Safety production equipment risk monitoring method and system

Abstract

The invention relates to a safety production equipment risk monitoring method and system, and belongs to the technical field of risk monitoring, and the method comprises the steps: obtaining historical operation data and real-time operation data of to-be-monitored safety production equipment; setting a dynamic threshold according to the historical operation data, predicting the range of the dynamic threshold according to preset target process operation data and the historical operation data through a bidirectional long-short-term memory network, and updating the range of the dynamic threshold in combination with the real-time operation data to obtain an updated dynamic threshold; the comprehensive risk index is calculated according to the updated dynamic threshold value and the real-time operation data through the analytic hierarchy process, the risk of the safety production equipment to be monitored is evaluated according to the comprehensive risk index, an evaluation result is obtained, dynamic monitoring and evaluation of the risk of the safety production equipment are achieved, and the safety production equipment risk evaluation efficiency is improved. The actual operation state of the equipment can be reflected more accurately, potential risks can be found in time, corresponding measures can be taken, and the operation safety and reliability of the equipment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk monitoring, in particular to a safety production equipment risk monitoring method and system. BACKGROUND

[0002] Safety production equipment is a key equipment in industrial production process for protecting personnel safety and preventing accidents, such as pressure vessels, piping systems, rotating machinery, etc. These devices bear complex mechanical and thermal loads during operation, and once a failure occurs, it may lead to serious safety accidents, production stoppage and economic losses. Therefore, real-time monitoring of equipment operating conditions, accurate risk assessment and timely warning are crucial to ensure industrial safety production.

[0003] Currently, the risk monitoring of safety production equipment mainly relies on traditional point sensors and fixed threshold alarm systems. Point sensors measure local strain, temperature and other parameters through discrete arrangement, while fixed threshold alarm systems trigger warnings based on pre-set safety limits. In addition, some industry standards provide qualitative management requirements, but lack dynamic quantitative indicators.

[0004] However, the existing technology has obvious defects: point sensors cannot achieve continuous monitoring, have blind spots, and have poor anti-interference ability; fixed threshold systems cannot distinguish between normal operating fluctuations and real failures, with high false alarm rate; at the same time, traditional methods lack multi-source data fusion and trend prediction capabilities, leading to delayed risk warning, which cannot meet the needs of modern industry for precise and intelligent safety monitoring. SUMMARY

[0005] In view of the deficiencies in the related art, the present application aims to provide a safety production equipment risk monitoring method to solve the technical problems that point sensors in the prior art cannot achieve continuous monitoring, have blind spots, and have poor anti-interference ability; fixed threshold systems cannot distinguish between normal operating fluctuations and real failures, with high false alarm rate; at the same time, traditional methods lack multi-source data fusion and trend prediction capabilities, leading to delayed risk warning, which cannot meet the needs of modern industry for precise and intelligent safety monitoring.

[0006] The present application provides a safety production equipment risk monitoring method, characterized in that it comprises the following steps: Data acquisition step: acquiring historical operating data and real-time operating data of the safety production equipment to be monitored; Threshold setting step: setting a dynamic threshold according to the historical operating data, predicting the range of the dynamic threshold based on pre-set target process operating data and the historical operating data through a bidirectional long short-term memory network, updating the range of the dynamic threshold based on the real-time operating data, and obtaining an updated dynamic threshold; The risk assessment step comprises: calculating a comprehensive risk index according to the updated dynamic threshold and the real-time operation data by using an analytic hierarchy process, evaluating the risk of the safety production equipment to be monitored according to the comprehensive risk index, and obtaining an evaluation result.

[0007] By obtaining historical operation data and real-time operation data, setting a dynamic threshold range, using a bidirectional long short-term memory network to predict and update the threshold range, and calculating a comprehensive risk index by using an analytic hierarchy process for evaluation, dynamic monitoring and evaluation of the safety production equipment risk are realized, the actual operation state of the equipment can be more accurately reflected, the defect that the traditional fixed threshold monitoring method cannot adapt to the change of the equipment operation state is avoided, and by real-time data updating and risk evaluation, potential risks can be found in time and corresponding measures can be taken, thereby improving the safety and reliability of the equipment operation.

[0008] In some embodiments of the present application, the threshold setting step specifically comprises: setting a static threshold according to a preset equipment actual standard, and setting a dynamic threshold according to the static threshold and the historical operation data by using a box plot method and normal distribution test.

[0009] By distinguishing the setting of the static threshold and the dynamic threshold, multi-level monitoring of the equipment risk is realized, the static threshold is set based on the equipment design standard, the basic safety boundary of the equipment operation is ensured, the dynamic threshold is obtained by analyzing the historical operation data, the state change of the equipment in the actual operation is reflected, the safety bottom line of the equipment operation is ensured, and flexible adjustment according to the actual operation condition is realized, the false alarm or missed alarm problem caused by single threshold setting is avoided, and the accuracy and adaptability of the risk monitoring are improved.

[0010] In some embodiments of the present application, the threshold setting step specifically comprises: abnormal values outside a preset multiple of quartile range in the historical operation data are removed by using a box plot method, and a data set is constructed; the data set is verified by using a normal distribution test to determine whether the data set conforms to a normal distribution; if the data set conforms to the normal distribution, a k value is configured based on a preset equipment type, a mean value and a standard deviation are calculated according to the k value, and the dynamic threshold is set according to the mean value and the standard deviation.

[0011] The historical operation data is processed by a box plot method and a normal distribution test, so as to ensure the data quality for setting the dynamic threshold, the box plot method can effectively eliminate abnormal values and avoid the influence of abnormal data on the setting of the threshold, and the normal distribution test ensures that the statistical characteristics of the data meet the expectation, so that the set dynamic threshold is more scientific and reasonable, can truly reflect the parameter fluctuation range of the equipment in the normal operation state, provides a reliable basis for subsequent risk assessment, and improves the stability and reliability of the whole monitoring system.

[0012] In some embodiments of the application, the threshold setting step specifically comprises: If the equipment type is a first risk type, the k value is set to a first numerical value, a first mean value and a first standard deviation are calculated according to the first numerical value, and the dynamic threshold is set according to the first mean value and the first standard deviation. If the equipment type is a second risk type, the k value is set to a second numerical value, a second mean value and a second standard deviation are calculated according to the second numerical value, and the dynamic threshold is set according to the second mean value and the second standard deviation.

[0013] By setting different k values according to the equipment risk type to calculate the dynamic threshold, the differential monitoring of equipment of different risk levels is realized, a smaller k value is used for low-risk equipment to improve the monitoring efficiency while ensuring basic safety, and a larger k value is used for high-risk equipment to ensure equipment safety with a more stringent standard, which can intelligently adapt to the safety needs of different types of equipment, optimizes resource allocation while ensuring safety, and improves the overall monitoring performance.

[0014] In some embodiments of the application, the threshold setting step further comprises: The range of the dynamic threshold in the preset future period is predicted by the bidirectional long short-term memory network, and a predicted dynamic threshold change range is obtained. In a preset period, the predicted dynamic threshold change range is updated according to the real-time operation data, and an updated dynamic threshold is obtained.

[0015] The range of the dynamic threshold in the future period is predicted by the bidirectional long short-term memory network, and the real-time operation data is updated, which realizes the dynamic adjustment and optimization of the threshold, can fully consider the historical change trend of the equipment operation parameter, makes a reasonable prediction on the future state, and the periodic updating mechanism ensures that the threshold can timely reflect the latest operation state of the equipment, can adapt to various changes of the equipment operation state, avoids the inadaptability of the traditional fixed threshold method when the equipment is aging or the working condition changes, and improves the accuracy and timeliness of the monitoring.

[0016] In some embodiments of the application, the threshold setting step further comprises: extracting a same-operation sample in the historical operation data according to the target process operation data; extracting a dynamic threshold record of the safety production equipment to be monitored in normal operation in the same-operation sample, and obtaining a threshold change rule according to the dynamic threshold record; predicting a range of the dynamic threshold according to the threshold change rule and an equipment parameter in the target process operation data, and obtaining a predicted dynamic threshold change range.

[0017] By extracting the same-operation sample in the historical operation data and analyzing the characteristics thereof, the change rule of the dynamic threshold can be predicted for a specific process operation, the influence difference of different process operations on the equipment operation parameter is fully considered, the parameter change trend under a specific operation can be more accurately predicted through analysis of the same sample, prediction deviation caused by mixing data under different operation modes is avoided, the adjustment of the dynamic threshold is more accurate, the risk monitoring demand under various process operation conditions can be better adapted, and the adaptation capability of the system to different production scenes is improved.

[0018] In some embodiments of the application, the threshold setting step further comprises: calculating a deviation value of the real-time operation data and the predicted threshold change range, judging whether the same-operation sample is re-extracted according to the deviation value, and updating the dynamic threshold change range to obtain an updated dynamic threshold.

[0019] By comparing the deviation of the real-time operation data and the predicted threshold range, and determining whether the sample is re-extracted and the threshold is updated according to the deviation value, dynamic verification and correction of the prediction result are realized, deviation between actual operation and expectation can be found in time, and the threshold is updated by re-matching the sample or adjusting the proportion, so that the threshold setting is always consistent with the actual operation state of the equipment, the error accumulation problem is effectively solved, risk misjudgment caused by prediction deviation is avoided, and the self-adaptation capability and reliability of risk monitoring are improved.

[0020] In some embodiments of the application, the risk assessment step specifically comprises: calculating a risk index according to the updated dynamic threshold and the real-time operation data; obtaining a comprehensive index through weighted calculation of a preset weight and the risk index; obtaining a corresponding risk level according to the comprehensive index, and responding to the safety production equipment to be detected according to the risk level.

[0021] The comprehensive index is obtained by weighting the risk indicators and preset weights, and the risk level and response measures are determined according to the comprehensive index, so that the quantitative evaluation and hierarchical response of the equipment risk are realized, the risk conditions of single parameters are considered, the influence degree of different parameters on the overall risk is reflected through weight distribution, the risk evaluation is more comprehensive and objective, the risk level is determined according to the range of the comprehensive index, and the corresponding measures are taken, so that a complete risk response mechanism is formed, and a more scientific decision basis is provided for equipment safety management.

[0022] In some embodiments of the application, the historical operation data or real-time operation data includes pressure data, strain data and temperature data, and the data acquisition step specifically includes: A distributed optical fiber is arranged on the safety production equipment to be monitored, and the safety production equipment to be monitored is monitored through the distributed optical fiber to obtain the pressure data, the strain data and the temperature data.

[0023] The pressure, strain and temperature data are acquired through the distributed optical fiber, continuous spatial monitoring of the equipment state is realized, the distributed optical fiber sensing technology can provide comprehensive strain and temperature distribution information of the equipment surface, avoids the possible monitoring blind area of the traditional point sensor, and the optical fiber sensing has the advantages of anti-electromagnetic interference and corrosion resistance, is suitable for long-term stable work in complex industrial environments, provides a more comprehensive and reliable original data basis for subsequent threshold setting and risk evaluation, and improves the accuracy and stability of the entire monitoring system.

[0024] The application further provides a safety production equipment risk monitoring system for realizing the safety production equipment risk monitoring method of the above-mentioned embodiments. The safety production equipment risk monitoring system includes: A data acquisition module is configured to acquire historical operation data and real-time operation data of the safety production equipment to be monitored. A threshold setting module is configured to set a dynamic threshold according to the historical operation data, predict a range of the dynamic threshold according to preset target process operation data and the historical operation data through a bidirectional long short-term memory network, update the range of the dynamic threshold according to the real-time operation data, and obtain an updated dynamic threshold. A risk evaluation module is configured to calculate a comprehensive risk index according to the updated dynamic threshold and the real-time operation data through an analytic hierarchy process, evaluate the risk of the safety production equipment to be monitored according to the comprehensive risk index, and obtain an evaluation result.

[0025] The function integration of data acquisition, threshold setting and risk assessment is realized through the modular design, a complete risk monitoring system is formed, the data acquisition module is responsible for collecting the basic data of the equipment operation, the threshold setting module realizes the dynamic prediction and update of the threshold, the risk assessment module completes the calculation and assessment of the comprehensive risk, so that the functions of the system are clear and cooperate closely, the integrity of the system is ensured and the independent optimization and maintenance of each part are facilitated, a high-efficiency and reliable technical platform is provided for the risk monitoring of the safety production equipment, and the safety monitoring demand in different industrial scenes can be met. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, specific embodiments of the present application will be described in detail below with reference to the drawings. For those skilled in the art, other drawings can also be obtained without creative labor on the premise of the drawings. Figure 1 A flow chart of safety production equipment risk monitoring provided by the embodiment of the present application; Figure 2 A structural schematic diagram of safety production equipment risk monitoring provided by the embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the present application more clear, the present application is described and explained below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not intended to limit the present application. Based on the embodiments provided by the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application. It should be noted that the terms used here are only for the purpose of describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form, and in addition, it should be understood that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices. In the traditional safety production equipment risk monitoring process, it mainly relies on traditional point sensors and fixed threshold alarm systems.

[0028] However, the traditional point sensors such as strain gauges and thermocouples need to be arranged discretely, which cannot realize continuous monitoring of the equipment, and is prone to cause missed detection of local risks such as weld micro-cracks.

[0029] And the point sensor wiring is complex, poor anti-electromagnetic interference ability, in chemical, metallurgical and other strong electromagnetic environment, often due to signal interference leading to monitoring failure. For example, in the large reactor monitoring of chemical enterprises, due to the discrete arrangement of strain gauges, there is a large area of monitoring blind area on the surface of the reactor, and once the small cracks are not found in time, they may expand rapidly with the operation of the equipment, causing serious safety accidents.

[0030] The point sensor measures local strain, temperature and other parameters by discrete arrangement, and the fixed threshold alarm system triggers early warning based on the preset safety limit. In addition, some industry standards provide qualitative management requirements, but lack dynamic quantitative indicators.

[0031] In addition, current equipment management standards, such as TSG11-2020 "Boiler Safety Technology Regulations", are mainly qualitative requirements, and lack dynamic quantitative indicators based on real-time data. From historical accidents, such as the explosion of the electric arc furnace in Huaye Foundry, it can be seen that the unreasonable setting of equipment risk threshold, early warning lag and other problems are prominent. In actual production, due to the lack of scientific quantitative standards, equipment managers are difficult to accurately judge the degree of equipment risk, and often can only take measures when the equipment appears obvious failure or approaches to dangerous state, causing significant economic loss and safety hazards.

[0032] The traditional fixed threshold alarm system cannot effectively distinguish between normal fluctuations and potential failures of the equipment, resulting in a high false alarm rate. At the same time, lacking the ability of multi-source data fusion and trend prediction, it is difficult to realize early warning of equipment failure. Taking power equipment monitoring as an example, affected by factors such as environmental temperature changes and equipment load fluctuations, the traditional threshold alarm system frequently issues false alarms, increasing the workload of workers while also reducing the credibility of the early warning system.

[0033] In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The technical solutions of the present application will be described in detail below with reference to specific embodiments and the accompanying drawings.

[0034] As Figure 1 shown, the present application provides a safety production equipment risk monitoring method, characterized in that it comprises the following steps: Data acquisition step S1: acquiring historical operation data and real-time operation data of the safety production equipment to be monitored; wherein the historical operation data or real-time operation data includes pressure data, strain data and temperature data; specifically, the historical operation data is pressure data, strain data and temperature data of the safety production equipment to be monitored running continuously for greater than or equal to 720 hours; In some embodiments, the data acquisition step S1 specifically comprises: A distributed optical fiber is arranged on the safety production equipment to be monitored, and the safety production equipment to be monitored is monitored through the distributed optical fiber to obtain pressure data, strain data and temperature data.

[0035] Specifically, through Brillouin optical time domain analysis (BOTDA) technology, based on the Brillouin scattering effect of light, a distributed optical fiber is selected, and the distributed optical fiber is spirally wound or surface-pasted on a key part of the safety production equipment to be monitored, such as a pressure vessel head weld or a pipe elbow; wherein the distributed optical fiber is an armored single-mode optical fiber. When light propagates in the optical fiber, photons collide with the optical fiber lattice inelastically, generating Brillouin scattered light with a frequency shift. According to the phonon theory, the frequency shift has a linear relationship with the strain and temperature of the environment in which the optical fiber is located, and this linear relationship is the core basis for realizing the measurement of the strain and temperature double parameters; The calculation formula of the frequency shift is: ; Wherein, is the frequency shift; is the reference frequency shift when there is no strain and no temperature change; is the strain sensitivity coefficient, which is determined by the photoelastic effect of the optical fiber material, and is about 0.05 MHz / με; is the strain of the optical fiber; is the temperature sensitivity coefficient, which is determined by the thermo-optic effect, and is about 1.0 MHz / ℃; is the current temperature measurement value; is the reference temperature; The Brillouin optical time domain analysis technology is used to realize distributed measurement, through the "pump-probe" double-beam scheme, the frequency difference between the pump light and the probe light is scanned to control the Brillouin gain spectrum, and the spatial position of the scattered light is determined by using the optical time domain reflection (OTDR) principle.

[0036] The pulsed laser emits narrow pulse width pump light into the sensing optical fiber, and the continuous probe light is injected from the other end of the optical fiber. The two and the backscattered Brillouin scattered light are coherent mixed, the photoelectric detector collects the beat frequency signal, and the frequency shift information is extracted by fast Fourier transform (FFT), and then the distance of the positioning point is calculated by combining the pulse transmission time, so as to realize the distributed measurement with a spatial resolution less than or equal to 1 meter.

[0037] The calculation formula of the distance of the positioning point is: ; Wherein, is the distance of the positioning point; is the speed of light; is the pulse transmission time; is the refractive index of the optical fiber; A single optical fiber can achieve continuous monitoring over tens of kilometers, with a strain measurement accuracy of ±2με and a temperature measurement accuracy of ±0.5℃, meeting the requirements of GB / T22577 "Fiber Optic Sensors". It realizes high-precision strain and temperature monitoring of continuous areas on the surface of the monitored production safety equipment, solves the monitoring blind spot problem of traditional point sensors, and captures early subtle deformations of the equipment (such as crack initiation and local stress concentration). The optical fiber has both sensing and communication functions, reducing external wiring and significantly improving system reliability.

[0038] By acquiring pressure, strain, and temperature data through distributed optical fibers, continuous spatial monitoring of equipment status is achieved. Distributed optical fiber sensing technology can provide comprehensive strain and temperature distribution information on the equipment surface, avoiding the potential monitoring blind spots of traditional point sensors. At the same time, optical fiber sensing has the advantages of anti-electromagnetic interference and corrosion resistance, making it suitable for long-term and stable operation in complex industrial environments. It provides a more comprehensive and reliable raw data basis for subsequent threshold setting and risk assessment, thereby improving the accuracy and stability of the entire monitoring system.

[0039] Threshold setting step S2: setting a dynamic threshold based on historical operation data, predicting the range of the dynamic threshold based on preset target process operation data and the historical operation data using a bidirectional long short-term memory network, and updating the range of the dynamic threshold in combination with real-time operation data to obtain an updated dynamic threshold; In some embodiments, the threshold setting step S2 is specifically as follows: The static threshold is set according to the actual standard of the preset equipment, and the dynamic threshold is set according to the static threshold and historical operation data through the box plot method and normal distribution test.

[0040] By distinguishing between the settings of static thresholds and dynamic thresholds, multi-level monitoring of equipment risks is achieved. The static threshold is set based on the equipment design standards to ensure the basic safety boundary of equipment operation, while the dynamic threshold is obtained through analysis of historical operation data. It can reflect the status changes of the equipment in actual operation, ensure the safety bottom line of equipment operation, and can be flexibly adjusted according to actual operating conditions, avoiding the false alarm or missed alarm problems that may be caused by a single threshold setting, and improving the accuracy and adaptability of risk monitoring.

[0041] Specifically, the static threshold is a rigid safety boundary based on standards, which can ensure that the equipment operating parameters do not exceed the mechanical limits of materials and industry regulations, and serve as the bottom line for risk warning. Static thresholds include strain thresholds, temperature thresholds, and pressure thresholds; The strain threshold is set according to GB / T150.1-2024 "Pressure Vessels Part 1: General Requirements", and the allowable strain is 0.2% of the material yield strength plastic strain criterion: ; wherein, is the allowable strain of the material; is the yield strength; is the elastic modulus of the material; For example, in Q345R steel material, is set to 345 MPa, and is set to the allowable strain of 206 GPa: .

[0042] The temperature threshold is set according to GB / T 4240-2021 “Stainless Steel Bar”, and the allowable temperature is taken as the 80% safety factor of the creep limit (10 million hour rupture strength) of the material: wherein, is the allowable temperature of the material; is the creep limit temperature of the material; For example, the creep limit allowable temperature of 304 stainless steel at 800°C is: .

[0043] The pressure threshold is set according to GB 150.1 “Pressure Vessels Part 1: General Requirements” and API 520 “Safety Valves and Bursting Discs”, and the pressure threshold is taken as 1.1 times the design pressure (non-flammable and explosive medium) or 1.05 times the design pressure (flammable and explosive medium), and does not exceed 0.8 times the burst pressure of the material; For example, an LPG storage tank made of Q345R steel material is a flammable and explosive medium, the design pressure is 1.6 MPa, and the burst pressure is 2.5 MPa, and the pressure threshold is: According to the design pressure: 1.6 MPa x 1.05 = 1.68 MPa; According to the burst pressure: 2.5 MPa x 0.8 = 2.0 MPa; Take the smaller value, and the final pressure threshold is 1.68 MPa.

[0044] When one or any combination of strain data, temperature data, or pressure data exceeds the static threshold, a red warning is triggered directly to ensure that the safety production equipment to be monitored does not enter a dangerous state.

[0045] In some embodiments, the dynamic threshold is based on the flexible fluctuation range of the data, and by setting the dynamic threshold, the complex working condition changes (such as medium composition fluctuation, environmental temperature and humidity change, equipment aging) of the chemical pressure-bearing equipment can be adapted, the normal operation fluctuation and abnormal trend can be accurately distinguished, the false positive rate can be reduced to below 15%, and the reliability of the early warning system can be improved.

[0046] wherein, the threshold setting step S2 is specifically: The abnormal values outside the preset multiple of the interquartile range are removed from the historical operation data by the box plot method to construct a data set; The data set is verified by normal distribution test to determine whether the data set conforms to the normal distribution; If the data set conforms to the normal distribution, the k value is configured based on the preset device type, the mean and standard deviation are calculated according to the k value, and the dynamic threshold is set according to the mean and standard deviation; Specifically, the box plot method is a method of identifying and removing abnormal values through data quantiles. When constructing a data set, the collected historical operation data is first sorted from small to large, and the value at the 25% position (Q1) and the value at the 75% position (Q3) are calculated. The difference between the two is the interquartile range (IQR).

[0047] The preset multiple is set to 3, and the data normal range boundary is calculated using the formula lower limit = Q1-k*IQR and upper limit = Q3+k*IQR. Data outside this upper and lower limit is identified as an abnormal value and is removed, and the remaining data constitutes the data set, providing reliable data for subsequent dynamic threshold calculation. The box plot method does not require data to conform to a specific distribution, can effectively exclude extreme data caused by device failure and sensor interference, and at the same time retains normal fluctuation data.

[0048] The data set is verified by normal distribution test to determine whether the data set conforms to the normal distribution; optionally, the normal distribution test is Shapiro-Wilk test, and the significance level a is set to 0.05; Verify whether the data set conforms to the normal distribution N(μ,σ 2 If the probability value p calculated by the test statistic is greater than the significance level a, the data set conforms to the normal distribution, and the data is confirmed to be usable; If the data set conforms to the normal distribution, the dynamic threshold is set based on the preset device type and the data set through the k value.

[0049] The historical operation data is processed by the box plot method and the normal distribution test to ensure the quality of the data used to set the dynamic threshold. The box plot method can effectively remove abnormal values and avoid the influence of abnormal data on threshold setting. The normal distribution test ensures that the statistical properties of the data meet the expectations, making the set dynamic threshold more scientific and reasonable, and truly reflecting the parameter fluctuation range of the device in the normal operating state, providing a reliable basis for subsequent risk assessment and improving the stability and credibility of the entire monitoring system.

[0050] In some embodiments, based on the normal distribution probability theory, the boundary between normal fluctuation and abnormality is quantified by the k value, the confidence interval is dynamically adjusted in combination with the preset device type, and the dynamic threshold is determined.

[0051] The threshold setting step S2 is specifically: If the device type is the first risk type, the k value is set to a first value, a first mean and a first standard deviation are calculated according to the first value, and the dynamic threshold is set according to the first mean and the first standard deviation; If the device type is the second risk type, the k value is set to a second value, a second mean and a second standard deviation are calculated according to the second value, and the dynamic threshold is set according to the second mean and the second standard deviation.

[0052] Specifically, when the device is a low-risk scenario such as a storage tank, the k value is set to 3, the 3σ principle is adopted, and 99.73% of the data under normal distribution is located in μ1±3σ1, balancing the monitoring sensitivity and operation efficiency; When the device is a high-risk scenario such as a high-pressure reactor, the k value is set to 5, the 5σ principle is adopted, and 99.99994% of the data under normal distribution is located in μ2±5σ2, to reduce the risk of false negatives and ensure the safety of critical equipment.

[0053] Learn from industrial control (such as SPC) and high-reliability fields (such as 6σ management) standards to achieve differentiated monitoring with low-risk efficiency first and high-risk safety first.

[0054] For example, a storage tank storing gasoline, the normal working pressure range is 0.5~0.8MPa, and the design pressure is 1.0MPa; The explanation for normal fluctuations is that the daytime ambient temperature rises, the gasoline in the tank volatilizes intensively, and the pressure may rise from 0.6MPa to 0.75MPa; the temperature drops at night, and the pressure drops to 0.55MPa. This small fluctuation (within ±0.15MPa) with temperature change is the normal fluctuation of the device operation, which is within the safety range.

[0055] The explanation for the abnormal boundary according to the k value is: The above-mentioned storage tank storing gasoline is a low-risk type, and the k value is set to 3. According to statistics, the normal pressure mean μ=0.6MPa, and the standard deviation σ=0.05MPa.

[0056] According to μ+3σ, the upper limit boundary is 0.6+3×0.05=0.75MPa; According to μ-3σ, the lower limit boundary is 0.6-3×0.05=0.45MPa.

[0057] When the pressure suddenly rises to 0.8MPa, exceeding the upper limit or drops to 0.4MPa, below the lower limit, it breaks the abnormal boundary, which may be tank leakage or safety valve failure, and the system will alarm; And for high-risk equipment such as high-pressure reactor, set k value to 5, assuming normal pressure mean μ = 5.0 MPa, standard deviation σ = 0.1 MPa.

[0058] The upper limit boundary is 5.0 + 5 x 0.1 = 5.5 MPa, and since even a small overpressure of high-pressure equipment can cause an explosion, an alarm is immediately reported once the upper limit boundary is exceeded.

[0059] By setting different k values according to the risk type of the equipment, the dynamic threshold is calculated, which realizes the differentiated monitoring of equipment of different risk levels. For low-risk equipment, a smaller k value can ensure basic safety while improving monitoring efficiency, and for high-risk equipment, a larger k value can ensure equipment safety with stricter standards, which can intelligently adapt to the safety needs of different types of equipment, optimize resource allocation while ensuring safety, and improve overall monitoring efficiency.

[0060] In some embodiments, the threshold setting step S2 further comprises: The range of the dynamic threshold in the preset future period of time is predicted by the bidirectional long short-term memory network, and a predicted dynamic threshold change range is obtained; optionally, by deploying the bidirectional long short-term memory network BiLSTM, the range of the threshold in the future 6 hours is predicted, and the prediction error is controlled within ± 3%, and the root mean square error RMSE is less than or equal to 0.05 MPa; The threshold updating step S24: within a preset period, the predicted dynamic threshold change range is updated according to the real-time running data, and an updated dynamic threshold is obtained; optionally, every 30 minutes, the mean μ and the variance σ are dynamically adjusted according to the real-time running data, and the predicted dynamic threshold change range is updated synchronously, adapting to the gradual aging of the equipment and the changes of the environment temperature and humidity.

[0061] The threshold range in the future period of time is predicted by the bidirectional long short-term memory network, and is updated in combination with the real-time running data, which realizes the dynamic adjustment of the threshold, can fully consider the historical change trend of the equipment running parameters, make reasonable prediction on the future state, and the periodic updating mechanism ensures that the threshold can timely reflect the latest running state of the equipment, so that the risk monitoring system can adapt to various changes of the equipment running state, avoid the inadaptability of the traditional fixed threshold method when the equipment ages or the working condition changes, and improve the accuracy and timeliness of the monitoring.

[0062] In some embodiments, the threshold setting step S2 further comprises: According to the target process operation data, the same type of operation samples in the historical running data are extracted; The dynamic threshold record of the safety production equipment to be monitored in the normal operation of the same type of operation samples is extracted, and the threshold change law is obtained according to the dynamic threshold record; According to the threshold change rule and the device parameters in the target process operation data, the range of the dynamic threshold value is predicted, and a predicted dynamic threshold value change range is obtained.

[0063] Specifically, the dynamic threshold value record of the safety production equipment to be monitored in normal operation is extracted from the same operation sample, and the threshold value change rule (such as the synchronous adjustment range of the upper limit of the threshold value in the feeding stage with the increase of the pressure) is determined. Combined with the device parameters (such as the aging coefficient and the structural strength parameter) in the target process operation data, the threshold value change rule is corrected, and the dynamic threshold value change range in the next 6 hours is obtained, so as to ensure covering the normal pressure fluctuation driven by the process.

[0064] For example, according to the target process operation data, the same operation sample is extracted from the historical database. The target process operation data includes process operations such as feeding, discharging and temperature rising, and associated parameters such as feeding amount, operation time and material properties. The conditions for extracting the same operation sample are: The process operation types are consistent (such as intermittent feeding); The working condition parameters are similar (the deviation of the environmental temperature and humidity is less than or equal to ±2℃, the deviation of the device aging coefficient is less than or equal to ±5%, and the deviation of the material viscosity is less than or equal to 10%); The time dimension is matched (including the complete pressure curve from 1 hour before operation to 5 hours after operation, covering the prediction period of 6 hours); The extracted same operation sample is statistically analyzed, and the pressure change characteristics in 0-6 hours under the target process operation are extracted, including: The average value and fluctuation range of the pressure in each stage (such as the pressure rising rate in the feeding stage for 0-2 hours and the pressure fluctuation amplitude in the stable stage for 2-6 hours); Based on the 95% confidence interval of the historical sample, the upper limit value and the lower limit value of the pressure change are determined.

[0065] According to the actual operation parameters (such as the actual feeding amount and the expected operation time) of the process operation, 30 groups of data most similar to the same operation sample are matched, and the pressure threshold value change range in the next 6 hours is predicted; wherein the real-time operation data including the actual operation parameters are obtained by distributed optical fiber.

[0066] The predicted pressure threshold value change range is compared with the calculated dynamic threshold value change range, and the state prediction result of the equipment in the next 6 hours is output.

[0067] By extracting similar operation samples in historical operation data and analyzing their characteristics, the variation law of the dynamic threshold for a specific process operation can be predicted, fully considering the influence of different process operations on equipment operation parameters. Through analysis of similar samples, the parameter variation trend under a specific operation can be more accurately predicted, avoiding prediction deviation caused by mixing data under different operation modes, making the adjustment of the dynamic threshold more accurate, better adapting to risk monitoring needs under various process operation conditions, and improving the adaptability of the system to different production scenarios.

[0068] In some embodiments, the threshold setting step S2 further comprises: calculating the deviation value of the real-time operation data and the predicted threshold variation range, judging whether to re-extract similar operation samples according to the deviation value, and updating the dynamic threshold variation range to obtain the updated dynamic threshold.

[0069] Specifically, the actual operation parameters (actual feed amount, operation rate) are compared with the production plan parameters in the target process operation data to determine the operation deviation (such as feed rate deviation, temperature control deviation).

[0070] Calculate the deviation value of the pressure data in the real-time operation data and the predicted pressure threshold variation range and the predicted pressure. If the deviation value is less than or equal to 5%, there is no need to re-extract similar operation samples, and only the pressure variation range in the subsequent stage needs to be fine-tuned. If the deviation value is greater than 5%, similar operation samples are re-extracted (preferably matching samples consistent with the current operation deviation), and the prediction result of the pressure threshold variation range is updated.

[0071] Based on the corrected pressure threshold variation range, the dynamic threshold variation range is adjusted by a preset proportion (such as 80% of the pressure prediction increase), ensuring that the threshold always adapts to the normal fluctuations driven by the process, and avoiding false positives.

[0072] By adjusting the threshold fluctuation range through the BiLSTM bidirectional long short-term memory network, the future trend is predicted to make the threshold more consistent with the actual operation state of the equipment.

[0073] For example, the pressure threshold in the static threshold of a certain storage tank is 0.5MPa to 0.8MPa, but in actual operation, the pressure will be affected by factors such as feed amount and environmental temperature. In the next 6 hours, the temperature may drop sharply due to heavy rain, and the pressure will probably be lower than 0.5MPa (which is a normal fluctuation caused by temperature drop, not a fault). At this time, if the static threshold is used, it will misreport an abnormality; while the BiLSTM bidirectional long short-term memory network prediction can adjust the threshold in advance to avoid false positives.

[0074] The pressure, ambient temperature, and feed quantity are input into a BiLSTM bidirectional long short-term memory network for prediction, and a pressure threshold change range for the next 6 hours (360 time steps) is output. For example, the pressure gradually decreases from 0.62 MPa to 0.60 MPa, and then to 0.58 MPa; the ambient temperature continuously decreases from 25 degrees Celsius to 22 degrees Celsius, and then to 18 degrees Celsius; and the feed quantity is stable at 50 m 3 / h, without mutation. The output pressure threshold change range is: the predicted mean is 0.52 MPa; the prediction error is ±3%, i.e. 0.52 x 3% ≈ 0.0156 MPa, combined with the root mean square error (RMSE) less than or equal to 0.05 MPa, the final prediction range is 0.48 MPa to 0.56 MPa, rather than the non-static 0.5 MPa to 0.8 MPa.

[0075] By comparing the deviation of real-time running data and predicted threshold range, and deciding whether to re-extract samples and update the threshold according to the deviation value, dynamic verification and correction of the prediction result are realized, the deviation between actual operation and expectation can be found in time, and the threshold is updated by re-matching samples or adjusting the proportion, ensuring that the threshold setting always keeps consistent with the actual running state of the equipment, effectively solving the problem of possible error accumulation, avoiding risk misjudgment caused by prediction deviation, and improving the adaptive ability and reliability of risk monitoring.

[0076] Risk assessment step S3: calculate a comprehensive risk index according to the updated dynamic threshold and real-time running data by the analytic hierarchy process, evaluate the risk of the safety production equipment to be monitored according to the comprehensive risk index, and obtain an evaluation result.

[0077] By obtaining historical running data and real-time running data and setting a dynamic threshold range, using a bidirectional long short-term memory network to predict and update the threshold range, and then calculating a comprehensive risk index by the analytic hierarchy process for evaluation, dynamic monitoring and evaluation of the safety production equipment risk are realized, the actual running state of the equipment can be more accurately reflected, the defect that the traditional fixed threshold monitoring method cannot adapt to the change of the equipment running state is avoided, and through real-time data updating and risk assessment, potential risks can be found in time and corresponding measures can be taken, improving the safety and reliability of equipment operation.

[0078] In some embodiments, the risk assessment step S3 specifically includes: calculating a risk index according to the updated dynamic threshold and real-time running data; obtaining a comprehensive index by weighted calculation of a preset weight and the risk index; obtaining a corresponding risk level according to the comprehensive index, and responding to the safety production equipment to be detected according to the risk level.

[0079] Specifically, experts in the fields of equipment operation and maintenance, process safety, and fault analysis directly score the importance of pressure data, strain data, and temperature data, and then average or weightedly aggregate the scores to obtain a weight of 0.4 for pressure data, a weight of 0.3 for temperature data, and a weight of 0.3 for strain data.

[0080] According to the updated dynamic threshold and real-time operation data, a pressure risk index, a strain risk index, and a temperature risk index are calculated, and the calculation formulae are as follows: wherein, is the pressure risk index; is the temperature risk index; is the strain risk index; is real-time operation data; is pressure data; is temperature data; is strain data; is an updated dynamic threshold; According to the above calculation formulae, the pressure risk index, the strain risk index, and the temperature risk index are converted into risk scores of 0 to 100; The calculation formula of the comprehensive index is as follows: wherein, is the comprehensive index; is the weight of pressure data; is the weight of temperature data; is the weight of strain data; According to the above calculation formula of the comprehensive index, a quantitative evaluation of 0 to 100 for the safety production equipment risk is finally realized; wherein, when , the corresponding risk level is a safe state, and the response strategy adopted is to perform routine monitoring; when , the corresponding risk level is a yellow pre-warning, and the response strategy adopted is to encrypt data collection and perform trend analysis; when , the corresponding risk level is an orange pre-warning, and the response strategy adopted is to start local shutdown and perform expert diagnosis; when , the corresponding risk level is a red pre-warning, and the response strategy adopted is to urgently shut down the process and start an emergency plan; ​​The comprehensive index is obtained by weighting the risk indicators and preset weights, and the risk level and response measures are determined according to the comprehensive index, so as to realize the quantitative evaluation and hierarchical response of the equipment risk. The risk evaluation is more comprehensive and objective by considering the risk status of a single parameter and reflecting the influence degree of different parameters on the overall risk through weight distribution. According to the range of the comprehensive index, the risk level is determined and the corresponding measures are taken, forming a complete risk response mechanism and providing a more scientific decision basis for equipment safety management.

[0081] In some embodiments, the safety production equipment risk monitoring method also provides human-computer interaction, remote monitoring and intelligent operation and maintenance functions. Among them, in terms of human-computer interaction, based on WebGL technology, a digital twin model of equipment is constructed using Three.js engine, and real-time rendering of 100,000 monitoring points is realized through GPU acceleration, with frame rate greater than or equal to 60FPS, ensuring smooth interface.

[0082] The risk level is visually coded using colors in accordance with GB / T 2893.1-2015 "Graphic Symbols-Safety Colors and Safety Signs" (yellow #FFEB3B represents lower risk, orange #FFC107 represents medium risk, and red #F44336 represents high risk), and real-time monitoring data, risk level and fault location are displayed to provide intuitive and clear equipment operation status information for workers.

[0083] In terms of remote monitoring, real-time operation data is pushed to the cloud platform through MQTT protocol, and mobile phone APP and PC remote viewing and control are supported.

[0084] No matter where the workers are, they can master the equipment operation status in real time through mobile devices or computers, realize remote management, and improve work efficiency and flexibility of equipment management.

[0085] In terms of intelligent operation and maintenance, a state space S containing operation parameters, operation time, historical maintenance records and other information is constructed, an action space A covering inspection, lubrication, component replacement and other operations is constructed, and a reward function R with maintenance cost reduction and downtime reduction as the target is constructed, forming a Markov decision process model, which is trained using deep Q network DQN, the experience replay buffer capacity is set to 100,000, and the objective function is:

[0086] Among them, is the objective function; is the expectation operator; is the immediate reward; is the discount factor, and its value is set to 0.95; is the next action; is the state; is a current state; is a current action; Through continuous training optimization, the maintenance cost is reduced by more than 30%, and the equipment operation and maintenance efficiency and economy are effectively improved.

[0087] As Figure 2 shown, the embodiment of the present application also provides a safety production equipment risk monitoring system, which is used to realize the safety production equipment risk monitoring method of the above-mentioned embodiment. The safety production equipment risk monitoring system comprises: A data acquisition module 1 is configured to acquire historical operation data and real-time operation data of the safety production equipment to be monitored. A threshold setting module 2 is configured to set a dynamic threshold according to the historical operation data, to predict a range of the dynamic threshold according to preset target process operation data and the historical operation data by using a bidirectional long short-term memory network, to update the range of the dynamic threshold in combination with real-time operation data, and to obtain an updated dynamic threshold. A risk assessment module 3 is configured to calculate a comprehensive risk index according to the updated dynamic threshold and the real-time operation data by using an analytic hierarchy process, to evaluate the risk of the safety production equipment to be monitored according to the comprehensive risk index, and to obtain an evaluation result.

[0088] The modular design realizes the functional integration of data acquisition, threshold setting and risk assessment, forms a complete risk monitoring system, the data acquisition module is responsible for collecting the basic data of equipment operation, the threshold setting module realizes the dynamic prediction and update of the threshold, and the risk assessment module completes the calculation and evaluation of the comprehensive risk, so that the functions of the system are clear and cooperate closely, the integrity of the system is ensured, and the independent optimization and maintenance of each part are facilitated, which provides an efficient and reliable technical platform for the risk monitoring of safety production equipment and can meet the safety monitoring requirements in different industrial scenes.

[0089] It should be noted that the above is a reference mode of a safety production equipment risk monitoring method, and the present application is not limited thereto.

[0090] The embodiments of the present invention realize dynamic monitoring and assessment of risks of production safety equipment, can more accurately reflect the actual operating status of the equipment, avoid the defect of traditional fixed threshold monitoring methods that cannot adapt to changes in the operating status of the equipment, and at the same time, through real-time data updates and risk assessments, can timely discover potential risks and take corresponding measures, thereby improving the safety and reliability of equipment operation, and solving the problems of existing point sensors that cannot achieve continuous monitoring, have blind spots, and have poor anti-interference capabilities; fixed threshold systems are difficult to distinguish between normal operating fluctuations and real faults, and have a high false alarm rate; at the same time, traditional methods lack multi-source data fusion and trend prediction capabilities, resulting in delayed risk warnings and an inability to meet the needs of modern industry for precise and intelligent safety monitoring.

[0091] Finally, it should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other. The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the same. Although the present invention has been described in detail with reference to preferred embodiments, persons skilled in the art should understand that the specific implementation methods of the present invention may still be modified or some technical features may be replaced by equivalents without departing from the spirit of the technical solutions of the present invention, and all of these should fall within the scope of the technical solutions claimed for protection by the present invention.

Claims

1. A method for monitoring risks of production safety equipment, characterized in that: The steps include: Data acquisition step: obtaining historical and real-time operating data of the production safety equipment to be monitored; Threshold setting step: setting a dynamic threshold based on the historical operation data, predicting a range of the dynamic threshold based on preset target process operation data and the historical operation data using a bidirectional long short-term memory network, and updating the range of the dynamic threshold in combination with the real-time operation data to obtain an updated dynamic threshold; Risk assessment step: calculating a comprehensive risk index based on the updated dynamic threshold and the real-time operation data by using the hierarchical analysis method, and assessing the risk of the production safety equipment to be monitored based on the comprehensive risk index to obtain an assessment result.

2. The method for monitoring production safety equipment risks according to claim 1, characterized in that: The threshold setting steps are specifically as follows: A static threshold is set according to the actual standard of the preset equipment, and a dynamic threshold is set according to the static threshold and the historical operation data through a box plot method and a normal distribution test.

3. The method for monitoring production safety equipment risks according to claim 2, characterized in that: The threshold setting steps are specifically as follows: Outliers outside the interquartile range of a preset multiple of the historical operating data are eliminated by a box plot method to construct a data set; Verifying the data set through a normal distribution test to determine whether the data set conforms to a normal distribution; If the data set conforms to a normal distribution, a k value is configured based on a preset device type, a mean and a standard deviation are calculated according to the k value, and the dynamic threshold is set according to the mean and the standard deviation.

4. The method for monitoring production safety equipment risks according to claim 3, characterized in that: The threshold setting steps are specifically as follows: If the device type is a first risk type, setting the k value to a first value, calculating a first mean and a first standard deviation based on the first value, and setting the dynamic threshold based on the first mean and the first standard deviation; If the device type is the second risk type, the k value is set to a second value, a second mean and a second standard deviation are calculated based on the second value, and the dynamic threshold is set based on the second mean and the second standard deviation.

5. The method for monitoring production safety equipment risks according to claim 4, characterized in that: The threshold setting step further includes: Predicting the range of the dynamic threshold for a preset future time period using the bidirectional long short-term memory network to obtain a predicted dynamic threshold variation range; Within a preset period, the predicted dynamic threshold variation range is updated according to the real-time operation data to obtain an updated dynamic threshold.

6. The method for monitoring production safety equipment risks according to claim 5, characterized in that: The threshold setting step further includes: Extracting similar operation samples from the historical operation data according to the target process operation data; Extracting dynamic threshold records of the production safety equipment to be monitored during normal operation from the similar operation samples, and obtaining a threshold change rule based on the dynamic threshold records; The range of the dynamic threshold is predicted according to the threshold variation rule and the equipment parameters in the target process operation data to obtain a predicted dynamic threshold variation range.

7. The method for monitoring production safety equipment risks according to claim 6, characterized in that: The threshold setting step further includes: Calculate the deviation between the real-time operation data and the predicted threshold value variation range, determine whether to re-extract similar operation samples based on the deviation value, and update the dynamic threshold value variation range to obtain an updated dynamic threshold value.

8. The method for monitoring production safety equipment risks according to claim 7, characterized in that: The risk assessment steps specifically include: Calculating a risk indicator based on the updated dynamic threshold and the real-time operation data; A comprehensive index is obtained by weighted calculation of the preset weight and the risk index; A corresponding risk level is obtained according to the comprehensive indicator, and a response is made to the production safety equipment to be inspected according to the risk level.

9. The method for monitoring production safety equipment risks according to any one of claims 1 to 8, characterized in that: The historical operation data or real-time operation data includes pressure data, strain data and temperature data, and the data acquisition step specifically includes: A distributed optical fiber is set on the safety production equipment to be monitored, and the safety production equipment to be monitored is monitored through the distributed optical fiber to obtain the pressure data, the strain data and the temperature data.

10. A safety production equipment risk monitoring system, characterized in that: include: Data acquisition module: obtains historical and real-time operating data of the safety production equipment to be monitored; Threshold setting module: sets a dynamic threshold based on the historical operation data, predicts the range of the dynamic threshold based on preset target process operation data and the historical operation data through a bidirectional long short-term memory network, and updates the range of the dynamic threshold in combination with the real-time operation data to obtain an updated dynamic threshold; Risk assessment module: Calculates a comprehensive risk index based on the updated dynamic threshold and the real-time operation data through the hierarchical analysis method, and assesses the risk of the production safety equipment to be monitored based on the comprehensive risk index to obtain an assessment result.

Images