An explosive production whole-process analysis and intelligent regulation and control system based on multi-parameter correlation

Abstract

The application discloses a kind of based on multi-parameter association's explosive production whole process analysis and intelligent regulation and control system, it is related to explosive production control technical field, the present application is coupled with the characteristics of complex degree characteristics of the phase coupling characteristics of multiple process parameters, constructs phase coupling characteristic library;Virtual test scene is constructed, verify phase coupling characteristic library, and respectively construct chain reaction characteristic library and early control scheme library;In actual production, early risk is identified using phase coupling characteristic library, risk transmission path is predicted in combination with chain reaction characteristic library, and optimal control scheme is matched based on early control scheme library and is executed;Continuously monitor and track control effect, form closed-loop management and control, break through the limitation of traditional threshold alarm, realize early risk early identification, greatly advance warning time, guarantee production safety, realize early risk early prevention and control.

Description

Technical Field

[0001] This invention relates to the field of explosives production control technology, specifically to an intelligent control system for the entire explosives production process based on multi-parameter correlation. Background Technology

[0002] Explosives production involves many high-risk processes such as raw material preparation, emulsification, sensitization, and loading. There are numerous process parameters and complex coupling relationships. Moreover, the evolution of risks is characterized by strong nonlinearity, concealment, suddenness, and chain transmission. Based on multi-parameter correlation, the analysis and intelligent control of the entire explosives production process can achieve early warning and precise intervention of risks, which is irreplaceable for ensuring the safety of explosives production.

[0003] Existing technologies, such as the invention patent application with publication number CN115385763A, describe a method for quantitatively predicting the pressing process and density of press-loaded mixed explosives based on the AdaBoost algorithm. This method introduces the AdaBoost algorithm into the study of explosive process parameters and performance, leveraging the advantage of machine learning in uncovering hidden relationships within data. Process parameters such as specific pressure, pressing temperature, and holding time are used as sample features, and the charge density is used as the sample label. Supervised regression learning is performed using the AdaBoost algorithm to obtain a charge density prediction model. Using the trained model, an inverse prediction program is written to quantitatively predict the pressing process and density of press-loaded mixed explosives. This invention, through machine learning methods, establishes the relationship between the process parameters and charge density of press-loaded mixed explosives with high precision and automation, providing guidance for the selection of process parameters.

[0004] Existing technologies, such as the invention patent application with publication number CN120722726B, entitled "One-Click Production Control Method and System for Mixed Emulsion Explosives," solve the problems of low efficiency, high cost, quality fluctuations, and process uncertainty in traditional mixed emulsion explosive production through automated and intelligent control methods. It utilizes a quality sensor array and PID control algorithm to achieve precise raw material feeding, and ensures stable process parameters through multi-sensor data fusion and advanced control algorithms, thereby improving product quality consistency. Simultaneously, it establishes a feedback control loop and a continuous learning mechanism to achieve closed-loop control of the production process, optimize process parameters, ensure production stability and reliability, significantly improve production efficiency and product quality, and reduce labor costs.

[0005] The above technical solutions specifically disclose process optimization in explosive production and independent monitoring and post-event early warning of single parameters in production, but there are still at least the following problems: 1. The signal amplitude in the early stage of risk in explosive production is extremely low and is often submerged in noise. Moreover, the relationship between parameters is highly nonlinear and coupled. Traditional threshold alarms are difficult to capture the small parameter drift and phase coupling anomalies in the early stage of risk. Alarms can only be triggered when the risk evolves to obvious anomalies. However, at this time, the risk has already acquired a certain transmission capability and is very likely to trigger a chain reaction, resulting in delayed early warning, missing the best intervention opportunity, and making it difficult to achieve early risk prevention and control.

[0006] 2. Existing technologies adopt a single-parameter independent monitoring mode, which ignores the inherent coupling relationship between multiple process parameters in the entire explosive production process. It analyzes each parameter separately and cannot identify the hidden risks caused by the coordinated fluctuation and cross-influence of multiple parameters, resulting in serious blind spots in safety management.

[0007] 3. Post-event warnings can only issue alerts after a risk has occurred. They lack the ability to depict and predict the chain transmission pattern of risks and cannot clearly identify the transmission path of risks. As a result, operators can only passively deal with risks that have already occurred and cannot actively block the transmission of risks, making it difficult to prevent the spread of risks.

[0008] 4. Existing technologies lack a standardized library of chain reaction characteristics and an early control scheme library. Control strategies lack specificity and scientific rigor, and are prone to problems such as over-control, delayed control, or control deviation. Not only can they not effectively suppress risks, but they may also trigger new parameter anomalies or secondary risks. At the same time, they are difficult to adapt to dynamic changes under different production processes and operating conditions, and have poor versatility and adaptability. Summary of the Invention

[0009] To address the aforementioned technical shortcomings, the present invention aims to provide a system for analyzing and intelligently controlling the entire explosives production process based on multi-parameter correlation.

[0010] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a full-process analysis and intelligent control system for explosive production based on multi-parameter correlation, including the following modules: a risk feature construction module, used to obtain historical production data from different early risk production and normal production samples in different production processes from the historical production records of explosives, and construct a phase coupling feature library.

[0011] The virtual scenario testing module is used to construct a virtual test scenario for the explosives production line. The phase coupling feature library is verified in the virtual test scenario. After successful verification, the risk chain reaction is tested to determine the chain reaction feature library. Then, the control test is carried out to determine the early control scheme library.

[0012] The explosives production monitoring module is used to monitor the explosives production process in the current production line, acquire the production process and sensor signals corresponding to each process parameter in the current production line, identify the existence of early risks using the phase coupling feature library, predict the chain reaction of risks based on the chain reaction feature library, determine the early control scheme according to the early control scheme library, and carry out control.

[0013] The continuous monitoring module is used to continuously monitor the sensor signals corresponding to each process parameter, analyze whether explosive production has returned to safety, and then provide corresponding early warning feedback.

[0014] The beneficial effects of this invention are as follows: 1. This invention provides a multi-parameter correlation-based full-process analysis and intelligent control system for explosives production. It constructs a phase coupling feature library based on the phase coupling and complexity characteristics of multiple process parameters; it constructs a virtual test scenario to verify the phase coupling feature library, and separately constructs a chain reaction feature library and an early control scheme library; in actual production, it uses the phase coupling feature library to identify early risks, combines the chain reaction feature library to predict risk transmission paths, matches the optimal control scheme based on the early control scheme library, and executes it; it continuously monitors and tracks the control effect, forming a closed-loop management system, breaking through the limitations of traditional threshold alarms, achieving early risk identification, significantly advancing the warning time, ensuring production safety, and realizing early risk prevention and control.

[0015] 2. This invention combines multi-parameter correlation and coupling logic, breaking the limitations of independent monitoring of single parameters. It can accurately identify hidden risks caused by multi-parameter coordinated fluctuations and chain reactions, clarify the coupling relationship between parameters, comprehensively cover the entire process of explosive production safety management and control scenarios, and reduce the omission of hidden risks.

[0016] 3. By accurately characterizing the risk transmission pattern through the chain reaction feature database, the risk transmission path and superposition effect can be predicted in advance, realizing the transformation from post-event early warning to proactive prevention and control, which can effectively block the chain spread of risks and reduce the accident rate. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, an intelligent control system for the entire process of explosive production based on multi-parameter correlation includes the following modules: a risk feature construction module, which is used to obtain historical production data from different early risk production and normal production samples in different production processes from the historical production records of explosives, and construct a phase coupling feature library.

[0021] In a specific embodiment, the construction process of the phase coupling feature library in the risk feature construction module is as follows: For each production process, obtain each set of process parameters from historical production data in different early risk production and normal production samples. The set includes sensor signals and timestamps. Through time alignment, obtain the sensor signals of each process parameter at the same time. Using the adaptive complete set empirical mode decomposition method, adaptively decompose the sensor signals into multiple intrinsic mode functions, i.e., IMFs. Calculate the permutation entropy and fuzzy entropy for each IMF. Use Hilbert transform to extract the instantaneous frequency of each IMF. Calculate the phase synchronization index between multiple parameters based on the instantaneous frequency.

[0022] The phase synchronization index calculated using the phase-locked value method described above is existing technology and will not be elaborated further here.

[0023] Process parameters refer to quantitative indicators that can be collected in real time by sensors throughout the entire explosive production process and are used to reflect the production status, including emulsion temperature, aqueous phase density, sensitizer flow rate, emulsion pressure, and matrix viscosity.

[0024] IMF components with a phase synchronization index greater than a preset phase synchronization index threshold are selected as valid IMF components. A phase coupling matrix between multiple process parameters is constructed, with each matrix element corresponding to the phase synchronization index of any two process parameter IMF components. At the same time, the permutation entropy and fuzzy entropy of each valid IMF component are used as complexity features and marked next to the phase coupling matrix. According to different production processes, different early risk scenarios, and normal production scenarios, the phase coupling matrices and complexity features between multiple process parameters of all scenarios are used to form a phase coupling feature library.

[0025] It should be noted that the preset phase synchronization index threshold is set by professionals according to the needs of explosives production, and no specific numerical limit is specified here.

[0026] The virtual scenario testing module is used to construct a virtual test scenario for the explosives production line. The phase coupling feature library is verified in the virtual test scenario. After successful verification, the risk chain reaction is tested to determine the chain reaction feature library. Then, the control test is carried out to determine the early control scheme library.

[0027] In a specific embodiment, the verification process of the phase coupling feature library in the virtual scene testing module is as follows: normal production scenarios and each early risk production scenario under different production processes are simulated in the virtual test scenario, and then simulated production is carried out. During production, random fluctuations of process parameters are adjusted, and sensor signals and timestamps of each process parameter are acquired at the same time.

[0028] Production process refers to the complete process flow and operation in the entire process of explosive production, from raw material preparation to finished product output.

[0029] By utilizing sensor signals and timestamps for each process parameter under normal production scenarios and each early risk production scenario under different production processes, we screen out each effective IMF component, the phase coupling matrix between multiple process parameters, and the complexity features under the test, and compare them with the phase coupling feature library.

[0030] The verification is considered successful when all of the following conditions are met: (1) The matching degree between the effective IMF component under the test and the effective IMF component in the phase coupling feature library is greater than the preset matching degree threshold; (2) The difference between the matrix element in the phase coupling matrix between multiple process parameters under the test and the matrix element in the phase coupling matrix between process parameters in the phase coupling feature library is less than the preset element difference; (3) The difference between the complexity feature of each effective IMF component under the test and the complexity feature of the corresponding effective IMF component in the phase coupling feature library is less than the preset complexity feature difference threshold; If any condition is not met, the verification is considered to have failed and a failure feedback is given.

[0031] It should be noted that the matching degree is the result of dividing the number of valid IMF components in the test that are the same as the number of valid IMF components in the phase coupling feature library by the number of valid IMF components in the phase coupling feature library.

[0032] The setting process for the preset matching degree threshold, preset element difference, and preset complexity feature difference threshold is the same as the setting process for the preset phase synchronization index threshold.

[0033] Failure feedback: Send a text message indicating verification failure to the user's device to provide feedback on the failure.

[0034] In another specific embodiment, the chain reaction feature library is constructed as follows: when simulating each early risk production scenario under different production processes in a virtual test scenario, no safety control measures are activated, the simulated production continues, and then the sensor signals and timestamps of each process parameter are collected in real time.

[0035] Using a validated phase coupling feature library, each cascading risk in the continuous simulation production process is identified. Simultaneously, based on the temporal changes of sensor signals, the start time of each cascading risk is determined as the cascading start time. Furthermore, by combining each parameter, the transmission path of each cascading risk is traced.

[0036] Preferably, the process for determining each cascading risk and its start time is as follows: The sensor signals and timestamps of each process parameter collected in real time are analyzed according to the phase coupling feature library construction process to obtain each effective IMF component, the phase coupling matrix between multiple process parameters, and the complexity features. Then, these are compared with the effective IMF components, phase coupling matrices between multiple process parameters, and complexity features under the same early-risk production scenario in the successfully verified phase coupling feature library. If all verification conditions are met with the effective IMF components, phase coupling matrices between multiple process parameters, and complexity features under a certain early-risk production scenario, then the cascading risk is determined to exist. The time corresponding to the first satisfaction of all verification conditions is obtained as the start time of the cascading risk. This process is used to determine each cascading risk existing in the continuous simulated production process and its start time.

[0037] Preferably, the process of tracing the transmission path of each chain risk is as follows: extract each effective IMF component, the phase coupling matrix and complexity features between multiple process parameters in the same chain risk production scenario and normal production scenario under the same production process in the successfully verified phase coupling feature library, screen out effective IMF components that are different from each effective IMF component in the normal production scenario or effective IMF components whose complexity feature difference with the corresponding effective IMF component in the normal scenario is ≥ the preset complexity feature difference threshold, and matrix elements that are different from the matrix elements in the phase coupling matrix in the normal production scenario, and respectively take them as each abnormal effective IMF component and each abnormal matrix element.

[0038] Each abnormal effective IMF component and each abnormal matrix element are compared with the real-time effective IMF component and phase coupling matrix. If the real-time effective IMF component or phase coupling matrix first satisfies the following condition: the matching degree with a certain abnormal effective IMF component or a certain abnormal matrix element is greater than or equal to the preset matching threshold, and the corresponding process parameters continue to show abnormal fluctuations within the subsequent preset time, then the process parameters corresponding to the effective IMF component or phase coupling matrix at that time are obtained as the starting point of the chain risk transmission path.

[0039] Continuous abnormal fluctuations: The effective IMF component amplitude and complexity characteristics corresponding to the process parameters are always greater than the preset effective IMF component amplitude difference threshold and complexity characteristic difference threshold with the normal production scenario within a preset time, or the phase synchronization index is always in the abnormal range without obvious return to the normal range; the preset time is 5-10 seconds.

[0040] Then, using the elements of each anomaly matrix, the phase synchronization relationship between the starting process parameters and other process parameters is analyzed, and each associated parameter with a phase synchronization index deviation ≥ a preset deviation threshold is selected as an intermediate node for risk transmission. At the same time, the time when the complexity feature of each intermediate node's IMF component first differs from the corresponding complexity feature in the chain risk by less than a preset complexity feature difference threshold is extracted as the propagation time of each intermediate node. The nodes are connected according to the chronological order and phase coupling association logic to clarify the transmission order of each intermediate node.

[0041] Continuously trace the associated parameters corresponding to each intermediate node until all abnormal valid IMF components or process parameters with complexity characteristics greater than the complexity characteristic threshold are found. This process parameter is the endpoint of the chain risk transmission. The complete link from the starting point to the intermediate node to the endpoint is traced.

[0042] It should be noted that the setting process for the effective IMF component amplitude difference threshold, the preset matching threshold, and the complexity feature threshold is the same as the setting process for the preset phase synchronization index threshold.

[0043] Each chain of risks in each early-stage risk production scenario is sequentially linked according to its start time, and the transmission path is marked to form a chain reaction chain corresponding to each early-stage risk production scenario. The chain reaction chains of all early-stage risk production scenarios are classified and archived according to different production processes and different early-stage risk scenarios to form a chain reaction feature library.

[0044] In another specific embodiment, the process of determining the early control scheme library is as follows: First, for each early risk production scenario under different production processes, multiple initial early control schemes are preset. When simulating production in each early risk production scenario under different production processes, each initial early control scheme is executed respectively. Then, the sensor signals and timestamps of each process parameter are collected in real time, and the test is repeated.

[0045] Using a validated phase coupling feature library, the sensor signals and timestamps of each process parameter in each repeated test are analyzed to obtain the execution effect of each initial early control scheme. Several initial early control schemes with good execution effects are selected as qualified early control schemes. The suppression feature values ​​of each qualified early control scheme against cascading risks are obtained, and then sorted in descending order to obtain the order of each qualified early control scheme. Finally, the schemes are classified and archived according to different production processes, different early risks, and orders to obtain an early control scheme library.

[0046] Preferably, the process of obtaining the execution effect of each initial early control scheme is as follows: the sensor signals and timestamps of each process parameter collected in real time during each repeated test after the execution of each initial early control scheme are time-aligned, the phase coupling matrix and complexity features in real time during each repeated test are extracted, and compared with the phase coupling matrix and complexity features of the normal production scenario corresponding to the same early risk production scenario under the same production process in the phase coupling feature library, to determine whether the early risk has been eliminated in each repeated test, and the elimination time is recorded.

[0047] It should be noted that the elimination time is the duration from the start of early risk control to the time when the risk is eliminated.

[0048] The process for determining whether early risks have been eliminated in each repeated test is the same as the process for determining chain risks and their start times. If all verification conditions are not met, it indicates that the risk has been eliminated.

[0049] Simultaneously, the real-time phase coupling matrix and complexity features in each repeated test are compared with the phase coupling matrix and complexity features of different early risk production scenarios corresponding to the same early risk production scenario under the same production process in the phase coupling feature library to determine whether new chain risks are generated in each repeated test. If so, the generation duration is recorded. At the same time, based on the chain reaction feature library, it is analyzed whether each chain risk is suppressed in each repeated test. If it is eliminated, the corresponding suppression duration is recorded.

[0050] It should be noted that the process for determining new cascading risks is the same as that for each cascading risk and its start time. If a new early risk exists that meets all verification conditions, it is determined that a new cascading risk has occurred. The process for determining whether a cascading risk has been suppressed is the same as that for determining whether an early risk has been eliminated. The suppression duration is the time between the start of control measures and the time when the cascading risk is eliminated. The occurrence duration is the time between the start of control measures and the time when the new cascading risk occurs.

[0051] For each initial early control plan, the early risk elimination rate, average elimination time, new chain risk generation rate, average generation time, chain risk suppression rate, and average suppression time are statistically analyzed. Then, the data is normalized, and the early risk effect characteristic value and chain risk suppression characteristic value of each initial early control plan are calculated. Finally, the execution effect characteristic value of each initial early control plan is calculated. If the execution effect characteristic value is greater than the preset execution effect characteristic value threshold, the execution effect is judged to be good; otherwise, the execution effect is judged to be poor.

[0052] It should be noted that the early risk elimination rate is the number of early risk eliminations in repeated tests divided by the total number of repeated tests. The average elimination time is the average of the early risk elimination times across multiple repeated tests. The new cascading risk generation rate, average generation time, cascading risk suppression rate, and average suppression time are calculated using the same methods as the early risk elimination rate and average elimination time.

[0053] The early risk elimination rate and average elimination time are normalized. The characteristic value of early risk effect = normalized value of early risk elimination rate × 0.6 + (1 - normalized value of average elimination time) × 0.4.

[0054] The new chain risk generation rate, average generation duration, chain risk suppression rate, and average suppression duration are normalized. The suppression characteristic value of chain risk is calculated as follows: [(1 - normalized value of new chain risk generation rate) × 0.6 + average generation duration × 0.4] × 0.5 + [normalized value of chain risk suppression rate × 0.6 + (1 - normalized value of average suppression duration) × 0.4] × 0.5.

[0055] The performance effect characteristic value = early risk effect characteristic value × 0.6 + suppression of cascading risks characteristic value × 0.4.

[0056] It should be noted that the process for setting the preset execution effect feature value threshold is the same as the process for setting the preset phase synchronization index threshold.

[0057] The explosives production monitoring module is used to monitor the explosives production process in the current production line, acquire the production process and sensor signals corresponding to each process parameter in the current production line, identify the existence of early risks using the phase coupling feature library, predict the chain reaction of risks based on the chain reaction feature library, determine the early control scheme according to the early control scheme library, and carry out control.

[0058] In a specific embodiment, the specific process of the explosive production monitoring module is as follows: extract the production process of explosive production in the current production line as the current production process, use the sensor devices deployed on the production line to collect sensor signals and timestamps of each process parameter in real time, perform time alignment, extract the real-time phase coupling matrix and complexity features in the current production process, and then compare them with the phase coupling matrix and complexity features corresponding to different early risk scenarios and normal production scenarios of the current production process in the phase coupling feature library to determine whether there are early risks in the current production process.

[0059] Among them, the production process of explosives production in the current production line is extracted from the production control center.

[0060] If there is only one early risk, the corresponding chain reaction chain is matched based on the chain reaction feature library and used as the current risk transmission path; the qualified early control plan ranked first in the early control plan library is called as the early control plan and control is implemented.

[0061] If there are at least two early risks, firstly, based on the chain reaction feature library, match the chain reaction chain of each early risk, compare each early risk with its chain reaction chain, determine whether each early risk is related, and determine the current risk transmission path and confirm the early control plan according to the correlation of each early risk.

[0062] Preferably, the process of determining the current risk transmission path is as follows: multiple related early risks are taken as risk groups, the chain reaction chains within the risk groups are obtained, and the paths of the intersection nodes and their upstream and downstream are merged, starting from each early risk. In the merged graph, the betweenness number of each intermediate node is calculated, and the node with the highest betweenness number is the risk convergence point.

[0063] Betweenness is the number of paths that pass through that node.

[0064] Based on the real-time phase coupling matrix and complexity characteristics of the current production process, the abnormal correlation of risk convergence points is verified, and it is confirmed whether the risk convergence points are the core hubs for the transmission of each early risk. False risk convergence nodes are eliminated. Taking each early risk as the starting point and the risk convergence point as the core hub, the relationship between all nodes in the merged path is sorted out. The top 3 key intermediate nodes in terms of the centrality of the intermediate number of each path are retained, and the rest are eliminated to construct the current risk transmission path of the risk group.

[0065] The chain reaction of multiple unrelated early risks is taken as an independent risk transmission path, and the current risk transmission path is composed of the current risk transmission path of the risk group and each independent risk transmission path.

[0066] In the above, the abnormal correlation of risk convergence points is verified by comparing the real-time phase synchronization index and the complexity characteristics of the effective IMF component of the process parameters corresponding to the risk convergence point with the phase synchronization index and the complexity characteristics of the effective IMF component of the process parameters of an early risk production scenario in the phase coupling feature library. The difference between the phase synchronization index and the complexity characteristics is calculated. If the difference between the phase synchronization index and the complexity characteristics is less than or equal to the preset threshold, the risk convergence point is confirmed as the core hub of the early risk transmission. Otherwise, it is determined as a false risk convergence node. This is used to confirm whether the risk convergence point is the core hub of each early risk transmission.

[0067] The process for setting the phase synchronization index difference threshold is the same as the process for setting the preset phase synchronization index threshold.

[0068] Preferably, the process of confirming the early control scheme is as follows: First, extract each new chain risk and each suppressed chain risk corresponding to each qualified early control scheme in the current production process risk group from the chain reaction feature library. Then, compare each new chain risk with the early risks of each node in the current risk transmission path of the risk group. If there are any similarities, they are recorded as superimposed risks. Count the number of superimposed risks corresponding to each qualified early control scheme.

[0069] The suppressed chain risks are compared with the early risks at each node in the current risk transmission path of the risk group. If there are similarities, they are recorded as eliminated risks. The number of eliminated risks corresponding to each qualified early control plan is counted.

[0070] Calculate the priority value corresponding to each qualified early control scheme, select the qualified early control scheme with the highest priority value as the main control scheme, select multiple secondary control schemes from the remaining qualified early control schemes, and then superimpose the main control scheme and multiple secondary control schemes to obtain the risk group control scheme.

[0071] In the above, the number of superimposed risks and the number of eliminated risks are normalized. The priority value = normalized value of eliminated risks - normalized value of superimposed risks. The priority value ranges from [-1, 1].

[0072] It should be noted that the new chain risks and suppressed chain risks corresponding to each qualified early control scheme within the risk group of the current production process in the chain reaction feature library are all new chain risks and all chain risks that have been successfully suppressed at least once in repeated tests.

[0073] In this application, all normalization processes are linear normalization, which uniformly maps the index values ​​to the [0,1] interval.

[0074] For example, the selection process of the secondary control scheme is as follows: the secondary control scheme is selected if all of the following conditions are met: (1) the number of superimposed risks is less than the preset number of superimposed risks; (2) the number of superimposed risks is the same as the number of superimposed risks in the main control scheme is less than the preset threshold for the same number of superimposed risks; (3) the number of risks eliminated is greater than the preset threshold for the number of risks eliminated; (4) the number of risks eliminated is the same as the number of risks eliminated in the main control scheme is less than the preset threshold for the same number of risks eliminated.

[0075] It should be noted that all the above preset parameters are set in the same way as the preset phase synchronization index threshold.

[0076] Retrieve each new chain risk corresponding to each independent risk transmission path from the early control scheme database, compare each new chain risk with the early risks of each node in the current risk transmission path of the risk group, if there are any similarities, record them as superimposed risks, count the number of superimposed risks corresponding to each qualified early control scheme, and select the qualified early control scheme with the smallest number of superimposed risks as the independent control scheme.

[0077] The early control plan consists of the risk group control plan and the independent control plan for each independent risk transmission path.

[0078] The continuous monitoring module is used to continuously monitor the sensor signals corresponding to each process parameter, analyze whether explosive production has returned to safety, and then provide corresponding early warning feedback.

[0079] It should be noted that the continuous monitoring method is the same as that in the explosives production monitoring module. When no early risks are identified, it is determined that explosives production has returned to safety; otherwise, it is not considered safe, and a text message indicating that production has not returned to safety is sent to the user terminal. At the same time, the explosives production monitoring module is executed again for adjustment. If safety is still not restored after adjustment, a manual intervention prompt is sent to the user terminal.

[0080] The examples described in this invention are not limited to the specific embodiments listed above. The examples are merely illustrative to facilitate understanding of the invention and do not constitute a limitation on the scope of protection of this invention. Any modifications, equivalent substitutions, etc., made within the spirit and principles of this invention should be included within the scope of protection.

[0081] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A system for analyzing and intelligently controlling the entire process of explosive production based on multi-parameter correlation, characterized in that, Includes the following modules: The risk feature construction module is used to obtain historical production data from different early risk production and normal production samples in different production processes from the historical production records of explosives, and to construct a phase coupling feature library. The virtual scenario testing module is used to construct a virtual test scenario for the explosives production line. The phase coupling feature library is verified in the virtual test scenario. After successful verification, the risk chain reaction is tested to determine the chain reaction feature library. Then, the control test is carried out to determine the early control scheme library. The explosives production monitoring module is used to monitor the explosives production process in the current production line, acquire the production process and sensor signals corresponding to each process parameter in the current production line, identify the existence of early risks using the phase coupling feature library, predict the chain reaction of risks based on the chain reaction feature library, determine the early control scheme according to the early control scheme library, and carry out control. The continuous monitoring module is used to continuously monitor the sensor signals corresponding to each process parameter, analyze whether explosive production has returned to safety, and then provide corresponding early warning feedback.

2. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 1, characterized in that, The construction process of the phase coupling feature library in the risk feature construction module is as follows: For each production process, a set of process parameters is obtained from historical production data in different early risk production and normal production samples. The set includes sensor signals and timestamps. Through time alignment, the sensor signals of each process parameter at the same time are obtained. Using the adaptive complete set empirical mode decomposition method, the sensor signals are adaptively decomposed into multiple intrinsic mode functions, i.e., IMFs. The permutation entropy and fuzzy entropy are calculated for each IMF. The instantaneous frequency of each IMF is extracted using Hilbert transform. Based on the instantaneous frequency, the phase synchronization index between multiple parameters is calculated. IMF components with a phase synchronization index greater than a preset phase synchronization index threshold are selected as valid IMF components. A phase coupling matrix between multiple process parameters is constructed, with each matrix element corresponding to the phase synchronization index of any two process parameter IMF components. At the same time, the permutation entropy and fuzzy entropy of each valid IMF component are used as complexity features and marked next to the phase coupling matrix. According to different production processes, different early risk scenarios, and normal production scenarios, the phase coupling matrices and complexity features between multiple process parameters of all scenarios are used to form a phase coupling feature library.

3. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 1, characterized in that, The verification process of the phase coupling feature library in the virtual scene testing module: In the virtual test scenario, normal production scenarios and each early risk production scenario under different production processes are simulated. Then, simulated production is carried out, and random fluctuations of process parameters are adjusted during production. At the same time, sensor signals and timestamps of each process parameter are acquired. By using sensor signals and timestamps of each process parameter under normal production scenarios and each early risk production scenario under different production processes, we screen out each effective IMF component, the phase coupling matrix between multiple process parameters and the complexity features under the test, and compare them with the phase coupling feature library. The verification is considered successful when all of the following conditions are met: (1) The matching degree between the effective IMF component under the test and the effective IMF component in the phase coupling feature library is greater than the preset matching degree threshold; (2) The difference between the matrix elements in the phase coupling matrix between multiple process parameters under test and the matrix elements in the phase coupling feature library between process parameters is less than the preset element difference; (3) The difference between the complexity feature of each effective IMF component under the test and the complexity feature of the corresponding effective IMF component in the phase coupling feature library is less than the preset complexity feature difference threshold. If any condition is not met, the verification is deemed to have failed, and a failure feedback is sent.

4. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 1, characterized in that, The construction process of the chain reaction feature library is as follows: In the virtual test scenario, when simulating each early risk production scenario under different production processes, no safety control measures are activated, and the simulated production continues. Then, the sensor signals and timestamps of each process parameter are collected in real time. Using the successfully validated phase coupling feature library, each cascading risk in the continuous simulation production process is identified. At the same time, based on the temporal changes of sensor signals, the start time of each cascading risk is determined as the cascading start time. And by combining each parameter, the transmission path of each chain of risks can be traced; Each chain of risks in each early-stage risk production scenario is sequentially linked according to its start time, and the transmission path is marked to form a chain reaction chain corresponding to each early-stage risk production scenario. The chain reaction chains of all early-stage risk production scenarios are classified and archived according to different production processes and different early-stage risk scenarios to form a chain reaction feature library.

5. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 1, characterized in that, The process of determining the early regulation scheme library: First, for each early risk production scenario under different production processes, multiple initial early control schemes are preset. During the simulated production of each early risk production scenario under different production processes, each initial early control scheme is executed. Then, the sensor signals and timestamps of each process parameter are collected in real time, and the test is repeated. Using a validated phase coupling feature library, the sensor signals and timestamps of each process parameter in each repeated test are analyzed to obtain the execution effect of each initial early control scheme. Several initial early control schemes with good execution effects are selected as qualified early control schemes. The suppression feature values ​​of each qualified early control scheme against cascading risks are obtained, and then sorted in descending order to obtain the order of each qualified early control scheme. Finally, the schemes are classified and archived according to different production processes, different early risks, and orders to obtain an early control scheme library.

6. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 5, characterized in that, The process of obtaining the execution effect of each initial early control scheme is as follows: After each initial early control scheme is executed, the sensor signals and timestamps of each process parameter collected in real time during each repeated test are time-aligned. The phase coupling matrix and complexity features in real time during each repeated test are extracted and compared with the phase coupling matrix and complexity features of the normal production scenario corresponding to the same early risk production scenario under the same production process in the phase coupling feature library. It is then determined whether the early risk has been eliminated in each repeated test and the elimination time is recorded. Simultaneously, the real-time phase coupling matrix and complexity features in each repeated test are compared with the phase coupling matrix and complexity features of different early risk production scenarios corresponding to the same early risk production scenario under the same production process in the phase coupling feature library to determine whether new chain risks are generated in each repeated test. If so, the generation duration is recorded. At the same time, based on the chain reaction feature library, it is analyzed whether each chain risk is suppressed in each repeated test. If it is eliminated, the corresponding suppression duration is recorded. For each initial early control plan, the early risk elimination rate, average elimination time, new chain risk generation rate, average generation time, chain risk suppression rate, and average suppression time are statistically analyzed. Then, the data is normalized, and the early risk effect characteristic value and chain risk suppression characteristic value of each initial early control plan are calculated. Finally, the execution effect characteristic value of each initial early control plan is calculated. If the execution effect characteristic value is greater than the preset execution effect characteristic value threshold, the execution effect is judged to be good; otherwise, the execution effect is judged to be poor.

7. The explosives production process analysis and intelligent control system based on multi-parameter correlation according to claim 1, characterized in that, The specific process of the explosives production monitoring module is as follows: The production process of explosives production in the current production line is extracted as the current production process. Using the sensor devices deployed on the production line, the sensor signals and timestamps of each process parameter are collected in real time and time-aligned. The real-time phase coupling matrix and complexity features of the current production process are extracted and then compared with the phase coupling feature library corresponding to different early risk scenarios and normal production scenarios of the current production process to determine whether there are early risks in the current production process. If there is only one early risk, the corresponding chain reaction chain is matched based on the chain reaction feature library and used as the current risk transmission path. The qualified early control plan ranked first in the early control plan library is selected as the early control plan and control is implemented. If there are at least two early risks, firstly, based on the chain reaction feature library, match the chain reaction chain of each early risk, compare each early risk with its chain reaction chain, determine whether each early risk is related, and determine the current risk transmission path and confirm the early control plan according to the correlation of each early risk.

8. The explosives production process analysis and intelligent control system based on multi-parameter correlation according to claim 7, characterized in that, The process of determining the current risk transmission path: Multiple related early risks are grouped into a risk group. The chain reaction chains within the risk group are obtained. Starting from each early risk, the intersection nodes and the paths upstream and downstream of the intersection nodes are merged. In the merged graph, the betweenness of each intermediate node is calculated. The node with the highest betweenness is the risk convergence point. Based on the real-time phase coupling matrix and complexity characteristics of the current production process, the abnormal correlation of risk convergence points is verified, it is confirmed whether the risk convergence points are the core hubs of each early risk transmission, and false risk convergence nodes are eliminated. Starting with each early risk and with the risk convergence point as the core hub, we sort out the relationships between all nodes in the merged path, retain the top 3 key intermediate nodes in terms of intermediary centrality for each path, and remove the rest to construct the current risk transmission path of the risk group. The chain reaction of multiple unrelated early risks is taken as an independent risk transmission path, and the current risk transmission path is composed of the current risk transmission path of the risk group and each independent risk transmission path.

9. The explosives production process analysis and intelligent control system based on multi-parameter correlation as described in claim 8, characterized in that, The process for confirming the early control plan is as follows: First, extract each new chain risk and each suppressed chain risk corresponding to each qualified early control scheme within the risk group of the current production process from the chain reaction feature library. Then, compare each new chain risk with the early risks of each node in the current risk transmission path of the risk group. If there are any similarities, they are recorded as superimposed risks. Count the number of superimposed risks corresponding to each qualified early control scheme. Compare the suppressed chain risks with the early risks at each node in the current risk transmission path of the risk group. If there are similarities, they are recorded as eliminated risks. Count the number of eliminated risks corresponding to each qualified early control plan. Calculate the priority value corresponding to each qualified early control scheme, select the qualified early control scheme with the highest priority value as the main control scheme, select multiple secondary control schemes from the remaining qualified early control schemes, and then superimpose the main control scheme and multiple secondary control schemes to obtain the risk group control scheme. Retrieve each new chain risk corresponding to each independent risk transmission path from the early control scheme library, compare each new chain risk with the early risks of each node in the current risk transmission path of the risk group, if there are any similarities, record them as superimposed risks, and count the number of superimposed risks corresponding to each qualified early control scheme. Select the qualified early intervention scheme with the minimum number of superimposed risks as the independent intervention scheme; The early control plan consists of the risk group control plan and the independent control plan for each independent risk transmission path.

10. The explosives production process analysis and intelligent control system based on multi-parameter correlation according to claim 9, characterized in that, The selection process of the secondary control scheme: The following conditions must be met for a secondary control scheme to be selected: (1) The number of superimposed risks is less than the preset number of superimposed risks; (2) The number of superimposed risks that are the same as the number of superimposed risks in the main control scheme is less than the preset threshold for the same number of superimposed risks; (3) The number of risks eliminated is greater than the preset threshold for the number of risks eliminated. (4) The number of risks eliminated is the same as the number of risks eliminated in the main control plan, which is less than the preset threshold for the same number of risks eliminated.

Images