Anti-blocking belt conveyor automatic multi-point sampling system

Abstract

The application discloses an anti-blocking belt conveyor automatic multi-point sampling system, and relates to the technical field of material sampling equipment. The system comprises a parameter acquisition module, a sampling decision module, an anti-blocking decision module, a cooperative control module and a cooperative execution module. The system collects real-time speed, load, vibration spectrum, material granularity distribution and moisture content data of the belt. The sampling decision module determines the optimal sampling point and sampling time along the belt by means of an improved adaptive algorithm which combines the material flow uniformity model and the vibration energy spectrum characteristics. The anti-blocking decision module generates adaptive control parameters for each point according to the material parameters. After the relevant instructions are uniformly issued, the distributed sampling units cooperatively complete the fixed-point and fixed-time sampling, and the corresponding anti-blocking measures are simultaneously enabled. The system can adapt to the on-site conveying conditions to carry out dynamic sampling, improve the representativeness of the collected samples, reduce the material blocking problem at the sampling points and stably complete the multi-point material sampling operation.

Description

Technical Field

[0001] This invention belongs to the technical field of material sampling equipment, specifically an automatic multi-point sampling system for anti-clogging belt conveyors. Background Technology

[0002] Most existing material sampling operations on belt conveyors adopt a fixed-point, fixed-time sampling mode. The sampling points and sampling times are predetermined, and the sampling rhythm is controlled solely based on the basic belt speed. Very little data on equipment vibration or the physical properties of the material itself is collected and used in the sampling logic. This conventional sampling mode can only meet the basic material sampling needs, and the overall operation logic is relatively simple, making it impossible to make flexible adjustments based on the actual operating conditions of the belt conveyor.

[0003] Fixed sampling arrangements are difficult to replicate the actual material flow distribution along belt conveyor lines. The density of material transport varies across different sections of the belt, and samples taken from fixed points cannot fully reflect the actual state of the entire batch of transported material, resulting in limited sampling coverage. Furthermore, various transported materials exhibit differences in particle size and moisture content. Conventional sampling structures lack differentiated control methods tailored to the specific characteristics of each material. When materials with larger particle sizes or higher moisture content enter the sampling structure, they tend to aggregate and accumulate, obstructing the flow of material at the sampling ports.

[0004] When sampling operations are carried out simultaneously at multiple locations on site, each sampling point cannot be autonomously adapted and adjusted according to the actual material conditions at its location. The unified operation mode is difficult to adapt to the material conveying conditions in different areas, and the sampling operation process is prone to interruption due to material blockage. It is difficult to maintain the overall smoothness of the on-site material sampling operation. It is necessary to establish a dynamic sampling point determination calculation method, and at the same time build a point-specific control logic that adapts to different material properties to optimize the overall sampling operation mode. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art; Therefore, this invention proposes an automatic multi-point sampling system for anti-clogging belt conveyors, comprising: The parameter acquisition module acquires the belt conveyor operating status parameters and material characteristic parameters. The belt conveyor operating status parameters include the real-time belt speed, real-time load, and vibration spectrum. The material characteristic parameters include the estimated particle size distribution and moisture content of the currently conveyed material. The sampling decision module processes the belt conveyor's operating status parameters using an improved adaptive sampling point decision algorithm, calculating multiple optimal sampling points and corresponding sampling times along the belt length direction. The improved adaptive sampling point decision algorithm optimizes the sampling decision based on the belt conveyor's material flow uniformity model and vibration energy spectrum characteristics. The anti-clogging decision module, based on material characteristic parameters, calls the anti-clogging decision model to generate anti-clogging control parameters corresponding to each optimal sampling point; The collaborative control module sends the multiple optimal sampling points, their corresponding sampling times, and the anti-blocking control parameters to the multi-point sampling collaborative controller. The collaborative execution module coordinates multiple distributed sampling execution units through the multi-point sampling collaborative controller to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters, so as to obtain representative samples within the multi-point sampling cycle and avoid sampling point blockage.

[0006] Furthermore, an improved adaptive sampling point decision algorithm is applied to the belt conveyor's operating state parameters to calculate multiple optimal sampling points and corresponding sampling times along the belt length direction, including: Using the real-time speed and real-time load of the belt, the material flow rate passing through the cross-section of the belt per unit time is calculated, and combined with historical data, a material flow uniformity model of the belt conveyor is constructed to predict the uniformity of material distribution on the belt in the near future. Frequency domain analysis is performed on the vibration spectrum to identify the main frequency band energy related to the stability of material flow. The vibration energy spectrum characteristics are characterized by the stability of the main frequency band energy. The improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity and is constrained by the stability of vibration energy spectrum characteristics. In the two-dimensional space formed by the belt length direction and the time dimension, it searches for a set of multiple sampling points, so that the sampling points in the set are spatially dispersed and uniformly distributed in time, and the vibration energy spectrum characteristics corresponding to each sampling point are in the stable range. The improved adaptive sampling point decision algorithm outputs the coordinates of multiple optimal sampling points that satisfy the above conditions along the belt length direction, as well as the belt running time corresponding to each optimal sampling point. This belt running time is the sampling time corresponding to the optimal sampling point.

[0007] Furthermore, the improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity and is constrained by the stability of vibration energy spectrum characteristics. It searches for a set of multiple sampling points in a two-dimensional space defined by the belt length direction and the time dimension, including: Define belt length and time as two-dimensional decision variables; An objective function is established, which is calculated on a selected set of multiple sampling points to determine the degree to which the collected samples represent the overall material distribution, i.e., the representativeness of the material distribution uniformity. Establish constraints that require the vibrational energy spectrum characteristics of each selected sampling point at the corresponding sampling time to fall within a preset stable interval; A heuristic optimization algorithm is used to search for the set of sampling points that maximize the objective function value among all sampling point sets that satisfy the constraints, and this set is used as the final set of multiple optimal sampling points and their corresponding sampling times.

[0008] Furthermore, the heuristic optimization algorithm is a hybrid particle swarm optimization-simulated annealing optimization algorithm.

[0009] Furthermore, based on material characteristic parameters, the anti-clogging decision model is invoked to generate anti-clogging control parameters corresponding to each optimal sampling point, including: The anti-clogging control parameters include pulse airflow pressure, guide vane angle, and vibrator frequency. The estimated particle size distribution and moisture content of the currently conveyed material are input into the pre-trained anti-clogging decision model; The anti-clogging decision model includes a material clogging tendency assessment sub-network, which outputs a quantified value of clogging risk level based on the combination of the estimated particle size distribution and moisture content of the currently conveyed material. Based on the quantified value of the blockage risk level, a preset anti-blockage control parameter mapping table is queried. The anti-blockage control parameter mapping table defines the recommended pulse airflow pressure range, guide vane angle range, and vibrator frequency range for different blockage risk levels. Within the range of recommended parameters obtained from the query, and combined with the local information of the optimal sampling point, the specific pulse airflow pressure, guide vane angle, and vibrator frequency are calculated and used as the anti-clogging control parameters.

[0010] Furthermore, the material blockage tendency assessment subnetwork outputs a blockage risk level quantification value based on the estimated particle size distribution and moisture content of the currently conveyed material, including: The material blockage tendency assessment subnetwork is a deep neural network. Its input layer receives the feature vector encoded with the estimated particle size distribution and moisture content of the currently conveyed material as input features. The deep neural network performs nonlinear transformations on the input features through multiple hidden layers to learn the complex characteristics related to the adhesion, flowability and clogging risk of materials under the combined effect of particle size distribution and moisture content. The output layer of the deep neural network maps the learned features into a continuous numerical value. After standardization, this numerical value falls within a preset range of quantified values ​​for congestion risk level, which is then used as the quantified value for congestion risk level.

[0011] Furthermore, the deep neural network is a long short-term memory network that includes an attention mechanism.

[0012] Furthermore, the multiple optimal sampling points, their corresponding sampling times, and the anti-blocking control parameters are sent to the multi-point sampling collaborative controller, including: Each optimal sampling point and its corresponding sampling time are encapsulated into a sampling task instruction unit. Each sampling task instruction unit includes the target location coordinates, the trigger timestamp, and the associated sampling execution unit identifier. Each optimal sampling point is encapsulated into an anti-blocking control instruction unit, and logically bound to the sampling task instruction unit corresponding to the same sampling point. All the encapsulated and bound instruction units are arranged into a unified control instruction sequence according to the order of sampling time, and then sent to the multi-point sampling collaborative controller.

[0013] Furthermore, the multi-point sampling coordination controller coordinates multiple distributed sampling execution units to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters, including: The multi-point sampling cooperative controller internally maintains a global clock and a state list of all sampling execution units; When the global clock reaches the trigger timestamp specified in a certain sampling task instruction unit, the multi-point sampling coordination controller sends a sampling action start command to the sampling execution unit specified in the sampling task instruction unit; Simultaneously, the multi-point sampling collaborative controller sends the anti-blocking control command unit to the anti-blocking device control unit associated with the current optimal sampling point. The anti-blocking device control unit drives the corresponding pulse airflow valve, guide plate actuator and vibrator to perform actions based on the received pulse airflow pressure, guide plate angle and vibrator frequency. Upon receiving a start command, the sampling execution unit drives the sampling robotic arm or sampling shovel to move to the optimal sampling point coordinates, performs a complete sampling action, and obtains a material sample at that point.

[0014] Furthermore, the anti-clogging device control unit drives the corresponding pulse airflow valve, guide plate actuator, and vibrator to operate based on the received pulse airflow pressure, guide plate angle, and vibrator frequency, including: The anti-clogging device control unit includes three independent control channels, which are used to control the pulse airflow valve, the guide plate actuator, and the vibrator, respectively. For the control channel of the pulse airflow valve, the received pulse airflow pressure parameters are converted into corresponding air pressure regulating valve opening control signals and sent to the target pulse airflow valve so that it sprays airflow at a specified pressure before and after the sampling action is executed; For the control channel of the guide vane actuator, the received guide vane angle parameters are converted into corresponding motor rotation angle or hydraulic cylinder stroke control signals and sent to the target guide vane actuator so that it adjusts the guide vane to the specified angle; For the control channel of the vibrator, the received vibrator frequency parameters are converted into the corresponding frequency control signal of the inverter output and sent to the target vibrator to make it vibrate at the specified frequency.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Based on real-time belt speed, real-time load, vibration spectrum, and estimated particle size distribution and moisture content of materials, an adaptive sampling point decision-making logic is built by combining the belt conveyor material flow uniformity model and vibration energy spectrum characteristics. This logic can accurately arrange multiple sampling points along the entire belt extension path and match the precise sampling time corresponding to different locations. The arrangement of sampling points is no longer limited by a fixed layout, and the sampling time can be autonomously adjusted according to the real-time operating status of the equipment. The changes in material density and fluctuations in material flow in different sections of the entire belt conveyor can be covered by the sampling action, and the actual material conveying pattern throughout the entire material conveying process can be fully integrated into the sampling process.

[0016] Based on the material's estimated particle size distribution and moisture content, two inherent material properties, an independent anti-clogging decision model is invoked to process the data. For each confirmed sampling point, corresponding anti-clogging control parameters are generated individually. The physical states of materials in different sampling areas can be accurately distinguished, and each sampling point can be matched with an operational adjustment mode that suits its own material state. When materials enter the sampling execution structure, the material arrangement and passage are rationally adjusted, gradually reducing the phenomenon of material aggregation and accumulation within the sampling structure.

[0017] The multi-point sampling collaborative controller can simultaneously receive sampling point information, sampling time information, and various point-specific anti-blocking control parameters. While the distributed sampling execution unit is carrying out sampling actions, the corresponding anti-blocking related operations are started simultaneously. Multiple work processes cooperate and advance in sync. The material sampling process carried out in parallel in multiple areas can maintain a stable and continuous operating state, and large-scale multi-point material sampling operations on site can be carried out in an orderly and continuous manner. Attached Figure Description

[0018] Figure 1 This is a timing diagram of an automatic multi-point sampling system for anti-clogging belt conveyors according to the present invention; Figure 2 A flowchart for calculating the optimal sampling point and corresponding sampling time for the improved adaptive sampling point decision algorithm; Figure 3 This is a flowchart for searching a set of sampling points in two-dimensional space. Detailed Implementation

[0019] The technical solution of the present invention will be clearly and completely described below with reference to 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Example See Figure 1 An automatic multi-point sampling system for anti-clogging conveyor belts is disclosed, the specific implementation of which is as follows: The system includes a parameter acquisition module, a sampling decision module, an anti-clogging decision module, a collaborative control module, and a collaborative execution module. The parameter acquisition module collects real-time operating status parameters and material characteristic parameters of the conveyor belt. The operating status parameters include real-time belt speed, real-time load, and vibration spectrum. The material characteristic parameters include the estimated particle size distribution and moisture content of the currently conveyed material. After receiving the above parameters, the sampling decision module processes the conveyor belt operating status parameters using an improved adaptive sampling point decision algorithm, calculating multiple optimal sampling points and corresponding sampling times along the belt length. This improved adaptive sampling point decision algorithm optimizes the sampling decision based on the conveyor belt material flow uniformity model and vibration energy spectrum characteristics. Based on the material characteristic parameters, the anti-clogging decision module calls the anti-clogging decision model to generate anti-clogging control parameters corresponding to each optimal sampling point. The collaborative control module sends the calculated multiple optimal sampling points and corresponding sampling times, as well as the anti-clogging control parameters, to the multi-point sampling collaborative controller. The collaborative execution module coordinates multiple distributed sampling execution units through a multi-point sampling collaborative controller to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters, thereby acquiring representative samples within the multi-point sampling cycle and avoiding sampling point blockage.

[0021] In one embodiment of the present invention, see [reference] Figure 2An improved adaptive sampling point decision algorithm is applied to the operating parameters of the belt conveyor to calculate multiple optimal sampling points and corresponding sampling times along the belt length. Using the real-time belt speed and load, the material flow rate across the belt cross-section per unit time is calculated. Combined with historical data, a material flow uniformity model is constructed to predict the material distribution uniformity on the belt in the near future. Frequency domain analysis of the vibration spectrum identifies the dominant frequency band energy related to material flow stability; the vibration energy spectrum characteristics are characterized by the stability of the dominant frequency band energy. The improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity, constrained by the stability of the vibration energy spectrum characteristics. In a two-dimensional space comprising the belt length direction and the time dimension, it searches for a set of multiple sampling points that are spatially dispersed, temporally uniformly distributed, and whose vibration energy spectrum characteristics are all within a stable range.

[0022] The aforementioned optimal sampling point refers to the sampling point obtained by optimizing the calculation in the two-dimensional space of belt length-time according to a defined objective function and constraints through an improved adaptive sampling point decision algorithm.

[0023] The improved adaptive sampling point decision algorithm outputs the coordinates of multiple optimal sampling points that satisfy the above conditions along the belt length direction, as well as the belt running time corresponding to each optimal sampling point. This belt running time is the sampling time corresponding to the optimal sampling point.

[0024] In practical implementation, when processing the belt conveyor operating status parameters using the improved adaptive sampling point decision algorithm, the material flow rate across the belt cross-section per unit time is calculated using the real-time belt speed and real-time load. The material flow rate is calculated using the following formula: ; Where: Q represents the material flow rate across the belt cross-section per unit time, v represents the real-time belt speed, and λ represents the real-time belt load, i.e., the mass of material carried per unit length of the belt. In some embodiments, this material flow rate is combined with historical data to construct a belt conveyor material flow uniformity model. This model predicts the uniformity of material distribution on the belt in the near future using a time series prediction method. The uniformity is characterized by the coefficient of variation of the mass of the material along the belt length. Optionally, the historical data includes the real-time belt speed sequence, the real-time belt load sequence, and the corresponding measured material distribution values ​​recorded over several past sampling periods.

[0025] In practical implementation, when performing frequency domain analysis on the vibration spectrum, the vibration spectrum acquired by the parameter acquisition module is transformed to the frequency domain through a Fast Fourier Transform (FFT) to identify the dominant frequency band energy related to the stability of material flow. This dominant frequency band energy corresponds to the characteristic frequency range of the conveyor belt idlers and conveyor belt under tension. The vibration energy spectrum characteristics are characterized by the stability of the dominant frequency band energy, which is obtained by calculating the ratio of the standard deviation to the mean of the dominant frequency band energy within a continuous time window. It can be understood that when the material flow is stable, the fluctuation amplitude of the dominant frequency band energy is small, and the ratio of the standard deviation to the mean of the dominant frequency band energy within a continuous time window is lower than a preset threshold; when material accumulates or segregates, the fluctuation amplitude of the dominant frequency band energy increases, and the ratio of the standard deviation to the mean of the dominant frequency band energy within a continuous time window exceeds the preset threshold. In some embodiments, the real-time values ​​of the vibration energy spectrum characteristics are used to determine whether the current sampling point is suitable for sampling.

[0026] In its implementation, the improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity, constrained by the stability of vibration energy spectrum characteristics. It searches for a set of multiple sampling points in a two-dimensional space defined by the belt length direction and the time dimension. Specifically, the representativeness of material distribution uniformity is quantified by the mutual information between the predicted material distribution uniformity value corresponding to the sampling point set and the overall predicted material distribution uniformity value of the belt. The constraint of stable vibration energy spectrum characteristics requires that the vibration energy spectrum characteristics corresponding to each sampling point in the set are within a stable interval, defined by the low and high quantiles of the historical data of the vibration energy spectrum characteristics. Optionally, when searching in the two-dimensional space, the algorithm ensures that the sampling points in the resulting set are spatially dispersed and temporally uniform. Spatially dispersed means that the distance between any two adjacent sampling points along the belt length direction is not less than a preset minimum spacing threshold, and temporally uniform means that the deviation between the sampling time interval of any two adjacent sampling points and the theoretical average sampling interval does not exceed the allowable error range.

[0027] In practical implementation, the improved adaptive sampling point decision algorithm outputs the coordinates of multiple sampling points satisfying the above conditions along the belt length direction, as well as the belt running time corresponding to each sampling point. This belt running time is the corresponding sampling moment. It can be understood that the belt running time starts counting from the moment the belt conveyor starts. When the belt reaches a certain coordinate position and the count reaches the corresponding sampling moment, the sampling execution unit is triggered.

[0028] In one embodiment of the present invention, see [reference] Figure 3The improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity, constrained by the stability of vibration energy spectrum characteristics. It searches for a set of multiple sampling points in a two-dimensional space defined by the belt length and time dimension. This includes defining belt length and time as two-dimensional decision variables, establishing an objective function that calculates the degree to which samples collected from the selected set of sampling points represent the overall material distribution, i.e., the representativeness of material distribution uniformity. Simultaneously, constraints are established, requiring that the vibration energy spectrum characteristics of each selected sampling point at the corresponding sampling time fall within a preset stable interval. A hybrid particle swarm optimization-simulated annealing algorithm is used as a heuristic optimization algorithm. Among all possible sets of sampling points satisfying the constraints, the set of sampling points that maximizes the objective function value is searched, serving as the final set of multiple optimal sampling points and their corresponding sampling times.

[0029] In practical implementation, the improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity and is constrained by the stability of vibration energy spectrum characteristics. When searching for a set of multiple sampling points in a two-dimensional space consisting of the belt length direction and the time dimension, the belt length and time are defined as two-dimensional decision variables, where the belt length is denoted as... The time is recorded as Then each point in the two-dimensional decision space Represents the coordinates along the length of the belt. and corresponding belt running time One of the candidate sampling points.

[0030] In some embodiments, an objective function is established. Used to calculate a set of selected sampling points The degree to which the collected samples represent the overall material distribution, i.e., the representativeness of the material distribution uniformity, is the objective function. The specific form is as follows: ; in: Represents the set of sampling points The total number of sampling points in the middle, Indicates the first The weighting coefficients for each sampling point Indicates the first The mutual information between the material distribution uniformity at each sampling point and the overall material distribution uniformity of the conveyor belt. Optional, weighting coefficients. The material flow rate at each sampling point is allocated according to its proportion to the total material flow rate, so that areas with higher material flow rates correspond to higher weighting coefficients.

[0031] In practical implementation, constraints are established requiring that the vibration energy spectrum characteristics at each selected sampling point at a given time must fall within a preset stable interval. Specifically, the vibration energy spectrum characteristics are denoted as... , indicating at time The main frequency band energy stability index extracted from the vibration spectrum of the belt conveyor. The stable interval is denoted as... ,in The 5th percentile of historical vibrational energy spectrum characteristic data. The 95th percentile is taken from historical vibrational energy spectrum characteristic data. The constraint condition is expressed as for the set... Each sampling point in Its corresponding vibrational energy spectrum characteristics satisfy It is understood that only candidate sampling points that meet this constraint are allowed to enter the subsequent search process. In some embodiments, when the vibrational energy spectrum characteristics... When the signal exceeds the stable range, it indicates that the conveyor belt is in a non-stable operating state at that moment. Sampling at this moment will not yield representative samples, so the algorithm automatically removes the corresponding candidate sampling points.

[0032] In its implementation, a hybrid particle swarm optimization (PSO)-simulated annealing (SUS) algorithm is used as the heuristic optimization algorithm. It searches for the set of sampling points that maximize the objective function value among all possible sampling point sets satisfying the constraints, resulting in the final set of multiple optimal sampling points and their corresponding sampling times. Specifically, the PSO-SUS optimization algorithm is used for global search, while simulated annealing is used for local perturbation to avoid getting trapped in local optima. Optionally, during algorithm initialization, several sampling point sets are randomly generated as particles in the swarm, with each particle corresponding to a candidate sampling point set. Optionally, in each iteration, the PSO algorithm updates the velocity and position of each particle based on its individual optimal position and the swarm's optimal position, while the simulated annealing algorithm accepts a degraded solution with a certain probability, which gradually decreases as the number of iterations increases. It can be understood that when the algorithm reaches the preset maximum number of iterations or the improvement in the objective function value is less than a preset threshold, the algorithm terminates and outputs the currently found optimal sampling point set.

[0033] In one embodiment of the present invention, when generating anti-clogging control parameters for each optimal sampling point based on material characteristic parameters using an anti-clogging decision model, the anti-clogging control parameters include pulse airflow pressure, guide vane angle, and vibrator frequency. The estimated particle size distribution and moisture content are input into a pre-trained anti-clogging decision model, which includes a material clogging tendency assessment sub-network. This sub-network is a long short-term memory network incorporating an attention mechanism. Its input layer receives feature vectors encoded with the estimated particle size distribution and moisture content. Through multiple hidden layers, it performs nonlinear transformations on the input features, learning the complex characteristics related to the material's adhesion, flowability, and clogging risk under the combined influence of particle size distribution and moisture content. The output layer maps the learned features to a continuous numerical value. This value, after standardization, falls within a preset clogging risk level quantification range, serving as the clogging risk level quantification value. Based on this clogging risk level quantification value, a preset anti-clogging control parameter mapping table is consulted. This mapping table defines the recommended pulse airflow pressure range, guide vane angle range, and vibrator frequency range for different clogging risk levels. Within the range of recommended parameters obtained from the query, and combined with the local information of the optimal sampling point, the specific pulse airflow pressure, guide vane angle, and vibrator frequency are calculated as anti-clogging control parameters.

[0034] In practical implementation, the output layer of the material blockage tendency assessment subnetwork maps the learned features into a continuous numerical value. After standardization, this value falls within a preset range of blockage risk level quantification values, which are then used as the blockage risk level quantification values. The standardization process uses the following formula: ; in: This represents the standardized quantitative value of the congestion risk level. This represents the raw, continuous numerical value directly output by the output layer of a deep neural network. and These represent the minimum and maximum values ​​of the original output values ​​in the training dataset, respectively. and These represent the lower and upper limits of a preset range of quantified congestion risk levels, respectively. In some embodiments, the preset range of quantified congestion risk levels is... ,at this time , It is understandable that, after standardization, the quantified value of the congestion risk level... Always fall Within the interval, The closer the value is to 1, the higher the risk of congestion. The closer to 0, the lower the risk of congestion.

[0035] In practical implementation, a preset anti-clogging control parameter mapping table is consulted based on the quantified value of the clogging risk level. This mapping table defines the recommended pulse airflow pressure range, guide vane angle range, and vibrator frequency range for different clogging risk levels. Optionally, the anti-clogging control parameter mapping table uses the tenths interval of the quantified clogging risk level as an index, for example... , ... Each index interval corresponds to a set of recommended parameter ranges. Within the range of recommended parameters obtained from the query, and combined with the local information of the optimal sampling point, the specific pulse airflow pressure, guide vane angle, and vibrator frequency are calculated as anti-clogging control parameters. In specific implementations, the local information includes the belt conveyor structure clearance size at the sampling point, the relative distance between the sampling execution unit and the belt surface, and the historical sampling frequency. In some embodiments, when the clogging risk level quantification value... Falling When the interval is specified, the recommended range for pulsed airflow pressure is: The recommended range for the deflector angle is: The recommended frequency range for the vibrator is: The final specific parameter values ​​are determined by combining the belt conveyor structure gap size in the local information (the smaller the gap, the larger the airflow pressure and the larger the guide plate angle).

[0036] In practical implementation, the material blockage tendency assessment subnetwork is a long short-term memory network (LSTM) incorporating an attention mechanism. This attention mechanism assigns different weights to features of different dimensions in the input feature vector, allowing the network to focus on particle size distribution ranges and moisture content values ​​that contribute significantly to blockage risk. Optionally, the training data for the LSM network incorporating the attention mechanism consists of experimental material blockage data measured in the laboratory under different combinations of particle size distribution and moisture content. The training label is the measured quantified value of the blockage risk level. It can be understood that when the estimated particle size distribution contains more than 30% fine particles (particle size less than 0.074 mm) and the moisture content is between 8% and 12%, the raw value output by the LSM network incorporating the attention mechanism will be... The quantified value of the congestion risk level is relatively large and obtained after standardization. Close to 1.

[0037] In one embodiment of the present invention, when multiple optimal sampling points, their corresponding sampling times, and anti-blocking control parameters are sent to the multi-point sampling collaborative controller, each optimal sampling point and its corresponding sampling time are encapsulated into a sampling task instruction unit. Each sampling task instruction unit includes the target position coordinates, a trigger timestamp, and an associated sampling execution unit identifier. The anti-blocking control parameters corresponding to each optimal sampling point are encapsulated into an anti-blocking control instruction unit and logically bound to the sampling task instruction unit corresponding to the same sampling point. All encapsulated and bound instruction units are arranged into a unified control instruction sequence according to the order of sampling times and sent to the multi-point sampling collaborative controller.

[0038] In practical implementation, when multiple optimal sampling points, their corresponding sampling times, and anti-blocking control parameters are sent to the multi-point sampling collaborative controller, each optimal sampling point and its corresponding sampling time are encapsulated into a sampling task instruction unit. Each sampling task instruction unit includes the target position coordinates, the trigger timestamp, and the associated sampling execution unit identifier, where the target position coordinates are represented as follows: Two-dimensional coordinates in form Indicates the position coordinates along the length of the belt. Indicates the position coordinates perpendicular to the belt width direction, trigger timestamp. Represents the absolute time value calculated from the start time of the belt conveyor; sampling execution unit identifier. This is a unique number pre-assigned to each distributed sampling execution unit. In some embodiments, the data structure of the sampling task instruction unit adopts a fixed-length binary frame format, with the frame header byte identifying the instruction type and the target position coordinates stored sequentially in the frame body. , Floating-point values, trigger timestamp Unsigned integer values ​​and sampling execution unit identifiers The byte value. Optionally, when there are multiple optimal sampling points, the corresponding multiple sampling task instruction units are arranged according to the trigger timestamp. They are generated in ascending order.

[0039] In practical implementation, the anti-blocking control parameters corresponding to each optimal sampling point are encapsulated into an anti-blocking control instruction unit, and logically bound to the sampling task instruction unit corresponding to the same sampling point. Logical binding is achieved by appending the index number of the target sampling task instruction unit to the header of the anti-blocking control instruction unit. Specifically, the data structure of the anti-blocking control instruction unit includes a pulse airflow pressure field, a guide vane angle field, and a vibrator frequency field. The pulse airflow pressure field stores a floating-point value in megapascals, the guide vane angle field stores an integer value in degrees, and the vibrator frequency field stores an integer value in hertz. In some embodiments, the logical binding is implemented by adding a binding identifier field to the anti-blocking control instruction unit. The value of this field is equal to the position number of the corresponding sampling task instruction unit in the instruction sequence. It can be understood that through logical binding, when the sampling task instruction unit is sent to the multi-point sampling co-controller, the anti-blocking control instruction unit bound to it is simultaneously identified as an associated instruction, thereby ensuring the synchronous execution of sampling actions and anti-blocking measures.

[0040] In practice, all encapsulated and bound instruction units are arranged into a unified control instruction sequence according to the sampling time sequence, and then sent to the multi-point sampling collaborative controller. Specifically, the sampling time sequence is determined by comparing the trigger timestamps in each sampling task instruction unit. The numerical value determines the trigger timestamp. Smaller instruction units are placed at the beginning of the sequence, triggering timestamps. Instruction units with larger values ​​are placed at the end of the sequence. For any two sampling task instruction units, if the trigger timestamp... If they are equal, then the target position coordinates will be used. The values ​​are sorted in ascending order. Optionally, the unified control command sequence is stored in a circular queue buffer. Each unit in the buffer stores a complete command frame. The header of the command frame contains a type flag to distinguish between sampling task command units and anti-blocking control command units. It can be understood that after the control command sequence is arranged, the entire sequence is sent to the multi-point sampling co-controller in the form of data blocks via industrial Ethernet or fieldbus. After receiving the sequence, the multi-point sampling co-controller parses it sequentially and stores it in its internal command cache. In some embodiments, the total length of the control command sequence is calculated using the following formula: ; in: This indicates the total number of bytes in the control instruction sequence. This represents the total number of optimal sampling points. Indicates the byte length of a single sampling task instruction unit. Indicates the byte length of a single anti-blocking control instruction unit.

[0041] In one embodiment of the present invention, based on the sampling time and spatial coordinates output by the above algorithm, the system generates a sampling action start command to drive the field hardware to complete the actual sampling action. This command includes the target sampling point coordinates (position coordinates along the belt length direction and position coordinates perpendicular to the belt width direction) and the execution parameters of the sampling action (sampling shovel insertion depth, sampling duration). Optionally, when the current working state of the specified sampling execution unit is faulty or offline, the multi-point sampling coordination controller automatically skips the sampling task corresponding to the trigger timestamp and records the event in its internal log. By coordinating multiple distributed sampling execution units through the multi-point sampling coordination controller to execute the sampling action at the optimal sampling point at the corresponding sampling time, and synchronously activating the anti-blocking measures corresponding to the anti-blocking control parameters, the multi-point sampling coordination controller internally maintains a global clock and a status list of all sampling execution units. When the global clock reaches the trigger timestamp specified in a sampling task instruction unit, the multi-point sampling coordination controller sends a sampling action start command to the sampling execution unit specified in that instruction unit.

[0042] Simultaneously, the multi-point sampling coordination controller sends anti-clogging control commands to the anti-clogging equipment control unit associated with the current sampling point. The anti-clogging equipment control unit contains three independent control channels, used to control the pulse airflow valve, the guide vane actuator, and the vibrator, respectively. For the pulse airflow valve control channel, the received pulse airflow pressure parameters are converted into corresponding air pressure regulating valve opening control signals and sent to the target pulse airflow valve, causing it to spray airflow at a specified pressure before and after the sampling action. For the guide vane actuator control channel, the received guide vane angle parameters are converted into corresponding motor rotation angle or hydraulic cylinder stroke control signals and sent to the target guide vane actuator, causing it to adjust the guide vane to a specified angle. For the vibrator control channel, the received vibrator frequency parameters are converted into corresponding inverter output frequency control signals and sent to the target vibrator, causing it to vibrate at a specified frequency. After receiving the start command, the sampling execution unit drives its sampling robotic arm or sampling shovel to move to the optimal sampling point coordinates, performs a complete sampling action, and obtains the material sample at that point.

[0043] In practical implementation, a multi-point sampling coordinating controller coordinates multiple distributed sampling execution units to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters. The multi-point sampling coordinating controller internally maintains a global clock and a status list of all sampling execution units. The global clock is implemented as a counter in milliseconds and is synchronized with the system time of the belt conveyor main control system. The status list of each distributed sampling execution unit contains a unique identifier, its current working state (idle, running, fault, offline), and the timestamp of its most recent sampling action. In some embodiments, the status list is stored using a hash table structure, with the unique identifier of the sampling execution unit as the key and the status structure as the value. Optionally, the multi-point sampling coordinating controller proactively sends a status query frame to all distributed sampling execution units at preset time intervals. Upon receiving the frame, the distributed sampling execution units return a status response frame containing their current status information. The multi-point sampling coordinating controller updates the corresponding entries in the status list based on the received status response frame.

[0044] In implementation, when the global clock reaches the trigger timestamp specified in a sampling task instruction unit, the multi-point sampling co-controller sends a sampling action start command to the sampling execution unit specified in that sampling task instruction unit. Specifically, the trigger timestamp is represented by an absolute time value. The comparator inside the multi-point sampling co-controller compares the current count value of the global clock with the trigger timestamp of each sampling task instruction unit one by one. When the difference between the global clock count value and the trigger timestamp is less than a preset time synchronization error threshold, the trigger timestamp is determined to have arrived. The sampling action start command includes the coordinates of the target sampling point (position coordinates along the belt length direction and position coordinates perpendicular to the belt width direction) and the execution parameters of the sampling action (sampling shovel insertion depth, sampling duration). Optionally, when the current working state of the specified sampling execution unit is faulty or offline, the multi-point sampling co-controller automatically skips the sampling task corresponding to that trigger timestamp and records the event in the internal log.

[0045] In specific implementations, the multi-point sampling co-controller sends anti-clogging control command units to the anti-clogging device control unit associated with the current sampling point. The anti-clogging device control unit includes three independent control channels, used to control the pulse airflow valve, the guide vane actuator, and the vibrator, respectively. The pulse airflow valve control channel includes one digital output port and one analog output port; the guide vane actuator control channel includes one pulse width modulation output port or one analog output port; and the vibrator control channel includes one inverter communication interface or one analog output port. In some embodiments, the anti-clogging device control unit and the multi-point sampling co-controller communicate via a controller area network bus, and the anti-clogging control command units are transmitted on the bus in a standard frame format. It can be understood that after receiving the anti-clogging control command unit, the anti-clogging device control unit immediately parses the pulse airflow pressure value, the guide vane angle value, and the vibrator frequency value.

[0046] In practical implementation, for the pulse airflow valve control channel, the received pulse airflow pressure parameters are converted into corresponding air pressure regulating valve opening control signals and sent to the target pulse airflow valve, causing it to inject airflow at a specified pressure before and after the sampling action. Specifically, the unit of the pulse airflow pressure parameters is megapascals, and the air pressure regulating valve opening control signal uses a 4 to 20 mA current signal. The conversion relationship satisfies the linear mapping formula: ; in: This indicates the output current control signal. Indicates the minimum output current. Indicates the maximum output current. This indicates the received pulse airflow pressure parameter. This indicates the minimum settable value of the pulse airflow pressure. This indicates the maximum settable value of the pulsed airflow pressure. For the guide vane actuator control channel, the received guide vane angle parameters are converted into corresponding motor rotation angle or hydraulic cylinder stroke control signals and sent to the target guide vane actuator to adjust the guide vane to the specified angle. The unit of the guide vane angle parameters is degrees, and the motor rotation angle control signal uses a 0-10 volt voltage signal, with each volt corresponding to 9 degrees. For the vibrator control channel, the received vibrator frequency parameters are converted into corresponding inverter output frequency control signals and sent to the target vibrator to vibrate at the specified frequency. The unit of the vibrator frequency parameters is Hertz, and the inverter output frequency control signal uses the frequency setting register value in the serial communication protocol (modular bus communication protocol).

[0047] In practical implementation, after receiving a start command, the sampling execution unit drives its sampling robotic arm or sampling shovel to move to the optimal sampling point coordinates, executes a complete sampling action, and obtains a material sample at that point. Specifically, the sampling execution unit includes a multi-degree-of-freedom robotic arm and a sampling shovel installed at the end of the robotic arm. The sampling shovel is made of wear-resistant steel and has a serrated edge machined at the shovel mouth. The motion trajectory of the sampling robotic arm or sampling shovel is calculated in real time by the built-in motion controller based on the target sampling point coordinates. The motion controller uses a Cartesian linear interpolation algorithm to ensure that the sampling shovel reaches the target position along the shortest path. Optionally, the complete sampling action includes the following steps: the sampling shovel accelerates downward from the initial standby position to a preset safe distance above the belt surface, cuts into the material flow at a constant speed, stays in the material flow for a preset holding time, and then rises and retracts to the initial standby position at a relatively fast speed. During the retraction process, the movable bottom plate inside the sampling shovel automatically opens, allowing the material sample to fall into the collection container below. It can be understood that the depth to which the sampling shovel cuts into the material flow is determined by the sampling shovel insertion depth parameter carried in the sampling action start command. In some embodiments, while performing a sampling action, the sampling execution unit sends a sampling action execution status feedback signal to the multi-point sampling coordination controller. The feedback signal includes a sampling start timestamp, a sampling end timestamp, and a flag indicating whether the sampling was successful or failed.

[0048] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An automatic multi-point sampling system for anti-clogging belt conveyors, characterized in that, The system includes: The parameter acquisition module acquires the belt conveyor operating status parameters and material characteristic parameters. The belt conveyor operating status parameters include the real-time belt speed, real-time load, and vibration spectrum. The material characteristic parameters include the estimated particle size distribution and moisture content of the currently conveyed material. The sampling decision module processes the belt conveyor's operating status parameters using an improved adaptive sampling point decision algorithm, calculating multiple optimal sampling points and corresponding sampling times along the belt length direction. The improved adaptive sampling point decision algorithm optimizes the sampling decision based on the belt conveyor's material flow uniformity model and vibration energy spectrum characteristics. The anti-clogging decision module, based on material characteristic parameters, calls the anti-clogging decision model to generate anti-clogging control parameters corresponding to each optimal sampling point. The anti-clogging decision model contains a material clogging tendency assessment sub-network, which outputs a quantified value of clogging risk level based on the combination of the estimated particle size distribution and moisture content of the currently conveyed material. The collaborative control module sends the multiple optimal sampling points, their corresponding sampling times, and the anti-blocking control parameters to the multi-point sampling collaborative controller. The collaborative execution module coordinates multiple distributed sampling execution units through the multi-point sampling collaborative controller to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters, so as to obtain representative samples within the multi-point sampling cycle and avoid sampling point blockage. The improved adaptive sampling point decision algorithm aims to maximize the representativeness of material distribution uniformity and is constrained by the stability of vibration energy spectrum characteristics. It searches for a set of multiple sampling points in a two-dimensional space defined by the belt length direction and the time dimension. This set ensures that the sampling points are spatially dispersed, temporally uniformly distributed, and that the vibration energy spectrum characteristics corresponding to each sampling point are within a stable range, including: Define belt length and time as two-dimensional decision variables; An objective function is established, which is calculated on a selected set of sampling points to determine the degree to which the collected samples represent the overall material distribution, i.e., the representativeness of the material distribution uniformity. for: ； in, Represents the set of sampling points The total number of sampling points in the middle, Indicates the first The weighting coefficients for each sampling point Indicates the first The mutual information between the material distribution uniformity corresponding to each sampling point and the overall material distribution uniformity of the belt; Establish constraints that require the vibrational energy spectrum characteristics of each selected sampling point at the corresponding sampling time to fall within a preset stationary range; The hybrid particle swarm optimization-simulated annealing optimization algorithm is adopted as the heuristic optimization algorithm. Under the constraints, the set of sampling points that maximizes the objective function value is searched among all sampling point sets, and this set is used as the final set of multiple optimal sampling points and corresponding sampling times. The particle swarm optimization algorithm is used for global search, and the simulated annealing algorithm is used for local perturbation to avoid getting trapped in local optima. During the initialization of the hybrid particle swarm-simulated annealing optimization algorithm, several sets of sampling points are randomly generated as particles in the particle swarm, and each particle corresponds to a candidate set of sampling points. The material flow rate across the belt cross-section per unit time is calculated using the real-time belt speed and real-time load. The material flow rate is calculated using the following formula: ； Where Q represents the material flow rate across the cross-section of the belt per unit time, v represents the real-time speed of the belt, and λ represents the real-time load of the belt, i.e. the mass of material carried per unit length of the belt. The material flow uniformity model of the belt conveyor is constructed by combining the material flow rate with historical data. The material flow uniformity model of the belt conveyor predicts the uniformity of material distribution on the belt in the near future based on the time series prediction method. The uniformity of distribution is characterized by the mass variation coefficient of the material along the length of the belt. The historical data includes the real-time speed sequence of the belt, the real-time load sequence of the belt, and the corresponding measured values ​​of material distribution recorded in several past sampling periods.

2. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 1, characterized in that, An improved adaptive sampling point decision algorithm is applied to the belt conveyor's operating state parameters to calculate multiple optimal sampling points and corresponding sampling times along the belt length direction, including: Using the real-time speed and real-time load of the belt, the material flow rate passing through the cross-section of the belt per unit time is calculated, and combined with historical data, a material flow uniformity model of the belt conveyor is constructed to predict the uniformity of material distribution on the belt in the near future. Frequency domain analysis is performed on the vibration spectrum to identify the main frequency band energy related to the stability of material flow. The vibration energy spectrum characteristics are characterized by the stability of the main frequency band energy. The improved adaptive sampling point decision algorithm outputs the coordinates of multiple optimal sampling points that satisfy the above conditions along the belt length direction, as well as the belt running time corresponding to each optimal sampling point. This belt running time is the sampling time corresponding to the optimal sampling point.

3. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 1, characterized in that, Based on material characteristic parameters, the anti-clogging decision model is invoked to generate anti-clogging control parameters corresponding to each optimal sampling point, including: The anti-clogging control parameters include pulse airflow pressure, guide vane angle, and vibrator frequency. The estimated particle size distribution and moisture content of the currently conveyed material are input into the pre-trained anti-clogging decision model; Based on the quantified value of the blockage risk level, a preset anti-blockage control parameter mapping table is queried. The anti-blockage control parameter mapping table defines the recommended pulse airflow pressure range, guide vane angle range, and vibrator frequency range for different blockage risk levels. Within the range of recommended parameters obtained from the query, and combined with the local information of the optimal sampling point, the specific pulse airflow pressure, guide vane angle, and vibrator frequency are calculated as the anti-clogging control parameters. The local information includes the belt conveyor structure gap size at the sampling point, the relative distance between the sampling execution unit and the belt surface, and the historical sampling frequency.

4. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 3, characterized in that, The material blockage tendency assessment subnetwork outputs a blockage risk level quantification value based on the estimated particle size distribution and moisture content of the currently conveyed material, including: The material blockage tendency assessment subnetwork is a deep neural network. Its input layer receives the feature vector encoded with the estimated particle size distribution and moisture content of the currently conveyed material as input features. The deep neural network performs nonlinear transformation on the input features through multiple hidden layers to learn the characteristics related to the adhesion, flowability and clogging risk of materials under the combined effect of particle size distribution and moisture content. The output layer of the deep neural network maps the learned features into a continuous numerical value. After standardization, this numerical value falls within a preset range of quantified values ​​for congestion risk level, which is then used as the quantified value for congestion risk level.

5. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 4, characterized in that, The deep neural network is a long short-term memory network that includes an attention mechanism.

6. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 1, characterized in that, Sending multiple optimal sampling points, their corresponding sampling times, and the anti-blocking control parameters to the multi-point sampling collaborative controller includes: Each optimal sampling point and its corresponding sampling time are encapsulated into a sampling task instruction unit. Each sampling task instruction unit includes the target location coordinates, the trigger timestamp, and the associated sampling execution unit identifier. Each optimal sampling point is encapsulated into an anti-blocking control instruction unit, and logically bound to the sampling task instruction unit corresponding to the same sampling point. All the encapsulated and bound instruction units are arranged into a unified control instruction sequence according to the order of sampling time, and then sent to the multi-point sampling collaborative controller.

7. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 6, characterized in that, The multi-point sampling coordination controller coordinates multiple distributed sampling execution units to perform sampling actions at the optimal sampling point at the corresponding sampling time, and synchronously activates anti-blocking measures corresponding to the anti-blocking control parameters, including: The multi-point sampling cooperative controller internally maintains a global clock and a state list of all sampling execution units; When the global clock reaches the trigger timestamp specified in a certain sampling task instruction unit, the multi-point sampling coordination controller sends a sampling action start command to the sampling execution unit specified in the sampling task instruction unit; Simultaneously, the multi-point sampling collaborative controller sends the anti-blocking control command unit to the anti-blocking device control unit associated with the current optimal sampling point. The anti-blocking device control unit drives the corresponding pulse airflow valve, guide plate actuator and vibrator to perform actions based on the received pulse airflow pressure, guide plate angle and vibrator frequency. Upon receiving a start command, the sampling execution unit drives the sampling robotic arm or sampling shovel to move to the optimal sampling point coordinates, performs a complete sampling action, and obtains a material sample at that point.

8. The anti-clogging automatic multi-point sampling system for belt conveyors according to claim 7, characterized in that, The anti-clogging device control unit drives the corresponding pulse airflow valve, guide plate actuator, and vibrator to operate based on the received pulse airflow pressure, guide plate angle, and vibrator frequency, including: The anti-clogging device control unit includes three independent control channels, which are used to control the pulse airflow valve, the guide plate actuator, and the vibrator, respectively. For the control channel of the pulse airflow valve, the received pulse airflow pressure parameters are converted into corresponding air pressure regulating valve opening control signals and sent to the target pulse airflow valve so that it sprays airflow at a specified pressure before and after the sampling action is executed; For the control channel of the guide vane actuator, the received guide vane angle parameters are converted into corresponding motor rotation angle or hydraulic cylinder stroke control signals and sent to the target guide vane actuator so that it can adjust the guide vane to the specified angle; For the control channel of the vibrator, the received vibrator frequency parameters are converted into the corresponding frequency control signal of the inverter output and sent to the target vibrator to make it vibrate at the specified frequency.

Images