length of the sequence is determined based on the actual sampling period and data volume to ensure a complete reflection of the node's feature change trend over a period of time. For example, for nodes related to slurry viscosity, the feature response sequence will include viscosity gradient change data at different times, while the sequence for nodes related to particle size distribution consists of particle size distribution parameters at each time point.

[0071] When calculating the morphological similarity metric of feature response sequences using the dynamic time warping algorithm, the two sequences to be compared are first preprocessed. Preprocessing includes removing outliers from the sequences, which may be caused by transient sensor malfunctions or abrupt changes in slurry flow. Data points exceeding the normal fluctuation range are identified and replaced using a sliding window method. Subsequently, a time warping path is constructed for the sequences. This path finds the optimal match between the two sequences through dynamic programming, allowing the sequences to non-linearly scale on the time axis to accommodate delays or advances in the feature responses of different nodes. During the calculation, path constraints are set to limit the slope range of the warped path, avoiding excessively distorted matching results.

[0072] When calculating the morphological similarity metric, the cumulative distance is the sum of the squared differences between corresponding points on the regular path of two sequences. The smaller the cumulative distance, the higher the morphological similarity between the two sequences. After calculating the cumulative distance, it is standardized into a morphological similarity metric. The standardization process is adjusted in conjunction with the sequence length to enable effective similarity comparisons between sequences of different lengths.

[0073] When mapping morphological similarity metrics to cross-dimensional association strength weight coefficients, the first step is to determine the range of values ​​for the mapping, typically set within the [0,1] interval. The mapping rules are formulated based on the specific requirements for association strength in practical applications. When the morphological similarity metric is small, a higher weight coefficient corresponds to a stronger association between the two nodes in their respective dimensions; conversely, when the metric is large, the weight coefficient decreases, reflecting a weaker association. A non-linear transformation function is introduced during the mapping process to assign more sensitive weight changes to regions with high similarity, highlighting the differences between strongly associated nodes, while smoothing out regions with low similarity to avoid drastic fluctuations in the weight coefficients.

[0074] When parsing the entropy value of the feature vector in the mineral quality feature screening module, the dispersion coefficient of mineral quality distribution for each rheological dimension is first obtained. To calculate the dispersion coefficient, the mean and standard deviation of all feature data within that dimension are first calculated. The mean reflects the central tendency of the feature data, while the standard deviation reflects the degree of dispersion. The ratio of the two is the dispersion coefficient of mineral quality distribution. For dimensions with large datasets, a segmented calculation method is used. The data is divided into several segments according to time or space, and the dispersion coefficient of each segment is calculated separately before averaging to reduce the impact of extreme values ​​on the results. For example, when processing pipeline flow velocity fluctuation data, the data is divided into multiple time periods according to the sampling time. The mean and standard deviation of the flow velocity are calculated for each time period, and then the overall dispersion coefficient for that dimension is obtained.

[0075] When calculating the statistical weight distribution of the internal feature components of an eigenvector for each dimension, the compositional structure of the eigenvector is first clarified. Each eigenvector consists of multiple interrelated feature components, which reflect the mineral characteristics of that dimension from different perspectives. The calculation of the statistical weight distribution is based on the information contribution of each component in the eigenvector. By analyzing the frequency of occurrence, degree of variation, and correlation with other components, the weight value of each component is determined. Components that are active in changes in slurry characteristics are assigned higher weights, while components with slow changes or small impacts have correspondingly lower weight values. An iterative adjustment mechanism is used in the calculation process. Through multiple iterations, the weight allocation is optimized so that the weight distribution can more accurately reflect the actual importance of each component.

[0076] When generating feature vector entropy quantification results based on the dispersion coefficient of mineral distribution and the statistical weight distribution, both are used as input parameters for comprehensive calculation. First, the dispersion coefficient of mineral distribution is standardized to align with the value range of the statistical weight distribution. Then, the two are fused using a weighted combination to form a preliminary entropy quantification result. During the fusion process, the weight ratios of the two are adjusted according to the characteristics of different rheological dimensions. For dimensions where dispersion significantly impacts mineral characteristics, the weight of the dispersion coefficient is increased; for dimensions with significant differences between feature components, the influence ratio of the statistical weight distribution is increased. Finally, the preliminary result is normalized to obtain the final feature vector entropy quantification result. This result comprehensively reflects the uncertainty and information richness of the feature vector, providing a quantitative basis for subsequent mineral feature selection.

[0077] Example 3: See Figure 4 The process of filtering the core entropy sequence of mineral features by the mineral feature screening module begins with extracting the cross-dimensional association strength weight coefficients of all key mineral feature nodes in the global rheological feature topology map. In the global rheological feature topology map, each key mineral feature node is associated with multiple other nodes. Extracting the cross-dimensional association strength weight coefficients requires traversing the association records of each node with all other nodes. For each node, its cross-dimensional association strength values ​​with other nodes are collected. These values ​​are values ​​within the [0,1] interval, previously calculated and mapped using the dynamic time warping algorithm. The arithmetic mean of these values ​​is taken to obtain the average association strength of the node. This average association strength is the cross-dimensional association strength weight coefficient of the key mineral feature node, and its magnitude reflects the overall degree of association between the node and other dimensional nodes in the global topology.

[0078] The eigenvector entropy quantization result is weighted and fused with the corresponding cross-dimensional association strength weight coefficient. The eigenvector entropy quantization result is calculated using the mineral distribution dispersion coefficient and statistical weight distribution, reflecting the amount of feature information of a single node within its own rheological dimension; the cross-dimensional association strength weight coefficient reflects the association importance of a node in the global topology. During the weighted fusion process, appropriate weight ratios need to be assigned to both, and these ratios are dynamically adjusted according to the actual characteristics of the slurry. For gold concentrate slurries with high viscosity, the proportion of the cross-dimensional association strength weight coefficient is appropriately increased to highlight the mutual influence between nodes; for slurries with relatively good fluidity, the proportion of the eigenvector entropy quantization result is appropriately increased. The following formula is used for fusion:

[0079]

[0080] in, This represents the comprehensive evaluation value of the features. The weighting coefficients represent the quantization results of the eigenvector entropy values. This represents the quantization result of the eigenvector entropy value. The weighting coefficient represents the weighting coefficient of the cross-dimensional association strength. This represents the weighting coefficient for the cross-dimensional correlation strength, and .

[0081] Highly significant mineral characteristic nodes are selected by pre-setting an entropy threshold. The entropy threshold is set by referring to the distribution of comprehensive evaluation values ​​in historical data and considering the representativeness requirements of mineral samples in actual production. First, the comprehensive evaluation values ​​of all key mineral characteristic nodes over a period of time are collected, and their distribution histograms are plotted to analyze the central tendency and dispersion of the data. Based on the distribution characteristics of the histogram, an appropriate value is selected as the threshold. This threshold should be able to distinguish nodes with high comprehensive evaluation values ​​and significant impact on the overall characteristics of the slurry. When the comprehensive evaluation value of a key mineral characteristic node is higher than this threshold, it is determined to be a highly significant mineral characteristic node; otherwise, it is considered a low-significance node and is not included in the core entropy sequence.

[0082] All highly significant mineral feature nodes are sorted in descending order of entropy value to generate a mineral feature core entropy sequence. The sorting process is primarily based on the comprehensive feature evaluation value, arranging all highly significant nodes from highest to lowest comprehensive feature evaluation value. If nodes have the same comprehensive feature evaluation value, their eigenvector entropy quantification results are further compared, with nodes having higher entropy values ​​ranked higher. If the eigenvector entropy quantification results are also the same, their cross-dimensional correlation strength weight coefficients are compared, with nodes having higher coefficients ranked higher. After sorting, an ordered node sequence is formed, which is the mineral feature core entropy sequence.

[0083] After generating the core entropy sequence, its stability needs to be verified. During verification, multiple sets of slurry sample data are continuously collected, and the core entropy sequence corresponding to the mineral characteristics of each set is calculated. The changes in the order of nodes in different sequences are compared. If the relative positions of the nodes remain stable across multiple sequences, the core entropy sequence has good reliability. If significant fluctuations occur, the entropy threshold or the weighting ratio of the weighted fusion needs to be readjusted until the sequence stabilizes. Furthermore, the length of the core entropy sequence needs to be determined based on actual needs. An excessively long sequence may contain too much redundant information, increasing the complexity of subsequent processing; an excessively short sequence may miss important nodes, affecting the accuracy of sampling anomaly detection. Typically, based on the accuracy requirements for mineral sample analysis in production, the first 20%-30% of the nodes in the sequence are retained as the core entropy sequence. The specific proportion can be dynamically adjusted according to changes in the viscosity of the slurry.

[0084] The selected mineral characteristic core entropy sequence needs to be updated in real time to adapt to changes in slurry characteristics. The update frequency is consistent with the sampling frequency of the ore flow state sensing module to ensure that the core entropy sequence can reflect the latest state of the slurry in a timely manner. During each update, only newly added slurry data is processed, a new comprehensive feature evaluation value is calculated, compared with the existing core entropy sequence, nodes that no longer meet the high significance criteria are replaced, and newly emerging high significance nodes are added to the sequence and reordered. Through this dynamic update mechanism, the mineral characteristic core entropy sequence can continuously and accurately reflect the core characteristics of the slurry, providing a reliable basis for judgment for the sampling anomaly detection module.

[0085] Example 4: See Figure 5 The sampling anomaly detection module includes a sampling execution feedback unit, an instruction optimization unit, and an instruction fusion unit. The sampling execution feedback unit collects the motion state parameters of the sampling robotic arm in real time and generates motion state feedback signals. The displacement trajectory deviation data in the motion state feedback signals is acquired jointly by an encoder and a vision positioning device installed at the joints of the robotic arm. The encoder records the actual rotation angle of each joint of the robotic arm and calculates the theoretical position of the end effector by combining it with the kinematic model of the robotic arm. The vision positioning device captures images of the end effector in the workspace and analyzes the coordinates of feature points in the images to obtain the actual position. The difference between the two is the displacement trajectory deviation data, which is in millimeters and includes deviation components in the X, Y, and Z directions.

[0086] Hydraulic pressure fluctuation data is collected by pressure sensors integrated into the hydraulic lines. The sensor's sampling frequency is set according to the hydraulic system's response speed, typically collecting data every millisecond. It records the pressure changes of the hydraulic oil during different actions, with pressure fluctuation data expressed in megapascals (MPA), forming a pressure curve over time. Sampling container filling rate data is acquired through a flow sensor installed at the sampling pipe inlet. This sensor, based on electromagnetic induction, measures the volume of slurry flowing into the sampling container per unit time. Filling rate data is expressed in liters per minute (L / min), and its value varies with the viscosity of the slurry and the rhythm of the sampling robotic arm's movements. Mechanical vibration spectrum data is collected by accelerometers fixed to the robotic arm base and end effector. The sensors capture the vibration acceleration of the robotic arm during movement, converting the time-domain signal to a frequency-domain signal to obtain vibration amplitudes at different frequencies, covering a range from 0 to 1000 Hz, with amplitudes expressed in meters per second squared.

[0087] The instruction optimization unit connects to the sampling execution feedback unit, fusing the mineral characteristic core entropy sequence and the action state feedback signal to generate dynamic compensation instructions. During the fusion process, each node value in the mineral characteristic core entropy sequence is first correlated with various parameters in the action state feedback signal to determine which node changes are related to fluctuations in action state parameters. For example, when the node value reflecting slurry viscosity in the mineral characteristic core entropy sequence increases, its correlation with hydraulic execution pressure fluctuation data is analyzed. If a significant synchronous change is found, the hydraulic pressure is adjusted accordingly for the change in that viscosity node when generating the dynamic compensation instruction.

[0088] The generation of dynamic compensation commands adopts a layered processing approach. For displacement trajectory deviation data, the angle increment that each joint of the robotic arm needs to be adjusted is calculated based on the direction and magnitude of the deviation. For hydraulic actuation pressure fluctuation data, the opening adjustment amount of the hydraulic valve is determined based on the amplitude and frequency of the pressure fluctuation. For sampling container filling rate data, the inflow velocity of the slurry is changed by adjusting the valve opening on the sampling pipeline. For mechanical vibration spectrum data, the robotic arm component corresponding to the main vibration frequency is identified, and the vibration is suppressed by adjusting the motion parameters of that component.

[0089] The instruction fusion unit superimposes and integrates sampling action control instructions and dynamic compensation instructions to output the final sampling control signal. The sampling action control instructions are basic control instructions generated based on the core entropy sequence of mineral characteristics, including basic parameters such as the robotic arm's motion path, sampling time, and sampling quantity. The dynamic compensation instructions are adjustment instructions made to address deviations in the action state feedback signal. During superposition and integration, both types of instructions need to be converted to the same parameter space. For example, the robotic arm's angle adjustment and displacement compensation are uniformly converted into the number of pulses in the joint motors, and the hydraulic pressure adjustment is converted into the control current of the hydraulic valves.

[0090] During the integration process, different adjustment commands for the same control parameter are processed using a priority ranking method. When a parameter in the sampling action control command conflicts with a parameter in the dynamic compensation command, the priority is determined based on the degree of influence of the parameter on the sampling accuracy. The parameter with a greater impact is adopted first, while the other parameter is appropriately modified to reduce conflict. The final generated sampling control signal is a series of digital pulses and analog voltage signals, which are transmitted to the control system of the sampling robotic arm via cable, directly driving each actuator to complete the sampling operation.

[0091] The following is an example of the acquisition of some parameters in the action status feedback signal:

[0092]

[0093] The table records some motion state parameters at five consecutive sampling times. These parameters allow us to observe the changes in the robotic arm's motion over a short period of time, providing real-time feedback information to the instruction optimization unit.

[0094] Example 5: The system includes an adaptive sampling actuator connected to a sampling anomaly detection module, which receives sampling action control commands and executes physical sampling operations. The rotary cutting sampling head, vacuum negative pressure stabilization unit, multi-stage filtration separation chamber, and sample packaging robot in the adaptive sampling actuator work together to complete the entire process from obtaining samples from the slurry to sample packaging.

[0095] The rotary cutting sampling head is installed at the sampling port of the slurry pipeline. Its cutting action is driven by a servo motor, and the operating parameters can be adjusted according to the sampling action control commands. The cutting speed gradient refers to the rate of change of the sampling head's rotational speed over time from startup to reaching the preset rated speed. This rate of change is set based on the viscosity characteristics of the slurry. When the slurry viscosity is high, a lower speed gradient is used to allow the sampling head to accelerate slowly, avoiding sample splashing or damage to the sampling head due to excessive instantaneous impact force. When the slurry viscosity is low, a higher speed gradient can be used to shorten the time it takes for the sampling head to reach the rated speed. The cutting angle offset is the angular deviation between the cutting plane of the sampling head and the pipeline axis. By adjusting this angle, the sampling head can adapt to the flow direction of the slurry within the pipeline. When the slurry flows in a spiral pattern within the pipeline, the cutting angle offset is adjusted accordingly based on the spiral direction to ensure that the sampling head can cut into the core area of ​​the slurry flow. The reciprocating motion frequency is the number of times the sampling head travels back and forth between the sampling position and the initial position after completing one cutting action. This frequency is set according to the sampling interval requirements. When multiple consecutive samplings are required, the reciprocating motion frequency is increased to shorten the sampling interval. The cutting depth increment is the change in the depth to which the sampling head extends into the pipe during each cut. The initial cutting depth is determined based on the pipe diameter and slurry flow rate. For each subsequent cut, the depth increment is adjusted based on the previous sampling amount. If the previous sampling amount was insufficient, the cutting depth increment is increased; otherwise, it is decreased.

[0096] The vacuum negative pressure stabilization unit is connected to a rotary cutting sampling head via a pipeline, sucking in the cut mineral sample and transporting it to a multi-stage filtration and separation chamber. The negative pressure intensity threshold is the minimum negative pressure value required to maintain stable mineral sample transport. This value is determined based on the particle size and viscosity of the mineral sample. For mineral samples with larger particles and higher viscosity, a higher negative pressure intensity threshold is set to ensure smooth passage through the transport pipeline; for mineral samples with smaller particles and lower viscosity, a lower negative pressure intensity threshold is set to prevent excessive evaporation of liquid components due to excessive negative pressure. The airflow pulsation suppression coefficient is achieved by adjusting the opening of the pressure regulating valve in the vacuum system. When airflow pulsation is detected in the transport pipeline, the suppression coefficient is adjusted according to the amplitude of the pulsation. The larger the coefficient, the greater the adjustment range of the pressure regulating valve, to quickly suppress airflow pulsation, ensuring uniform force on the mineral sample during transport and preventing sedimentation of the mineral sample in the pipeline due to pulsation. The air pressure balance value of the sample transmission pipeline refers to the air pressure difference between the pipeline inlet and outlet. This value is set according to the length and diameter of the transmission pipeline. When the pipeline is long and the diameter is small, a larger air pressure balance value is set to overcome pipeline resistance; when the pipeline is short and the diameter is large, a smaller air pressure balance value is set to reduce energy consumption.

[0097] The multi-stage filtration separation chamber consists of multiple sequentially connected filtration units. Each filtration unit contains a screen with a different aperture size. After the mineral sample enters, it passes through each level of screen for filtration, achieving the separation of particles of different sizes. The screen aperture combination sequence is a sequence of the aperture sizes of each level of screen arranged in the filtration order. This sequence is determined based on the expected particle size distribution of the mineral particles in the sample, typically arranged from largest to smallest, ensuring that larger particles are separated first, while smaller particles enter subsequent filtration units. The vibration separation amplitude curve is a curve showing the change in vibration amplitude over time during the filtration process. The shape of the curve is set according to the viscosity of the mineral sample and the particle agglomeration. For mineral samples with high viscosity and easy particle agglomeration, a curve with gradually increasing amplitude is used to break up particle agglomeration through continuously increasing vibration; for mineral samples with low viscosity and good particle dispersion, a stable vibration amplitude curve is used to reduce unnecessary energy consumption. The mineral residue cleaning cycle is the time interval for regularly cleaning the screen. It is determined based on the degree of screen blockage. When the filtration efficiency of the screen is detected to have dropped to a certain level, the cleaning cycle is triggered. The cleaning methods include reverse airflow purging and mechanical scraping to remove the mineral particles remaining on the screen and restore the filtration performance of the screen.

[0098] A sample packaging robot is installed at the outlet of the multi-stage filtration and separation chamber. Once the filtered mineral sample enters the designated location, the robot loads it into a sample container and packages it. The robot's motion parameters include gripping force, moving speed, and packaging pressure. These parameters are adjusted according to the material of the sample container and the characteristics of the mineral sample. For fragile containers, a smaller gripping force and packaging pressure are used; for heavier mineral samples, the moving speed is increased to shorten the packaging time. After packaging, the sample packaging robot places the packaged sample in a designated storage location, awaiting further processing.

[0099] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0100] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An automated sampling system for viscous gold concentrate based on intelligent sensing, characterized in that, include: The mineral flow state sensing module is configured to collect dynamic parameters of rheological characteristics of viscous gold concentrate slurry in real time through a multi-source sensor array, perform heterogeneous data fusion processing on the dynamic parameters of rheological characteristics, and output a slurry rheological parameter matrix. The dynamic parameters of rheological characteristics included in the mineral flow state sensing module specifically cover slurry viscosity gradient data, mineral particle size distribution data, solid-liquid two-phase concentration data, pipeline flow velocity pulsation data, and temperature field distribution data. The mineral sample characteristic fusion module is connected to the mineral flow state sensing module and is configured to perform multi-dimensional feature decoupling on the mineral slurry rheological parameter matrix, generate a mineral slurry feature tensor, and construct a local correlation topology for each rheological dimension based on the mineral slurry feature tensor. The mineral topology map module is configured to extract key mineral feature nodes in each rheological dimension, calculate the cross-dimensional correlation strength between the key mineral feature nodes, and integrate all local correlation topologies to form a global rheological feature topology map of viscous gold concentrate. The mineral feature screening module is configured to parse the feature vector entropy value of each rheological dimension from the slurry feature tensor, and combine it with the cross-dimensional correlation strength in the global rheological feature topology map to screen out the core entropy sequence of mineral features. The sampling anomaly detection module is connected to the mineral feature screening module and is configured to trigger the sampling anomaly determination of viscous gold concentrate based on the core entropy sequence of the mineral features, and generate sampling action control instructions. The sampling anomaly detection module also includes: The sampling execution feedback unit is configured to collect the motion status parameters of the sampling robotic arm in real time and generate motion status feedback signals. The instruction optimization unit, connected to the sampling execution feedback unit, is configured to fuse the core entropy sequence of mineral features with the action state feedback signal to generate a dynamic compensation instruction. The instruction fusion unit is configured to superimpose and integrate the sampling action control instruction and the dynamic compensation instruction to output the final sampling control signal; The operation of parsing the feature vector entropy value by the mineral feature screening module includes: Obtain the dispersion coefficient of mineral distribution for each rheological dimension; Calculate the statistical weight distribution of the internal feature components for each dimension of the feature vector; Based on the dispersion coefficient of the mineral distribution and the statistical weight distribution, a quantization result of the entropy value of the feature vector is generated; The steps of the mineral feature screening module in screening the core entropy sequence of mineral features include: Extract the cross-dimensional correlation strength weight coefficients of all key mineral feature nodes in the global rheological feature topology map; The quantization result of the feature vector entropy value is weighted and fused with the corresponding cross-dimensional correlation strength weight coefficient; Highly significant mineral feature nodes are selected by setting a preset entropy threshold. The core entropy sequence of the mineral features is generated by sorting all highly significant mineral feature nodes in descending order of entropy value.

2. The automated sampling system for viscous gold concentrate based on intelligent sensing according to claim 1, characterized in that, The mineral sample characteristic fusion module performs multi-dimensional feature decoupling operations, including: The slurry rheological parameter matrix is ​​subjected to noise suppression and data normalization to generate a standardized slurry rheological parameter matrix; The feature decoupling engine separates the independent feature components of each dimension in the standardized slurry rheological parameter matrix; The slurry feature tensor is generated by aggregating the independent feature components of all dimensions.

3. The automated sampling system for viscous gold concentrate based on intelligent sensing according to claim 1, characterized in that, The process by which the mineral topology map module calculates the cross-dimensional correlation strength includes: Select any two key mineral feature nodes and extract the feature response sequences of the key mineral feature nodes in the slurry feature tensor; The morphological similarity measure of the feature response sequence is calculated using a dynamic time warping algorithm; The morphological similarity metric is mapped to a cross-dimensional association strength weighting coefficient.

4. The automated sampling system for viscous gold concentrate based on intelligent sensing according to claim 1, characterized in that, The action status feedback signal specifically includes displacement trajectory deviation data of the sampling robot arm, hydraulic actuation pressure fluctuation data, sampling container filling rate data, and mechanical vibration spectrum data.

5. The automated sampling system for viscous gold concentrate based on intelligent sensing according to claim 1, characterized in that, The system further includes: An adaptive sampling actuator, connected to the sampling anomaly detection module, is configured to receive the sampling action control command and perform physical sampling operations; The adaptive sampling actuator includes a rotary cutting sampling head, a vacuum negative pressure stabilization unit, a multi-stage filtration separation chamber, and a sample packaging robot.

6. The automated sampling system for viscous gold concentrate based on intelligent sensing according to claim 5, characterized in that, The operating parameters of the rotary cutting sampling head include cutting speed gradient, cutting angle offset, reciprocating motion frequency, and cutting depth increment; The parameters that the vacuum negative pressure stabilization unit adjusts in real time include the negative pressure intensity threshold, the airflow pulsation suppression coefficient, and the air pressure balance value of the sample transmission pipeline. The separation parameters configured in the multi-stage filtration separation chamber include the screen aperture combination sequence, vibration separation amplitude curve, and mineral residue cleaning cycle.

Images