A real-time fusion method and system for multi-modal and multi-parameter monitoring data of process production safety

Abstract

The present invention discloses a real-time fusion method and system for multi-modal and multi-parameter monitoring data of process production safety, which belongs to the field of production safety monitoring. This method can make up for the defect that the safety risk warning in the current process production scene is often a single parameter monitoring and analysis. It can realize real-time data fusion analysis based on the combination relationship of multi-risk characteristic parameters for different safety risk modes of process production activity objects, as well as hierarchical and graded warning functions of safety risk status for parameter layer, modality layer and object layer; through the collection, transmission and analysis of safety risk monitoring data, the real-time data utilization rate of process production is greatly improved, and the safety risk supervision efficiency of process production site is effectively improved; digital and intelligent technology is used to empower process production safety risk warning equipment, reduce the application threshold of safety production prediction and warning technology services, and broaden new ways of safety management.

Description

Technical Field

[0001] The present invention belongs to the field of production safety monitoring, and more specifically, relates to a real-time fusion method and system for multi-modal and multi-parameter monitoring data of process production safety. Background Art

[0002] Danger warning based on multi-parameter monitoring data is a key technical means for safe production in all walks of life. Process production safety includes multiple safety risk modes, and each safety risk mode includes multiple types of parameters. Each parameter has a stable change pattern in the production process cycle, and the production cycle is relatively fixed. Once a parameter is abnormal, it means that the probability of risk occurrence increases, and the occurrence of risk often has signs of multi-parameter abnormalities. The characteristics of this multi-modal and multi-parameter safety in the process production scenario lead to the need to determine the parameter combination mode and abnormal change pattern of each unsafe risk mode in the risk state identification process, and monitor and evaluate the early warning risk in real time.

[0003] The main problems of existing risk status identification technology include: (1) the contradiction between the relatively complete monitoring of production safety status and the insufficient comprehensive evaluation of production safety status analysis in real time; (2) the contradiction between the intelligent monitoring and the reliance on manpower for production safety management and control decisions; (3) the contradiction between the multi-parameter risk of production safety monitoring and multi-modal real-time monitoring and early warning. Therefore, there is an urgent need for a real-time monitoring data fusion method and system that can adapt to production activities with multi-variable parameters and can select multiple risk parameter combinations. Summary of the invention

[0004] In response to the above defects or improvement needs of the prior art, the present invention provides a method and system for real-time fusion of multi-modal and multi-parameter monitoring data of process production safety, aiming to solve the problems of real-time monitoring, fusion analysis and risk classification and early warning of multi-modal and multi-parameter safety, improve the intelligence and versatility of safety monitoring instruments, reduce the cost of production safety supervision, deepen the scenario application of safety production prediction and early warning technology services, and promote the digitalization of process safety production.

[0005] To achieve the above object, according to a first aspect of the present invention, a method for real-time fusion of multi-modal and multi-parameter monitoring data of process production safety is provided, comprising:

[0006] S1, obtain each risk characteristic parameter F i The monitoring fluctuation values ​​Δd at the latest l moments in the current monitoring cycle (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficients at the latest l moments Where i = 1, 2, 3, ..., m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple historical complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient;

[0007] S2, for ω (i，j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient according to The numerical range in which F is determined i In T p The risk status level is determined and a graded warning is issued; wherein, in the latest l moments, T p The weight coefficient is the largest, and p The closer the moment is, the greater the weight coefficient is;

[0008] S3, for each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient λ (k，p) , according to λ (k，p) The numerical range in which M is located determines k In T p The risk status level is determined and graded warning is issued;

[0009] S4, weighted calculation of the risk status evaluation coefficients of all safety risk modes of the monitored process production activity object A is performed to obtain the risk status evaluation coefficients of A at T p The multimodal fusion risk state coefficient θ p , according to θ p The numerical range in which A is located determines the p The risk status level is determined and graded warning is issued;

[0010] The numerical intervals correspond to the risk status levels one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level.

[0011] According to a second aspect of the present invention, a real-time fusion system for multi-modal and multi-parameter monitoring data of process production safety is provided, comprising:

[0012] Data fusion analysis module, used to obtain each risk characteristic parameter F i The monitoring fluctuation values ​​Δd at the latest l moments in the current monitoring cycle (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficients at the latest l moments Where i = 1, 2, 3, ..., m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple historical complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient; for ω (i,j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient For each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient of (k,p) ; The risk state evaluation coefficients of all safety risk modes of the monitored process production activity object A are weighted and calculated to obtain A at T p The multimodal fusion risk state coefficient θ p ;

[0013] Risk status classification module is used to The numerical range in which F is determined i In T p The risk status level of (k,p) The numerical range in which M is located determines k In T p The risk status level of p The numerical range in which A is located determines the p wherein the numerical intervals correspond to the risk status levels one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level;

[0014] Multi-level hazard warning module is used to i In T p The risk status level is graded and early warning is carried out according to M k In T p The risk status level is determined and a graded warning is given. p The risk status level can be determined and graded warning can be issued.

[0015] According to a third aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method described in the first aspect.

[0016] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0017] (1) To make up for the defect that the safety risk warning in the current process production scenario is often a single parameter monitoring and analysis, the method and system provided by the present invention can realize real-time data fusion analysis based on the combination relationship of multiple risk feature parameters for different safety risk modes of process production activity objects, as well as hierarchical and graded safety risk status warning functions for the parameter layer, mode layer, and object layer.

[0018] (2) Through the collection, transmission and analysis of safety risk monitoring data, the real-time data utilization rate of process production can be greatly improved, and the safety risk supervision efficiency of process production sites can be effectively improved.

[0019] (3) Empower process production safety risk warning equipment with digital and intelligent technologies, lower the application threshold of production safety prediction and warning technology services, and broaden new ways of safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of a process production safety multi-modal multi-parameter real-time monitoring data fusion method provided by an embodiment of the present invention;

[0021] Figure 2 An example of a method for setting a security risk status classification threshold interval provided in an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of the updating process of the monitoring value of the characteristic parameter at the current moment provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of a calculation process for parameter risk level classification provided by an embodiment of the present invention;

[0024] Figure 5A functional structure diagram of a process production safety multi-modal and multi-parameter real-time monitoring data fusion analysis and risk status classification early warning system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0026] An embodiment of the present invention provides a method for real-time fusion of multi-modal and multi-parameter monitoring data of process production safety. Before introducing the method, the definitions of relevant terms are first explained.

[0027] Risk characteristic parameters: refers to the variables that can reflect the changing characteristics of the safety risk status of the process production activity object by monitoring the changing patterns of data during the process production process.

[0028] Safety risk mode: refers to a certain type of process safety accident. A safety risk mode corresponds to one or more risk characteristic parameters, that is, the risk status of a safety risk mode is jointly determined by the risk status of one or more corresponding risk characteristic parameters. Each risk characteristic parameter has its own weight, which can be set according to the actual application scenario.

[0029] Process production activity object: refers to the monitoring object, such as a production equipment or a production process. Usually, a process production activity object corresponds to multiple safety risk modes, that is, the risk status of the process production activity object is jointly determined by the risk status of one or more corresponding safety risk modes. Each safety risk mode has its own weight, which can be set according to the actual application scenario.

[0030] Taking the production process of ladle turret rotation and pouring operation as an example, the process production activity object is the ladle turret, the safety risk mode is collapse, explosion, and fire, and the risk characteristic parameters include ladle weight, ladle support arm surface strain, turret inclination, rotation speed, combustible gas concentration, ambient temperature, ambient humidity, etc. Among them, the risk characteristic parameters corresponding to collapse are ladle weight, ladle support arm surface strain, turret inclination, and rotation speed; the risk characteristic parameters corresponding to explosion are combustible gas concentration and ambient temperature, and the risk characteristic parameters corresponding to fire are combustible gas concentration, ambient temperature, and ambient humidity.

[0031] The embodiment of the present invention provides a method for real-time fusion of multi-modal and multi-parameter monitoring data of process production safety, such as Figure 1 As shown, including:

[0032] S1, obtain each risk characteristic parameter F i The monitoring fluctuation value Δd of the latest l moments in the current monitoring cycle under the current state (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficients at the latest l moments Where i = 1, 2, 3, ... m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient.

[0033] Specifically, S1 includes S11-S13.

[0034] S11, for the real-time monitoring data of several cycles of the safety status of the process production, based on the safety data sequence set of all risk characteristic parameters obtained in several complete monitoring cycles, a time-space standard data matrix of multi-parameter combination is constructed. The row represents the parameter dimension of the object space feature, the column represents the time dimension, and the element is the monitoring time T of the characteristic parameter in several complete monitoring cycles. j` The corresponding normal data statistical mean represents the numerical safety characteristics of the parameter at the corresponding moment.

[0035] That is, for real-time monitoring of process production safety risks, real-time monitoring data of several safety status cycles are selected, and a multi-parameter combination of spatiotemporal standard data matrix S is constructed based on the safety data sequence set of all risk characteristic parameters obtained in several complete monitoring cycles. T , its general form is:

[0036]

[0037] S T The rows represent the spatial feature parameter dimension, the columns represent the time dimension, and the elements is the characteristic parameter F i The monitoring time T within several complete monitoring cycles j` The corresponding normal data statistical mean represents the numerical safety characteristics of the parameter at the corresponding moment. Each complete monitoring cycle covers m risk characteristic parameters, namely F 1 ,F2 ,…,F i ,…,F m (i=1,2,3,…,m); each characteristic parameter contains n monitoring time points, namely T 1 ,T 2 ,…,T j` ,…,T n (j`=1,2,3,…,n).

[0038] S12, for real-time monitoring of process production safety, a monitoring variable time series value matrix composed of the original monitoring values ​​of the risk characteristic parameters in the current monitoring cycle is constructed, and the columns of the matrix are the spatial characteristic parameter dimension and the behavioral time dimension. A monitoring variable time series value matrix composed of the monitoring values ​​of all parameters is formed in the period from the initial moment to the current moment in the current monitoring cycle, and the matrix elements represent the monitoring values ​​of the characteristic parameters at a certain moment in the current monitoring cycle.

[0039] That is, the real-time monitoring of process production safety risks is built at the initial time T in the current monitoring cycle. 1 To the current time T p In the time period (p≤n), the monitoring matrix D composed of all parameter monitoring values new , where element d (i,j) Indicates the characteristic parameter F i At time T in the current monitoring cycle j The monitoring value of . The row vector d i =[d (i，1) …d (i，2) …d (i，j) …d (i , p) ], i = 1, 2, 3, ..., m; column vector d' j =[d (1,j) d (2,j) …d (i,j) …d (m,j) ] T , j = 1, 2, 3, ..., p. D new The general form is:

[0040]

[0041] S13, for real-time monitoring of process production safety, constructs a monitoring variable fluctuation characteristic value matrix composed of all parameter monitoring values ​​in the period from the initial moment to the current moment in the current monitoring cycle. The matrix elements represent the time series fluctuation values ​​of the monitoring parameters of the characteristic parameters in the current monitoring cycle.

[0042] That is, the real-time monitoring of process production safety risks is constructed from the initial time T 1To the current time T p The monitoring variable fluctuation characteristic matrix Δ composed of all parameter monitoring values ​​in the period of time, where the elements j=1,2,3,…,p, represents the characteristic parameter F i At time T in the new monitoring cycle j The general form of Δ is:

[0043]

[0044] S2, for ω (i，j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient according to The numerical range in which F is determined i In T p The risk status level is determined and graded warning is carried out, thereby realizing graded warning at the parameter level; wherein, in the latest l moments, T p The weight coefficient is the largest, and p The closer the time is, the greater the weight coefficient is.

[0045] Specifically, S2 includes S21-S22.

[0046] S21, for real-time monitoring of process production safety, calculates T in the current monitoring cycle 1 To the current time T p The risk time-varying matrix of all characteristic parameters in the period of time includes:

[0047] S211, setting the interval threshold for characteristic parameter fluctuation state classification.

[0048] The characteristic parameter fluctuation state can be divided into multiple levels according to the type of specific production process monitored, and the embodiment of the present invention does not make a unique limitation to this.

[0049] Taking the example of classifying the characteristic parameter fluctuation state into three levels, namely: normal state, low-amplitude abnormal fluctuation and high-amplitude abnormal fluctuation, the characteristic parameter fluctuation state classification intervals include: normal state interval, low-amplitude abnormal fluctuation interval and high-amplitude abnormal fluctuation interval.

[0050] Preferably, the interval threshold of the characteristic parameter fluctuation state classification is set based on the data statistical distribution characteristics of all parameters in the historical complete monitoring cycle under the process production safety state in S11, that is:

[0051] Normal state interval (d` (i,j`) -h (ij`,0) ,d` (i,j`) +h ( 'ij`,0) );

[0052] Low abnormal fluctuation range (d` (i,j`) -h (ij`,1) ,d` (i,j`) -h (ij`,0) ]∪[d` (i,j`) +h ( ' ij`,0) ,d` (i,j`) +h ( ' ij`,1) );

[0053] High abnormal fluctuation range

[0054] Among them, Coming soon (i,j`) Take it as the statistical mean, based on the “3σ rule” of statistics, h (ij`,0) 、h ( ' ij`,0) Take M standard deviations, h (ij`,1) 、h ( ' ij`,1 ) is taken as N standard deviations, M<N, and M and N are both positive integers; for example: h (ij`,0) , h′ (ij`,0) Take 2 standard deviations, h (ij`,1) , h′ (ij`,1) Taken as 3 standard deviations;

[0055] Preferably, the interval threshold of the characteristic parameter fluctuation state classification can also be set based on the accident cause parameter threshold of each risk characteristic parameter, that is: let d` (i,j`) is the accident cause parameter threshold (can be a value or a numerical range), h (ij`,0) , h′ (ij`,0) For d` (i,j`) B%,h (ij`,1) , h′ (ij`,1) Take C% of the threshold, B<C, and B and C are both positive integers; for example, h (ij`,0) , h′ (ij`,0) Taken as 10% of the threshold, h (ij`,1) , h′ (ij`,1) It is taken as 20% of the threshold, and the upper or lower limit should be set according to the actual situation.

[0056] S212, calculating the state fluctuation coefficient of the characteristic parameter at the monitoring time.

[0057] That is, calculate the characteristic parameter F i At monitoring time T j The state fluctuation coefficient ω (i，j) .like Figure 2As shown, the new monitoring fluctuation value Δd of the characteristic parameter (i,j) The smaller it is, the smaller its fluctuation coefficient is; conversely, the more the new monitoring value deviates from the normal state interval, the stronger the abnormal fluctuation of the parameter is and the larger its fluctuation coefficient is.

[0058] Taking the volatility coefficients as 0, 1, and 2, ω (i,j) The value formula is:

[0059]

[0060] It is understandable that the fluctuation coefficient can also be 1, 2, 3 or other values, as long as it meets the above principle of "the more the new monitoring value deviates from the normal state range, the stronger the abnormal fluctuation of the parameter, and the greater its fluctuation coefficient."

[0061] S213, constructing a full characteristic parameter risk time-varying matrix composed of state fluctuation coefficients in the period from the initial moment to the current moment of the current monitoring cycle, wherein the columns of the matrix are parameter dimensions and the behavior time dimension. As time increases, the number of columns of the matrix increases.

[0062] That is, construct the current monitoring cycle from the initial time T 1 To the current time T p In the period of time, the state fluctuation coefficient ω (i，j) The total characteristic parameter risk time-varying matrix R (i,j) ,i=1,2,3,…,m,j=1,2,3,…,p,R (i,j) The general form is:

[0063]

[0064] S22, parameter layer: discriminant feature parameter F i At the current time T p The specific process is as follows:

[0065] S221, according to the value conditions of the parameter state fluctuation coefficient, set the state assessment time shift window length of the parameter risk identification to l, and construct the current time T p The full parameter risk discriminant matrix I (m,l) , that is, the risk level is calculated based on the parameter state fluctuations at the latest l moments, and the value of l is generally not less than 3 and satisfies l≤p. (m,l) The general form is Satisfies p-j+1=l.

[0066] Take m=3, l=5 as an example, Figure 3 It shows that as the production activity time cycle progresses, the three parameters at the current time T pThe update process of the state assessment time-shift window length l represents the interception of the l monitoring moments closest to the current moment. The risk discrimination matrix I composed of the three parameter state fluctuation coefficients in the state assessment time-shift window (3,5) The form is

[0067] S222, extract the parameter risk transfer row vector composed of the state fluctuation coefficients at the current moment and the nearest several monitoring moments. That is, extract the characteristic parameter F i At the current time T p (p≥3) and the first l-1 monitoring moments immediately adjacent to it, the parameter risk transfer row vector r composed of its state volatility coefficient (i,j) , the dimension is the state evaluation time-shift window length l, and j = 1, 2, 3, ..., l, r (i，j) The general form is:

[0068] r (i,j) =[ω (i,p-l+1) ω (i,p-l+2) … ω (i,p-j+1) … ω (i,p-1) ω (i,p) ].

[0069] S223, determine the weight values ​​of different monitoring moments to the risk state of the parameter at the current moment, and construct a weight column vector. The principle of setting the weight value is that the closer the moment monitoring value is to the current moment, the greater the weight value is.

[0070] That is, determine the weight values ​​of different monitoring moments for the risk state of the current parameter, and construct the weight column vector v (i,j``) , the dimension is the length of the state evaluation time-shift window l. The principle is that the closer to the current time T p The greater the weight value is, the greater the time point is.

[0071] v (i,j``) The general form is [v (i，p-l+1) v (i，p-l+2) … v (i，p-j``+1) … v( i，p-1) v (i，p) ] T , where v (i，p-j``1+) ∈[0,1] and j``=1, 2, 3,…, l.

[0072] It is understandable that the values ​​of the weight coefficients at different monitoring moments can be set arbitrarily, and the present invention does not impose a unique limit on this. As long as the above “the closer to the current moment T p The greater the weight value, the greater the (i,p-j``+1) ∈[0,1] and The principle is enough.

[0073] For example, one way to set the weights of different monitoring moments is Among them, the monitoring time importance index a p-j``+1 Satisfy a p-j``+1 ∈[0,1] and a p-l+1 <a p-l+2 <…<a p-j+1 <…<a p-1 <a p (j``=1,2,3,…,l). That is, the weight column vector v at different monitoring times (i,j) A specific form of is:

[0074]

[0075] S224, calculating the parameter risk state evaluation coefficient of the characteristic parameter at the current moment, the parameter being obtained by multiplying the parameter risk transfer row vector by the weight column vector.

[0076] That is, the characteristic parameter F i At the current time T p Parameter risk state evaluation coefficient

[0077] S225, construct risk status classification rules for characteristic parameters, based on The calculated value of the characteristic parameter F i At the current time T p risk status.

[0078] Among them, the characteristic parameter risk status can be divided into multiple levels according to the type of specific production process monitored, and the embodiment of the present invention does not make a unique limitation on this. Similarly, the specific value of the characteristic parameter risk status level interval can be set according to the type of specific production process monitored, and the embodiment of the present invention does not make a unique limitation on this.

[0079] Taking the risk status classification into four levels as an example, namely: safe state, low risk state, medium risk state and high risk state, a classification rule of characteristic parameter risk state level (FRSL) is as follows:

[0080]

[0081] Among them, NS (Normal Safety) indicates a safe state, LR (Low-Risk) indicates a low-risk state, MR (Medium-Risk) indicates a medium-risk state, and HR (High-Risk) indicates a high-risk state.

[0082] More preferably, Figure 4 Shows the current time Tp Parameter risk state evaluation coefficient The calculation process and parameter risk status classification results. In this example, the user sets the importance index a for the latest l = 3 monitoring moments. p-2 =0.25, a p-1 =0.5, a p =1, we can get the weight column vector v (i，p) =[1 / 7 2 / 7 4 / 7] T There are 27 combinations of state volatility coefficients at three moments, namely, the parameter risk transfer row vector r (i,p) There are 27 forms.

[0083] S3, for each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient λ (k,p) , according to λ (k,p) The numerical range in which M is located determines k In T p The risk status level can be determined and graded warning can be carried out, thus realizing graded warning at the modal level.

[0084] Step S3 is the modality layer: identifying the security risk modality M k At the current time T p The specific process includes:

[0085] S31, construct row vector u k , which is used to select all characteristic parameters contained in the security risk mode.

[0086] Preferably, u k The general form is u k =[u 1 u 2 … u i … u m ], where u i =0 or 1.

[0087] S32, constructing a parameter weight row vector for setting the weights of each selected feature parameter.

[0088] That is, construct the row vector v k , which is used to set the weights of each characteristic parameter contained in the security risk mode.

[0089] Preferably, v k The general form is v k =[v 1 v 2 … v i… v m ], where v i ∈(0,1] and satisfies

[0090] S33, a parameter combination vector is calculated from the parameter selection row vector and the parameter weight row vector, and is used to quantify the parameter combination relationship of the safety risk mode.

[0091] That is, calculate the parameter combination vector C k , used to quantify the security risk mode M k The parameter combination relationship.

[0092] Preferably, C k The general calculation form is C k =u k ⊙v k =[u 1 v 1 u 2 v 2 … u i v i … u m v m ].

[0093] S34, extracting the p , column vector of all parameter risk state evaluation coefficients

[0094] Preferably, The general form is

[0095] S35, calculating the modal risk state evaluation coefficient of the security risk mode at the current moment, which is used to evaluate its risk state. The coefficient is obtained by multiplying the parameter combination vector and the column vector of the full parameter risk state evaluation coefficient at the current moment.

[0096] That is, calculate the security risk mode M k At the current time T p The modal risk state evaluation coefficient λ (k,p) , which is used to evaluate its risk status.

[0097] λ (k,p) The general calculation form is And satisfy λ (k,p) ∈[0,2].

[0098] S36, construct the security risk modal risk state level discrimination rule, according to λ (k,p) The calculated value of determines the risk status of the security risk mode at the current moment.

[0099] The security risk modal risk state can be divided into multiple levels according to the type of specific production process being monitored, and the embodiment of the present invention does not make any unique restrictions on this. Similarly, the specific value of the security risk modal risk state level interval can be set according to the type of specific production process being monitored, and the embodiment of the present invention does not make any unique restrictions on this. Taking the classification of risk status into four levels as an example, namely: safe state, low-risk state, medium-risk state and high-risk state, a classification rule for the security risk modal risk state level (MRSL) is as follows:

[0100]

[0101] S4, weighted calculation of the risk status evaluation coefficients of all safety risk modes of the monitored process production activity object A is performed to obtain the risk status evaluation coefficients of A at T p The multimodal fusion risk state coefficient θ p , according to θ p The numerical range in which A is located determines the p The risk status level of the object can be determined and graded warning can be carried out, thereby realizing graded warning at the object level;

[0102] The numerical intervals correspond to the risk status levels one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level.

[0103] Step S4 is the object layer, which determines the monitored process production activity object A at the current time T p , the overall risk status level of security multi-modal fusion, the specific process includes:

[0104] S41, for secure multimodal M 1 、M 2 …M k …M q (q is the total number of security multi-modalities, k = 1, 2, 3, ..., q), the user sets the security risk modalities M for multi-modal fusion k The weight row vector w is used to determine the impact of each safety risk mode.

[0105] Preferably, w is of the general form w=[w 1 w 2 … w k … w q ], where w k ∈(0,1] and satisfies

[0106] S42, based on step S35, constructs a p , the risk state column vector λ of secure multimodal fusion p .

[0107] Preferably, λ p The general form of p =[λ (1，p) λ (2，p) … λ (k，p) … λ (q，p) ] T , satisfying λ (k，p) ∈[0,2].

[0108] S43, calculate the activity risk state coefficient of the monitored process production activity object at the current moment by the safety multimodal fusion, which is used to determine the overall risk state level at the current moment. The coefficient is obtained by multiplying the modal weight column vector and the safety multimodal fusion risk state column vector at the current moment.

[0109] That is, the production activity object A of the process monitored by the calculation is at the current time T p The safe multimodal fusion risk state coefficient θ p , used to determine A at the current time T p The overall risk status level.

[0110] θ p The general calculation form is Satisfy θ p ∈[0,2].

[0111] S44, construct the overall risk status level discrimination rule of the monitored process production activity object A, according to θ p The calculated value of object A at the current time T p risk status.

[0112] The risk status of a process production activity object can be divided into multiple levels according to the type of specific production process being monitored, and the embodiment of the present invention does not make any unique restrictions on this. Similarly, the specific value of the risk status level interval of a process production activity object can be set according to the type of specific production process being monitored, and the embodiment of the present invention does not make any unique restrictions on this. Taking the classification of risk status into four levels as an example, namely: safe state, low-risk state, medium-risk state and high-risk state, a classification rule for the overall risk status level (ARSL) of process production activity objects is as follows:

[0113]

[0114] After completing the above-mentioned division of characteristic parameter risk status level (FRSL), safety risk modal risk status level (MRSL), and process production activity object overall risk status level (ARSL), by constructing risk grading warning models at different levels, it is used to realize the graded warning function of the parameter layer (S2), modal layer (S3) and object layer (S4) of the monitored production process.

[0115] The risk classification warning mode can be divided into multiple levels according to the type of specific production process being monitored, and the embodiment of the present invention does not limit this to uniqueness. Taking the risk classification warning mode as an example, the levels are divided into: safety, level I warning, level II warning, and level III warning. At the current time T p , characteristic parameter F i A specific model for risk warning level classification is:

[0116]

[0117] Current time T p , security risk mode M k A specific model for risk warning level classification is:

[0118]

[0119] Current time T p 、Process production activity object A O A specific model for the overall risk warning level classification is:

[0120]

[0121] The method provided by the present invention can be used for the fusion of multi-modal and multi-parameter real-time monitoring data of process production safety in a multi-level digital and intelligent safety monitoring system for process operations as shown in the patent "A multi-level digital and intelligent safety monitoring system for process operations" (application number 2022111303888), effectively improving the efficiency of real-time supervision of process production safety risks.

[0122] The embodiment of the present invention provides a real-time fusion system for multi-modal and multi-parameter monitoring data of process production safety, including:

[0123] Data fusion analysis module, used to obtain each risk characteristic parameter F i The monitoring fluctuation values ​​Δd at the latest l moments in the current monitoring cycle (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficients at the latest l moments Where i = 1, 2, 3, ..., m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient; for ω (i,j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient For each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient of (k,p) ; The risk state evaluation coefficients of all safety risk modes of the monitored process production activity object A are weighted and calculated to obtain A at T p The multimodal fusion risk state coefficient θ p ;

[0124] Risk status classification module is used to The numerical range in which F is determined i In T p The risk status level of (k,p) The numerical range in which M is located determines k In T p The risk status level of p The numerical range in which A is located determines the p wherein the numerical interval corresponds to the risk status level one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level.

[0125] Multi-level hazard warning module is used to i In T p The risk status level is graded and early warning is carried out according to M k In T p The risk status level is determined and a graded warning is given. p The risk status level can be determined and graded warning can be issued.

[0126] Furthermore, the system further comprises:

[0127] The full parameter data collection module is used to collect the change data of all safety risk modal risk characteristic parameters covering the process production activity objects selected by the user;

[0128] A data signal sending module, used to send the collected parameter data to the data signal receiving module using a wireless network communication protocol;

[0129] A data signal receiving module, used to receive monitoring data of various risk characteristic parameters from the signal sending module;

[0130] The periodic data storage module is used to store the monitoring data of all risk characteristic parameters in the complete process operation cycle with parameter type and monitoring time as identification;

[0131] The user basic setting module is used for users to select the production activity objects and safety risk mode types to be monitored, and to set the risk characteristic parameters and the weights of each parameter according to the production process requirements.

[0132] Figure 5 The functional structure diagram of a multi-modal and multi-parameter real-time monitoring data fusion analysis and risk classification early warning system for process production safety is provided, including: a full-parameter data acquisition module, a data signal sending module, a data signal receiving module, a periodic data storage module, a user basic setting module, a data fusion analysis module, a risk status classification module, and a multi-level danger early warning module;

[0133] The full parameter data acquisition module is used to collect change data of all safety risk modal risk characteristic parameters covering the process production activity object selected by the user, including various sensing elements that can sense changes in the physical and chemical states of parameters, such as sensors, etc.;

[0134] The data signal sending module is used to send the collected parameter data to the real-time data signal receiving module using a wireless network communication protocol, including serial port, Wifi, 4G, 5G, etc.;

[0135] The data signal receiving module is used to receive the monitoring data of each risk characteristic parameter from the signal sending module;

[0136] The periodic data storage module is used to store the monitoring data of all risk characteristic parameters in the complete process operation cycle with parameter type and monitoring time as identification;

[0137] The user basic setting module is used for users to select production activity objects and safety risk modal types to be monitored, and set risk characteristic parameters and the weights of each parameter according to production process requirements, including: activity object selection submodule, safety risk modal selection submodule, modal weight setting submodule, characteristic parameter selection submodule, parameter weight setting submodule;

[0138] The data fusion analysis module includes a standard period data matrix construction unit, a full parameter monitoring matrix construction unit, a full parameter fluctuation characteristic matrix construction unit, a full parameter risk coefficient time-varying matrix calculation unit, a parameter risk state evaluation coefficient calculation unit, a modal risk state evaluation coefficient calculation unit, and an activity risk state evaluation coefficient calculation unit, which are used to realize the data fusion analysis calculation function in the above steps S1 to S4, and output the risk evaluation coefficient of the production activity, safety risk mode, and characteristic parameter selected by the user at the current moment to the risk state classification module;

[0139] The data fusion analysis module also has the function of updating the standard data matrix using the safety data of the current monitoring cycle;

[0140] The risk status classification module is used to classify the risk status level of the production activities, safety risk modes, and characteristic parameters selected by the user at the current moment, and realize the multi-level risk status classification function in the above steps S225, S36, and S44; preferably, low risk corresponds to a first-level warning, medium risk corresponds to a second-level warning, and high risk corresponds to a third-level warning;

[0141] The multi-level danger warning module is used to issue danger warnings for the identified unsafe risk status, and implement the hierarchical real-time warning function in the above step S8 for different levels of departments such as on-site workers, safety supervisors, and management decision-makers;

[0142] Preferably, the safety status is continuously monitored and controlled, and the unsafe status is warned of danger, which is divided into three levels from low to high: low-risk status corresponds to level I warning, medium-risk status corresponds to level II warning, and high-risk status corresponds to level III warning;

[0143] Preferably, according to the patent "A real-time intelligent monitoring mother machine, method and system for safety status" (application number: 2022114489681), a full-parameter data acquisition module and a data signal sending module are integrated to manufacture an intelligent sensor. It is used to sense parameter status values ​​in real time;

[0144] Preferably, according to the patent "A safe state multi-modal real-time intelligent control mother machine, method and system" (patent number: 2022114412779), the data signal receiving module, periodic data storage module, user basic setting module, data fusion analysis module, risk state classification module, and multi-level danger warning module are integrated to manufacture an intelligent control instrument. It is used for real-time monitoring and warning of multi-modal and multi-parameter safety risks in the control scene.

[0145] An embodiment of the present invention provides a real-time fusion system for multi-modal and multi-parameter monitoring data of process production safety, including: a computer-readable storage medium and a processor;

[0146] The computer-readable storage medium is used to store executable instructions;

[0147] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the method described in any of the above embodiments.

[0148] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method described in any of the above embodiments.

[0149] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time fusion method for multi-modal and multi-parameter monitoring data of process production safety, characterized in that: include: S1, obtain each risk characteristic parameter F i The monitoring fluctuation values ​​Δd at the latest l moments in the current monitoring cycle (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficient ω at the latest l moments (i,j) ; Where i = 1, 2, 3, ..., m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple historical complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient; S2, for ω (i,j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient according to The numerical range in which F is determined i In T p The risk status level is determined and a graded warning is issued; wherein, in the latest l moments, T p The weight coefficient is the largest, and p The closer the moment is, the greater the weight coefficient is; S3, for each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient λ (k,p) , according to λ (k,p) The numerical range in which M is located determines k In T p The risk status level is determined and graded warning is issued; S4, weighted calculation of the risk status evaluation coefficients of all safety risk modes of the monitored process production activity object A is performed to obtain the risk status evaluation coefficients of A at T p The multimodal fusion risk state coefficient θ p , according to θ p The numerical range in which A is located determines the p The risk status level is determined and graded warning is issued; The numerical intervals correspond to the risk status levels one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level.

2. The method according to claim 1, characterized in that The fluctuation range includes a normal range, a low-amplitude abnormal fluctuation range and a high-amplitude abnormal fluctuation range.

3. The method according to claim 2, characterized in that The normal interval is (d` (i,j`) -h (ij`,0) ,d` (i,j`) +h' (ij`,0) ); The low-amplitude abnormal fluctuation range is (d` (i,j`) -h (ij`,1) ,d` (i,j`) -h (ij`,0) ]∪[d` (i,j`) +h ( ' ij`,0) ,d` (i,j`) +h' (ij`,1) ); The high-amplitude abnormal fluctuation range is (-∞,d` (i,j`) -h (ij`,1) ]∪[d` (i,j`) +h' (ij`,1) ,+∞); in, F i The monitoring mean of multiple historical complete monitoring cycles in a safe state, j`=1,2,3,…,n, n is the total number of moments in the complete monitoring cycle, h (ij`,0) 、h' (ij`,0) Take M·σ,h (ij`,1) 、h' (ij`,1) Take N·σ, where σ is F i The standard deviation of the monitoring values ​​of multiple historical complete monitoring cycles in a safe state is M<N, and both M and N are positive integers.

4. The method according to claim 2, characterized in that The normal interval is (d` (i,j`) -h (ij`,0) ,d` (i,j`) +h' (ij`,0) ); The low-amplitude abnormal fluctuation range is (d` (i,j`) -h (ij`,1) ,d` (i,j`) -h (ij`,0) ]∪[d` (i,j`) +h ( ' ij`,0) ,d` (i,j`) +h' (ij`,1) ); The high-amplitude abnormal fluctuation range is Among them, d` (i,j`) F i The accident cause threshold, h (ij`,0) 、h' (ij`,0) For d` (i,j`) B%,h (ij`,1) 、h' (ij`,1) Take C% as the threshold, B<C, and both B and C are positive integers.

5. The method according to claim 1, characterized in that The weight coefficient v of each moment in the latest l moments (i,p-j``+1) ∈[0,1] and 6. The method according to claim 5, characterized in that a p-j``+1 ∈[0,1]anda p-l+1 <a p-l+2 <…<a p-j+1 <…<a p-1 <a p .

7. The method according to claim 1, characterized in that The risk status levels include safe status, low risk status, medium risk status and high risk status; The low-risk state, medium-risk state, and high-risk state correspond to level I, II, and III warnings, respectively.

8. A real-time fusion system for multi-modal and multi-parameter monitoring data of process production safety, characterized in that: include: Data fusion analysis module, used to obtain each risk characteristic parameter F i The monitoring fluctuation values ​​Δd at the latest l moments in the current monitoring cycle (i,j) , according to Δd (i,j) The fluctuation range determines F i The fluctuation coefficient ω at the latest l moments (i,j) ; Where i = 1, 2, 3, ..., m, m is the number of risk characteristic parameters, j = p-l+1, p-l+2, ..., p, p is the total number of moments in the current monitoring cycle, p ≥ 3, l ≤ p; Δd (i,j) F i The monitoring mean d of the time corresponding to the latest l time in multiple historical complete monitoring cycles in a safe state (i,j) With F i The latest monitoring value at the moment in the current monitoring cycle The fluctuation range corresponds to the fluctuation coefficient one by one, and the minimum value of the fluctuation range with a large fluctuation coefficient is greater than the maximum value of the fluctuation range with a small fluctuation coefficient; for ω (i,j) Perform weighted calculation to get F i At the current time T p Risk status evaluation coefficient For each security risk mode M k The risk status evaluation coefficients of all risk characteristic parameters are weighted and calculated to obtain M k In T p The risk status evaluation coefficient λ (k,p) ; The risk state evaluation coefficients of all safety risk modes of the monitored process production activity object A are weighted and calculated to obtain A at T p The multimodal fusion risk state coefficient θ p ; Risk status classification module is used to The numerical range in which F is determined i In T p The risk status level of (k,p) The numerical range in which M is located determines k In T p The risk status level of p The numerical range in which A is located determines the p wherein the numerical intervals correspond to the risk status levels one by one, and the minimum value of the numerical interval with a high risk status level is greater than the maximum value of the numerical interval with a low risk status level; Multi-level hazard warning module is used to i In T p The risk status level is graded and early warning is carried out according to M k In T p The risk status level is determined and a graded warning is given. p The risk status level can be determined and graded warning can be issued.

9. The system according to claim 8, characterized in that Also includes: The full parameter data collection module is used to collect the change data of all safety risk modal risk characteristic parameters covering the process production activity objects selected by the user; A data signal sending module, used to send the collected parameter data to the data signal receiving module using a wireless network communication protocol; A data signal receiving module, used to receive monitoring data of various risk characteristic parameters from the signal sending module; The periodic data storage module is used to store the monitoring data of all risk characteristic parameters in the complete process operation cycle with parameter type and monitoring time as identification; The user basic setting module is used for users to select the production activity objects and safety risk mode types to be monitored, and to set the risk characteristic parameters and the weights of each parameter according to the production process requirements.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method according to any one of claims 1 to 6.

Images