Intelligent early warning method for falling risk of high-altitude operation in railway station canopy transformation

Abstract

The invention provides an intelligent early warning method for high-altitude operation falling risks of railway station canopy transformation. Comprising the following steps: inputting canopy component parameters, reference values and weight coefficients, and starting all sensors to carry out data calibration so as to remove abnormal values; the real-time environment temperature, the cumulative sunshine duration of the day, the component surface humidity and the real-time wind speed of an operation point are collected through the multi-dimensional environment sensing module, and the component actual measurement thickness and the component vibration speed are collected through the canopy component state monitoring module. The total load, the actual operation height and the single-time operation duration of the operation platform are collected through the operation platform state monitoring module, and the respiratory frequency variation coefficient of the operation personnel is collected through the personnel physiology monitoring module; inputting the acquired real-time environment temperature, the accumulated sunshine duration on the day, the surface humidity of the component, the actually measured thickness of the component and the vibration speed of the component into an intelligent early warning calculation module; reliable safety guarantee can be provided for railway station canopy transformation aloft work in regions with remarkable temperature differences.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk early warning, and particularly relates to a railway station canopy reconstruction high-altitude operation falling risk intelligent early warning method. BACKGROUND

[0002] In the railway station canopy reconstruction project, high-altitude operation falling risk early warning is a key link to ensure construction safety. At present, the mainstream early warning scheme in the industry is mainly designed around conventional environmental and working condition parameters such as wind speed and operation height, and risk prompts are realized by setting fixed thresholds. Such schemes can meet the basic early warning needs under conventional meteorological conditions and have been widely used in various high-altitude operation scenarios. However, for the specific scenario of railway station canopy reconstruction, the existing scheme has significant technical defects: it does not consider the deformation of canopy metal components due to changes in environmental temperature and the cascading risk between the friction coefficient of the operation platform. In particular, in regions such as Shanxi where the diurnal temperature difference is large and the seasonal extreme temperature is obvious, low temperatures in winter easily lead to the shrinkage deformation of canopy steel components, while summer sun exposure causes the components to expand, both types of deformation change the contact state of the operation platform and the component support point, directly leading to the attenuation of the contact friction coefficient. The existing scheme only relies on the initial calibrated friction coefficient for risk assessment and does not dynamically associate the influence of deformation on the friction state, making the platform sliding risk severely underestimated. At the same time, the existing scheme ignores the coupling effect of the physiological state of the operating personnel and the environmental risk under low temperature, strong wind and other environmental factors, further leading to a significant decline in early warning accuracy in extreme scenarios, frequent false positives and false negatives, and the inability to provide reliable safety protection for canopy reconstruction high-altitude operations, Based on the above problems, an intelligent early warning technical scheme is needed to solve the cascading risk assessment deficiency. SUMMARY

[0003] The purpose of the present application is to solve the shortcomings in the prior art and to propose a railway station canopy reconstruction high-altitude operation falling risk intelligent early warning method, which comprises: Enter the canopy component parameters, reference values and weight coefficients, start all sensors for data calibration to eliminate abnormal values; Collect real-time environmental temperature, daily cumulative sunshine duration, component surface humidity and operation point real-time wind speed through the multi-dimensional environmental perception module, collect component measured thickness and component vibration speed through the canopy component state monitoring module, collect operation platform total load, actual operation height and single operation duration through the operation platform state monitoring module, and collect the respiratory rate variation coefficient of the operating personnel through the personnel physiological monitoring module; Input the collected real-time environmental temperature, daily cumulative sunshine duration, component surface humidity, component measured thickness and component vibration speed into the intelligent early warning calculation module, and calculate the canopy metal component temperature deformation coefficient by the intelligent early warning calculation module; The contact friction coefficient between the operation platform and the canopy truss is calculated in combination with the temperature deformation coefficient of the canopy metal member, the total load of the operation platform, and the member vibration speed, and the falling risk early warning index is calculated in combination with the contact friction coefficient, the coefficient of variation of the operation personnel's breathing frequency, the real-time wind speed at the operation point, the actual operation height, and the single operation time length; The hierarchical early warning stage triggers the corresponding level of early warning action according to the falling risk early warning index; The closed-loop optimization stage stores the early warning data and actual risk events of the day, and corrects the weight coefficient by a gradient descent algorithm to improve the subsequent early warning accuracy.

[0004] Preferably, the canopy member parameters entered in the data preprocessing stage include the thermal expansion coefficient of the canopy member material and the member corrosion amount, the reference values include the member reference temperature, the low temperature threshold, and the basic friction coefficient, and the weight coefficients include the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-operation time weight coefficient, and the low temperature compensation coefficient. The sensor calibration duration is set to 10 minutes, and the out-of-range data caused by sensor failure is excluded during calibration; The thermal expansion coefficient of the canopy member material is determined according to the actual metal material used in the canopy, the member corrosion amount is obtained by member detection before operation and entered into the system, the member reference temperature is set to 25℃, the low temperature threshold is set to 5℃ according to the conventional low temperature value in winter in Shanxi, and the basic friction coefficient is calibrated to 0.6 according to the steel-steel dry contact characteristics experiment.

[0005] Further preferably, the multi-dimensional environment perception module comprises an infrared temperature sensor, a photosensitive sensor, a capacitive humidity sensor, and a miniature wind speed sensor, the infrared temperature sensor is used to collect the real-time environment temperature, the photosensitive sensor is used to accumulate and count the daily cumulative sunshine duration, the capacitive humidity sensor is used to collect the component surface humidity, and the miniature wind speed sensor is used to collect the real-time wind speed at the work point; the canopy component state monitoring module comprises an optical fiber displacement sensor and a piezoelectric vibration sensor, the optical fiber displacement sensor is used to collect the measured thickness of the component, and the piezoelectric vibration sensor is used to collect the vibration speed of the component; the work platform state monitoring module comprises a pressure sensor, a laser ranging sensor, and a timing unit, the pressure sensor is used to collect the total load of the work platform, the laser ranging sensor is used to collect the actual working height, and the timing unit is used to count the single working time; the personnel physiological monitoring module adopts a non-contact millimeter wave radar, and the non-contact millimeter wave radar is used to collect the respiratory rate variation coefficient of the worker; the intelligent early warning calculation module adopts an STM32H743 embedded chip and carries an edge computing algorithm, and the edge computing algorithm is used to perform calculation of the canopy metal component temperature deformation coefficient, the contact friction coefficient, and the falling risk early warning index in the dynamic calculation stage.

[0006] Further preferably, the early warning actions triggered in the hierarchical early warning stage include a first-level early warning, a second-level early warning, and a safe state; in the first-level early warning, a high-frequency alarm signal is sent out through an audible and visual alarm, the work platform is controlled to stop running through a work platform emergency stop controller, and work point positioning data and real-time risk data are sent to a management and control center through a 4G / 5G communication module; in the second-level early warning, a medium-frequency alarm signal is sent out through the audible and visual alarm, and a speed reduction work prompt information is output to the worker at the same time; in the safe state, a green indicator light is kept on, and the real-time environment temperature, the component surface humidity, the component vibration speed, and the falling risk early warning index calculated are uploaded to a storage unit for archiving every 5 minutes through the 4G / 5G communication module; the frequency of the high-frequency alarm signal is set to 2 Hz, and the frequency of the medium-frequency alarm signal is set to 1 Hz.

[0007] Further preferably, the canopy metal component temperature deformation coefficient is calculated by the following formula: ; Wherein, λ m represents the thermal expansion coefficient of the canopy component material; T real represents the real-time environment temperature, T ref represents the component reference temperature, t sun represents the daily cumulative sunshine duration, d comp represents the measured thickness of the component, and δrust represents the amount of member rust, k env represents the humidity correction coefficient, H surf represents the member surface humidity; the humidity correction coefficient is calibrated as 0.02 according to the dry environment characteristics of Shanxi region.

[0008] Further preferably, the contact friction coefficient of the operation platform and the canopy truss is calculated by the following formula: ; wherein, μ base represents the base friction coefficient, k α represents the deformation influence coefficient, α T represents the canopy metal member temperature deformation coefficient, k F represents the load correction coefficient, F load represents the total load of the operation platform, k H represents the humidity attenuation coefficient, H surf represents the member surface humidity, v vib represents the vibration speed of the canopy member; the deformation influence coefficient is calculated by α T The attenuation curve fitting of the friction coefficient is 120, the load correction coefficient is set as 0.03, and the humidity attenuation coefficient is set as 0.015.

[0009] Further preferably, the falling risk early warning index is calculated by the following formula: ; wherein, A represents the friction-height weight coefficient, μ contact represents the contact friction coefficient of the operation platform and the canopy truss, h work represents the actual operation height, B represents the physiological-wind speed weight coefficient, C Vresp represents the operation personnel breathing frequency variation coefficient, v wind represents the operation point real-time wind speed, C represents the deformation-operation time length weight coefficient, α T represents the canopy metal member temperature deformation coefficient, t work represents the single operation time length, D represents the low temperature compensation coefficient, T real represents the real-time environmental temperature, T low represents the low temperature threshold, R warn represents the falling risk early warning index; the friction-height weight coefficient is set as 0.85, the physiological-wind speed weight coefficient is set as 0.72, the deformation-operation time length weight coefficient is set as 0.45, and the low temperature compensation coefficient is set as 0.05.

[0010] Further preferably, the closed-loop optimization stage stores the daily early warning data, including the real-time environmental temperature collected by the multi-dimensional environmental perception module, the daily cumulative sunshine duration, the component surface humidity, and the real-time wind speed at the work point, the component measured thickness and the component vibration speed collected by the canopy component state monitoring module, the total load of the work platform, the actual work height, and the single work duration collected by the work platform state monitoring module, the work personnel respiratory frequency coefficient of variation collected by the personnel physiological monitoring module, and the canopy metal component temperature deformation coefficient, the contact friction coefficient, and the falling risk early warning index obtained by the dynamic calculation stage; the actual risk events include component abnormal vibration records, work personnel physiological state abnormal records, and work platform operation state abnormal records during work; the gradient descent algorithm iteratively corrects the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-work duration weight coefficient, and the low-temperature compensation coefficient by minimizing the deviation value of the falling risk early warning index calculated value and the actual risk event corresponding to the day; the corrected weight coefficients are automatically updated to the intelligent early warning calculation module for coefficient and index calculation in the next day's dynamic calculation stage.

[0011] Further preferably, all data collection operations of the dynamic calculation stage are executed in a 1-second cycle; the infrared temperature sensor collects the real-time environmental temperature 3 times per cycle and takes the arithmetic mean value as the value of Treal, the photosensitive sensor updates the daily cumulative sunshine duration according to the light intensity, the capacitive humidity sensor collects the component surface humidity once per cycle, and the micro wind speed sensor collects the real-time wind speed at the work point 5 times per cycle and takes the effective value as the value of vwind; the optical fiber displacement sensor collects the component measured thickness 2 times per cycle, and the piezoelectric vibration sensor collects the component vibration speed once per cycle; the pressure sensor collects the total load of the work platform once per cycle, the laser ranging sensor collects the actual work height 2 times per cycle and takes the arithmetic mean value as the value of hwork, and the timing unit updates the single work duration according to the time increment; the non-contact millimeter wave radar collects the work personnel respiratory frequency coefficient of variation once per cycle; the normal value range of the work personnel respiratory frequency coefficient of variation is set to 5% to 15%, and greater than 20% is determined as a physiological state abnormality.

[0012] Further preferably, the falling risk early warning index threshold value adopted by the hierarchical early warning stage is calibrated by pilot data of high-altitude operations of canopy reconstruction at three different railway stations in Shanxi Province; the falling risk early warning index threshold value corresponding to the first-level early warning is set to be greater than 80, the falling risk early warning index threshold value corresponding to the second-level early warning is set to be greater than 60 and less than or equal to 80, and the falling risk early warning index threshold value corresponding to the safe state is set to be less than or equal to 60; the pilot data includes environmental data such as the real-time environmental temperature, the component surface humidity, and the component vibration speed in different seasons and at different time periods, component state data such as the component actual thickness, personnel physiological data such as the respiratory frequency variation coefficient of the operating personnel, and actual operation risk records; during the calibration process, the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-operation time weight coefficient, and the low-temperature compensation coefficient are adjusted so that the early warning accuracy rate is greater than 92% and the false positive rate is controlled to be less than 11.5%.

[0013] Technical effects: The present application collects data through multi-module linkage, constructs temperature and deformation coefficients, contact friction coefficients, and progressive calculation logic of falling risk early warning index, and further adds a mechanism for closed-loop optimization of weight coefficients, accurately solves the core problem of missing risk assessment of the interlocking of temperature and deformation of canopy metal components and the contact friction coefficient of the operation platform in the background technology, and simultaneously correlates personnel physiological state and environmental risk to avoid the situation that the friction coefficient is underestimated in extreme scenarios, greatly improves the early warning accuracy rate, reduces false positives and omissions, and provides reliable and safe protection for high-altitude operations of railway station canopy reconstruction in regions with significant temperature differences. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The present application is a railway station canopy reconstruction high-altitude operation falling risk intelligent early warning method flowchart. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0016] Traditional railway station canopy reconstruction high-altitude operation early warning only focuses on conventional factors such as wind speed and operation height, lacks multi-dimensional data cooperative collection and progressive risk calculation, and has no closed-loop optimization mechanism, resulting in low early warning accuracy rate and high false positive rate in extreme scenarios.

[0017] Based on this, please refer to Figure 1 , the railway station canopy reconstruction high-altitude operation falling risk intelligent early warning method, comprising: S1: Enter the canopy component parameters, reference values and weight coefficients, start all sensors to calibrate data to eliminate abnormal values; S2: Collect real-time environmental temperature, daily cumulative sunshine duration, component surface humidity and working point real-time wind speed through the multi-dimensional environment perception module, collect component measured thickness and component vibration speed through the canopy component state monitoring module, collect working platform total load, actual working height and single working time through the working platform state monitoring module, and collect the working personnel respiratory frequency variation coefficient through the personnel physiological monitoring module; S3: Input the collected real-time environmental temperature, daily cumulative sunshine duration, component surface humidity, component measured thickness and component vibration speed into the intelligent early warning calculation module, and calculate the canopy metal component temperature deformation coefficient by the intelligent early warning calculation module; S4: Calculate the contact friction coefficient of the working platform and the canopy truss in combination with the canopy metal component temperature deformation coefficient, the working platform total load and the component vibration speed, and calculate the falling risk early warning index in combination with the contact friction coefficient, the working personnel respiratory frequency variation coefficient, the working point real-time wind speed, the actual working height and the single working time; S5: The grading early warning stage triggers the corresponding level of early warning action according to the falling risk early warning index; S6: The closed-loop optimization stage stores the daily early warning data and actual risk events, and corrects the weight coefficients by the gradient descent algorithm to improve the subsequent early warning accuracy.

[0018] The technical scheme builds a complete early warning logic through four stages: in the data preprocessing stage, the basic parameters are first determined and the sensors are calibrated to avoid the influence of initial data errors on subsequent calculations; in the dynamic calculation stage, data is collected through the linkage of environment, component, platform and personnel modules, and then calculated in a progressive manner according to the temperature deformation coefficient, contact friction coefficient and falling risk early warning index to ensure the quantitative correlation of hidden factors; in the grading early warning stage, differentiated actions are triggered according to the index to avoid response confusion of a single alarm mode; in the closed-loop optimization stage, the weight coefficients are corrected by the gradient descent algorithm to realize continuous iteration of the early warning capability. The whole scheme breaks the traditional static threshold early warning mode, integrates material mechanics, physiology and intelligent algorithms, and forms a multi-dimensional and dynamic early warning system.

[0019] The technical effects of the above embodiments include covering multiple implicit risk factors, improving early warning accuracy, reducing false alarm rate and realizing continuous optimization of early warning capability.

[0020] The existing early warning scheme does not explicitly indicate the specific content and acquisition method of the canopy component parameters, the reference value and the weight coefficient in the data preprocessing stage, and there is no unified standard for sensor calibration, which leads to insufficient accuracy of the basic data and directly affects the reliability of the subsequent risk calculation results.

[0021] Therefore, the canopy component parameters entered in the data preprocessing stage include the thermal expansion coefficient of the canopy component material and the component corrosion amount, the reference value includes the component reference temperature, the low temperature threshold and the basic friction coefficient, and the weight coefficient includes the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-operation time weight coefficient and the low temperature compensation coefficient; the sensor calibration time is set to 10 minutes, and the out-of-range data caused by sensor failure is removed during calibration; the thermal expansion coefficient of the canopy component material is determined according to the actual metal material used in the canopy, the component corrosion amount is obtained by detecting the component before operation and is entered into the system, the component reference temperature is set to 25℃, the low temperature threshold is set to 5℃ according to the regular low temperature value in winter in Shanxi, and the basic friction coefficient is calibrated to 0.6 according to the steel-steel dry contact characteristic experiment.

[0022] The technical scheme focuses on the parameter specification and calibration rigor in the data preprocessing stage: first, the specific composition of the three types of core parameters is explicitly indicated to avoid calculation deviation caused by parameter ambiguity; in the canopy component parameters, the thermal expansion coefficient of the material needs to match the actual metal to ensure that it fits the real physical properties of the canopy component, and the component corrosion amount is obtained by detecting before operation instead of estimation to avoid the interference of corrosion on component deformation calculation; the reference value setting combines the regional characteristics of Shanxi, the low temperature threshold of 5℃ corresponds to the regular low temperature in winter in Shanxi, the component reference temperature of 25℃ is the standard environment temperature calibrated in the laboratory, and the basic friction coefficient of 0.6 is obtained through the steel-steel dry contact experiment to ensure the accuracy of the initial value of the friction calculation; the weight coefficient covers friction-height, physiological-wind speed, deformation-time and low temperature compensation to provide differentiated weight basis for subsequent risk index calculation. At the same time, the sensor calibration is set to 10 minutes, which not only ensures the sufficiency of calibration, but also avoids the time-consuming affecting the operation efficiency, and the out-of-range data is explicitly removed to further ensure the effectiveness of the collected data.

[0023] The technical effects of the above embodiments include: explicitly indicating the source of the preprocessing parameters and the calibration standard to ensure the accuracy of the basic data and lay a reliable foundation for subsequent risk calculation.

[0024] The existing early warning system does not explicitly indicate the hardware composition, sensor type and data interaction relationship of each module, which leads to the fact that the scheme cannot be implemented, and there is a lack of non-contact physiological monitoring means, which is easy to interfere with the operation of the operator.

[0025] Based on this, the multi-dimensional environment perception module includes an infrared temperature sensor, a photosensitive sensor, a capacitive humidity sensor, and a micro wind speed sensor, the infrared temperature sensor is used to collect the real-time environment temperature, the photosensitive sensor is used to accumulate the daily cumulative sunshine duration, the capacitive humidity sensor is used to collect the component surface humidity, and the micro wind speed sensor is used to collect the real-time wind speed at the work point; the canopy component state monitoring module includes an optical fiber displacement sensor and a piezoelectric vibration sensor, the optical fiber displacement sensor is used to collect the measured thickness of the component, and the piezoelectric vibration sensor is used to collect the vibration speed of the component; the work platform state monitoring module includes a pressure sensor, a laser ranging sensor, and a timing unit, the pressure sensor is used to collect the total load of the work platform, the laser ranging sensor is used to collect the actual work height, and the timing unit is used to count the single work duration; the personnel physiological monitoring module adopts a non-contact millimeter wave radar, and the non-contact millimeter wave radar is used to collect the respiratory rate variation coefficient of the worker; the intelligent early warning calculation module adopts an STM32H743 embedded chip and carries an edge computing algorithm, and the edge computing algorithm is used to perform the calculation of the canopy metal component temperature deformation coefficient, the contact friction coefficient, and the falling risk early warning index in the dynamic calculation stage.

[0026] The technical scheme realizes the specific implementation mode and data flow of each module through the cooperative design of hardware and software: the multi-dimensional environment perception module selects four types of special sensors, the infrared temperature sensor ensures the real-time environment temperature collection accuracy, the photosensitive sensor reflects the influence of sunshine on the component by accumulating the illumination time, the capacitive humidity sensor accurately captures the component surface humidity, and the micro wind speed sensor focuses on the local wind speed at the work point; the canopy component state monitoring module adopts an optical fiber displacement sensor and a piezoelectric vibration sensor; the work platform state monitoring module realizes real-time load change sensing through a pressure sensor, accurately measures the work height through a laser ranging sensor, and counts the work duration through a timing unit; the personnel physiological monitoring module adopts a non-contact millimeter wave radar, which can collect the respiratory rate variation coefficient without contacting the worker, avoiding cable connection interference operation.

[0027] The intelligent early warning calculation module selects an STM32H743 embedded chip, the computing power is adapted to the edge computing demand, the carried edge algorithm directly performs the calculation of the three core coefficients, ensures the real-time data processing, and the data collected by each module is transmitted to the calculation module, forming a complete hardware link of collection and calculation.

[0028] The technical effects of the above embodiment include: the hardware composition and interaction logic of each module are clear, non-contact physiological monitoring is realized, the scheme can be implemented, and the real-time data collection and calculation are improved.

[0029] The existing hierarchical early warning only simply distinguishes early warning and a safe state, and does not clearly indicate specific device actions, communication modes and data archiving mechanisms corresponding to different early warning levels, resulting in chaotic early warning responses and the inability to trace historical early warning data, which is not conducive to subsequent analysis and optimization.

[0030] Based on this, the early warning actions triggered in the hierarchical early warning stage include a first-level early warning, a second-level early warning and a safe state; in the first-level early warning, a high-frequency alarm signal is sent out through a sound-light alarm, a work platform is stopped from running through a work platform emergency stop controller, and work point positioning data and real-time risk data are sent to a control center through a 4G / 5G communication module; in the second-level early warning, a medium-frequency alarm signal is sent out through the sound-light alarm, and a speed reduction work prompt is output to a worker; in the safe state, a green indicator light is kept on, and the real-time environmental temperature, the member surface humidity, the member vibration speed and the calculated falling risk early warning index collected every 5 minutes are uploaded to a storage unit for archiving; the frequency of the high-frequency alarm signal is set to 2Hz, and the frequency of the medium-frequency alarm signal is set to 1Hz.

[0031] The technical scheme constructs a differentiated and executable hierarchical early warning system, and clearly indicates device linkage and data management in each state: the first-level early warning corresponds to a high-risk scene, a 2Hz high-frequency sound-light alarm signal is used, a work platform emergency stop controller is used to directly control the platform to stop, and a 4G / 5G communication module is used to send positioning and real-time risk data to a control center; the second-level early warning corresponds to a medium-risk scene, a 1Hz medium-frequency sound-light alarm signal is used, and a speed reduction work prompt is output; in the safe state, a green indicator light is kept on, and key data is uploaded to a storage unit every 5 minutes, which not only avoids waste of communication resources caused by real-time uploading, but also ensures that historical data can be traced, thereby providing data support for closed-loop optimization. Each early warning action is associated with a specific hardware device, and the signal frequency, data uploading content and period are clearly indicated, thereby avoiding ambiguous descriptions.

[0032] The technical effects of the above embodiments include: realizing differentiated responses of hierarchical early warning, ensuring executable early warning actions, and realizing historical data archiving to provide support for subsequent optimization.

[0033] The existing rain shelter member temperature deformation calculation does not consider the comprehensive influence of the sunshine duration in Shanxi, the member corrosion amount and the surface humidity, and the formula parameters do not have clear dimensions and calibration basis, resulting in large calculation deviation of the deformation coefficient and the inability to accurately reflect the real deformation state of the member.

[0034] Based on this, the rain shelter metal member temperature deformation coefficient is calculated by the following formula: ; λ m represents the thermal expansion coefficient of the rain shelter member material, and has a dimension of 1 / ℃; Treal represents the real-time ambient temperature, with the dimension of ℃ (Celsius); T ref represents the component reference temperature, with the dimension of ℃; t sun represents the daily cumulative sunshine duration, with the dimension of h; d comp represents the measured thickness of the component, with the dimension of mm; δ rust represents the corrosion amount of the component, with the dimension of mm; k env represents the humidity correction coefficient, dimensionless; H surf represents the humidity of the component surface, with the dimension of %RH.

[0035] The technical solution precisely calculates the temperature deformation coefficient through a multi-factor coupling formula, and each parameter is designed in combination with the regional characteristics of Shanxi and the physical characteristics of the component: the formula molecule part, (T real -T ref ) reflects the difference between the real-time temperature and the reference temperature, which is the core temperature difference factor driving the deformation of the component, multiplied by t sun 0.6 times, the longer the sunshine duration, the more heat the component absorbs, and the more significant the deformation, 0.6 times reflects the nonlinear influence of sunshine; the denominator part, d comp and δ rust reflect the actual effective thickness of the component, the thicker the component and the greater the corrosion amount, the more difficult the deformation occurs, and the larger the denominator, the smaller the deformation coefficient, which conforms to the physical law; λ m as a material inherent property, directly determines the sensitivity of the component to temperature changes; finally multiplied by the humidity correction term of k , k env is calibrated to 0.02 in the dry environment of Shanxi to avoid exaggeration of the additional influence of high humidity on deformation, H surf 0.3 times reflects the weak nonlinear correlation of humidity to deformation.

[0036] Each parameter has a clear dimension, ensuring the physical reasonableness of the formula calculation logic and avoiding calculation errors caused by dimension confusion.

[0037] The technical effects of the above embodiments include: calculating the deformation coefficient in combination with the regional and component characteristics, the parameter dimensions are clear, which improves the accuracy of the deformation coefficient calculation and reflects the real deformation state of the component.

[0038] The existing operation platform contact friction coefficient calculation does not associate the temperature deformation coefficient, and does not consider the synergistic attenuation effect of load, vibration and humidity, the formula parameters have no fitting basis and clear dimensions, resulting in that the friction coefficient cannot accurately reflect the actual contact stability of the platform and the component.

[0039] Therefore, the contact friction coefficient of the operation platform and the rain shed truss is calculated by the following formula: ; denotes the base friction coefficient, dimensionless; denotes the deformation influence coefficient, dimensionless; a T denotes the canopy metal member temperature deformation coefficient, dimensionless; k F denotes the load correction coefficient, dimensionless; F load denotes the total load of the operation platform, dimension of kN; k H denotes the humidity attenuation coefficient, dimensionless; H surf denotes the member surface humidity, dimension of %RH; v vib denotes the vibration speed of the canopy member, dimension of mm / s.

[0040] The technical scheme calculates the contact friction coefficient through a multi-factor correction formula, and the core lies in the synergistic effect of temperature deformation and other influencing factors: the formula takes μ base as the benchmark, first multiplies the deformation correction term of a , the larger the a T , the more significant the member deformation, and the more the friction coefficient decays, =120 through experimental fitting to ensure that the decay amplitude conforms to the actual situation; then multiplies the load correction term of F , the larger the F load , the larger the denominator, and the smaller the friction coefficient, and the 0.8 power reflects the nonlinear decay of the load, k F =0.03 avoids over-amplification of the load effect; then multiplies the humidity correction term of v , which adopts an exponential form to reflect the rapid decay characteristics of humidity on friction, =0.015 ensures that the friction coefficient decays reasonably by 10%RH for each increase in humidity; and finally multiplies the vibration correction term of H , the larger the H , the more intense the member vibration, and the smaller the friction coefficient, and the 0.01 coefficient reflects the linear decay effect of vibration, which conforms to the physical law that vibration leads to unstable contact. The dimensions of each parameter and the fitting basis are clear, ensuring that the friction coefficient calculation conforms to the actual contact scenario.

[0041] The technical effects of the above embodiment include: correlating temperature deformation with multi-factor synergistic correction, improving the accuracy of friction coefficient calculation, and reflecting the actual contact stability of the platform and the member.

[0042] The existing fall risk early warning index calculation does not couple the physiological state of personnel with the Shanxi low temperature factor, the formula parameter weight has no basis and the dimension is chaotic, which leads to that the risk index cannot quantitatively reflect the actual fall risk, and the early warning threshold lacks scientific support.

[0043] Therefore, the fall risk early warning index is calculated by the following formula: ; A represents a friction-height weight coefficient, dimensionless, and is set to 0.85; μ contact represents a contact friction coefficient between the work platform and the canopy truss, dimensionless; h work represents an actual working height, dimension of m; B represents a physiological-wind speed weight coefficient, dimensionless, and is set to 0.72; C Vresp represents a working personnel breathing frequency variation coefficient, dimension of %; v wind represents a real-time wind speed at the working point, dimension of m / s; C represents a deformation-working time length weight coefficient, dimensionless; α T represents a temperature deformation coefficient of the canopy metal member, dimensionless; t work represents a single working time length, dimension of h; D represents a low-temperature compensation coefficient, dimensionless, and is set to 0.05); T real represents a real-time environmental temperature, dimension of ℃; T low represents a low-temperature threshold, dimension of ℃; R warn represents a falling risk early warning index, dimensionless.

[0044] The technical scheme quantifies the risk index through progressive coupling formulas, and the core lies in fusing four risk factors of environment, member, platform and personnel: the numerator part is composed of three risk factors weighted, the first term , indicates that the smaller the friction coefficient is, the higher the risk is, indicates that the higher the height is, the risk increases nonlinearly, A=0.85 gives it the highest weight; the second term , indicates that the larger the breathing frequency variation coefficient is, the more unstable the breathing is, the more likely the operation is to fail, v wind indicates that the larger the wind speed is, the higher the risk is, B=0.72 gives it a secondary weight; the third term , α T indicates that the larger the deformation is, the more significant the deformation is, t work indicates that the longer the working time length is, the more tired the personnel is, the higher the risk is, C=0.45 gives it an auxiliary weight. The denominator part is a low-temperature compensation term, T real <T low , the denominator decreases, R warn increases, which indicates the characteristics that low temperature aggravates the risk, D=0.05 avoids that the influence of low temperature is excessively enlarged.

[0045] Finally, multiplying by 100 quantifies the index to 0-100, which is convenient for setting the graded early warning threshold. The weight and dimension of each parameter are clear, which ensures that the risk index is quantified scientifically.

[0046] The technical effects of the above embodiment include: fusing four-dimensional risk factors and low-temperature compensation, quantifying the risk index, providing scientific support for setting the graded early warning threshold, and improving the risk assessment accuracy.

[0047] The existing early warning system does not have a clear closed-loop optimization mechanism, and the specific content and weight coefficient correction logic of the stored data are not clear, which leads to the inability to continuously improve the early warning ability according to the actual operation data, and the accuracy is prone to decrease after long-term use.

[0048] Therefore, the early warning data stored in the closed-loop optimization stage includes the real-time environmental temperature collected by the multi-dimensional environment perception module, the daily cumulative sunshine duration, the component surface humidity, and the real-time wind speed of the operation point, the component measured thickness and the component vibration speed collected by the canopy component state monitoring module, the total load of the operation platform, the actual operation height, and the single operation duration collected by the operation platform state monitoring module, the operation personnel respiratory rate variation coefficient collected by the personnel physiological monitoring module, and the canopy metal component temperature deformation coefficient, the contact friction coefficient, and the falling risk early warning index obtained by the dynamic calculation stage; the actual risk event includes component abnormal vibration record, operation personnel physiological state abnormal record, and operation platform operation state abnormal record during operation; the gradient descent algorithm iteratively corrects the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-operation duration weight coefficient, and the low-temperature compensation coefficient by minimizing the deviation value of the falling risk early warning index calculation value and the actual risk event; the corrected weight coefficient is automatically updated to the intelligent early warning calculation module for coefficient and index calculation in the next day.

[0049] The technical scheme constructs the data storage, deviation calculation, weight correction, and new application processing flow, and ensures the continuous iteration of the early warning ability: firstly, the stored early warning data of the day covers the collection and calculation of the whole process, including the original collection data of each module and the intermediate coefficients and final early warning index obtained by calculation, ensuring data integrity; the actual risk event focuses on three key abnormalities, providing an actual risk benchmark for deviation calculation; the gradient descent algorithm iteratively corrects the four types of weight coefficients, friction-height, physiological-wind speed, deformation-time, and low-temperature compensation, avoiding the subjectivity of manual adjustment; the corrected weight coefficient is automatically updated to the intelligent early warning calculation module and directly applied to the next day calculation, forming a closed loop of daily optimization and next day application, ensuring that the early warning model continuously adapts to the actual operation data and avoiding the decrease in accuracy after long-term use.

[0050] The technical effects of the above embodiments include: constructing a complete closed-loop optimization mechanism, dynamically correcting the weight coefficient, and ensuring that the early warning ability continuously improves with the operation data.

[0051] The data collection of the existing early warning system does not clearly define the cycle and data processing method, resulting in discontinuous and large fluctuation of the collected data, which cannot provide stable and reliable input for risk calculation, and the determination standard of physiological abnormality of personnel is not clear, affecting the accurate identification of physiological risk. At this point, all data collection operations of the dynamic calculation stage are executed in a 1-second cycle; the infrared temperature sensor collects the real-time environmental temperature 3 times per cycle and takes the arithmetic mean as the value of T real The photosensitive sensor accumulates and updates the daily cumulative sunshine duration according to the light intensity per cycle, the capacitive humidity sensor collects the surface humidity of the component once per cycle, the miniature wind speed sensor collects the real-time wind speed of the work point 5 times per cycle and takes the effective value as the value of v wind The optical fiber displacement sensor collects the measured thickness of the component 2 times per cycle, the piezoelectric vibration sensor collects the vibration speed of the component once per cycle; the pressure sensor collects the total load of the work platform once per cycle, and the laser ranging sensor collects the actual working height 2 times per cycle and takes the arithmetic mean as the value of h work The timing unit updates the single work duration according to the time increment per cycle; the non-contact millimeter wave radar collects the respiratory rate coefficient of variation of the worker once per cycle; the normal value range of the respiratory rate coefficient of variation of the worker is set to 5% to 15%, and greater than 20% is determined as a physiological state abnormality.

[0052] The technical scheme ensures stable and reliable input data through standardized collection cycle and data processing method, and clearly defines the physiological abnormality determination standard: setting 1 second as the collection cycle ensures data real-time and avoids risk lag caused by long cycle, and avoids resource waste caused by high-frequency collection; different processing methods are designed according to the characteristics of different sensors: the infrared temperature sensor collects 3 times and takes the average to reduce the influence of temperature fluctuation, the miniature wind speed sensor collects 5 times and takes the effective value to smooth the instantaneous fluctuation of wind speed, the laser ranging sensor collects 2 times and takes the average to improve the height measurement accuracy, the photosensitive sensor accumulates and updates the sunshine duration to ensure continuous statistics of sunshine influence, and other sensors collect 1 time per cycle to balance accuracy and efficiency; the timing unit updates the work duration according to the cycle increment to ensure continuous and uninterrupted duration statistics; the normal range and abnormal threshold of the respiratory rate coefficient of variation are clearly defined to avoid subjective and fuzzy physiological state determination and ensure accurate identification of personnel operation stability risk. The standardized collection and processing process provides high-quality data input for subsequent risk calculation.

[0053] The technical effects of the above embodiments include: standardized collection cycle and data processing method, clear physiological abnormality determination standard, stable and reliable input data, and improved physiological risk identification accuracy.

[0054] The existing pre-warning threshold setting does not combine the actual operation scene of the railway station in Shanxi region for pilot calibration, only uses the general threshold, which leads to the mismatch of the threshold and the scene of low temperature in winter and sunburn in summer in Shanxi, and the pre-warning accuracy and false alarm rate cannot meet the actual demand.

[0055] Therefore, in the railway station canopy reconstruction high-altitude operation falling risk intelligent pre-warning method, the falling risk pre-warning index threshold used in the grading pre-warning stage is obtained through pilot data calibration of canopy reconstruction high-altitude operation of three different railway stations in Shanxi region; the falling risk pre-warning index threshold corresponding to the first-level pre-warning is set to be greater than 80, the falling risk pre-warning index threshold corresponding to the second-level pre-warning is set to be greater than 60 and less than or equal to 80, and the falling risk pre-warning index threshold corresponding to the safe state is set to be less than or equal to 60; the pilot data includes environmental data such as real-time environmental temperature, member surface humidity and member vibration speed, member state data such as member actual thickness, personnel physiological data such as respiratory frequency variation coefficient of the operation personnel, and actual operation risk record; in the calibration process, the friction-height weight coefficient, the physiological-wind speed weight coefficient, the deformation-operation time weight coefficient and the low temperature compensation coefficient are adjusted, so that the pre-warning accuracy is greater than 92% and the false alarm rate is controlled to be less than 11.5%.

[0056] The technical scheme ensures that the pre-warning threshold fits the actual scene through pilot calibration in Shanxi region, and the core lies in threshold optimization based on localized data: three different railway stations in Shanxi are selected as pilots, covering the differences in canopy structure and operation environment of different stations, to ensure the universality of the threshold; the pilot data covers different seasons and time periods, and includes three types of core data, i.e. environment, member and physiology, and actual risk records, to ensure the representativeness of the data; in the calibration process, the four types of core weight coefficients, i.e. friction-height, physiological-wind speed, deformation-time and low temperature compensation, are adjusted, so that the pre-warning accuracy and false alarm rate reach the target value, avoiding the scene mismatch problem of the general threshold; finally, the three-level threshold is determined, which not only clearly divides the risk levels, but also ensures that the pre-warning actions corresponding to each level match the actual risk, for example, the first-level threshold > 80 corresponds to high risk, triggering the emergency stop action, which meets the emergency demand of high-risk scene in Shanxi.

[0057] The technical effects of the above embodiments include that the threshold is fitted to the actual scene in Shanxi through regional pilot calibration, ensuring that the pre-warning accuracy and false alarm rate meet the actual operation demand.

[0058] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the disclosed technical content into equivalent embodiments with equivalent changes, and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the technical solution content of the present application, still falls within the protection scope of the present application.

Claims

1. A method for intelligent early warning of fall risk during high-altitude operations in the renovation of railway station canopies, characterized in that, include: Enter the canopy component parameters, baseline values, and weighting coefficients, and start all sensors to perform data calibration to eliminate outliers; The multi-dimensional environmental perception module collects real-time ambient temperature, cumulative sunshine duration, component surface humidity and real-time wind speed at the work site; the canopy component status monitoring module collects the measured thickness and vibration velocity of the components; the work platform status monitoring module collects the total load of the work platform, the actual working height and the duration of a single work session; and the personnel physiological monitoring module collects the coefficient of variation of the breathing frequency of the workers. The collected real-time ambient temperature, the cumulative sunshine duration of the day, the surface humidity of the component, the measured thickness of the component, and the vibration velocity of the component are input into the intelligent early warning calculation module, which calculates the temperature deformation coefficient of the canopy metal component. The contact friction coefficient between the work platform and the canopy truss is calculated by combining the temperature deformation coefficient of the canopy metal components, the total load of the work platform, and the vibration velocity of the components. Then, the fall risk warning index is calculated by combining the contact friction coefficient, the breathing frequency variation coefficient of the workers, the real-time wind speed at the work point, the actual work height, and the duration of a single work session. The tiered early warning stage triggers a corresponding level of early warning action based on the fall risk early warning index; The closed-loop optimization phase stores the daily warning data and actual risk events, and corrects the weight coefficients using a gradient descent algorithm to improve the accuracy of subsequent warnings.

2. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The parameters of the canopy components entered in the data preprocessing stage include the thermal expansion coefficient of the canopy component material and the amount of corrosion of the component. The benchmark values ​​include the component benchmark temperature, low temperature threshold and basic friction coefficient. The weighting coefficients include friction-height weighting coefficient, physiological-wind speed weighting coefficient, deformation-operation time weighting coefficient and low temperature compensation coefficient. The sensor calibration time is set to 10 minutes, and out-of-range data caused by sensor malfunctions are removed during the calibration process; The thermal expansion coefficient of the canopy component material is determined based on the actual metal material used in the canopy. The amount of corrosion of the component is obtained through component inspection before operation and entered into the system. The reference temperature of the component is set to 25℃. The low temperature threshold is set to 5℃ based on the normal low temperature value in Shanxi winter. The basic friction coefficient is calibrated to 0.6 based on the steel-steel dry contact characteristic experiment.

3. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The multi-dimensional environmental perception module includes an infrared temperature sensor, a photosensor, a capacitive humidity sensor, and a miniature wind speed sensor. The infrared temperature sensor is used to collect the real-time ambient temperature, the photosensor is used to accumulate and count the cumulative sunshine duration of the day, the capacitive humidity sensor is used to collect the surface humidity of the component, and the miniature wind speed sensor is used to collect the real-time wind speed at the work site. The canopy component status monitoring module includes a fiber optic displacement sensor and a piezoelectric vibration sensor. The fiber optic displacement sensor is used to collect the measured thickness of the component, and the piezoelectric vibration sensor is used to collect the vibration velocity of the component. The work platform status monitoring module includes a pressure sensor, a laser rangefinder, and a timing unit. The pressure sensor is used to collect the total load of the work platform, the laser rangefinder is used to collect the actual working height, and the timing unit is used to count the duration of a single operation. The personnel physiological monitoring module uses a non-contact millimeter-wave radar, which is used to collect the respiratory rate variation coefficient of the workers; the intelligent early warning calculation module uses an STM32H743 embedded chip and is equipped with an edge computing algorithm, which is used to calculate the temperature deformation coefficient of the canopy metal component, the contact friction coefficient, and the fall risk warning index in the dynamic calculation stage.

4. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The warning actions triggered in the graded warning stage include Level 1 warning, Level 2 warning, and a safe state. During Level 1 warning, a high-frequency alarm signal is emitted via an audible and visual alarm, the work platform is stopped via an emergency stop controller, and work site location data and real-time risk data are sent to the control center via a 4G / 5G communication module. During Level 2 warning, a medium-frequency alarm signal is emitted via the same audible and visual alarm, and a speed reduction prompt is sent to the workers. During the safe state, a green indicator light remains constantly lit, and every 5 minutes, the collected real-time ambient temperature, component surface humidity, component vibration velocity, and the calculated fall risk warning index are uploaded to the storage unit for archiving via the 4G / 5G communication module. The high-frequency alarm signal frequency is set to 2Hz, and the medium-frequency alarm signal frequency is set to 1Hz.

5. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The temperature deformation coefficient of the metal components of the canopy is calculated using the following formula: ； Where, λ m T represents the coefficient of thermal expansion of the material of the canopy components; real T represents the real-time ambient temperature. ref The reference temperature of the component is represented by t. sun d represents the total daily sunshine duration. comp δ represents the measured thickness of the component. rust k represents the amount of corrosion on a component. env H represents the humidity correction factor. surf This indicates the surface humidity of the component; the humidity correction factor is calibrated to 0.02 based on the dry environment characteristics of Shanxi region.

6. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 5, characterized in that, The coefficient of contact friction between the working platform and the canopy truss is calculated using the following formula: ； Where, μ base k represents the basic coefficient of friction. α α represents the deformation influence coefficient. T k represents the temperature deformation coefficient of the metal components of the canopy. F F represents the load correction factor. load k represents the total load of the work platform. H H represents the humidity attenuation coefficient. surf Indicates the surface humidity of a component, v vib The deformation influence coefficient represents the vibration velocity of the canopy components; α is used to represent the deformation influence coefficient. T The friction coefficient decay curve was fitted to 120, the load correction coefficient was set to 0.03, and the humidity decay coefficient was set to 0.

015.

7. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 6, characterized in that, The fall risk warning index is calculated using the following formula: ； Where A represents the friction-height weighting coefficient, μ contact h represents the coefficient of contact friction between the work platform and the canopy truss. work B represents the actual operating height, C represents the physiological-wind speed weighting coefficient, and B represents the actual operating height. Vresp V represents the coefficient of variation of the worker's respiratory rate. wind The real-time wind speed at the work site is represented by C, which represents the deformation-work duration weighting coefficient, and α is the real-time wind speed at the work site. T The coefficient of thermal deformation of the metal components of the canopy is represented by t. work Indicates the duration of a single operation, D represents the low-temperature compensation coefficient, and T represents the duration of a single operation. real T represents the real-time ambient temperature. low R represents the low temperature threshold. warn The fall risk warning index is represented by the following: the friction-height weighting coefficient is set to 0.85, the physiological-wind speed weighting coefficient is set to 0.72, the deformation-working time weighting coefficient is set to 0.45, and the low temperature compensation coefficient is set to 0.

05.

8. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The daily early warning data stored in the closed-loop optimization stage includes the real-time ambient temperature, the cumulative sunshine duration, the surface humidity of the component, and the real-time wind speed at the work site collected by the multi-dimensional environmental perception module; the measured thickness and vibration velocity of the component collected by the canopy component status monitoring module; the total load of the work platform, the actual working height, and the duration of a single work session collected by the work platform status monitoring module; the respiratory rate variation coefficient of the workers collected by the personnel physiological monitoring module; and the temperature deformation coefficient, the contact friction coefficient, and the fall risk warning index of the canopy metal component obtained in the dynamic calculation stage. The actual risk events include abnormal vibration records of components during operation, abnormal physiological state records of workers, and abnormal operating status records of the work platform; the gradient descent algorithm iteratively corrects the friction-height weighting coefficient, the physiological-wind speed weighting coefficient, the deformation-operation duration weighting coefficient, and the low temperature compensation coefficient by minimizing the deviation between the calculated value of the fall risk warning index for the day and the value corresponding to the actual risk events; the corrected weighting coefficients are automatically updated to the intelligent warning calculation module for coefficient and index calculation in the dynamic calculation stage the next day.

9. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, All data acquisition operations in the dynamic calculation phase are executed in a loop with a period of 1 second. The infrared temperature sensor collects the real-time ambient temperature three times per cycle and takes the arithmetic mean as the Treal value. The photosensitive sensor updates the cumulative sunshine duration of the day based on the light intensity each cycle. The capacitive humidity sensor collects the surface humidity of the component once per cycle. The miniature wind speed sensor collects the real-time wind speed at the work point five times per cycle and takes the effective value as the vwind value. The fiber optic displacement sensor collects the measured thickness of the component twice per cycle. The piezoelectric vibration sensor collects the vibration velocity of the component once per cycle. The pressure sensor collects the total load of the work platform once per cycle. The laser rangefinder collects the actual working height twice per cycle and takes the arithmetic mean as the hwork value. The timing unit updates the single work duration based on the time increment each cycle. The non-contact millimeter-wave radar collects the respiratory rate variation coefficient of the worker once per cycle. The normal range for the respiratory rate variation coefficient of the worker is set to 5% to 15%, and a value greater than 20% is considered an abnormal physiological state.

10. The intelligent early warning method for fall risk during high-altitude operations in railway station canopy renovation according to claim 1, characterized in that, The fall risk warning index threshold used in the graded early warning stage was calibrated using pilot data from high-altitude operations involving the renovation of canopies at three different railway stations in Shanxi Province. The fall risk warning index threshold corresponding to the first-level warning was set to be greater than 80, the fall risk warning index threshold corresponding to the second-level warning was set to be greater than 60 and less than or equal to 80, and the fall risk warning index threshold corresponding to the safe state was set to be less than or equal to 60. The pilot data included environmental data such as real-time ambient temperature, component surface humidity, and component vibration velocity in different seasons and time periods, component status data such as measured component thickness, physiological data such as the respiratory rate variation coefficient of the workers, and actual operation risk records. During the calibration process, the friction-height weighting coefficient, the physiological-wind speed weighting coefficient, the deformation-operation duration weighting coefficient, and the low temperature compensation coefficient were adjusted to achieve an early warning accuracy rate of over 92% and a false alarm rate of less than 11.5%.

Images