A method and system for intelligent regulation of the breaking strength of a blasting bead

Abstract

The present application relates to the technical field of intelligent control, in particular to a kind of burst ball break strength intelligent control method and system, by tension connection clustering, state evolution modeling, abnormal disturbance collection and non-homogeneous region identification, realize burst ball node strength grade division and control task allocation, output accurate break strength control result.In the present application, by grouping clustering analysis to the direction vector of burst ball shell node tension, tension size and connection duration, the directional consistency feature in tension conduction structure is accurately revealed, which provides a basis support for subsequent identification of the tension variation law between nodes.With the help of time series construction and hidden Markov model, the transition process of node tension state is analyzed, the system captures the main path of tension evolution and transition probability, realizes the trend identification and prediction ability of tension state change.Further, by detecting the difference of tension change rate, abnormal section is constructed, the sensitivity and accuracy of node structure stability judgment are strengthened.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a method and system for intelligently controlling the bursting intensity of bouncy beads. Background Technology

[0002] The field of intelligent control technology encompasses the overall control process of using sensors, actuators, and control algorithms to perceive, judge, and adjust the system state in real time, achieving precise management of the operating state of complex systems through automation. The core content of this field includes sensor information acquisition, control logic setting, execution behavior adjustment, and feedback mechanism establishment.

[0003] Among them, the intelligent control method for the burst strength of the capsule refers to a method for identifying and adjusting the burst strength of the capsule contained in tobacco products during the smoking process. It obtains the suction data during the current use through the suction resistance detection method, establishes its strength response model based on the mechanical parameters of the capsule, and then uses the preset critical suction parameters to judge the strength and output control commands in real time.

[0004] Current technologies primarily rely on suction resistance detection, which only indirectly identifies pod breakage behavior at the macroscopic suction force data level. They lack fine-grained analysis of tension transmission mechanisms and node state evolution at the microscopic structural level, leading to delays and misjudgments in identifying breakage strength. The absence of time-series modeling and structural state transition probability assessment makes it impossible to predict potential anomalies in tension evolution. Non-homogeneous structures in the production process are difficult to distinguish, resulting in blind spots in spatial selection of control strategies and often leading to the generalized application of control commands that fail to adapt to structurally differentiated needs. Furthermore, current technologies fail to establish a dynamic linkage between disturbance response behavior and structural stress evolution, making control judgments highly dependent on a single suction threshold setting and lacking precise response capabilities to individual node differences. For example, in practical applications, if some nodes in a pod structure exhibit weakened tension transmission capabilities, current technologies, under uniform suction conditions, can easily lead to premature breakage or failure, affecting breakage consistency and production yield. These shortcomings, coupled with the increasing demand for high-precision control, are gradually revealing bottlenecks such as slow response, coarse identification, and lagging control. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose a method and system for intelligent control of the bursting strength of boulders.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for intelligent control of the bursting intensity of explosive beads, comprising the following steps:

[0007] S1: Obtain the tension connection parameter set between nodes of the popping bead shell, and perform tension direction consistency clustering on the tension connection parameter set to generate a tension connection clustering feature set;

[0008] S2: Input the tension connection clustering feature set into the hidden Markov model to analyze the tension state transition of nodes of the same type of popping bead shell and construct the tension evolution path sequence;

[0009] S3: Filter the abnormal tension fluctuation sections of the popping bead shell nodes in the tension evolution path sequence, load suction disturbance signals onto the abnormal sections, perform response behavior acquisition, and obtain the disturbance response behavior dataset;

[0010] S4: Perform cross-analysis on the disturbance response behavior dataset and the tension evolution path sequence to identify the non-homogeneous bursting bead shell nodes and form a structural non-homogeneous region identification dataset;

[0011] S5: Based on the non-homogeneous region identifier dataset, the burst intensity control level of the capsule is divided, and the control task allocation is performed according to the control level to obtain the burst intensity control result of the capsule.

[0012] As a further aspect of the present invention, the tension connection clustering feature set includes tension direction distribution information, connection stability grouping information, and the topological structure of the relationship between the bursting bead shell nodes; the tension evolution path sequence includes tension transfer stage sequence, state change probability distribution, and path continuity characteristics of the bursting bead shell nodes; the disturbance response behavior dataset includes displacement response peak information, response duration characteristics, and disturbance action direction records; the structural non-homogeneous region identification dataset includes abnormal bursting bead shell node numbers, local tension distribution deviation characteristics, and path stability disturbance markers; and the bursting bead rupture intensity control result includes control task type, execution operation instructions, and target region identifier.

[0013] As a further aspect of the present invention, the step of obtaining the tension connectivity clustering feature set specifically includes:

[0014] S111: Obtain the nodal tension direction vector, tension magnitude and connection duration in the nodal region of the popping bead shell using a fiber optic strain sensor to obtain the original set of tension parameters;

[0015] S112: Based on the tension direction vector in the original tension parameter set, calculate the cosine angle of the tension direction between the nodes of the bursting bead shell, determine the consistency of the direction, and group them according to the principle of similar vector angles to obtain the tension direction relationship group;

[0016] S113: Extract the tension direction relationship group, extract the tension direction distribution information, connection duration and tension magnitude within each group, and construct a tension connection cluster feature set.

[0017] As a further aspect of the present invention, the step of obtaining the tension evolution path sequence specifically includes:

[0018] S211: Arrange and combine the tension connection clustering feature set in chronological order, set the observation state and hidden state code corresponding to each popping bead shell node, and generate a time series of tension state of popping bead shell node.

[0019] S212: Based on the tension state sequence of the popping bead shell node, a hidden Markov model is used to compare the tension change and state duration between each state, and the transition frequency and time ratio of the state pairs are statistically analyzed to obtain the tension state transition probability.

[0020] S213: Based on the tension state transition probability, select the transition path with the highest probability in each state sequence as the main path, extract the tension direction change value sequence of continuous states in the path, and record the corresponding offset direction change trend according to the node number of the burst bead shell to obtain the tension evolution path sequence.

[0021] As a further aspect of the present invention, the steps for obtaining the disturbance response behavior dataset are specifically as follows:

[0022] S311: Based on the tension evolution path sequence, detect the continuous tension change rate of the bursting bead shell node, calculate the tension change rate difference between two consecutive time nodes, and determine whether there is a sequence interval in the continuous difference that is greater than the tension change rate abrupt change benchmark value, and integrate it into the tension fluctuation abnormal segment.

[0023] S312: Based on the abnormal tension fluctuation section, apply pulsed negative pressure disturbance loading to the corresponding popping bead shell node position, record the time difference between the disturbance loading time and the node response start time, and obtain the node response time sequence;

[0024] S313: Collect the maximum displacement response value, displacement direction change path and displacement fluctuation amplitude of the corresponding node in the node response time series to obtain the disturbance response behavior dataset.

[0025] As a further aspect of the present invention, the steps for obtaining the structurally heterogeneous region identifier dataset are as follows:

[0026] S411: Match the perturbation response behavior dataset with the number of the bursting bead shell node in the tension evolution path sequence, extract the timestamp information of the two types of data under the same number, calculate the time difference between the displacement response duration and the tension transfer cycle, and obtain the node response and transfer difference value.

[0027] S412: Based on the node response and transfer difference value, the density distribution of the nodes of the bursting bead shell is judged, the standard deviation of the response difference value between each node and the adjacent node in the corresponding tension path is calculated, and the anomaly identification threshold is set according to the density distribution outlier rate judgment standard to obtain density outlier abnormal nodes.

[0028] S413: Based on the density outlier nodes, extract the outlier node number, corresponding path segment number and location sequence number, summarize the distribution of outlier nodes by grouping them according to the shell structure region of the popping bead, mark the non-homogenization trend index in each region, and establish a structural non-homogenization region identification dataset.

[0029] As a further aspect of the present invention, the step of obtaining the burst strength control result of the explosive beads is specifically as follows:

[0030] S511: Based on the non-homogeneous region identifier dataset of the structure, extract two indicators, tension direction offset amplitude and displacement response amplitude, of the popping bead shell node and take a weighted average to obtain the joint score of the popping bead shell node;

[0031] S512: Based on the joint score of the bursting bead shell nodes, calculate the average and standard deviation of all score values, divide the level labels according to the numerical interval, and obtain the bursting bead strength level distribution results.

[0032] S513: Based on the distribution results of the bursting intensity level of the bursting beads, combine and organize the bursting bead shell node number and level label corresponding to each level, establish a task instruction mapping, and output the bursting intensity control result.

[0033] A smart control system for the burst intensity of explosive beads, the system being used to execute the above-mentioned smart control method for the burst intensity of explosive beads, the system comprising:

[0034] The tension parameter clustering module obtains the tension connection parameter set between nodes of the popping bead shell, and performs tension direction consistency clustering processing on the tension connection parameter set to generate a tension connection clustering feature set;

[0035] The tension evolution modeling module inputs the tension connection clustering feature set into the hidden Markov model to analyze the tension state transition of nodes of the same type of popping bead spherical shell and constructs a tension evolution path sequence.

[0036] The disturbance acquisition and response module filters the abnormal tension fluctuation sections of the popping bead shell nodes in the tension evolution path sequence, applies a suction disturbance signal to the abnormal sections, and performs response behavior acquisition to obtain a disturbance response behavior dataset.

[0037] The non-homogeneity identification module performs cross-analysis on the disturbance response behavior dataset and the tension evolution path sequence to identify non-homogeneous bursting bead shell nodes and form a structural non-homogeneous region identification dataset.

[0038] The intensity control grading module divides the intensity control levels of the burst beads based on the non-homogeneous region identifier dataset, performs control task allocation according to the control level, and obtains the burst bead intensity control result.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0040] In this invention, by grouping and clustering the tension direction vector, tension magnitude, and connection duration of the nodes in the bursting bead spherical shell, the directional consistency characteristics in the tension transmission structure are accurately revealed, providing a foundation for subsequent identification of tension change patterns between nodes. Using time series analysis and Hidden Markov Models, the transition process of node tension states is analyzed, and the system captures the main path of tension evolution and the transition probability, achieving the ability to identify and predict trends in tension state changes. Furthermore, abnormal sections are constructed by detecting differences in tension change rates, and pulsed negative pressure disturbance signals are applied while collecting node response behavior. A linkage response chain is established between the disturbance and the natural state, enhancing the sensitivity and accuracy of node structural stability judgment. Through cross-analysis of response amplitude and tension offset direction, non-homogeneous nodes are identified and structurally heterogeneous regions are delineated, completing the spatial localization of abnormal trends at the structural level. Finally, by combining node scoring and grading mechanisms, the control task is precisely mapped to the target node region, effectively promoting the refinement, grading, and automation of fracture strength regulation. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0042] Figure 2 This is a flowchart of step S1 of the present invention;

[0043] Figure 3 This is a flowchart of step S2 of the present invention;

[0044] Figure 4 This is a flowchart of step S3 of the present invention;

[0045] Figure 5 This is a flowchart of step S4 of the present invention;

[0046] Figure 6 This is a flowchart of step S5 of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0048] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0049] Please see Figure 1 This invention provides a technical solution: a method for intelligently controlling the bursting strength of bouncy beads, comprising the following steps:

[0050] S1: Obtain the tension connection parameter set between nodes of the popping bead shell, and perform tension direction consistency clustering on the tension connection parameter set to generate a tension connection clustering feature set;

[0051] S2: Input the tension connection clustering feature set into the hidden Markov model to analyze the tension state transition of nodes of the same type of popping bead spherical shell and construct the tension evolution path sequence;

[0052] S3: Filter the abnormal tension fluctuation sections of the popping bead shell nodes in the tension evolution path sequence, apply suction disturbance signals to the abnormal sections and perform response behavior acquisition to obtain the disturbance response behavior dataset;

[0053] S4: Cross-analysis of the disturbance response behavior dataset and tension evolution path sequence is performed to identify non-homogeneous bursting bead shell nodes and form a dataset identifying structural non-homogeneous regions.

[0054] S5: Based on the structurally heterogeneous region identifier dataset, the burst intensity control levels of the capsule are divided, and the control task allocation is performed according to the control level to obtain the burst intensity control results.

[0055] The tension connection clustering feature set includes tension direction distribution information, connection stability grouping information, and topological structure of the node relationship between the bursting bead and the shell. The tension evolution path sequence includes tension transfer stage sequence, state change probability distribution, and path continuity characteristics of the bursting bead and the shell. The disturbance response behavior dataset includes displacement response peak information, response duration characteristics, and disturbance direction records. The structural heterogeneity region identification dataset includes abnormal bursting bead and shell node numbers, local tension distribution deviation characteristics, and path stability disturbance markers. The bursting bead rupture intensity control results include control task type, execution operation instructions, and target region identification.

[0056] Please see Figure 2 The specific steps for obtaining the tensile connectivity clustering feature set are as follows:

[0057] S111: Obtain the nodal tension direction vector, tension magnitude and connection duration in the nodal region of the popping bead shell using a fiber optic strain sensor to obtain the original set of tension parameters;

[0058] First, a fiber optic strain sensor array is pre-positioned on the burst bead structure. Each sensor is precisely attached to a structural node, recording the node strain by measuring changes in laser wavelength reflection. The acquired fiber optic reflection spectrum is converted into micro-strain data using a demodulator. Then, based on the micro-strain at the node and the Young's modulus and Poisson's ratio of the shell material, the tension value is calculated using the stress-strain relationship. The tension direction vector is determined by the position difference of the sensor response points in the three-dimensional coordinate change. For example, if nodes A and B show positive and negative strain changes along the x-axis respectively, the tension direction is along the x-axis, and the tension magnitude is the difference in strain between the two points multiplied by the modulus. This method is used to obtain the direction vector and magnitude between multiple nodes. Simultaneously, the initial change time and the final stabilization time are recorded; the time difference between the two times is the connection duration. In implementation, if the burst bead diameter is 120mm, the shell thickness is 2mm, the fiber optic grating sensing range is ±2000με, the initial change in the reflected wavelength of a certain node is 1.5nm, corresponding to a strain of 1500με, and the material modulus is 70GPa, the tension magnitude is:

[0059] σ=E·∈=70×10 9 ×1500×10 -6 =105×10 6 Pa;

[0060] Where σ: tension at the node (in Pa), E: Young's modulus of the material (in Pa, here taken as 70 GPa), ∈: strain measured by the fiber grating (in με, here 1500 με = 1.5 × 10⁻⁶). -3 ), the direction vector is set to The connection duration ranges from t=2.1s to t=3.9s, a total of 1.8s. The above process is repeated at multiple nodes to obtain the original parameter set.

[0061] S112: Based on the tension direction vector in the original tension parameter set, calculate the cosine angle of the tension direction between the nodes of the bursting bead shell, determine the consistency of the direction, and group them according to the principle of similar vector angles to obtain the tension direction relationship group;

[0062] Based on the obtained tension direction vectors of each node, the tension vector of each node is first standardized into a unit vector. Then, the directions of each pair of nodes are compared, specifically using the cosine angle calculation method, that is, assuming the unit vectors of the tension directions of two nodes are... and Calculate the cosine of the included angle:

[0063]

[0064] Where θ: the angle between the vectors, in radians or degrees. The unit tension direction vectors at the two nodes. The dot product of two vectors is represented by the formula a. x b x +a y b y +a z b z , The vector magnitude, since it has been standardized to unit vectors and is always 1, has a cosine value equal to the dot product of the two vectors. If the cosine value is ≥0.95 (i.e., the included angle is ≤18.19°), the directions are considered approximately consistent. A grouping threshold of 18° is set for the included angles, and within each group, no two included angles exceed this threshold. For example: Suppose the unit vectors of the tension directions at three nodes are... but Those that meet the requirements should be grouped together. The included angle reaches 24.2°, which does not meet the grouping criteria. The final output is as follows: Tension direction relationship group is This completes the tension-direction polymerization and grouping process.

[0065] S113: Extract tension direction relationship groups, extract tension direction distribution information, connection duration and tension magnitude within each group, and construct tension connection cluster feature set;

[0066] After obtaining the tension direction relationship groups, feature extraction is performed on the member nodes within each group. First, the distribution direction of each vector in three-dimensional space is extracted to form a statistical distribution matrix, listing the average value and standard deviation of the direction vector components. Then, the average value is extracted and the maximum and minimum values ​​are filtered in conjunction with the tension magnitude corresponding to each node. Next, the occurrence and duration of tension at each node are recorded according to time to construct a time span sequence. For example, in group 1, there are 3 nodes with tension magnitudes of σ1 = 105 MPa, σ2 = 110 MPa, and σ3 = 98 MPa, and the direction vectors are... Given durations t1 = 1.8s, t2 = 2.1s, and t3 = 1.5s, the statistical mean vector of the tension direction distribution is:

[0067]

[0068] The average tension is

[0069] The average duration is

[0070] The constructed result is that the tensile connectivity cluster feature set is {direction mean = (0.96, 0.223, 0.02), tension mean = 104.33 MPa, time mean = 1.8 s}.

[0071] Please see Figure 3 The specific steps for obtaining the tension evolution path sequence are as follows:

[0072] S211: Arrange and combine the tension connection cluster feature set in chronological order, set the observation state and hidden state code corresponding to each popping bead shell node, and generate the tension state time series of popping bead shell node.

[0073] First, extract the mean value of the tension direction vector, the mean value of the tension magnitude, and the mean value of the tension duration for each popping bead shell node constructed in the preprocessing stage, and record the corresponding time t. i Sort the data in ascending order and reassemble the time series. Then, assign a state code to each sorted node. The state code consists of observed state and latent state. Observed states are classified according to the numerical range of tension magnitude: tension between 0 and 90 MPa is designated as "A", tension between 90 and 110 MPa as "B", and tension greater than 110 MPa as "C". Latent states are determined based on the maximum projection dimension of the principal component of the tension direction: if the x-component is the largest, it is marked as "X", the y-component as "Y", and the z-component as "Z". This process is used to classify a node as a direction vector. For the implicit state “X”, if the tension is 108MPa, the corresponding observed state is “B”, and the encoding combination is “BX”. Repeat this method to generate the state encoding sequence in sequence. For example, after arranging according to time t = {2.1, 4.5, 6.7}, the state sequence {AX, BX, CY} is obtained, which completes the serialization representation of the tension state of each node.

[0074] S212: Based on the tension state sequence of the popping bead shell node, a hidden Markov model is used to compare the tension change and state duration between each state, and the transition frequency and time ratio of the state pairs are statistically analyzed to obtain the tension state transition probability.

[0075] Based on the constructed node tension state time series, extract any two consecutive state pairs s i →s j For each state pair, the evolutionary behavior of the execution tension is analyzed, and the tension magnitude σ is extracted respectively. i σ j With the start and end time t of the state i t j The tension change Δσ was calculated. i,j =σ j -σ i With state duration Δt i,j =t j -t i We construct the state transition representation and establish a hidden Markov transition probability model, using the following improved formula for probability modeling:

[0076]

[0077] The parameters are explained as follows: P i,j : indicates from state s i Transition to state s j The transition probability (dimensionless) is used to measure the likelihood that a state pair will be an actual transition path in the overall state sequence; σ i : state s i The corresponding nodal tension value, in MPa (megapascals), is usually derived from the mean tension value of the tension cluster feature set at the node; σ j : state s j The corresponding nodal tension value, in MPa; t i : state s i The starting time point, measured in seconds (s), is typically derived from the moment when the node states in the tension state time series are recorded; t j : state s j The starting time point, in seconds (s); Δt i,j =t j -t i: Indicates the state from s i Transfer to s j The time taken is in seconds (s). The tension change rate term, in MPa / s, is used to measure the intensity of the tension change per unit time between states; State transition rate, in seconds. -1 γ1: Tension change rate penalty factor, in seconds per megapascal (s / MPa), used to adjust the degree of suppression of transition probability by drastic tension changes. It is usually set in the range of 0.02–0.1 s / MPa, specifically based on the tension response bandwidth of the exploding spherical shell in the experiment. For example, if the structural strain stability range is ±15MPa, the expected maximum change rate is no more than 10MPa / s, so the value is set to no more than 0.1. γ2: State transition penalty factor, in seconds (s), used to control the influence of frequent state jumps in a short time. It is generally set in the range of 0.05–0.2 s, based on the shortest stable period of the node response in actual testing. If the shortest node response period is 1 second, states with changes less than this threshold are given stronger attenuation to avoid misjudgment. Z: Normalization factor, in dimensionless form, represents the sum of the unnormalized probabilities of all possible states, such that the final transition probability of each state satisfies ∑ j P i,j =1, and its value is calculated from all state pairs and then normalized.

[0078] Let state pair s i =BX,s j =CX, and their nodal tension values ​​are σ i =100MPa, σ j =112MPa, time point t i =4.5s, t j = 7.0s, the time difference is Δt i,j =2.5s, set the penalty factors γ1 = 0.05s / MPa, γ2 = 0.1s, then the calculations are as follows:

[0079] Tension change rate term:

[0080] State transition rate item:

[0081] Expanding the exponent:

[0082] Calculate the index value:

[0083] Let the sum of the unnormalized weights of all current state pairs be Z = 2.25, then after normalization: The results indicate that when the state transition time is moderate (2.5 s) and the tension change is not drastic (12 MPa), the transition probability remains at a moderately high level and can be used as a reference for main path selection. This probability will be ranked together with other state pairs for path construction and trend identification. The setting logic of parameters γ1 and γ2 is highly consistent with the experimental period and tension change characteristics, avoiding over-response to non-physical abrupt changes, while maintaining the repeatability and dimensional integrity of the model prediction results.

[0084] S213: Based on the tension state transition probability, select the transition path with the highest probability in each state sequence as the main path, extract the tension direction change value sequence of continuous states in the path, and record the corresponding offset direction change trend according to the node number of the popping bead shell to obtain the tension evolution path sequence.

[0085] Based on all the states obtained in the previous stage, P i,j In the entire node sequence, the maximum connection path is extracted according to probability as the main path. Let the state sequence of the main path be {BX, CX, CY, BZ}. Extract the sequence of tension direction changes between consecutive states from it. For each adjacent state pair s i →s i+1 Extract its tension direction vector Calculate the change in direction Determine the positive or negative value of each dimensional component to identify the main offset direction, such as assuming... but This indicates that the tension direction shift tends towards +x, -y, +z, and so on. The directional changes of each segment on the main path are calculated sequentially, and the shift trend is recorded using the node number as an index. For example, if the node number is #03, its direction evolution is x↑, y↓, z↑. The final tension evolution path sequence is constructed for subsequent anomaly detection and path reproduction. The complete sequence format is: Node Number → State Sequence → Tension Change → Direction Shift Trend, for example, "03→BX→CX→CY→BZ→Tension Increase → Direction X↑Y↓Z↑".

[0086] Please see Figure 4 The specific steps for obtaining the disturbance response behavior dataset are as follows:

[0087] S311: Based on the tension evolution path sequence, detect the continuous tension change rate of the popping bead shell node, calculate the tension change rate difference between two specified consecutive time nodes, and determine whether there is a sequence interval in the continuous difference that is greater than the tension change rate abrupt change benchmark value, and integrate it into the tension fluctuation abnormal segment.

[0088] Based on the time series data of each node in the tension evolution path, a continuous analysis of the tension response rate of each node in the bursting bead spherical shell structure is required. First, the tension value and timestamp of two adjacent time points in the node tension measurement data are extracted. The tension value at time k is denoted as "the kth tension measurement value" in megapascals (MPa), and the corresponding time point is denoted as "the kth tension measurement time" in seconds (s). The tension change rate of the kth segment is calculated as "the difference between two adjacent tension measurements divided by the corresponding time difference," with the result in megapascals per second (MPa / s). For example, if the third tension measurement value is 91 MPa and the corresponding time is 4.5 seconds, and the fourth tension measurement value is 96 MPa and the corresponding time is 5.0 seconds, then the tension change rate is... The difference between adjacent rates is then taken to form the "tension change rate difference," the unit of which is megapascals per square second (MPa / s). 2 This difference is used to determine whether there is a sudden change in the tension response process. Continuing with the previous example, if the fifth tension measurement is 108 MPa and the time is 5.6 seconds, then the rate of change is... The rate difference is 20-10=10MPa / s 2 The value is compared with a preset benchmark value for the sudden change in tension rate, which is defined as "40% of the maximum steady-state tension rate". Assuming the maximum steady-state rate is 5 MPa / s, the benchmark value is 5 × 0.4 = 2.0 MPa / s. 2 Because the calculated value is 10 MPa / s 2 Higher than 2.0 MPa / s 2 Therefore, this time period is marked as an abnormal tension fluctuation zone and stored in the list of abnormal time periods.

[0089] S312: Based on the abnormal tension fluctuation section, apply pulsed negative pressure disturbance loading to the corresponding popping bead shell node position, record the time difference between the disturbance loading time and the node response start time, and obtain the node response time series;

[0090] Based on the identified abnormal tension fluctuation time periods, perturbation loading was applied to nodes located in the abnormal section of the spherical shell structure. The perturbation method was in the form of instantaneous negative pressure pulses, acting on the shell surface of the node area. The time point of this perturbation was called the "perturbation loading start time," denoted as a specific time value in seconds (s). The perturbation process was executed by a negative pressure generator, with a set negative pressure value of -50 kPa and a loading duration not exceeding 0.2 seconds. The perturbation parameters were derived from the local strain limit test range of the structural shell. Simultaneously with loading, sensors attached to the nodes continuously collected structural response information and detected... The moment when the displacement first changes is called the "node response start time". The difference between the two is defined as the "node response delay time", which is in seconds (s). For example, on node number 7, the disturbance time is 6.40 seconds and the node's first response time is 6.48 seconds. Then the response delay time is 6.48-6.40=0.08s. In this way, the loading time and response start time of all disturbed nodes are recorded separately, and then the response time difference of each node is calculated in a unified manner to form a set of node response time data, which reflects the actual response time of the node to external disturbances in the abnormal tension change range.

[0091] S313: Collect the maximum displacement response value, displacement direction change path and displacement fluctuation amplitude of the corresponding node in the node response time series to obtain the disturbance response behavior dataset;

[0092] For all nodes that underwent perturbation experiments and obtained response times, three-dimensional displacement data during the response phase was continuously collected. This displacement data included the spatial movement distance of the nodes at multiple time points. First, the "maximum response displacement value of the node" was extracted, which is the maximum displacement that occurred during the entire period from the perturbation loading to the end of the response, in millimeters (mm). This displacement was calculated by taking the maximum value after taking the spatial displacement magnitude recorded by the node sensors at each moment. Second, the continuity of the direction during the displacement change was analyzed. The difference between two adjacent displacement data points was calculated, and the coordinate axis (e.g., X-axis, Y-axis, Z-axis) to which the main direction of change belonged was determined from the difference. The direction of change was then represented by a sign (e.g., positive or negative), forming a "displacement" data structure. The displacement direction change path is recorded. For example, if the displacement direction components of a node at four time points increase sequentially along the positive x-axis and then increase along the y-axis, then its direction path is recorded as "x positive, x positive, y positive". Finally, the displacement fluctuation amplitude of the node throughout the entire response period is recorded, defined as "the difference between the maximum displacement value and the minimum displacement value", in millimeters. If a node records displacements of {0.18mm, 0.26mm, 0.24mm, 0.31mm}, then the maximum is 0.31mm, the minimum is 0.18mm, and the fluctuation amplitude is 0.31-0.18=0.13mm. The above three items—maximum response displacement, displacement direction change path, and displacement fluctuation amplitude—are uniformly constructed into a disturbance response characteristic data item.

[0093] Please see Figure 5 The specific steps for obtaining the dataset of structurally heterogeneous regions are as follows:

[0094] S411: Match the perturbation response behavior dataset with the node numbers of the bursting bead spherical shell in the tension evolution path sequence, extract the timestamp information of the two types of data under the same number, calculate the time difference between the displacement response duration and the tension transfer cycle, and obtain the node response and transfer difference value.

[0095] The node numbers recorded in the disturbance response behavior dataset are compared with the known node numbers in the tension evolution path sequence. First, the node number fields in both datasets are read sequentially, and node data pairs with the same number are identified through a matching operation. In each matched dataset, the disturbance loading start time and node response start time are extracted from the disturbance data; these are called "loading time" and "response time," respectively. The difference between these two values ​​yields the displacement response duration of the node, denoted as "response duration." Then, the start and end times of the first significant tension change at the node are extracted from the tension data, denoted as "tension start time." The difference between the "start time" and the "tension termination time" is defined as the "tension transfer cycle". The difference between these two durations is calculated to obtain the "response and transfer difference value". For example, if the disturbance loading time of node #08 is 6.10 seconds and the node response time is 6.35 seconds, then the response duration is 0.25 seconds. At the same time, its tension change period is from 6.12 seconds to 6.50 seconds, then the tension cycle is 0.38 seconds. The calculated difference value is 0.25-0.38=-0.13 seconds. After repeating the above operation to process all node data, a three-column data record containing node number, response duration, tension cycle and difference value can be constructed.

[0096] S412: Based on the difference between node response and transfer value, the density distribution of the nodes of the bursting bead shell is judged, the standard deviation of the response difference value between each node and the adjacent node in the corresponding tension path is calculated, and the anomaly identification threshold is set according to the density distribution outlier rate judgment standard to obtain density outlier abnormal nodes.

[0097] After obtaining the response and transfer difference values ​​corresponding to the node numbers, all nodes are first sorted in ascending order by number. Then, a local difference analysis window is constructed with each node as the center point. Two adjacent nodes are selected to form a data window of length 5. The standard deviation of the response difference values ​​is calculated sequentially. The specific steps are: calculate the average of the 5 difference values ​​in the window, then square the difference between each difference value and the average, sum them up, and divide by the number of nodes to obtain the variance. The square root of the variance is then taken to obtain the standard deviation. The standard deviation value of each node is recorded as a measure of the fluctuation of the response difference in the local tension path. Taking node number 25 as an example, its adjacent node numbers are {23, 24, 25, 26, 27}, and its corresponding "response and transfer difference values" are:

[0098] {x 23 ,x 24 ,x 25 ,x 26 ,x 27 = {0.028, 0.034, 0.032, 0.036, 0.029} seconds;

[0099] Calculate the average:

[0100] Calculate the square of the difference between each value and the mean:

[0101] Item 1: (0.028-0.0318) 2 =(-0.0038) 2 =1.444×10 -5 Second 2 ;

[0102] Item 2: (0.034-0.0318) 2 =(0.0022) 2 =4.84×10 -6 Second 2 ;

[0103] Item 3: (0.032-0.0318) 2 =(0.0002) 2 =4.0×10 -8 Second 2 ;

[0104] Item 4: (0.036-0.0318) 2 =(0.0042) 2 =1.764×10 -5 Second 2 ;

[0105] Item 5: (0.029-0.0318) 2=(-0.0028) 2 =7.84×10 -6 Second 2 .

[0106] Calculate the variance and take the square root to obtain the standard deviation:

[0107]

[0108] Assume the mean of the standard deviations for all nodes is μ. σ =0.0025 seconds, population standard deviation is: σ σ =0.001 seconds, set the anomaly detection threshold to the mean plus twice the standard deviation: threshold = μ σ +2·σ σ =0.0025 + 2·0.001 = 0.0045 seconds. If the local standard deviation of a certain node (such as #32) is σ 32 =0.0053 seconds. Since 0.0053 > 0.0045, node #32 is determined to be a density outlier.

[0109] Furthermore, the kernel density estimation method is used to determine the density distribution of each node in the response space based on the above differences, as shown in the following formula:

[0110]

[0111] Where: ρ i : Represents the density estimate of node i; h: Bandwidth parameter, controlling the smoothness of the kernel function, usually set to 1.06 times the sample standard deviation of the response difference; x i : The difference between the response and transition values ​​of the i-th node; x j : The difference between the response and transition of the j-th node in the sample; n: The total number of nodes involved in the density estimation; exp: The natural exponential function, representing the Gaussian kernel decay function; The normalization constant term of the normal distribution.

[0112] Let the difference value of the i=30th node be: x 30 =0.030 seconds, select the difference values ​​of the remaining 5 nodes in the sample: x j ={0.028,0.031,0.027,0.035,0.033}, assuming the standard deviation of the total sample difference is σ = 0.002 seconds, then the bandwidth is set as: h = 1.06·σ = 1.06·0.002 = 0.00212 seconds.

[0113] Calculate the Gaussian kernel function value term by term:

[0114] Item 1:

[0115] Item 2:

[0116] Item 3:

[0117] Item 4:

[0118] Item 5:

[0119] Summation: S = 0.640 + 0.895 + 0.367 + 0.062 + 0.367 = 2.331.

[0120] Substitute into the formula to calculate the density: This result indicates that the node is located in a high-density region within the response difference space. Conversely, if the density value is significantly lower than half the mean of its neighboring nodes, it is considered a low-density outlier. Combining this with the local standard deviation results forms a joint criterion for labeling outlier nodes. Local standard deviation analysis and density estimation analysis are complementary mechanisms; the two methods can label outliers independently or jointly. A node may satisfy only one of these criteria or both. Once the outlier criterion defined by either method is met, the node can be added to the density outlier node set.

[0121] S413: Based on density outlier nodes, extract outlier node numbers, corresponding path segment numbers and location numbers, summarize the distribution of outlier nodes by grouping them according to the popping bead shell structure region, mark the non-homogenization trend indicators in each region, and establish a structural non-homogenization region identification dataset.

[0122] First, for each node identified as anomalous, extract its corresponding path segment number from the original tension path record. This number reflects the path location to which the node's tension evolution belongs. For example, the tension path number corresponding to node #30 is path segment #12. Next, obtain the spatial position number of the node in the spherical shell structure model. This number is usually the local mesh number in the three-dimensional coordinate system or the sector number in the polar coordinate system. For example, node #30 is located in region "B3", which is the middle region of the third quadrant on the spherical shell model. Then, the three items "anomalous node number + path segment number + structural region number" are... The data is aggregated into structural localization triples, recorded in the form of (30, 12, B3). This process is performed on all anomalous nodes sequentially, forming a complete dataset of anomalous node structural distribution. Then, all anomalous nodes are grouped according to their structural region numbers, establishing a set of anomalous nodes within each region. The number of anomalous nodes in each region's set is counted, and the total number of nodes in that region is obtained. The percentage of anomalous nodes in that region is calculated as an indicator of the region's non-uniformity trend. For example, region B3 contains 10 nodes with node numbers 26–35, of which 4 are anomalous nodes; therefore, the non-uniformity index is... For example, if region C2 has 12 nodes, and only one is identified as an anomalous node, then the non-homogenization trend indicator is: After repeating the above calculation process for each structural region, a set of data is formed: "Region ID - Number of Abnormal Nodes - Total Number of Nodes in the Region - Non-uniformity Index". Finally, this dataset is used to establish structural non-uniformity region identification information, forming identification fields such as: Region C2, non-uniformity index 0.083, containing abnormal node #41; Region B3, non-uniformity index 0.40, containing abnormal nodes #30, #31, #32, and #34. This process ensures that the response consistency status of each structural region is traceable and quantifiable in the data.

[0123] Please see Figure 6 The specific steps for obtaining the burst strength control results of the menthol capsules are as follows:

[0124] S511: Based on the structurally heterogeneous region identification dataset, extract two indicators, tension direction offset amplitude and displacement response amplitude, of the popping bead shell node and take a weighted average to obtain the joint score of the popping bead shell node;

[0125] Based on the node numbers recorded in the structural heterogeneous region identification dataset, two indicators, tension direction offset amplitude and displacement response amplitude, are extracted for each node. Tension direction offset amplitude is obtained by comparing the change angle of the principal direction of the node's tension vector before and after perturbation loading. Displacement response amplitude is calculated by the difference between the maximum and minimum displacement modulus of the node throughout the perturbation response process, expressed in angle and millimeters, respectively. For example, node #15 has a horizontal tension direction before perturbation loading and deflects to a 60-degree direction after loading, resulting in a tension offset amplitude of 60 degrees. The maximum displacement of this node during displacement is 0.38 mm, and the minimum displacement is 0.14 mm, resulting in a displacement response amplitude of 0.24 mm. Next, weighted values ​​need to be assigned to these two indicators. The influence of tension direction offset and displacement response on bursting behavior needs to be balanced within the structural stress characteristics. Based on previous loading experiments, it was found that tension offset affects the local strength changes of the structure. The influence of the direction of tension offset accounts for more than 60%, while the influence of displacement response accounts for about 40%. Therefore, the weight of the direction of tension offset is set to 0.6, and the weight of displacement response amplitude is set to 0.4. This ratio is verified by multiple sets of experimental node data. Among them, when the direction of tension offset changes abruptly in experiments E17 to E26, the probability of structural instability increases by 23% to 28%, while the average increase when displacement amplitude increases is 12%. Based on this, the weight setting refers to this influence ratio relationship and approaches the integer ratio to obtain a unified weight scheme. Then, for each node, the operation of multiplying the tension offset by 0.6 and the displacement response by 0.4 is performed and accumulated. For example, for node #15, the tension offset is 60 degrees and the displacement response is 0.24 mm. Then, 60 × 0.6 = 36 and 0.24 × 0.4 = 0.096 are performed. After merging, the joint score of the node is 36.096. This operation is performed on all nodes in turn, and finally a set of joint score values ​​containing all non-homogenized nodes is generated.

[0126] S512: Based on the joint score of the bursting bead shell nodes, calculate the average and standard deviation of all score values, divide the level labels according to the numerical interval, and obtain the distribution results of the bursting bead strength level.

[0127] First, the arithmetic mean and standard deviation of all scores are calculated as basic distribution statistical parameters. The statistical mean is used to measure the central trend of the overall node response intensity, and the standard deviation is used to assess the overall dispersion of node response fluctuations. In this implementation, a node score set is set as {36.096, 21.44, 28.72, 19.66, 32.82, 25.40}, with an average of 27.69 and a calculated standard deviation of 6.45. Based on this, three level division intervals are set: Level I nodes are those with a joint score greater than the average plus one standard deviation, i.e., nodes with a score greater than 34.14; Level II nodes are those with a joint score greater than the average plus one standard deviation; and Level II nodes are those with a joint score greater than the average plus one standard deviation. For nodes with scores between 21.24 and 34.14, Level III consists of nodes with scores below 21.24. Substituting the scores of each node into the judgment criteria, node #15 has a score of 36.096, which is higher than 34.14, so it is classified as Level I. Node #07 has a score of 19.66, which is less than 21.24, so it is classified as Level III. Node #13 has a score of 28.72, which is in the middle range, so it is classified as Level II. All nodes undergo the above classification process and the node number, score value and its corresponding level label are recorded. Finally, a burst intensity level distribution result set containing three columns of data: "node number - score value - level" is formed.

[0128] S513: Based on the distribution results of the burst strength level of the burst beads, combine and organize the burst bead shell node number and level label corresponding to each level, establish a task instruction mapping, and output the burst strength control result of the burst beads.

[0129] All node numbers are organized by level, and all node numbers under each level are extracted and combined with the corresponding level labels to form an instruction mapping list. Node numbers under each level are arranged in ascending order to form task partitions, creating combined data items with the structure "level label: node number list". For example, level I nodes are {#15, #18}, level II nodes are {#09, #13, #22}, and level III nodes are {#07, #12}. These are sequentially written into the instruction table, and a preset task response strategy number is assigned according to the level label. Level I nodes are marked as "Stable Intervention Priority-A". The Level II node is labeled "Periodic Monitoring-B", and the Level III node is labeled "Low-Level Intervention-C". These labels are set through the configuration table in the preset strategy of the structural control system. The corresponding parameters are the response frequency, the number of pressurization adjustments, and the pre-inspection cycle. For example, the intervention cycle for Level I node is 1 day and the maximum number of pressurization adjustments is 3 times. The cycle for Level III node is set to 5 days and the number of pressurization adjustments is only 1 time. The node number, level label, task type, and control parameters are then uniformly organized and output to form a set of results for the bursting intensity control of the bead, thus completing the final node control mapping relationship.

[0130] A smart control system for the burst intensity of explosive beads, the system being used to execute the aforementioned smart control method for the burst intensity of explosive beads, the system comprising:

[0131] The tension parameter clustering module obtains the tension connection parameter set between nodes of the popping bead shell, and performs tension direction consistency clustering processing on the tension connection parameter set to generate a tension connection clustering feature set;

[0132] The tension evolution modeling module inputs the tension connection clustering feature set into the hidden Markov model to analyze the tension state transition of nodes of the same type of popping bead spherical shell and constructs the tension evolution path sequence.

[0133] The disturbance acquisition and response module filters the abnormal tension fluctuation sections of the popping bead shell nodes in the tension evolution path sequence, applies suction disturbance signals to the abnormal sections, performs response behavior acquisition, and obtains the disturbance response behavior dataset.

[0134] The non-homogeneity identification module performs cross-analysis on the disturbance response behavior dataset and the tension evolution path sequence to identify non-homogeneous bursting bead shell nodes and form a structural non-homogeneous region identification dataset.

[0135] The intensity regulation and grading module divides the intensity regulation levels of the burst beads based on the structurally heterogeneous region identifier dataset, performs control task allocation according to the regulation level, and obtains the burst bead intensity regulation results.

[0136] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for intelligent regulation of the breaking strength of a blasting bead, characterized in that, The method comprises the following steps: S1: Obtain the tension connection parameter set between the balloon shell nodes, and perform tension direction consistency clustering processing on the tension connection parameter set to generate a tension connection clustering feature set; S2: Input the tension connection clustering feature set into a hidden Markov model, analyze the tension state transition of balloon shell nodes of the same type, and construct a tension evolution path sequence; S3: Screen the tension fluctuation abnormal section of the balloon shell node in the tension evolution path sequence, load the response behavior data set of the abnormal section to the disturbance signal, and obtain the disturbance response behavior data set; S4: Cross analyze the disturbance response behavior data set and the tension evolution path sequence, identify the non-uniform balloon shell nodes, and form a structure non-uniform region identification data set; S5: Divide the balloon rupture strength regulation level based on the structure non-uniform region identification data set, perform control task allocation according to the regulation level, and obtain the balloon rupture strength regulation result.

2. The method of claim 1, wherein the method further comprises: The tension connection clustering feature set includes tension direction distribution information, connection stability grouping information, and balloon shell node relationship topology structure; the tension evolution path sequence includes tension transition stage sequence, state change probability distribution, and balloon shell node path continuity feature; the disturbance response behavior data set includes displacement response peak value information, response duration feature, and disturbance action direction record; the structure non-uniform region identification data set includes abnormal balloon shell node number, local tension distribution deviation feature, and path stability disturbance mark; and the balloon rupture strength regulation result includes regulation task type, execution operation instruction, and target region identification.

3. The method of claim 1, wherein the method further comprises: The acquisition step of the tension connection clustering feature set is specifically: S111: Obtain the node tension direction vector, tension size, and connection duration of the balloon shell node region through the fiber grating strain sensor to obtain a tension original parameter set; S112: Calculate the tension direction cosine angle between the balloon shell nodes based on the tension direction vector in the tension original parameter set, judge the direction consistency relationship, group according to the principle of similar vector angle, and obtain the tension direction relationship group; S113: Extract the tension direction relationship group, extract the tension direction distribution information, connection duration, and tension size in each grouping group, and construct the tension connection clustering feature set.

4. The method of claim 3, wherein the method further comprises: The acquisition step of the tension evolution path sequence is specifically: S211: Arrange and combine the tension connection clustering feature set in time sequence, set the observation state and hidden state code corresponding to each balloon shell node, and generate a balloon shell node tension state time sequence; S212: Based on the balloon shell node tension state sequence, compare the tension change amount and state duration between each state by using a hidden Markov model, count the transition frequency and time proportion of the state pair, and obtain the tension state transition probability; S213: According to the tension state transition probability, the most probable transition path in each group of state sequences is screened as the main path, the tension direction change value sequence of the continuous states in the path is extracted, and the corresponding offset direction change trend is recorded according to the node number of the balloon shell structure, and the tension evolution path sequence is obtained.

5. The method of claim 4, wherein the method further comprises: The acquisition step of the perturbation response behavior data set is specifically: S311: Based on the tension evolution path sequence, the continuous tension change rate of the balloon shell node is detected, the tension change rate difference of two consecutive time nodes is calculated, and it is judged whether there is a sequence interval greater than the tension change rate mutation reference value in the continuous difference value, which is integrated as a tension fluctuation abnormal section; S312: According to the tension fluctuation abnormal section, pulse negative pressure disturbance loading is implemented on the corresponding balloon shell node position, the time difference between the disturbance loading time and the node response starting time is recorded, and the node response time sequence is obtained; S313: The maximum displacement response value, displacement direction change path and displacement fluctuation amplitude of the corresponding node in the node response time sequence are collected, and the perturbation response behavior data set is obtained.

6. The method of claim 5, wherein the method further comprises: The acquisition step of the structure non-uniform region identification data set is specifically: S411: The perturbation response behavior data set and the node number of the balloon shell structure in the tension evolution path sequence are matched, the time stamp information of the two types of data under the same number is extracted, the time difference between the displacement response duration and the tension transition period is calculated, and the node response and transition difference value is obtained; S412: Based on the node response and transition difference value, the density distribution of the balloon shell node is judged, the response difference value standard deviation of each node and the adjacent node in the corresponding tension path is calculated, the abnormal recognition threshold is set according to the density distribution outlier rate judgment standard, and the density outlier abnormal node is obtained; S413: According to the density outlier abnormal node, the outlier node number, corresponding path section number and position serial number are extracted, the abnormal node distribution in each region is grouped according to the balloon shell structure region, the non-uniform trend index in each region is marked, and the structure non-uniform region identification data set is established.

7. The method of claim 6, wherein the method further comprises: The acquisition step of the balloon rupture strength regulation result is specifically: S511: Based on the structure non-uniform region identification data set, the tension direction offset amplitude and displacement response amplitude of the balloon shell node are extracted and weighted average is taken, and the balloon shell node joint score is obtained; S512: According to the balloon shell node joint score, the average value and standard deviation of all score values are calculated, the numerical interval is divided into grade labels, and the balloon rupture strength grade distribution result is obtained; S513: According to the balloon rupture strength grade distribution result, the balloon shell node number and grade label corresponding to each grade are combined and sorted, the task instruction mapping is established, and the balloon rupture strength regulation result is output.

8. A system for intelligent control of the burst strength of a blasting capsule, characterized in that The balloon rupture strength intelligent regulation method according to any one of claims 1-7, the system comprises: The tension parameter clustering module acquires the tension connection parameter set between the balloon shell nodes, and performs tension direction consistency clustering processing on the tension connection parameter set to generate a tension connection clustering feature set; The tension evolution modeling module inputs the tension connection clustering feature set into a hidden Markov model, analyzes the tension state transition of the same type of balloon shell node, and constructs a tension evolution path sequence; The disturbance collection response module screens the tension fluctuation abnormal section of the balloon shell node in the tension evolution path sequence, loads the suction disturbance signal on the abnormal section to perform response behavior collection, and obtains a disturbance response behavior data set; The non-uniform identification module cross-analyzes and processes the disturbance response behavior data set and the tension evolution path sequence to identify non-uniform balloon shell nodes and form a structure non-uniform region identification data set; The strength regulation grading module divides the balloon rupture strength regulation grade based on the structure non-uniform region identification data set, allocates control tasks according to the regulation grade, and obtains the balloon rupture strength regulation result.

Images